Structural Concrete 01/2016 free sample copy

1Volume 17 March 2016 ISSN 1464-4177

- Shear strength of SCC beams with few stirrups- Scatter in shear capacity- Method for computing RC beam arch capacity against collapse- Behaviour of RC beams with CFRP-strengthened openings- Evaluation of safety formats for non-linear FEA- Transfer lengths in precast pretensioned concrete members- Concrete fatigue in composite dowels- ASR effects on properties of concretes with different aggregates- ASR and sulphate performance of mortar including waste- Nano-indentation and XCT for investigating sulphate damage- Properties of concrete with blended cement/ceramic powder- Ultrafine fly ash effect on concrete with aggregates made from waste

��

��

March 2016ISSN 1464-4177 (print)ISSN 1751-7648 (online)

3Bautechnik 81 (2004), Heft 1

Contents

Structural Concrete Vol. 17 / 1

Editorial

1 Hans BeushausenPredicting the behaviour of concrete structures – modelling or testing?

Technical Papers

3 Thomás Lima de Resende, Lidia da Conceição Domingues Shehata, Ibrahim Abd El Malik ShehataShear strength of self-compacting concrete beams with small stirrups ratios

11 Filippo Sangiorgio, Johan Silfwerbrand, Giuseppe ManciniScatter in the shear capacity of slender RC members without web reinforcement: an overview study

21 Reza Abbasnia, Foad Mohajeri NavA theoretical method for calculating the compressive arch capacity of RC beams against progressive collapse

32 Siew Choo Chin, Nasir Shafiq, Muhd Fadhil NuruddinBehaviour of RC beams with CFRP-strengthened openings

44 Mattias Blomfors, Morten Engen, Mario PlosEvaluation of safety formats for non-linear finite element analyses of statically indeterminate concrete structures subjected to different load paths

52 Sun-Jin Han, Deuck Hang Lee, Sang-Heum Cho, Soon-Beum Ka, Kang Su KimEstimation of transfer lengths in precast pretensioned concrete members based on a modified thick-walled cylinder model

63 Martin Classen, Joerg GallwoszusConcrete fatigue in composite dowels

74 Okpin Na, Yunping Xi, Edward Ou, Victor E. SaoumaThe effects of alkali-silica reaction on the mechanical properties of concretes with three different types of reactive aggregate

84 Ana Mafalda Matos, Joana Sousa-CoutinhoASR and sulphate performance of mortar containing industrial waste

96 Chunxiang Qian, Yanfeng Nie, Tianji CaoSulphate attack-induced damage and micro-mechanical properties of concrete characterized by nano-indentation coupled with X-ray computed tomography

105 Tereza Kulovaná, Eva Vejmelková, Martin Keppert, Pavla Rovnaníková, Zbynek Keršner, Robert CernýMechanical, durability and hygrothermal properties of concrete produced using Portland cement-ceramic powder blends

116 Faiz ShaikhEffect of ultrafine fly ash on the properties of concretes containing construction and demolition wastes as coarse aggregates

125 fib-news

A5 Products and Projects

The Tsubasa Bridge in Neak Loeung is a three span stay cable bridge crossing the Mekong River in Cambodia. With a main span of 640 m the bridge is not only a new landmark of the region, but also Cambodia´s longest bridge. Construction of this bridge was due to the in-creasing congestion on the important National Highway 1 in Cambodia, which connects the capital of Phnom Penh with the Bavet border crossing to Vietnam. DYWIDAG Stay Cables as well as Strand Post-Tensioning Systems and Bar Post-Tensioning Systems support the construction of the Tsubasa Bridge. (© DSI, see p. A5–A6)

fédération internationale du bétonInternational Federation for Structural Concrete www.fib-international.org

Journal of the fib

Peer reviewed journalSince 2009, Structural Concrete is indexed in Thomson Reuter’s Web of Knowledge (ISI Web of Science).

Impact Factor 2014: 1.492

www.ernst-und-sohn.de/structural-concrete

http://wileyonlinelibrary.com/journal/suco

The journal “Structural Concrete”, the official journal of the International Federation for Structural Concrete (fib – fédération internationale du béton), provides conceptual and procedural guidance in the field of concrete construction, and features peerreviewed papers, keynote research and industry news covering all aspects of the design, construction, performance in service and demolition of concrete structures.

“Structural Concrete” is published four times per year completely in English.

Except for a manuscript, the publisher Ernst & Sohn purchases exclusive publishing rights. Only works are accepted for publication, whose content has never been published before. The publishing rights for the pictures and drawings made available are to be obtained from the author. The author undertakes not to reprint his article without the express permission of the publisher Ernst & Sohn. The “Notes for authors” regulate the relationship between author and editorial staff or publisher, and the composition of articles. These can be obtained from the publisher or in the Internet at www.ernstundsohn.de/en/journals.

The articles published in the journal are protected by copyright. All rights, particularly that of translation into foreign languages, are reserved. No part of this journal may be reproduced in any form without the written approval of the publisher. Names of brands or trade names published in the journal are not to be considered free under the terms of the law regarding the protection of trademarks, even if they are not individually marked as registered trademarks.

Manuscripts can be submitted via ScholarOne Manuscripts at www.ernst-und-sohn.de/suco/for_authors

Discussions on papers published in Structural Concrete can be submitted. Please read our author guideline for more information.

If required, special prints can be produced of single articles. Requests should be sent to the publisher.

Publisherfib – International Federation for Structural ConcreteCase Postale 88, CH1015 Lausanne,Switzerlandphone: +41 (0)21 693 2747, fax: +41 (0)21 693 6245email: [email protected], Website: www.fibinternational.org

Publishing houseWilhelm Ernst & SohnVerlag für Architektur und technische Wissenschaften GmbH & Co. KGRotherstraße 2110245 Berlin/Germanyphone: +49 (0)30/47031200fax: +49 (0)30/47031270email: info@ernstundsohn.de, Website: www.ernstundsohn.de

Managing editorFrancisco Velasco, Verlag Ernst & SohnRotherstraße 21, D10245 Berlinphone: +49 (0)30/47031277, fax: +49 (0)30/47031227email: [email protected]

Advertising managerFred Doischer, Verlag Ernst & Sohnphone: +49 (0)30/47031234

AdvertisingJohannes Krätschell, Verlag Ernst & SohnRotherstraße 21, D10245 Berlinphone: +49 (0)30/4 70 312 42, fax: +49 (0)30/4 70 312 30email: [email protected]

Layout and typesetting: TypoDesign Hecker GmbH, Leimen Printing: ColorDruck Solutions GmbH, Leimen

Editorial boardEditor-in-Chief Luc Taerwe (Belgium), email: [email protected]

Deputy Editor Steinar Helland (Norway), email: [email protected]

Members György L. Balázs (Hungary) Josée Bastien (Canada) Oguzhan Bayrak (USA) Mikael Braestrup (Denmark) Michael Fardis (Greece) David FernándezOrdóñez (Secretary General fib) Stephen Foster (Australia) Sung Gul Hong (Korea) Tim Ibell (UK) S.G. Joglekar (India) Akio Kasuga (Japan) Daniel A. Kuchma (USA) Giuseppe Mancini (Italy) Pierre Rossi (France) Guilhemo Sales Melo (Brazil) Tamon Ueda (Japan) Yong Yuan (China)

Current pricesThe journal Structural Concrete has four issues per year. In addition to “Structural Concrete print”, the PDF version “Structural Concrete online” is available on subscription through the online service Wiley Online Library.

print print print + online print + online Single copy (personal) (institutional) (personal) (institutional) (print)

184.00 € 764.00 € 221.00 € 917.00 € 52.00 €

Other currencies and bulk discounts are available on request.Members of the fib receive the journal Structural Concrete as part of their membership.Prices exclusive VAT and inclusive postage, errors and omissions excepted. Subject to change without notice. Prices are valid until 31 August 2016.A subscription lasts for one year. It can be terminated in writing at any time with a period of notice of three months to the end of the calendar year. Otherwise, the subscription extends for a further year without written notification.

Bank detailsCommerzbank AG Mannheimaccount number 751118800bank sort code 67080050SWIFT: DRESDEFF670Structural Concrete, ISSN 14644177, is published quarterly. US mailing agent: SPP, PO Box 437, Emigsville, PA 17318. Periodicals postage paid at Emigsville PA.Postmaster: Send all address changes to Structural Concrete, John Wiley & Sons Inc., c/o The Sheridan Press, PO Box 465, Hanover, PA 17331.

Service for customers and readersWileyVCH Customer Service for Ernst & SohnBoschstrasse 12, D69469 WeinheimTel.: +49 (0)800 1800 536 (within Germany)Tel.: +44 (0)1865476721 (outside Germany)Fax: +49 (0)6201 606184cs[email protected]

Quicklink: www.wileycustomerhelp.com

© 2016 Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin

Imprint

Structural Concrete 17 (2016), No. 1

Inserts in this issue: World Building Congress 2016; Verlag Ernst & Sohn GmbH & Co. KG, 10245 Berlin

A5Responsible for Products & Projects: Publishing House Ernst & Sohn Structural Concrete 17 (2016), No. 1

Products & Projects

The Tsubasa Bridge: DYNA Grip® StayCables support Cambodia’s longest Bridge

With a length of approx. 167 km, National Highway 1 in Cam-bodia connects the capital of Phnom Penh with the Bavetborder crossing to Vietnam. Due to the increasing congestion

on this important route, a decision was made to construct theTsubasa Bridge in Neak Loeung 60 km south-east of PhnomPenh. The structure is a three span stay cable bridge crossingthe Mekong River.

The project also included the construction of two 900 m and675 m long approach bridges. Both structures consist of

35 spans, each of which is 45 m long. Theprecast concrete girders were fabricated in aconstruction yard and installed using twosets of erection girders. The precast girderswere post-tensioned using Type 12S 15.2DYWIDAG Strand Tendons, and Type3S 15.2 DYWIDAG Strand Tendons wereused for transverse post-tensioning. The main bridge is 640 m long and dividedinto three spans with lengths of 155 m,330 m and 155 m. The two pylons are 121 mhigh measured from the pile cap and consistof two individual pillars on the outside ofthe bridge that are connected by severalcross beams. The pylons rest on cast-in-place, 2.5 m ∅ concrete piles.The main girder has a cross-section of 17 mand a height of 1.8 m. The bridge segmentsadjacent to the pylons were built first so thatthe stay cables could be installed at both py-lons and anchored at the bridge deck. Thisway, the stay cables supported the weight ofthe form traveler during the construction ofthe bridge sections between the pylons.These were constructed using an underslung

form traveler in a 10 day construction cycle. In total, 930 t ofType DG-P61, 55, 37, 31 and 22 DYNA Grip® Stay Cables wereinstalled. The stay cables are galvanized, waxed and inserted inHDPE ducts. In the main span area, transverse post-tensioningwas carried out using 170 t of 32 mm ∅ DYWIDAG Bar Ten-dons. The individual cross beams were stressed using 510 t ofType 12S 15.2 horizontal DYWIDAG Strand Tendons.Furthermore, 140 t of Type 19S15.2 DYWIDAG Strand Post-Ten-sioning Systems and 200 t of Types 5S 15.2, 7S 15.2 and 9S 15.2DYWIDAG Strand Post-Tensioning Systems were used in the py-lons. Additionally, 180 t of 36 mm ∅ DYWIDAG Bar Tendonswere installed in the pylons. The Tsubasa Bridge is not only anew landmark of the region, but also the longest bridge in Cambodia.

Construction board:Owner: Ministry of Public Works and Transport, Kingdom ofCambodia, Cambodia

Fig. 2. View from the top of the pylon

Fig. 3. Installation of the Stay Cables

Fig. 1. General view

Fig. 4. Main span of the bridge (© DSI)

General Contractor: Sumitomo Mitsui Construction Co., Ltd.,JapanArchitect: Chodai Co., Ltd and Oriental Consultants Co., Ltd,JapanConsulting: Chodai Co., Ltd and Oriental Consultants Co., Ltd,JapanDSI Units: DYWIDAG-Systems International GmbH, GBU,Germany/Sumitomo (SEI) Steel Wire Corp., JapanSubcontractor: DYWITECH Co. Ltd., Taiwan

DSI and Sumitomo Scope: Production, supply, installation, engi-neering services, technical supportDYWIDAG Products: Type DG-P61, 55, 37, 31 and 22 DYNAGrip® Stay Cables; Type 12S 15.2, 3S 15.2, 19S 15.2, 5S 15.2,7S 15.2 and 9S 15.2; DYWIDAG Strand Tendons; 32 mm and36 mm ∅ DYWIDAG Bar Tendons

www.dywidag-systems.com

A6 Structural Concrete 17 (2016), No. 1 Responsible for Products & Projects: Publishing House Ernst & Sohn

Products & Projects

The new PUMIs with stepless support and other innovations at bauma 2016At bauma 2016 in hall B6 booth 100, Putzmeister will presentthe next generation of its tried and proven truck mixer con-crete pumps: the PUMI 25-4 and the 28-4 NEW Generation.The most important advancements include the unique ste-pless support, the 4-arm distributing boom, the new S-pump,the Ergonic 2.0 control and technologies for reduced opera-ting costs and exhaust and noise emissions.

The PUMI 25-4 and the 28-4 NEW Generation are causing a stirwith numerous innovations. The stepless support, which offersincredible flexibility on constricted sites in particular, is an ab-solute novelty. Just like the 4-arm distributing boom with opti-mised slip characteristics and maximum working range. Easier and safer operation, thanks not least to the extensivelighting concept and the new computer-assisted Ergonic 2.0 con-trol. The newly developed S-piston pump with large hopper, ide-al accessibility and optimised performance data was developedspecifically for PUMI applications.In order to counteract wear, Putzmeister has further refined thedesign of its machines. The torsion-resistant integral frame andshape-optimised mixer drum ensure a consistent axle load distri-bution, preserving the machine over the long term and keepingoperating costs low. The new PUMIs are proving to be eco-friendly and economic thanks to lower operating fluid consump-tion and reduced exhaust and noise emissions. As usual, the PUMI 25-4 and PUMI 28-4 NEW Generation areavailable with either piston or rotor pump.

Innovative stepless support The stepless support allows a flexible and safe machine setup.The ESC (Ergonic® Setup Control), Putzmeister safety system,guarantees permanent control at all times through the interac-tion between support, boom movements and pump function. Inaddition to the full support, the machine can also be setup withone side support. The 4-arm distributing boom with Z-fold offers optimised slipcharacteristics. The perfected kinematics enlarges the spatialworking range and prevents “dead space”. This increases the ef-fectively attainable operating reach. A 125 mm diameter deliveryline supplies the required amount of concrete.The direct response characteristics of the boom control, theminimised boom vibrations, the arrangement of the deliverylines and the rigidity of the steel structure guarantee extremelyprecise concrete placement.

Operation and service – simple and safe The PUMI 25-4 and the 28-4 NEW Generation are the firsttruck mixer concrete pumps to be fitted with the Ergonic® 2.0control. An operating concept developed specifically for the ma-

chines makes operation of new PUMIs intuitive and hence verysimple. The machine support is controlled by the Ergonic®

Setup Control (ESC) safety system. The support ranges are dis-played to the operator, thereby minimising operating errors.Other operator-friendly details, such as the new extensive lightconcept for illuminating the operating positions and the supportranges, allow work to be completed more conveniently andfaster. The new models also offer extensive equipment for activesupport during use and afterwards.The latest generation models are recognisable even by their al-tered appearance. The shape-optimised mixer drum aids opti-mised axle load distribution. Additionally, the torsion-resistantintegral frame ensures force is optimally transferred from boomto frame, from where it is conducted directly into the stabilisers.The incredibly stability and minimum movements during thepumping operation minimise wear and hence the associatedcosts.The S-piston pump, designed specifically for the new PUMIs, isanother innovative development. A large hopper, optimised ac-cessibility to the water box and top performance data ensure aconcrete pump that is ideally suited to the new PUMIs.

Eco-friendliness pays offLower exhaust and noise emissions, as well as reduced dieseland hydraulic oil consumption are easy on the budget and onthe environment. This is made possible by a new drive conceptand larger hydraulic pump. So the machine operates at a lowerengine speed while the pump is operating. The new filter con-cept, which significantly reduces the number of oil changes, is also eco-friendly.

www.putzmeister.de

The stepless support allows a flexible and safe machine setup, especiallyon constricted sites. (© Putzmeister)

Special offer 2016Selected Ernst & Sohn books at a special rate!

20 % discount

Order code: EUS20

Ernst & Sohn is offering a discount on selected English books. A selection of the available titles is shown here. To fi nd a complete overview visit the given website.

You can take advantage of 20% off the regular price for all orders until 30th of June 2016. Please be advised to enter the order code.

10

88

16

6_d

p

Title overview and orders: www.wiley.com/go/eus20

Customer Service: Wiley-VCH

Boschstraße 12

D-69469 Weinheim

Ernst & Sohn

Verlag für Architektur und technische

Wissenschaften GmbH & Co. KG

Tel. +49 (0)800 1800 536

Fax +49 (0)6201 606-184

[email protected]

Foz Tua Dam, Vila Real – Alijó, Portugal

PERI devised and supplied an optimised and safe formworkand scaffolding solution for construction of the Foz Tua Damand the associated pumped-storage power station. The manu-factorer’s specialists supported the construction firm, Bar-ragem de Foz Tua, ACE, to ensure the build was completedon schedule and with precision, providing an elaborate con-cept, as well as logistical and technical services. This includedcontinual coordination on site by the company’s Project Man-ager in addition to the engineers’ close collaboration with allproject participants.

The dam in northern Portugal will reach a height of 108 m,while the dam crest measures 275 m long. From a formwork per-spective, the particular challenge is caused through the complexdesign of the double curvature reinforced concrete structurewith the 5-m-wide dam crest and its integrated overflow mecha-nism. The hydraulic circuit covers over 700 m along the right-hand bank of the river and includes 2 independent tunnels, aswell as the subterranean power station with 2 reversible units.

Material and support from one sourceWith a complex project of this size the issue is to coordinate thediverse requirements of the build especially with a tight scheduleand high material requirements. In addition to planning theformwork and scaffolding solution for complex geometries and

high loads, one of the biggest challenges is strict compliancewith the construction phase and cost plan for the gigantic damproject. To support the build PERI also deploys a Project Man-ager to the construction site, who acts as liaison between theTechnical Office and the enormous construction team. Whilekeeping in continual contact, this Project Manager makes sureall processes and works run smoothly in relation to formworkand scaffolding technology – especially for supply logistics aswell. Thus, the profitability of the construction site is not onlyassured by the project-specific optimised equipment, but also bythe continual controls and adjustment to material volumes onsite. All processes relating to planning, logistics and formworkassembly are scheduled in detail by the company’s Project Man-ager and coordinated in depth with the actual construction se-quence.This PERI overall solution of planning, material provision andproject management removes many interfaces and reduces fric-tional losses in the construction sequence. The contractor bene-fits from increased profitability in terms of time and costs.

The PERI formwork and scaffolding solutionPERI’s engineers devised a solution of the SCS Climbing Systemand the VARIO GT 24 Wall Formwork for the double curvature

A8 Structural Concrete 17 (2016), No. 1 Responsible for Products & Projects: Publishing House Ernst & Sohn

Products & Projects

1009116_dp

Do you require special prints of selected articles from the journals at Ernst & Sohn?Please contact: Janette SeifertVerlag Ernst & Sohn Rotherstraße 21, 10245 BerlinTel +49(0)30 47031-292Fax +49(0)30 47031-230E-Mail [email protected]

www.ernst-und-sohn.de/en/reprints

Fig. 1. PERI supplied a project-specific formwork and scaffolding solutionand supported with comprehensive engineering services for the construc-tion of the 108-m-high complex dam on the River Tua in Portugal.

Fig. 2. With the SCS Climbing System the loads from the fresh concretepressure are transferred through the bracket into the climbing anchor ofthe previous concreting step. The horizontal alignment of the working plat-forms on the PERI SCS Climbing System ensure the highest working safetywhen used in inclined situations.

Fig. 3. An extremely complex formwork and falsework design is requiredfor the dam’s overflow system. PERI’s engineers devised a concept basedon the system components of the VARIOKIT Engineering Construction Kit,the SB Brace Frame and various falsework elements. The solution was es-pecially cost-effective as all these components are available in the PERIrental park.

dam. This combination is the ideal solu-tion for the massive dam, as it facilitatesboth the single-face load transfer and thehigh finish requirements.SCS enables single-face concreting bytransferring the loads from the fresh con-crete pressure without formwork anchorsthrough the bracket into the climbing an-chor of the previous concreting step.Thanks to the modular concept with mul-ti-piece brackets, the SCS Climbing Sys-tem can easily be adapted to the structur-al geometry. Large climbing units are pos-sible due to the special load capacity ofthe brackets and high load factor of theanchoring, which ensures rapid progresson site. The working platforms with auseful width of 1.90 m remain horizontaldespite an inclined position on the dam,creating extremely secure working condi-tions for users.The highly flexible VARIO GT 24 GirderFormwork makes units with the platformsthat can be moved by crane. The plannedproject-specific wall formwork also en-sures the best architectural concrete re-sults thanks to the planned arrangementof joints.PERI also supplied a modified climbingand falsework solution for the highlycomplex geometry, such as the overflowarea of the dam crest. The well-engi-neered design consists of systems compo-nents from the manufactorer’s rentalpark, making it particularly cost-effective:a combination of SB Brace Frames andsystem components of the VARIOKITEngineering Construction Kit is used to

form large platforms anchored in thehardened dam wall. These serve as an in-stallation area for MULTIPROP ShoringTowers and SLS Heavy-Duty Props,which in turn support the inclined form-work units based on the VARIO GT 24Girder Formwork.The tried-and-tested TRIO Panel Form-work is the fast and best solution for theless complex wall structures of the dam,such as in the area of the dam galleries.The universal formwork system is de-signed for simple shuttering and to re-duce shuttering times.Moreover, the PERI UP Modular Scaffoldis being used for diverse falsework andworking scaffold tasks and for access so-lutions. The scaffold system with its met-ric grid arrangement and a multitude ofpractical details ensures secure erectionand dismantling, as well as high safetyduring use. PERI UP Stairs with installa-tion heights up to 55 m also ensure fastand secure accessibility to the working ar-eas.PERI Portugal also devised the rightformwork solution for the reinforced con-crete pipes of the hydraulic circuit. Thetwo 700-m-long tunnels have varyingcross sections from 5.50 m to 7.50 m. Astandardised, self-supporting steel frameis being used as the basis for the twoformwork carriages. The steel formliningis adapted based on the tunnel cross sec-tion. The equipment for manual process-es and hydraulic control were planned inaccordance with the specific constructionsite requirements. The steel formworkcarriage enables the construction team toachieve the best concrete finish. The ex-tremely robust structure is also very cost-effective thanks to the high number ofuse cycles.

About the Foz Tua pumped-storagepower stationThe Foz Tua pumped-storage power sta-tion is being built on the River Tua, animportant tributary of the Douro. As partof the Portuguese national energy planand with an investment volume of 370mio. €, the power station is intended toradically increase energy availability inthe region. Construction work on thedam began in 2011 and is expected totake 5 years. Commissioning of thepumped-storage power station is alreadyplanned for 2016. The main investor EDPEnergias de Portugal developed a de-tailed concept in collaboration with theexperienced project offices COBA,QUADRANTE and architect SoutoMoura, to minimise the environmentalimpact of this significant building project.

www.peri.de

Responsible for Products & Projects: Publishing House Ernst & Sohn

Products & Projects

Structural Analysis and Design

Up-to-Date Information...

Free Trial Version atwww.dlubal.com

FurtherInformation:

Dlubal Software GmbHAm Zellweg 2, D-93464 TiefenbachTel.: +49 9673 9203-0Fax: +49 9673 [email protected]

DESIGN according to Eurocode 2,DIN 1045, ACI 318, SIA 262, GB 50010RF-/DYNAM Pro: New add-on modules for analysis of natural vibrations, analysis of forced vibrationsand seismic analyses

The Ultimate FEA Program

Steel Construction

Solid Construction

Co

lum

n B

ases

3D Finite Elements

BIM

/ Eu

roco

des

Stab

ility

an

d D

ynam

ics

© www.ssp-muc.com

3D Frameworks

Cro

ss-S

ecti

on

s

Follo

w u

s on

:

The Structural Beam Analysis Program

Bridge Construction

© www.wtm-engineers.de

© w

ww

.mgm

-ki.p

l

Fig. 4. One of the biggest challenges is to stickto the construction phase and cost plan. For thisreason, work also continued through the nighton this complex project of immense proportions.(© PERI)

A10 Structural Concrete 17 (2016), No. 1

software

Dlubal Software GmbHAm Zellweg 293464 TiefenbachPhone +49 (0) 96 73 92 03-0Fax +49 (0) 96 73 92 03-51Mail: [email protected]: www.dlubal.de

stay cables

DYWIDAG-Systems International GmbHMax-Planck-Ring 140764 Langenfeld/GermanyPhone +49 (0)21 73/7 90 20Mail: [email protected]: www.dywidag-systems.de

vibration isolation

BSW GmbHAm Hilgenacker 24 D-57319 Bad Berleburg Phone +49(0)2751 803-126 Mail: [email protected] Web: www.bsw-vibration-technology.comunder-screed impact sound insulation with European Technical Approval, PUR foam & PUR rubber materials for vibration isolation

reinforcement technologies

HALFEN Vertriebsgesellschaft mbHLiebigstraße 14D-40764 LangenfeldPhone +49 (0) 21 73 9 70-0Fax +49 (0) 21 73 9 70-2 25Mail: [email protected]: www.halfen.deconcrete: fixing systems facade: fastening technology framing systems: products and systems

Max Frank GmbH & Co. KGTechnologies for the construction industryMitterweg 194339 Leiblfing GermanyPhone +49 (0)94 27/1 89-0 Fax +49 (0)94 27/15 88 Mail: [email protected] Web: www.maxfrank.com

sealing technologies

Max Frank GmbH & Co. KGTechnologies for the construction industryMitterweg 194339 Leiblfing GermanyPhone +49 (0)94 27/1 89-0 Fax +49 (0)94 27/15 88 Mail: [email protected] Web: www.maxfrank.com

Provider directoryproducts & services

bridge accessories

Maurer AGFrankfurter Ring 193 D-80807 München Phone +49(0)89 32394-341 Fax +49(0)89 32394-306 Mail: [email protected] Web: www.maurer.euStructural Protection Systems Expansion Joints Structural Bearings Seismic Devices Vibration Absorbers

literature

Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KGRotherstraße 21 10245 Berlin Phone +49 (0) 30 4 70 31-2 00 Fax +49 (0) 30 4 70 31-2 70 E-mail: [email protected] Web: www.ernst-und-sohn.de

fastening technology

HALFEN Vertriebsgesellschaft mbHLiebigstraße 14D-40764 LangenfeldPhone +49 (0) 21 73 9 70-0Fax +49 (0) 21 73 9 70-2 25Mail: [email protected]: www.halfen.deconcrete: fixing systems facade: fastening technology framing systems: products and systems

post-tensioning

DYWIDAG-Systems International GmbHMax-Planck-Ring 140764 Langenfeld/GermanyPhone +49 (0)21 73/7 90 20Mail: [email protected]: www.dywidag-systems.de

prestressed concrete

Paul Maschinenfabrik GmbH & Co. KGMax-Paul-Straße 188525 Dürmentingen/GermanyPhone +49 (0)73 71/5 00-0Fax +49 (0)73 71/5 00-1 11Mail: [email protected]: www.paul.eu

1© Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin · Structural Concrete 17 (2016), No. 1

Judging by the abstracts submitted for this year’s fib Symposium (November 2016), international research on concrete materials and concrete structural technology is primarily concerned with high-performance solutions for concrete mix design, reinforc-ing materials and structural systems. As concrete materials and structural systems become more complex, more “modern”, so their behaviour also becomes increasingly difficult to predict with conventional materials models or analytical methods. The new edition of the fib Model Code for Concrete Structures 2010 makes allowance for this as it explicitly includes performance evaluation philosophy as well as conventional prescriptive design methods and established analytical models.

For the research and development of concrete materials and structural systems, performance assessment through experi-mental investigations represents a well-accepted and frequently used approach. However, in practice, many design engineers prefer to rely on established analytical models and tend to avoid approaches that involve experimental investigations. While this might be acceptable and practical for standard design situations, experimental investigations become necessary for the design and conformity assessment of new materials and systems.

One of the most critical foundations for the application of performance approaches in practice is the development of reliable test methods that deliver reproducible results and relate to the behaviour of the as-built concrete structure. The testing of me-chanical material properties is generally based on well-established methods and linked to accepted analytical models. In contrast, considerable work is still needed in order to develop reliable test methods and interpretation criteria for concrete durability properties. For example, despite the fact that reinforcement cor-rosion is probably the most significant threat to the durability and structural performance of the built infrastructure, associated test methods and service life models still contain an abundance of uncertainties.

The prediction of reinforcement corrosion is based on mod-elling chloride ingress and carbonation in concrete. Considering chloride ingress, the expected material performance is typically based on diffusion models and the determination of diffusion coef-ficients by testing. However, diffusion is just one of several trans-port mechanisms responsible for chloride ingress into concrete and may not describe the performance of the material adequately. For application in practice, the associated service life models rely on ambiguous assumptions for chloride surface concentra-tions and environmental exposure conditions. In addition, they do not take sufficient account of other parameters that influence reinforcement corrosion, such as the concrete’s electrical resistiv-ity, cover depth, cracking characteristics, moisture and oxygen availability, temperature, etc. This has resulted in practising engi-

Predicting the behaviour of concrete structures – modelling or testing?

Editorial

Hans Beushausen

neers and concrete producers frequently voicing criticism of exist-ing approaches to durability design. In turn, researchers often re-spond by developing ever more complicated models and applying increasingly sophisticated statistical approaches (which, however, do not necessarily improve the accuracy of the models).

As research progresses, so it becomes increasingly clear just how complex the task of durability modelling really is. We will probably soon conclude that it is time to simplify our approaches, considering that the sophisticated models have not really im-proved the reliability of durability predictions. However, safe simplifications can only be made once the fundamental material behaviour has been understood.

The simplified modelling of complex material behaviour represents one of the most powerful tools for engineering design in practice. This is well demonstrated by the oldest and most widely used performance-based approach for the specification and conformity assessment of concrete material properties – the compressive strength test. This test was developed in the first half of the previous century and initially widely criticized for not representing the material property of the as-built structure. It was argued that a small cubic or cylindrical concrete specimen subjected to short-term loading between two steel platens cannot represent the complex loading regimes and stress distributions of structural members subjected to real loads. However, despite its obvious limitations, this test has been successfully used worldwide for decades. The underlying aim of this test is to contribute to the design and construction of structurally safe structures, which has so far been a great success. Similarly, the ultimate aim of studies of concrete deterioration and service life modelling is to increase the durability of our built environment. And in this respect, major improvements have been made in recent years.

Many of the more than 350 abstracts submitted for the fib Symposium 2016 relate to performance assessment, with top-ics covering structural behaviour, concrete durability and also strengthening and repair. I would like to use this opportunity to extend a wholehearted invitation to you to attend the fib Sympo-sium in Cape Town and participate in the exchange of knowledge in the aforementioned fields of concrete materials and structural concrete.

Associate Professor Hans BeushausenUniversity of Cape Town, Department of Civil EngineeringOrganizing Committee fib Symposium 2016, Cape Town, South Africa

2 Structural Concrete 17 (2016), No. 1

Editorial

3© 2016 Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin · Structural Concrete 17 (2016), No. 1

Technical Paper

DOI: 10.1002/suco.201400084

In comparison with a vibrated concrete (VC) of the same strength class, self-compacting concrete (SCC) typically has a lower coarse aggregate content and, possibly, a smaller maximum aggregate size. This may result in reduced aggregate interlock between the fracture surfaces of a SCC. Since aggregate inter-lock plays an important role in the shear strength of slender beams, SCC beams may have a shear strength lower than that of similar VC beams, but studies on that subject are still limited.This article summarizes an experimental programme that in-cludes beams of high-strength SCC and transverse reinforcement ratios around the minimum given by different codes – a case that had not been investigated so far. The shear strengths of those SCC beams are compared with those of VC beams with similar concrete compressive strength and small ratios of transverse reinforcement and also compared with beams calculated accord-ing to different code procedures.

Keywords: self-compacting concrete beams, shear strength, minimum stirrups ratio, slender beams

1 Introduction

The use of self-compacting concrete (SCC) is steadily in-creasing, mainly in the precast industry, and a large amount of research has been conducted on the fresh and hardened properties of SCC. However, relatively little re-search has been carried out on the structural behaviour of SCC.

As far as shear behaviour is concerned, results from different research projects show contradictory results. Some show that SCC and vibrated concrete (VC) beams with the same characteristics have a similar shear strength, whereas, according to others, SCC beams have a lower shear strength.

That is probably due to the different parameters that affect the shear strength of beams and also the different possible SCC compositions. In order to obtain the neces-sary flowability of the concrete, it is more usual to opt for increasing the powder content and reducing the coarse

aggregate content and for rounder aggregates or a smaller maximum aggregate size. If a viscosity-modifying admix-ture is used, however, SCC may have a coarse aggregate content of the same order as that of VC, but the use of that admixture in SCC is not common practice among ready-mix concrete suppliers in Brazil.

In comparison with a vibrated concrete of the same strength class, the reduction in coarse aggregate content and, possibly, the maximum aggregate size in SCC may result in reduced aggregate interlock between fracture sur-faces. However, this also depends on the paste and inter-facial transition zone characteristics, which tend to be denser and more uniform than in VC.

The ultimate nominal shear stress of slender beams without transverse reinforcement depends mainly on the concrete strength, the aggregate interlock between sur-faces of cracks, the effective depth (size effect) and the longitudinal reinforcement ratio. Aggregate interlock is affected by the roughness of the crack interfaces, which depends on the type and size of the aggregate, as well as crack width. In high-strength concrete beams, cracks can pass through the aggregates instead of propagating around them, thus reducing the roughness of crack interfaces and, consequently, the interlocking capacity [1, 2]. In order to avoid non-conservative predictions when using high-strength concrete beams, some code procedures limit the concrete strength or maximum aggregate size to be con-sidered in shear strength equations. The UK National An-nex to Eurocode 2 [3] limits fck to 50 MPa unless other-wise justified and the fib Model Code for Concrete Structures 2010 [4] considers aggregate maximum size to be equal to zero when the concrete strength exceeds 70 MPa.

Beams with a higher percentage of longitudinal rein-forcement have a higher shear capacity, which can be at-tributed to a combination of additional dowel action and smaller crack widths, resulting in increased aggregate in-terlock, and a larger concrete compression zone.

For the same longitudinal reinforcement ratio and aggregate size, larger beams have wider cracks and aggre-gate interlock becomes less effective. Transverse reinforce-ment itself contributes to shear strength and enhances the contribution of other shear transfer mechanisms, restricts the widening of shear cracks and may mitigate that size effect on the shear strength of beams, but not suppress it [5].

Thomás Lima de ResendeLidia da Conceição Domingues Shehata*Ibrahim Abd El Malik Shehata

DOI: 10.1002/suco.201400084

* Corresponding author: [email protected]

Submitted for review: 22 September 2014; revision: 27 March 2015; accepted for publication: 28 March 2015. Discussion on this paper must be submitted within two months of the print publication. The discussion will then be published in print, along with the authors’ closure, if any, approximately nine months after the print publication.

Shear strength of self-compacting concrete beams with small stirrups ratios

4

T. Lima de Resende/L. da Conceição Domingues Shehata/I. Abd El Malik Shehata · Shear strength of self-compacting concrete beams with small stirrups ratios


The shear force carried by aggregate interlock tends to decrease with increasing concrete strength and beam depth. However, experimental studies including SCC beams with great depth and high concrete strength were not found.

This paper summarizes the results of an experimen-tal study on the shear strength of reinforced high-strength SCC beams (fc = approx. 70 MPa), 500 mm deep and with small amounts of transverse reinforcement – a case that had not been studied so far. The variables of the beams were the longitudinal tensile reinforcement ratio (2.0 or 2.5 %), the diameter (4.2 or 6.3 mm) and the transverse reinforcement index (ρwfyw = 0.508–0.975 MPa). The com-parison of the shear strength of those beams with the those of VC beams with similar compressive strength and coarse aggregates tested by Garcia [24] shows possible dif-ferences between the shear strength of SCC and VC beams with small web reinforcement indexes. A detailed description of the study is given by Resende [6].

2 Experimental programme2.1 Properties of materials

The SCC was provided by a local ready-mixed concrete supplier and has been used in structures in Rio de Janeiro. Details of the concrete mix proportions were not given; the only information provided was that the mix included silica fume as a mineral admixture (6 % by mass of ce-ment), the water/cement ratio was 0.30, the volume frac-tion of coarse aggregate was about 33 % and half of it had the nominal maximum size of 9.5 mm (crushed granite) and the other half 19 mm (crushed syenite). No viscosity-modifying agent was used.

Concrete cylinders (150 mm dia. × 300 mm high) were cast simultaneously with the beams in order to deter-mine the hardened concrete properties. The fresh and hardened concrete properties are given in Table 1. The fresh concrete tests, which evaluate the self-compactabili-ty of the concrete, were carried out at the start of casting the beams. According to slump flow, t500 and V-funnel tests, the SCC concrete classes are SF2, VS1 and VF1 re-spectively. The concrete cylinders were cured under poly-ethylene sheeting for one day and then stripped and moist-cured for 12 days. Following casting, the beams were wetted and kept under plastic sheeting for 13 days, then removed from their formwork and left under room conditions with the control cylinders.

The mean SCC compressive strength after one month was 65 MPa; the properties given in Table 1 (com-pressive strength, splitting tensile strength, tangent modu-lus of elasticity at origin of stress-strain diagram) were ob-

The database for SCC and VC beams [6] tested by other authors who investigated the shear strength of SCC beams with a shear span-to-effective depth ratio a/d ≥ 2 [7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19] shows that the majority of the beams had depths smaller than normal for beams (h ≤ 300 mm). Comparing the experimental shear strength of those beams Vu with the strength calculated using code provisions VR [4, 21, 22, 23], Resende [6] found no relevant differences between the means and medians of Vu/VR values for SCC and VC beams. From the beams analysed, 95 were SCC (60 without transverse reinforce-ment, 35 with transverse reinforcement) and 35 VC (21 without transverse reinforcement, 14 with transverse rein-forcement). A higher coefficient of variation of Vu/VR was found for the SCC beams, possibly due to the difference between the number of SCC and VC beams. For beams with shear reinforcement, no value of Vu/VR < 1.0 was found, but this was not the case for the beams without shear reinforcement, where the percentage of cases with Vu/VR < 1.0 was greater for the SCC beams.

From the groups of SCC and VC beams tested with a/d ≥ 2 and without transverse reinforcement found in the literature, only those of Hassan et al. [8, 9] and Arezou-mandi and Voltz [18] had a depth > 350 mm. The amount of coarse aggregate in the SCC used by Hassan et al. [8, 9] was 20 % less than that in the VC, and the SCC beams exhibited lower shear strengths than the similar VC beams. The difference for those with greater depth and smaller longitudinal reinforcement ratio (h = 750 mm, a/d = 2.8, fc ≅ 45 MPa, ρ ≅ 1.1 %) was 17 %. The SCC and VC used in the beams tested by Arezoumandi and Voltz [18] had the same coarse aggregate content and the flowability of SCC was only achieved by using high-range water-re-ducing and viscosity admixtures (chemically based SCC), a procedure that is not normally used in practice. From a comparison of the shear strengths of the SCC and VC beams tested with those VC beams in the database of Reineck et al. [20], it was concluded that the SCC and VC results fall within a 95 % confidence interval of a non-lin-ear regression curve fitted to the VC database.

Cuenca et al. [11], Lin and Chen [16] and Arezou-mandi and Volz [19, with a correction confirmed in cor-respondence with the authors] tested SCC and VC beams with transverse reinforcement and depth > 350 mm. The SCC beam tested by Cuenca et al. [11] performed similarly to the VC beam (SCC coarse aggregate content about 10 % less than in VC, h = 700 mm, a/d = 3.2, fc ≅ 50 MPa, ρ ≅ 3 %, ρwfyw = 0.85 MPa). The SCC beams of Lin and Chen [16] had two different types of SCC that had maxi-mum aggregate size smaller than that of the VC; one had the same coarse aggregate content as the VC and the oth-er had 14 % less coarse aggregate. According to Lin and Chen [16], the SCC beams with a smaller coarse aggregate content tended to have a lower shear strength. Those beams had h = 360 mm, a/d from 2.5 to 3.5, fc from about 30 to 49 MPa, ρ ≅ 4.5 % (higher than used in real cases) and ρwfyw from 1.4 to 1.8 MPa. Arezoumandi and Volz [19] tested two pairs of (chemically based) SCC and VC beams (h = 460 mm, a/d ≅ 3.0, fc = approx. 35 MPa, ρ ≅ 2.7 %, ρwfyw = 0.73 MPa) and the average shear strength of the SCC beams was 18 % lower than that of the similar VC beams.

Table 1. Properties of SCC

Fresh state Hardened state

Slump flow (mm)

t500 (s)

690

1.5

fc (MPa) 71.6

V-funnel (s) 7.8 fct,sp (MPa) 4.80

Column segregation index (%) 0 Ec (GPa) 33.7

5



according to the codes and shows that ρw,minfywk values from different codes can be quite different. The stirrup spacing ranged from approx. 0.3d to 0.5d.

Table 3 summarizes the beam data. Beams V3 and V5 had the same value of ρwfyw and different values of ρ; beams V1 and V6 had approximately the same value of ρwfyw but different stirrup diameter and spacing. The dif-ference between V1, V2, V3 and V4 was ρwfyw.

2.3 Instrumentation and testing procedures

The beams were supported on a roller and a pinned sup-port placed on 145 mm wide × 25 mm thick steel plates and loaded through a steel plate 95 mm wide × 25 mm thick, as shown in Fig. 2. The critical shear span (with greater shear force) was 1250 mm, corresponding to a shear span-to-effective depth ratio of 2.8.

Displacement transducers were used to measure the deflections at the cross-section of maximum bending mo-ment and strain gauges were used to measure the strains in the reinforcement and the concrete. The strain gauges were installed on the lower layer of the bottom longitudi-nal reinforcement at the cross-section of maximum bend-ing moment and at mid-height of some stirrups (Fig. 2). Concrete strains were measured at the section located in the critical shear span and at a distance of 130 mm from

tained at the age of about four months, when the beams were tested. A visual inspection of the failure surfaces of the SCC cylinders tested for obtaining the splitting tensile strength detected some fractured coarse aggregates.

Table 2 gives the mean reinforcing bar properties obtained from tests of samples: yield stress, tensile strength and tensile strength-yield stress ratio. The 4.2 mm diame-ter bars were the only ones that had a stress-strain rela-tionship without yield plateau and low fst/fy ratio.

2.2 Beam characteristics

The six beams tested had a rectangular cross-section (bw = 175 mm, h = 500 mm, d = 443 mm) and a total length of 5 m. In five beams the longitudinal tensile reinforcement ratio ρ was 2.5 % (2 φ25 mm and 3 φ20 mm bars) and in the other one it was 2.0 % (5 φ20 mm bars). On the com-pression side, all the beams had 2 bars 16 mm diameter. All longitudinal bars were continued straight to the ends. The diameter of the stirrups was 4.2 mm (five beams) or 6.3 mm (one beam).

The beams were designed to fail in shear, with no yielding of longitudinal reinforcement, and their main variable was the transverse reinforcement ratio ρw. The values of the web reinforcement index ρwfyw, with fyw tak-en as the yield stress given in Table 2, were chosen in such a way that they covered the range of ρw,minfywk given by ACI 318:2011 [21], EN 1992-1-1:2004 [22], fib Model Code 2010 [4] and ABNT NBR 6118:2014 [23]. Fig. 1 compares the ρwfyw values of the beams with ρw,minfywk

Table 2. Steel bar properties

φ (mm) fy or fyw (MPa) fst (MPa) fst/fy or fst/fyw

4.20 770 795 1.03

6.30 567 711 1.25

16.0 510 561 1.10

20.0 505 609 1.21

25.0 589 735 1.25

Table 3. Beam data

Beam

Transverse reinforcementLongitudinal reinforcement

φ(mm)

s(mm)

ρw(%)

ρwfyw (MPa) ρwfst (MPa) ρ (%)

V1

4.2

125 0.127 0.975 1.01

2.5V2 150 0.106 0.813 0.839

V3 185 0.086 0.659 0.680

V4 240 0.066 0.508 0.524

V5 185 0.086 0.659 0.680 2.0

V6 6.3 210 0.170 0.962 1.21 2.5

bw = 175 mm; h = 500 mm; d = 443 mm; a/d = 2.8

Fig. 1. Comparison of ρwfyw values of beams and ρw,minfywk values of codes as a function of the concrete strength

6



The maximum concrete strain measured varied from –0.53 × 10–3 (beam V5, with smaller ρ and failure load) to –1.8 × 10–3 (beam V6, with greater ρwfst and failure load).

The strain measured in the longitudinal reinforce-ment at failure was about 2.2 × 10–3 (about 87 % of the yielding strain) for beams V1 and V6; in the other beams it did not exceed 1.6 × 10–3 (about 63 % of the yielding strain).

3 Comparison of test results with shear provisions of codes of practice

Many formulae for calculating the shear strength of beams have been developed since the beginning of the 20th cen-tury, but there is no consensual approach to this. Different design codes use distinct approaches to calculate beam shear strength, which can lead to quite different results.

The experimental shear strengths of the beams were compared with the shear provisions of ACI 318:2011

the cross-section of maximum bending moment, at 10 mm and 30 mm from the top.

The tests were carried out with displacement con-trol. Load, displacements and strains were recorded by a computer-controlled data acquisition system. For safety reasons, critical diagonal crack width was measured using an optical comparator only until it was about 1.0 mm. Load was applied until failure of each beam.

2.4 Test results

All the beams failed in shear prior to yielding of the longi-tudinal reinforcement. Except for beam V6, which had stirrups with greater diameter and more ductile behav-iour, failure occurred with rupture of the stirrups crossed by the critical diagonal crack. Beam V6, with about the same values of ρwfyw and Vu as beam V1, also underwent shear failure, but without rupture of the stirrups.

The critical diagonal crack developed as an exten-sion of the flexure-shear crack closest to the support when it reached about mid-depth of the beam. Table 4 lists the shear force when the critical diagonal crack became visi-ble and at failure. Photographs of the cracking patterns at failure are shown in Fig. 3.

Fig. 4 shows Vu as a function of ρwfyw. Beam V5, with ρ = 2.0 %, had the lowest shear strength, and was 15 % less resistant than beam V3, with ρ = 2.5 % and the same ρwfyw.

The load-deflection responses of beams V2, V3, V4, V5 and V6 are presented in Fig. 5. Beams with lower ρwfyw values tended to have a more pronounced increase in deflection and concrete and stirrup strains when the critical diagonal crack formed. As soon as this crack be-came visible, its width was already at least 0.2 mm.

Fig. 2. Dimensions (mm) of and loading on beams, reinforcement and strain gauge location

Table 4. Variable parameters and critical diagonal crack and ultimate shear forces of SCC beams

Beam ρwfyw (MPa) ρwfst (MPa) ρ (%) Vcr (kN) Vu (kN)

V6 0.962 1.21 2.5 126 253

V1 0.975 1.01 2.5 126 251

V2 0.813 0.839 2.5 112 174

V3 0.659 0.680 2.5 112 151

V5 0.659 0.680 2.0 105 128

V4 0.508 0.524 2.5 126 158

7



ABNT NBR 6118:2014 [23] gives the greatest values of ρw,minfywk and values of Vc greater than those of ACI318:2011 [21] and the Level III approximation of fib Model Code 2010 [4]. Table 5 presents the ratio of ex-perimental to code-predicted shear capacity (Vu/VR). For calculating VR, the materials and shear resistance factors were taken as equal to one and mean concrete and steel strengths (Tables 1 and 2) were considered instead of characteristic ones.

[21], EN 1992-1-1:2004 [22], fib Model Code 2010 [4] and ABNT NBR 6118:2014 [23]. For the shear strength of beams with transverse reinforcement, the procedures of ACI 318:2011 [21], the Level III approximation of fib Model Code 2010 [4] and ABNT NBR 6118:2014 [23] consider VR = Vc + Vs (“concrete” and steel web contribu-tions), whereas EN 1992-1-1:2004 [22] and the Level I approximation of fib Model Code 2010 [4] consider only the transverse reinforcement contribution (VR = Vs).

V6 (ρwfyw = 0.962 MPa; Vu = 253 kN)

V2 (ρwfyw = 0.813 MPa; Vu = 174 kN)

V5 (ρwfyw = 0.659 MPa; Vu = 128 kN)

V1 (ρwfyw = 0.975 MPa; Vu = 251 kN)

V3 (ρwfyw = 0.659 MPa; Vu = 151 kN)

V4 (ρwfyw = 0.508 MPa; Vu = 158 kN)

Fig. 3. Views of beams after failure

8



longitudinal reinforcement ratio, even if ρwfyw ≅ ρw,minfywk. For the two beams with the highest values of ρwfyw, the only ones that have ρwfyw > ρw,minfywk accord-ing to ABNT NBR 6118:2014 [23], no code resulted in Vu/VR < 1.

4 Comparison of shear strengths of VC and SCC beams

The ultimate nominal shear stresses of the SCC beams are compared with those of VC beams tested by Garcia [24] in Fig. 7.

The VC had 40 % coarse aggregate with a nominal maximum size of 19 mm (crushed syenite), silica fume as a mineral admixture (10 % by mass of cement), a water/ce-ment ratio of about 0.30 and a 90 mm slump.

For calculating Vs according to EN 1992-1-1:2004 [22] and the Level I approximation of fib Model Code 2010 [4], the minimum permissible angle between the concrete com-pression struts and the beam axis of the truss model was used and, for Vc of ACI 318:2011 [21] and ABNT NBR 6118:2014 [23], the simplest formula was used.

Fig. 6 compares Vu with VR graphically. The lines connect the VR values and are shown dashed where ρwfyw < ρw,minfywk. This figure shows that the Level I ap-proximation of fib Model Code 2010 [4] and EN 1992-1-1:2004 [22] procedures are the most conservative, since they consider VR = Vs. ACI 318:2011 [21] and the Level III approximation of fib Model Code 2010 [4] lead to VR values close to each other and may give non-conser-vative estimates, particularly for beams with a smaller

Fig. 4. Experimental shear resistance as a function of ρwfyw

Fig. 5. Deflection at section of maximum bending moment for different values of shear at critical shear span

Table 5. Ratios of experimental to calculated shear strengths of beams

Codeρw,minfywk(MPa)

Vu/VR

V6 V1 V2 V3 V5 V4

ACI 318:2011 0.525 1.37 1.34 0.994 0.926 0.785 1.05

EN 1992-1-1:2004

0.677

1.51 1.48 1.23 1.31 1.11 (1.78)

fib MC2010 (level I) 2.18 2.13 1.77 1.90 1.61 (2.57)

fib MC2010 (level III) 1.34 1.31 0.978 0.910 0.821 (1.03)

NBR 6118:2014 0.926 1.17 1.15 (0.841) (0.766) (0.650) (0.849)

Values in brackets refer to beams with ρwfyw < ρw,minfywk; the others refer to beams with ρwfyw ≅ ρw,minfywk or ρwfyw > ρw,minfywk

Fig. 6. Comparison between experimental and calculated shear strengths

Fig. 7. Comparison of ultimate shear stresses of high-strength SCC beams (d = 443 mm, ρ = 2.5 %, a/d = 2.8) and VC beams (d = 405 mm, ρ = 2.6 %, a/d = 3.0) with fc ≅ 70 MPa and small values of ρwfyw

9



Vs shear strength provided by transverse reinforcementVu experimental ultimate shear forceφ steel bar diameterρ tensile longitudinal reinforcement ratioρw transverse reinforcement ratioρw,min minimum transverse reinforcement ratio

References

1. Regan, P. E., Kennedy-Reid, I. L., Pullen, A. D., Smith, D. A.: The influence of aggregate type on the shear resistance of reinforced concrete. The Structural Engineer, 2005, 83, No. 23/24, pp. 27–32.

2. Sagaseta, J., Vollum, R. L.: Influence of beam cross-section, loading arrangement and aggregate type on shear strength. Magazine of Concrete Research, 2011, 63, No. 2, pp. 139–155.

3. British Standards Institution (BSI): NA to BS EN 1992-1-1:2004, UK National Annex to Eurocode 2: Design of con-crete structures – Part 1-1: General rules and rules for build-ings, London, 2009.

4. International Federation for Structural Concrete: fib Model Code for Concrete Structures 2010, Wilhelm Ernst & Sohn, Berlin, 2013.

5. Yu, Q., Bazant, Z. P.: Can stirrups suppress size effect on shear strength of RC beams? Journal of Structural Engineer-ing, 2011, 137, No. 5, pp. 607–617.

6. Resende, T. L.: Shear strength of self-compacting concrete beams. MSc thesis, COPPE-UFRJ, Rio de Janeiro, 2014 (in Portuguese).

7. Lachemi, M., Hossain, K. M. A., Lambros, V.: Shear resis-tance of self-consolidating concrete beams – experimental investigations. Canadian Journal of Civil Engineering, 2005, 32, No. 6, pp. 1103–1113.

8. Hassan, A. A. A., Hossain, K. M. A., Lachemi, M.: Behavior of full scale self-consolidating concrete beams in shear. Ce-ment & Concrete Composites, 2008, 30, No. 7, pp. 588–596.

9. Hassan, A. A. A., Hossain, K. M. A., Lachemi, M.: Strength, cracking and deflection performance of large-scale self-con-solidating concrete beams subjected to shear failure. Engi-neering Structures, 2010, 32, No. 5, pp. 1262–1271.

10. Beygi, M. H. A., Amiri, J. V., Moazen, A. R., Malidareh, N. R., Mazandaran, M. H.: The investigation of effect of steel fiber on the shear behavior of self compacting concrete beams with normal and high strength. Proc. of Conf. on our World in Concrete & Structures, Singapore, 2008.

11. Cuenca, E., Serna, P., Pelufo, M. J.: Structural behavior of self-compacting and fiber reinforced concrete under shear loading. Proc. of Intl. Association for Shell and Spatial Structures Symposium, Valencia, 2009, pp. 2920–2931.

12. Boel, V., Helincks, P., Desnerck, P., De Schutter, G.: Bond behaviour and shear capacity of self-compacting concrete. Proc. of SCC 2010, Montreal, 2010, pp. 343–353.

13. Abed, A. H.: Shear Behavior of Self Compacting R.C. I-Beams. Journal of Engineering and Development, 2012, 16, No. 4, pp. 1–16.

14. Atshan, A. F.: Shear Behavior of self compacting concrete. Journal of Engineering and Development, 2012, 16, No. 2, pp. 289–305.

15. Safan, M. A.: Shear strength of concrete beams cast with self-compacting concrete containing different fillers and coarse aggregates. Canadian Journal of Civil Engineering, 2012, 39, No. 7, pp. 760–770.

16. Lin, C., Chen, J.: Shear behavior of self-consolidating con-crete beams. ACI Structural Journal, 2012, 109, No. 3, pp. 307–316.

17. Salman, M. M., Jarallah, H. K., Delef, A. N.: Experimental study for shear behavior of hybrid self-compacting concrete

The VC beams had a rectangular cross-section (bw = 150 mm, h = 450 mm, d = 405 mm), ρ = 2.6 %, fc and fct,sp of about 70 and 4.3 MPa respectively, a ρwfyw from zero to 1.16 MPa and were tested in three-point bending with a/d = 3.0. Apart from one, the beams failed in shear with fracture of the stirrups, and the two with higher fail-ure load exhibited yielding of the longitudinal reinforce-ment.

As shown in Fig. 7, the SCC beams had ultimate shear stresses lower than those of the VC beams.

The differences between the ρ and a/d values of the groups of SCC and VC beams do not justify the differenc-es in the order of 30 %, nor the differences between the effective depths, since the stirrups tend to diminish the size effect [5].

5 Conclusions

On the basis of available test results, it can be concluded that the ultimate shear stress of SCC and VC slender beams with similar characteristics may or may not be alike, depending on the concrete compositions and strengths and the beam depths, and possibly also depend-ing on their shear reinforcement ratio.

The comparison of the ultimate shear stress of high-strength SCC beams with small transverse reinforcement indexes of this study with those of the VC beams tested by Garcia [24] indicates that the difference might not be neg-ligible.

Not all shear code provisions can safely predict the shear capacity of beams with a low transverse reinforce-ment index, even if the ρwfyw is greater than the ρw,minfywk given by the code. This is especially true for beams with a lower longitudinal reinforcement ratio, since the longitu-dinal reinforcement stress at the time of shear failure has a significant effect on the shear capacity of a beam.

Acknowledgments

The authors would like to thank CAPES, CNPq, FAPERJ and ENGEMIX for their support.

Notation

a shear spanbw width of webd effective depthfc mean cylinder compressive strength of concretefct,sp mean splitting tensile strength of concretefy mean yield strength of longitudinal reinforcing steelfyw mean yield strength of transverse reinforcing steelfywk characteristic yield strength of transverse reinforcing

steelfst mean tensile strength of reinforcing steelh depths spacing of stirrupsEc mean tangent modulus of elasticity at origin of

stress-strain diagramVc shear strength provided by concrete webVcr experimental shear force corresponding to critical

shear crackVR calculated shear strength

10



Thomás Lima de ResendeLecturer, Universidade Federal dos Vales do Jequitinhonha e MucuriInstituto de Ciência, Engenharia e Tecnologia, Teófilo Otoni, MG, BrasilTel: + 55 33 35224873Email: [email protected]

Lidia da Conceição Domingues ShehataProfessor, Universidade Federal Fluminense and COPPE – Universidade Federal do Rio de JaneiroCaixa Postal 68506, CEP 21945-970, Rio de Ja-neiro, RJ, BrasilTel: + 55 21 999640940Email: [email protected] ; [email protected]

Ibrahim Abd El Malik ShehataProfessor, COPPE – Universidade Federal do Rio de JaneiroCaixa Postal 68506, CEP 21945-970, Rio de Ja-neiro, RJ, Brasil,Tel: + 55 21 999829713Email: [email protected]

beams. Journal of Engineering and Development, 2013, 17, No. 2, pp. 97–118.

18. Arezoumandi, M., Volz, J. F.: Shear strength of chemically-based self-consolidating concrete beams – Fracture mechan-ics approach vs. modified compression field theory. ASCE Journal of Materials in Civil Engineering, 2014, 26, No. 4, pp. 713–720.

19. Arezoumandi, M., Volz, J. F.: An experimental study on shear strength of chemically-based self-consolidating con-crete. International Journal of Concrete Structures and Ma-terials, 2013, 7, No. 4, pp. 273–285.

20. Reineck, K., Kuchma, D. A., Kim, K. S., Marx, S.: Shear da-tabase for reinforced concrete members without shear rein-forcement. ACI Structural Journal, 2003, 100, No. 2, pp. 240–249.

21. American Concrete Institute: ACI 318-11: Building Code Requirements for Structural Concrete – and Commentary, Farmington Hills, Michigan, 2011.

22. European Committee for Standardization: EN 1992-1-1:2004: Design of concrete structures – Part 1-1: General rules and rules for buildings, Brussels, 2004.

23. Associação Brasileira de Normas Técnicas: ABNT NBR 6118:2014: Design of concrete structures – Procedures, Rio de Janeiro, 2014 (in Portuguese).

24. Garcia, S. L. G.: Minimum shear reinforcement ratio in rein-forced concrete beams. DSc thesis, COPPE/UFRJ, Rio de Janeiro, 2002 (in Portuguese).


All researchers who have tested the shear capacity of RC mem-bers without stirrups have observed a large scatter in the results.The objective of this paper is to conduct an overview study of the causes of the great shear failure scatter of RC beams without stirrups. Thirteen groups of shear tests on comparable experi-ments, extracted from the ACI-DAfStb evaluation database, are considered. The amount of data available is increased numeri-cally. To this end, based on Eurocode 2 equations for shear re-sistance and shrinkage strain, a full probabilistic model is defined according to the JCSS Probabilistic Model Code. A multivariate statistical evaluation of outcomes is then performed.The investigation highlights the fact that both the tensile strength of concrete and high shrinkage values may be usefully consid-ered for more in-depth studies of the phenomenon, whereas geo-metrical properties and concrete compressive strength seem to be less important or can even be neglected.

Keywords: shear strength, shrinkage, reinforced concrete members, probabilistic models, Monte Carlo simulation

1 Introduction1.1 General

The shear strength of reinforced concrete (RC) members without web reinforcement, which is regarded in this paper as maximum load, is a subject that has generated many controversies and debates since the beginning of the last century; Rebeiz [1] has provided a brief and peda-gogical historical account. All the researchers who have tested the shear capacity of RC members without web reinforcement have observed a large scatter in the results; even simple members cast simultaneously from the same concrete batch may show significant differences (Silfwerbrand [2] measured 15 % in tests on overlaid concrete beams, for example).

As a result, a pronounced scatter may involve the risk of underestimating – with serious consequences such as undersizing and the possibility that shear failure will

occur in the RC members. On the other hand, a conspicu-ous safety margin may run the risk that a large majority of RC structures will be oversized, with consequences such as poor economy for the owner and excessive depletion of natural resources.

The objective of this paper is to conduct an overview study of the causes of the great shear failure scatter for normal-strength RC slender members without stirrups subjected to shear and bending. The outcomes of the study will be of importance for further, more in-depth, investigations of the phenomenon, highlighting which pa-rameters may be useful and which of them, if any, are less important or can even be neglected.

1.2 Shear transfer mechanism in RC members without web reinforcement

There is general agreement among researchers regarding the mechanisms that participate in carrying shear loads over the cross section, Fig. 1. In the light of the findings of the state-of-the-art reports by the joint ASCE-ACI Com-mittee 426 [3] and 445 [4], the shear transfer mechanisms involve the following components:1) the shear stresses in uncracked concrete, e.g. the flex-

ural compression zone,2) the interface shear transfer, often called aggregate in-

terlock or crack friction,3) the dowel action of the longitudinal reinforcing bars,4) the arch action, and5) the residual tensile stresses transmitted directly across

cracks.

In order to understand these mechanisms better, let us consider a slender RC member loaded in flexure without axial force, Fig. 1. Immediately, flexure and shear com-bine to create a biaxial state of stress. Cracks form when the principal tensile stresses exceed the tensile strength of the concrete. In a region of large bending moments, these stresses are greatest at the extreme tensile fibre of the member and are responsible for the initiation of flexural cracks perpendicular to the axis of the member. In a region of high shear force, significant principal ten-sile stresses, also referred to as diagonal tension, may be generated at approx. 45° to the axis of the member. These may result in inclined (diagonal tension) cracks. The majority of diagonal cracks, in most beams, form as

Technical Paper

Scatter in the shear capacity of slender RC members without web reinforcement: an overview study

Filippo Sangiorgio*Johan SilfwerbrandGiuseppe Mancini

DOI: 10.1002/suco.201400107


Submitted for review: 17 November 2014; revision: 25 March 2015; accepted for publication: 10 May 2015. Discussion on this paper must be submitted within two months of the print publication. The discussion will then be published in print, along with the authors’ closure, if any, approximately nine months after the print publication.

12

F. Sangiorgio/J. Silfwerbrand/G. Mancini · Scatter in the shear capacity of slender RC members without web reinforcement: an overview study


2.1 Evaluation of comparable experiments extracted from the ACI-DAfStb shear database

Researchers who investigate the causes of shear failure scatter need reported test results containing tests on al-most identical beams.

A total of 13 groups of shear tests on comparable slender RC beams without stirrups extracted from the ACI-DAfStb evaluation database (the reader should refer to [17] for details) and grouped by Sangiorgio et al. [18] were identified and considered for the subsequent analy-sis.

Unfortunately, the amount of data available from each group of comparable tests is not sufficient to extract valid statistical information. Therefore, it becomes neces-sary to increase the amount of data via a numerical study that generates a large number of virtual tests from existing ones via probabilistic investigations, as explained in more detail in section 2.5.

2.2 Choice of resistance model

It has to be emphasized that the choice of resistance mod-el is extremely important because it may substantially in-fluence the results. It should be chosen to be as simple as possible, but must describe all the important conclusions pertaining to the phenomenon investigated. As a result, it is always a trade-off between accuracy, complexity and speed of computation.

The presence of high uncertainty in the shear trans-fer mechanism is doubtless reflected in the quality of the different predictive models, which implies that, at present, no advanced and sophisticated formulations exist which are able to predict the shear strength of reinforced con-crete beams without web reinforcement with high preci-sion and accuracy. In this regard, painstaking research activities were carried out in the past [17], [19]–[21]. One promising method of investigation might be to use a suit-ably calibrated non-linear FEM analysis; however, this procedure is also complicated and time-consuming.

Similarly, the calculation of shrinkage is a difficult task, because shrinkage, degree of restraint, modulus of elasticity, Poisson’s ratio, creep, concrete age and con-crete quality all influence the stress. In accordance with [22], owing to the large number of influencing parameters, an accurate prediction of shrinkage stresses is almost impossible in an arbitrary case and, if detailed knowledge of actual concrete properties and environmental effects is lacking, the use of complicated models can hardly be motivated.

extensions of flexural cracks. After formation of diagonal cracks, the resistance mechanisms gradually change from beam action to arch action with increasing load intensity. Once a crack is formed, provided that shear displace-ments occur at the two faces of the crack, a number of coarse aggregate particles projecting across such a crack will provide resistance against slip and shear stresses will therefore be generated (aggregate interlock). Clearly, the width and coarseness of the crack, the shear displacement and the strength of embedment (i.e. concrete strength) are likely to be the most important variables that affect this component. The same shear displacement will also cause the rebars to bear against the cover concrete, inducing dowel forces across the flexural reinforcement. Normally, dowel action is not significant in members without trans-verse reinforcement because the maximum shear in a dowel is limited by the tensile strength of the concrete cover supporting the dowel. Moreover, a clean break does not occur in the crack and small pieces of concrete bridge the crack and continue to transmit tensile force until the crack width is sufficiently enlarged (residual tensile stresses across cracks). See [4] and [5] for a more detailed explanation.

1.3 Shrinkage of concrete and effects on RC members

When concrete loses moisture, it shrinks. Shrinkage is the time-dependent linear strain measured in an unloaded and unrestrained specimen at constant temperature. Usu-ally, concrete shrinkage strain is categorized as independ-ent of the stress conditions in the material; unfortunately, the real life situation is not so simple, see [6]. Moreover, the shrinkage strain in concrete is composed of different components. An in-depth discussion of concrete shrink-age can be found in [7].

Sophisticated creep and shrinkage prediction mod-els are given in, for example, [8]–[16].

Finally, it has been shown that the final shrink-age strain may vary greatly, being generally in the range 0.2–0.6 mm/m, but sometimes as much as 1.0 mm/m [5].

2 The methodology

The method basically consists of the following five steps:A) evaluation of the comparable experiments extracted

from the ACI-DAfStb shear database,B) choice of the resistance model,C) assessment of the model uncertainty,D) full probabilistic model, andE) statistical and probabilistic investigation.

Fig. 1. Shear transfer mechanism for RC members without web reinforcement

13



is εcs = εcd + εca. Relaxation of shrinkage strains due to creep of concrete is not taken into consideration.

2.3 Assessment of model uncertainty

Models, descriptive or predictive, are the basic vehicles by which we reflect and express our understanding of some aspect of reality, a particular unknown of interest. As it is virtually impossible to grasp any situation in its entire complexity, models are always partial representations of reality. In other words, what we know about the true nature of the unknown of interest is generally incomplete, resulting in a state of uncertainty. Accordingly, the uncer-tainties in model predictions arise from uncertainties in the values assumed by the model parameters, para meter uncertainty, and the uncertainties and errors associ-ated with the structure of the model, model uncertainty, which stem from abstractions, assumptions and approxi-mations [27].

The assessment procedure for qR went through a pro-cess of comparison with experimental results. Engineering judgment was also used. The ACI-DAfStb evaluation da-tabase [17] was considered. The raw database was filtered so that all experiments performed outside of the following range were excluded: 2.5 ≤ a/dtest ≤ 7, and ρl,test ≤ 2 %. Experiments lacking recorded fct,test values were also ignored. Only 159 tests remained from the original 784 and these were compared directly with model A, B and C predictions using the various researchers’ fc,test (for model A) or fct,test (for models B and C).

For each model, the model safety factor, defined here as gmod = Vtest/VR,c, was computed and the results visually explored, as shown in Figs. 3 and 4. The two fig-ures show the scatter plots of gmod with respect to, respec-tively, i) the main geometric and ii) the main mechanical and concrete composition parameters of the beams tested. The main geometric parameters are summarized in bw,test, dtest, a/dtest and As,test. The main mechanical and concrete composition parameters are characterized by ρl,test, Φa,test, fc,test and fct,test.

Experimental and numerical results differ moder-ately. Looking in more detail, a careful analysis of Fig. 3b indicates that the different gmod values are sensitive to the variation in the effective depth of the cross-section and, for beams with dtest ≤ 500 mm (approximately), the model predictions tend to underestimate the measurements (a higher mean value for gmod) with a decrease in the scatter (in the authors’ opinion, this seems to be a manifestation of the size effect). Consequently, as the comparable ex-periments considered fit the case mentioned, the authors believe that qR for the three predictive models can be well represented by a lognormal distribution with a mean value μ = 1.1 and coefficient of variation CoV = 0.10, which is quite in accordance with what is suggested in the JCSS Probabilistic Model Code [28].

Considering the effects of qR, Fig. 5 shows that, for each of the three models, predictions cover the range of observed shear strength in a satisfactory manner. In particular, the figure shows the comparison of the box plots of Vtest measured for groups 3, 5 and 12 of compa-rable experiments (notation according to [18]) and VR,c,q evaluated with the three models A, B and C. A box plot

Therefore, since “the purpose of computing is in-sight, not numbers” [23], a simple formulation for both ultimate shear resistance and shrinkage strain, such as the empirical one suggested by Eurocode 2 [24], which is, however, able to explain this phenomenon qualitatively, was considered sufficient for the aim of the present study. Shear resistance and shrinkage strain equations were then combined to take into account shrinkage effects.

As a limitation of this strategy, it should be noted that influences of other factors, such as i) aggregates and their location in the cross-section and in relation to rebars, ii) microcracks and their random pattern and iii) surface area of the rebars, are not captured by the model formulation, despite the fact that they can be significant (see [25] and [26], for example).

Three different predictive models were considered and assessed through comparison with experimental re-sults so that the effects of different parameters (both geo-metric and mechanical properties, and shrinkage) could be gradually taken into account. From the simplest to the more complex, they are characterized as follows:

Model A: Shear resistance model entirely according to Eurocode 2 [24] (with CR,c = 0.18) and shrinkage effects neglected:

Model B: Same as model A but considering the square root of the tensile strength of concrete multiplied by a cor-rective coefficient given by the ratio 1/0.3 instead of the cube root of the compressive strength (over many simula-tions this will not affect the mean value of the distribution of the shear capacity, but it will consider the greater scat-ter of the tensile strength of concrete with respect to the compressive strength):

Model C: Same as model B but including shrinkage ef-fects:

The following assumptions were made for Model C: t = 28 days, ts = 7 days, RH = 60 % and N strength class of cement.

According to [5], shrinkage effects were modelled as a tensile restraining force, due to the bonded reinforce-ment, applied to the concrete at the level of the centroid of the steel (see Fig. 2). In the figure, consistent with Eu-rocode 2 [24], the total unrestrained shrinkage of concrete

[ (100 ) ] (1)R,c R,C l c1/3V C K f b dwρ=

[ (100 ) ( /0.3) ] (2)R,c R,C l1/3

ct1/2V C K f b dwρ=

[ (100 ) ( /0.3) ] (3)R,c R,C l1/3

ct1/2

1 csV C K f k b dwρ= + σ

Fig. 2. Shrinkage effects

14



the concrete compressive strength due to the special plac-ing, curing and hardening conditions, concrete age and variations in the tensile strength due to gravel type and size, cement composition and climatic conditions were also taken into account. Model uncertainty is modelled in terms of uncertain perturbations by introducing a lognor-mal distributed stochastic variable qR with expected mean value and standard deviation properties estimated on the basis of the model assessment procedures described in section 2.3. The model uncertainty qR is then multiplied by the outcomes of the predictive models. The random-ness of shrinkage is introduced by assuming that εcd,0 is an uncorrelated random variable with known mean value (the predicted one), CoV = 30 % and lognormal prob-

[29]–[31] is a statistical technique used to summarize and compare groups of data visually. In each box, the central mark is the median, the edges of the box are the 25th and 75th percentiles, the whiskers extend to the most extreme data points not considered outliers and outliers are plot-ted individually.

2.4 Full probabilistic model

A full probabilistic model was defined according to [28], able to describe the mechanical properties of concrete and reinforcing steel, the reinforcement area, the geo-metric properties of the cross-section and the resistance model uncertainty. The effects of additional variations in

Fig. 3. Scatter plots of model A, B and C safety factors vs. the main geometrical parameters of the beams tested

Fig. 4. Scatter plots of model A, B and C safety factors vs. the main mechanical and concrete composition parameters of the beams tested

15



DAfStb database [17]. Barely three groups of comparable experiments remained from the original collection: groups 3, 5 and 12.

In order to increase the number of tests numerically, the Monte Carlo method was used to generate 50 000 samples for each of the three identified groups of compa-rable experiments from the mean values of their physical parameters, covering a wide range of geometric and me-chanical properties.

For example, let us consider group 12 of the compa-rable experiments. This group contains six tests performed by Chana [32] and characterized by the following mean values of the parameters affecting the predictive models: bwm,group = 203 mm, hm,group = 406 mm, dm,group = 356 mm, Asm,group = 1256.6 mm2 and fcm,group = 34 N/mm2. For

ability distribution; no randomness is associated with εca. Details of the probabilistic model are given in Table 1. The table shows:a) all the basic random variables with their symbols/equa-

tions,b) distribution types,c) units,d) mean values μ,e) standard deviations σ, andf) coefficients of variation CoV.

2.5 Statistical and probabilistic investigation

The 13 groups of comparable experiments mentioned [18] were in turn filtered as previously done for the raw ACI-

Fig. 5. Box plots of the shear capacities Vtest and VR,c,q for groups 3, 5 and 12 of the comparable experiments

Table 1. Probabilistic model

Basic random variable Dist. Unit. μ σ CoV

fc = α(t,τ) fc,0λ Y1 – MPa – – –

α(t,τ) = 0.8 ∙ [0.6 + 0.12 ln(t)] D – 1.42 – –

t D day 28 – –

fc,0 LGN MPa fcm,group σ(fcm,group) –

λ D – 0.96 – –

Y1 LGN – 1.00 – 0.06

fct = 0.3 fc2/3 Y2 – MPa – – –

Y2 LGN – 1.00 – 0.30

As N mm2 Asm,group – 0.02

h, bw N mm Xm,group 4 + 0.006 ∙ Xm,group ≤ 10 –

d N mm dm,group 10.00 –

qR LGN – 1.10 – 0.10

εcd,0 LGN % 0.38 – 0.30

ts D day 7 – –

RH D % 60 – –

Note: D = deterministic value; N = normal distribution; LGN = lognormal distribution

16



quality of a linear relationship (linear least squares fitting) between the different relevant parameters of the problem. An important implication of correlation between those parameters is that it is possible to discern the design vari-ables having the greatest effect on the shear strength and, therefore, on its scatter. However, it is noteworthy that the presence of any correlation between two parameters may merely mean that they tend to fluctuate together; it does not necessarily mean that one causes the other to fluctuate.

The cells in the correlation matrices contain the cor-relation coefficients r (sometimes denoted r = √R2, where R2 is the coefficient of determination [33]), also known as the product-moment coefficients of correlation or Pearson’s correlations [34]. Correlations are interpreted by squaring the value of the correlation coefficients. The squared values represent the proportion of variance of

more detail, the reader is referred to [17] and [18]. Based on those values and according to the guidelines given in [28], the probabilistic model was developed after the man-ner of Table 1.

The statistical population included a total of 150 000 virtual tests. For each simulation setting, the shear strength was computed by the three predictive models (A, B and C). Finally, both a visual examination and multivariate statistical analysis of the outcomes were performed.

3 Computational results

The matrices of correlation coefficients for VR,c,q and all the parameters of the three models bw, h, d, fc, Y1, As, qR, fct, Y2 and εcs, relating to the samples generated from group 3 of the comparable experiments, are presented in Table 2. The correlation matrices give information on the

Table 2. Correlation matrices of Pearson between the shear capacity VR,c,q and all the parameters of the three models bw, h, d, fc, Y1, As, qR, fct, Y2 and εcs relating to the samples generated from group 3 of the comparable experiments

MODEL A VR,c,q bw h d fc Y1 As qR fct Y2 εcs

VR,c,q 1.00

bw 0.38 1.00

h 0.03 0.00 1.00

d 0.50 0.01 0.07 1.00

fc 0.42 0.00 0.00 0.00 1.00

Y1 0.13 0.00 0.01 0.00 0.31 1.00

As 0.00 0.00 –0.01 0.00 0.00 0.00 1.00

qR 0.66 0.01 0.00 0.00 0.00 0.00 0.00 1.00

MODEL B

VR,c,q 1.00

bw 0.27 1.00

h 0.03 0.00 1.00

d 0.35 0.00 0.07 1.00

fc 0.30 0.00 0.00 0.00 1.00

Y1 0.10 0.00 0.00 0.00 0.31 1.00

As 0.00 0.00 0.01 –0.01 –0.01 –0.01 1.00

qR 0.46 –0.01 0.00 –0.01 0.00 0.00 0.00 1.00

fct 0.76 0.00 0.00 0.00 0.39 0.13 0.00 0.00 1.00

Y2 0.70 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.91 1.00

MODEL C

VR,c,q 1.00

bw 0.28 1.00

h 0.05 0.01 1.00

d 0.29 0.00 0.06 1.00

fc 0.31 0.00 0.00 0.00 1.00

Y1 0.10 0.00 0.00 0.00 0.31 1.00

As –0.02 0.00 0.01 0.01 0.00 0.00 1.00

qR 0.39 0.00 0.00 0.00 0.00 0.00 –0.01 1.00

fct 0.78 0.00 0.00 0.00 0.39 0.12 0.00 0.00 1.00

Y2 0.71 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.91 1.00

εcs –0.26 0.00 0.00 0.00 0.00 0.00 0.00 0.00 –0.01 –0.01 1.00

17



the linear regression line (solid line) and the non-linear fit (dashed line), the latter using, in the former case, a power function of the form f(x) = a ∙ xb and, in the latter, a rational function of the sort f(x) = a/(x + b), and the corresponding r, R2 and root-mean-square error (RMSE)

one variable that can be predicted from the other variable. A rule of thumb for interpreting correlation coefficients has been established from experimental studies [35]:i) 0.0 ≤ r < 0.2, very weakii) 0.2 ≤ r < 0.4, weakiii) 0.4 ≤ r < 0.7, moderateiv) 0.7 ≤ r < 0.9, strongv) 0.9 ≤ r ≤ 1.0, very strong

Values in bold in the table show the most significant cor-relations. The results for the other groups of comparable experiments are omitted because of similarity.

Fig. 6 shows the scatter plots of VR,c,q evaluated with model A for the set of random samples numerically gener-ated from group 3 of the comparable experiments versus bw, d and fc.

Fig. 7 shows the scatter plots and the fitting of a least squares line (in red), at the same time with the correlation coefficients r for VR,c,q obtained from the three predictive models as a function of qR for the set of random samples numerically generated from group 3 of the comparable experiments.

Figs. 8 and 9 present the scatter plot of VR,c,q evalu-ated with models B and C respectively for the set of ran-dom samples numerically generated from group 3 of the comparable experiments versus fct and εcs together with

Fig. 6. Scatter plots of the shear capacity VR,c,q for group 3 of the comparable experiments, obtained from model A, plotted against bw, d and fc

Fig. 7. Scatter plots and fitting of a least squares line of the shear capacity VR,c,q for group 3 of the comparable experiments evaluated with the three pre-dictive models, plotted against qR

Fig. 8. Scatter plot of the shear capacity VR,c,q for group 3 of the compara-ble experiments, evaluated with predictive model B, plotted against fct

18



sions suggest). However, the explanation lies in the fact that this study deals with probabilistic analysis and the degree of natural variation of these variables is therefore small.

– As a consequence of the previous two points, the con-crete cover, given by h–d, seems to be of little relevance since both variables have shown a weak correlation with the outcome of the predictive models.

– On the other hand, it is possible to see that fct affects the predictions of both model B and model C in a relatively strong way (r = approx. 76 %). As shown in Fig. 8, the relationship between the two variables can be equally described by both a linear and a non-linear regres-sion model (the relatively moderate value of the coef-ficient of determination for the non-linear regression, R2 = 0.58, is equivalent to a correlation coefficient r = 0.76); the equivalence of both regression models is also highlighted by the similar RMSE values (about 1.85). However, this fact can be attributed to the empirical formula used and might be misleading. The relatively high value of r = 76 % indicates that both regression models are able to predict almost 60 % of the variance of the outcomes. Owing to the probabilistic model used, fct is, in turn, strongly correlated (r = approx. 91 %) with Y2. Moreover, this finding provides insight into the strict relationship between diagonal cracking loads and shear strength. Diagonal cracking is in fact a tension phenomenon that mainly occurs when the principal ten-sile stresses exceed the diagonal tensile strength of the member and, therefore, in RC beams it is in turn firmly correlated to the tensile strength of concrete.

– Shrinkage deserves particular attention. Although the correlation matrix related to model C indeed highlights a weak negative correlation between εcs and the shear strength, no reasonable (linear or non-linear) relation-ships are observed between the two parameters, as shown in Fig. 9. The general pattern of data distribution seems to have a central core with a pronounced unsymmetri-cal scatter that can hardly be described by a regression model. In the light of the results, it is possible to argue that, generally, shrinkage does not appear to contribute to the scatter in shear capacity but, returning to Fig. 5, only to reduce its magnitude slightly. Despite the use of an empirical approach, this finding is in accordance with previous studies [38]–[40]. An exception seems to be nec-essary for high values of εcs (approx. > 0.7 mm/m) which, in the case of restraint, can lead to an increase in existing crack widths or to the formation of new cracks; here, scatter and εcs appear to be moderately linked.

5 Concluding remarks

In summary, the causes of the great scatter in the shear capacity of reinforced slender normal-strength concrete members without web reinforcement were studied in a statistical-probabilistic way.

The following conclusions have been drawn from the results:1. The scatter of the shear capacity seems to be mainly

due to the randomness of the tensile strength of con-crete which, in turn, is due to the presence of flaws

of both statistical models. RMSE is a measure of the vari-ability remaining after fitting a model (see [36] for more details). The interpretation of RMSE is often controversial since its magnitude is not relative and it is useful only for the purpose of comparison with other models.

4 Discussion

First of all, we must keep in mind that we have used an empirical approach to the problem and, therefore, to arrive at a rough estimate of the importance of the role played by the variables governing it. Throughout this process, results have not always conformed to the phe-nomenon itself and may be influenced by the particular methodology used.

Getting back to the data analysis, in the authors’ opinion, it is worth highlighting the importance of using supplementary information such as scatter plots before interpreting correlation coefficients. The correlation coef-ficient is in fact a numerical summary and, as such, it can be reported as a measure of association for any batch of numbers, irrespective of the data structure [37]. Therefore, an examination of Table 2 in conjunction with Figs. 6, 7, 8 and 9 reveals – within the limits of our study – the follow-ing main features:– Model A predictions are moderately affected by qR with

a coefficient of correlation of 0.66, which covers almost 45 % of the variability of the outcomes. This makes the statistical analysis less precise and the interpretation of the data somewhat difficult. On the other hand, qR dras-tically loses importance when a more complex model involving further parameters (models B and C) is con-sidered (see Fig. 7).

– It is possible to see that both h and As do not affect the shear strength – or at best only very weakly. In fact, the results indicate that their natural variations produced a negligible CoV (< 5 %) in outcomes.

– Likewise, only a weak correlation (20 % ≤ r < 40 %) is observed between bw, d, fc (considering its additional variation due to special placing, curing and hardening conditions, and concrete age) and shear capacity. Model A represents a slight exception. However, as previously discussed, model A does not lend itself to a good under-standing of results. The poor influence of both bw and d on the shear strength may appear odd (considering that it is counter to everything that shear design provi-

Fig. 9. Scatter plot of the shear capacity VR,c,q for group 3 of the compara-ble experiments, evaluated with predictive model C, plotted against εcs

19



h depth of cross-sectionhtest depth of cross-section (experimental test)hm,group mean value of grouped htestN = εcs ∙ Es ∙ As axial tensile force on concrete due to re-

strained shrinkage deformationsRH ambient relative humidityR2 coefficient of determinationr correlation coefficientt age of concretets age of concrete at onset of drying shrink-

ageVR,c shear resistanceVR,c,q = VR,c ∙ qR shear resistance with model uncertaintyVtest shear resistance (experimental test)x distance between reinforcement and cen-

troid of cross-sectionY1 variable representing the additional vari-

ation in fc due to special placing, curing and hardening conditions

Y2 variable that mainly reflects variations due to factors not well accounted for by fc (e.g. gravel type and size, composition of cement and other constituents, cli-matic conditions)

α(t,τ) function to take into account the effects of concrete age

gmod model safety factorεca autogenous shrinkage strainεcd drying shrinkage strainεcd,0 total unrestrained drying shrinkageεcs total unrestrained shrinkage strainqR model uncertaintyλ factor taking into account the systematic

variation in fcρ l geometric percentage of longitudinal re-

inforcementρ l,test geometric percentage of longitudinal re-

inforcement (experimental test)σcs shrinkage stressτ duration of loadingΦa,test maximum aggregate size (experimental

test).

References

1. Rebeiz, K. S.: Shear Strength Prediction for Concrete Mem-bers. ASCE Journal of Structural Engineering, 1999, 125, No. 3, pp. 301–308.

2. Silfwerbrand, J.: Samverkan mellan delvis nedbilad betong-platta och pågjutning. Balkförsök. Meddelande nr 142, insti-tutionen för byggnadsstatik, KTH, Stockholm, 1984.

3. ASCE-ACI Committee 426: The Shear Strength of Rein-forced Concrete Members. ASCE Journal of Structural Divi-sion, 1973, 99, No. 6, pp. 1091–1187.

4. ASCE-ACI Committee 445 on Shear and Tension: Recent Approaches to Shear Design of Structural Concrete. ASCE Journal of Structural Engineering, 1998, 124, No. 12, pp. 1375–1417.

5. Park, R., Paulay, T.: Reinforced Concrete Structures, Wiley, New York, 1975.

6. Pickett, G.: The Effect of Change in Moisture Content on the Creep of Concrete Under a Sustained Load. ACI Journal, 1942, 38, pp. 333–356.

or microcracks that may vary in dimensions and ori-entation depending on different factors (e.g. gravel type and size, composition of cement, high rate of restrained shrinkage, climatic and curing conditions). High shrinkage values can also play an important role.

2. Other parameters such as a) depth of cross-section, b) area of reinforcing steel, c) width of cross-section in tension zone, d) effective depth of cross-section, e) con-crete cover and f) concrete compressive strength seem to be of only minor importance.

A likely explanation for the unusual conclusion that the scatter is mostly due to variations in the tensile strength of concrete instead of its compressive strength is that the two variables are of course correlated (being the tensile strength defined as a function of the compressive strength). However, this correlation is estimated to be only weak (r = 39 %) and, therefore, it is less relevant with respect to other factors.

Finally, in the authors’ opinion, the scatter in shear is a complex problem subjected to randomness which can be better understood through the use of strain-based models, as a non-linear suitably calibrated FEM analysis coupled with a probabilistic approach. Together, these will provide a long-term perspective on the question. Un-fortunately, this procedure is also complicated and time-consuming. As a consequence, in order to speed up the survey, these findings will be useful for further research, giving an idea as to which parameters may be convenient for consideration and which of them, if any, are less im-portant or can even be neglected.

Acknowledgements

The authors would like to thank Prof. K.H. Reineck at the University of Stuttgart, Germany, for his time and kindness in providing data and experience on the subject.

Notation

As area of reinforcing steelAs,test area of reinforcing steel (experimental

test)Asm,group mean value of grouped As,testa/dtest shear-to-span ratio (experimental test)bw width of cross-section in tension zonebw,test width of cross-section in tension zone

(experimental test)bwm,group mean value of grouped bw,testd effective depth of cross-sectiondtest effective depth of cross-section (experi-

mental test)dm,group mean value of grouped dtestEs Young’s modulus of steelfc in situ concrete compressive strengthfc,0 concrete compressive strength of groupfc,test concrete compressive strength (experi-

mental test)fcm,group mean value of grouped fc,test fct tensile strength of concretefct,test tensile strength of concrete (experimental

test)

20



28. JCSS Probabilistic Model Code. JCSS-OSTL/DIA/VROU-10-11-2000.

29. McGill, R., Tukey, J. W., Larsen, W. A.: Variations of Box-plots. The American Statistician, 1978, 32, No. 1, pp. 12–16.

30. Velleman, P. F., Hoaglin, D. C.: Applications, Basics, and Computing of Exploratory Data Analysis, Duxbury Press, PacificGrove, 1981.

31. Nelson, L. S.: Evaluating Overlapping Confidence Intervals. Journal of Quality Technology, 1989, 21, pp. 140–141.

32. Chana, P. S.: Some aspects of modelling the behaviour of reinforced concrete under shear loading, Tech. Rep. No.543, Cement and Concrete Association, Wexham Springs, 21, 1981.

33. Rao, C. R.: Linear Statistical Inference and its Applications, 2nd ed. Wiley, New York, 1973.

34. Snedecor, G. W., Cochran, W. G.: Statistical Methods, 6th ed., Iowa State Univ. Press, Ames, 1967.

35. Garcia, E.: A Tutorial on Correlation Coefficients, http://www.miislita.com, 2010 – simmons.edu

36. Willmott, C. J., Matsuura, K.: Advantages of the mean abso-lute error (MAE) over the root mean square error (RMSE) in assessing average model performance. Clim. Res., 2005, 30, pp. 79–82.

37. Dallal, G. E.: Correlation Coefficients, http://www.jerrydal-lal.com/LHSP/corr.htm, 1999.

38. Sato, R., Kawakane, H., Kawamoto, T., Ushio, R., Maruyama, I.: Shrinkage Effect on Shear Strength of Reinforced, High Strength Concrete Beams. ACI Structural Implications of Shrinkage and Creep of Concrete, 2007, pp. 261–278.

39. Gebreyouhannes, E., Maekawa, K.: Numerical Simulation on Shear Capacity and Post-Peak Ductility of Reinforced High-Strength Concrete Coupled with Autogenous Shrink-age. Journal of Advanced Concrete Technology, 2011, 9, No. 1, pp. 73–88.

40. Hyodo, H., Sato, R., Kawai, K., Nakayama, H.: Effects of Drying Shrinkage on Shear Tension Strength of Reinforced Concrete Beams. FraMCoS-8, VIII Intl. Conf. on Fracture Mechanics of Concrete and Concrete Structures, 10–14 March 2013, Toledo, Spain.

Filippo Sangiorgio(corresponding author)Division of Structural Engineering and BridgesKTH Royal Institute of TechnologyBrinellvägen 23 – 10044 Stockholm, SwedenPhone: +46 8 790 9047Fax.: +46 8 790 [email protected]

Prof. Dr. Johan SilfwerbrandDivision of Structural Engineering and BridgesKTH Royal Institute of TechnologyBrinellvägen 23 – 10044 Stockholm, Sweden

Prof. Dr. Giuseppe ManciniDepartment of Structural, Geotechnical and Building EngineeringPolitecnico di TorinoCorso Duca degli Abruzzi, 24 – 10129 Torino, Italy

7. Neville, A. M., Brooks, J. J.: Concrete Technology, 2nd ed., Pearson, New Jersey, 2010.

8. CEB Bulletin d’Information n. 142/142 bis: Design Manual on Structural Effects of Time-Dependent Behaviour of Con-crete. Georgi Pub. Co. (Saint-Saphorin, Switzerland & Brookfield, Vt.), 1984.

9. Bažant, Z. P., Kim, J. K., Panula, L.: Improved Prediction Model for Time-Dependent Deformations of Concrete: Part 1 – Shrinkage. Mater. Struct., 1991, 24, pp. 327–345.

10. Bažant, Z. P., Kim, J. K.: Improved Prediction Model for Time-Dependent Deformations of Concrete: Part 2 – Basic Creep. Mater. Struct., 1991, 24, pp. 409–421.

11. Bažant, Z. P., Kim, J. K.: Improved Prediction Model for Time-Dependent Deformations of Concrete: Part 3 – Creep at Drying. Mater. Struct., 1992, 25, pp. 21–28.

12. Bažant, Z. P., Kim, J. K.: Improved Prediction Model for Time-Dependent Deformations of Concrete: Part 4 – Tem-perature Effects. Mater. Struct., 1992, 25, pp. 84–94.

13. Bažant, Z. P., Kim, J. K.: Improved Prediction Model for Time-Dependent Deformations of Concrete: Part 5 – Cyclic Load and Cyclic Humidity. Mater. Struct., 1992, 25, pp. 163–169.

14. Bažant, Z. P., Panula, L., Kim, J. K., Xi, Y.: Improved Predic-tion Model for Time-Dependent Deformations of Concrete: Part 6 – Simplified Code-Type Formulation. Mater. Struct., 1992, 25, pp. 219–223.

15. Bažant, Z. P., Xi, Y., Baweja, S.: Improved Prediction Model for Time-Dependent Deformations of Concrete: Part 7 – Short Form of BP-KX Model, Statistics, and Extrapolation of Short-Time Data. Mater. Struct., 1993, 26, pp. 567–574.

16. ACI Committee 209: Prediction of Creep, Shrinkage, and Temperature Effect on Concrete Structures. American Con-crete Institute, Detroit, 1992.

17. Reineck, K.H., Bentz, E. C., Fitik, B., Kuchma, D. A., Bayrak, O.: ACI-DAfStb Database of Shear Tests on Slender Reinforced Concrete Beams without Stirrups. ACI Structur-al Journal, 2013, 110, No. 5, pp. 867–875.

18. Sangiorgio, F., Silfwerbrand, J., Mancini, G.: Assessment of the ACI-DAfStb Database of Shear Tests on Slender Rein-forced Concrete Beams without Stirrups for Investigations on the Shear Capacity Scatter. Athens Journal of Technolo-gy, 2014, No. 3, pp. 181–197.

19. Bresler, B., Scordelis, A. C.: Shear Strength of Reinforced Concrete Beam. ACI Journal, 1963, 60, No. 1, pp. 51–72.

20. Cho, S.H.: Shear Strength Prediction by Modified Plasticity Theory for Short Beams. ACI Structural Journal, 2003, 100, No. 1, pp. 105–112.

21. Song, J., Kang, W. H.: Probabilistic shear strength models for reinforced concrete beams without shear reinforcement. Struc-tural Engineering and Mechanics, 2010, 34, No. 1, pp. 15–38.

22. Silfwerbrand, J.: Differential Shrinkage in Normal and High Strength Concrete Overlays. Nordic Concrete Research – Publications, 1996, 19, pp. 55–68.

23. Hamming, R. W.: Numerical Methods for Scientists and En-gineers, McGraw-Hill, New York, 1962.

24. Eurocode 2 (EN 1992-1-1): Design of concrete structures – Part 1-1: General rules and rules for buildings, 2004.

25. Sherwood, E. G., Bentz, E. C., Collins, M. P.: Effect of aggre-gate size on beam-shear strength of thick slabs. ACI Struc-tural Journal, 2007, 104, No. 2, pp. 180–190.

26. Zandi Hanjari, K., Flansbjer, M., Lindqvist, J. E., Silfwerbrand, J.: Structural analysis of concrete members with shear failure. Proc. of fib Symp. on Concrete Structures for Sus-tainable Community, 11–12 June 2012, Stockholm, pp. 165–168.

27. Droguett, E. L. Mosleh, A.: Bayesian Methodology for Mod-el Uncertainty Using Model Performance Data. Risk Analy-sis, 2008, 28, No. 5, pp. 1457–1476.


Compressive arch action is one of the main resistance mecha-nisms against progressive collapse in reinforced concrete (RC) buildings. Hence, many studies have investigated the develop-ment of arching action in RC beams and frames but less attention has been paid to calculating the corresponding enhancement in structural capacity. In the present study, a theoretical method is introduced in order to calculate the arching capacity of RC beams and also to obtain a quantitative assessment regarding structural robustness against progressive collapse. The proposed method is validated using the experiments in the literature. The evaluation indicates that the procedure introduced here could establish a reliable foundation for estimating the arching capacity of beams and also structural robustness.

Keywords: compressive arch action, reinforced concrete beams, structural robustness, arching capacity, progressive collapse, structural analysis

1 Introduction

Progressive collapse is caused by the initial local failure in a structure and its spread to other parts, which, in the end, can lead to disproportionate collapse of the structure. In recent years, many researchers have studied the progres-sive collapse of reinforced concrete (RC) structures in order to understand and also improve resistance mecha-nisms such as compressive arch action (CAA), frame (Vierendeel) action, suspension action and also catenary action. Compressive arch action develops at early stages after the initial failure and hence plays a significant role in controlling the spread of failures and also the redistribu-tion of unbalanced forces within the building. Therefore, studying the ensuing effects along with the development of arching action in RC structures were the main focus of several recent research projects. Experimental research work performed by FarhangVesal et al. [1], Lew et al. [2], Su et al. [3], Choi and Kim [4], Yu and Tan [5], [6], Yu et al. [7] and Yi et al. [8] are important studies where the struc-tural responses of RC sub-assemblages including three

columns and two beams were tested for progressive col-lapse. Among all the studies mentioned, only Lew et al. [2] performed real-scale tests and the rest of the experiments focused on scaled specimens. Following these studies, Gu [9] and Qian et al. [10] investigated the effects of transverse beams and slabs. Magnusson et al. [11] discussed arching action in RC beams under dynamic out-of-plane loadings. They focused on the enhancement of the shear capacity of RC beams subjected to extreme loads. In addition to sub-assemblages, Stinger and Orton [12] tested a larger part of an RC frame and investigated the performance of different resistance mechanisms, including compressive arch ac-tion, against progressive collapse. Sasani et al. [13]–[18] performed comprehensive experimental programmes on real RC structures. Breugel [19] discussed different aspects of a protective RC building against extreme events that could lead to disproportionate collapse of the structure.

In order to obtain a quantitative assessment of RC structures, Jian and Zheng [20] introduced a theoretical model to predict the general behaviour of RC sub-assem-blages against progressive collapse. In their model, com-pressive arch capacity is computed based on the classic flexural resistance of RC sections while neglecting the en-hancement due to arching action. Su et al. [3] calculated compressive arch capacity based on a repetitive procedure that was first presented by Park and Gamble [21] for RC slabs. Yu and Tan [22] introduced a theoretical model in order to calculate the compressive arch capacity of RC sub-assemblages based on a repetitive procedure. Accord-ing to a comprehensive review of the progressive collapse of structures by Yagob et al. [23], efficient and inexpensive design methods are needed in order to provide structures resistant to abnormal loading events. More recent stud-ies, [20] and [22], also demonstrate the same trend among researchers. Although many studies have developed theo-retical methods for estimating the catenary capacity of RC sections, there are no similar approaches for compressive arch resistance due to the complexities and also consider-able interaction of arching and flexural actions. Hence, the present study introduces a theoretical approach for estimating the arching capacity of RC beams under large deformations caused by progressive collapse. The present method is developed based on the procedure introduced previously by Rankin and Long [24] for one-way slabs. The approach introduced here estimates compressive arch capacity based on the geometrical and structural

Technical Paper

A theoretical method for calculating the compressive arch capacity of RC beams against progressive collapse

Reza AbbasniaFoad Mohajeri Nav*

DOI: 10.1002/suco.201400119


Submitted for review: 22 December 2014; revision: 20 March 2015; accepted for publication: 10 May 2015. Discussion on this paper must be submitted within two months of the print publication. The discussion will then be published in print, along with the authors’ closure, if any, approximately nine months after the print publication.


R. Abbasnia/F. Mohajeri Nav · Method for computing RC beam arch capacity against collapse

details of beams and without any need to perform a repeti-tive procedure. It also determines structural robustness based on the arching capacity of beams and specifies whether or not collapse continues beyond the arching ac-tion. The proposed method is validated by experiments in the literature. Results indicate that the present approach could achieve a reliable prediction of the arching capacity of RC beams against progressive collapse.

2 Development of the theoretical method 2.1 General concept

Once a column is removed, cracking of the concrete hap-pens at the joint interfaces (above the removed column) and leads to movement of the neutral axis. Consequently, the beams connected to the removed column tend to elon-gate, but the restraint provided by other elements leads to the development of compressive arch action within the beams. Hence, a consequential resistance enhancement would be expected. Fig. 1a demonstrates the arching ac-tion within the beams connected to the removed column.

Deformation of unreinforced masonry walls that are surrounded by two rigid supports was studied by McDowell et al. [25]. Rankin and Long [24] developed this method for one-way slabs. In the present study, this meth-od is modified so that it can be used for RC beams under-going progressive collapse within an RC sub-assemblage. The idealized model of removed column and connected beams forms a sub-assemblage that is depicted in Fig. 1b.

Resistance of beams against progressive collapse up to the end of the arching stage is due to two mechanisms: 1) flexural action and 2) arching action. Therefore, in the present study, the capacity of beams based on each action is calculated separately and then added together to obtain the ultimate strength of the beams. The development of the present method is based on the following assumptions:a) The maximum arching moment happens after initial

yielding of the longitudinal bars.b) Flexural deformations are neglected in comparison to

large displacements due to the progressive collapse.c) Crack patterns and failure modes on both sides of the

removed column are symmetrical.d) The depths of the compression zones at both ends of the

beams connected to the removed column are similar.e) Only the role of the beams is considered in the present

study and the effects of floor or slab are neglected.

Since compressive arch capacity occurs at a vertical displacement of 0.18h to 0.46h at the joint above the

removed column (h = depth of beam) [6], logically, this would be after the first yielding of the longitudinal bars, as is stated in the first assumption. In addition, since column removal causes much larger displacements in comparison to flexural deformations, the second assumption is also acceptable.

The third assumption will be valid only if the re-moved member is an internal or external column and the beam spans on both sides are equal. According to experi-ments [11]–[18], in the situation mentioned, the develop-ment of the failure on both sides of the removed column is almost symmetrical and so the third assumption will be satisfied. Only corner columns or different beam spans can jeopardize symmetric failure modes.

Top longitudinal bars in the beams connected to the removed column experience different forces at their two ends. Whereas they are carrying compressive forces next to the removed column, tensile forces are dominant at the other end of the beam. Bottom reinforcement faces a similar situation. Since the longitudinal bars do not have the same types of force along the beam, they cannot have a significant effect on the arching action and concrete plays the main role. Therefore, as long as the concrete withstands the compression, the development of arching strength continues. Once the concrete crushes, the com-pressive arch strength decreases [6]. In addition, regarding small deflections at the arching stage in comparison to the ultimate displacement, the axial compressive forces can be assumed constant along the beam [3], [6]. Hence, con-cerning the main role of the concrete in the development of these forces, the fourth assumption, which presumes a similar depth for the compression zones at both ends of the beams, will be reasonable.

Owing to the high costs and complexity of laborato-ry studies, most of the experiments are performed on bare frames and assemblies that do not consider the effects of floors or slabs. Hence, in the present study, the influence of a floor is neglected. However, since the present method is extended based on the slab theories, where sufficient experimental data is available for validation study, the method is easily extended for considering the slab effects.

2.2 Theoretical basis

Fig. 2a depicts the essential parameters for defining strain at the contact surfaces at the two ends of the beam. Ow-ing to the symmetry, a similar strain distribution applies on both sides of the removed column. In addition, it is presumed that beams can transfer only the compressive

Fig. 1. Compressive arch action: a) in RC beams, b) in an RC sub-assemblage

23



forces to the supports (adjacent columns). Deformation of beams is idealized based on the rotation of two rigid parts about the axes passing through the joint interfaces on both sides of the removed column and at both ends of the beams. As Fig. 2b demonstrates, it is also assumed that point A is the centre of rotation at both ends of the beams.

Point B′ in Fig. 2a marks the end of the contact surface at the joint interfaces above the removed column. Therefore, point B′ should be located on the axis pass-ing through the beam-column interface and the distance between B and B′ represents the vertical displacement of the joint above the removed column. Thus, as B and B′ are located in the same vertical plane, the centre of rotation, point A, should be placed halfway between these two points (BB′ = 4a). Hence, the maximum strain eb is calculated based on a linear strain distribution over the beam depth:

where:δ maximum deflection at joint above removed column

(s = BB′)d1 half the depth of the arching sectionL beam span

In addition, an elasto-plastic stress-strain relationship previously defined by Rankin and Long [24] is used to describe the stress distribution at the beam ends. Eq. (2) defines this plastic strain ec:

where fc (MPa) is the compressive strength of the con-crete. Accordingly, concrete in the compressive zone at both ends of the beam, the shaded area in Fig. 2a, can be in two different conditions: the maximum strain eb either smaller or larger than the plastic strain. Therefore, stress and strain distributions in these two states can be defined as is shown in Fig. 2c and the flexural moments at beam

2 2( ) tan 2

4(1)1

2 1LL

d a

L Ldb bε

θε δ δ= ⋅ ∆ = ⋅

−⇒ = −

( 400 60 0.33 ) 10 (2)2 6f fb c cε = − + − ⋅ −

sections are calculated separately based on the depicted distributions. The lever arm for the first case is shown in Fig. 2d. Using a similar method, the lever arm in the lat-ter case can be easily calculated. Based on the calculation mentioned, moment relations are extracted for both cases. McDowell et al. [25] used dimensionless parameters in or-der to summarize and categorize the equations obtained. Eqs. (3) to (5) define these parameters:

where Mu and Mr are the arching moment and arching moment ratio respectively. Table 1 lists the calculated equations based on these parameters.

In order to obtain the maximum arching moment ra-tio, the equations presented are differentiated with respect to u. The first equation has a root u = 0.31, but in the sec-ond equation, u is obtained as a function of R. Based on these results, Rankin and Long [24] obtained a polynomial relationship between Mr and R according to Eq. (6):

The boundary between these two relations is computed by replacing the value u = 0.31 in Eqs. (1) and (4) and comparing with the plastic strain ec. Using the procedure mentioned, the plastic strain in the concrete for R > 0.26 is not exceeded and the plastic strain in the concrete for 0 < R < 0.26 is exceeded. Noting that the maximum stress in the concrete is sc = 0.85fc , the maximum arching moment Mu max per unit width of the beam is calculated using Eq. (7):

4(3)

2

12

RL

dcε

=

2(4)

1

udδ=

4(5)

12

MM

dr

u

cσ=

( ) 0.260.3615

( ) 0 0.26 4.3 16.1 (6)

3.3 10 0.12434

i R MR

ii R M

R

r

r

> ⇒ =

< < ⇒ = −

⋅ +−

Fig. 2. Geometry of deformations and stress distributions at both ends of the beams



According to the calculations, the maximum value for u is 0.31, which means a vertical displacement δ = 0.62d1 at the middle joint above the removed column. By assuming 2d1 = h, the resulting vertical displacement is δ = 0.31h, which is approximately in the range of the experimental results (0.18h to 0.46h). In addition, according to Chris-tiansen [26], the available depth that takes part in arching action is equal to the depth that remains after removing the zone required for balancing compressive and tensile forces. Thus, the depth mentioned, 2d1, is calculated using Eq. (8):

Both ends of a beam in an RC frame are restrained against axial displacement, which is obviously not a rigid re-straint. Hence, the situation of the beams after removing the connected column can be described by a three-hinged arch and a linear elastic spring, as is depicted in Fig. 3a. The spring represents the restraint of the beam against ax-ial displacement. A larger stiffness for the spring leads to a higher degree of restraint. At the end of the arching stage, the concrete is crushed completely and cannot participate in carrying vertical loads. Hence, a tensile force develops in the longitudinal reinforcement and catenary action be-gins. In an actual two-hinged arch, formation of the third hinge leads to an unstable arch which ends with structural collapse. In contrast, in an RC beam, the catenary action of the longitudinal reinforcement makes a different three-hinged arch that could still withstand collapse.

0.85

4(7)max

12

Mf d

Muc

r=

2 ( )0.85

(8)1d hf d

fy

c

ρ ρ= − + ′

Regarding the equilibrium of a three-hinged arch, if the span of an equivalent rigidly restrained beam Lr is equal to Eq. (9), the load-deflection response of elastic and rigidly restrained beams will be the same.

where:le net span of each beam with elastic restraintsA area of equivalent compression zoneEc elastic modulusK stiffness of restraining spring

Owing to the similar geometry and boundary conditions, the equation mentioned is modified so that it can be used in RC beams. Hence, the static equilibrium of three-hinged arches with elastic and rigid restraint relates the arching moments according to Eq. (10):

where Mumax is the maximum arching moment in a three-hinged arch with rigid restraint and MCAA the correspond-ing value for an arch with elastic restraint. The equivalent vertical load corresponding to the compressive arch ac-tion PCAA is determined based on the classic structural analysis according to Eq. (11):

where ln and ln′ are the net spans of the beams. Hence, the ultimate capacity of the beams is computed using Eq. (12):

where Pf is the capacity due to flexural action. Flexure is the dominant mechanism in beams under normal loading. Therefore, in order to obtain the ultimate flexural capac-ity, the bending strength of beams is calculated and Pf is computed based on the vertical equilibrium:

where Mp1, Mp2, Mp1′ and Mp2′ are the plastic moments at the two ends of the beams. Using Eq. (9) needs a value

1 (9)3L lE A

Klr ec

e

= +

(10)maxMl

LMCAA

e

ru=

4(11)P

M

l lCAACAA

n n

=+ ′

(12)P P Pu f CAA= +

l(13)1 2 1 2P

M M

l

M Mf

P P

n

P P

n

=+

++

′ ′

′

Fig. 3. Equivalent three-hinged arch: a) elastic restraint, b) rigid restraint

Table 1. Analysis of arching moment [25]

Equation No.

Strain condition

Strain distribution

Stress distribution σσ

==MMdr

u

c

4

12

A eb max < ec

−

−

83

12

1 54

2uR

u u

B eb max ≥ ec+ + − −

4 12

34

2 23

2

2R u u R

u

25



for lateral stiffness of beams K. Hon et al. [27] developed a method for estimating the lateral stiffness of RC slabs based on a simplified model of the structural system. This method is modified here so that it can be used for RC beams. In their method, the axial stiffness of a restrained slab is defined using a set of springs where the stiffness of each spring depends on the width of the slab. In order to modify this method, the axial stiffness of each beam is modelled using a separate spring. Adjacent members of the sub-assemblage (beams and slabs) are also considered according to Fig. 4, which demonstrates the general con-cept of the method. A horizontal unit load is applied at both ends of the beams and the corresponding displace-ment is computed based on a linear elastic analysis.

2.3 Structural robustness

This section focuses on the development of a practical tool in order to determine the status of the building based on the compressive arch capacity of the beams. Following this goal, a non-dimensional parameter λ is defined ac-cording to Eq. (14):

where:N axial force in removed column before eliminationPu

(i) compressive arch capacity of RC beams at ith storeyn number of storeys above removed columnφ reduction factor (0.9)λ arching ratio

Since all the beams above the removed column in a two-dimensional frame will take part in resisting progressive collapse, they are considered in the formulations present-ed here. According to the experimental observations [14], [17], the vertical displacement above the removed column decreases in upper storeys due to column elongation and, necessarily, the ultimate capacity of the beams in different

(14)( )

N

Pui

i l

n

∑λ

φ

=

=

storeys does not occur simultaneously. In other words, whereas beams in upper storeys could sustain more load, the beams of lower storeys have reached their arching ca-pacity due to the larger vertical displacements. Therefore, some of the beams will not reach their ultimate capacity and the summation of flexural moments developed at the ends of these beams could be estimated at about 90 % of the expected theoretical capacity. This is obtained based on an evaluation performed on the extracted moment diagrams for RC frames available in the literature [14], [17]. Hence, a reduction factor of 0.9 is utilized. Accord-ing to the formulation developed, λ < 1 indicates that the building is able to resist progressive collapse based on the compressive arch capacity of the beams, and λ > 1 denotes that failure continues beyond the arching action.

It should be noted that in the development of the present method, only the resistance of the beams in the plane of the frame is considered. However, the framework presented could be utilized separately for calculating the arching capacity of the transverse beams and the ad-ditional strength could be added to obtain the ultimate resistance of the building. This procedure neglects the ef-fects of the slabs.

2.4 Summary of the method

Regarding the methodology mentioned, the CAA capacity of the beams can be obtained by following the procedure shown in Fig. 5 and, consequently, structural robustness against progressive collapse can be also assessed based on the arching ratio.

3 Validation study

A complete set of experiments from the literature was used to evaluate the capability of the proposed method. These experiments were performed on RC sub-assemblag-es including two beams and three columns. Table 2 lists these sub-assemblages and compares the theoretical pre-dictions with the experimental results. Lateral stiffnesses values obtained based on the analytical method are also

Fig. 4. Determination of the axial stiffness of the sub-assemblage: a) original plan, b) equivalent model for an elastic analysis



presented in Table 2. In order to quantify the strength of the relationship between theoretical and experimental results, the correlation coefficient is also calculated.

Theoretical and experimental results are also com-pared graphically in Fig. 6. As is shown in Fig. 6 and con-firmed by the correlation coefficient, a limited discrepancy existed between theoretical and experimental capacities.

According to the results, the theoretical predictions of the models belonging to the study by Choi and Kim [4] are larger than the experimental capacities. One of the columns in these sub-assemblages is an external column in the original building and the others (including the removed one) are internal columns. Typically, the internal columns in a building are stiffer than the external columns and external joints thus experience larger rotations and displacements. Therefore, the dimensions of one of the lat-eral columns (an internal column in the original building) are twice those of the other lateral column (an external column in the original building). The design described causes a relative asymmetry in the sub-assemblages and, consequently, leads to damage being concentrated on the weaker side. This phenomenon challenges the third as-sumption of the proposed method, which necessitates a symmetric failure mode in the sub-assemblage; hence, the observed difference between experimental and theoretical capacities is obtained.

Fig. 7 presents positive errors where the theoretical predictions are larger than the experimental ones. Accord-ing to Fig. 7, the mean and maximum percentage errors are 4.4 and 14.6 % respectively. The models of Choi and Kim [4] have error values > 3 % and the reason for this has been discussed before. In addition, according to the re-sults reported by Lew et al. [2], during the development of compressive arch action in the IMF model, concentration of damage and cracking on one side of the removed col-umn contravenes the symmetric conditions in the beams on both sides of the middle joint. Fig. 8a demonstrates the situation described in the IMF model, whereas according to Fig. 8b, almost symmetric crack patterns in the SMF model lead to a precise theoretical prediction. In real buildings where beams are working as parts of a frame, when a column is removed, the axial deformation of the beams and rotation of the joint above the removed column Fig. 5. Procedure for calculating ultimate arching capacity and structural robustness

Fig. 6. Comparison of measured and calculated results for loading capacity under compressive arch action

27



are restricted due to the other members of the frame. In ad-dition, experimental research [5], [6] indicates that in an RC sub-assemblage without a rotational restraint at the middle joint, more cracks develop on one side of the middle joint in comparison to the other side due to the flawed geometry and material non-uniformity. Hence, the cracks in the sub-sequent loading process would concentrate on the weaker side. Consequently, the lack of a rotational restraint in the sub-assemblages helps the middle joint to rotate towards the more severely cracked side. Experiments performed on

real buildings [13]–[18] and multi-storey frames [8] indicate that rotation of the joint above the removed column is restricted by the columns of upper storeys. Hence, the mid-dle joint cannot rotate freely and the symmetric crack pat-tern generally happens in reality. So this difference is not expected in real RC frames and regarding the fact that the present method assumes a symmetric failure mode in the beams, it can be claimed that the proposed procedure pre-sents a precise prediction of the compressive arch capacity of beams with symmetric crack patterns.

Table 2. Theoretical (Theo.) results in comparison to experimental (Exp.) responses

Model Reference ln/h(1) Long. reinft. ratio (2) (%)

Scale Restraint stiffness (kN/m)

Ultimate capacity (kN)

Theo./Exp.

Top Bottom Exp. Theo.

S1

Yu & Tan [5], [6]

11.00 0.90 0.49

1/2

1.06E + 05 41.6 41.9 1.008

S2 11.00 0.73 0.49 1.06E + 05 38.4 38.7 1.008

S3 11.00 1.24 0.49 4.29E + 05 54.5 49.1 0.902

S4 11.00 1.24 0.82 4.29E + 05 63.2 55.2 0.872

S5 11.00 1.24 1.24 4.29E + 05 70.3 62.4 0.887

S6 11.00 1.87 0.82 4.29E + 05 70.3 68.7 0.977

S7 8.60 1.24 0.82 4.29E + 05 82.8 74.3 0.897

S8 4.60 1.24 0.82 4.29E + 05 121.3 107.6 0.886

A1

Su et al. [3]

4.08 0.55 0.55

1/3

1.00E + 06 168.0 152.3 0.907

A2 4.08 0.83 0.83 1.00E + 06 221.0 181.7 0.822

A3 4.08 1.13 1.13 1.00E + 06 246.0 212.9 0.865

A4 4.08 0.55 0.38 1.00E + 06 147.0 136.1 0.926

A5 4.08 0.83 0.55 1.00E + 06 198.0 165.8 0.838

A6 4.08 1.13 0.75 1.00E + 06 226.0 190.8 0.844

B1 6.58 1.13 1.13 1.00E + 06 125.0 100.2 0.802

B2 9.08 1.13 1.13 1.00E + 06 82.9 70.5 0.850

B3 9.08 1.13 0.75 1.00E + 06 74.7 65.9 0.883

V1

FarhangVesali et al. [1]

11.72 0.51 0.51

2/5

1.00E + 06 40.5 35.2 0.868

V2 11.72 0.51 0.51 1.00E + 06 35.7 34.2 0.959

V3 11.72 0.51 0.51 1.00E + 06 41.4 35.0 0.846

V4 11.72 0.77 0.51 1.00E + 06 40.1 39.2 0.977

V5 11.72 0.77 0.51 1.00E + 06 41.6 40.2 0.965

V6 11.72 0.77 0.51 1.00E + 06 39.4 40.3 1.023

IMFLew et al. [2]

10.77 0.64 0.411/1

5.06E + 05 296.0 339.1 1.146

SMF 7.96 0.68 0.59 5.06E + 05 903.0 916.1 1.015

P1Qian et al. [10]

10.56 0.92 0.921/4

1.00E + 06 32.0 29.7 0.928

P2 9.29 1.51 1.51 1.00E + 06 36.0 30.9 0.857

M Stinger & Orton [12] 9.44 1.17 0.88 3/8 1.00E + 06 25.8 24.1 0.933

B1A

Gu [9]

12.00 0.86 0.86

1/4

1.06E + 06 18.0 17.4 0.967

MB1(3) 12.00 0.86 0.86 1.06E + 06 17.5 16.9 0.967

MB2 (4) 9.00 1.49 0.83 1.06E + 06 14.7 17.3 0.948

5S

Choi & Kim [4]

7.78 1.16 0.46 1/2.70 1.00E + 06 39.9 41.2 1.031

5G 9.46 0.58 0.58 1/2.70 1.00E + 06 22.8 24.4 1.072

8S 8.46 1.46 0.87 1/2.85 1.00E + 06 54.1 57.3 1.060

8G 10.31 0.82 0.82 1/2.85 1.00E + 06 23.7 24.5 1.033

Mean value of theoretical to experimental ratios 0.936Correlation coefficient 0.996

(1) Span-to-depth ratio(2) Longitudinal reinforcement ratio(3) Known as B1 in the original research(4) Known as B2 in the original research



Fig. 9 demonstrates the negative errors, meaning smaller theoretical values in comparison to the experi-mental ones. The mean percentage error of 10.1 % and the maximum percentage error of 19.8 % indicate the capabil-ity of the proposed method when it comes to predicting the ultimate arching capacity of RC beams. The main reason for obtaining smaller theoretical values is related to hardening of the longitudinal steel bars, which happens

while the ultimate capacity is obtained. This hardening, which occurs after yielding of the tension reinforcement, is neglected in the proposed method and, therefore, small-er theoretical predictions are obtained. According to the results and discussions, it can be claimed that in the case of having a symmetric situation in the beams, the theoreti-cal prediction of the ultimate arching capacity lies below the real strength with an acceptable accuracy.

Fig. 8. Crack patterns during compressive arch action: a) IMF model with unsymmetric pattern, b) SMF model with symmetric pattern

Fig. 9. Negative difference between experimental and theoretical predictions

Fig. 7. Positive difference between experimental and theoretical predictions

29



Fig. 10 depicts the theoretical and experimental arching capacity of all the models investigated which are normalized separately with respect to the flexural capacity of the respective beam. The general trend of normalized arching capacities is shown based on the span-to-depth ra-tio. The span-to-depth ratio for each model was presented previously in Table 2. As expected, by increasing the span-to-depth ratio, the arching capacity of each beam de-creases. Beams with larger spans need stronger restraints in order to restrict the horizontal and rotational deforma-

Fig. 10. Equivalent three-hinged arch: a) elastic restraint, b) rigid restraint

tions, but such conditions are not generally provided and so the ultimate arching capacity decreases. According to Fig. 10, the general trend of the normalized arching ca-pacities is similar for both theoretical and experimental results and a general reduction is obtained. The mean and maximum values of capacity enhancement based on the proposed method are 32.2 and 140.8 % respectively; the corresponding values based on the experimental results are 43.1 and 159.1 % respectively. As the experiments indicate and the proposed method also confirms, the de-

Table 3. Resistance to progressive collapse

Model Loads (kN/m) Lr(a) (m) n (b) N(c) (kN) Structural robustness (λ)

Dead Live Dead (roof) Live (roof) Theoretical Experimental

S1

6.33 7.20 6.33 1.81 3.000 5 186.8

1.06 1.00

S2 1.17 1.08

S3 0.89 0.76

S4 0.77 0.66

S5 0.66 0.59

S6 0.54 0.59

S7 0.59 0.50

S8 0.42 0.34

V1

4.56 4.8 4.56 1.38 2.200 7 136.6

0.62 0.54

V2 0.63 0.61

V3 0.62 0.52

V4 0.55 0.54

V5 0.54 0.52

V6 0.54 0.55

IMF34.00 21.90 29.58 5.49 6.096 7 2258.4

1.11 1.21

SMF 0.41 0.40

5S11.36 12.74 11.36 3.65 1.852 5 206.3

1.11 1.15

5G 1.88 2.01

8S10.19 12.05 10.19 3.46 1.755 8 258.2

0.63 0.66

8G 1.46 1.51

(a) Removed column’s bearing length(b) Number of floors above removed column (including roof)(c) Estimated axial force in removed column



velopment of arching capacity is directly related to the de-velopment of axial forces and also the geometric specifica-tions of RC members (such as longitudinal reinforcement ratio and span-to-depth ratio). Based on these factors, a significant development equal to 140.8 % is obtained for the ultimate capacity of model A4, whereas in another cases, such as model M, a capacity enhancement of only 3.7 % is observed.

In order to investigate the robustness of the RC frames where these sub-assemblages are a part of such frames, they are evaluated using the arching ratio intro-duced here. Since the real value of the column’s axial force in these structures is not known, it is calculated based on the reported gravity loads in the references [1]–[6], [9], [10], [12]. Hence, for those where the details are not reported, this evaluation is not performed. It is assumed that all the beams of the upper storeys have the same dimensions and reinforcement. According to Table 3, n indicates the number of storeys for which their load should be redistrib-uted after column removal. Consequently, the axial force in the removed column is calculated and the resistance of the structures is computed based on the arching ratio. Table 3 lists the results using theoretical and experimental predictions separately. According to the results based on the theoretical predictions, S3 to S8, V1 to V6, SMF and 8S could resist progressive collapse. However, in others the arching capacity of the beams is not sufficient to with-stand failure and collapse continues beyond the arching stage. The corresponding values based on the experimen-tal results also demonstrate the same trend.

4 Conclusions

In the present study, a theoretical approach is developed in order to predict the ultimate arching capacity of RC beams. The proposed method is extended based on the theories presented for calculating the compressive mem-brane capacity of RC slabs. It also expresses the structural robustness with regard to progressive collapse based on the arching capacity of beams. A collection of 35 experimental studies in the literature is used to validate the method pre-sented. The following conclusions can be drawn:– Where there are symmetric situations on both sides of the

removed column, the proposed method can supply a precise prediction of the ultimate arching capacity of the beams.

– The arching capacity of symmetric sub-assemblages is estimated precisely (mean error < 10 %) and the results are on the safe side in comparison to experiments.

– Smaller capacity predictions are related to hardening, which occurs after yielding of the tension bars but is neglected in the proposed method.

– If symmetry does not exist on both sides of the removed column, the theoretical predictions will be larger than the actual values. The difference between theoretical and ac-tual capacities depends on the degree of asymmetry, with more asymmetric situations leading to larger differences.

– A larger theoretical prediction is related to the concen-tration of damage at weaker points of the structure, which occurs due to asymmetric conditions. In symmet-ric beams, these failures are distributed over different points of the structure and, hence, the actual arching capacity of the beams increases.

– Based on the evaluation study, by increasing the span-to-depth ratio of the beams, the arching capacity de-creases, which is consistent with experimental results.

– The proposed method quantifies the structural robust-ness based on the arching capacity of the beams.

– The proposed method is simple, practical and easy to understand, and can be used for designing and retrofit-ting RC buildings.

– The evaluation study shows that the proposed method could establish a reliable foundation for predicting arch-ing capacity in RC beams.

According to the evaluations performed, the proposed meth-od supplies precise results for symmetric situations. Since the existence of asymmetric conditions is probable in RC buildings, it is necessary to extend the proposed method for asymmetric sub-assemblages. In addition, considering the effects of floors in the ultimate capacity of structures is also important and helps to obtain more realistic predictions.

References

1. Farhang Vesali, N., Valipour, H., Samali, B., Foster, S.: Deve-lopment of arching action in longitudinally-restrained rein-forced concrete beams. Construction and Building Materi-als, 2013, 47, pp. 7–19.

2. Lew, H. S., Bao, Y., Pujol, S., Sozen, M. A.: Experimental Study of Reinforced Concrete Assemblies under Column Re-moval Scenario. ACI Structural Journal, 2014, vol. 111, No. 4, pp. 881–892.

3. Su, Y., Tian, Y., Song, X.: Progressive Collapse Resistance of Axially-Restrained Frame Beams. ACI Structural Journal, 2009, vol. 106, No. 5, pp. 600–607.

4. Choi, H., Kim, J.: Progressive collapse-resisting capacity of RC beam–column sub-assemblage. Magazine of Concrete Research, 2011, 63(4), pp. 297–310.

5. Yu, J., Tan, K. H.: Experimental and numerical investigation on progressive collapse resistance of reinforced concrete beam column sub-assemblages. Engineering Structures, 2011, vol. 55, pp. 90–106.

6. Yu, J., Tan, K.: Structural Behavior of RC Beam-Column Sub-assemblages under a Middle Column Removal Scenario. Journal of Structural Engineering (ASCE), 2013, 139(2), pp. 233–250.

7. Yu, J., Rinder, T., Stolz, A., Tan, K., Riedel, W.: Dynamic Pro-gressive Collapse of an RC Assemblage Induced by Contact Detonation. Journal of Structural Engineering (ASCE), 2014, 140(6), 04014014.

8. Yi, W. J., He, Q. F., Xiao, Y., Kunnath, S. K.: Experimental Study on Progressive Collapse-Resistant Behavior of Rein-forced Concrete Frame Structures. ACI Structural Journal, 2008, vol. 105, No. 4, pp. 433–439.

9. Gu, X.: Progressive Collapse Resistance of Reinforced Con-crete Frame Structures. National Natural Science Foundati-on of China and Tongji University, 2012, Report No. 90715004.

10. Qian, K., Li, B., Ma, J. X.: Load-Carrying Mechanism to Re-sist Progressive Collapse of RC Buildings. Journal of Structu-ral Engineering (ASCE), 2014, 04014107.

11. Magnusson, J., Hallgren, M., Ansell, A.: Shear in concrete structures subjected to dynamic loads. Structural Concrete, 2014, 15, No. 1, pp. 55–65.

12. Stinger, S. M., Orton, S. L.: Experimental Evaluation of Dis-proportionate Collapse Resistance in Reinforced Concrete Frames. ACI Structural Journal, 2013, vol. 110, No. 3, pp. 521–530.

31



13. Sasani, M., Bazan, M., Sagiroglu, S.: Experimental and Ana-lytical Progressive Collapse Evaluation of an Actual Rein-forced Concrete Structure. ACI Structural Journal, 2007, vol. 104, No. 6, pp. 731–739.

14. Sasani, M., Sagiroglu, S.: Progressive Collapse Resistance of Hotel San Diego. Journal of Structural Engineering (ASCE), 2008, vol. 134, No. 3, pp. 474–488.

15. Sasani, M., Sagiroglu, S.: Progressive Collapse of RC Struc-tures: A Multi-hazard Perspective. ACI Structural Journal, 2008, vol. 105, No. 1, pp. 96–103.

16. Sasani, M.: Response of a Reinforced Concrete Infilled-fra-me Structure to Removal of Two Adjacent Columns. Engi-neering Structures, 2008, vol. 30, No. 9, pp. 2478–2491.

17. Sasani, M., Kazemi, A., Sagiroglu, S., Forest, S.: Progressive Collapse Resistance of an Actual 11-story Structure Subjec-ted to Severe Initial Damage. Journal of Structural Enginee-ring (ASCE), 2011, vol. 137, No. 9, pp. 893–902.

18. Sasani, M., Sagiroglu, S.: Gravity Load Redistribution and Progressive Collapse Resistance of a 20-story RC Structure Following Loss of an Interior Column. ACI Structural Jour-nal, 2010, vol. 107, No. 6, pp. 636–644.

19. Breugel, K. V.: Structural systems for protection against extre-me events. Structural Concrete, 2011, 12, No. 1, pp. 5–12.

20. Jian, H., Zheng, Y.: Simplified Models of Progressive Collap-se Response and Progressive Collapse-Resisting Capacity

Curve of RC Beam-Column Substructures. Journal of Perfor-mance of Constructed Facilities (ASCE), 2014, 28(4), 04014008.

21. Park, R., Gamble, W. L.: Reinforced Concrete Slabs, 2nd ed., Wiley & Sons, 2000.

22. Yu, J., Tan, K. H.: Analytical model for the capacity of com-pressive arch action of reinforced concrete sub-assemblages. Magazine of Concrete Research, 2014, 66(3), pp. 109–126.

23. Yagob, O.,Galal, K., Naumoski, N.: Progressive collapse of reinforced concrete structures. Structural Engineering and Mechanics, 2009, vol. 32, No. 6.

24. Rankin, G. I. B., Long, A. E.: Arching strength enhancement in laterally-restrained slab strips. Proc. of Institution of Civil Engineers, Structures and Buildings, 1997, vol 122, pp. 461–467.

25. McDowell, E. L., McKee, K. E., Sevin, E.: Arching action theory of masonry walls. Journal of the Structural Division(ASCE), 1956, vol. 82(ST2), pp. 915-1–915-18.

26. Christiansen, K. P.: The effect of membrane stresses on the ultimate strength of an interior panel in a reinforced concre-te slab. Structural Engineer, 1963, 41, No. 8, pp. 261–265.

27. Hon, A., Taplin, G., Al-Mahaidi, R. S.: Strength of Rein-forced Concrete Bridge Decks Under Compressive Membra-ne Action. ACI Structural Journal, 2005, vol. 102, No. 3, pp. 1–11.

Reza AbbasniaAssociate Professor, Civil Engineering DepartmentIran University of Science and TechnologyTehran 1684613114, IranTel.: +98-9122186903Fax: [email protected]

Foad Mohajeri NavPhD Candidate, Civil Engineering DepartmentIran University of Science and TechnologyTehran 1684613114, IranTel.: +98-9111851943Fax: [email protected]

32 © 2016 Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin · Structural Concrete 17 (2016), No. 1

Technical Paper

DOI: 10.1002/suco.201400111

A detailed investigation was conducted to study the behaviour of reinforced concrete (RC) beams with large openings strength-ened by externally bonded carbon fibre-reinforced polymer (CFRP) laminates. A total of six simply supported beams consist-ing of two solid beams and four beams with openings were cast and tested under four-point bending. Each beam had a cross-section of 120 × 300 mm and length of 2000 mm. Each beam had a large opening placed symmetrically at mid-span. Test parameters included the opening shape and size as well as the strengthening configuration for the CFRP laminates. The study was conducted by way of both experimental testing and finite element analysis. The experimental results show that including a large opening at mid-span reduces the beam capacity to about 50 %. In the experi-mental results, strength gain due to strengthening with CFRP lam-inates was in the range 80–90 %. The finite element and experi-mental results were compared.

Keywords: large opening, strengthening, CFRP, reinforced concrete beam

1 Introduction

Openings are usually found in floors due to stairs, eleva-tors, ducts and pipes. On top of that, openings are fre-quently provided through floor beams to facilitate the passage of utility pipes and service ducts. Such ducts ac-commodate essential services such as conduits, power supplies, water and drainage pipes, ventilation systems, air-conditioning and telecommunication system or even allow access for inspection purposes in beam structures. However, the provision of openings may change the beam’s structural behaviour to a complex one.

In the past, various investigations were conducted to study the behaviour of reinforced concrete (RC) beams with openings. Most of the experiments focused more on the effects due to shear. Large openings provided near to the support where the shear region is predominant have been the subject of many investigations conducted in the past [1-4]. Researchers have also conducted tests on beams with no additional reinforcement provided in the areas above and below the opening [5] and have found

that the beams failed prematurely due to sudden forma-tion of a diagonal crack in the compression chord [6].

Proper design and the provision of adequate rein-forcement are possible if the location of the opening is known. The introduction of a suitable scheme consisting of additional bars near the top and bottom faces of the top and bottom chords as well as short stirrups in both chords was studied [6,7]; failure eventually occurred in a gradual manner. The design of reinforcement around large open-ings proposed by Mansur et al. [4] was in accordance with the ACI code [8].

However, openings might be provided in existing beams due to unexpected circumstances, e.g. changes in services, rehabilitation and other reasons. The drilling of an opening in a RC beam will reduce the cross-sectional area of the beam and this might lead to a reduction in the original structural capacity of the beam and have detri-mental effects for the beam. To reinstate the beam’s origi-nal structural capacity in such a situation, external strengthening is the most suitable method and works by bonding the strengthening material to the concrete sur-face externally. The most common type of external strengthening material is fibre-reinforced polymer (FRP).

Based on the research gaps identified, most of the studies investigated the behaviour of RC beams with openings; however, the studies on strengthening of RC beams with openings are rather limited. This paper focus-es on the strengthening of RC beams with large openings; rectangular and rounded rectangular openings were pro-vided at mid-span of each beam. The methodology of this investigation encompasses both experimental testing and numerical analyses. The results of the study include crack pattern, failure mode and load-deflection relationship.

2 Numerical analysis

In this study, two-dimensional (2D) modelling using a non-linear finite element program, ATENA, was adopted to simulate the behaviour of all beams. Various strength-ening configurations were designed and modelled in AT-ENA based on the crack pattern results obtained from the analysis of RC beams with openings. The most effective strengthening configuration was selected and applied to the experimental programme. The finite element analysis results were compared with the experimental results [9,10].

Siew Choo Chin*Nasir ShafiqMuhd Fadhil Nuruddin


Submitted for review: 27 November 2014; revision: 24 May 2015; accepted for publication: 14 June 2015. Discussion on this paper must be submitted within two months of the print publication. The discussion will then be published in print, along with the authors’ closure, if any, approximately nine months after the print publication.

Behaviour of RC beams with CFRP-strengthened openings

33

S. Choo Chin/N. Shafiq/M. Fadhil Nuruddin · Behaviour of RC beams with CFRP-strengthened openings


sured in the experimental study. These values were then used in the finite element modelling. A Poisson’s ratio of 0.3 was used for steel reinforcement. The bond between steel reinforcement and concrete was assumed to be a perfect bond. A linear elastic orthotropic constitutive rela-tion was assumed for the FRP composites. A rupture point on the stress-strain curve for the fibre direction defines the ultimate stress and strain of the FRP.

2.2 Geometrical model

In this study, concrete was represented by the SBETA ma-terial model in which two-dimensional plane stress ele-ments are used. The tensile behaviour of concrete was modelled by a combination of non-linear fracture mechan-ics and the crack band method in which the smeared crack concept was adopted. The steel reinforcement, stir-rups and carbon fibre-reinforced polymer (CFRP) lami-nates were modelled by a single straight line in a discrete manner by bar reinforcement elements. The bond be-tween steel reinforcement and concrete was assumed to be perfect. A bond-slip relation between CFRP and con-crete was defined and assigned in the properties of the discrete reinforcement.

2.3 CFRP/concrete interface

A bond-slip model between CFRP composite and concrete interface as shown in Fig. 3 was considered in the analysis [9,10]. The model developed by Lu et al. [13] was consid-ered as an accurate bond-slip model that can be incorpo-rated into the finite element (FE) analysis [14-17]. The me-chanical behaviour of the CFRP/concrete interface was modelled as a relationship between the local shear stress τ and the relative displacement s between the CFRP laminate and the concrete. The τ-s relationship is given by [16]:

exp[– / – 1)] if s (2)max 0 0s s sτ = τ α ( ≥

The maximum bond strength τmax and the corresponding slip s0 are governed by the tensile strength of the concrete ft and a width ratio parameter βw as follows:

/ if s (1)max 0 0s s sτ = τ ≤

2.1 Material models

The concrete model used for the numerical investigation was the constitutive model of the ATENA finite element package [11]. In this approach, the elastic constants are derived from a stress-strain function known as the equiva-lent uniaxial law, which covers the complete range of the plane stress behaviour in tension and compression. For the stress-strain relationship of the concrete in compres-sion, the formula recommended by CEB-FIP Model Code 90 [12] was adopted for the ascending branch. The elastic limit of the maximum concrete compressive strength is reached and then followed by a non-linear behaviour until the maximum concrete strength is reached. The softening law in compression is linearly descending. In this case the behaviour of concrete softens until concrete crushing oc-curs. An ascending-descending behaviour was used for the concrete in tension. The slope of the ascending branch is equal to the concrete’s modulus of elasticity. In the de-scending branch of the stress-strain curve, a fictitious crack model based on a crack-opening law and fracture energy was used in which the plane of failure occurs per-pendicular to the principal stress direction. Poisson’s ratio for concrete was assumed to be 0.2. Fig. 1 shows the uni-axial stress-strain law for concrete. The material proper-ties of concrete are given in Table 1.

The steel was represented by a multi-linear law which consists of four lines as shown in Fig. 2. This law allows linear modelling of all four stages of steel behav-iour: elastic state, yield plateau, hardening and fracture. The stress and strain of the steel reinforcement were mea-

Fig. 1. Uniaxial stress-strain diagram for concrete

Table 1. Material properties of concrete

Material type SBeta material

Elastic modulus Ec 32.29 GPa

Poisson’s ratio N 0.2 -

Compressive strength fc 29.75 MPa

Tensile strength ft 2.568 MPa

Type of tension softening exponential

Fracture energy Gf 64.2 N/m

Crack model rotated

Fig. 2. Multi-linear stress-strain law for reinforcement

34



The parameter βw is defined in terms of CFRP laminate width bf and beam width bc as follows:

The area under the τ-s curve indicates the interfacial frac-ture energy Gf, which corresponds to the energy per unit bond area required for complete debonding and is calcu-lated as follows:

The difference in relative displacement between the con-crete and the CFRP laminate represents the slip at the in-terface. The width of CFRP bw used in this study was 80 mm, with a maximum bond stress of 3.50 MPa and a slip of 4.56E-05 m.

3 Experimental procedure

In the experimental programme, a total of six RC beams were tested to failure under four-point bending in order to investigate the beam’s structural behaviour including crack patterns, failure mode and load-deflection behav-

1.5 (3)max w tfτ = β

0.0195 (4)0 w ts f= β

2.25 – /

1.25 /(5)w

b b

b bf c

f fβ =

+

0.308 (6)f w2

tG f= β

iour. The RC beams included two solid beams as reference beams, two beams with a large opening at mid-span with-out strengthening and two CFRP-strengthened beams [9].

3.1 Material characteristics

The concrete used in the experimental study was ready-mixed concrete with a compressive strength of 35 MPa at 28 days; the compressive strength at testing was in the range 36–37 N/mm2. The water/cement ratio was 0.54. The coarse aggregate was 20 mm crushed granite. Fine sand was used as fine aggregate. The longitudinal steel re-inforcement consisted of deformed steel bars with a nomi-nal yield strength of 460 MPa. The web reinforcement was mild steel with a nominal yield strength of 275 MPa. The CFRP laminates were unidirectional with a width of 100 mm and a thickness of 1.4 mm. The nominal tensile strength and tensile modulus of elasticity for the CFRP laminates were 2200 MPa and 170 GPa respectively. The CFRP laminates were attached after the beams were cast. To ensure a suitable surface for bonding, the beam sur-faces were brushed and cleaned before attaching the CFRP laminates with epoxy resin. The thickness of a cured CFRP laminate bonded to the specimen was typi-cally 3 mm.

3.2 Test specimens

A schematic diagram of the test specimen showing the re-inforcement details is depicted in Fig. 4. The test specimen was 2000 mm long with a rectangular cross-section meas-uring 120 × 300 mm. The effective depth to the main rein-forcement was 280 mm and the effective span of the beam was 1800 mm. The tension steel reinforcement consisted of two 12 mm diameter deformed steel bars each having a nominal cross-sectional area A = 113 mm2. The compres-sion steel reinforcement consisted of two 10 mm diameter deformed steel bars with A = 79 mm2 for each bar. The stirrups consisted of 6 mm diameter plain bars with A = 28 mm2 at a spacing of 300 mm centre to centre. Stirrups were not provided in the top and bottom chords at the opening so that strengthening is due entirely to the CFRP laminates.

Large rectangular and rounded rectangular openings were considered in this study. Each opening was 140 mm high × 800 mm long. The ratio of the opening size to the beam’s effective depth was 0.50, so it may be considered as a large opening [7,18]. The beams were cast in a hori-zontal position using plywood formwork. The openings

Fig. 3. Bond-slip model

Fig. 4. Beam geometry and reinforcement details of control beam

Bond stress(MPa)

Slip(mm)

Debonding

initiation

Before

debondτmax

Gf

so smax

ks

1

Complete

debonding

35



were formed by inserting a box fabricated from plywood at mid-span of each beam. The beam specimens are listed in Table 2.

3.3 Test setup

All beams were tested to failure under four-point loading with a static load using a 500 kN universal testing ma-chine (UTM). The test setup is shown in Fig. 5. A spreader beam was used to transfer the load to the test specimen through two loading points 500 mm apart. The beam de-flection was monitored by a number of linear variable displacement transducers (LVDTs) placed on the soffit of

the beam. Crack development and propagation were marked and the mode of failure was recorded.

3.4 CFRP strengthening system

The CFRP laminate strengthening configuration for the experimental beams SBEO and SBRO was based on the strengthening configuration designed in the ATENA finite element program [9]. From several strengthening scheme options, the most effective strengthening configuration was selected and applied to the experimental beams. The strengthening configurations for beams SBEO and SBRO are depicted in Figs. 6 and 7 respectively. As shown in Fig. 6, four longitudinal CFRP laminates 400 mm long with fi-bres parallel to the longitudinal axis of the beam were bonded to the top and bottom chords of the large rounded rectangular opening. Reinforcing with horizontal lami-nates along the top and bottom opening chords resists the tensile force in the bottom chord and the compression force in the top chord. Furthermore, these laminates can control the propagation of cracks in the top and bottom chords.

Fig. 7 illustrates the use of four 400 mm long longitu-dinal CFRP laminates with fibres parallel to the longitudi-nal axis of the beam which were bonded to the top and bottom chords of the large rectangular opening. This strengthening configuration prevents crushing of the con-crete at the corners of the opening due to the high stress concentration. The two 140 mm vertical CFRP laminates with fibres perpendicular to the longitudinal axis of the

Table 2. Test specimens

No. Beam specimenOpening

StrengtheningShape Size (mm) Location

1. CB1 – – – –

2. CB2 – – – –

3 BRO rectangular 140 × 800

at beam mid-span

without strengthening

4 BEO rounded rectangular 140 × 800 (semicircle ∅140) without strengthening

5. SBRO rectangular 140 × 800 CFRP strengthening

6. SBEO rounded rectangular140 × 800(semicircle ∅140)

CFRP strengthening

Fig. 5. Test setup for control beam

Fig. 6. Strengthening configuration for beam SBEO (dims. in mm)

Section A-A

Section A-A

400400

400 400

80

140

80

36



beam inside the rectangular opening were included to help the top and bottom chords resist concrete crushing.

The selected strengthening configurations were found to be supported by Madkour [19], who clarified the non-linear behaviour of strengthened reinforced concrete beams with rectangular web openings. The CFRP lami-nates were modelled in the vertical and horizontal direc-tion with respect to the beam axis. According to Madkour [19], the results from the horizontal strengthening tech-nique are potentially recommended rather than the verti-cal system due to the capability of the horizontal lami-nates to resist both the tensile force in the bottom chord and the compression force in the top chord. Although this system can control local cracks resulting from shear stresses in the opening chords, the efficiency is reduced. The gain in ultimate load capacities from this scheme was about 20 %, compared with only 13 % for the enhance-ment with the vertical strengthening technique.

In contrast, a study of FRP shear strengthening of RC beams using the FE method reported that 90-degree FRP alignment was found to be the most effective configu-ration in terms of strength gain, whereas the 0-degree alignment only helped with controlling the propagation of shear cracks to some extent and not the strength [20].

4 Experimental results and discussion4.1 Behaviour of RC beams with large openings4.1.1 Crack pattern and failure mode

The crack pattern and failure mode of beam BEO are shown in Fig. 8. The large opening at mid-span divided the beam into upper and lower horizontal chords. Owing to

that, early cracks were observed at mid-span in the soffit of the beam which later propagated at the left side of the opening in the lower chord, with the cracks running diago-nally from the bottom to one-third of the beam depth. Once adequate deflection was reached, the solid segments of the beam to the left and right of the opening exerted axial com-pression and tension forces on the upper and lower chords respectively, which caused deflection in the chord mem-bers similar to the Vierendeel action with points of contraf-lexure located at mid-span of each chord (strictly valid when the chord members are symmetrically reinforced). Formation of a mechanism with four hinges in the chord members, one at each corner of the opening, happened dur-ing the final failure of the beam, and this supported the hy-pothesis by Mansur [7]. The beam failed in shear at the up-per chord due to localized shear failure with a diagonal crack near the point load. On the other hand, concrete crushing in the lower chord was seen after the main local shear failure of the upper chord. This happened when the lower chord suffered under flexure and there was an exces-sive stress concentration near the left opening boundary.

The effects of a large rectangular opening at mid-span on failure mode and crack patterns are shown in Fig. 9. Early cracks were found at mid-span in the soffit of the beam which then appeared on the left side of opening in the bottom chord; such cracks were diagonal. The solid parts of the beam to the left and right of the opening ap-plied axial compression and tension forces to the upper and lower chords respectively. Failure of the beam in the top chord due to localized shear failure was traced with a diagonal crack, which penetrated at the opening corner because of the high stress concentration.

Fig. 7. Strengthening configuration for beam SBRO (dims. in mm)

Fig. 8. Crack pattern and failure mode of beam BEO

Section A-A

Section A-A

400

140

400

400 400

80

140

80

37



4.1.2 Load-deflection relationship

A large rounded rectangular opening measuring 140 mm high × 800 mm long was provided in the middle of the beam elevation of specimen BEO. The semicircles at both ends had a diameter of 140 mm. The rounded rectangular opening resulted in a 24 % loss in area and 47 % loss in beam depth, as listed in Table 3. On the other hand, the large 140 mm high × 800 mm long rectangular opening provided at beam mid-span of specimen BRO resulted in a 21 % loss in area and 47 % loss in beam depth.

Table 3 summarizes the results of the load-deflection curves. Beam BEO was cast and compared with control beam CB1 from the same casting batch. The large round-ed rectangular opening in beam BEO reduced the beam capacity by about 39 % compared with that of the original beam capacity shown by beam CB1. As shown in Fig. 10, the yield load Py of beam BEO was obtained as 32 kN at 2.0 mm deflection, whereas the ultimate load Pu was found to be 49 kN at 3.3 mm deflection. The load Py for BEO was less than 50 % of Py for beam CB1. One of the reasons for this is that the 800 mm long opening turned the top and bottom parts of the beam into chords. So when load was applied, it affected the top chord through main local shear failure. Concrete crushing of the lower chord under compression was seen after failure of the top chord. From the load-deflection curve in Fig. 10, it can be seen that beam BEO exhibited a sharp reduction in load after failure. This is due to the stress concentration in the rounded region of the rounded rectangular opening [21].

Beam BRO was cast and then compared with its control beam, CB2, from the same casting batch. The opening resulted in a decrease in beam capacity of about 59 % compared with that of beam CB2. In Fig. 11, the

yield load was obtained as 34 kN at 2.5 mm deflection, whereas the ultimate load of 39 kN was observed at 3 mm deflection. Compared with the yield load Py of beam CB2, the yield load Py of beam BRO was found to be < 59 %. This is due to the similar phenomenon as observed in beam BEO, where the 800 mm long × 140 mm high open-ing separates the beam into two segments consisting of upper and lower chords. Hence, when load was applied through the top chord, the top chord failed before the bot-tom chord due to localized shear failure. Referring to Fig. 11, beam BRO exhibited a sharp decrease in load after the ultimate load was reached. One of the reasons for this is the discontinuity in the beam cross-section due to the opening, where the stress concentration occurs at its cor-

Fig. 9. Crack pattern and failure mode of beam BRO

Table 3. Load-deflection results of control beams and beams with opening

BeamYielding load Py (kN)

∆y (mm)

Ultimate load Pu (kN)

∆u (mm)

OpeningareaBEA

100%

( )×

BCDTBD

100%

( )×

CB1 77 8.2 80 9.1 – –

CB2 86 8.0 96 36.4 – –

BEO 32 2.0 49 3.3 24 47

BRO 34 2.5 39 3.0 21 47

Note: BEA = beam elevated area; BCD = beam cut depth; TBD = total beam depth

Fig. 10. Comparison of load-deflection curves between beams BEO and SBEO and control beam CB1

crushing due to stressconcentration

38



ners. The stress concentration resulted in excessive crack-ing and caused premature failure of the beam [22].

4.2 Behaviour of RC beams with openings strengthened by CFRP laminates

4.2.1 Crack pattern and failure mode

The failure mode and crack pattern of the effects of a sin-gle large rounded rectangular opening at mid-span of beam SBEO are shown in Fig. 12. The initial cracks were observed at mid-span in the soffit of the beam between two longitudinal CFRP laminates. Owing to the presence of CFRP laminates, these cracks were eventually observed on the left and right sides of the CFRP laminates away from mid-span; the cracks were diagonal and penetrated

from bottom to top. When the beam reached adequate deflection, the numbers of early cracks at mid-span and diagonal cracks remote from the CFRP laminates in-creased and later failure occurred due to localized shear failure in the top chord in which a diagonal crack with crushing of the concrete was observed. This caused the longitudinal CFRP laminates to delaminate partially from the concrete surface, whereas vertical cracks were notice-able at mid-span near the right-hand longitudinal CFRP laminates. A possible method for improving and avoiding this type of failure is to attach the CFRP laminates with their fibres perpendicular to the beam axis.

CFRP laminates in this strengthening arrangement directed the early cracks to form at mid-span in the beam soffit between the two longitudinal CFRP laminates and prevented crushing of the concrete in the bottom chord as the beam reached adequate deflection.

The effects of a large rectangular opening at mid-span on the failure mode and crack patterns of beam SBRO are shown in Fig. 13. Early cracks were found at mid-span in the soffit of the beam between two longitudi-nal CFRP laminates. These cracks then propagated on the left and right sides away from the CFRP laminates and penetrated diagonally from the bottom. When the beam reached sufficient deflection, more cracks were observed at mid-span between the two CFRP laminates and diago-nal cracks propagated from the bottom towards the top near the CFRP laminate as shown in Fig. 13. Failure hap-pened due to shear by crushing of the concrete in the top chord together with shearing of the CFRP laminate fibres in the middle. Similarly to beam SBEO, this type of failure could be avoided by placing the CFRP laminates with the direction of their fibres perpendicular to the beam axis.

Fig. 11. Comparison of load-deflection curves between beams BRO and SBRO and control beam CB2

Fig. 12. Crack pattern and failure mode of beam SBEO

Fig. 13. Crack pattern and failure mode of beam SBRO

39



5 Comparison of numerical and experimental results5.1 Unstrengthened beams with openings 5.1.1 Load-deflection behaviour

The accuracy of the FE models when it comes to predict-ing the load-deflection response and crack patterns was examined by comparing their predictions with the experi-mental results. Fig. 14 shows the comparison of load-de-flection curves for the experimental control beam, EXP-CB1, and the beam with an unstrengthened rounded rectangular opening, EXP-BEO, with their finite element results, FE-CB and FE-BEO. From the comparison of the control beams, the trends in the load-deflection curves are very similar. The ultimate loads obtained from EXP-CB1 and FE-CB1 were, respectively, 80 kN at 9.5 mm deflec-tion and 86.7 kN at 7.8 mm deflection. On the other hand, the load-deflection curves for beams FE-BEO and EXP-BEO were almost similar, with the ultimate load attained as, respectively, 49 kN at a deflection of 3.3 mm and 45 kN at a deflection of 2.2 mm. Evidently, there is a good correlation between the numerical and experimental load-deflection curves. The finite element models were able to predict the load capacities of the simulated beams.

The results of experimental and FE analysis in terms of the load-deflection relationship of the beam with rec-tangular opening, EXP-BRO and FE-BRO, were compared with their respective control beams, EXP-CB2 and FE-CB, as shown in Fig. 15. A similar trend in the load-deflection

The CFRP laminates in this configuration managed to control cracks at mid-span and on the left and right sides away from the CFRP laminate in the bottom chord. Although crushing of the concrete occurred in the top chord, which caused shearing of the CFRP laminate fi-bres, the bond between CFRP laminate and concrete sur-face was still intact.

4.2.2 Load-deflection relationship

Table 4 summarizes the load-deflection curve results of strengthened beams with an opening. The CFRP laminate configuration around the large rounded rectangular open-ing, which covered about 20 % of the beam’s exposed ar-ea, managed to restore 96 % of the capacity of its corre-sponding control beam, CB1. In Fig. 10, the yield load of beam SBEO was 62 kN at 3.7 mm deflection, whereas the ultimate load of 77 kN was observed at 5.3 mm deflection. A sharp decrease in load was observed after the ultimate load was achieved, as illustrated in Fig. 14. This indicates that the beam failed by crushing of the concrete due to shear failure [23].

Referring to Table 4, CFRP laminates covering ap-prox. 22 % of the exposed area around the opening in beam SBRO caused the beam to regain 86 % of the beam capacity of CB2. As shown in Fig. 11, the yield strength was obtained as 72 kN at 6.6 mm deflection, whereas the ultimate load was found to be 83 kN at 10.4 mm deflec-tion. After the ultimate load was attained, the load de-creased sharply, thus indicating a brittle failure due to high stress concentrations at the corners of the opening which later failed due to crushing of the concrete.

In general, strengthening of beams with a large open-ing in flexure using CFRP laminates could regain and re-store approx. 80–90% capacity of their respective control beams depending on the strengthening configurations. Based on the load-deflection results of each type of beam, it was found that CFRP laminates bonded around the opening significantly increase the beam capacity to almost the same as that of the control beams. The increase in ca-pacity in these beams was caused by the presence of CFRP laminates that interrupt the natural path of crack propaga-tion, hence requiring greater energy to redirect the path of cracks through the unreinforced region. Thus, strengthen-ing the opening region enhances the deflection behaviour and controls cracks around the opening.

Table 4. Load-deflection results of control beams and strengthened beams with opening

BeamYielding load Py (kN)

∆y(mm)

Ultimate loadPu (kN)

∆u(mm)

BEA (mm2)

TotalCFRP Area

BEA

100%

( )×

CB1 77 8.2 80 9.1 – –

CB2 86 8.0 96 36.4 – –

SBEO 62 3.7 77 5.3 1 296 000 20

SBRO 72 6.6 83 10.4 1 296 000 22

Note: BEA = beam exposed area

Fig. 14. Comparison of load-deflection curves between experimental and FE results of control beams and beams with rounded rectangular opening

40



curves was obtained when comparing the experimental and FE control beams. It was seen that the ductility in control beam EXP-CB2 is quite high compared to FE-CB. This may be due to inaccuracy during beam preparation and testing. The ultimate load and deflection of beams EXP-CB2 and FE-CB was obtained as 96 kN at 31 mm and 86.7 kN at 7.8 mm respectively, which shows a slight difference in ultimate load of about 9 %. A comparison of the load-deflection behaviour of beams EXP-BRO and FE-BRO was found to be similar, with both attaining an ulti-mate load of about 50 kN with a variation in deflection of about 20 %.

5.1.2 Crack pattern

A comparison of the FE analysis and experimentally ob-served crack patterns of the beam with large rounded rec-tangular opening is shown in Fig. 16. Similarly to that ob-served in the experimental beam, the numerical results of beam SBEO exhibited vertically aligned flexural cracks in the bottom chord from the edge of the beam which were distributed along the length of the opening. Diagonally oriented shear cracks were noticed at the point of curve

formation to the beam support where the number of cracks increased. On the other hand, diagonal cracks were seen to propagate from the points of the curve to-wards the loading points at the top chord, which eventu-ally led to localized shear failure, as observed in the ex-perimental beam. The crack pattern of the FE analysis matches well with that of the observed beam.

5.2 Beams strengthened with CFRP laminates5.2.1 Load-deflection behaviour

The load-deflection curves of the strengthened beams with rounded rectangular opening for both experimental and FE analysis are plotted in Fig. 17. The comparison re-vealed dissimilarity in the curve trend between EXP-SBEO and FE-SBEO. It was found that the ultimate load attained in FE-SBEO was lower than that of the experi-mental beam, EXP-SBEO: 64 kN at 1.5 mm deflection and 77 kN at 5.3 mm deflection respectively.

Fig. 18 shows the comparison of load-deflection curves for strengthened beams with rectangular openings for both experimental and FE results. From the curve trend, a similar case was observed as for beam FE-SBEO, with the load-deflection curve trend of FE-SBRO lower than experimental beam EXP-SBRO by about 30 %. This might be due to the FE models of the strengthened beams which incorporated the interfacial bond-slip action be-tween the concrete and CFRP which failed by debonding once the FE simulation terminated due to a divergence.

5.2.2 Crack pattern

Similarly, the crack patterns of a beam in flexure with a large rectangular opening, SBRO, as shown in Fig. 19, were found to be similar to the beam with rounded rectan-gular opening. The vertical cracks from the numerical analysis were initiated from the bottom edge of the beam along the length of the opening in the mid-span. Diago-nally oriented cracks were found at the opening corners which propagated to the beam support. FE analysis also captured diagonal and horizontal cracks at both corners of the opening at the top chord, which resulted in local-

Fig. 15. Comparison of load-deflection curves between experimental and FE results of control beams and beams with rectangular opening

Fig. 16. Comparison of EXP and FEM crack pattern results for beam SBEO

41



ized shear failure at the right corner of the opening, as traced in the experimental beam.

Table 5 summarizes and compares the ultimate loads from the experimental work and predicted loads from the FE analysis. It is significant that the ratio obtained for con-

trol beams and unstrengthened beams BEO and SBEO was in the range of 0.9–1.1, which indicates that the results are comparable between the FE and experimental results. Con-trary to this, the FE-to-experiment ultimate load ratio of beams SBEO and SBRO was 0.83 and 0.69, respectively. The calculated average ratio and standard deviation of FE-to-experiment are 0.76 and 0.1, respectively. This signifies that the correlation is not in good agreement.

6 Conclusions

Based on the results obtained, the following conclusions can be drawn:1. It was found that large openings at beam mid-span,

both rectangular and rounded rectangular, suffered about 50 % loss in capacity because such openings went beyond the loading points.

2. FE analysis was used to strengthen the openings with the selected strengthening configurations, showing that 80–90 % of the lost capacity could be restored. The strengthening options from the FE analysis were veri-fied using experimental testing and almost similar re-sults were obtained. From this conclusion, openings with lengths of about 30–40 % of the effective span are not advisable until and unless the upper and lower chords are properly designed.

3. In terms of suitable opening shapes and sizes, the study showed, overall, that the opening with sharp edges and corners caused stress concentrations and also disturbed the load/flow pattern within the beam. Openings with cracks at rounded edges, such as the rounded rectangu-lar opening, have proved to provide higher capacity for a similar load/flow pattern. Hence, such shapes are preferable.

4. The load-deflection behaviour of FE analysis and ex-perimental results for both unstrengthened beams with openings were found to be comparable, which confirms the validity of the FE models developed and the reliabil-ity of the FE simulation. However, the agreement be-tween experimental and FE results for strengthened beams with openings did not achieve a good correla-

Fig. 17. Comparison of load–deflection curves between experimental and FE results of control beams and strengthened beams with rounded rectan-gular opening

Fig. 18. Comparison of load–deflection curves between experimental and FE results of control beams and beams with strengthened rectangular opening

Fig. 19. Comparison of EXP and FEM crack pattern results for beam SBRO

42



tion. This may be due to the incorporation of interfacial bond-slip action between the concrete and the CFRP in the FE analysis which failed by debonding once the FE simulation terminated due to a divergence.

7 Recommendations

The following are the main recommendations for further work:1. To study the dynamic behaviour of RC beams due to the

effects of large openings (single or multiple) of various shapes and sizes subjected to critical shear, bending, torsion and combined loading in simply supported beams, T-beams, continuous beams and deep beams.

2. To investigate the effects of strengthening configura-tions using CFRP laminates on the dynamic behaviour of beams.

3. To study the failure modes of CFRP laminates around openings of various shapes, sizes and locations by way of both numerical analysis and experimental investigation.

4. To simulate/validate the beams using 3D FE analyses.

Acknowledgments

The authors wish to express their gratitude and sincere ap-preciation to Universiti Teknologi PETRONAS (UTP) for financing this research work. The authors would also like to thank the laboratory technologists of the Concrete and Structural Laboratory, Civil Engineering Department, UTP for their assistance throughout the experimental work. Last but not least, the authors would like to thank Dr. Su Kong Ngien for his contribution in proofreading this paper.

References

1. K.W. Nasser, A. Acavalos, and H.R. Daniel, “Behavior and Design of Large Openings in Reinforced Concrete Beams,” ACI Journal, Proceedings, vol. V.64, 1967, pp. 25–33.

2. G.B. Barney, J.M. Corley, W.G., Hanson, and R.A. Parmelee, “Behaviour and Design of Prestressed Concrete Beams with Large Web Openings,” PCI Journal, vol. 22, 1977, pp. 32–61.

3. H.S. Ragan and J. Warwaruk, “Tee Members with Large Web Openings,” Journal of the Prestressed Concrete Institu-te, vol. 12, 1967, pp. 52–65.

4. M.A. Mansur, K.H. Tan, and S.L. Lee, “Design Method for Reinforced Concrete Beams with Large Openings,” ACI Journal, vol. 82, 1985, pp. 517–524.

5. W.B. Siao and S.F. Yap, “Ultimate Behaviour of Un-Strengthened Large Openings Made in Existing Concrete Beams,” Journal of the Institution of Engineers, Singapore, vol. 30, 1990, pp. 51–57.

6. M.A. Mansur, “Effect of Openings on the Behaviour and Strength of R/C Beams in Shear,” Cement and Concrete Composites, vol. 20, 1998, pp. 477–486.

7. Mansur M.A., “Design of Reinforced Concrete Beams with Web Openings,” Proceedings of the 6th Asia-Pacific Struc-tural Engineering and Construction Conference (ASPEC 2006), 5-6 September 2006, Kuala Lumpur, Malaysia., 2006, pp. 104–120.

8. ACI Committee 318, “Building Code Requirements for Rein-forced Concrete (ACI 318-95) and Commentary (ACI 318R-95),” American Concrete Institute,, 1995, p. 369.

9. S.C. Chin, “Strengthening of Reinforced Concrete Beams with Openings Using CFRP Laminates,” PhD Thesis, Uni-versiti Tekonologi Petronas, Malaysia. 2013.

10. S.C. Chin, N. Shafiq, and M.F. Nuruddin, “Strengthening of RC Beams with Large Openings in Shear by CFRP Lamina-tes: Experiment and 2D Nonlinear Finite Element Analysis,” Research Journal of Applied Sciences, Engineering and Technology, vol. 4, 2012, pp. 1172–1180.

11. V. Cervenka, L. Jendele, and J. Cervenka, “ATENA Program Documentation,” Part 1 Theory, 2010, pp. 1–248.

12. CEB-FIP Model Code 1990, Design Code, Thomas Telford, 1993.

13. J.G. Lu, X.Z., Teng, L.P. Ye, and J.J. Jiang, “Bond – Slip Mo-dels for FRP Sheets / Plates Bonded to Concrete,” Enginee-ring Structures, vol. 27, 2005, pp. 920–937.

14. A. Godat, K.W. Neale, and P. Labossiere, “Towards Model-ling FRP Shear-Strengthened Reinforced Concrete Beams,” FRPRCS-8, 2007, pp. 1–10.

15. Y.T. Obaidat, S. Heyden, and O. Dahlblom, “The effect of CFRP and CFRP/concrete interface models when modelling retrofitted RC beams with FEM,” Composite Structures, vol. 92, May. 2010, pp. 1391–1398.

16. H.A. Kotynia, R., Baky, K.W. Neale, and U.A. Ebead, “Flexu-ral Strengthening of RC Beams with Externally Bonded CFRP Systems : Test Results and 3D Nonlinear FE Analysis,” Jour-nal of Composites for Construction, 2008, pp. 190–201.

17. A. Godat, P. Labossière, and K.W. Neale, “Numerical Inve-stigation of the Parameters Influencing the Behaviour of FRP Shear-Strengthened Beams,” Construction and Buil-ding Materials, vol. 32, 2012, pp. 90–98.

Table 5. Comparison of EXP and FEM results

EXP FEM

Ult. FEM/Ult. EXPBeam

Yielding loadPy (kN)

∆y(mm)


∆u(mm)

Yielding loadPy (kN)

∆y(mm)


∆u(mm)

CB1 77 8.2 80 9.1 77 3.9 87 7.8 1.09

CB2 86 8.0 96 36.4 77 3.9 87 7.8 0.91

BEO 32 2.0 49 3.3 26 0.8 45 2.2 0.92

BRO 34 2.5 39 3.0 23 0.9 40 2.3 1.03

SBEO 62 3.7 77 5.3 36 0.7 64 1.5 0.83

SBRO 72 6.6 83 10.4 38 0.8 57 1.8 0.69

43



18. A. Pimanmas, “Strengthening R/C Beams with Opening by Externally Installed FRP Rods: Behavior and Analysis,” Composite Structures, vol. 92, Jul. 2010, pp. 1957–1976.

19. H. Madkour, “Non-linear Analysis of Strengthened RC Be-ams with Web Openings,” Proc. Inst. Civ. Eng. Struct. Build., vol. 162, no. 2, pp. 115–128, 2009.

20. I. Elyasian, N. Abdoli, And H. R. Ronagh, “Evaluation Of Parameters Effective In Frp Shear Strengthening Of Rc Be-ams Using FE Method,” Civ. Eng., vol. 7, no. 3, pp. 249–257, 2006.

21. J.D. Hsu, J.W. Michael, and J.R. Fisk, AAOS Atlas of Ortho-ses and Assistive Devices, Mosby Elsevier, 2008.

22. M.A. Mansur, “Combined Bending and Torsion in Rein-forced Concrete Beams with Rectangular Openings,” Con-crete International, 1983, pp. 51–58.

23. P. Balaguru, A. Balaguru, P.N., Nanni, and J. Giancaspro, FRP Composites for Reinforced and Presstressed Concrete Structures: a guide to fundamentals and design for repair and retrofit, Taylor and Francis, 2008.

Siew Choo Chin, Senior Lecturer, Universiti Malaysia Pahang Faculty of Civil Engineering & Earth Resources, Lebuhraya Tun Razak, 26300 Gambang, Kuantan, Pahang, Malaysia. Tel:+6095492933 Email: [email protected], [email protected]

Nasir Shafiq, Professor, Universiti Teknologi PETRONAS, Department of Civil Engineering, Faculty of Engineering, Bandar Seri Iskandar, 31750 Tronoh, Perak, Malaysia [email protected]

Muhd Fadhil Nuruddin, Professor, Universiti Teknologi PETRONAS, Department of Civil Engineering, Faculty of Engineering, Bandar Seri Iskandar, 31750 Tronoh, Perak, Malaysia [email protected]


Technical Paper

To increase the efficiency of new structures and perform safety evaluations of existing structures, it is necessary to model and analyse the non-linear behaviour of reinforced concrete. The applicability of the safety formats in present design codes is un-clear for indeterminate structures subjected to loading in several directions. The safety formats in fib Model Code 2010 have been evaluated for a reinforced concrete frame subjected to vertical and horizontal loading and the influence of load history studied. Basic reliability methods were used together with response sur-faces to assess the failure probabilities and one safety format did not meet the intended safety level. The results indicate the importance of load history and it is concluded that more research is required regarding how load history influences the safety level of complex structures.

Keywords: reinforced concrete frame, non-linear finite element analysis, safety formats, response surface, first-order reliability method, fib Model Code 2010

1 Introduction

Large concrete structures exposed to a variety of loads acting in different directions, e.g. dams and offshore con-crete structures, must be designed to meet requirements regarding structural capacity and reliability for their de-sign life. The design must also be feasible with respect to constructability. In engineering practice, the design of large concrete structures is verified using global linear finite element analyses (LFEA) [1]. Large solid elements are used in the analyses to reduce the number of elements and, consequently, the computation time. Owing to the nature of linear analysis, load effects caused by various loads can be superimposed. Loads are typically combined using post–processing software, which also designs the reinforcement in the structure.

In reality, reinforced concrete structures show non-linear material behaviour when subjected to increasing

load. Cracking of concrete and yielding of reinforcement change the stiffness properties of the material and stress redistributions can occur within the structure [2]. This behaviour is not represented by an LFEA.

An analysis that recognizes the non-linear behaviour of reinforced concrete, i.e. a non-linear finite element analysis (NLFEA), is required in order to capture the stress distribution accurately for the load level studied. Efforts are being directed towards finding a general, ro-bust and stable solution strategy for NLFEA, comprising choices regarding force equilibrium, kinematic compat-ibility and material model [3], [4].

The traditional design process using LFEA involves controlling the utilization ratio (UR) locally in each design section to ensure safety, with the material strength being scaled down and the load effect scaled up using appropri-ate partial factors to obtain design values. However, a global NLFEA reveals the global capacity of the structure – to which all sections contribute – and the safety should be verified accordingly [2]. The safety formats available in the fib Model Code for Concrete Structures 2010 [5] have been applied to a reinforced concrete frame subjected to both vertical and horizontal loading using two different load histories. Based on the results from these analyses, the safety formats are checked to see if they lead to their intended safety level. It will be shown that subjective choices have to be made regarding the measure of struc-tural resistance and that the load history has a significant effect on the results.

2 Reliability assessment of structures 2.1 General

The behaviour and safety of a structure can be described in terms of physical properties such as material strengths, dimensions, loads, weight, etc. These properties are called basic variables. They are random in nature and described using probability distributions, which are fitted based on observations and measurements [6], [7].

The probability of failure is a central part of a reli-ability assessment. For a structure with a certain resist-ance R subjected to a certain load effect S, which are both independent random variables, the probability of failure can be written as

( ) ( 0) (1)p P R S P R Sf = ≤ = − ≤

Evaluation of safety formats for non-linear finite element analyses of statically indeterminate concrete structures subjected to different load paths

Mattias Blomfors*Morten EngenMario Plos

DOI: 10.1002/suco.201500059


Submitted for review: 28 April 2015; revision: 31 July 2015; accepted for publication: 05 September 2015. Discussion on this paper must be submitted within two months of the print publication. The discussion will then be published in print, along with the authors’ closure, if any, approximately nine months after the print publication.

45

M. Blomfors/M. Engen/M. Plos · Evaluation of safety formats for non-linear FEA


where:pf probability of failureR resistanceS load effect

or more generally as

where:G limit state functionfRS(r,s) bivariate (joint) probability density function of

resistance and load effectD failure domain

For an actual structure, it is complicated to evaluate the above expression since many basic variables are includ-ed in both the resistance and the load terms.

Furthermore, uncertainties originating from all im-portant sources must be evaluated and included in the model of basic variables. The uncertainties can be char-acterized as follows [6]:– Inherent physical or mechanical uncertainty, e.g. ma-

terial and geometry uncertainties– Statistical uncertainty, e.g. when few observations

form the basis for design decisions– Model uncertainties

Material and geometry uncertainties can be taken into account by modelling the material and geometrical prop-erties as random variables using suitable distributions. All types of uncertainty where statistical estimators, e.g. sample mean and standard deviation, are used to find suitable probability density functions are also subjected to statistical uncertainty. This is because the observations do not represent the actual variable perfectly.

Modelling uncertainty is an important part when describing that a predicted solution obtained via a com-putational model will generally differ from the true solu-tion. This might be due to lack of knowledge or simplify-ing assumptions. Some common simplifying assumptions concern, for example, idealizations of the structure, mate-rial behaviour and load application. The modelling un-certainty can be evaluated by comparing analysis results with experiments or observations. However, it was found by Schlune et. al. [8] that model uncertainty is hard to assess for difficult-to-model failure modes. Ideally, when the model uncertainty parameters are implemented in the computational model, the model should, on average, predict the solutions correctly. It should be noted that many models in the design codes are intended to yield conservative results, i.e. they have an intentional bias that should be considered when comparing results obtained using different methods.

The generalized reliability index is typically used as a reliability measure. The definition corresponds to a cer-tain probability of failure based on a standardized normal distribution. It can be described as the number of stand-ard deviations away from the mean value corresponding to a studied probability of failure. It is written as [6]

( , ) 0 ( , ) (2)p P G R S f r s drdsf RSD∫∫= ≤ =

( ) (3)1 pfβ = − Φ−

where pf is the probability of failure and Φ–1 the inverse Gaussian distribution. The probability of failure should be calculated from the standardized joint distribution of the basic variables, including modelling and statistical uncertainty.

Neither analytical nor conventional numerical in-tegration techniques are generally feasible for evaluating the failure probability inherent in practical applications [7]. Other methods have been developed for integration in many dimensions, which also accommodates any type of distribution for the random input variables. These numeri-cal solution techniques are commonly known as Monte Carlo simulations.

In many practical applications, such as finite ele-ment analysis, the limit state is given implicitly. In other words, the limit state surface is not explicitly described by a set of equations, instead the boundary between the safe and failure domains must be revealed by several numeri-cal analyses with different input values. A suitable func-tion can be fitted to a set of results obtained from analyses performed with different deterministic input values for the basic variables. The closed form function fitted to the im-plicit limit state function, cf. G in Eq. (2), is differentiable, now represents a tractable approximation of the failure surface and is known as a response surface. A procedure for constructing a response surface is presented in [9].

Response surfaces enable reliability assessment using the first/second–order reliability method (FORM/SORM) based on results from FEA. In these methods, the limit state function G(X) = 0 is transformed into standardized normal space and the basic variables are transformed into standard normal variables. The safety index is now found as the distance between the origin and the closest point on the limit state function, known as the design point [7]. The difference between FORM and SORM lies in the degree of approximation of the limit state function around the design point, where a first- and second-order polyno-mial is used respectively.

2.2 Safety formats for non-linear analysis according to fib Model Code 2010

In order to verify the safety of a structure, it is necessary to check that the structural resistance is greater than or equal to the load effect. As stated previously, traditional checks at section level have limited applicability when it comes to non-linear analysis since the global behaviour of the structure is analysed and significant stress redistribu-tion can occur. For NLFEA the most probable, i.e. mean, resistance of the structure should typically be used as a reference for the safety evaluation. The purpose of the safety formats is to account for the uncertainties in the basic variables and yield a design resistance in agreement with the chosen safety level [5].

The general design condition used is as follows:

where:Fd design value of actionsRd design resistance

, (4)*

F R RR

d d dm

R Rdγ γ≤ =



Rm mean resistanceγ*

R global safety factorγRd model uncertainty factor

There are several safety formats described in fib Model Code 2010 which can be used to evaluate Rd. The formats studied in the present paper are the two global resist-ance factor methods and the partial safety factor method, which are presented below.

2.2.1 Global resistance factor method (GRFm)

In the GRFm modified mean material properties are used for calculating the resistance. The design resistance is formulated as

where function r represents the resistance obtained from non-linear analysis with modified mean input material parameters. The safety factors for resistance and model uncertainty, γR and γRd, are given here as 1.2 and 1.06 respectively. The mean steel reinforcement properties can be approximated from the characteristic values as

The modified mean concrete properties to be used in the analysis can be calculated as

where fcf is the concrete strength parameter. The reduc-tion originates from the common global safety factor, here used for both steel and concrete. Since the steel and concrete partial factors are formulated for the same failure probability but the concrete has a larger variability, con-crete requires a larger safety margin. This is adjusted for in the expression above. Furthermore, 0.85 ≤ αcf ≤ 1.0 is a fac-tor accounting for sustained load and unfavourable load application effects, and fcfk is the characteristic concrete property.

2.2.2 Method of estimating a coefficient of variance of resistance (ECOV)

This safety format is based on research suggesting that the randomness of the resistance can be described as a lognor-mal distribution with two parameters: the mean resistance and the variance of the resistance. Two NLFEAs are con-ducted, one performed with mean input parameters for the material data and one with characteristic values. An underlying assumption is that an analysis performed us-ing mean material parameters yields the mean resistance, and likewise for the characteristic parameters. Using these resistances it is possible to estimate the coefficient of vari-ance of resistance based on the assumed distribution. This can be expressed in mathematical terms as

( , ...)(5)R

r fd

m

R Rdγ γ=

1.1 (6)f fym yk=

1.1 (7)f fcf cf cfks

c

αγγ

=

( , ...), ( , ...) (8)R r f R r fm m k k= =

where r is the resistance from non-linear analyses, with mean and characteristic values as material input param-eters respectively, and fm here denotes the actual mean input parameters, in contrast to the adjusted mean input parameters in Eq. (5). Describing the resistance as a two-parameter lognormal distribution enables the coefficient of variation of resistance to be determined as:

where VR is the coefficient of variation of resistance for a characteristic value with 5 % probability. The factor for global resistance γR can be calculated as follows:

where αR is a weight factor for the reliability index β , tak-ing into account that only the resistance side of the design condition is treated. The design value for the resistance can be calculated as

where 1.06 is suggested for the model uncertainty γRd for numerical models with well-documented validity. A higher value should be used for the model uncertainty if this requirement is not fulfilled.

2.2.3 Partial factor method

According to the partial factor method, a non-linear analysis is performed using the deterministic design val-ues derived from the random material parameters. Using significantly reduced material parameters, this approach might not capture the global safety and might not describe the structural response of the structure very well. It can also cause the structure to fail in a distorted mode, not re-flecting the actual behaviour [2]. Nevertheless, case stud-ies suggest that the partial factor method can be used as a rough but safe estimate. In mathematical terms, the design resistance is obtained using

where r represents the non-linear analysis for the design value fd of the input parameters. It should be noted that the model uncertainty factor γRd is not included in the above expression. This is because it is typically included in the partial safety factor γM for the materials and, conse-quently, in the design values.

3 Evaluation of safety formats in fib Model Code 2010

In order to assess the safety level of the safety formats in fib Model Code 2010 [5], they were applied to a re-inforced concrete frame subjected to loading in two directions. Following the approach suggested by Engen et. al. [3], a frame was designed for the stresses obtained by LFEA using Multiconsult’s in-house post processor MultiCon [1], which accounts for non-linear section resist-

11.65

ln( ) (9)VR

RRm

k

=

exp( ) (10)VR R Rγ α β=

(11)RR

dm

Rd Rγ γ=

( , ...) (12)R r fd d=

47



ance. The design was conducted in compliance with the provisions given in Eurocode 2 [10] using concrete strength classes C45/55 and steel reinforcement grade B500. Fig. 1 shows the main frame dimensions and point loads with directions.

The frame was subjected to both vertical and hori-zontal loading, considered as permanent and variable loads respectively. Two load sequences were used: one de-noted the main load history, where the vertical load was applied first followed by the horizontal load, and another, reverse case, denoted the inverse load history. The loads were applied in 20 increments, each applying 5 % of the characteristic load level. Once both characteristic loads had been applied, they were scaled up simultaneously according to their load factors until failure. The load ap-plication is presented in Fig. 2.

3.1 Solution strategy for NLFEA

The DIANA v. 9.4.4 FE software [11] was used for the analy-ses. The analyses were force-controlled and benchmarked with good agreement against a portal frame experiment conducted by Kotsovos and Pavlovic [12]. The solution

strategy, consisting of choices regarding force equilibrium, kinematic compatibility and material modelling, was se-lected based on the guidelines for NLFEA of Concrete Structures presented in [13] and a comparative study per-formed by the second author [4]. A fixed crack model was used for the concrete with exponential tension softening and parabolic behaviour for the crushing of the concrete. The lateral influence of cracking and the influence of lateral confinement were accounted for according to [14] and [15]. The aforementioned guidelines [13] propose the use of a variable shear retention factor (SRF). How-ever, in the comparative study [4], where NLFEAs were performed with large finite elements, the use of a varia-ble SRF resulted in excessively low estimated capacities. Therefore, a constant SRF with a relatively low value of 0.1 was used for the present study. Poisson’s ratio was re-duced depending on the crack opening and the reinforce-ment was modelled with tensile hardening after yielding.

Furthermore, 20-node isoparametric solid brick ele-ments with quadratic interpolation were used together with standard embedded truss elements for the reinforce-ment. The elements were relatively large (three elements over the thickness), full integration was used and geomet-ric non-linearities were included. The loads were applied in relatively large increments in order to be feasible in an engineering setting. The convergence criteria were set to either a relative energy variation of 0.1 % or a relative out-of-balance force of 1 %, whichever occurred first. Deter-mining the ultimate capacity also involved checking the failure mode and concrete and steel stresses. Furthermore, all material parameters for the analyses were estimated using relevant expressions from fib Model Code 2010 [5] and NS 3576–3 [16].

3.2 Response surface

A second-order polynomial response surface was fitted to the analysis results, essentially following the procedure suggested in [17]. However, the procedure was altered to suit the coarseness of the present case. The limit state function G(X) was chosen as the total load-carrying ca-pacity from NLFEA minus the design resistance obtained from the different safety formats:

where R(X) is the resistance obtained from NLFEA as a function of the basic variable X and RSF is the design re-sistance for the safety format studied. The measure of the structural resistances is chosen as the sum of the vertical and horizontal load. The basic variables included are the com-pressive strength of the concrete and the yield strength of the steel reinforcement, both assumed to follow lognormal distributions. The coefficients of variation for the concrete and reinforcement material parameters were taken as 0.20 and 0.10 respectively from [8], [18]. Response surfaces were constructed for the main and the inverse load history for each of the design loads according to the safety formats to-gether with the LFEA design load – a total of eight surfaces. The FORM and SORM were applied to the eight polynomial limit state functions using a MATLAB script [19], yielding the probabilities of failure obtained by the safety formats.

( ) ( ) (13)XX XXG R RSF= −

Fig. 1. Main dimensions of frame and applied point loads

Fig. 2. Loading sequences: the loads are first applied one by one to their characteristic levels and then scaled up simultaneously according to their load factors.



4 Results

The crack pattern for the frame loaded according to the main load history analysed using mean material param-eters is plotted on the deformed shape in Fig. 3. The loads are applied up to their characteristic levels.

The deformed shape of the main and the inverse load histories is similar, whereas the crack pattern differs, see Fig. 4. Note the difference between the frames in the

upper left corner, which originates from the change of load history.

The design resistances of the structure corresponding to an intended probability of failure of 0.1 % are presented for the three safety formats and two different load histories. The frame was designed – based on linear elastic load effects and non-linear sectional capacity – for a total load of 2093 kN. The design resistances obtained with the three safety formats in fib Model Code 2010 are presented in Table 1.

Fig. 3. Deformed shape and crack pattern for the frame loaded with both loads to their characteristic level according to main load history

Fig. 4. Deformed shape and crack pattern for the frame loaded with both loads to their characteristic level according to inverse load history

Table 1. Design resistances calculated according to the safety formats applied to main and inverse load histories

Rm [kN] Rk [kN] VR γR γRd Rd [kN]

Main load history ECOV 2864 2614 0.055 1.18 1.06 2284GRFm 2492 – – 1.2 1.06 1959PSFm – – – – – 1967

Inverse load history ECOV 2923 2402 0.119 1.437 1.06 1920GRFm 2491 – – 1.2 1.06 1959PSFm – – – – – 1939

49



pacity between the analyses using mean and characteristic material strength parameters leads to a low coefficient of variation, yielding a low factor of safety and a high design load.

The response surface for the main load history us-ing the ECOV safety format is shown together with data points in Fig. 5. The difference between the safety formats is simply a translation of the response surface in the G-direction, i.e. a vertical shift.

Contour plots of the response surfaces for the ECOV safety format applied to the main and inverse load histo-ries, together with the FORM design points, are presented in Figs. 6 and 7 respectively.

The resulting safety indexes and corresponding failure probabilities obtained by FORM and SORM for the safety formats are presented in Table 2. The safety for the load used in the LFEA-based frame design is also presented for the two load histories. The intended safety index for the safety formats is β = 0.8 × 3.8 = 3.04, corresponding to a probability of failure of 10–3. The ECOV safety format failed to reach the intended safety level for the main load history.

5 Discussion

The design resistance obtained using ECOV did not reach the intended safety index when the vertical load was applied first followed by the horizontal load, and the calculated design resistance was found to be significantly higher with this safety format. The small difference in ca-

Fig. 5. Response surface for main load history with some data points visible as dots

Fig. 6. Contour plot of response surface for main load history obtained us-ing the ECOV safety format, with FORM design point indicated by a diamond

Fig. 7. Contour plot of response surface for inverse load history obtained us-ing the ECOV safety format, with FORM design point indicated by a diamond

Table 2. Safety indexes and probabilities of failure for main and inverse load histories based on the safety formats. The safety of the original design loads is also given.

βFORM (pf) βSORM (pf)

Main load history ECOV 2.47 (6.71 × 10–3) 2.45 (7.09 × 10–3)GRFm 4.06 (2.45 × 10–5) 4.06 (2.48 × 10–5)PSF 4.02 (2.92 × 10–5) 4.02 (2.97 × 10–5)LFEA 3.39 (3.50 × 10–4) 3.38 (3.64 × 10–4)

Inverse load history ECOV 3.72 (1.01 × 10–4) 3.67 (1.24 × 10–4)GRFm 3.60 (1.61 × 10–4) 3.53 (2.04 × 10–4)PSF 3.66 (1.27 × 10–4) 3.60 (1.58 × 10–4)LFEA 3.14 (8.51 × 10–4) 3.02 (1.27 × 10–3)



The ultimate capacity determined from the analyses depends on the convergence criteria used for the equilib-rium iterations. A relatively strict convergence criterion was used and the ultimate capacity of the frame was de-termined as the resistance for the last converged load step, without regard to possible close-to-converged, sub-sequent steps before divergence. It could be argued that the structure has not failed until divergence occurs in the iterations, since it is in fact close to equilibrium even if the out-of-balance energy and forces exceed the convergence criterion. However the unconverged load steps may not be realistic with respect to structural response at section level and could include spurious deformations etc., even if global equilibrium is almost fulfilled. It remains unclear as to whether global equilibrium could be fulfilled with a physically feasible response at section level; hence, these load increments are excluded.

Limited information is provided in fib Model Code 2010 [5] regarding how the structural resistance should be defined when used in the safety formats. It depends on the current loading situation of the structure, but the resistance is also affected by the previous load history. Even with the simple load situation presented in this paper, deciding on the measure of structural resistance to be used for the safety formats is not a trivial matter. If cyclic loading or time-de-pendent aspects are included in the analyses, the definition of resistance becomes even more complex.

The order in which the loads were applied was found to be important. In general, the load-carrying capac-ity was lower for the inverse load history compared with the main load history. For the inverse load history, apply-ing the horizontal load first causes a horizontal deforma-tion. When the vertical load is then applied, it is carried in a way that further increases the concrete stress on the inside of the upper right corner of the frame. The con-crete strength may therefore be expected to influence the resistance to a greater extent for the inverse load history, which can also be observed when comparing the contour lines in Figs. 6 and 7. This observation explains the find-ing presented in Table 1, i.e. that between the main and inverse load history, Rm increases while Rk decreases for the ECOV analyses. This demonstrates the importance of considering the load history in non-linear analyses. The load history should preferably include all the loads that the structure has been subjected to during its life, includ-ing the construction phase. Since the aim is to increase the use of non-linear analyses in engineering applications, there should be more descriptive guidelines. These should treat aspects such as how to define the design resistance as well as acceptable simplifications regarding structural model and loading history.

The response surface was fitted using a second-order polynomial in the safety assessment. The failure mode was compression failure in the concrete in the upper right corner for all the analyses close to the design point. Thus, the ultimate capacity was highly dependent on the con-crete strength. A higher-order polynomial failure surface was regarded as unnecessary owing to the simple ultimate limit behaviour.

Only two basic variables were included for the safety assessment: concrete compressive strength and steel yield strength. The other material parameters were calculated

based on these values and provided as inputs in the analy-ses. This is similar to assuming full correlation between the strength properties and is considered to be an ad-equate simplification in this safety assessment.

6 Conclusions

Even though NLFEAs of reinforced concrete structures have been performed in research settings for many years, they are yet to be used extensively in engineering practice. There are several reasons for this, e.g. the inability to superimpose loads, leading to vast computational costs, but also the lack of safety formats leading surely to the intended safety level as well as sufficient guidance.

The ECOV safety format did not meet the intended safety level for one of the load histories studied. This points out the importance of load history considerations when it comes to NLFEA. Therefore, it would be interest-ing to assess the safety level of the safety formats applied to more complex cases, e.g. a shell structure subjected to a broad variety of loading conditions. Different types of failure modes should be included and also their effect on the modelling uncertainty. Other safety formats that include the failure mode and modelling uncertainty more objectively are perhaps needed [20], since the generality of a safety format might lead to excessive conservatism in some cases.

Furthermore, a framework of engineering guidelines for NLFEA is desired. These should treat all aspects im-portant for conducting an accurate NLFEA. Questions have been raised regarding how to treat the load history but also how to define the design resistance. If the guide-lines are to be accessible for engineers so they can use them in practice, it is of great importance, but also greatly challenging, to present simplifications that can be used in the models and analyses. Substantial research is required to prove which simplifications can be made conservative-ly without degrading the accuracy to such an extent that the purpose of an NLFEA is negated.

Acknowledgements

This article is based on a Master’s thesis project con-ducted at Multiconsult AS in Oslo, Norway, during the autumn of 2014 [21]. The project supervisor was Morten Engen, Research Fellow at Multiconsult AS. The thesis′ examiner was Mario Plos, Associate Professor/Head of the Structural Engineering Division at the Department of Civil and Environmental Engineering, Chalmers Univer-sity of Technology, Gothenburg, Sweden.

References

1. Brekke, D.-E., Åldstedt, E., Grosch, H.: Design of Offshore Concrete Structures Based on Postprocessing of Results from Finite Element Analysis (FEA): Methods, Limitations and Accuracy. Proc. of 4th Intl. Offshore & Polar Enginee-ring Conference, Osaka, 1994, pp. 318–328.

2. Cervenka, V.: Reliability-based non-linear analysis according to fib Model Code 2010. Structural Concrete, 2013, vol. 14, No. 1, pp. 19–28.

3. Engen, M., Hendriks, M. A. N., Øverli, J. A., Åldstedt, E.: Application of NLFEA in the Design of Large Concrete

51



Structures. Proc. of XXII Nordic Concrete Research Sympo-sium, Reykjavik, Iceland, 2014.

4. Engen, M., Hendriks, M. A. N., Øverli, J. A., Åldstedt, E.: Solution strategy for non-linear Finite Element Analyses of large reinforced concrete structures. Structural Concrete, 2015, Vol. 16, No. 3, pp. 389–397.

5. fib: Chapter 7 – Design, fib Model Code for Concrete Struc-tures 2010, Ernst & Sohn GmbH & Co., Hoboken, New Jer-sey, USA, 2013.

6. Joint Committee on Structural Safety: Part 1 – Basis of De-sign. Probabilistic Model Code, Zurich, Switzerland, 2001.

7. Melchers, R. E.: Structural Reliability Analysis and Predic-tion, John Wiley & Sons, Hoboken, New Jersey, USA, 1990.

8. Schlune, H.: Safety Evaluation of Concrete Structures with Nonlinear Analysis. PhD thesis, Chalmers University of Technology, pub. No. 3232, Gothenburg, Sweden, 2011.

9. Bucher, C. G., Bourgund, U.: A fast and efficient response surface approach for structural reliability problems. Structu-ral Safety, 1990, vol. 7, No. 1, pp. 57–66.

10. European Committee for Standardization: European Stan-dard, EN 1992–1–1, Eurocode 2: Design of concrete structu-res – Part 1–1: General rules and rules for buildings. Brus-sels, Belgium, 2004.

11. TNO DIANA: DIANA User’s Manual, release 9.4.4, TNO DIANA BV, Delft, The Netherlands, 2012.

12. Seraj, S., Kotsovos, M., Pavlovic, M.: Application of the com-pressive-force path concept in the design of reinforced con-crete indeterminate structures: A pilot study. Structural Engi-neering and Mechanics, 1995, Vol. 3, No. 5, pp. 475–495.

13. Hendriks, M. A. N., den Uijl, J. A., de Boer, A., Feenstra, P. H., Belletti, B., Damoni, C.: Guidelines for Nonlinear Finite Element Analysis of Concrete Structures, RTD 1016:2012. Technical report, Rijkswaterstaat – Ministerie van Infrastruc-tuur en Milieu, 2012.

14. Selby, R. G., Vecchio, F. J.: Three-dimensional constitutive re-lations for reinforced concrete. Pub. No. 93–02, technical re-port, University of Toronto, Dept. of Civil Engineering, 1993.

15. Vecchio, F. J., Collins, M. P.: Compressive Response of Crak-ked Reinforced Concrete. Journal of Structural Engineering, 1993, 119(12), pp. 3590–3610.

16. Norwegian Standard: NS3576–3: Steels for reinforcement of concrete – Dimensions and properties – Part 3. Standard Norge, Lilleaker, Oslo, 2012.

17. Zhao, W., Qiu, Z.: An efficient response surface method and its application to structural reliability and reliability-based optimization. Finite Elements in Analysis and Design, 2013, Vol. 67, pp. 34–42.

18. Pimentel, M., Bruhwiler, E., Figueiras, J.: Safety examination of existing concrete structures using the global resistance safety factor concept. Engineering Structures, 2014, Vol. 70, pp. 130–143.

19. Kostandyan, E. E., Sørensen, J. D.: Reliability Assessment of IGBT Modules Modeled as Systems with Correlated Compo-nents. Proc. of 2013 Annual Reliability & Maintainability Symposium, 2013, pp. 1–6.

20. Schlune, H., Plos, M., Gylltoft, K.: Safety formats for non-li-near analysis of concrete structures. Magazine of Concrete Research, 2012, Vol. 64, pp. 563–574.

21. Blomfors, M.: Global Safety Assessment of Concrete Structu-res using Nonlinear Finite Element Analysis. Master’s thesis, Chalmers University of Technology, pub. No. 163, Gothen-burg, Sweden, 2014.

Mattias Blomfors, MScindustrial PhD candidateDepartment of Civil and Environmental Engineering, Structural EngineeringChalmers University of TechnologySven Hultins gata 8412 58 GothenburgSwedenTel.: +46 7273 568 [email protected]

Morten Engen, MScindustrial PhD candidateDepartment of Structural EngineeringNorwegian University of Science and TechnologyRich. Birkelandsvei 1A7491 TrondheimNorway & Multiconsult ASPostboks 265 Skøyen0213 OsloNorwayTel.: +47 402 115 [email protected]

Mario PlosAssociate Professor & Head of DivisionCivil and Environmental Engineering, Structural EngineeringChalmers University of TechnologySven Hultins gata 8 412 58 GothenburgSwedenTel.: +46 3177 222 [email protected]


Technical Paper

DOI: 10.1002/suco.201500049

In pretensioned concrete members, prestress is introduced by the bond mechanism between prestressing tendon and surround-ing concrete. Therefore, to secure the intended level of effective prestress in the tendon, sufficient bond stresses between the prestressing tendon and the concrete should be developed at release, for which a certain length from the end of the preten-sioned concrete member is required, and this required distance is defined as the transfer length of the prestressing tendon. In this study, the prestress introduction mechanism between concrete and prestressing tendon was mathematically formulated based on thick-walled cylinder theory (TWCT). On this basis, an analyti-cal model for estimating the transfer length was presented. The proposed model was also verified through comparison with test results collected from the literature. It was confirmed that the proposed model can accurately evaluate the effects of influential factors – such as diameter of prestressing tendon, compressive strength of concrete, concrete cover thickness and magnitude of initial prestress – on the transfer lengths of prestressing tendons in various types of pretensioned concrete member.

Keywords: transfer length, pretensioned concrete member, anchorage, prestressed concrete, bond stress, thick-walled cylinder model

1 Introduction

In pretensioned concrete members, prestress is intro-duced by the bond mechanism developed between pre-stressing tendon and surrounding concrete [1–7] In other words, the diameter of the prestressing tendon, which is initially reduced by the Poisson effect when the tendon is tensioned and anchored to the buttress of the pretension-ing bed, expands simultaneously with the prestress trans-fer. This mechanism induces wedge action, the so-called Hoyer effect, which introduces the prestress into the con-crete. The ACI 318-11 [8] and AASHTO-LRFD [9] design codes express the transfer length required to secure the intended level of effective prestress simply as a function of the diameter of the prestressing tendon dp and the effec-tive prestress fpe. Eurocode 2 (EC2) [10] includes a design

equation for calculating transfer length as a function of the tensile strength of concrete at release fctd, the mag-nitude of initial prestress fpj and the prestressing tendon diameter dp. The EC2 equation also can reflect the various factors influencing the transfer length, such as type of ten-don, release method and bond condition between tendon and concrete, using simple coefficients for these factors, i.e. a1, a2, hp1 and h1. In addition, recent studies [11–13] have reported that the compressive strength of concrete at transfer fci, thickness of concrete cover C, magnitude of initial prestress fpj and diameter of prestressing tendon dp significantly affect the transfer length.

In this study, the transfer mechanism of the tendons in the pretensioned concrete member was mathemati-cally formulated based on thick-walled cylinder theory (TWCT). The bond stress t developed between the tendon and the concrete and the expansion pressure p induced in the concrete due to the Hoyer effect were estimated in the analysis. On this basis, the stress and strain profile curves of the prestressing tendon and the surrounding concrete in the longitudinal direction were determined and the distance from the end of the anchorage to the intersec-tion point between the strain profile curves of tendon and concrete was defined as the transfer length of the prestressing tendon. To verify the proposed model, test results collected from the literature were compared with those estimated by the proposed approach as well as the current design codes.

2 Review of previous research

Table 1 summarizes the transfer length estimation models proposed by previous researchers and the current design codes [8–10]. As shown in Fig. 1, Guyon [14] proposed the relationship between the end slip of the prestressing tendon d and the transfer length lt as follows:

where:d end slip between prestressing tendon and surround-

ing concrete (estimated by integrating the hatched area in Fig. 1)

epi initial prestrain in prestressing tendona shape coefficient of strain profile curve

(1)ltpi

α δε

=

Estimation of transfer lengths in precast pretensioned concrete members based on a modified thick-walled cylinder model

DOI: 10.1002/suco.201500049


Submitted for review: 14 April 2015; revision: 22 June 2015; accepted for publication: 02 July 2015. Discussion on this paper must be submitted within two months of the print publication. The discussion will then be published in print, along with the authors’ closure, if any, approximately nine months after the print publication.

Sun-Jin HanDeuck Hang LeeSang-Heum ChoSoon-Beum KaKang Su Kim*

53

S.-J. Han/D. H. Lee/S.-H. Cho/S.-B. Ka/K. S. Kim · Estimation of transfer lengths in precast pretensioned concrete members based on a modified thick-walled cylinder model


strength of concrete at transfer fci, magnitude of effective prestress fpe and diameter of prestressing tendon dp. Rus-sell and Burns [27] reported that the transfer length could be reduced due to debonding of the prestressing tendon in the end region of a pretensioned concrete member, and that the transfer length at the cut end was generally larger than that at the dead end.

Oh et al. [28] proposed a theoretical model that ide-alized the prestressing tendon and the concrete as solid and hollow cylinders respectively in order to calculate the expansion pressure of the prestressing tendon caused by the Hoyer effect at prestress transfer. They also formulated force equilibrium equations and compatibility conditions between the idealized elements in a polar coordinate system. This analytical model is quite significant in that it can theoretically explain the bond mechanism between the prestressing tendon and the concrete at transfer. However, it requires complex numerical analyses when the concrete cylinder surrounding the tendon is cracked due to expansion of the tendon at transfer, because their model adopted anisotropic constitutive laws to reflect the stiffness reductions. Recently, Park and Cho [7] proposed a transfer length estimation model that assumed the strain profile curve of the prestressing tendon within the transfer length to be a simple predefined quadratic function. The application of this model would be advantageous in that it does not require an iterative calculation. However, his model cannot consider the effect of splitting crack dam-age due to the expansion pressure induced by the Hoyer effect in the concrete, and so it would be difficult to es-timate the transfer length of the pretensioned concrete members cracked in the tangential direction of the section due to insufficient concrete cover thickness.

3 Radial pressure due to the Hoyer effect3.1 Radial pressure before concrete cracking

The diameter of a prestressing tendon in the pretensioned concrete members is reduced during the jacking operation due to the Poisson effect, and expands later upon release of the prestressing tendon. Fig. 2 shows the expansion pressure p in the surrounding concrete at transfer. In Fig. 2:

Ri initial radius of prestressing tendonRj radius of tensioned prestressing tendon before re-

leaseRc distance from centroid of prestressing tendon to

crack tipRo distance from centroid of prestressing tendon to

edge of concrete cover thicknessuj deformation in radial direction of prestressing ten-

don due to prestress release

If the tensile strain in the concrete in the tangential di-rection eq = uj/Rj induced by the expansion pressure p exceeds the tensile cracking strain ecr = ft/Ec, it is assumed that the cracks occur inside a concrete cylinder. The radius of the prestressing tendon reduced by the jacking force Rj can be calculated by multiplying the longitudinal strain in the prestressing tendon by its Poisson’s ratio np as follows:

Several researchers [12, 15–23] have proposed the shape coefficient a of the strain profile curve based on test results, and its approximate range was estimated to be between 1.5 and 3.0. Marin and Scott [24] proposed a simple bilinear relationship between the stress of the pre-stressing tendon and the transfer length, based on test re-sults reported by Hanson and Kaar [25]. Zia and Mostafa [26] also proposed empirical expressions for estimating the transfer length considering the prestress introduction methods, i.e. sudden release or gradual release, based on their test results. Mitchell et al. [11] proposed a transfer length calculation reflecting the effect of the compressive

Transfer length zone

Distance from end

Stra

in c

hang

e

= ∆cz pzε ε

pzεTendon

Concrete

Initial prestress

Slip ( )

piε

δ

Fig. 1. Schematic representation of strain change in transfer length zone

Table 1. Existing models for transfer length

Code or researcher Equation

ACI 318-11 [8] lt = 0.145 (fpe/3)dp (mm)

ASSHTO LRFD [9] lt = 60dp (mm)

Eurocode 2 [10]

Guyon [14]

Marin & Scott [24]

Zia & Mostafa [26]

Mitchell et al. [11]

Russell & Burns [27] lt = 0.0725 fpedp (mm)

(mm)1 21 1

lf

fdt

pj

p ctdpα α

η η=

(mm)ltpi

α δε

=

80135 31 (for 80 , mm)

135 0.39(for 80 , mm)

1/6

1/6

fld d

l d

fd

ld

l d

px

p px p

pp

x

px p

= +

≤

= + ≥

1.5 – 4.6 (sudden release, inch)

1.3 – 2.3 (gradual release, inch)

lf

fd

lf

fd

tpj

cip

tpj

cip

=

=

0.048 20/ (mm)l f d ft pj p ci=

54



where:fpj jacking stress of prestressing tendonEp modulus of elasticity of prestressing tendonnp Poisson’s ratio of prestressing tendon

According to Oh et al. [28], the radial displacement of the prestressing tendon in a polar coordinate system at trans-fer uj can be estimated as follows:

where:Ec modulus of elasticity of concretenc Poisson’s ratio of concrete, taken to be 0.167 in this

studyfcz longitudinal stress in concrete at level of prestress-

ing tendonp radial pressure induced in the concrete cylinder

due to the expansion of the prestressing tendon at release

The latter two can be calculated as follows:

where:fpz stress in prestressing tendonAp cross-sectional area of prestressing tendon

1 – (2)Rf

ERj

pj

pp iν=

–

(1/ – 1/ )

(1 – ) (1 )– (3)

2 2 2 2u

pR

E R R R R

f R

Ejj

c o j

c

o

c

j

c cz j

c

ν ν ν= +

+

1 (4)f f AA

eI

ycz pz p= +

(1 – / ) – (1 – / )

(1 – ) / [ – ( )/( – )] /(5)

2 2 2 2p

R f E R f E

R E R R R R R Ei p pz p j c cz c

p i p c j o j o j c

ν νν ν

=+ +

A cross-sectional area of concrete gross sectionI moment of inertia of concrete gross sectione eccentricity of prestressing tendon from centroid of

concrete section

The tangential strain in the concrete at the boundary between concrete and prestressing tendon eq(Rj) can be calculated by using the radial displacement uj calculated with Eq. (3). When the tangential strain exceeds the crack-ing strain in the concrete ecr = ft/Ec, the analysis of the cracked concrete section applies, which is explained in the following section.

3.2 Radial pressure in partially cracked concrete section

The authors proposed a thick-walled cylinder model (TWCM) in their previous research [29]. They assumed the displacement field of the concrete cylinder to be linear elastic and calculated the radial displacement u(r) and the tangential strain eq(r) as follows:

In the case where the tangential strain in the concrete at the boundary between concrete and prestressing tendon eq(Rj) calculated with Eq. (3) exceeds the cracking strain ecr, the following mathematical relation can be obtained by using Eqs. (3) and (6a):

From this relationship, the radius from centroid of pre-stressing tendon to crack tip Rc can be easily calculated. Further, the distribution of the tangential strain in the cracked concrete cylinder can be obtained using Eq. (6b). In this study, the tension softening curve of concrete [29–30] shown in Fig. 3 was adopted to consider the influence of cracks in the concrete cylinder on the magnitude of the expansion pressure as follows:

where e1 and eu were taken as 0.0003 and 0.002 respec-tively [29]. As shown in Fig. 4a, the equilibrium condition among the expansion pressure p, the confining stress in the concrete at the crack tip pc and the residual tensile

( )( / ) 1

( / ) 1(6a)

2

2u r

fE

rR r

R Rt

c

o

o c

=++

( )( / ) 1

( / ) 1(6b)

2

2r

fE

R rR R

t

c

o

o c

ε =++θ

–

(1/ – 1/ )

(1 – ) (1 )–

( / ) 1

( / ) 1(7)

2 2 2 2

2

2

pR

E R R R R

f R

E

fE

RR R

R R

j

c o j

c

o

c

j

c cz j

c

t

cj

o j

o c

ν ν ν+

+

=

+

+

( ) ( ) when ( ) (8a)r E r rc crσ ε ε ε= ≤θ θ θ

( ) 1 – 0.85( ) –

–when ( ) (8b)

11r f

rrt

cr

crcrσ

ε εε ε

ε ε ε=

< ≤θ

θθ

( ) 0.15( )

–when ( ) (8c)

11r f

rrt

u

uuσ

ε εε ε

ε ε ε=<

< ≤θθ

θ

( ) 0 when ( ) (8d)r ruσ ε ε= <θ θ

Fig. 2. Illustrations of radial displacement and pressure upon prestress transfer

Prestressingforce

a) Idealized thick-walled cylinder

b) Effect of radial pressure on concrete at prestress transfer

55



of the tangential strain when the radial cracks have fully propagated to the edge of the concrete cover can then be obtained using Eq. (11) and the residual tensile stress of the cracked concrete cylinder sq can be obtained with Eq. (8). Since there is no confining stress provided by the concrete pc at the fully cracked stress state, Eq. (9) can be simply rewritten as follows:

Eq. (13) can be used to calculate the expansion pressure of the prestressing tendon p for the fully cracked state.

3.4 Determination of transfer length using bond stress distribution

The bond stress t between prestressing tendon and con-crete can be estimated by using the expansion pressure p calculated in the previous section as follows:

where m is the friction coefficient, which depends on the surface condition between prestressing tendon and concrete. Burgueño and Sun [31] reported that the pre-stressing tendon friction coefficients vary between 0.23

( ) (13)p · R r drjR

R

j

c

∫ σ= θ

· (14)pτ µ=

stress in the cracked concrete in the tangential direction sq(r) can be expressed as follows:

Since the tangential tensile stress at the crack tip sq(Rc) should be identical with the tensile strength of the con-crete ft, the confining stress in the concrete at the crack tip pc can be calculated as follows:

By substituting the confining stress pc estimated with Eq. (10) into Eq. (9), the expansion pressure p can be calcu-lated for the partially cracked state.

3.3 Radial pressure in fully cracked concrete section

Fig. 4b shows the stress condition at which the cracks have fully propagated to the edge of the concrete cover, i.e. a fully cracked concrete section. The following re-lationships can be obtained by substituting Ro for Rc in Eq. (6):

where eqc is the tangential strain at the edge of the con-crete cover. Further, the following relationship can also be obtained from Eqs. (3) and (11a):

Thus, the tangential strain at the edge of the concrete cover eqc can be estimated by Eq. (12). The distribution

( ) (9)pR p R r drj c cR

R

j

c

∫ σ= + θ

–(10)

2 2

2 2p f

R RR Rc t

o c

o c

=+

( )( / ) 1

2(11a)

2u r r

R rc

oε=+

θ

( )( / ) 1

2(11b)

2r

R rc

oε ε=+

θ θ

–

(1/ – 1/ )

(1 – ) (1 )–

( / ) 1

2(12)

2 2 2 2

2

pR

E R R R R

f R

E

RR R

j

c o j

c

o

c

j

c cz j

c

c jo j

ν ν ν

ε

++

=

+θ

θσ

θεuε1εcrε

0.15 tf

tf

cE

Fig. 3. Stress-strain relationship of concrete in tension

Fig. 4. Equilibrium condition at boundary between tendon and concrete

a) Equilibrium for partially cracked concrete section

b) Equilibrium for fully cracked concrete section

56



zero, the analysis will be carried out with the initial value fpz,n = 0. The calculation procedure for the transfer length of the pretensioned concrete member can be summarized as follows: 1. Calculate Rj with Eq. (2). 2. Set fpz,n = 0. 3. Calculate fcz with Eq. (4). 4. Calculate p with Eq. (5). 5. Check if eq(Rj) ≥ ecr. 6. If step 5 is satisfied, calculate Rc with Eq. (7); other-

wise, go to step 11. 7. Calculate p with Eq. (9). 8. Check if Rc ≥ Ro. 9. If step 8 is satisfied, input Rc = Ro; otherwise, go to

step 11.10. Calculate eqc and p with Eqs. (12) and (13).11. Calculate t with Eq. (14).12. Calculate Dfpz and fpz,n with Eqs. (15) and (16).13. Calculate Depz,n and ecz,n with Eqs. (17) and (18).14. Check if Depz,n = ecz,n.15. If step 14 is satisfied, lt can be determined; otherwise,

repeat steps 3–14.

4 Verification of the proposed model

To verify the analytical model proposed in this study, the transfer lengths measured on 97 test specimens were col-lected from previous studies [7, 11–12, 18, 36–37]. Their dimensional and material properties are summarized in Table 2, which includes a wide range of factors that influ-ence the transfer lengths of pretensioned members. The diameter of the prestressing tendon dp ranges from 9.5 to 15.2 mm, the compressive strength of the concrete at transfer fci from 19.2 to 50.0 MPa, the concrete cover thickness C from 30.0 to 57.2 mm, the eccentric ratio of the prestressing tendon e/h from 0.15 to 0.32 and the magnitude of the initial prestress from 0.48 fpu to 0.8 fpu.

In the case where multiple tendons were provided in the pretensioned concrete test specimens, the effective cover thickness suggested by Den Uijl [15] was used to consider the influence of the strand spacing (i.e. multiple strands effect) as follows:

where:C clear cover thicknessnp number of tendonssp spacing of tendons

4.1 Determination of friction coefficient

As mentioned above, previous studies [31–35] reported different friction coefficients, which implies that it is dif-ficult to identify the friction coefficient between prestress-ing tendon and concrete. In this study the friction coef-ficient m was therefore selected from the analysis results considering a range of m values from 0.3 to 0.7 so that the proposed method can provide the closest estimation of the transfer lengths of the test specimens collected from

2 1.5 ( – 1)

2(19)C

C n s

neffp p

p=

+

and 0.7; other researchers [32–35] have proposed different values ranging from 0.4 to 1.4. In this study, the friction coefficient m was selected to be 0.5, based on a parametric study of m values ranging from 0.3 to 0.7, which will be addressed later in the verification section. As shown in Fig. 5, the pretensioned concrete member can be discre-tized into n elements with length Dz, and the stress incre-ment of the prestressing tendon Dfpz due to the accumula-tion of the bond stress t in an element can be calculated as follows [28]:

Since the stress in the prestressing tendon at the end of the pretensioned concrete member (i.e. z = 0) should be zero, the strain change in the tendon Depz is identical with the initial prestrain epi. Thus, the stress and strain changes in the prestressing tendon, fpz,n and Depz,n, at the arbitrary nth element can be calculated as follows:

From Eq. (4), the concrete strain at the nth element ecz,n can be calculated as follows:

Thus, the transfer length lt can be determined as the dis-tance from the end anchorage to the point shown in Fig. 1 where the strain change in the prestressing tendon Depz,n obtained from Eq. (17) is equal to the concrete strain ecz,n obtained from Eq. (18), i.e. Depz,n = ecz,n.

3.5 Solution algorithm

At prestress transfer, since the stress in the prestressing tendon at the end anchorage of the member (i.e. z = 0) is

(15)fd

Azpz

p

p

π τ∆ = ∆

(16)0

f fpz,n pzk

n

∑= ∆=

– (17)f

Epz,n pipz,n

pε ε∆ =

1 (18)f A

E AeI

ycz,npz,n p

cε = +

Fig. 5. Free body diagram between point k and k + 1

57



Table 2. Summary of test specimens and analysis results

Researchers Specimen bw(mm)

h(mm)

dp(mm)

C(mm)

fci(Mpa)

fpj(Mpa)

e(mm)

lt,exp(mm)

lt,cal(mm)

Oh & Kim [12] M12-N-C3-1&2 112.7 200 12.7 30 33.6 1402 63.7 851 810

M12-N-C4-1&2 112.7 200 12.7 40 35 1391 53.7 669 693

M12-N-C5-1&2 112.7 200 12.7 50 33.6 1402 43.7 589 690

M12-H-C3-1&2 112.7 200 12.7 30 44.7 1359 63.7 692 654

M12-H-C4-1&2 112.7 200 12.7 40 46.3 1375 53.7 568 597

M12-H-C5-1&2 112.7 200 12.7 50 44.7 1394 43.7 513 594

M15-N-C3-1&2 115.2 200 15.2 30 35 1377 62.4 1084 1056

M15-N-C4-1&2 115.2 200 15.2 40 33.6 1392 52.4 839 867

M15-N-C5-1&2 115.2 200 15.2 50 35 1393 42.4 698 792

M15-H-C3-1&2 115.2 200 15.2 30 46.4 1357 62.4 888 819

M15-H-C4-1&2 115.2 200 15.2 40 44.7 1364 52.4 722 720

M15-H-C5-1&2 115.2 200 15.2 50 45.6 1384 42.4 574 684

T12-N-S3 150.8 200 12.7 30 34 1398 43.7 808 798

T12-N-S4 150.8 200 12.7 40 35.5 1418 43.7 674 753

T12-N-S5 150.8 200 12.7 50 37.3 1389 43.7 632 711

T12-H-S3 150.8 200 12.7 30 44.2 1374 43.7 695 681

T12-H-S4 150.8 200 12.7 40 43.2 1377 43.7 595 672

T12-H-S5 150.8 200 12.7 50 46.3 1392 43.7 558 648

T15-N-S3 160.8 200 15.2 30 37.6 1357 42.4 997 909

T15-N-S4 160.8 200 15.2 40 34.8 1361 42.4 840 876

T15-N-S5 160.8 200 15.2 50 33.4 1381 42.4 782 873

T15-H-S3 160.8 200 15.2 30 47.2 1376 42.4 889 795

T15-H-S4 160.8 200 15.2 40 46.9 1400 42.4 725 762

T15-H-S5 160.8 200 15.2 50 43.9 1377 42.4 662 738

Park et al. [36] S1&2&4 150 150 15.2 75 31.5 1348 – 1158 846

Park & Cho [7] C45-S-A70-CS15-CT8-FB 150 150 15.2 75 31.5 1302 – 661 780

Thomas.et al. [37] T3UN 88.9 88.9 9.5 39.7 29.9 1490 – 897 555

S&F3UN 102.4 152.4 9.5 46.1 29.9 1490 25.4 864 567

T5UN 102.4 102.4 12.7 44.9 29.9 1490 25.4 1753 762

S&F5UN 127 203.2 12.7 57.2 29.9 1490 30.6 1267 780

S&F6UN 152.4 254 15.2 68.6 29.9 1490 38.1 1499 921

Russell & Burns [18] SS150-1&2 102 127 12.7 44.7 19.2 1406 – 1473 969

SS150-3&4 102 127 12.7 44.7 26 1299 – 775 774

SS160-1&2 102 127 15.2 43.4 24.3 1239 – 1182 936

SS160-3&4 102 127 15.2 43.4 30.2 1311 – 1181 858

SS160-5&6 102 127 15.2 43.4 46.9 1287 – 1042 870

SS160-7&8 102 127 15.2 43.4 46.9 1239 – 826 834

58



4.2 Parametric study of influential factors

To investigate the effect of the various parameters – such as cover thickness of concrete C, concrete compressive strength at transfer fci, magnitude of initial prestress fpj and diameter of prestressing tendon dp – that influence the transfer length, parametric analyses were performed using the proposed method. The dimensional and material properties used for the parametric study are summarized in Table 4.

Fig. 7a shows the effect of concrete cover thick-ness on the transfer lengths of the pretensioned concrete members estimated by the proposed model. For the PSC members with small concrete cover depths less than about 30 mm, there is a significant decreasing trend of transfer

the literature [7, 11–12, 18, 36–37]. As shown in Fig. 6 and Table 3, m = 0.5 provided the most accurate analysis result with the mean and the COV of the transfer length ratios lt,cal/lt,exp being 1.029 and 0.225 respectively. It should be noted that the COVs of the analysis results, which represent the accuracy of the proposed model, are not in-fluenced by the friction coefficients, whereas the averages and the standard deviations are greatly modified by them.

Table 2. Continued

Researchers Specimen bw(mm)

h(mm)

dp(mm)

C(mm)

fci(Mpa)

fpj(Mpa)

e(mm)

lt,exp(mm)

lt,cal(mm)

Mitchell et al. [11] 9.5/31-1200 100 200 9.5 45.3 21 1219 50 506 603

9.5/43-1350 100 200 9.5 45.3 27 1240 50 584 552

9.5/43-1000 100 200 9.5 45.3 27 1240 50 482 552

9.5/65-800 100 200 9.5 45.3 48 1192 50 303 417

9.5/75-950&700 100 200 9.5 45.3 50 1230 50 406 417

9.5/89-825&575 100 200 9.5 45.3 50 1234 50 415 417

13/31-1200 150 225 12.7 43.7 21 1274 62.5 811 930

13/43-1600&1250 100 200 12.7 43.7 27 1217 50 584 711

13/65-850 150 225 12.7 43.7 48 1315 62.5 506 594

13/75-1100 100 200 12.7 43.7 50 1303 50 507 543

13/75-950 100 200 12.7 43.7 50 1303 50 405 543

13/89/950 125 175 12.7 43.7 50 1329 37.5 387 567

13/89/650 125 175 12.7 43.7 50 1329 37.5 495 567

16/31-1865 200 250 15.2 42.4 21 1220 75 872 1098

16/31/1500 200 250 15.2 42.4 21 1220 75 912 1098

16/65/1150 200 250 15.2 42.4 48 1176 75 528 687

16/65/725 200 250 15.2 42.4 48 1176 75 536 687

16/89-975 125 175 15.2 42.4 50 871 37.5 306 525

16/89-675 125 175 15.2 42.4 50 871 37.5 465 525

Fig. 6. Effect of friction coefficients on analysis results

Table 3. Summary of analysis results for friction coefficients

Friction coefficient

Mean(lt,cal/lt,exp)

SD(lt,cal/lt,exp)

COV(lt,cal/lt,exp)

m = 0.3 1.715 0.387 0.225

m = 0.4 1.287 0.290

m = 0.5 1.029 0.232

m = 0.6 0.857 0.193

m = 0.7 0.734 0.165

59



lengths since the confining stresses induced by the sur-rounding concrete become larger as the concrete cover thickness increases. When the concrete cover thickness is greater than about 40 mm, however, its effect on the mag-nitude of the transfer lengths becomes marginal. This is because sufficient confining stresses can be developed by the surrounding concrete with a certain cover thickness. Fig. 7b illustrates the effect of the compressive strength of concrete at release on the magnitude of the transfer length. The analysis results show that the transfer lengths decrease as the compressive strength of concrete at re-lease increases, which is because the confining stresses developed by the uncracked concrete cylinder pc and the residual tensile stress in the cracked cylinder sq increase as the compressive strength increases. Consequently, the bond stresses between prestressing tendon and concrete increase, and so the transfer lengths decrease. For the magnitude of the initial prestress and the diameter of the prestressing tendon as shown in Figs. 7c and 7d, the transfer lengths increase as both factors become larger. This is because the prestressing force itself increases in proportion to the magnitude of the initial prestress and the diameter of the prestressing tendon.

The proposed model provided reasonable trends for the transfer lengths of PSC members according to the various influential factors. Its accuracy for estimating the transfer lengths of test specimens will be described in the following section.

4.3 Verifications

Fig. 8 shows a comparison of the transfer lengths of speci-mens reported by Oh et al. [12] and those estimated by the proposed model. The longitudinal bond stress distribution of the test specimens calculated by the proposed model is shown for reference at bottom right of each graph. Note that the initial prestress of the tendon is about 0.7 fpu for all test specimens. As shown in Figs. 8a and 8b, the transfer length of specimen M12-N-C4-1 was calculated to be smaller than that of specimen M12-N-C3-2, which is due to the larger concrete cover thickness C of specimen M12-N-C4-1 than that of specimen M12-N-C3-2. In other words, the pretensioned concrete member with a larger concrete cover thickness can provide higher confine-ment, and thus its transfer length is surely smaller than that of a member with small concrete cover thickness. In addition, as shown in Figs. 8b and 8c, the transfer length of specimen M12-H-C4-1 was calculated to be smaller than that of specimen M12-N-C4-1, and this tendency was also observed in the test results. This is because even though the concrete cover thickness C of both specimens was 40.0 mm, the tensile strength of the concrete ft of specimen M12-H-C4-1 was greater than that of specimen M12-N-C4-1. Therefore, specimen M12-H-C4-1 showed a relatively higher bond strength, as shown in the bond

Table 4. Dimensional and material properties used for parametric study

bw (mm)

h (mm)

fpu (MPa)

C (mm)

fct (MPa)

fpj (MPa)

db (mm)

200 200 1860 20–100 20–100 0.4–0.8fpu 9.5–15.2

Fig. 7. Effect of various parameters on transfer length

b) Effect of concrete compressive strength at release

d) Effect of prestressing tendon diameter

c) Effect of magnitude of initial prestress

a) Effect of concrete cover thickness

60



stress distribution calculated by the proposed model. The estimated transfer length of specimen M15-N-C3-1 is larg-er than that of specimen M12-N-C3-2, as shown in Figs. 8a and 8d. This is because the amount of prestressing tendon Ap provided in specimen M15-N-C3-1 was greater than that in specimen M12-N-C3-2, which means that larger compressive stresses acting on the concrete section was expected for specimen M15-N-C3-1. Therefore, splitting crack damage would affect specimen M15-N-C3-1 more significantly than specimen M12-N-C3-2, and the bond stress developed between the concrete and the prestress-ing tendon of specimen M15-N-C3-1 is also smaller than that of specimen M12-N-C3-2.

Fig. 9 shows a comparison between the analysis results of the collected test data [7, 11–12, 18, 36–37] esti-mated by the proposed model and those estimated by the current code equations [8–10] (see also Table 2). The value dpfpj/C √−−

fci of the test specimens ranged from 36 to 118. The average (mean) and the coefficient of variation (COV) of the calculated-to-observed transfer length ratios lt,cal/lt,exp of the proposed model were 1.029 and 0.225 respec-tively, which is considered to represent good accuracy. The code equations of ACI 318-11 [8] and AASHTO-LR-FD [9] provided conservative results compared with the proposed model, and such conservatism mainly resulted from not reflecting the important influential factors, such as concrete cover thickness C and compressive strength of concrete at transfer fci. The use of the EC2 equation [10] is advantageous in that the various factors influencing the transfer length can be reflected in a simple manner, but the effect of concrete cover thickness is not yet consid-ered in the EC2 model. The accuracy of the EC2 model was comparable to the ACI 318-11 code model, but if the effect of concrete cover thickness is included in the EC2 model, it seems that its accuracy could be improved.

5 Conclusions

This study proposed an analytical model for estimating transfer length which could reflect the influence of split-ting cracks in the radial direction of the section as well as the expansion pressure induced by the Hoyer effect at pre-stress release. In addition, a large number of test results Fig. 8. Verification of the proposed model by comparison with test results

a) Specimen M12-N-C3-2

b) Specimen M12-N-C4-1

c) Specimen M12-H-C4-1

d) Specimen M15-N-C3-1

Fig. 9. Comparison of code equations and proposed model

61



u(r) radial displacement for whole concrete cylinderDfpz stress increment of prestressing tendonDz length of discretized element a shape coefficient of strain profile of tendon and

concreteepi magnitude of initial prestrain in prestressing tendon Depz,n strain change in tendon in nth elementecr cracking strain corresponding to tensile strength of

concreteecz,n concrete strain in nth elementeu concrete strain corresponding to zero tensile stress

in concreteeqc unknown concrete hoop tensile strain at concrete

surfaceeq(r) hoop tensile strain in concrete for whole concrete

cylindere1 concrete strain corresponding to tensile stress of

concrete (= 0.15ft)m friction coefficient between prestressing tendon

and concretesq(r) tensile stress corresponding to hoop tensile straint bond stress between prestressing tendon and con-

cretenc Poisson’s ratio of concretenp Poisson’s ratio of prestressing tendon

References

1. Lee, J. Y., Lee, D. H., Hwang, J. H., Park, M. K., Kim, K. S., Kim, H. Y.: Investigation on Allowable Compressive Stresses in Pretensioned Concrete Members at Transfer. KSCE Jour-nal of Civil Engineering, 2013, 17, No. 5, pp. 1083–1098.

2. Lee, D. H., Hwang, J. H., Kim, K. S., Kim, J. S., Chung, W. S., Oh, H. S.: Simplified Strength Design Method for Allowable Compressive Stresses in Pretensioned Concrete Members at Transfer. KSCE Journal of Civil Engineering, 2014, 18, No. 7, pp. 2209–2217.

3. Lee, D. H., Park, M. K., Oh, J. Y., Kim, K. S., Im, J. H., Seo, S. Y.: Web-Shear Capacity of Prestressed Hollow-Core Slab Unit with Consideration on the Minimum Shear Reinforce-ment Requirement. Computers and Concrete, 2014, 14, No. 3, pp. 211–231.

4. Lee, D. H., Kim, K. S.: Flexural Strength of Prestressed Con-crete Members with Unbonded Tendons. Structural Enginee-ring and Mechanics, 2011, 38, No. 5, pp. 675–696.

5. Kim, K. S., Lee, D. H.: Flexural Behavior Model for Post-Tensioned Concrete Members with Unbonded Tendons. Computers and Concrete, 2012, 10, No. 3, pp. 241–258.

6. Kim, K. S., Lee, D. H.: Nonlinear Analysis Method for Conti-nuous Post-Tensioned Concrete Members with Unbonded Ten-dons. Engineering Structures, 2012, 40, No, 1, pp. 487–500.

7. Park, H., Cho, J. Y.: Bond-Slip-Strain Relationship in the Transfer Zone of Pretensioned Concrete Elements. ACI Structural Journal, 2014, 111, No. 3, pp. 503–514.

8. American Concrete Institute Committee 318: Building Code Requirements for Structural Concrete and Commentary, Far-mington Hills, 2011.

9. American Association of state Highway and Transportation Officials (AASHTO): AASHTO LEFD Bridge Design Speci-fications: Customary U.S. Units, 4th ed., AASHTO, Washing-ton, D.C., 2007.

10. European Committee for Standardization: Eurocode 2: De-sign of Concrete Structures – Part 1-1: General Rules and Rules for Buildings, Brussels, 2004.

11. Mitchell, D., Cook, W. D., Khan, A. A., Tham, T.: Influence of High Strength Concrete on Transfer and Development

were collected and used to verify the proposed model. The following conclusions could be drawn from this study:1. An analytical model for estimating transfer length was

proposed in this study based on thick-walled cylinder theory, which can provide not only the transfer length, but also the expansion pressure of the prestressing ten-don and the bond stress between tendon and concrete.

2. The transfer lengths measured on 97 test specimens were compared with the analysis results estimated by the proposed model. This showed that the proposed model can accurately assess the transfer length of the pretensioned concrete specimens.

3. The analytical model proposed in this study reason-ably reflected the influences of compressive strength of concrete at transfer fci, magnitude of initial prestress fpj, concrete cover thickness C and diameter of prestress-ing tendon dp.

4. The proposed model requires some iterative calcula-tions and so a simplified approach needs to be devel-oped for its practical application.

Acknowledgments

This research was supported by the Basic Science Re-search Program through the National Research Founda-tion of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (NRF-2013R1A1A1A05007001).

Notation

A gross sectional area of concreteAp cross-sectional area of tendonC clear cover thicknessCeff effective cover thicknessdp nominal diameter of prestressing tendonEc elastic modulus of concreteEp elastic modulus of prestressing tendone eccentricity of tendonfci compressive strength of concrete at prestress releasefctd tensile strength of concrete at releasefcz compressive stress of concrete in z directionfpj magnitude of initial prestress (or jacking stress) in

prestressing tendonfpz tensile stress in tendon in z directionfpz,n tensile stress in tendon in z direction of nth elementft tensile strength of concreteI moment of inertia of gross sectionlt,cal transfer length calculated by proposed modellt,exp transfer length measured from experimentsnp number of prestressing tendonsp expansion pressure of prestressing tendonpc confining stress of concrete at crack tipRc radius from centroid of prestressing tendon to

crack tipRi initial radius of prestressing tendonRj radius of prestressing tendon reduced by jacking

forceRo radius from centroid of prestressing tendon to edge

of concrete cover thickness (= C + dp/2)sp spacing of prestressing tendonsuj deformation in radial direction of prestressing ten-

don at prestress release

62



32. Wang, X. H., Liu, X. L.: Modeling bond strength of corroded reinforcement without stirrups. Cement and Concrete Re-search, 2004, 34, No. 8, pp. 1331–1339.

33. Tepfers, R.: Cracking of Concrete Cover Along Anchored De-formed Reinforcing Bars. Magazine of Concrete Research, 1979, 31, No. 106, pp. 3–12.

34. Laldji, S., Young, A. G.: Bond between Steel Strand and Ce-ment Grout in Ground Anchorages. Magazine of Concrete Research, 1988, 31, No. 106, pp. 3–12.

35. Arab, A. A., Badie, S. S., Manzari, M. T.: A Methodological Approach for Finite Element Modeling of Pretensioned Con-crete Members at the Release of Pretensioning. Engineering Structures, 2011, 33, No. 6, pp. 1918–1929.

36. Park, H., Din, Z. U., Cho, J. Y.: Methodological Aspects in the Measurement of Strand Transfer Length in Pretensioned Concrete. ACI Structural Journal, 2012, 109, No. 5, pp. 625–634.

37. Thomas, E. C., David, W. J., Paul, Z.: Transfer Length of Epoxy-Coated Prestressing Strand. ACI Materials Journal, 1990, 87, No. 3, pp. 193–203.

Sun-Jin HanGraduate StudentDepartment of Architectural EngineeringUniversity of Seoul163 Siripdaero, Dongdaemun-guSeoul 130-743, KoreaTel: 82-2-6490-5576Fax: [email protected]

Deuck Hang Lee, Ph.D.Research ScholarDepartment of Civil and Environmental EngineeringUniversity of Illinois at Urbana-Champaign205 N. Mathews Ave. UrbanaIllinois 61801, United StatesTel : 1-618-380-1066Fax: [email protected]

Sang-Heum ChoGraduate StudentDepartment of Architectural EngineeringUniversity of Seoul163 Siripdaero, Dongdaemun-guSeoul 130-743, KoreaTel: 82-2-6490-5576Fax: [email protected]

Soon-Beum KaGraduate StudentDepartment of Architectural EngineeringUniversity of Seoul163 Siripdaero, Dongdaemun-guSeoul 130-743, KoreaTel: 82-2-6490-5576Fax: [email protected]

Kang Su Kim, Ph.D.ProfessorDepartment of Architectural EngineeringUniversity of Seoul163 Siripdaero, Dongdaemun-guSeoul 130-743, KoreaTel: 82-2-6490-2762Fax: [email protected]

Length of Pretensioning Strand. PCI Journal, 1993, 38, No. 3, pp. 52–66.

12. Oh, B. H., Kim, E. S.: Realistic Evaluation of Transfer Lengths in Pretensioned, Prestressed Concrete Members. ACI Structural Journal, 2000, 97, No. 6, pp. 821–830.

13. Oh, B. H., Lim, S. N., Choi, Y. C.: Finite Element Analysis of Transfer Length in Pretensioned Prestressed Concrete Mem-bers. Journal of the Korea Concrete Institute, 2004, 16, No. 3, pp. 293–302.

14. Guyon, Y.: Pretensioned Concrete: Theoretical and Experi-mental Study, Paris, 1953.

15. Den Uijl, J. A.: Bond Modelling of Prestressing Strand. Bond and Development of reinforcement – A tribute to Dr. Peter Gergely, SP-180-7, American Concrete Institute, Farmington Hills, 1998, pp. 145–169.

16. Lopes, S. M. R., do Carmo, R. N. F.: Bond of Prestressed Strands to Concrete: Transfer Rate and Relationship bet-ween Transfer Length and Tendon Draw-in. Structural Con-crete, 2002, 3, No. 3, pp. 117–126.

17. Brooks, M. D., Gerstle, K. H., Logan, D. R.: Effect of Initial Strand Slip on the Strength of Hollow-Core Slabs. PCI Jour-nal, 1988, 33, No. 1, pp. 90–111.

18. Russell, B. W., Burns, N. H.: Measured Transfer Lengths of 0.5 and 0.6 in. Strands in Pretensioned Concrete. PCI Jour-nal, 1996, 44, No. 5, pp. 44–65.

19. Logan, D. R.: Acceptance Criteria for Bond Quality of Strand for Pretensioned Prestressed Concrete Applications. PCI Journal, 1997, 42, No. 2, pp. 52–90.

20. Steinberg, E., Beier, J. T., Sargand, S.: Effects of Sudden Prestress Force Transfer in Pretensioned Concrete Beams. PCI Journal, 2001, 46, No. 2, pp. 64–75.

21. Wan, B., Harries, K. A., Petrou, M. F.: Transfer Length of Strands in Prestressed Concrete Piles. ACI Structural Jour-nal, 2002, 99, No. 5, pp. 577–585.

22. Olesniewicz, A.: Statistical Evaluation of Transmission Length of Strand, Research and Design Centre for Industrial Building (BISTYP), Warsaw, 1975.

23. Jonsson, E.: Anchorage of Strands in Prestressed Extruded Hollow Core Slabs. Proc. of Intl. Symp. on Bond in Concre-te: From Research to Practice, Riga Technical University and CEB (eds.), Riga, Latvia, 1992, pp. 2.20–2.28.

24. Marin, L. D., Scott, N.: Development of Prestressing Strand in Pretensioned Members. ACI Journal, 1976, 73, No. 8, pp. 453–456.

25. Hanson, N., Kaar, P.: Flexural Bond Tests of Pretensioned Hollow Core units. ACI Journal, 1959, 55, No. 7, pp. 783–803.

26. Zia, P., Mostafa, T.: Development Length of Prestressing Strand. PCI Journal, 1977, 22, No. 5, pp. 54–65.

27. Russell, B. W., Burns, N. H.: Measurement of Transfer Lengths on Pretensioned Concrete Elements. Journal of Structural Engineering, ASCE, 1997, 123, No. 5, pp. 541–549.

28. Oh, B. H., Kim, E. S., Choi, Y. C.: Theoretical Analysis of Transfer Lengths in Pretensioned Prestressed Concrete Members. Journal of Engineering Mechanics, ASCE, 2006, 132, No. 10, pp. 1057–1066.

29. Han, S. J., Lee, D. H., Kim, K. S., Seo, S. Y., Moon, J. H., Monteiro, P. J. M.: Degradation of Flexural Strength in Rein-forced Concrete Members Caused by Steel Corrosion. Cons-truction and Building Materials, 2014, 54, No. 1, pp. 572–583.

30. Bhargava, K., Ghosh, A. K.: Analytical model of corrosion-induced cracking of concrete considering the stiffness of re-inforcement. Structural Engineering and Mechanics, 2003, 16, No. 6, pp. 749–769.

31. Burgueño, R., Sun, Y.: Stress transfer characteristics of shea-thed strand in prestressed concrete beams: computational study. PCI Journal, 2014, 59, No. 3, pp. 95–109.


In modern bridge construction, steel-concrete composite struc-tures with composite dowels are being built more than ever, es-pecially for small and medium spans. In contrast to headed studs, in which initial steel cracks occur after only a few load cycles [1], [2], the lifetime of composite dowels is characterized by the com-pression of the multi-axially stressed concrete in front of the composite dowel. Here, plastic compression strains occur in the concrete and accumulate over load cycles, leading to a cyclic in-crease in relative displacements in the connection. Certain pro-portions of these relative displacements, called inelastic slip, re-main in the connection, even after the loading is relieved. The inelastic slip changes the characteristics of the static dowel curve. The initially rigid connection degrades over its lifetime, leading to redistributions of internal forces, which may be critical for fatigue design. In order to consider the degradation of the composite connection, a cyclic dowel curve can be used, which may be developed from the static dowel curve by introducing the inelastic slip. This paper presents the results of cyclic shear tests on different composite dowel geometries. The effect of load-de-pendent parameters (upper load level and load range) was inves-tigated. Furthermore, an engineering model for determining the cyclic dowel curve is presented, which was developed based on the results of experimental and numerical investigations.

Keywords: composite construction, fatigue of concrete, composite dowels, cyclic shear tests, multi-axial stress state, inelastic slip, cyclic dowel curve

1 Introduction

In steel-concrete composite girders, innovative composite dowels can be used to transfer shear forces between the concrete slab and the steel section. Composite dowels ha-ve been developed as an alternative to headed stud con-nectors. They are produced by plasma or laser cutting. The cutting torch burns regular open (e.g. puzzle- or clo-thoid-shaped [3]) or closed recesses (e.g. Perfobond shape [4]) immediately into the web of a steel beam or into a steel strip that is subsequently welded to the upper flange of the steel beam. After encasing the strip in concrete, the

vertically embedded steel dowels and the interstitial con-crete dowels ensure a structural, interlocked connection. The innovative composite dowels combine high shear ca-pacity and sufficient deformation capacity, even in high- and ultra-high-performance concretes.

The use of composite dowels is currently not embo-died in EC4 [5], but in Germany a National Technical Approval for composite dowels ([6], [7]) regulates the use of clothoid- (CL-shape) and puzzle-shaped (PZ-shape) do-wels in normal-strength concrete composite beams (Fig. 1, right). Possible configurations of composite beams with composite dowels are shown in Fig. 1 (left).

Fig. 2 is an overview of possible failure modes for composite dowels. Under static loading, shear failure of the concrete dowel (Fig. 2a), pry-out failure (b) and verti-cal concrete splitting (c) can occur, whereas the steel do-wel may fail due to combined shear and bending stresses (d) [8]. Besides their use in building construction ([9]–[13]), where they are mainly exposed to static loading, composi-te dowels are particularly prevalent in engineering structu-res and bridge construction ([14]–[16]), where the cyclic traffic loads may cause fatigue processes in the composite connection. The few existing studies documented in the literature ([17], [18]) have dealt mainly with the fatigue processes of the steel dowel, whereas the concrete fatigue behaviour has been disregarded. So far, these studies have been conducted on beam tests, where the global fatigue behaviour was investigated. By way of a contrast, the pre-sent paper focuses on the local concrete fatigue processes in composite dowels, which may lead either to cyclic con-crete pry-out (Fig. 2e) if the fatigue tensile strength is ex-ceeded, or to the release of pulverized concrete in cracked concrete slabs (f), or to the evolution of cyclic slip in the

Technical Paper

Concrete fatigue in composite dowels

Martin Classen*Joerg Gallwoszus

DOI: 10.1002/suco.201400120


Submitted for review: 22 December 2014; revision: 24 February 2015; accepted for publication: 28 March 2015. Discussion on this paper must be submitted within two months of the print publication. The discussion will then be published in print, along with the authors’ closure, if any, approximately nine months after the print publication.

Fig. 1. Composite dowel applications in composite structures (left) and do-wel geometries investigated (right)

composite connection (g) [19]. Since cyclic pry-out (e) on-ly occurs with very small concrete covers, which are not common in bridge construction, and failure mode (f) only arises in negative bending regions, mode (g) has the high-est practical relevance and is analysed in detail below.

In the recesses of the composite dowel, where shear forces are transferred between the steel and the concrete, the concrete is exposed to multi-axial compression stress-es. Plastic compression strains occur in the concrete, which accumulate over the load cycles and lead to a cyclic increase in relative displacements in the composite con-nection (Fig. 2g). Certain proportions of these relative dis-placements remain in the connection, even after the exter-nal loading is relieved. These permanent displacements are called inelastic slip din (Fig. 3a). When inelastic slip occurs, the assumption of a rigid connection between the steel and the concrete, as specified in bridge construction, no longer applies. Inelastic slip leads to the degradation of the composite effect (Fig. 3b) and an increase in the parti-al bending moments in the steel beam and the concrete slab. Consequently, higher local stresses occur in the steel section, which may contribute to premature fatigue crack-

64

M. Classen/J. Gallwoszus · Concrete fatigue in composite dowels


ing in the steel. Owing to these complex interactions bet-ween steel and concrete, which have not been investiga-ted so far, current design provisions tend to exclude the effect of cyclic slip increase, defining very conservative li-mits for the acceptable multiaxial compressive stresses in the concrete at the ultimate limit state [6], [7].

However, in the present paper, approaches that expli-citly consider concrete fatigue in composite dowels are in-troduced for the first time. Here, the transformation of the static dowel curve into a cyclic dowel curve, considering the effect of inelastic slip in the connection, is crucial. In practical engineering, linear elastic FE calculations of com-posite structures are frequently used to determine the rele-vant steel stresses for fatigue design, with shear connectors typically modelled as springs coupling the composite mem-bers (steel and concrete). In such calculations, the concrete fatigue and the increase in inelastic slip can be considered by replacing the static spring characteristics of the shear connectors by the corresponding cyclic dowel curve (Fig. 3c,d). The cyclic dowel curve at a specified number of load cycles n is determined by offsetting the static dowel curve by the inelastic slip at n load cycles (Fig. 3c).

Fig. 2. Failure modes of composite dowels under static loading (left) and fatigue loading (right)

Fig. 3. Inelastic slip, cyclic dowel curve and its application in FE calculations

65



2 Characteristics of concrete fatigue

Fatigue in composite dowels is characterized by the com-plex fatigue characteristics of the two materials from which they are made – steel and concrete. The fatigue of concrete under uniaxial stress with a constant stress range can be described by means of a three-phase diagram of strain over lifetime (or, similarly, load cycles). The first phase is characterized by rapidly growing micro-scale cracks, the second phase by the continuous growth of tho-se cracks and the third phase by unstable growth of the cracks and failure caused by fracture surfaces (Fig. 4, left) [20], [21]. For normal-strength concrete, phases I and III represent about 20 % of the overall life cycle.

The fatigue of the concrete in the recesses of the composite dowel is characterized by a multi-axial stress state (Fig. 4, right). Applied lateral compression stresses up to approx. 80 % of the uniaxial compression strength lead to a significant increase in the number of load cycles to failure, compared with tests without lateral pressure [23], [24]. This holds for both actively introduced lateral pressure as well as passive lateral pressure resulting from concrete confinement (e.g. by stirrup reinforcement). The present paper focuses on the fatigue processes in the con-crete of the composite dowel; steel cracking is not con sid-ered here as the fatigue behaviour of steel is comprehensi-vely documented in the literature (e.g. [25]).

3 Experimental investigations3.1 General

Cyclic shear tests were performed with different test se-tups in order to investigate the cyclic slip increase in composite dowels with puzzle and clothoid shapes. The test programme was limited to the investigation of load-dependent parameters (load range and upper loads). All geometrical and material parameters were kept constant. The effect of varying load ranges was investigated by maintaining a constant lower load level while adjusting the upper load. To test the impact of the upper load, the load range was kept constant while adjusting the lower load.

The aims of the experimental investigations were to investigate the cyclic slip increase and the concrete fatigue behaviour in phases I and II. Determining the number of

cycles to failure that the dowels could withstand or the explicit failure state of the composite connection (phase III) was not the subject of the investigations.

3.2 Test setup and performance

The test setups used for the experimental investigations are shown in Fig. 5 (top). The push-out standard test (POST) according to EN 1994-1-1 [5] (Fig. 5, left) consists of a composite beam section under shear forces in which the (average) dowel force-slip relation of the shear con-nectors is measured. Under cyclic loading, the POST se-tup turned out to be quite sensitive to unintended mis-alignments of the concrete slabs, which may result from small inaccuracies during the installation of the test speci-men. As the shear force distribution in the dowels can be significantly distorted due to small, unavoidable asymme-tries, additional shear tests with single shear connectors were performed. Fig. 5 (right) shows the setup for a single push-out test (SPOT) in which only a single shear connec-tor is tested under almost pure shear loading. As the lines of action of the external forces are (almost) identical, the specimen can be tested without bending. As this is a stati-cally determinate test setup, the connector’s shear force is equal to the applied testing force, applied by tensile bars. Here, inevitable installation accuracies are negligible and may be tolerated as they do not affect irregular shear force distributions in the dowels. As with the measurements in the POST setup, the slip between the steel dowel and the concrete slab is measured with inductive displacement transducers close to the concrete dowels. The comparabi-lity of the two setups (POST and SPOT) is documented in detail in [19], [26].

Steel dowels made from HEB 600 sections (S355, tw = 15.5 mm) were used in all the tests. The geometry of the composite dowels was burned into the web by flame cutting. The concrete slab was made of normal-strength concrete (C30/37) and included highly ductile B500 rein-forcing steel. The reinforcement in the specimens is shown in Fig. 5 (bottom). The cyclic loading of the speci-men was applied with controlled force according to the loading scheme depicted in Fig. 6. After each series, con-sisting of 100 000 load cycles, the specimen was complete-ly unloaded then statically reloaded to the upper load in order to observe possible variations in the static dowel

Fig. 4. Evolution of concrete compressive strains under fatigue (left); transfer of shear forces in concrete near composite dowels [22] (right)

stiffness Cstat (Fig. 6, left). Furthermore, the cyclic compo-site stiffness, referred to as hysteresis stiffness Chys, was evaluated. For this purpose, a linear relation between slip and load was assumed for the upper and lower load within the hysteresis cycle (Fig. 6, right).

3.3 Test results3.3.1 Phenomenology of the cyclic slip evolution

In order to illustrate the fundamental questions, Fig. 7 (left) shows the measured force-slip curves (blue) for test

66



S-C-N-E6 after 0, 0.1 million and 1 million load cycles as well as the idealized cyclic dowel curves (orange). The slip consists of an elastic part del and an inelastic part din. Whereas the elastic slip recovers completely after unload-ing, the inelastic slip remains in the composite connec-tion. Furthermore, Fig. 7 shows the evolution of the slip over the dowel’s lifetime. As the number of load cycles increases, so the inelastic slip develops approximately equivalent to the cyclic concrete compression strains as shown in Fig. 4. Consequently, different phases of con-crete fatigue can be identified in the slip evolution. The first phase includes approx. 300 000 load cycles and is characterized by a notable increase in slip. The slip in-crease during this first phase can be explained by the micro-scale cracking inside the specimen and the com-paction of imperfections in the contact surfaces between the steel and the concrete dowel. The second phase of the damage process is characterized by a continuous increase in relative displacements. The concrete matrix immediate-ly in front of the steel dowels is strongly compacted, so that the concrete matrix is pulverized (Fig. 7, right). Owing to the stable growth of the micro-cracks, the con-crete powder relaxes in the surrounding concrete matrix, yielding permanent relative displacements in the com-posite connection. The third phase of concrete fatigue, which is characterized by unstable cracking (Fig. 4) and excessive slip increase, was not reached in the test.

Fig. 5. Test setup for push-out standard test (POST, top left) and single push-out test (SPOT, top right) and reinforcement in the specimens (bottom)

Fig. 6. Testing procedure for determining the static stiffness (left) and hys-teresis stiffness (right)

67



3.3.2 Impact of upper load level

The impact of the upper load is explained by comparing tests S-C-N-E2 and S-C-N-E6, which involved identical load ranges but different upper load levels. The loads were related to the connector’s calculated shear capacity (lo = Po/PRk; lu = Pu/PRk), where the shear capacity was determined by the formulas given in [6], [7]. Fig. 8 (top) shows the slip evolutions for both tests as a function of the load cycles. In test S-C-N-E6, the higher upper load (lo,E6 = 0.63) caused a significantly higher initial elastic slip. In test S-C-N-E2, with a lower upper load (lo,E2 = 0.31), the first phase of concrete fatigue was al-ready completed after approx. nE2 = 15 000 load cycles, whereas the higher upper load caused more distinctive micro-cracking, which did not finish until approx. 300 000 load cycles (S-C-N-E6). Here, the second fatigue phase was characterized by a stable slip increase, whereas in the test with the smaller upper load (S-C-N-E2), no further slip increase was observed during the second fatigue phase. Fig. 8 (centre) shows the evolution of the slip range for both experiments. The recurring peaks in the curve are due to the testing procedure and have no relevance for the concrete fatigue process described here. Obviously, the different upper loads with identical load ranges (Dl = 0.22) have no impact on the slip range. Fig. 8 (bottom) illustrates the evolution of the static stiffness (dashed line) and the continuously measured hysteresis stiffness (solid

line). Aside from small, scattered differences, the static stiffnesses as well as the hysteresis stiffnesses in both tests (S-C-N-E2, S-C-N-E6) show similar developments, such that the influence of the upper load on the stiffness can be deemed insignificant. Both stiffness curves show nearly constant stiffness values over the specimens’ entire life-times (Fig. 8, bottom).

3.3.3 Impact of load range

To investigate the impact of the load range, the results of tests S-C-N-E4 and S-C-N-E6, tested under identical upper loads (lo = 0.63) but different load ranges, are compiled in Fig. 8. The slip evolution of both tests is mostly congruent (Fig. 8, top). Fig. 8 (centre) illustrates the evolution of the slip range. As expected, for test S-C-N-E4 with a higher load range (Dl = 0.36), a higher slip range was measured than for S-C-N-E6. Fig. 8 (bottom) shows the evolution of the stiffnesses of S-C-N-E4 and S-C-N-E6 over their life-times. With an increasing number of load cycles, the stiff-ness ordinates of S-C-N-E4 and S-C-N-E6 remained almost unchanged, so that constant stiffnesses (Chys and Cstat) can be assumed over the whole lifetime. However, the compa-rison shows that an increase in the load range from Dl = 0.22 (S-C-N-E6) to 0.36 (S-C-N-E4) causes a significant decrease in the hysteresis stiffness. The static stiffness is independent of the load range, and is nearly identical for both experiments.

Fig. 7. Load-slip characteristic (cyclic dowel curve) and slip evolution of specimen S-C-N-E6; sections through concrete dowels (right)

68



3.3.4 Overview of test results

Table 1 provides an overview of all cyclic push-out stan-dard test (POST) and single-push-out test (SPOT) results. Along with the related upper (lo) and lower (lu) loads as well as the load ranges (Dl), the results of the concrete material tests (fc,cube,150 and Ecm) are given. The name of each test includes the test setup (P: POST, S: SPOT), the dowel geometry (C: CL, P: PZ) and the type of concrete used (N: normal concrete).

In addition, the table contains the composite stiff-nesses Cstat determined by regular static unloading and reloading, Chys calculated from the hysteresis loops at around 200 000 cycles and the number of load cycles Ntot when the test was stopped. All tests were conducted until a minimum load cycle number of 1 million cycles. For practical reasons, the tests subsequently had to be stop-ped because of their excessive duration. As a conse-quence, a cyclic concrete failure of the composite dowels (phase III of concrete fatigue) was not observed in any of these tests. This fact is acceptable, as the aim of the docu-mented investigations was to analyse the evolution of the inelastic slip in fatigue phases I and II covering, pre-sumably, 90 % of the connector’s lifetime. Instead of in-vestigating the explicit failure state of the composite con-nection (phase III), it was authors’ intention to develop approaches for simulating the fatigue behaviour in phases I and II in order to derive cyclic dowel curves, which may be subsequently used for the fatigue design of the com-posite girder. Determining the number of cycles to failure that the dowels could withstand was not subject of the investigations.

The test results compiled in Table 1 confirm the ob-servations presented above (sections 3.3.1–3.3.3). In the tests, the upper load significantly affected the elastic slip, and the characteristics of inelastic slip increase in the dif-ferent fatigue phases. High upper loads led to high magni-tudes of elastic slip, a delayed transition between the fa-tigue phases and an enhanced inelastic slip increase in phase II. However, the load range influenced the com-posite dowel’s stiffness within the hysteresis. Here, large load ranges led to a decrease in hysteresis stiffness (Fig. 9, left) and to a greater increase in inelastic slip, whereas static stiffness was unaffected by the load range. The

Fig. 8. Cyclic evolution of slip (top), slip range (centre) and stiffness (bottom)

Table 1. Overview of test results

Series lo lu Dl fc,cube,150 Ecm Cstat Chys NI-II Ntot [–] [–] [–] [N/mm²] [N/mm²] [kN/cm] [kN/cm] million [–] million [–]

P-C-N-E 0.65 0.28 0.37 33.7 24 150 – 9740 0.3 4.5

P-P-N-E 0.30 0.09 0.21 38.4 23 650 – 11 590 0.2 4.5

S-P-N-E1 0.25 0.05 0.20 39.7 23 800 8950 16 900 0.15 2.0

S-P-N-E2 0.28 0.08 0.20 41.7 23 800 10 650 22 600 0.02 2.0

S-C-N-E1 0.40 0.18 0.22 42.3 24 900 5450 17 300 0.25 2.0

S-C-N-E2 0.31 0.09 0.22 40.0 24 500 19 280 23 130 0.1 2.0

S-C-N-E3 0.40 0.09 0.31 43.7 25 500 5890 6200 0.1 2.0

S-C-N-E4 0.63 0.27 0.36 46.7 26 400 4400 15 100 0.1 1.8

S-C-N-E5 0.54 0.18 0.36 46.8 26 300 5440 13 600 0.25 1.0

S-C-N-E6 0.63 0.40 0.22 46.8 26 300 3760 24 050 0.3 1.6

69



upper load did not impact on the composite dowel’s stiff-nesses (Chys and Cstat) (Fig. 9, right). Furthermore, all tests confirmed that almost constant stiffnesses (Chys and Cstat) can be assumed over the dowel’s entire lifetime.

4 Engineering model for determining cyclic dowel curves4.1 General

The experimental determination of cyclic dowel curves re-quires immense technical and experimental input. There-fore, the following sections serve to present approaches utilizing simple models that allow the calculation of cyclic dowel curves at a specific number of load cycles. The cyclic dowel curve of a composite dowel after n load cycles can be determined by offsetting the static dowel curve by the inelastic slip din,nLC. An approach for calcu lat ing the in-elastic slip and a subsequent engineering model for de-termining the static stiffness of composite dowels, relevant for the static dowel curve, is developed below.

4.2 Model for cyclic increase in inelastic slip

The two-phase evolution of inelastic slip over the dowel’s lifetime depends on the upper load level and the load range. To merge the impact of both, parameter do,hys (Fig. 10, left) is calculated from the ratio of the upper load Po to the hysteresis stiffness Chys:

(1)

PCo,hys

o

hysδ =

The hysteresis stiffness required depends on the load range, as illustrated in Fig. 9, and can be calculated from the static stiffness (Eq. (2a)). At Dlstat = 0.48, static stiff-ness Cstat is assumed to be equal to hysteresis stiffness Chys (Eq. (2b) and Fig. 9, left).

Chys = Cstat + 64 000 · (Dlstat – Dl) for Dl ≤ Dlstat (2a)

Chys = Cstat for Dl ≤ Dlstat (2b)

The statistical evaluation of Eqs. (2a) and (2b) and the test results lead to a slight overestimation of the stiffness by the theoretical approach (mean value 1.05) and a coeffi-cient of variation COV of 0.21. However, the reliability of these statistical measures is severely limited owing to the very small database of only 14 tests.

By using do,hys and constants c1 and a, the slip evolu-tion can be formulated as a function of the number of load cycles n:

din(n) = c1 · do,hys · na (3)

CL shape: c1 = 0.65, a = 0.12PZ shape: c1 = 0.40, a = 0.10

Fig. 9. Influence of relative load range Dl and upper load l0 on dowel stiffness Cstat and Chys

Fig. 10. Determination of cyclic dowel curve (left); comparison between experimental and calculated slip development (right)

The values listed for constants c1 and a were determined from the experimental results given in section 3. The com-parison of the test results of a CL-shaped composite dowel with the calculated slip over the load cycles in Fig. 10 (right) illustrates the overall sufficient correlation between the model and tests.

4.3 Model for static composite dowel stiffness4.3.1 General

The static composite dowel stiffness Cstat is needed not only for understanding the evolution of inelastic slip, but also to construct the elastic branch of the cyclic dowel curve (Fig. 10, left). In the following, a numerical parame-tric study for the identification of relevant parameters for the dowel stiffness is first presented. Afterwards, a simple spring model is developed, allowing for an easy estimate of the dowel’s static stiffness through hand calculations.

4.3.2 Numerical investigations for composite dowel stiffness

The experimental investigations of local fatigue behaviour were limited to load-dependent parameters. In order to in-vestigate how geometrical and material parameters affect static dowel stiffness, finite element calculations have been performed using ABAQUS [27]. The simulation of composite dowel shear connectors has been described ex-tensively in [28]–[31]. An identical modelling approach has been implemented in the present paper, with the normal-strength concrete being modelled by the material model “Concrete Damaged Plasticity” (CDP) [32], [33]. A stress-crack opening relationship according to [34], [35] was ap-plied for concrete in tension (fctm = 2.5 N/mm²). The ap-proach by Sargin [36] served as a mathematical description of the behaviour of concrete in compression (fcc = 34.8 N/mm²). All CDP parameters (Kc = 0.66, ex = 0.1, Ψ = 35°, fb0/fc0 = 1.16) used in the calculations were chosen according to the recommended values given in [27]. Stress-strain relations with elastic-plastic material behaviour and hardening in tension and compression have been used for both the structural steel (fy/fu =

70



422/517 N/mm²) and the reinforcement components (fy/fu = 585/668 N/mm²). The contact between the steel dowel and concrete was implemented as a “Hard-Con-tact” in the normal direction, whereas frictional contact was assumed in the tangential direction. Here, a friction coefficient value of m = 0.3 has been used according to [37], where values between 0.3 and 0.5 have been recom-mended as a result of a comprehensive parametric study. The reinforcement was modelled with truss elements and completely embedded in the concrete, so that all transla-tional degrees of freedom were coupled with the surround-ing concrete.

Fig. 11 (left) illustrates the setup, support and load introduction of the FE model as well as its discretization in the area of the composite connection. A comparison between static reference tests and the FE simulation per-formed (Fig. 11, right) shows sufficient correlation, espe-cially in the load range relevant for fatigue behaviour (permissible upper load according to [5] is 70 % of charac-teristic load capacity). The poor correlation at the ulti-mate load level may either result from the chosen material parameters (default settings), or the modelling used for interactions between concrete and steel dowel or between concrete and reinforcement. Since the well-simulated elas-tic region is the focus of the research, the chosen model is used for further analysis despite these discrepancies at the ultimate load level.

The concrete strength, concrete cover, doweling re-inforcement, stirrup reinforcement and web thickness pa-rameters were varied within the numerical studies. Here, one single parameter was varied within the given limits of the National Technical Approval [6], whereas the other parameters were held constant. The stiffness was evalu-ated at 70 % of the calculated characteristic shear capaci-ty for all parameter calculations. The studies reveal that the static stiffness of the composite dowel mainly depends on the concrete strength (Young’s modulus of concrete) and the web thickness of the steel dowel, whereas all other parameters have only a minor influence [19]. The relative increase in the stiffness Cstat between concrete ty-pes C20/27 and C60/70 was + 23 % for the CL shape and

Fig. 11. FE model (left) and comparison between static reference tests and FE simulation (right)

71



+ 25 % for the PZ shape. Enlarging the web thickness tw from 10 to 21 mm led to an increase of + 20 % for the CL shape. Fig. 13 shows the results of the parametric studies along with the results of the engineering model.

4.3.3 Engineering model for static composite dowel stiffness

Based on the numerical results from the parameter study, a mechanical spring model was developed for calculating the static dowel stiffness Cstat. Here, three springs are con-nected in series in order to consider bending and shear in the steel dowel as well as compression in the concrete dowel (Fig. 12). Consequently, the static stiffness of the composite dowel Cstat can be calculated with Ct (shear in steel dowel), Cflex (bending in steel dowel) and Cc (com-pression in concrete dowel):

(4)

1 1 1–1

CC C Cstat

r flex c= + +

To determine the spring stiffness of the steel dowel under bending Cflex, the dowel is idealized as a cantilever (Fig. 12). Here, the length of the cantilever is set as the distance of the resultant dowel force from the dowel base. The position of the resultant force is an adequate approxi-mation of the point of maximum contact stress between steel and concrete dowel. In the FE simulation this was found to be hCL = 0.18 ex (clothoid shape) and hPZ = 0.067 ex (puzzle shape). To simplify the calculations, the effective second moment of area of the steel dowel Ieff is determined at the point of the smallest dowel width beff

(0.325 ex or 0.437 ex), where ex is the mutual distance be-tween adjoining dowels. Consequently, the following equations can be used:

(5)

3 · · 3 · ··

12 1.47 · ·3

3

3C

E I

h

Eb t

hE tflex,CL

s eff seff,CL w

CLs w= = =

(6)

3 · ·3

CE I

hflex,PZs eff

= ∞

If the geometry of the composite dowel is set according to [5], all dimensions are scalable over the dowel spacing ex. Eqs. (5) and (6) can then be summarized in short equati-ons depending on the Young’s modulus of steel Es and the strip thickness tw. As the distance between the dowel base and resultant dowel force is very small for the puzzle-shaped dowel, the influence of bending may be neglected (see Eq. (6)).

The spring stiffness of the steel dowel under shear forces can be determined by assuming a parabolic shear force distribution in the dowel cross-section (Eqs. (7) and (8)):

(7)

· · 2/3 · ·1.20 · ·C

G A

h

G b t

hG tr,CL

s r,CL s eff,CL w

CLs w= = =

(8)

· · 2/3 · ·4.35 · ·C

G A

h

G b t

hG tr,PZ

s r,PZ s eff,PZ w

PZs w= = =

For the time being, Eqs. (7) and (8) are limited to compos-ite dowels that are immediately burned into the web of a steel beam. Eqs. (7) and (8) do not cover the use of com-posite dowels welded to the upper flange of the beam, since an entirely homogeneous weld between dowel and flange cannot be guaranteed.

The calculation of the spring stiffness for the con-crete dowel Cc depends on the average Young’s modulus of the concrete Ecm, the effective area of the concrete spring Aeff,c and the length of the concrete spring l0. Fig. 12 shows the dimensions of the concrete spring. The spring length l0 is defined as 0.68 ex for the CL shape and 0.56 ex for the PZ shape, and the distance between dowel base and the upper rounding of the steel dowel is defined as the effective concrete dowel height (hCL = 0.31 ex and hPZ = 0.18 ex). The average width of the concrete dowel is assumed to be 0.22 ex + tw (CL shape) and 0.16 ex + tw (PZ shape). This results in the following equations:

(9)

· · (2 · 0.11 · ) · 0.31 ·0.68 ·

0.46 · · (0.22 · )0

CE A

lE e t e

eE e t

c,CLc eff,c,CL c x w x

x

c x w

= =+

= +

Cc,PZ = 0.32 · Ec · (0.16 · ex + tw) (10)

The spring model described enables a simple and quick estimation of the static dowel stiffness Cstat depending on the relevant parameters. Fig. 13 demonstrates the good correlation between the engineering model and the results of the numerical parameters.

Fig. 12. Spring model for determining static dowel stiffness (values in brackets for puzzle shape)

5 Conclusion and perspectives

Fatigue processes in composite dowels are characterized by compression of the multi-axially stressed concrete in front of the composite dowel, where shear forces are transferred between steel and concrete. Here, plastic con-crete compression strains occur, which accumulate over load cycles and lead to a cyclic increase in the relative displacements in the composite connection.

The impact of the upper load and load range for composite dowels with puzzle and clothoid shapes was in-vestigated in 14 cyclic shear tests. The upper load level had a significant impact on the elastic slip and on the charac-teristics of the inelastic slip increase during the fatigue process, but it did not influence the composite dowel’s stiffness. High upper loads caused high magnitudes of elastic slip, a delayed transition between fatigue phases I and II and a greater increase in inelastic slip in phase II. The load range influenced the composite dowel’s stiffness within the hysteresis. Here, large load ranges led to a de-crease in the hysteresis stiffness and to a greater increase in inelastic slip, whereas the static stiffness was unaffected.

For the calculative consideration of the connector’s degradation, cyclic dowel curves are needed. Engineering models for determining the increase in inelastic slip and the static dowel stiffness in phases I and II were derived in the present paper. Those models allow the easy construc-tion of cyclic dowel curves for a given number of load cycles. These cyclic dowel curves can be used subsequent-ly for the fatigue design of composite girders.

The experimental results of this paper represent a relatively small database. In order to validate the pro-posed models, further cyclic shear tests are absolutely vi-tal. Furthermore, the experimental investigations and proposed approaches in this paper focus exclusively on the composite dowel fatigue processes in phases I and II; the explicit failure state (phase III) was not considered. Consequently, the proposed approaches are neither able to determine the permissible number load cycles to failure (“lifetime prediction”) nor calculate the remaining shear capacity after a given number of load cycles. Therefore, future shear tests should not be stopped at a specific num-ber of cycles, instead need to be continued until explicit fatigue failure of the connector occurs. Such investigati-ons have been documented in [38] for composite dowels under cyclic pull-out loading, and a corresponding model

72



for lifetime prediction has been proposed. In future, this model should be transferred to shear-carrying behaviour.

References

1. Gattesco, N., Giuriani, E., Gubana, A.: Low-cycle fatigue test on stud shear connectors. Journal of Structural Enginee-ring, vol. 123 (1997), No. 2, pp. 145–150.

2. Hanswille, G., Porsch, M., Üstündag, C.: Neue Untersu-chungen zum Ermüdungsverhalten von Kopfbolzendübeln. Stahlbau, vol. 75 (2006), No. 4, pp. 303–316.

3. Verissimo, G., Valente, M., Paes, J., Cruz, P., Fakury, R.: Design and experimental analysis of a new shear connector for steel and concrete composite structures. 3rd Intl. Conf. on Bridge Maintenance, Safety & Management, 2006.

4. Iwasaki, H., Fujii, K., Fukada, K., Toyota, T., Nakamura, H.: A consideration on slip test methods for Perfobond shear connector focusing on concrete confinements. 1st Intl. Conf. on Advances in Experimental Structural Engineering, 2005, pp. 871–876.

5. EN 1994-1-1 (2010): Eurocode 4: Design of composite steel and concrete structures – Part 1-1: General rules and rules for buildings, Dec 2010.

6. Deutsches Institut für Bautechnik: Allgemeine bauaufsicht-liche Zulassung der Verbunddübelleiste, Zulassung-Nr. Z-26.4-56, Berlin, 13 May 2013.

7. Gündel, M., Kopp, M., Feldmann, M., Gallwoszus, J., Heg-ger, J., Seidl, G.: Design of composite dowels according to the new national technical approval. Stahlbau, vol. 83 (2014), No. 2, pp. 112–121.

8. Heinemeyer, S., Gallwoszus, J., Hegger, J.: Verbundträger mit Puzzleleisten und hochfesten Werkstoffen. Stahlbau, vol. 81 (2012), No. 8, pp. 595–603.

9. Hegger, J., et al.: Multifunctional composite slab system with integrated building services – Studies on the load bearing and fire behavior, thermal efficiency and sustainability of a novel composite floor system. Stahlbau, vol. 83 (2014), No. 7, pp. 452–460.

10. Dressen, T., Classen, M.: Deformation of reinforced and prestressed concrete beams with large web openings. Beton- und Stahlbetonbau, vol. 108 (2013), No. 7, pp. 462–474.

11. Classen, M., Dressen, T.: Experimental investigations on prestressed concrete beams with openings. ACI Structural Journal, vol. 112 (2015), pp. 221–232.

12. Dressen, T., Classen, M.: Experimentelle Untersuchungen an Spannbetonträgern mit großen Stegöffnungen. Bauingeni-eur, vol. 89 (2014), No. 9, pp. 359–369.

13. Classen, M., Gallwoszus, J., Hegger, J.: Load-bearing behav-ior of an integrated composite floor system. Bauingenieur, vol. 89 (2014), No. 3, pp. 91–101.

Fig. 13. Influence of dowel thickness (left), average Young’s modulus of concrete (centre) and dowel spacing (right) on the static stiffness

73



14. Hechler, O., Berthellemy, J., Lorenc, W., Seidl, G., Viefhues, E.: Continuous shear connectors in bridge construction, ASCE Conf. Proc., vol. 396, No. 7, 2008, DOI:10.1061/41142.

15. Feldmann, M., et al.: Eisenbahnüberführung Simmerbach. Stahlbau, vol. 81 (2012), No. 10, pp. 737–747.

16. Schmidt, V., Seidl, G., Hever, M., Zapfe, C.: Verbundbrücke Pöcking – Innovative VFT-Träger mit Betondübeln. Stahl-bau, vol. 73 (2004), No. 6, pp. 387–393.

17. Lorenc, W., Kozuch, M., Rowinski, S.: The behaviour of puzzle-shaped composite dowels – Part I: Experimental stu-dy. Journal of Constructional Steel Research, vol. 101 (2014), pp. 482–499.

18. Dudzinski, W., Pekalski, G., Harnatkiewicz, P., Kopczynski, A., Lorenc, W., Kozuch, M., Rowinski, S.: Study on Fatigue Cracks in Steel-Concrete Shear Connection with Composite Dowels. Archives of Civil and Mechanical Engineering, vol. 11 (2011), No. 4, pp. 839–858.

19. Gallwoszus, J.: Zur Ermüdung von Verbundkonstruktionen mit Verbunddübelleisten. Dissertation, Institut für Massiv-bau, RWTH Aachen, 2014.

20. Lohaus, L., Oneschkow, N., Wefer, M.: Design model for the fatigue behaviour of normal-strength, high-strength and ultra-high-strength concrete. Structural Concrete, vol. 13 (2012), No. 3, pp. 182–192.

21. Holmen, J. O.: Fatigue of concrete by constant and variable amplitude loading. Bulletin No. 79-1, Division of Concrete Structures, NTH-Trondheim, 1979.

22. Wurzer, O.: Zur Tragfähigkeit von Betondübel. Dissertation, Institut für Konstruktiven Ingenieurbau, Universität der Bundeswehr, Munich, 1997.

23. Takhar, S. S., Jordaan, I. J., Gamble, B. R.: Fatigue of con-crete under lateral confining pressure. Abeles Symposium on Fatigue of Concrete, ACI Pub. SP, vol. 41, 1974, pp. 59–69.

24. Hooi, T. T.: Effects of passive confinement on fatigue proper-ties of concrete. Magazine of Concrete Research, vol. 52, No. 1, 2000, pp. 7–15.

25. Bäumel, A.: Experimentelle und numerische Untersuchung der Schwingfestigkeit randschichtverfestigter eigenspan-nungsbehafteter Bauteile. Dissertation, Technische Hoch-schule Darmstadt, 1991.

26. Gallwoszus, J., Classen, M.: Resistance of composite dowels to fatigue. Eurosteel 2014, Naples.

27. Hibitt, Karlsson, Sorensen, Pawtucket: ABAQUS Online Documentation. Version 6.11, Dassault Systèmes, SIMULIA, 2011.

28. Fink, J., Ondris, L.: New Shear Connectors for Composite Girders – Experiences with ABAQUS Push-Out Test Simula-tions. Computational Civil Engineering 2007, Intl. Symp., Romania.

29. Classen, M., Gallwoszus, J., Hegger, J.: Influence of transver-sal cracking on the shear load bearing behavior of composite dowels in slender concrete chords. Beton- und Stahlbeton-bau, vol. 109 (2014), No. 12, pp. 882–894.

30. Classen, M., Hebrand, M.: Shear behavior of composite do-wels in transversely cracked concrete. Structural Concrete, vol. 16 (2015), DOI: 10.1002/suco.201400100.

31. Classen, M., Hegger, J.: Anchoring behavior of composite dowels in slender concrete chords. Bautechnik, vol. 91 (2014), No. 12, pp. 869–883.

32. Lee, J., Fenves, G.: Plastic-Damage Model for Cyclic Loading of Concrete Structures. Journal of Engineering Mechanics, vol. 124, 1998, pp. 892–900.

33. Lubliner, J., et al.: A Plastic-Damage Model for Concrete. International Journal of Solids and Structures, vol. 25, 1989, pp. 299–326.

34. Hillerborg, A., Modéer, M., Peterson, P.-E.: Analysis of Crack Formation and Growth in Concrete by Means of Frac-ture Mechanics and Finite Elements. Cement and Concrete Research, vol. 6, No. 6, 1976, pp. 773–781.

35. Feenstra, P., de Borst, R.: A composite Plasticity Model for Concrete, International Journal of Solids and Structures, vol. 33, 1996, pp. 707–730.

36. Sargin, M.: Stress-Strain Relationships for Concrete and the Analysis of Structural Concrete Sections. Solid Mechanics Division, University of Waterloo, Canada, 1971.

37. Fink, J., Petraschek, T., Ondris, L.: Push-Out Test Parametric Simulation Study of a New Sheet-Type Shear Connector. Vienna: projects to central application servers, Reports 2006, Central computer science (ZID) of Technical Univer-sity of Vienna, 2007, http://www.zid.tuwien.ac.at/projekte.

38. Gallwoszus, J., Classen, M.: Fatigue of composite dowels in UHPC under cyclic pull out loading. Bautechnik, vol. 92, No. 7, 2015, pp. 509–521.

Dipl.-Ing. Martin ClaßenRWTH AachenInstitute of Structural ConcreteMies-van-der-Rohe-Str. 152074 [email protected]

Dipl.-Ing. Joerg GallwoszusRWTH AachenInstitute of Structural ConcreteMies-van-der-Rohe-Str. 152074 [email protected]


Technical Paper

DOI: 10.1002/suco.201400062

This paper investigates the degradation of the mechanical prop-erties of concretes made with three types of aggregate affected by alkali-silica reaction (ASR). Three standard testing methods – ASTM C289, JASS 5N T-603 and ASTM C1260 – were used to identify the reactivity of ASR of the three aggregates selected. The test results show that all three aggregates are potentially deleterious. A new acceleration method based on JASS 5N T-603 and ASTM C1260 was proposed for concrete specimens. In the acceleration method, cylindrical concrete specimens were used, additional alkali material was added to the concrete mixture and the specimens were stored under conditions similar to ASTM C1260. The preconditioned concrete specimens were then used for evaluating the mechanical properties of the ASR-affected concrete in terms of strength and stiffness. The test results show that special attention must be paid to the effects of two opposing mechanisms on the strength and stiffness of concrete: hydration reactions and ASR. Hydration reactions enhance the mechani-cal properties, whereas ASR does the opposite. The changes in length of concrete specimens were also measured, which showed that the basic trends for change in length and mechani-cal properties may be different. It is better to examine the effect of ASR on both change in length and mechanical properties. The size and reactivity of the aggregate are very important factors for the mechanical properties of ASR-affected concretes. Within the two-month testing period, the reactive fine aggregate might cause ASR expansion and the reactive coarse aggregates might not.

Keywords: deterioration, ASR, reactive aggregate, aggregate size, concrete

1 Introduction

Solid concrete structures exposed to severe environmen-tal conditions, e.g. dams, pavements, coastal concrete structures, are vulnerable to attacks by various environ-mental loadings such as temperature fluctuation, moisture variation and aggressive chemicals. Alkali-silica reaction (ASR) of concrete is one of the chemical reactions in concrete that can cause severe damage. In a high-moisture

environment, the product of ASR is expansive, which is detrimental to concrete structures [1, 2].

ASR is a chemical reaction between the reactive silica in the aggregate and the alkalis (Na2O and K2O) in Portland cement. This chemical reaction produces alkali-silica gel swelling with the absorption of the moisture from the surrounding cement paste. The expansive gel can cause cracking in the concrete. Therefore, the neces-sary conditions for the expansive ASR gel to form in the concrete are a sufficiently high alkali concentration in the cement, high moisture content in the concrete and reac-tive aggregates.

In order to control or prevent the occurrence of ASR in concrete, most countries have adopted a limit to the equivalent percentage of alkalis (such as Na2O) in Portland cement. A commonly used limit is 0.6 %, below which the cement is considered to be low-alkali cement. Low-alkali cement has been considered relatively safe with respect to ASR damage. Recently, however, it has become evident that the critical value of 0.6 % equivalent alkalis was not exactly sufficient to prevent ASR damage. For example, concrete structures in the USA developed deleterious ASR where alkali levels in the cements were as low as 0.45–0.50 % equivalent Na2O [3]. This means that the equivalent critical alkali level is not always ap-plicable. This is because alkalis do not come exclusively from Portland cement; they may come from other sources such as mineral admixtures, chemical admixtures and ag-gregate.

It is not only the mineralogical composition of reac-tive aggregate that is important; the size of the aggregate is also important. The reactive aggregate size corresponding to the largest ASR expansion is called the pessimum size. To investigate the effect of the pessimum size of aggre-gates, the accelerated mortar bar test (ASTM C1260) was used with mortar bars made with waste glass particles of various sizes [4, 5, 6, 7]. The test results show that a de-crease in particle size causes an increase in volume expan-sion and damage due to ASR. However, when the particle size decreased to a certain level, the fine aggregates did not result in any excessive ASR expansion. This means that the aggregate size plays a key role in expansion due to ASR [8, 9].

In order to predict the ASR volume expansion math-ematically when considering the pessimum size effect of reactive aggregate, a theoretical model based on the modi-

The effects of alkali-silica reaction on the mechanical properties of concretes with three different types of reactive aggregate

Okpin Na*Yunping XiEdward OuVictor E. Saouma

DOI: 10.1002/suco.201400062


Submitted for review: 28 July 2014; revision: 13 July 2015; accepted for publication: 19 July 2015. Discussion on this paper must be submitted within two months of the print publication. The discussion will then be published in print, along with the authors’ closure, if any, approximately nine months after the print publication.

75

O. Na/Y. Xi/E. Ou/V. E. Saouma · The effects of alkali-silica reaction on the mechanical properties of concretes with three different types of reactive aggregate


ferent sources: CS, GJ and WM. The CS aggregate came from the Colorado Springs region, Colorado. The rock type for the CS aggregate was granite, one of the igneous rocks. The GJ aggregate was acquired from the Grand Junction region, Colorado. It was a basalt or andesite, a type of volcanic rock. The WM aggregate came from the state of Wyoming and its geological information was not released by the supplier. Fig. 1 indicates the grading curvse for the three types of coarse aggregate and the fine aggregate used in the present study.

2.2 Three standard testing methods and specimen preparation

The expansive ASR gel that forms in concrete ultimately contributes to the ASR expansion and the formation of internal microcracks in the concrete. In order to inves-tigate the ASR expansion effectively and determine the experimental parameters of ASR-affected concretes, three standard testing methods – JASS 5N T-603, ASTM C1260 and ASTM C289 – were used to examine the reactivity of the three aggregates selected. The specimen preparation and testing methods are different for the three methods. They are outlined here for the reader’s convenience.

The JASS 5N T-603 testing method is for measur-ing the change in length of concrete specimens affected by ASR. The Portland cement used for this study was an ASTM type I/II cement with a high alkali content of 0.79 % as shown in Table 1. This testing method requires that granular sodium hydroxide of 98 % fineness be added to the concrete mixes. Adding the high concentration of alkali material to the concrete mix accelerates the ASR. Three different amounts of NaOH were used. The suf-fixes 12, 18 and 24 for the concrete specimens indicate the equivalent weight of NaOH 1.2, 1.8 and 2.4 kg/m3 respectively. Concrete admixtures were not used because the focus of this study was to investigate the effect of ASR on concrete made with reactive aggregates without admix-tures. The mix design is shown in Table 2 and the concrete mixing procedure is specified in ASTM C 192. Table 3 lists the properties of the fresh concrete mixes. The prism moulds used for making specimens were 2 inch (50 mm) wide × 2 inch high × 10 inch (100 mm) long. Length mea-surement was conducted using an SPI digital indicator.

fied version of the generalized self-consistent theory was developed [10]. The model can predict the pessimum size of ASR in mortar bars and was verified with the test data of mortar bars obtained using ASTM C1260.

Many accelerated experimental approaches have been developed to capture the ASR-induced damage in a short period of time. ASTM C1260 and C1293 are meth-ods commonly used in the USA. The National Aggregates Association (NAA) screened aggregates for ASR suscep-tibility and adopted the following critical values for ASR expansion: 0.2 % for ASTM C1260 and 0.04 % for ASTM C1293 [11].

This paper focuses mainly on evaluating the degra-dation of the mechanical properties of concrete due to ASR. First, most standard testing methods were developed for mortar bars and aggregate samples, and very few for concrete. It is important to study the effect of ASR on concrete because concrete is the structural material used in construction. Thus, in addition to mortar bars, con-crete specimens were made with three different types of aggregate for testing the effect of ASR. Second, most of standard testing methods evaluate the change in length (linear expansion) of mortar bars for ASR damage; very few evaluate the mechanical properties of concrete. It is important to study the effect of ASR on the mechani-cal properties of concrete because the performance of concrete structures depends directly on the mechanical properties of concrete. Therefore, in addition to measur-ing change in length for standard testing methods, the mechanical properties of ASR-affected concrete, such as strength and stiffness, were systematically studied.

In order to make sure that the three aggregates se-lected were ASR-reactive, standard testing methods were used to test the reactivity of the three aggregates. There are many standard testing methods available for ASR, and three of them were used in this study: JASS 5N T-603, ASTM C1260 and ASTM C289. In order to accelerate the ASR process in the concrete samples, a new accelera-tion method was proposed to precondition the concrete specimens. The new method was based on ASTM C1260 and JASS 5N T-603. The concrete samples made with the three reactive aggregates were preconditioned using the new acceleration method, and the mechanical properties of the ASR-affected concretes were tested and analysed in terms of strength and stiffness. Change in length was also measured. We would like to point out that the pre-sent work is an attempt to study the effect of ASR on the mechanical properties of concrete, the new acceleration method was proposed to facilitate the experimental study and more work needs to be done in order to develop a new standard testing method for ASR-affected concrete.

2 Experimental plan2.1 Properties of coarse and fine aggregates

Various kinds of aggregate are found in the USA. Basalt and similar igneous rocks such as andesite and shoshonite are scattered throughout the state of Colorado. Granite, one of the most commonly used igneous rocks in the USA, is widespread from Colorado Springs to Denver and the Colorado mountain region [12, 13]. The three coarse aggregates used in this study were acquired from three dif-

Fig. 1. Grading curves for coarse and fine aggregates

76



× 279.4 mm) and at least three specimens were made for each mix design using one type of Portland cement with a specific alkali content. After demoulding, the mortar bar specimens were cured in a fog room for 24 h (temperature 23 °C, relative humidity 100 %). Immediately after curing, the specimens were placed in a container with 1 N NaOH solution at 80 °C for 14 days. The changes in length were recorded with the same equipment as used in the JASS 5N T-608 test. For the two weeks of the testing period, the change in length of every specimen was measured once a day.

The ASTM C289 testing method is for testing the chemical reactivity of aggregates. No Portland cement was used in this test. The specimens prepared from the three different aggregates were placed in 1 M NaOH so-lution at 80 °C for 24 h. A chemical method was used to determine the potential reactivity of the aggregate.

3 Test results and discussion of the three standard testing methods

Fig. 3 shows the ASTM C289 test results, which provides quantitative information about deleterious aggregates, not numeric values of the potential ASR of aggregates. The Sc and Rc results for each aggregate are shown in Fig. 3 to compare with the calibration curves. In Fig. 3, the blue dot is for the results of CS aggregate, the red dot for GJ aggregate and the black dot for WM aggregate. It can be seen clearly from Fig. 3 that CS and WM aggregates can be considered deleterious, whereas GJ aggregate is poten-tially deleterious. Therefore, all three aggregates used in this study are potentially deleterious.

Fig. 4 shows the change in length of concrete speci-mens measured in accordance with JASS 5N T-603. The ASR expansion of the specimens was recorded for six months. All alkali contents in concrete specimens were much higher than 0.6 %, the critical alkali content for low-alkali Portland cement. This was because, as described earlier, large amounts of additional NaOH were added to the concrete mixes. The scattering of the test data is quite large in Fig. 4a; the effect of alkali content on ASR expan-

Its measuring range is 0.0–1.0 inch (0.0–25.4 mm) and its resolution is 0.00005 inch (0.00127 mm). The testing period was six months and the length variations were measured once a month.

The ASTM C1260 testing method is for measuring the change in length of mortar bars. The quantities of materials for making mortar bar specimens are 440 g of cement and 990 g of fine aggregate. Therefore, coarse aggregates were not used in this test. The testing period was 14 days. The fine aggregates were produced by crush-ing coarse aggregates as shown in Fig. 2. The portions of fine aggregates with different particle sizes are prescribed in Table 4 and the required water-cement ratio is 0.47 by mass. Different Portland cements of type I/II were used. They have different alkali contents of 0.59, 0.79 and 0.9 %. The mortar mixing procedure was Practice C 305. The specimen size was 1 × 1 × 11 inch (25.4 × 25.4

Table 1. Chemical composition (percent by mass)

CaO SiO2 Al2O3 Fe2O3 MgO Na2O K2O Na2Oeq SO3 LOI

% 62.5 20.1 5.6 2.0 3.1 0.2 0.9 0.79 3.2 1.7

LOI: loss on ignition.

Table 2. Mix design for JASS 5N T-603

Water Cement Fine aggregate Coarse aggregate kg/m3 kg/m3 kg/m3 kg/m3

165 330 754 1041

Table 3. Properties of fresh concrete mixes

Type of aggregate Air content (%) Slump (mm)

CS 4.5 76.2

GJ 4.2 69.9

WM 4.3 73.0

Fig. 2. Crushed fine aggregates

Table 4. Portions retained on sieves

Sieve size Mass

Passing Retained on % g

4.75 mm (No. 4) 2.36 mm (No. 8) 10 99.0

2.36 mm (No. 8) 1.18 mm (No. 16) 25 247.5

1.18 mm (No. 16) 600 μm (No. 30) 25 247.5

600 μm (No. 30) 300 μm (No. 50) 25 247.5

300 μm (No. 50) 150 μm (No. 100) 15 148.5

77



0.5591 %. The other test results in Figs. 5b and 5c also followed the same trend. It is important to point out that the increment in the alkali content of 34 % (from 0.59 to 0.79 %) resulted in a much higher percentage of ASR expansion – up to 230 %. However, when the equivalent alkali content reached 0.9 %, the ASR expansions did not increase significantly over those for 0.79 %. So it seemed that, at a fixed time, the ASR expansion increases with in-creasing alkali content in the cement, but only to a certain value, say 0.8 to 0.9 %, thereafter the ASR expansion does not increase with increasing alkali content. These results are limited to the short testing period used in this study. Apparently, more research should be conducted on this topic. Fig. 5d presents a summary of the final expansions of concrete specimens with different alkali contents and different types of aggregate. It is clear from the figure that GJ aggregate is much more reactive in terms of ASR than the other two aggregates.

4 Acceleration method based on ASTM C1260 and JASS 5N T-603 for concrete specimens

As indicated earlier, the JASS 5N T-603 testing method is for concrete specimens, but, it takes 180 days to obtain

sion is not clearly shown. However, the trend shown in the test data of JASS 5N T-603 is very consistent. The concrete specimens with GJ aggregate obviously swelled with additional alkali, whereas CS and WM aggregates did not provide clear evidence as to whether the aggre-gates are reactive or not. In this testing method, the criti-cal ASR expansion of concrete specimens is considered to be 0.1 % over the six months. As a comparison, the final ASR expansion of each concrete specimen is marked with dots in Fig. 4b. The critical amount of additional alkali in concrete may be predicted based on the test data shown in Fig. 4b. It can be seen that the critical alkali contents of GJ and WM aggregates can be estimated as 3.8 and 4.3 kg/m3 respectively, corresponding to the 0.1 % expan-sion.

Fig. 5 illustrates the expansions of mortar bars mea-sured according to ASTM C1260. It can be clearly seen that the alkali contents (0.59, 0.79 and 0.9 %) in the Port-land cement had important effects on the ASR expansion of the mortar bars. In Fig. 5a, the ASR expansion data of GJ specimens with the 0.59 % alkali content (slightly lower than the critical alkali content) are between 0.1 and 0.2 %. However, in Figs. 5b and 5c, the expansions of CS and WM specimens with 0.59 % alkali content is below 0.1 %. This means that GJ aggregate is more reactive than CS and WM aggregate. Nevertheless, with increasing al-kali content in mortar bar specimens, the ASR expansion increased in all specimens significantly, which indicates that alkali content in Portland cement is an extremely important factor. High alkali content can activate ASR ex-pansion of aggregate which may be not reactive with low-alkali cement. In Fig. 5a, GJ aggregates with the cement with 0.59 % alkali content resulted in 0.1667 % expansion during the testing period, while the same aggregates with the alkali contents of 0.79% and 0.9% expanded up to

Fig. 3. ASTM C289 test data and the division between innocuous and del-eterious aggregates

Fig. 4. ASR expansion test data and critical amount of NaOH in JASS 5N T-603: a) change in length test data for six months, and b) critical alkali amount for ASR expansion in concrete made with three different aggre-gates

a)

b)

78



of aggregate were prepared and cured under the standard conditions for 28 days (23 °C and 100 % relative humid-ity), then submerged in an alkaline solution (same concen-tration and same temperature as ASTM C1260) for about two months.

The purpose of this new acceleration method is to precondition the concrete specimens so that the mechani-cal properties such as strength and stiffness of ASR-affect-ed concrete can be tested. To examine the variation in strength and stiffness of the concrete specimens, compres-sion tests were carried out after 2, 4, 6 and 8 weeks.

5 Test results and discussion of ASR-affected concrete preconditioned by new acceleration method

5.1 Strength and stiffness of ASR-affected concrete

The compressive strength was obtained from the maxi-mum stress in a stress-strain curve. The stress-strain curves of the ASR-affected concrete specimens were tested at dif-ferent times (different acceleration periods). Fig. 6 shows the test results of the mechanical properties of the ASR-affected concrete specimens in terms of the immersion times and reactive aggregates. Fig. 7 shows the strains corresponding to the peak stresses at different immersion times. During the first four weeks of the testing period, the concrete strength gradually increased and then decreased,

results; the ASTM C1260 takes only 14 days, but it is for mortar bars not concrete specimens; and ASTM C289 is for aggregate not concrete. Based on the test data of the three standard testing methods, an acceleration method was proposed in this study to accelerate ASR in concrete samples so that the deterioration of mechanical proper-ties of ASR-affected concrete can be tested. Several im-portant parameters for accelerating ASR, such as adding NaOH to concrete mixes, using high-alkali cement and extending the testing period, were discussed in previous sections. Moreover, using a high testing temperature is another method commonly used to accelerate a chemi-cal reaction in a material. Among the three standard test methods, the testing conditions used in ASTM C1260 were adopted and modified to accelerate ASR in concrete specimens. In order to capture clearly the effect of higher alkali content on ASR expansion in concrete, Portland cement with a high equivalent alkali content was used as shown in Table 1. In addition, granular sodium hydroxide (NaOH) amounting to 5.6 kg/m3 was added during the mixing process in a similar way to JASS 5N T-603. The quantity of additional alkali is increased up to 50 % of the critical alkali amount used for GJ aggregate as shown in Fig. 4b. Cylindrical concrete specimens were used with a size of 3 inch (76.2 mm) diameter × 6 inch (152.4 mm) high. Concrete specimens made with the same three types

Fig. 5. Change in length of reactive aggregates according to alkali content (equivalent alkali levels of 0.59, 0.79 and 0.90 %): a) GJ aggregate, b) CS aggre-gate, c) WM aggregate, and d) final ASR expansion for different alkali contents

a)

c)

b)

d)

79



used in this study. This procedure is described in ASTM C469 and the equation to calculate the modulus of elastic-ity using test data is:

––

2 1

2 1E

f fchord ε ε

=

where e1 = 50 microstrain, f1 is the stress corresponding to e1, f2 = 40 % of ultimate stress fc and e2 is the longitudinal strain at f2.

The trend for stiffness is similar to that for the strength. In the first four weeks, the slightly increased stiffness can be explained by the dominant hydration reac-tions, whereas the decreased stiffness after the first four weeks indicates that the detrimental effect of ASR became more significant. These results agree with other test results available in the literature [2,14,15].

5.2 External appearance of ASR-affected concrete

Fig. 9 shows the surface cracks and colour changes of the concrete specimens after being submerged in NaOH solu-tion for 56 days. It can be clearly seen that surface cracks occurred due to ASR expansions of coarse aggregate beneath the concrete surface. Moreover, coloured spots on the surface of the concrete specimens can be seen in Fig. 9: dark grey spots on GJ aggregate and yellow ones on CS and WM aggregates. Those are evidence of progressive ASR.

5.3 Interactive effect of hydration reactions and ASR

The mechanical properties of concrete are affected not only by the ASR-induced damage, but also by the hydra-tion reactions of the Portland cement. The former causes the reduction in the mechanical properties of concrete and the latter enhance them. In order to distinguish the interaction between the two opposing mechanisms, addi-tional tests were designed and conducted using the same GJ concrete specimens submerged in hot water (not in hot alkali solution). In this case the hydration reactions were the dominant effect and the ASR was kept to a minimum because there was no alkali solution involved. The addi-

although the total variation in compressive strength is not large for different immersion times. On the other hand, the strain at peak stress suddenly decreased and then increased continuously. The variations in the stress and strain of the ASR-affected concrete specimens may be explained by two mechanisms that occurred in the con-crete at the same time. One is the effect of the hydration reactions of Portland cement which contributed to the in-crease in compressive strength and stiffness. The other is the effect of ASR that led to the reduction in the mechani-cal properties. The compressive strength and stiffness of the concrete affected by ASR were under the combined effect of the two opposing mechanisms. For the strains corresponding to the peak stresses, the variation shown in Fig. 7 can also be explained by the two opposing mecha-nisms: at the beginning of the testing period, the effect of hydration reactions dominated and thus the strains de-creased (stiffness increased); thereafter, the effect of ASR became stronger, which resulted in the increased strains in the concrete (and thus decreased stiffness).

Fig. 8 shows the stiffness of concrete affected by ASR. The stiffness can be characterized by the modulus of elasticity of the concrete. To determine the modulus of elasticity experimentally, the chord method (Echord) was

Fig. 6. Compressive strength vs. immersion time

Fig. 7. Peak strain vs. immersion time at peak stress

Fig. 8. Elastic moduli vs. immersion time

80



After four weeks, the strength of the concrete specimens in hot water continuously increased up to 10 %, while that of the GJ specimens in alkali solution decreased because of ASR. The results of normalized stiffness in Fig. 11 further demonstrate this phenomenon: the normalized stiffness of the specimens gradually increased up to about 7 %, while those of GJ aggregates rose to 24 % for the first month because of the contribution of hydration reactions and then dropped by about 30 % due to ASR.

From the test results it seems that tri- and di-calcium silicate minerals (C3S and C2S) in the cement react with water in the alkali solution and produce calcium-silicate hydrate (C-S-H) gel and calcium hydroxide (Ca(OH)2). During the first month, C3S exerts an effect on the early strength of concrete up to about 80 %. After that, C2S plays the leading role in developing the strength of concrete as shown in Fig. 12. At the same time, reactive aggregates are in a hydroxyl-rich medium. By the second month, the ef-fect of ASR becomes more significant. The high tempera-ture of 80 °C accelerates the ASR, a swelling gel formed from ASR increases in volume with imbibing water. Even though the hydration reactions progress continuously, the ASR-induced damage can exhibit a more dominant effect on the mechanical properties of the concretes.

tional test results of normalized strength and stiffness are shown in Figs. 10 and 11 respectively, where “GJ” repre-sents the specimen made with GJ aggregate submerged in alkali solution and “Additional” the specimen made with GJ aggregate submerged in hot water.

In Fig. 10, for the first four weeks, both strength curves increased due to the effect of hydration reactions.

Fig. 10. Normalized stress of additional specimen to observe hydration effect

Fig. 11. Elastic moduli of additional specimen to observe hydration effect

Fig. 9. Surface cracks in three different specimens after 56 days in NaOH solution

(a) GJ (b) CS (c) WM

Fig. 12. Compressive strength development in paste of pure cement com-pounds [17]

81



ties of the sand particle sizes specified in ASTM C1260. Fig. 14 illustrates the test results for the change in length of concrete specimens; concrete specimens with reactive fine aggregates reacted quickly with the sodium hydroxide solution and expanded within the 14-day testing period. In the case of non-reactive fine aggregate, however, ASR expansion did not occur for the first two weeks under the same testing conditions. So the effect of aggregate size on ASR rate is very significant. Therefore, the long-term du-rability performance of concrete structures under severe environmental conditions such as high humidity, high temperature and extremely aggressive chemicals heavily depends on the reactivity and size of aggregates.

6 Conclusions

In order to examine the deterioration of the mechanical properties of concrete affected by ASR, three different types of aggregate (CS, GJ and WM) were selected and used in the present study. Three standard testing methods (ASTM C289, JASS 5N T-603 and ASTM C1260) were employed to examine the ASR reactivity of the aggre-

5.4 Change in length of concrete specimens

Fig. 13 shows the change in length of concrete cylindrical specimens under the new acceleration method. The test results indicate that the length of specimens decreased gradually over time, which is different from the continu-ous expansion in the test data of ASTM C1260 for mortar bars. Thus, the change in length results are important. The trend for the mechanical properties of ASR-affected con-crete and the trend for change in length may be different. The change in length might not exhibit a severe problem, but that does not mean there is no problem for the me-chanical properties. Hence, there is a need to examine the effect of ASR on both change in length and mechanical properties. There is also a need to find the reasons for the difference. One possible reason is the dimensional effect of specimens.

The aspect ratios of specimens used in the JASS 5N T-603 and ASTM C1260 methods are 1:5 and 1:11 respectively. The aspect ratio of the cylindrical concrete specimen was 1:2. The expansion of the long mortar bar used in ASTM C1260 can be considered as uniform over the cross-section because the alkali solution can penetrate from the surface to the centre of a mortar bar (distance = 0.5 inch [12.7 mm]) quickly, whereas the ASR expansion of the cylindrical specimen over the cross-section is not uniform since it takes a longer time for the alkali solu-tion to reach the centre of a cylinder (distance = 1.5 inch [38.1 mm]) and thus there is an alkali concentration dis-tribution in the radial direction of the cylinder. The alkali content at the edge of a cylinder is higher than that at the centre. The non-uniform alkali concentration over the cross-section can lead to non-uniform ASR expansion, which is more complicated than the 1D expansion of the thin mortar bars. The measurement of change in length was conducted at the centre of the cylinder. This topic should be studied in more depth in the future.

ASR expansion will eventually occur in the concrete specimen, but it did not occur during the testing period, which is due to the fact that only the coarse aggregate used in the concrete specimens was reactive aggregate, and the fine aggregate was not. A large particle size for a coarse aggregate corresponds to a small surface area, which determines the low ASR rate.

5.5 Effect of fine aggregates

As shown in the above test results, the strength and stiff-ness of ASR-affected concrete did not decrease during the first month or two [2], whereas the ASR expansion of mortar bars in ASTM C1260 can reach a significant level in 14 days. The difference is caused by the size of reactive aggregate [8, 9, 17], i.e. the larger the size of reactive aggre-gates, the slower the rate of ASR reactivity [18].

The concrete specimens used in the above tests were made with the three types of reactive coarse aggregate and commercial sand, which is not reactive. In order to capture the effect of fine aggregate, additional specimens were prepared. There were two additional testing groups: non-reactive fine aggregates and reactive fine aggregates. The reactive fine aggregates were prepared by crushing the reactive coarse aggregates in accordance with the quanti-

Fig. 13. Change in length of GJ concrete specimens affected by ASR

Fig. 14. Effect of early ASR expansion according to reactivity of fine aggregate

82



5) Aggregate size is a very important factor for the dete-rioration of the mechanical properties of ASR-affected concrete. Tests were conducted to show the effect of re-active fine aggregates on the mechanical properties of ASR-affected concrete. The concrete specimens made with both reactive fine and coarse aggregates reacted quickly with the alkali solution and expanded faster than the concretes made with non-reactive fine and reactive coarse aggregates. Therefore, the deterioration behaviour of concrete structures under the effect of ASR depends on the size distribution of aggregates.

References

1. Basheer, P. A. M., Chidiact, S. E., Long, A. E.: Predictive models for deterioration of concrete structures. Construc-tion and Building Materials, 1996, vol. 10, pp. 27–37.

2. Giaccio, G., Zerbino, R., Ponce, J. M., Batic, O. R.: Mechani-cal behavior of concretes damaged by alkali-silica reaction. Cement and Concrete Research, 2008, vol. 38, pp. 993–1004.

3. Stark, D.: Alkali-Silica Reaction and its Effects on Concrete. USCOLD 2nd Intl. Conf. on Alkali-Silica Reaction, Chatta-nooga, TN, 1995.

4. Meyer, C., Xi, Y.: Use of Recycled Glass and Fly Ash for Pre-cast Concrete. Journal of Materials in Civil Engineering, ASCE, 1999, vol. 11, pp. 89–90.

5. Jin, W., Suwito, A., Meyer, C., Xi, Y.: Theoretical Modeling on Expansion and Damage due to Alkali-Silica Reaction. Proc. of 12th Engineering Mechanics Conf.: Engineering Mechanics: A Force for the 21st Century, San Diego, CA, 1998, pp. 1175–1178.

6. Al-Akhras, N. M.: Performance of olive waste ash concrete exposed to alkali-silica reaction. Structural concrete, 2012, vol. 13(4), pp. 221–226.

7. Matos, A. M., Sousa-Coutinho, J.: ASR and sulphate perfor-mance of mortar containing industrial waste. Structural Concrete, accepted, 2015.

8. Zhang, C., Wang, A., Tang, M., Wu, B., Zhang, N.: Influence of aggregate size and aggregate size grading on ASR expan-sion. Cement and Concrete Research, 1999, vol. 29, pp. 1393–1396.

9. Ramyar, K., Topal, A., Andiç, Ö.: Effects of aggregate size and angularity on alkali–silica reaction. Cement and Con-crete Research, 2005, vol. 35, pp. 2165–2169.

10. Suwito, A., Jin, W., Xi, Y., Meyer, C.: A Mathematical Model for the Pessimum Effect of ASR in Concrete. Concrete Sci-ence and Engineering, 2002, vol. 4, pp. 23–34.

11. Lobo, C.: Challenging ASR predictive testing. The Concrete Producer, 1998.

12. U.S. Geological Survey Circular 1219: Planning for the Conservation and Development of Infrastructure Resources in Urban Areas-Colorado Front Range Urban Corridor, 2002.

13. Modreski, P.: Colorado Rocks, USGS, 2004.14. Smaoui, N., Bérubé, M. A., Fournier, B., Bissonnette, B.,

Durand, B.: Effects of alkali addition on the mechanical properties and durability of concrete. Cement and Concrete Research, 2005, vol. 35, pp. 203–212.

15. Tosun, K., Felekoglu, B., Baradan, B.: The effect of cement alkali content on ASR susceptibility of mortars incorporat-ing admixtures. Building and Environment, 2007, vol. 42, pp. 3444–3453.

16. Mindess, S., Young, J. F., Darwin, D.: Concrete, 2nd ed., Prentice Hall, Upper Saddle River, NJ, 2003.

gate. Based on the test data of the three standard testing methods, all three aggregates may be considered to be potentially deleterious. Afterwards, cylindrical concrete specimens were made with the three types of aggregate for studying the effect of ASR on the mechanical properties of the concretes. A new acceleration method was proposed based on ASTM C1260 and JASS 5N T-603 for acceler-ating ASR in concrete specimens, and the strength and stiffness of concrete specimens conditioned by the new method were tested every two weeks for two months.1) The results from ASTM C289 indicate that all aggre-

gates are potentially deleterious. The results from JASS 5N T-603 show that the trend for expansion of con-crete bars was very consistent, but the ASR expansion of the concretes was not clear. GJ aggregate obviously expanded, but not the other two. The critical alkali content of GJ coarse aggregates was estimated to be about 4 kg/m3 for 0.1 % expansion. In ASTM C1260, all types of aggregate reacted quickly, as indicated by the linear expansion, depending on the equivalent alkali content in the cement. GJ aggregate especially was more sensitive to ASR expansion than the other two aggregates. With increasing alkali content in the cement, ASR expansion increased tremendously up to 0.8–0.9 % of cement content. Above that, there is no significant increase in ASR expansion.

2) A new ASR acceleration method based on ASTM C1260 and JASS 5N T-603 was used to investigate the degradation of the mechanical properties of the con-cretes. Cylindrical specimens measuring 3 inch (76.2 mm) diameter × 6 inch (154.2 mm) high were prepared with additional alkali added in the concrete mixes, and then immersed in high-temperature, high-alkaline solution for 56 days. The concrete specimens were tested for strength and stiffness during the 56 days. The results show that the compressive strength varied over the testing period. The strength of concrete gradually increased over the first four weeks and then decreased. The same trend was observed in the test data for con-crete stiffness. There are two possible mechanisms responsible for this phenomenon: hydration reactions and ASR. The hydration reactions helped to enhance the strength and stiffness and the ASR was responsible for the reduction in the mechanical properties.

3) The interaction between the two opposing mechanisms of hydration reactions and ASR was also studied. Con-crete specimens were submerged in hot water without alkali. The test data were compared with the data ob-tained with alkali solution. The mechanical properties of the concrete gradually increased, whereas those of the same concrete submerged in alkali solution de-creased, which is evidence of the contribution of the hydration reactions without ASR.

4) The change in length of concrete specimens under the new acceleration method was measured. The results show that the concrete specimens did not expand as the mortar bars did during the testing period, which means that the basic trends of concrete specimens for change in length and mechanical properties might be different. As a result, all of concrete specimens should be tested to evaluate the effect of ASR.

83



Edward Ou, Ph.D.TranSystems CorporationBridge Design 3030 LBJ Freeway Suite 900 DallasTexas, USA 75234

Victor E. Saouma, Ph.D.University of Colorado at BoulderCivil, Environmental, and Architectural EngineeringColorado, USA

17. Multon, S., Cyr, M., Sellier, A., Leklou, N., Petit, L.: Coupled effects of aggregate size and alkali content on ASR expan-sion. Cement and Concrete Research, 2008, vol. 38, pp. 350–359.

18. Saccani, A., Bonora, V., Monari, P.: Laboratory short-term evaluation of ASR: A contribution. Cement and Concrete Research, 2001, vol. 31, pp. 739–742.

Okpin Na, Ph.D.Hyundai E&C102-4, Mabuk-dongGiheung-guYongin-siGyeonggi-do446-716, Korea

Professor Yunping Xi (corresponding author)University of Colorado at BoulderCivil, Environmental, and Architectural EngineeringColorado, USAEmail: [email protected]


Technical Paper

DOI: 10.1002/suco.201400095

“Greener” concrete using adequate industrial waste is a pre-ferred option for sustainable construction. Alkali-silica reaction (ASR) and sulphate attack (SA) on concrete can be minimized by the use of mineral additions, which are particularly interesting if derived from waste. Grits from the paper industry, waste glass and two types of biomass ash were used as 10 % cement re-placement in mortar and tested for ASR and SA. Results and scanning electron microscopy observations were compared with plain mortar and mortar containing commercial silica fume. All waste materials mitigated ASR compared with the control mortar. Resistance to sulphates was increased for one of the biomass ashes used and especially for glass powder, which surpassed silica fume. Therefore, two of these waste materials seem to be promising as partial replacement materials for cement, leading to enhanced durability and thus contributing to sustainable con-struction.

Keywords: ASR, sulphate attack, concrete, sustainability, additions

1 Introduction

Cementitious materials, mainly in the form of concrete, are the most successful materials in the world; every year more than 1 m3 is produced per person worldwide. There-fore, there is increasing pressure to innovate and improve sustainability [1]. In fact, sustainable building develop-ment includes wise management of resources, achieved by the use of industrial by-products and post-consumer, dis-carded materials, and a lower environmental impact, achieved through reduced mining of natural aggregates from quarries. Concrete could be a feasible solution to environmental problems since it is also possible to reuse solid by-products from other industries for concrete pro-duction. This would reduce the need to landfill these ma-terials and extract natural aggregates from quarries [3] (for aggregates and also clinker production) while still main-taining acceptable concrete quality.

The cement industry contributes over 6 % to global CO2 emissions, making this industry an important sector

for implementation of CO2 emission mitigation strategies, such as increased use of additions [2], [4].

In fact, experience all over the world, both histori-cally and during the past few decades, shows that the Portland cement + pozzolanic/cementitious materials composite system will always be inherently superior to the Portland cement system in terms of its microstructure and durability properties [5].

Reuse of post-consumer wastes and industrial by-products in concrete is necessary if we are to produce “greener” concrete. The use of industrial waste and by-products materials is now widely recognized as one of the major preferred options for achieving sustainable develop-ment. Recycling by-products is an environmentally friend-ly method for large quantities of materials that may other-wise pollute land, water and air [6]. The use of coal ash, rice husk ash, wood ash, blast-furnace slag, silica fume and other similar pozzolanic materials can reduce the use of manufactured Portland cement clinker and at the same time produce concrete that is more durable [5]. “Greener” concrete also improves air quality, minimizes solid waste and leads to sustainable cement and concrete industries [7].

This implies that there are strong technical and economic arguments and evidence supporting the modifi-cation of the fineness or specific surface of industrial by-products rather than that of Portland cement and thus ensure that concrete including these by-products are manufactured for durability rather than for strength [5].

Damage to concrete structures due to ASR and con-sequent expansion is being observed in more and more countries. Alkali hydroxides present in the concrete pore solution react with amorphous or poorly crystalline silica phases in aggregates, forming a gel that imbibes water and expands. ASR will develop only if certain conditions exist, such as a sufficient amount of alkalis in the concrete pore solution, sufficient moisture levels in the concrete and re-active aggregate (coarse or fine). In order to limit ASR ex-pansion, or even suppress it, engineers must act on one or more of the above factors [8]. Numerous papers have been published regarding the mitigation of ASR with additions. It is well known that additions, particularly silica fume, metakaolin, low-calcium fly ash, ground glass, high-calci-um fly ash and blast-furnace slag (these last two at higher dosages), are effective against ASR [9].

ASR and sulphate performance of mortar containing industrial waste

Ana Mafalda MatosJoana Sousa-Coutinho*

DOI: 10.1002/suco.201400095


Submitted for review: 23 October 2014; revision: 04 February 2015; accepted for publication: 16 February 2015. Discussion on this paper must be submitted within two months of the print publication. The discussion will then be published in print, along with the authors’ closure, if any, approximately nine months after the print publication.

85

A. M. Matos/J. Sousa-Coutinho · ASR and sulphate performance of mortar containing industrial waste


Sulphate attack is an important problem when it comes to the durability of concrete structures. Sulphate at-tack typically occurs in structures exposed to solutions high in sulphates, such as seawater, rivers, or some groundwa-ters [10]. Sodium sulphate reacts with calcium hydroxide to form calcium sulphate (gypsum). This reaction proceeds to a greater or lesser extent, depending on the conditions [11]. Owing to the penetration of sulphate ions into the con-crete, the calcium monosulfoaluminate crystals in the paste may convert to gypsum or ettringite; this results in a change to a larger molar volume. These volumetric changes cause expansion and internal stresses, which ultimately weaken and destroy the paste bonds, deteriorating the concrete [12]. Calcium hydroxide has several damaging attributes with regard to the sulphate resistance of concrete. It reacts directly with sulphate ions to produce gypsum; the acidic nature of sulphate solutions favours this reaction, resulting in softening of the cement paste [12]. To mitigate this at-tack, concrete codes recommend a concrete mix with a low water/cement ratio and the use of Portland cement con-taining pozzolanic materials. The lowered availability of

calcium hydroxide can reduce the damage caused by sul-phate attack due to a direct reduction in the quantity of ettringite that can form [13]. In fact, the addition of pozzo-lan reduces the calcium hydroxide in cement paste and im-proves the permeability of concrete. This helps to increase the resistance of concrete to the attack by sulphates and other harmful solutions [14]. However, the mechanism of action and optimum blends of mineral additions with ce-ment requires further research [15], [16].

In this study, the influence of different types of addi-tions from industrial waste materials on ASR and the sul-phate resistance of mortar was investigated with scanning electron microscopy and EDX, which proved to be crucial tools in understanding the behaviour of these waste mate-rials.

Ordinary Portland cement type I was partially re-placed with 10 % by weight of each waste material. Two control mortar types were also produced, one with 100 % Portland cement and another containing 10 % cement re-placement with commercially available silica fume, a well-known active pozzolan. The studies presented include the

Table 1. The chemical and physical properties of each waste material and AI for mortar containing 10 % cement replacement with each respective waste material

GBFA GWA GRT GP SF Cement

LOI VER 3.59 VER VER

Oxides (%) SiO2 18 69.5 0.4 70 SiO2 20.4(NP EN 196-2)

K2O 6.8 3.6 0.1 0.4 –

Na2O 2.5 1.4 3 17 –

Al2O3 2.1 4.2 < 1 – 4.8

Fe2O3 1.8 2 0.1 0.7 3.2

CaO 32 8.1 56 8.7 62.2

Free Lime 0.98

MgO 5.5 0.2 0.5 3.7 1.9

SO3 2.6 < 0.1 0.31 < 0.05 3.4

Cl 3 0.013 0.15 < 0.005 0.01

MnO 0.8 0 –

NaO2 eq 7.0 3.8 3.1 17.3 – –

Amorphous matter no no – yes yes –

Mineralogy 60 % CaCO3 SiO2 92 % CaCO3 21 % SiO2 KAlSi3O8 7 % Ca2Na2 – – – 7 % KCl Al3,80Mg3,15Fe1,05 (CO3)3 (Si1,75Al4,25O20)

Pozzolanicity test negative negative negative positive positive – (NP EN 196-5)

Particle size distribution d(10) (mm) 1.80 1.12 3.08 1.78 1.50 d(50) (mm) 16.17 6.72 18.01 10.19 – 11.63 d(90) (mm) 93.13 27.55 95.92 34.11 – 33.93

AI 28d 91 % 97 % 87 % 88 % 100 % –

AI 90d 100 % 107 % 86 % 88 % > 100 % –

86



chemical properties and physical characterization of waste materials used as additions and the expansion of mortar under sulphate solution and due to alkali-silica re-action. The results of the study contribute to knowledge regarding the use of these waste materials in terms of the durability of concrete.

2 Materials and experimental programme2.1 Materials

Typical commercial type I 42.5R Portland cement was employed (specific gravity 3.16 g/cm3) and commercially available silica fume (SF) was used as a reference pozzola-nic material (specific gravity 2.20 g/cm3).

Waste glass was obtained from a recycling glass plant where glass scrap, e.g. from car windscreens, is crushed and sold to bottle manufacturers. Fine waste glass from this plant was ground for 48 h in a ball mill in the laboratory [17] and will be called GP in this study.

Grits are a waste material produced during paper manufacture. The Kraft process corresponds to the tech-nology for converting wood into wood pulp, which con-sists of almost pure cellulose fibres for paper production. The process entails treating wood chips with a mixture of sodium hydroxide and sodium sulphide, which break down the bonds between the lignin and the cellulose. The recovery boiler enables the recovery and reuse of the inor-ganic pulping chemicals such that a Kraft mill is almost a closed cycle with respect to inorganic chemicals, apart from those used in the bleaching process, where grits are

produced. Grits from a paper pulp plant were ground in a ball mill in the laboratory [18] and are called GRT here.

In the paper pulp industry (Kraft process), only heartwood and sapwood are useful for making pulp. Bark, which contains relatively few useful fibres, is removed and can be used as fuel for the biomass boiler, which provides steam, and thus energy, for use in the pulp mill. Biomass fly ash from the biomass boiler, retrieved at the electric filters, is mostly dumped to landfill, causing economic and environmental problems [19]. Ground biomass fly ash (GBFA) from the same Portuguese paper pulp supplier from which the grits were obtained was used in the pre-sent work.

Wood bottom ash was obtained from a biomass power plant in Portugal where forest waste is incinerated in a biomass boiler to produce energy. Forest waste com-bustion produces two types of waste: fly ash and wood bottom ash. Wood bottom ash was ground in a ball mill in the laboratory after screening with a 4 mm sieve and named ground wood bottom ash (GWA) [20].

The physical and chemical properties of the materi-als are shown in Table 1. The activity index (AI) of a cer-tain percentage of cement replacement by an addition corresponds to comparing the strength of mortar pro-duced with that percentage replacement of addition with the strength of an equivalent mortar with no cement re-placement, at the same age and produced under exactly the same conditions.

Scanning electron microscopy (SEM) images are shown in Fig 1.

Fig. 1. Scanning electron microscopy (SEM) images of samples of cement and the five additions used

CEM I 42.5R

GP BFA GRT

SF GWA

87



2.2 Mix proportions, production and fresh properties

Six mortar types were prepared following the procedure described in NP EN 196-1 [21]: a control mix with 100 % cement (CTL), mixes with 10 % of each addition as ce-ment replacement and one mix with 10 % cement replace-ment with silica fume (SF). A superplasticizer (SP) was used.

Mortar workability was measured according to ASTM C 109/90 [22] and ASTM C230 [23], and superplas-ticizer was added when a significant decrease in workabil-ity was observed. The mix proportions and fresh proper-ties of the mortars are shown in Table 2.

2.3 Test procedure2.3.1 Expansion due to ASR

ASR experiments were carried out in accordance with ASTM 1567 [24]. Mortar bars were prepared in 25 × 25 × 285 mm moulds to determine the change in length of the different types of the blended cement mortar types pro-duced.

Two minor modifications were performed when fol-lowing the ASTM procedure. Instead of the 1:2.25:0.47 cement:aggregate:water ratio required in the standard, the ratio used was 1:3:0.5, which is the ratio considered in NP EN 196-1. ASTM C 1567 also requires a particle size distri-bution of (reactive) aggregates as shown in Table 3, but the distribution used was the one considered in NP EN 196-1 for CEN sand (Table 3). The CEN sand used was found to be reactive. ASTM C 1567 was followed for the remaining procedure. So, immediately after casting, the moulds were covered and demoulded 24 h later. They were then precon-ditioned for a further 24 h in water maintained at 80 °C. The lengths of these mortar bars after immersion in hot water, measured along the four faces of each specimen, were then taken as the initial readings (L0). The mortar bars were subsequently transferred to 1 N NaOH solution main-tained at 80 °C and measured periodically over 14 days.

Therefore, the actual length Lx of each specimen on day x is given by

41

4

LLi

xi∑

= =

The expansion of each specimen on day x is given by

250

·100%0eL Lx=

−

Expansion with time is shown in Fig. 2 and final expan-sion in Fig. 3.

Fig. 2. Change in length due to ASR

Table 2. Mortar mix proportions and workability

CTL SF GBFA GWA GRT GP

Cement (g) 450 405 405 405 405 405

Addition (g) – 45 45 45 45 45

Sand (g) 1350 1350 1350 1350 1350 1350

Sp (g) – 1.35 – – – –

w/c ratio 0.5 0.5 0.5 0.5 0.5 0.5

Workability (mm) 199 196 196 194 194 203

Table 3. Grading requirements in ASTM C 1567 and grading for CEN sand used

ASTM C 1567 grading requirements CEN sand

Retained on Mass (%) Retained on Mass (%) sieve size sieve size

2.36 mm (No. 8) 10 2.00 mm 0

1.18 mm (No. 16) 25 1.60 mm 7

0.60 mm (No. 30) 25 1.00 mm 26

0.30 mm (No. 50) 25 0.50 mm 34

0.15 mm (No. 100) 15 0.16 mm 20

88



2.3.3 Microstructure analyses

After completion of the ASR and sulphate tests, the micro-structures of some of the mortar bar samples were investi-gated using SEM analysis with EDX spectrums, see Fig. 7 to Fig. 11.

The SEM/EDS examination was performed using a high-resolution (Schottky) environmental scanning elec-tron microscope with X-ray microanalysis and electron

2.3.2 Resistance to sulphate attack

Resistance to external sulphate attack was evaluated ac-cording to Portuguese standard E-462 [25] on six (1, 2, 3, 4, 5 and 6) mortar prisms measuring 16 × 16 × 160 mm. Test specimens were immersed in calcium hydroxide solu-tion for 28 days, and the measurements along the four side faces of each specimen were taken as initial readings (L0). Test specimens 2, 4 and 6 were transferred to a sodium sulphate solution and the remaining specimens were kept in calcium hydroxide saturated solution. Readings were taken throughout 26 weeks and the sulphate solution was renewed every two weeks. The actual increase in length of each specimen on day x is given by

(1)

exp 01600

(specimens 2, 4 and 6)2 4Na SO Lt L= −

(2)

exp ( ) 01600

(specimens1, 3 and 5)2Ca OH Lt L= −

where Lt is the comparator reading of the specimen at age t, L0 is the initial comparator reading of the specimen and 1600 mm is the nominal gauge length.

The expansion due to sulphates for each mortar is given by

Expansion (t) = exp Na2SO4(t) – exp Ca(OH)2(t) (3)

Photos of the mortar bar specimens after the sulphate re-sistance test, i.e. after 26 weeks of immersion in sodium sulphate solution, are shown in Fig. 4.

Expansion over 26 weeks is shown in Fig. 6, as well as the expansion limit for classifying the binder as sul-phate-resistant or non-resistant according to [25].

Fig. 3. Expansion due to ASR

Fig. 4. Mortar bar specimens after sulphate resistance test (Na2SO4 immer-sion for 26 weeks)

CTL SF GBFA

GWA GRT GP

89



backscattered diffraction analysis: Quanta 400 FEG ESEM/EDAX Genesis X4M.

Samples were coated with an Au/Pd thin film by sputtering, using the SPI module sputter coater equip-ment.

3 Results and discussion3.1 Chemical and physical properties of waste materials

The chemical composition of GP is in accordance with results published by other authors using soda lime glass

waste in mortar and concrete [17]. Considering that no specific European standard covers material such as GP and that it is mainly composed of silica, GP properties were compared with the requirements for fly ash (NP EN 450-1 [27]), as shown in Table 4. It can be seen that glass powder is basically in accordance with these require-ments, except for Na2O content (17 %) – far above the limit (5 %) imposed in the standard for fly ash.

Regarding GBFA, researchers have proved the sig-nificant possibility of using biomass fly ash, another indus-trial waste material, for cement replacement [26]. Standard

Fig. 5. Expansion with time due to sulphate attack

Fig. 6. Final expansion due to sulphates

90



Fig. 7. SEM on CTL mortar after ASR test

a) Sand particle in CTL mortar (× 500) b) Surface of sand particle (× 2000) c) Grooved surface and gel (× 5000)

d) Spectrum Z1 in c) showing the siliceous nature of the sand particle

e) Z2: spectrum of gel formed during ASR on the surface of the siliceous sand particle

Fig. 8. SEM on GP mortar after ASR test

a) General View

b) Surface of sand particle in GP mortar (× 2000) showing gel (Z1)

forming over the surface (Z2)

c) Surface of the sand particle (× 5000) with slight grooving

d) Z1: Gel spectrum e) Z2: Spectrum of Z2 in b) showing the siliceous nature of the sand particle

91



NP EN-450-1 [27] includes fly ash obtained from co-com-bustion of specific co-combustion biomass materials up to 20 % by mass of the total fuel. However, owing to the wide range of biomass resources and combustion conditions, an upper limit (by weight) of alkali content, chloride and unburned carbon is enforced at 5, 0.1 and 5 % respectively [28]. The biomass fly ash considered in this work is far dif-ferent chemically from coal fly ash, with the sum of SiO2 + Al2O3 + Fe2O3 being less than a third of the required val-ue, as can be seen in Table 4. The biomass fly ash studied falls within the limits given by [27], except for loss on igni-tion (LOI). High percent ages of calcium carbonate are typically observed in this type of ash [30]. In fact, calcium carbonate leads to CaO and CO2 on heating below 1000 ºC, the temperature used for determining LOI. Al-though LOI is widely attributed to the amount of com-bustible matter in the sample (especially for coal fly ash), it might not represent the amount of unburned carbon in ash properly, but rather the volatile fraction [30]. This bio-mass fly ash contains no amorphous material, thus com-promising pozzolanic activity. GBFA exhibits an equiva-lent fineness to the cement used, as can be seen from the laser particle size distribution data presented in Table 1. The amount of chlorides is also a problem and must be re duced by prior treatment if this material is to be used as a partial cement replacement material in reinforced con-crete. GBFA exhibits calcite (CaCO3) and quartz (Table 1).

The main crystalline phases in GWA are quartz (SiO2) and microcline (KAlSi3O8). GWA oxide contents match the requirements for coal fly ash in accordance

with EN 450-1 [27], as can be seen in Table 4. Chlorides are well below the requirements, and LOI falls within the limits [27] for class A coal fly ash.

GRT materials are composed mainly of calcium and the main crystalline form is calcite. As can be seen, many characteristics do not comply with the requirements (Ta-ble 4). In fact, sodium is not compatible with concrete due to alkali-silica reaction and so equivalent sodium oxide must be low. Calcium oxide content is also too high due to the existence of mostly carbonates which also contrib-ute to the excessive LOI [18].

In terms of physical properties, all additions comply with requirements in NP EN 450-1 (Table 4).

3.2 Expansive reactions

In terms of ASR, reactive sand was used and a reduced expansion in mortar with 10 % cement replacement of each addition was observed, compared with the control mortar, especially for SF and GBFA mortar. The greatest expansion (0.190 %) was observed when 100 % Portland cement was used as a binder, indicating potential deleteri-ous expansion according to ASTM 1567 [24] (> 0.10 % af-ter 14 days in NaOH).

According to Wang and Baxter [28], biomass fly ash has much better performance than class C fly ash when it comes to mitigating ASR expansion, despite its much higher alkali content, which was confirmed in the present study. In fact, Wang and Baxter concluded that after 14 days and six months, ASR expansion was, respectively,

Fig. 9. SEM on CTL mortar after 26 weeks of immersion in sodium sulphate

a) Massive precipitation of gypsum crystals on CTL mortar (× 30000)

b) CTL mortar (× 30000) indicating Z2 and Z3 zones

c) General view (× 5000)

d) EDX Spectrum of Z1 in a) of mainly gypsum

e) EDX Spectrum of Z2 indicating probably thaumasite and ettringite

f) EDX Spectrum of Z3 indicating probably ettringite

92



0.18 and 0.28 % for high alkali cement, 0.02 and 0.16 % for 33 % cement replacement by class C fly ash, 0.02 and 0.05 % for 33 % cement replacement by fly ash including co-fired sawdust, and 0.02 and 0.06 % for 33 % cement replacement by fly ash including co-fired switch grass [28].

Moreover, in a study carried out by Esteves et al. [26], expansion results for mortar mixes containing reac-tive aggregate and 20 and 30 % biomass fly ash showed a clear reduction in the expansion compared with plain ce-ment mortar.

The increased amount of C-S-H gel due to the reac-tion between calcium hydroxide and pozzolanic mineral additions in the blended cement compositions might have decreased the ASR gel formation of cement-based com-posite materials. The pozzolanic reaction of the blended cement, SF and GP, seems to have increased the durabili-ty of mortar regarding ASR.

Regarding GP mortar, ASR testing confirmed that glass powder assisted in hindering expansion compared with control specimens, confirming conclusions by other authors [31], [32], [33], [34]. Frederico and Chidiac [35] re-ported results on ASR expansion relating to the particle size of glass and cement replacement dosage including work by Jin et al. [36] and Shao et al. [37], with which the present results are in accordance.

Adding GRT decreased ASR expansion even with particle sizes considerably greater than those of cement

and no pozzolanic reaction taking place. According to Hobbs [38], in concretes containing a reactive aggregate which have not shown deleterious expansion due to ASR, gel can often be found filling or partially filling air voids, so that expansion is not observed. Air content seems to reduce expansion but does not prevent deleterious expan-sion from occurring. Results for GBFA are similar.

SEM examinations of mortar after the ASR test showed the presence of a kind of gel (Fig. 7a and 7c) com-posed of silicon, sodium and calcium (Fig. 7b and 7d), which must be the gel formed during deleterious reaction. Furthermore, SEM samples showed the formation of al-kali grooved surfaces of reactive aggregate observed in Fig. 7b and 7c and Fig. 8c. EDX results for these areas (Fig. 5) confirmed that besides O, the gel formed contains Si and Na, and that the grooved surface contains only Si and O, thus corresponding to siliceous aggregate (Fig. 7d and Fig. 8e). In the CTL samples, alkali attack on the ag-gregate seemed more intense, with deeper grooves, thus explaining higher expansion in the ASR test.

Regarding sulphate resistance, CTL, GRT and GBFA mortar, expansion rates were low at the beginning and increased substantially after 18 weeks of immersion in Na2SO4 and 10 weeks for GBFA. At the end of the test, ex-pansion proved to be > 0.20 % for CTL, GWA and GRT, and 2.00 % for BFA, and therefore 2 and 20 times respec-tively above the limit of 0.1 % required in E-462 [25].

Fig. 10. SEM on GP mortar after 26 weeks of immersion in sodium sulphate

a) General view (× 2500)

b) Area where ettringite formation seems to have led to microcracking

(× 15000)

c) Probably ettringite crystals, according to f)

d) Area with varied crystals and gel (× 15000)

e) Enlarged area (× 30000) corresponding to b)

f) EDX Spectrum of Z1 indicating mainly ettringite

93



These specimens exhibited cracks and disintegration and loss of cohesion in the case of GBFA and GRT, as can be seen in Fig. 4.

No obvious expansion acceleration pattern was ob-served in blended cement containing GP or SF. GP mor-tar showed the best performance regarding sulphate resis-tance. These results agreed well with visual examinations, indicating no deterioration of GP and SF mortar, as shown in Fig. 4. The pozzolanic activity of GP and SF binds portlandite (CH) released in the hydration of calci-um silicates (C3S and C2S), so CH is no longer available for reaction with sulphates, in accordance with [39], [40]; this prevents the formation of gypsum. Pozzolanic reac-tion produces a secondary C-S-H that also decreases the capillary porosity of mortar and significantly enhances the paste-aggregate interface [39].

SEM observations showed massive precipitation of gypsum (Fig. 9a and Fig. 11a) in GBFA mortar and also in CTL, but not in GP. This confirms findings by Chindapra-sirt et al. [41], who carried out a similar study with rice husk ash and fly ash.

As can be seen in Fig. 11b and Fig. 7c for GBFA mor-tar, EDX Z2 and Z3 seem to indicate the existence of mixed ettringite and thaumasite due to silicon content in accor-dance with [41], [42]. Manu et al. [15] suggested that ettring-ite and thaumasite coexist in solid solution. According to Collepardi [44], thaumasite may form from the reaction of calcium carbonate, calcium sulphate, the C-S-H and water. In Na2SO4 solution, decomposition of C-S-H appears to be

the source of reactive silica and this is available in the pore solution to react with the calcium carbonate from GBFA (containing 60 %) and SO4

– to form thaumasite.Ettringite causes expansion, but the material still re-

mains sound and cohesive even after cracking. Converse-ly, thaumasite causes the mortar to become incoherent and non-resistant. This was observed at the end of the test with GBFA mortar which may be related to the excessive expansion observed.

4 Conclusions

SEM was an essential tool in understanding processes in-volved in ASR and sulphate testing of these waste materi-als in cementitious mortar.

Expansion due to ASR was mitigated with all addi-tions used. Ordinary Portland cement mortar proved to be the most potentially deleterious mortar mix. It seems that the incorporation of additions mitigates ASR expansion depending on the type of waste material used.

Non-sulphate-resistant Portland cement became sul-phate-resistant when used in a blend with either silica fume or glass powder, the only actual pozzolanic addition from waste used. Pozzolanic reactions of blended cements reduced the CH content necessary for the formation of gypsum. The best dimensional stability was obtained with blended cements containing GP.

Of all the waste materials studied, only GP – al-though NaOeq exceeded the limit – seems to be promising

Fig. 11. SEM on GBFA mortar after 26 weeks of immersion in sodium sulphate

a) Massive precipitation of gypsum crystals on GBFA mortar (× 30000)

b) According to EDX spectrums, area with gypsum (Z1), ettringite and

thaumasite

c) According to EDX spectrums, area with ettringite and thaumasite

d) Z1 of b) EDX Spectrum mainly gypsum

e) Z2 EDX Spectrum probably ettringite and thaumasite

f) Z3 EDX Spectrum, probably ettringite and thaumasite

94



as a partial replacement material for cement. No signifi-cant strength loss was observed in mortar, and in terms of durability, GP mortar seems to be sulphate-resistant and leads to a decrease in ASR. Other studies suggest that when GP is used as 20 % cement replacement, ASR ex-pansion is very low [17], [31], [33], [35].

Acknowledgements

The authors would like to thank Telma Ramos, Tiago Se-co Duarte, José Miguel Neves and Alvarim Almeida, LEMC, LABEST, Secil and Sika.

This work was financed by FEDER funds within the scope of the Operational Program Factors of Competive-ness – COMPETE and by National Funds within the scope of FCT – Foundation for Science and Technology through project PTDC/ECM/098117/2008 “Additions from waste materials for sustainable structural concrete”.

References

1. Scrivener, K. L., Kirkpatrick, R. J.: Innovation in use and re-search on cementitious material. Cement and Concrete Re-search, 38, 9, 2008.

2. Mazzoli, A., Moricono, G.: Particle size, size distribution and morphological evaluation of glass fiber reinforced plastic (GRP) industrial by-product. http://dx.doi.org/10.1016/j.mi-cron.2014.07.0007.

3. Intl. Conf. on Sustainable Construction Materials & Tech-nology, Chun, Y.-M. et. al (eds.), Taylor and Francis Publish-ers, UK, June 2001.

4. Worrell, E., Price, L., Martin, N., Hendriks, J.: Carbon Diox-ide Emissions from the Global Cement I, Industry, Annual

Review of Energy and Environment, C. & Meida, L. O., vol. 26, 2001, pp. 303–329.

5. Swamy, R. N.: Sustainable Concrete for the 21st Century – Concept of Strength through Durability. Indian Concrete Journal, vol. 81, Dec, 2007, pp. 7–15.

6. Mehta, P. K.: Mineral admixtures for concrete – an overview of recent developments. Proc. of Engineering Foundation Conference Advances in cement and concrete, University of Newhampshire, Durham. ASCE, 1994, pp. 243–256.

7. Naik, T. R.: Environmental-friendly concrete with industrial and post-consumer byproducts, Report No. CBU-2004-21, REP-570, Dec, UWM Center for By-Products Utilization, Dept. of Civil Engineering & Mechanics, University of Wis-consin-Milwaukee, Milwaukee, USA, 2004.

8. Karakurt, C., Topçu, I. B.: Effect of blended cements pro-duced with natural zeolite and industrial by-products on al-kali-silica reaction and sulphate resistance of concrete. Con-struction and Building Materials 25 (2011), pp. 1789–1795.

9. Cyr, M., Rivardb, P., Labrecque, F.: Reduction of ASR-ex-pansion using powders ground from various sources of reac-tive aggregates. Cement and Concrete Composites, vol. 31, No. 7, Aug 2009, pp. 438–446.

10. Irassar, E. F., Gonzalez, M., Rahhal, V.: Sulphate resistance of type V cements with limestone filler and natural pozzola-na. Cem Concr Compos, 2000, 22, pp. 361–368.

11. Neville, A.: The confused world of sulphate attack on con-crete. Cem Concr Res, 2004, 34, pp. 1275–1296.

12. Tikalsky, P. J., Roy, D., Scheetz, B., Krize, T.: Redefining ce-ment characteristics for sulphate-resistant Portland cement. Cem Concr Res, 2002, 32, pp. 1239–1246.

13. Santahanam, M., Cohen, M., Olek, J.: Differentiating seawa-ter and groundwater sulphate attack in Portland cement mortars. Cem Concr Res, 2006, 36, pp. 2132–2137.

14. Malhotra, V. M. (ed.): Supplementary cementing materials for concrete, CANMET. Canadian Government Publishing Centre, 1987.

Table 4. Requirements in standard EN 450-1

Fly Ash GBFA GWA GP GRT (NP EN 450-1)

Chemical requirements Cl ≤ 0.10 % 3.00 % < 0.005 0.15 %

SO3 ≤ 3.0 % 2.60 % – 0.31 %

Free CaO ≤ 2.5 % 8.70% 56 %

Reactive CaO ≤ 10 %

SiO2 ≥ 25 % 18 % 69.5 % 70 % 0.4 %

SiO2+Al2O3+Fe2O3 ≥ 70 % 21.90 % 75.70 % 71.82 %

NaO2eq ≤ 5.0 % 7 % 3.80 % 17 %

LOI Class A ≤ 5 % 29 % 3.59 % 0.92 % 42 %

Class B 2–7 %

Class C 4–9 %

Physical requirements Activity index ≥ 75 % at 28 days 91 % 97 % 88 % 87 %

≥ 85 % at 90 days 100 % 107% 88 % 86 %

Initial setting time ≤ 120 min plus OK OK OK OK initial setting time of reference paste

Soundness ≤ 10 mm OK OK OK OK

95



15. Al-Amoudi, O. S. B.: Attack on plain and blended cements exposed to aggressive sulphate environments. Cement Con-crete Comp, 2002, 24, pp. 305–316.

16. Manu, S., Menashi, D. C., Jan, O.: Sulphate attack research – whither now. Cement Concrete Res, 2001, 31, pp. 845–851.

17. Matos, A. M., Joana Sousa-Coutinho, J.: Durability of mor-tar using waste glass powder as cement replacement. Con-struction and Building Materials 36 (2012), pp. 205–215.

18. Da Luz Garcia, M., Sousa-Coutinho, J.: Grits and dregs for cement replacement – preliminary studies. 11th IC- NOCMAT, UK, 6–9 Sept 2009.

19. Da Luz Garcia, M., Sousa-Coutinho, J.: Durability using biomass fly ash as a partial cement replacement material. II Simposio Aprovechamiento de residuos agro-industriales como fuente sostenible de materials de construcción, Valen-cia, 8–9 Nov 2010.

20. Da Luz Garcia, M., Sousa-Coutinho, J.: Strength and dura-bility of cement with forest waste bottom ash. Construction and building Materials, 2012 (in press).

21. IPQ. EN 196-1 Methods of testing cement – Part 1: Determi-nation of strength. Lisbon, IPQ, 2006.

22. ASTM C109/C109M-11: Standard test method for compres-sive strength of hydraulic cement mortars (using 2 in. or [50 mm] cube specimens). West Conshohocken, 2011.

23. ASTM C230/C230M-08: Standard specification for flow ta-ble for use in tests of hydraulic cement. Philadelphia, 2008.

24. ASTM C1567-11: Standard Test Method for Determining the Potential Alkali Silica Reactivity of Combinations of Ce-mentitious Materials and Aggregate (Accelerated Mortar Bar Method). ASTM International, 2011.

25. LNEC. E-462: Resistencia dos cimentos ao ataque por sulfa-tos. Resistance of cements to sulphate attack. Lisbon, 2004.

26. Esteves, T. C., Rajamma, R., Soares, D., Silva, A. S., Ferreira, V. M., Labrincha, J. A.: Use of biomass fly ash for mitigation of alkali-silica reaction of cement mortars. Construction and Building Materials 26 (2012), pp. 687–693.

27. NP EN 450-1, European Committee for Standardization, Fly Ash for Concrete, Part 1, Definition, Specification and Con-formity Criteria, 2012.

28. Wang, S., Baxter, L.: Comprehensive study of biomass fly ash in concrete: Strength, microscopy, kinetics and durabili-ty. Fuel Processing Technology 88(2007), pp 1165–1170.

29. Naik, T. R., Kraus, R. N., Kumar, R.: Wood Ash: A new source of pozzolanic material, Report No. CBU-2001-10, REP-435. UWM Center for By-Products Utilization. Dept. of Civil Engineering & Mechanics, University of Wisconsin-Milwaukee, Milwaukee, USA, June 2001.

30. Poykio, R., Ronkkomaki, H., Nurmesniemi, H., Peramaki, P., Popov, K., Valimaki, T., Tuomi, T.: Chemical and physi-cal properties of cyclone fly ash from the grate-fired boiler incinerating forest residues at a small municipal district plant (6MW). Journal of Hazardous Materials, 2008.

31. Schwarz, N., Cam, H., Neithalath, N.: Influence of a fine glass powder on the durability characteristics of concrete and its comparison to fly ash. Cement & Concrete Compos-ites, 2008, 30, pp. 486–496.

32. Browne, R. D.: Field Investigations: Site & Laboratory Tests: Maintenance, Repair and Rehabilitation of Concrete Struc-tures. CEEC, Lisbon, 1991.

33. Saccani, A., Bignozzi, M. C.: ASR expansion behavior of re-cycled glass fine aggregates in concrete. Cement and Con-crete Research, 2010, 40, pp. 531–536.

34. Taha, B., Nounu, G.: Using lithium nitrate and pozzolanic glass powder in concrete as ASR suppressors. Cement & Concrete Composites, 2008, 30, pp. 497–505.

35. Frederico, L. M., Chidiac, S. E.: Waste glass as a supplemen-tary cementitious material in concrete – Critical review of treatment methods. Cement & Concrete Composites, 2009, 31, pp. 606–610.

36. Jin, W., Meyer, C., Baxter, S.: Glascrete – concrete with glass aggregate. ACI Mater J, 2000, 97 (2), pp. 208–213.

37. Shao, Y., Lefort, T., Moras, S., Rodriguez, D.: Studies on con-crete containing ground waste glasses. Cement and Concrete Research, 2000, 30, pp. 91–100.

38. Hobs, D. W.: Alkali-silica reaction in concrete, Thomas Tel-ford Ltd, London, 1988.

39. Sahmaran, M., Kasap, O., Duru, K., Yaman, I. O.: Effects of mix composition and water-cement ratio on the sulphate re-sistance of blended cements. Cem Concr Compos, 2007, 29, pp. 159–167.

40. Thidar, A., Oguchi, C. T.: Resistance of plain and blended cement mortar exposed to severe sulphate attacks. Constr Build Mater, 2011, 25, pp. 2988–2996.

41. Chindaprasirt, P., Kanchanda, P., Sathonsaowaphak, A., Cao, H. T.: Sulphate resistance of blended cements contain-ing fly ash and rice husk ash. Construction and Building Materials 21 (2007), pp. 1356–1361.

42. Irassar, E. F., Bonavetti, V. L., Gonzalez, M.: Microstructur-al study of sulphate attack on ordinary and limestone Port-land cement at ambient temperature. Cement and Concrete Research 33 (2003), pp. 31–41.

43. Rodrigues, A., Fernandes, I., Noronha, F.: Caracterização de produtos de deterioração do betão por métodos petrográfi-cos. GEOTIC – Sociedade Geológica de Portugal VIII Con-gresso Nacional de Geologia. ISSN 1645-0388, vol. 9, No. 21, 2010.

44. Collepardi, M.: Degradation and restoration of masonry walls of historical buildings. Mater. Struct. 23 (1990), pp. 81–102.

Ana Mafalda MatosResearcher at LEMC-FEUP,Department of Civil EngineeringFaculty of EngineeringUniversity of PortoRua do Dr. Roberto Frias, s/n4200-465 PortoPortugalTel: +351 225081936Fax: +351 [email protected]

Joana Sousa-CoutinhoAssociate Professor at FEUPResearcher at LABESTFEUP, Department of Civil EngineeringFaculty of EngineeringUniversity of PortoPortugal


Technical Paper

DOI: 10.1002/suco.201400123

Sulphate attack-induced damage and micro-mechanical properties of concrete characterized by nano-indentation coupled with X-ray computed tomography

Sulphate attack is a serious problem for concrete in marine environments. Sulphate attack can change the composition and microstructure of concrete and eventually influence the mechani-cal and durability performance. In this paper, the heterogeneity and mechanical properties of concrete exposed to sulphate is investigated from the microscopic to the mesoscopic scale. X-ray computed tomography (XCT) and nano-indentation were adopted to define the defect zone and establish the relationship between interfacial transition zone (ITZ) and concrete matrix (mortar). The experiments were based on concrete and mortar specimens of different strengths. The results of XCT have nano-indentation indicate that the specimens had similar degrees of damage re-gionally and good correlation exists between the elastic moduli of the ITZ and the mortar. The concrete can be partitioned into three parts: the cracked zone with heavy damage, damaged zone and undamaged zone. The elastic modulus of the mortar phase and the ITZ has a linear relation.

Keywords: concrete, sulphate attack, X-ray computed tomography, nano-indentation, indentation modulus

1 Introduction

The characteristics of damage, deterioration and failure in concrete have multi-scale phenomena in time and space. So far, numerous studies have been carried out with extensive, repetitive and indoor tests. However, the dam-age mechanism and evolutionary process of mechanics and durability in concrete, which is exposed to sulphate in actual service conditions, cannot be precisely formu-lated. For the purposes of providing effective methods of solving sulphate attack problems, researchers turn to investigating the damage evolution in microstructure and mesostructure, trying to establish a relationship between macroscopic and microscopic concrete. Wittmann [1] considered that concrete performance mainly depends on the characteristics of concrete in microcosm and macro-cosm. Zhongwei [2] emphasized that studies of concrete should be from microscopic to macroscopic scale, from

entirety to segment, and that the interrelation and inter-action of concrete structure at different scales are very important. In summary, studies of the microscopic and mesoscopic structure of concrete are essential for improv-ing the mechanics and durability of concrete.

X-ray computed tomography (XCT) and nano-inden-tation are efficient research techniques that can reflect the evolutional process of damage and the elastic properties of microscopic and mesoscopic structures in cement-based materials. Morgan [3] applied medical CT to studies related to concrete cracks, aggregates and mortar first of all. Stock [4] analysed the damage condition of sulphate attack by XCT based on paste specimens with different water-cement ratios. XCT provides a method for inves-tigating the volume distribution of defects in materials without any prior drying preparation. For cement-based materials, XCT can monitor the microstructural changes due to sulphate attack [5], fracture [6] and cement hy-dration [7]. Nano-indentation is mainly used to analyse the microstructure of each component in concrete indi-vidually and establish the relationship between elastic properties locally. Sorrentino et al. [8] reported on the nano-indentation results for the modulus of elasticity and hardness of the major clinker phases. Acker [9] provided the modulus of elasticity of hydration products including C-S-H gel. Constantinides [10] introduced some statistical indentation techniques to characterize the mechanical properties of hydration products at the nano-scale. How-ever, only rarely has nano-indentation been used to inves-tigate the influences of sulphate attack on microstructural properties. The problem that hinders the application of nano-indentation is that the regulation which partitions the damage region and evaluates the degree of sulphate attack damage is unknown.

In this paper, we focus on combining the advan-tages of both testing methods. Firstly, XCT is applied to partition concrete into regions with different degrees of damage by way of grey values in the CT image. Grey values can reflect the densification and degree of damage in concrete. After sulphate attack, the concrete begins to swell and craze, which leads to a decrease in the grey value in the CT image of the damage region. Variations in the mechanical properties of the components with different chemical composition are then investigated by nano-indentation. Finally, the relationship between the modulus of elasticity of the mortar and the ITZ is

Chunxiang Qian*Yanfeng NieTianji Cao


Submitted for review: 23 December 2014; revision: 22 May 2015; accepted for publication: 23 May 2015. Discussion on this paper must be submitted within two months of the print publication. The discussion will then be published in print, along with the authors’ closure, if any, approximately nine months after the print publication.

97

Q. Chunxiang/N. Yanfeng/C. Tianji · Sulphate attack-induced damage and micro-mechanical properties of concrete characterized by nano-indentation coupled with X-ray computed tomography


was used to investigate the deterioration process of con-crete and acquire fast experimental results in the labora-tory. The concentration of the test solution was 10 % (by mass) sodium sulphate which was used to provide sulphate ions. Specimens were exposed to dry-wet cycles. Firstly, specimens were dried at 60 °C for 20 h. Secondly, the specimens were chilled in air at room temperature for 4 h. Thereafter, the specimens were immersed in the corrosive solution for 48 h. Specimens were exposed to five and ten dry-wet cycles, which meant that two de-grees of sulphate attack degradation were investigated.

Sulphate attack influences the speed of ultrasonic transport inside cement-based materials. The ultrasonic velocity was tested on 100 × 100 × 400 mm specimens and the definition of degree of degradation of the speci-men is based on this. The characterization of degree of degradation d can be defined by the following equation:

where ν ′ and ν represent the ultrasonic velocity of con-crete attacked and not attacked by sulphate respectively.

2.4 XCT and main test parameters

Fig. 1 shows the principle of XCT. In XCT, a detector measures the resulting intensity of a cone X-ray beam with known intensity after threading and absorption by the sample, and for different directions of irradiation θ, called projections [11]. For each angle of projection θ, the resulting intensity I (y, z, θ) on each pixel of the de-tector (coordinates y, z) is given by

where I0 is the intensity of the beam before the sample and µ (x, y, z, θ) for a great number of angles θ is called the Radon transform of the attenuation coefficient of the sample. The attenuation coefficient mainly depends on the density and chemical composition of the materials constituting the sample [12].

νν

ν νν

= ∆ = − ′(1)d

∫θ = − µ θ( , , ) exp( ( , , , ) ) (2)0I y z I x y z dx

analysed at the nano-scale. This paper indicates that the relationship between the modulus of elasticity of mortar and ITZ is linear.

2 Experimental programmes 2.1 Materials

The cement used in this study was grade 52.5 Portland cement (according to Chinese National Standard GB 175-2007 [20]). The chemical composition and mineral original composition of the cement are given in Tables 1 and 2 respectively.

2.2 Mix proportion and specimen preparation

The composition of the concrete used is presented in Ta-ble 3.

Specimens were demoulded one day after casting and cured in a fog room (20 ± 2 °C, 95 % relative humid-ity) for 28 days. Three types of specimen were prepared for different tests: 70.7 × 70.7 × 220 mm specimens were for detecting the stress–strain curve, 100 × 100 × 400 mm specimens for the XCT test and 70.7 × 70.7 × 70.7 mm mortar specimens with the same water-cement ratio as the concrete for determining the modulus of elasticity by na-no-indentation. Each test contained three specimens and the final result took the average of the three specimens.

2.3 Corrosion scheme and degree of degradation

The chemical interaction of sulphate attack is a compli-cated and slow process. Hence, an accelerated method

Table 1. Chemical composition of cement

Component Content (%) Component Content (%)

SiO2 21.35 P2O5 0.10Al2O3 4.67 BaO 0.04Fe2O3 3.31 ZnO 0.06CaO 62.60 MnO 0.18MgO 3.08 SrO 0.13K2O 0.54 PbO 0.02Na2O 0.21 Cl 0.03TiO2 0.27 Ignition 0.95SO3 2.25 Total 99.79

Table 2. Mineral original composition of cement

C3S (%) C2S (%) C3A (%) C4AF (%) Gypsum (%)

55.5 19.1 6.5 10.1 5.0

Table 3. Mixing proportion of concrete

Group Cement(kg/m3)

Water(kg/m3)

Gravel(kg/m3)

Fine aggregate(kg/m3)

Plasticizer(kg/m3)

Compressive strength(MPa)

C30 368 195 1103 735 0 30.12C50 448 157 1121 672 3.36 52.63C80 566 122 1019 679 8.46 79.07

Fig. 1. Principle of XCT with flat panel detector

98



ν Poisson’s ratio of cement-based material under test

Ei modulus of elasticity of indenter (a diamond in-denter is used in this paper, Ei = 1140 GPa)

νi Poisson’s ratio of indenter (νi = 0.07 in this paper) ν = 0.2–0.3 for cement-based material

According to Eq. (6), the values of M and E are close, so the modulus of elasticity of cement-based material at the nano-scale can be acquired by nano-indentation.

The nano-indentation test process is illustrated in Fig. 3. The size of the small specimen was 70.7 × 70.7 × 20 mm. Based on the literature [15], two kinds of loading diagram were implemented. The aim of a trapezoidal load-ing diagram is to obtain the elastic moduli and volume fractions of solid phases, which is illustrated in Fig. 4. The load of indenter P rises to 2 mN in 30 s at a constant rate; the maximum load Pmax = 2 mN is then maintained for 30 s; finally, the indenter unloads over 30 s (Fig. 4a). The other loading diagram is shown in Fig. 4b: the load of the indenter is kept constant at P = 2 mN for 240 s and the rate of loading and unloading is fast, so these times can be neglected in the whole test.

In this paper, the mortar and ITZ intruded by sul-phate attack were analysed by nano-indentation. The sec-

The XCT specimens measured 100 × 100 × 400 mm. The XCT was supplied by the YXLON company in Ger-many. The voltage used for the XCT source was 200 kV, the rotation interval was 4.59°/s and the reconstruction software was vgstudiomax21.

2.5 Nano-indentation tests

Nano-indentation is a method for determining the modu-lus of elasticity in microcosm. The basic principle of nano-indentation is that material is loaded by a cusp indenter in order to obtain a press-displacement curve (P-h curve). The indentation hardness H and indentation modulus M of a material can be calculated based on the P-h curve and mechanics model at continuous scales [13]. The calcula-tion formulas are

where:S stiffness, usually from the fitting calculation of the

elasticity part in the P-h curveβ correction coefficient of indenter (a Berkovich in-

denter is used in this paper and β = 1.034)Ac maximum of loading area

Fig. 2 illustrates a classic P-h curve (a) and a nano-inden-tation test scheme (b).

In Fig. 2:hmax max. depth of indenter penetrationhf remnant depth of indenter penetration after unloadinghc contact depth of indenter penetration

For homogeneous material with isotropy, the relation between indentation modulus M and elasticity modulus E is [14]

where:

= (3)HPAc

πβ

=2

(4)MS

Ac

= =( / ) (5)max

S dP dh h h

ν ν= − +

−1 1 1(6)

2 2

M E Ei

i

Fig. 2. a) P-h curve and b) schematic diagram of nano-indentation test

Fig. 3. Nano-indentation test process

Fig. 4. Loading diagrams for nano-indentation

99



HD C-S-H gel (24.0–35.3 GPa), CH (33.0-44.5 GPa), gypsum (45.7 GPa), ettringite (40.0–52.0 GPa) and mi-rabilite (57.1 GPa). Among these substances, crystalli-zation of the mirabilite and gypsum in the dry-wet cycle is complicated and the modulus of elasticity of ettrin-gite changes with porosity. Nano-indentation can only distinguish the phases that have large discrepancies between the mechanical properties. Besides, mirabilite, gypsum and ettringite are corrosion products. Therefore, gypsum, ettringite and mirabilite crystals are assumed to have a similar modulus of elasticity of 45.0–52.0 GPa and are regarded as one type of phase in this study.

tion of mortar was tested in a 10 × 10 matrix and 20 μm between two test points. The section of ITZ was tested in a 6 × 6 matrix and 20 µm between two test points.

3 Results and discussion 3.1 XCT analysis and application

Grey values distribution of concrete from surface to interi-or of specimen in C30, C50 and C80 after 5 and 10 cycles of sulphate attack are shown in Figs. 5, 6 and 7 respec-tively, where d is the degree of degradation of specimens based on ultrasonic velocity variation. Specimens exposed to 5 dry-wet cycles are indicated in red, which reflects less damage and a low degree of degradation. Specimens with 10 dry-wet cycles are shown in blue, which reflects serious damage and a high degree of degradation.

Fig. 5 indicates that the grey values increase from the surface to the middle of the specimen. In a sulphate envi-ronment, concrete will expand due to the generation of corrosion products. Generally, the corrosion products of sulphate attack are soft material, voids, gypsum, ettringite and sodium sulphate crystal. As expansion accumulates, the structure of the concrete was destructed from surface to inside with the appearance of cracks and voids. The grey value from the XCT can reflect the densification of specimens. Therefore, during sulphate attack, the grey value of the specimen decreases from interior to surface since the most serious damage occurs on the surface of the specimen. Similarly, the change in grey value distribu-tion also reflects the characteristics of the region eroded by sulphate, which is uniform locally but non-uniform in the concrete as a whole. The trend of the grey value distri-bution is consistent with the damage model in separated layers which was put forward by Santhanam [16] and Ju [17]. Based on the grey value, the specimens are divided into regions with different degrees of damage: a cracked zone with the most severe damage and lowest grey value, a damaged zone with moderate grey value and less dam-age than the cracked area, and an undamaged area with the highest grey value and least destruction. The cracked, damaged and undamaged zones correspond, respectively, to groups 1, 2 and 3 in Figs. 5, 6 and 7. The samples for nano-indentation were prepared from Table 4.

3.2 Nano-indentation test results

Based on the research of Constantinides [18] and the literature [19] generally, the modulus of elasticity E and hardness of different phase were pores and soft mate-rial phase (0–12.2 GPa), LD C-S-H gel (12.6–24.1 GPa),

Fig. 5. Grey values distribution from surface to interior for C30 with differ-ent degrees of damage



Table 4. Range of specimen partition

Group d 1(mm)

2(mm)

3(mm)

C30 0.25 8 10 17.350.49 11 15 9.35

C50 0.27 6 10 19.350.48 10 14 11.35

C80 0.35 6 14 15.350.58 10 12 13.35

100



All nano-indentation results for the first zone (cracked zone with severe damage) are merged and the experimental results and theoretical probability density functions (PDF) of the samples are shown in Fig. 8.

From Fig. 8 it can be seen that the microstructure of the mortar and the ITZ changes significantly and the mechanics of the microstructure decrease seriously. As the degree of damage increases, obvious changes oc-cur in the four peaks of the elastic modulus-frequency

a) Probability distribution plot of elastic modulus of sound mortar for C50

c) Probability distribution plot of elastic modulus of mor-tar attacked by sulphate for C30 (d = 0.25)

d) Probability distribution plot of elastic modulus of ITZ attacked by sulphate for C30 (d = 0.25)

b) Probability distribution plot of elastic modulus of sound ITZ for C50

curve. For example, in Figs. 8a and 8b, the four peaks correspond to soft matte phase, LD C-S-H, HD C-S-H and CH respectively. With increasing damage, the four peaks change noticeably: the LD C-S-H and HD C-S-H gradually decrease; the CH phase gradually disappears in high w/c ratio mixes, but exhibits a similar gradual reduction in low w/c ratio mixes; voids and soft material increase remarkably; gypsum, sodium sulphate crystals and ettringite are produced. The sulfate attack induced

e) Probability distribution plot of elastic modulus of mor-tar attacked by sulphate for C50 (d = 0.27)

f) Probability distribution plot of elastic modulus of ITZ attacked by sulphate for C50 (d = 0.27)

101



surface to interior and leads to less absorption of X-ray. The nano-indentation fitting curve is consistent with the mechanism of sulphate attack.

In order to acquire the evolution of the mechanical characteristics of the microstructure of concrete exposed to sulphate, the intrinsic modulus of elasticity and vol-ume fractions of LD C-S-H , HD C-S-H, CH, voids and soft material and eroded product phases are obtained by statistical deconvolutions (Tables 5 and 6).

damage in concrete is caused by the combination of salt crystallizing and chemical erosion. At macroscopic level, concrete expands, desquamates and cracks. At micro-scopic level, pores and soft materials increases and ero-sion substances (gypsum, sodium sulphate crystals and ettringite) are produced. The variations correspond with the results obtained by XCT. A decrease in grey value was seen in XCT, caused by the changes in microstructure due to sulphate attack, which destructs the concrete from

k) Probability distribution plot of elastic modulus of mor-tar attacked by sulphate for C50 (d = 0.48)

h) Probability distribution plot of elastic modulus of ITZ attacked by sulphate for C80 (d = 0.35)

i) Probability distribution plot of elastic modulus of mor-tar attacked by sulphate for C30 (d = 0.49)

j) Probability distribution plot of elastic modulus of ITZ attacked by sulphate for C30 (d = 0.49)

l) Probability distribution plot of elastic modulus of ITZ attacked by sulphate for C50 (d = 0.48)

g) Probability distribution plot of elastic modulus of mor-tar attacked by sulphate for C80 (d = 0.35)

102



m) Probability distribution plot of elastic modulus of mor-tar attacked by sulphate for C80 (d = 0.58)

n) Probability distribution plot of elastic modulus of ITZ attacked by sulphate for C80 (d = 0.58)

Fig. 8. Elastic modulus frequency distribution and fitting results of mortar and ITZ for C30, C50 and C80 concretes for the different degrees of damage

Table 5 Nano-indentation test fitting results for the mortar of concrete attacked by sulphate

Group d

Mortar

Voids & soft materials

LD C-S-H HD C-S-H CHSulphate product

E/GPa V/% E/GPa V/% E/GPa V/% E/GPa V/% E/GPa V/%

C30 0.25 6.1 26 16.8 16 28.5 26 39.9 12 51.2 200.49 5.6 43 16.5 21 28.3 20 — — 50.4 16

C50 0.27 6.4 15 17.4 21 29.5 40 40.3 18 50.7 80.48 5.0 24 16.4 22 28.3 22 40.0 16 50.9 16

C80 0.35 5.8 11 16.9 22 29.0 37 38.0 18 49.6 110.58 4.4 18 15.5 20 29.0 23 38.5 7 51.7 31

Sulphate attack has almost no effect on the modulus of elasticity of hydration products, but the relative com-position of the microstructure changes dramatically in Tables 5 and 6. The biggest change is the large increase in relative volume of voids and soft material, which results in a significantly lower modulus of elasticity for the micro-structure. The relative volume of corrosion products, such as voids and sulphate product, decreases with the increase in concrete strength in the mortar. This is because the high-strength concrete has a compact microstructure that leads to better resistance to sulphate attack. In the ITZ, the relative volume of sulphate product increases with the en-

hancement of concrete strength, but the voids are similar. This is mainly due to the characteristic of the ITZ whose actual water-cement ratio is low and similar in concrete with different strengths, which indicated there are more voids in the ITZ. Other zones (damaged and undamaged areas) exhibit similar trends to Fig. 8, Table 5 and Table 6.

3.3 Relationship between ITZ and mortar after sulphate attack on concrete

Based on the nano-indentation data for the mortar and the ITZ, a linear relationship between the average modu-

Table 6. Nano-indentation test fitting results for the ITZ of concrete attacked by sulphate

Group d

ITZ

Voids & soft materials

LD C-S-H HD C-S-H CHSulphate product

E/GPa V/% E/GPa V/% E/GPa V/% E/GPa V/% E/GPa V/%

C30 0.25 3.1 65 16.7 15 25.3 4 34.9 8 50.4 80.49 4.2 82 15.9 6 26.9 6 — — 50.9 6

C50 0.27 3.5 61 16.4 19 24.7 7 33.8 8 53.2 50.48 3.7 67 15.2 11 24.9 9 33.5 7 53.9 7

C80 0.35 4.9 62 15.3 11 24.0 7 33.0 10 51.0 100.58 4.1 63 15.0 11 24.4 10 33.9 9 52.7 7

103



Fig. 9. Nano-indentation average elasticity modulus relation between ITZ and mortar after sulphate attack on concrete (C30, C50)

Fig. 10. Nano-indentation average elasticity modulus relation between ITZ and mortar after sulphate attack on concrete (C80)

2. Wu Zhongwei. Introspection of technology in concrete [J]. Concrete and cement production, 1988, 1(6), pp. 4–5.

3. Morgan, I. L., Ellinger, H., Klinksiek, R., et al.: Examination of concrete by computerized tomography [J]. ACI Journal, 1980, 77(1), pp. 23–27.

4. Stock, S. R., Naik, N. K., Wilkinson, A. P., et al.: X-ray mi-crotomography (microCT) of the progression of sulfate at-tack of cement paste [J]. Cement and Concrete Research, 2002, 32(10), pp. 1673–1675.

5. Naik, N. N., Jupe, A. C., Stock, S. R., et al.: Sulfate attack monitored by microCT and EDXRD: Influence of cement type, water-to-cement ratio, and aggregate [J]. Cement and Concrete Research, 2006, 36(1), pp. 144–159.

6. Landis, E. N., Nagy, E. N., Keane, D. T.: Microstructure and fracture in three dimensions [J]. J Eng Mech, 2003, 70(7), pp. 911–925.

7. Bentz, D. P., Mizell, S., Satterfield, S., et al.: The visible ce-ment data set [J]. J Res National Ins Standards Technol, 2002, 107(2), pp. 137–148.

8. Sorrentino, F., Velez, K., Maximilien, S., et al.: Determination by nano-indentation of elastic modulus and hardness of pure constituents of Portland cement clinker [J]. Cem Concr Res, 2001, 31(4), pp. 555–561.

9. Acker, P.: Swelling, shrinkage and creep: a mechanical ap-proach to cement hydration [J]. Mater Struct, 2004, 37(268), pp. 237–243.

10. Constantinides, G., Ul, M. F., Van Vliet, K.: On the use of nano-indentation for cementitious materials [J]. Mater Struct, 2003, 36(257), pp. 191–196.

11. Lu, S., Landis E. N., Keane, D. T.: X-ray microtomographic studies of pore structure and permeability in Portland ce-ment concrete [J]. Mater Struct, 2006, 39, pp. 611–620.

12. Rougelot, T., Burlion, N., Bernard, D., et al.: About micro-cracking due to leaching in cementitious composites: X-ray microtomography description and numerical approach [J]. Cem Concr Res, 2010, 40, pp. 271–283.

13. Oliver, W. C., Pharr, G. M.: An improved technique for deter-mining hardness and elastic modules using load and dis-placement sensing indentation experiments [J]. Material Re-search, 1992, 7(6), pp. 1564–1583.

14. Wang, X. H., Jacobsen, S., He, J. Y., et al.: Application of na-no-indentation testing to study of the interfacial transition zone in steel fiber reinforced mortar [J]. Cement and Con-crete Research, 2009, 39, pp. 701–715.

15. Jirˇí, N., Vít, Š., Lubomír, K.: Nano-indentation characteris-tics of alkali-activated aluminosilicate materials [J]. Cem Concr Com, 2011, 33, pp. 163–170.

16. Santhanam, M., Cohen, M. D., Olek, J.: Mechanism of sul-fate attack: A fresh look – Part 2: Proposed mechanisms [J]. Cement and Concrete Research, 2003, 33(3), pp. 341–346.

lus of elasticity of mortar and ITZ is obtained, which is illustrated in Figs. 9 and 10. The average modulus of the mortar and the ITZ is calculated based on the entire mor-tar or ITZ, including the cracked, damaged and undam-aged zones in one specimen.

Figs. 9 and 10 illustrate that the modulus of elasticity changes linearly between ITZ and mortar phases. The lin-ear coefficient of C30 and C50 concrete is 0.45 (slope of fitting curve = 2.220, linear coefficient = 1/2.220 = 0.45), whereas it is 0.50 for C80 concrete (slope of fitting curve = 1.990, linear coefficient = 1/1.990 = 0.50).

4 Conclusions

In this study, X-ray computed tomography (XCT) and nano-indentation were used to analyse quantitatively two-dimensional defect changes and the relationship between interfacial transition zone (ITZ) and mortar with increasing degree of damage. From these experiments and results, the following conclusions can be drawn:1) Using XCT scan analysis, the degree of damage to con-

crete exposed to sulphate attack is inhomogeneous. Based on the grey value, the damaged zone can be par-titioned into three zones with different degrees of dam-age: cracked area, damaged area and undamaged area.

2) Nano-indentation test results indicate that sulphate at-tack changes the phase composition; soft materials and the void volume increase dramatically, the average modulus of elasticity of the microstructure decreases significantly. The relative volume of corrosion products, such as void and sulphate product, decreases with in-creasing concrete strength in the mortar (from C30 to C80). In the ITZ, the relative volume of voids is similar in concrete with different strengths.

3) The indentation modulus relationship between mortar phase and ITZ is linear. The linear coefficient for C30 and C50 concrete is 0.45, whereas it is 0.50 for C80 con-crete.

References

1. Wittmann, F. H, Roelfstra, P. E., Sadouki, H.: Simulation and analysis of composite structures. Materials Science and Engineering, 1985, 68(2), pp. 239–248.

104



17. Ju, J. W., Weng, L. S., Mindess, S., Boyd, A.: Damage assess-ment and service life prediction of concrete subject to sul-fate attack. In: Skalny, J., Marchand, J. (eds.), Material Sci-ence of Concrete-Sulfate Attack Mechanisms. American Ceramic Society, Westervill, OH, 1999, pp. 265–282.

18. Constantinides, G., Ulm, G. F. J.: The nanogranular nature of C-S-H [J]. J. Mech Phys Soli, 2007, 55(1), pp. 64–90.

19. Haecker, C.-J., Garboczi, E. J., Bullard, J. W., et al.: Mode-ling the linear elastic properties of Portland cement paste [J]. Cem Concr Res, 2005, 35(10), pp. 1948–1960.

20. GB/T 175-2007, Common Portland Cement[S].China, 2007.

Chunxiang QianSchool of Material Science and EngineeringSoutheast University, Nanjing 211189, ChinaResearch Institute of Green Construction MaterialNanjing, 21189, [email protected]

Yanfeng NieYan Tai University, Institute of Civil Engineering Yantai, China

Tianji CaoSchool of Material Science and EngineeringSoutheast University, Nanjing 211189, ChinaResearch Institute of Green Construction MaterialNanjing, 21189, China


Blended Portland cement-ceramic powder binder containing up to 60 % fine-ground waste ceramics from a brick factory is used in concrete mix design as an environmentally friendly alternative to the commonly used Portland cement. The experimental analy-sis of basic physical characteristics, mechanical and fracture-mechanical properties, durability properties and hygrothermal characteristics shows that the optimal amount of ceramic powder in the mix is 20 % of the mass of blended cement. The decisive parameters in that respect are compressive strength, liquid water transport parameters and resistance to de-icing salts, which are not satisfactory for higher ceramics dosage in the blends. In the case of other parameters studied, the limits for the effective use of ceramic powder are higher: 40 % for effective fracture toughness and specific fracture energy, 60 % for frost resistance and chemical resistance to MgCl2, NH4Cl, Na2SO4, HCl and CO2. The water vapour diffusion coefficient is found to increase with increasing ceramics content, which for wet enve-lopes can be considered as a positive feature, but may have a negative effect for dry envelopes. The thermal conductivity of all mixes increases fast with growing moisture content; differences of up to 50 % between the dry and water-saturated state values are observed. This has to be taken into account in energy-related calculations.

Keywords: concrete, ceramic powder, compressive strength, frost resistance, chemical resistance, hygrothermal characteristics

1 Introduction

Pozzolan-active substances have been used as a part of cement-based binders for over a century [1], [2], but their share in overall concrete production was marginal for a long time. The interest in cements containing pozzolan in the building industry only began to increase over the last few decades when the role of pozzolans as supplemen-tary cementitious materials was generally accepted [3]–[6]. The motivation was both economical and environmental in most cases. Cement producers in Europe were sup-posed to comply with stricter ecological legislation, the

restrictions on emission limits in particular. The rising price of fuel for burning the clinker was another burden. Most pozzolans are cheaper than cement and their use to replace part of the cement decreases the overall CO2 consumption [7], [8]. Mixing pozzolans with Portland ce-ment could thus be considered as one way of dealing with increasing environmental and economical requirements. In addition, pozzolans were found to increase the strength and chemical resistance and reduce heat of hydration, permeability, porosity and alkali-silica expansion, which was another bonus [9].

The most frequently used pozzolans in the current concrete industry are industrial by-products such as fly ashes [10], silica fume [11], metallurgical slags [12] or waste glass [13]. Agricultural wastes, represented, for instance, by rice husk ash [14], bagasse ash [15] or sugar cane leaf [16], are being increasingly used, particularly in the regions where they are produced. Some extracted nat-ural materials can also exhibit pozzolanic properties, e.g. diatomite [17], volcanic ash [18], zeolitic tuff [19] or zeolite [20]. One possible application of lunar mineral materials for concrete production was discussed as well [21].

Some clays with a high content of clayey minerals react with lime even at normal temperatures. Such clays are used for the stabilization of soils, but with respect to the physical properties of reaction products (swelling, shrinkage during drying), these cannot be added to con-crete. When clayey minerals are heated to temperatures > 600 °C, imperfectly crystalline or amorphous dehydrated aluminosilicates are formed [9]. These aluminosilicates are pozzolan-active. Kaolinite, which is converted to me-takaolin, a pozzolan-active substance, at a temperature of approx. 600 °C, may serve as an example [22]. However, although the energy-related demand for metakaolin pro-duction (1.3 GJ per tonne) is lower than for cement (3.4–4.0 GJ per tonne), it is still considerable, which makes its price rather high. In fact, the price of metakaolin is mostly significantly higher than cement, which is partly due to its lower production amounts and partly due to the market strategies of the particular producers. Therefore, the main motivation for using metakaolin in concrete is more technological than environmental or economic [23]. The well-known high durability of metakaolin concrete has been a sound argument for its increasing applica-tion in the building industry in recent years. Metakaolin increases the sulphate resistance of concrete in particular

Technical Paper

Mechanical, durability and hygrothermal properties of concrete produced using Portland cement-ceramic powder blends

Tereza KulovanáEva VejmelkováMartin KeppertPavla RovnaníkováZbynek KeršnerRobert Cerný*

DOI: 10.1002/suco.201500029


Submitted for review: 05 March 2015; revision: 16 June 2015; accepted for publication: 16 July 2015. Discussion on this paper must be submitted within two months of the print publication. The discussion will then be published in print, along with the authors’ closure, if any, approximately nine months after the print publication.

106

T. Kulovaná/E. Vejmelková/M. Keppert/P. Rovnaníková/Zbynek Keršner/R. Cerný · Mechanical, durability and hygrothermal properties of concrete produced using Portland cement-ceramic powder blends


The compositions of the concrete mixes are shown in Table 2. Taking into account the prices of the particular components of the mixes which are common in the Czech Republic (cement CZK 2500/tonne, 0–4 mm aggregates CZK 200/tonne, 4–8 mm aggregates CZK 380/tonne, 8–16 mm aggregates CZK 300/tonne, Dynamon SX su-perplasticizer CZK 120/kg, water CZK 18/tonne, waste ceramic powder CZK 0/tonne) results in – after recalcula-tion (€1 ~ CZK 28) – a cost of €66.8/m3 for the reference concrete used in this paper (without shipping), whereas for the HPC mix used in [27], it was €81.6/m3. For the de-signed mix, the savings related to the use of waste ceramic powder instead of cement are €3.2/m3 of concrete for every 10 % Portland cement replacement.

The XRD analysis of concrete mixes based on the Portland cement-ceramic powder blends performed by a PANalytical Empyrean device for the characteristic case of concrete with 20 % replacement of Portland cement by ceramic powder showed the presence of quartz, mus-covite, amphibole, albite, orthoclase, portlandite, calcite, gypsum and a residue of non-hydrated clinker minerals (C3S, C2S and C3A). The CSH phases formed from amor-phous components of ceramic powder and portlandite were not identified by XRD analysis due to their amor-phous state.

because the replacement of cement by metakaolin reduces the concentration of C3A, which reacts with sulphates to form ettringite.

Ceramic powder obtained by milling production residues from the ceramic industry such as red-clay bricks or floor and roof tiles to a suitable fineness [24]–[26] can be considered as a cheaper but almost equivalent alterna-tive to metakaolin. As has been shown in [27], at lower Portland cement replacement levels (up to 10 %), the mechanical, fracture-mechanical and durability properties of concrete containing ceramic powder and metakao-lin were virtually the same. Nevertheless, the present concrete industry uses ceramic powder as a supplemen-tary cementitious material only exceptionally. The reasons may be found in its lower availability in the form of either a constituent in blended cement or a ready-to-use product. But the potential for future applications is considerable. Besides the red ceramics, for instance, ceramic sanitary ware [28] or porcelain stoneware tiles [29] can also be used as sources of ceramic powder. Another possibility is the application of milled ceramic demolition waste [30], although in this case, possible contamination has to be taken into account as it could have an adverse affect on quality.

In this paper we introduce cost-effective concrete mixes containing waste ceramic powder which have func-tional properties approaching those of high-performance concretes. A wide set of parameters is analysed – includ-ing basic material characteristics, mechanical and frac-ture-mechanical properties, frost resistance, de-icing salt resistance, chemical resistance to MgCl2, NH4Cl, Na2SO4, HCl and CO2, water and water vapour transport proper-ties, thermal conductivity and specific heat capacity – in order to obtain complex information on the concretes analysed.

2 Materials and mix design

The Portland cement-ceramic powder blends were pre-pared using CEM I 42.5 R and ceramic powder from a brick factory in the Czech Republic. The ceramic powder was a waste material from grinding advanced types of hollow brick. As it does not need any further processing and is normally landfilled, its current cost in the Czech Republic is equal to the shipping costs only. For a com-parison, the ordinary cost of fly ash (without shipping) in the Czech Republic is €3 per tonne, for metakaolin it is €357 per tonne and for cement CEM I 42.5 it is €90 per tonne. The amount of ceramic powder in the blends was within the range 10–60 % by mass. The particle size distribution of cement and ceramics is presented in Fig. 1 and the chemical composition is given in Table 1. The spe-cific surface of ceramic powder was 402 m2/kg. Crushed granodiorite was used as coarse aggregate (Olbramovice, 2640 kg/m3), sedimentary river psefites as fine aggregate (Žabcice, 2580 kg/m3).

The aim of the mix design was to achieve cost-effective mixes with functional properties approaching those characteristic of high-performance concretes (HPC). Therefore, significantly lower amounts of cement and plasticizer were used than in the case of common HPCs and the granulometric curve was adjusted accordingly.

Fig. 1. Particle size distribution of cement and ceramic powder

Table 1. Chemical composition of cement and ceramic powder

Component Cement [% by mass]

Ceramic powder [% by mass]

SiO2 18.89 66.40

Al2O3 4.24 13.39

Fe2O3 3.83 4.66

CaO 62.37 3.66

MgO 0.99 1.67

SO3 2.31 1.09

Na2O 0.12 0.92

K2O 1.14 2.44

TiO2 0.30 0.70

MnO 0.077 0.073

P2O5 0.12 0.13

Loss on ignition 1.52 0.00

107



about the same size, the porosimetry measurements were performed on samples without coarse aggregates.

3.2 Mechanical and fracture-mechanical properties

The compressive strength was measured with the VEB WPM Leipzig hydraulic testing device, which has a stiff loading frame and a capacity of 3000 kN. The tests were performed according to CSN EN 12390-3 [32] after 7 and 28 days of curing in a conditioned laboratory at a tem-perature of 22±1 °C and 25–30 % relative humidity, then after 90, 180 and 360 days under laboratory conditions.

A three-point bending test on a specimen having a central edge notch length of about 1/3 of the depth of the specimen was used in the measurement of fracture-mechanical parameters. The effective fracture toughness was determined using the Effective Crack Model [33] and the specific fracture energy was obtained by the RILEM method [34].

3.3 Durability properties

Frost resistance tests were carried out according to CSN 73 1322/Z1 [35]. The samples were tested after 28 days of concrete maturing and standard curing. The total test required 100 freeze-thaw cycles. One cycle consisted of 4 h freezing at –20 °C and 2 h thawing in 20 °C warm wa-ter. The frost resistance coefficient K was determined as the ratio of bending or compressive strength of specimens subjected to 100 freeze-thaw cycles to the strength of ref-erence specimens that did not undergo the frost resistance test. The compressive strength was determined according to CSN EN 12390-3 [32] and the bending strength was measured using the procedure described in CSN EN 12390-5 [36].

The resistance to de-icing salts of the concretes stud-ied was measured according to CSN 731326/Z1:1984 [37]. The test specimens were saturated with water and placed in a bath with 3 % NaCl solution. Freeze-thaw cycles were then applied. In one cycle, the test specimen was first cooled from +20 °C to –15 °C over 45 min in an automatic conditioning device, then it was left at –15 °C for 15 min before being subsequently heated to +20 °C over 45 min and left for 15 min at that temperature. After every 25 cycles, the specimens were removed from the

3 Test methods3.1 Basic physical characteristics

As fundamental physical material characteristics, bulk density ρb [kgm–3], open porosity ψ0 [vol. %] and matrix density ρmat [kgm–3] were determined using the water vacuum saturation method [31]. Each sample was dried in a drier at 105 °C for one week to remove the major-ity of the physically bound water. After that, the samples were placed in a desiccator with deaerated water. Air was evacuated from the desiccator with a vacuum pump for 3 h. The sample was then kept underwater for not less than 24 h.

From the mass of the dry sample md, mass of water-saturated sample mw and mass of immersed water-saturat-ed sample ma, the volume V of the sample was determined using the following equation:

=−

ρV

m m(1)w a

w

where ρw is the density of water. The open porosity, bulk density and matrix density were calculated using Eqs. (2)–(4):

ψ =−ρ

m mV

(2)0w d

w

mV

(3)dρ =

mV(1 – )

(4)matd

0ρ =

ψ

Characterization of the pore structure was performed by mercury intrusion porosimetry. The experiments were carried out using the Pascal 140 and Pascal 440 instru-ments (Thermo Scientific). Pascal 140 is a low-pressure station used for mercury filling and measurement up to atmospheric pressure. The measuring glass container then has to be moved to the Pascal 440, the high-pressure station that can reach pressures as high as 400 MPa; the corresponding pore diameter is 3 nm. Since the size of the specimens is restricted to a volume of approx. 1 cm3 and the materials studied contained some aggregates with

Table 2. Composition of concrete mixes

Component Composition [kgm–3]

C REF CC 10 CC 20 CC 40 CC 60

CEM I 42.5 R 360 324 288 216 144

Aggregates, 0–4 mm 910 910 910 910 910

Aggregates, 4–8 mm 225 225 225 225 225

Aggregates, 8–16 mm 755 755 755 755 755

Water 146 146 146 146 146

Dynamon SX 14 3.96 3.96 4.29 5.18 6.16

Ceramic powder – 36 (10 %) 72 (20 %) 144 (40 %) 216 (60 %)

108



where i [kg/m2] is the cumulative water absorption and t the time from the beginning of the water absorption experiment.

The water absorption coefficient was then used for calculating the apparent moisture diffusivity in accor-dance with Kumaran (Eq. (6)) [39]:

where wc [kgm–3] is the saturated moisture content and w0 [kgm–3] the initial moisture content.

The wet-cup and dry-cup methods [31] were em-ployed in the measurement of water vapour transport parameters (Fig. 3). The specimens were insulated water-proof and vapour-proof by epoxy resin on all lateral sides, put into the cup and sealed with industrial plasticine. The impermeability of the plasticine sealing was achieved by heating it first for better workability and subsequent cool-ing, which caused it to harden. In the wet-cup method, the sealed cup containing saturated K2SO4 solution (with 97.8 % equilibrium relative humidity above the solution) was placed in an air-conditioned room with 50 % relative humidity and weighed periodically. The measurements were carried out at 25±1 °C over a period of four weeks. The steady-state values of mass loss determined by linear

Aw – w

(6)appc 0

2

κ ≈

bath, their mass loss due to spalling of particles on the surface was determined, the NaCl solution replaced and specimens placed in the bath again.

The chemical resistance in various environments was tested according to the procedure developed at Brno University of Technology. The specimens were prepared in 100 × 100 × 400 mm moulds and placed in a climatic chamber with a 100 % relative humidity environment. After 24 h they were demoulded and stored in the same environment for another 27 days. The specimens were then cut into 100 × 100 × 50 mm blocks and placed in groups of three in the chemical environments specified in Table 3. The concrete carbonation test was performed in a desiccator where the CO2 concentration was kept at 65±5 vol. %. The carbonation took place in an environ-ment above saturated KNO3 solution (85±5 % relative hu-midity). The reason why such a high CO2 concentration was used in the measurements was the low permeable structure of the concrete studied. For lower CO2 concen-trations, the possible extent of carbonation could not be observed in the limited time scale of the experiment. The specimens marked “distilled water” and “drinking water” in Table 3 were kept in a distilled water or drinking water bath which was replaced every 10 days. The duration of the chemical tests was 60 days. Thereafter, the specimens were subjected to a compressive strength test. The coeffi-cient of chemical resistance Kcr was determined as the ra-tio of the compressive strength after 60 days in a chemical environment and the compressive strength after 60 days under laboratory conditions. Concrete carbonation was also analysed using two additional tests. The 100 × 100 × 50 mm specimens were first cut in the middle and the carbonation depth was measured by a digital length me-ter. The specimens were then crushed and the pH values determined from the leach by the phenolphthalein test.

3.4 Hygrothermal characteristics

The liquid water transport was characterized by the water absorption coefficient and apparent moisture diffusivity. The water absorption coefficient was measured using a free water intake experiment (Fig. 2, see [38] for details). The specimen was insulated waterproof and vapour-proof on the four lateral sides and the face side was immersed 1–2 mm in the water. The automatic balance allowed re-cording of the increase of mass. The water absorption co-efficient A [kgm–2s–1/2] was calculated using the formula

i A ∙ t (5)=

Table 3. Chemical environments used in the tests

Environment Concentration

Drinking water –

Distilled water –

MgCl2 [g ∙ L–1] 17.76

NH4Cl [g ∙ L–1] 2.97

Na2SO4 [g ∙ L–1] 14.79

HCl [mol ∙ L–1] 10–3

CO2 [vol. %] 65±5

Fig. 2. Experimental setup for measurement of water absorption coefficient

109



with the surface of the sample. Measurement was done depending on the moisture content from dry state to water saturation.

The specific heat capacity of dry specimens was de-termined by the Isomet 2104 as well. The volumetric heat capacity of wet specimens ρwetcwet was calculated using Eqs. (9) and (10), based on the linear theory of mixtures and assuming that this heat storage parameter is an addi-tive quantity [9]:

where:ρwet bulk density of wet materialρdry bulk density of dry materialρw density of waterw moisture content by volume, defined by Eq. (11):

Vw volume of water in the poresV volume of whole specimencwet, cdry, cw specific heat capacities of wet material, dry

material and water respectively

So the specific heat capacity vs. moisture content function could be written as follows:

The relation between the moisture content by mass u and moisture content by volume w can be expressed by Eq. (13):

Therefore, Eq. (12) can be rewritten in terms of the mois-ture content by mass, a quantity more often used in build-ing materials science, in the form of Eq. (14):

( )1

(14)c uc c u

uwetdry w=

++

4 Results and discussion4.1 Basic physical characteristics

Table 4 shows that the open porosity of the concretes analysed was similar for the mixes with up to 20 % ce-ramic powder in the blended cements (C REF, CC 10 and CC 20). For the 40 and 60 % dosage (CC 40 and CC 60), it increased by up to ~20 %. This corresponded with the changes in bulk density. The matrix density values were within about 3 % for all materials. The pore distribution of C REF, CC 10 and CC 20 (Fig. 4) was relatively flat in the range 10–500 nm, but a distinct peak appeared for CC 40 and CC 60 at ~200 nm and additional peaks were found for CC 40 at ~10 nm and for CC 60 at ~20 nm. The

(9)c c c wwet wet dry dry w wρ ρ ρ= +

(10)wwet dry wρ ρ ρ= +

(11)wVVw=

( ) (12)c wc c w

wwetdry dry w w

dry w

ρ ρρ ρ

=++

(13)u w w

dry

ρρ

=

regression for the last five readings were used for deter-mining the water vapour diffusion coefficient. In the dry-cup method, the sealed cup containing dried CaCl2 (with 5 % equilibrium relative humidity above the desiccant) was placed in an air-conditioned room with 50 % relative humidity. Otherwise, the measurement was carried out in the same way as in the wet-cup method.

The water vapour diffusion coefficient D [m2s–1] was calculated from the measured data using the following equation:

where:∆m amount of water vapour diffused through the sample

[kg]d sample thickness [m]S specimen surface [m2]τ period of time corresponding to the transport of mass

of water vapor ∆m [s]∆pp difference between partial water vapour pressure in

the air above and below the specific specimen surface [Pa]

R universal gas constantM molar mass of waterT absolute temperature [K]

On the basis of the diffusion coefficient D, the water va-pour diffusion resistance factor µ [–] was determined as follows:

where Da is the diffusion coefficient of water vapour in air.

The thermal conductivity was measured using the commercial Isomet 2104 device (Applied Precision, Ltd.). The measurement is based on the analysis of the tem-perature response of the material analysed to heat flow impulses. The heat flow is induced by electrical heating using a resistor heater having a direct thermal contact

D m ∙ d ∙ R ∙ TS ∙ ∙ M ∙ p

(7)p

= ∆τ ∆

DD

(8)aµ =

Fig. 3. Experimental setup of cup method for measuring water vapour transport properties

110



mostly up to 10 %. The most substantial strength increase was observed over the period up to 90 days. The compres-sive strength of the concretes with higher ceramic powder content was significantly lower (30% for CC 40 and three times for CC 60 after 7 days), but the difference decreased with time so that after 360 days it was ~20 % for CC 40 and ~40 % for CC 60 compared with the reference mix. This confirmed the existence of the 20 % limit for the amount of ceramic powder in the blended cement on the one hand and the importance of the pozzolanic reaction on the other. The 20 % limit for the effective use of ceram-ic powder in blended Portland cement-based binders was found before in the calorimetric measurements reported by Tydlitát et al. [40]; it was explained by the limited par-ticipation of ceramics in the hydration process for higher dosages.

The decrease in compressive strength with the in-creasing amount of ground waste clay bricks was also ob-served by Toledo Filho et al. [24] and O’Farrell et al. [41], where mortars with a similar compressive strength to the concretes in this paper were investigated. The data meas-ured by O’Farrell et al. [41] showed a much faster decrease in compressive strength in the range up to 20 % ceramic powder after 7 and 28 days, but after 90 days the results were similar to those obtained in this paper. The 28-day strength values reported by Toledo Filho et al. [24] showed a similar trend to our mixes up to 40 % replacement level. The compressive strength development of concrete with 20 % ceramic brick powder studied by Pacheco-Torgal and Jalali [26] up to 90 days showed a decreasing difference from the reference mix which was in good qualitative agreement with our results. In a comparison with the high-performance concrete mix studied in [27], the com-pressive strength after 28 days was only 10–15 % lower, which is a good result, taking into account the difference in the cement/aggregate ratio (1:5.25 against 1:3.5 in [27]) and the lower dosage of plasticizer.

The effective fracture toughness for up to 20 % ce-ramic powder in the blend was 5–10 % higher than for the reference mix; a 3 % decrease was observed for CC 40 and 10 % decrease for CC 60 (Table 6). The specific fracture energy exhibited a similar trend, but for CC 40 it was 15 % higher and for CC 60 some 7 % lower in a comparison with the reference concrete. Thus, for fracture-mechanical properties, the limit for the effective use of ceramic pow-der in the blended cement was 40 %.

The measurement of fracture-mechanical properties of concretes containing ceramic powder as supplementary cementitious material was reported in [27] only; no other references were found in common literature sources. The reference high-performance concrete mix in [27] had

observed changes in total open porosity and pore size dis-tribution indicated that a 20 % dosage of ceramic powder in the blend may present a limiting value for substantial changes to appear in the properties of the hardened con-crete mixes studied.

The porosity of the mixes designed in this paper (Table 4) was ~30 % higher than the values reported in [24] for mortars with similar compressive strength and the same w/c = 0.40. This was probably related to the lower cement/aggregate ratio (1:1.5 against 1:5.25 in this paper) and the absence of coarse aggregates. On the other hand, the porosity of 30–35 MPa concrete with 20 % ceramic brick powder studied by Pacheco-Torgal and Jalali [26] was 20–25 % higher compared with C REF and CC 20; this was apparently due to the higher w/c of 0.6.

4.2 Mechanical and fracture-mechanical properties

The compressive strength of the CC 10 and CC 20 mixes was similar to the reference concrete C REF over the whole 360-day period (Table 5); the differences were

Table 4. Basic physical properties

Material ρ ρmat ψ0

[kgm–3] [kgm–3] [%]

C REF 2234±47 2571±26 13.1±0.3

CC 10 2263±43 2614±16 13.4±0.1

CC 20 2258±36 2613±15 13.6±0.2

CC 40 2182±40 2581±20 15.5±0.3

CC 60 2194±39 2630±21 16.6±0.3

Fig. 4. Pore size distribution

Table 5. Compressive strength [MPa]

Material 7 days 28 days 90 days 180 days 360 days

C REF 46.2±2.1 56.9±2.8 59.5±2.5 61.5±2.4 56.5±2.4

CC 10 42.7±1.8 55.9±2.5 60.1±2.3 62.1±2.8 66.6±2.4

CC 20 43.9±1.9 49.6±2.0 61.0±2.3 57.8±2.7 59.5±2.3

CC 40 29.4±1.4 37.4±1.6 46.5±1.8 47.0±2.0 46.8±1.9

CC 60 16.8±1.3 22.2±1.0 31.4±1.3 36.5±1.6 35.4±1.5

111



pores. As the volume of the smallest pores < 0.1 µm was highest for CC 40 and CC 60 (Fig. 4), these materials were damaged in a more significant way than the others.

The chemical resistance to MgCl2, NH4Cl, Na2SO4 and HCl (Table 9) was excellent for all the concretes. The coefficient of chemical resistance Kcr after 60 days was always > 0.90. The CC 20 and CC 60 mixes showed the best overall performance with Kcr > 1 in all environments. The resistance to CO2 of all the concretes was also very good. The pH values of concrete leaches and depth of carbonation are shown in Table 10. Apparently, even the high concentration of CO2 used in the tests did not lead to complete carbonation. The lowest measured pH value was 9.05, whereas the saturated solution of CaCO3, a product of carbonation, has a pH of 8.3.

The durability properties of concrete containing ceramic powder were studied only rarely in the past. Sul-phate resistance was analysed by Toledo Filho et al. [24] using the ratio of tensile strengths and by O’Farrell et al. [41] with the help of expansion tests. The results were in qualitative agreement with the experiments performed in this paper. In a comparison with the high-performance concrete containing fine-ground ceramics studied in [27], the concrete in this paper had similarly good durability properties.

15–20 % higher values of effective fracture toughness and specific fracture energy than the corresponding mix in this paper, but the mixes with the blended binders exhibited worse fracture-mechanical properties in most cases. This provides further supporting arguments for the mix design presented in this paper to those furnished by the compres-sive strength results.

4.3 Durability properties

The frost resistance of all mixes met the standard criterion of K > 0.75, which guaranteed adequate performance (Table 7). The best results for compressive strength were achieved with CC 20 and CC 40, for bending strength with C REF and CC 10. The mix with 20 % ceramic pow-der, CC 20, gave the best resistance against de-icing salts (Table 8). The results obtained for the CC 40 and CC 60 mixes were much worse. Therefore, the limit for using ceramic powder was the same here as for the compressive strength. The reason why the de-icing salt resistance of the materials analysed was much worse than their frost resist-ance was probably the higher severity of the former test. The combination of ice formation and chloride binding in the pore system induced by the repeated exposure to the NaCl solution could result in faster destruction of finer

Table 6. Fracture-mechanical properties

Material Effective fracturetoughness [MPam1/2]

Specific fracture energy [Jm–2]

C REF 1.43±0.05 242±7

CC 10 1.58±0.07 261±12

CC 20 1.50±0.05 279±14

CC 40 1.39±0.06 276±12

CC 60 1.28±0.03 225±11

Table 7. Frost resistance

Material Frost resistance coefficient K

as ratio of compressive strengths

as ratio of bending strengths

C REF 0.86±0.03 1.17±0.05

CC 10 0.88±0.03 1.08±0.04

CC 20 0.98±0.03 0.98±0.04

CC 40 1.00±0.05 0.96±0.02

CC 60 0.96±0.02 0.96±0.01

Table 8. Loss of mass due to de-icing salts action

Material/ Number of cycles

C REF CC 10 CC 20 CC 40 CC 60

Loss of mass [gm–2]

25 703±14 1930±40 578±13 2590±110 5640±110

50 3110±130 4630±190 2192±63 5220±130 9920±180

75 8300±290 6930±220 6260±310 740±200 break-up

100 16 810±670 10 750±330 17 700±750 10 240±300 break-up

Table 9. Coefficient of chemical resistance Kcr

Material Drinking water

Distilled water

Na2SO4 MgCl2 NH4Cl HCl CO2

C REF 0.91±0.02 1.00±0.01 0.92±0.01 0.94±0.01 1.01±0.01 0.99±0.01 1.23±0.01

CC 10 0.94±0.01 0.97±0.01 0.96±0.01 0.92±0.01 0.98±0.01 1.01±0.01 1.18±0.01

CC 20 1.05±0.02 1.00±0.01 1.04±0.02 1.01±0.01 1.08±0.01 1.03±0.01 1.19±0.01

CC 40 0.92±0.01 0.98±0.01 0.96±0.01 0.92±0.01 0.94±0.0 1.00±0.01 1.25±0.02

CC 60 1.05±0.01 1.07±0.01 1.03±0.01 1.11±0.03 1.04±0.02 1.07±0.01 1.09±0.01

112



water vapour inside. Therefore, the real effect of water vapour diffusion on a specific structure under specific environmental conditions should always be assessed by computational modelling. The D values obtained for the dry-cup arrangement were always lower than for the wet-cup setup. This was caused by the transport of capillary condensed water in the specimens with higher relative humidity similarly to many other materials with a high hygroscopic moisture content [9]. The results obtained were in qualitative agreement with the measurements of oxygen permeability by Pacheco-Torgal and Jalali [26], who found it increased by 10 % for concrete with 20 % ce-ramic brick powder. It should be noted in this respect that water vapour transport properties are only rarely studied in concrete-related investigations. However, this kind of parameter is necessary as input data for any detailed simulations of the hygrothermal performance of concrete structures, which are indispensable for service life assess-ment and energy performance analyses. This was dem-onstrated in a number of ’past studies for other types of concrete containing pozzolan (see, for example, [42], [43]).

The thermal conductivity λ in the dry state decreased with increasing amount of ceramic powder in the blended cement (Fig. 5). This was in qualitative agreement with the open porosity results (Table 4). The specific heat capacity c in the dry state increased slightly with increasing ce-ramic powder dosage; the maximum difference was about 8 % compared with the reference mix (Fig. 6). The thermal conductivity increased remarkably with increasing mois-ture content (Fig. 5); in the water-saturated state, λ values were up to 50 % higher than for the dry state. This finding may have significant consequences for the thermal per-formance calculations of building envelopes constructed using the concretes studied. Energy-related calculations

4.4 Hygrothermal characteristics

The water absorption coefficient A of the hardened con-crete mixes analysed increased with increasing amount of ceramic powder (Table 11). The A values were still satisfactory up to 20 % ceramic powder in the blended ce-ment, where an increase in A of up to 20 % was observed; for CC 60, A was already approximately two times higher than for the reference mix. This agreed, in a qualitative way, with the results obtained for open porosity (Table 4) and compressive strength (Table 5). The values of appar-ent moisture diffusivity followed the same trend as the water absorption coefficient, which was used for their calculation.

Toledo Filho et al. [24] found the water sorptivity of mortar to decrease with increasing amount of ground brick for up to 40 % of mass of cement. Pacheco-Torgal and Jalali [26] observed a 10 % decrease in the water permeability of concrete with 20 % ceramic brick powder. This contradicts the results presented in this paper and in [27]. The differences can be explained by the different mix design technology in [24] (mortar specimens with sand aggregates only, 1:1.5 cement/aggregate ratio) and by the use of a different measuring method in [26] (standard manual test to LNEC E393).

The water vapour diffusion coefficient D increased with increasing content of ceramic powder in the mix (Table 12), which agreed with the open porosity measure-ments (Table 4). This finding is not unambiguous from the point of view of the hygrothermal performance of multi-layered concrete building envelopes. For a wet en-velope, higher diffusion coefficients are beneficial because water vapour is removed faster, but for dry envelopes, more intensive water vapour diffusion may result in more

Table 10. pH value and depth of carbonation

Material pH Depth of carbonation [mm]

C REF 11.91±0.10 1.3±0.2

CC 10 11.71±0.10 4.5±0.2

CC 20 11.69±0.10 5.5±0.2

CC 40 9.85±0.10 14.5±0.2

CC 60 9.05±0.10 24.8±0.2

Table 11. Water transport properties

Material A κ

[kgm–2s–1/2] [m2 s–1]

C REF 0.0066±0.0001 2.66E-09±0.10E-09

CC 10 0.0068±0.0001 2.56E-09±0.12E-09

CC 20 0.0079±0.0002 3.85E-09±0.16E-09

CC 40 0.01070±0.0003 5.50E-09±0.27E-09

CC 60 0.0135±0.0003 8.16E-09±0.39E-09

Table 12. Water vapour transport properties

Material 5/50 % 97/50 %

D µ D µ

[m2s–1] [–] [m2s–1] [–]

C REF 2.67E-07±0.10E-07 86.4±3.7 5.54E-07±0.24E-07 41.5±2.0

CC 10 2.77E-07±0.10E-07 84.3±3.0 5.83E-07±0.22E-07 39.5±1.9

CC 20 2.92E-07±0.12E-07 78.9±2.8 6.09E-07±0.25E-07 37.8±1.7

CC 40 3.18E-07±0.15E-07 72.4±2.0 7.44E-07±0.27E-07 31.2±1.3

CC 60 3.59E-07±0.17E-07 64.1±1.9 1.02E-06±0.05E-06 22.5±0.9

113



– The shape of the pore size distribution curve was changed and the open porosity increased significantly for the specimens with a ceramic powder content > 20 % in the blends. This indicated that a 20 % dosage of waste ceramics in the blend presents a limiting value for substantial changes to appear in the engineering properties of the hardened concrete mixes studied.

– The compressive strength of concrete mixes with up to 20 % ceramic powder in the blended cement was similar to the reference concrete over the whole 360-day period of the study. It decreased in a significant way for higher waste ceramics dosage. On the other hand, for fracture-mechanical properties, the limit for the effective use of ceramic powder in the blends was 40 %.

– The frost resistance and chemical resistance to MgCl2, NH4Cl, Na2SO4, HCl and CO2 were excellent for all the concrete mixes analysed. The resistance to de-icing salts was satisfactory only up to 20 % ceramic powder in the blend.

– The water absorption coefficient and apparent moisture diffusivity of the hardened concrete mixes analysed in-creased with increasing amount of ceramic powder; the values obtained were satisfactory up to 20 % ceramic powder in the blended cement.

– The water vapour diffusion coefficient increased with increasing content of ceramics, which for wet envelopes could be considered as a positive feature, but could have a negative effect for dry envelopes.

– The thermal conductivity of all the mixes analysed increased by up to 50 % with increasing moisture con-tent. This might have significant consequences in the hygrothermal performance calculations for building en-velopes constructed using the concretes studied. On the other hand, for the specific heat capacity, the maximum difference between the dry and water-saturated states was only 25 %.

The experimental results summarized above show that al-though from both the environmental and economic points of view the amount of waste ceramics in the concrete mixes should be as high as possible, a limit for its effective use in the blends is 20 %.

Acknowledgments

This research was supported by the Czech Science Foun-dation under project No 14-04522S.

References

1. Massazza, F.: Pozzolanic cements. Cement and Concrete Composites, 1993, 15, pp. 185–214.

2. Malhotra, V. M., Hemmings, R. T.: Blended cements in North America: A review. Cement and Concrete Compos-ites, 1995, 17, pp. 23–35.

3. Lothenbach, B., Scrivener, K., Hooton, R. D.: Supplementary cementitious materials. Cement and Concrete Research, 2011, 41, pp. 1244–1256.

4. Scrivener, K. L., Nonat, A.: Hydration of cementitious mate-rials, present and future. Cement and Concrete Research, 2011, 41, pp. 651–665.

are mostly performed with only constant λ values corre-sponding to the dry state, which may underestimate the real energy consumption within a range of at least 5–10 %, as has been demonstrated, for instance, in [44] for several building envelopes based on autoclaved aerated concrete. The specific heat capacity increased with increasing mois-ture as well (Fig. 6), which was caused by the high spe-cific heat capacity of water. The maximum difference in c values between the dry and water-saturated states was approx. 25 %.

5 Conclusions

An extensive set of parameters of hardened concrete mixes prepared using blended Portland cement-ceramic powder binders containing up to 60 % fine-ground waste ceramics from a brick factory was presented in this paper. The experimental analysis of basic physical characteris-tics, mechanical and fracture-mechanical properties, dura-bility properties and hygrothermal characteristics resulted in the following main findings:

Fig. 5. Thermal conductivity as a function of moisture content

Fig. 6. Specific heat capacity as a function of moisture content

114



5. Papadakis, V. G., Tsimas, S.: Supplementary cementing ma-terials in concrete: Part I. Efficiency and design. Cement and Concrete Research, 2002, 32, pp. 1525–1532.

6. Glavind, M.: Green concrete structures. Structural Con-crete, 2011, 12, pp. 23–29.

7. Benhelal, E., Zahedi, G., Shamsaei, E., Bahadori, A.: Global strategies and potentials to curb CO2 emissions in cement industry. Journal of Cleaner Production, 2013, 51, pp. 142–161.

8. Schneider, M., Romer, M., Tschudin, M., Bolio, H.: Sustain-able cement production – present and future. Cement and Concrete Research, 2011, 41, pp. 642–650.

9. Cerný, R., Rovnaníková, P.: Transport Processes in Con-crete, Spon Press, London, 2002.

10. Yildirim, H., Sümer, M., Akyüncü, V., Gürbüz, E.: Compari-son on efficiency factors of F and C types of fly ashes. Con-struction and Building Materials, 2011, 25, pp. 2939–2947.

11. Chung, D. D. L.: Improving cement-based materials by using silica fume. Journal of Materials Science, 2002, 37, pp. 673–682.

12. Maes, M., Gruyaert, E., De Belie, N.: Resistance of concrete with blast-furnace slag against chlorides, investigated by comparing chloride profiles after migration and diffusion. Materials and Structures, 2013, 46, pp. 89–103.

13. Matos, A. M., Sousa-Coutinho, J.: ASR and sulphate perfor-mance of mortar containing industrial waste. Structural Concrete, accepted manuscript, online: 20 Mar 2015, doi 10.1002/suco.201400095.

14. Carmona, V. B., Oliveira, R. M., Silva, W. T. L., Mattoso, L. H. C., Marconcini, J. M.: Nanosilica from rice husk: Extrac-tion and characterization. Industrial Crops and Products, 2013, 43, pp. 291–296.

15. Ganesan, K., Rajagopal, K., Thangavel, K.: Evaluation of bagasse ash as supplementary cementitious material. Ce-ment & Concrete Composites, 2007, 29, pp. 515–524.

16. Guzmán, A., Gutiérrez, C., Amigó, V., Mejía de Gutiérrez, R., Delvasto, S.: Pozzolanic evaluation of the sugar cane leaf. Materiales de Construcción, 2011, 61, pp. 213–225.

17. Yılmaz, B., Ediz, N.: The use of raw and calcined diatomite in cement production. Cement & Concrete Composites, 2008, 30, pp. 202–211.

18. Siddique, R.: Effect of volcanic ash on the properties of ce-ment paste and mortar. Resources, Conservation and Recy-cling, 2011, 56, pp. 66–70.

19. Yılmaz, B., Ucar, A., Oteyaka, B., Uz, V.: Properties of ze-olitic tuff (clinoptilolite) blended Portland cement. Building and Environment, 2007, 42, pp. 3808–3815.

20. Perraki, T., Kontori, E., Tsivilis, S., Kakali, G.: The effect of zeolite on the properties and hydration of blended cements. Cement & Concrete Composites, 2010, 32, pp. 128–133.

21. Wilhelm, S., Curbach, M.: Review of possible mineral mate-rials and production techniques for a building material on the moon. Structural Concrete, 2014, 15, pp. 419–428.

22. Siddique, R., Klaus, J.: Influence of metakaolin on the prop-erties of mortar and concrete: A review. Applied Clay Sci-ence, 2009, 43, pp. 392–400.

23. Vejmelková, E., Pavlíková, M., Keppert, M., Keršner, Z., Rovnaníková, P., Ondrácek, M., Sedlmajer, M., Cerný, R.: High Performance Concrete with Czech Metakaolin: Experi-mental Analysis of Strength, Toughness and Durability Char-acteristics. Construction and Building Materials, 2010, 24, pp. 1404–1411.

24. Toledo Filho, R. D., Gonçalves, L. P., Americano, B. B., Fairbairn, E. M. R.: Potential for use of crushed waste cal-cined-clay brick as a supplementary cementitious material in Brazil. Cement and Concrete Research, 2007, 37, pp. 1357–1365.

25. Lavat, A. E., Trezza, M. A., Poggi, M.: Characterization of ceramic roof tile wastes as pozzolanic admixture. Waste Management, 2009, 29, pp. 1666–1674.

26. Pacheco-Torgal, F., Jalali, S.: Compressive strength and dura-bility properties of ceramic wastes based concrete. Materials and Structures 2011, 44, pp. 155–167.

27. Vejmelková, E., Keppert, M., Rovnaníková, P., Ondrácek, M., Keršner, Z., Cerný, R.: Properties of high performance concrete containing fine-ground ceramics as supplementary cementitious material. Cement and Concrete Composites, 2012, 34, pp. 55–61.

28. Medina, C., Banfill, P. F. G., Sánchez de Rojas, M. I., Frías, M.: Rheological and calorimetric behaviour of cements blended with containing ceramic sanitary ware and con-struction/demolition waste. Construction and Building Ma-terials, 2013, 40, pp. 822–831.

29. Bignozzi, M. C., Bonduà, S.: Alternative blended cement with ceramic residues: Corrosion resistance investigation on reinforced mortar. Cement and Concrete Research, 2011, 41, pp. 947–954.

30. Katzer, J.: Strength performance comparison of mortars made with waste fine aggregate and ceramic fume. Construc-tion and Building Materials, 2013, 47, pp. 1–6.

31. Roels, S., Carmeliet, J., Hens, H., Adan, O., Brocken, H., Cerný, R., Pavlík, Z., Hall, C., Kumaran, K., Pel, L., Plagge, R.: Interlaboratory Comparison of Hygric Properties of Po-rous Building Materials. Journal of Thermal Envelope and Building Science, 2004, 27, pp. 307–325.

32. CSN EN 12390-3: Testing of hardened concrete – Part 3: Compressive strength. Czech Standardization Institute, Prague, 2002.

33. Karihaloo, B. L.: Fracture Mechanics of Concrete, Longman Scientific & Technical, New York, 1995.

34. RILEM Committee 50-FMC: Determination of the Fracture Energy of Mortar and Concrete by Means of Three-Point Bend Test on Notched Beams. Materials and Structures, 1985, 18, pp. 258–290.

35. CSN 73 1322/Z1:1968: Concrete testing – Hardened con-crete – Frost resistance. Czech Standardization Institute, Prague, 2003.

36. CSN EN 12390-5: Testing of hardened concrete – Part 5: Bending strength. Czech Standardization Institute, Prague, 2007.

37. CSN 731326/Z1:1984: Determination of the resistance of the surface of concrete against water and de-icing salts. Czech Standardization Institute, Prague, 2003.

38. Vejmelková, E., Pavlíková, M., Jerman, M., Cerný, R.: Free Water Intake as Means of Material Characterization. Jour-nal of Building Physics, 2009, 33, pp. 29–44.

39. Kumaran, M. K.: Moisture Diffusivity of Building Materials from Water Absorption Measurements. Journal of Thermal Envelope and Building Science, 1999, 22, pp. 349–355.

40. Tydlitát,V., Zákoutský, J., Volfová, P., Cerný, R.: Hydration heat development in blended cements containing fine-ground ceramics. Thermochimica Acta, 2012, 543, pp. 125–129.

41. O’Farrell, M., Sabir, B. B, Wild, S.: Strength and chemical resistance of mortars containing brick manufacturing clays subjected to different treatments. Cement and Concrete Composites, 2006, 28, pp. 790–799.

42. Madera, J., Kocí, J., Vejmelková, E., Cerný, R., Rovnaníková, P., Ondrácek, M., Sedlmajer, M.: Influence of material char-acteristics of concrete and thermal insulation on the service life of exterior renders. WIT Transactions on Modelling and Simulation, 2009, 48, pp. 13–23.

43. Kocí, V., Jerman, M., Madera, J., Cerný, R.: Effect of Zeolite Admixture on Freeze/thaw Resistance of Concrete Exposed

115



Pavla RovnaníkováInstitute of ChemistryFaculty of Civil EngineeringBrno University of TechnologyŽižkova 17, 602 00 BrnoCzech Republic

Zbynek KeršnerInstitute of Structural MechanicsFaculty of Civil EngineeringBrno University of TechnologyVeverí 95, 602 00 BrnoCzech Republic

Robert CernýDepartment of Materials Engineering and ChemistryFaculty of Civil EngineeringCzech Technical University in PragueThákurova 7, 166 29 Prague 6Czech [email protected]

to the Dynamic Climatic Conditions. Advanced Materials Research, 2014, 982, pp. 27–31.

44. Kocí, V., Madera, J., Korecký, T., Jerman, M., Cerný, R.: Effect of moisture dependent thermal and hygric parameters on the

Tereza KulovanáDepartment of Materials Engineering and ChemistryFaculty of Civil EngineeringCzech Technical University in PragueThákurova 7, 166 29 Prague 6Czech Republic

Eva VejmelkováDepartment of Materials Engineering and ChemistryFaculty of Civil EngineeringCzech Technical University in PragueThákurova 7, 166 29 Prague 6Czech Republic

Martin KeppertDepartment of Materials Engineering and ChemistryFaculty of Civil EngineeringCzech Technical University in PragueThákurova 7, 166 29 Prague 6Czech Republic

moisture and temperature fields in multi-layered systems of building materials. WIT Transactions on Modelling and Simulation, 2013, 79, pp. 91–102.


Technical Paper

Faiz Shaikh DOI: 10.1002/suco.201500030

This paper presents the preliminary results of the effect of ul-trafine fly ash (UFFA) on the properties of concretes containing recycled coarse aggregates (RCA) originating from construction and demolition (C&D) wastes. The effect of 10 % UFFA on the compressive strength, tensile strength, sorptivity and chloride ion permeability of concretes containing 25 and 50 % RCA is evaluated at 7, 28 and 56 days. The addition of UFFA increased the compressive strength of recycled aggregate concretes at all ages up to 56 days. However, a slight reduction in the ten-sile strength of recycled aggregate concretes was observed. Concrete containing 25 % RCA and 10 % UFFA achieved 94 % of the control concrete’s compressive strength at 56 days. In both recycled aggregate concretes tested, the sorptivity and chloride ion permeability are much lower at all ages due to the addition of 10 % UFFA. This is because it serves to promote hydration and block the large capillary pores within the concrete.

Keywords: ultrafine fly ash, recycled aggregates, construction and demolition wastes, mechanical and durability properties, concrete

1 Introduction

One of the most widely used materials in the construc-tion industry is concrete, which plays an important role in the advancement of civilization. It also consumes large amounts of natural resources and energy as aggregates and limestone (the main ingredient for cement) are mined from the ground. The construction industry in Australia is undoubtedly a fast-growing industry due to the increas-ing demand for infrastructure as a result of the drastic increase in population growth in recent and coming years, predicted to be 72 % between 2007 and 2030 [1], as well as the anticipated growth in and demand for infrastruc-ture in the future. The global use of concrete is also rising for the same reasons. Coarse aggregate is one of the main constituents of concrete, accounting for more than two-thirds of the concrete’s volume and playing an important role in concrete’s strength [2].

Construction and demolition (C&D) waste accounts for about 40 % of Australia’s solid waste materials, with 160 million tonnes of virgin aggregates being mined every year [3]. In 2008 a total of 19 million tonnes of C&D waste was disposed of in Australia. Only 10.5 million tonnes, or 55 %, of this waste was recycled or recovered, leaving 45 %, or 8.5 million tonnes, of waste sent to be sent to landfills around Australia. When looking at West-ern Australia alone, only 29 % of its C&D waste was re-covered in 2008, with a target recovery rate of 100 % by the year 2015 [4]. With increasing target recovery of C&D wastes in future and the shortage of natural aggregates in some parts of the world, there is a need to develop recy-cled aggregates from C&D waste as an alternative source of aggregates.

Extensive studies have been conducted around the world on the properties of concretes containing recycled coarse aggregates (RCA) and recycled fine aggregates (RFA) [5–10]. In every case, the results show that the prop-erties of those concretes are inferior to those of concretes made with natural virgin aggregates. Several reasons for this have been identified, e.g. the presence of porous old mortar/cement paste adhering to the aggregate’s surface, the existence of microcracks in the aggregates which develop during the crushing process, the formation of a weak interfacial transition zone between the old mortar/paste and the new paste/mortar, etc. [11–12]. Studies have also been conducted to improve the properties of con-cretes containing recycled aggregates through the use of supplementary cementing materials in the mix as partial replacement for cement. Kou et al. [13–14] evaluated the effect of class F fly ash as a partial replacement for cement and as an addition to cement in concretes containing recycled coarse aggregates originating from broken old concrete parts. That study used 25 and 35 % fly ash ce-ment replacement, and in another study 25 % fly ash was added to the cement. The results show that in the case of cement replacement by fly ash, the compressive strength, tensile strength and elastic modulus of concretes contain-ing different RCA after up to 28 days curing are lower than those containing no fly ash. However, those strength properties increase after 28 days of curing. In the case of fly ash addition, strength and durability properties are improved in recycled aggregates concretes for both early and later ages. In another study, Berndt [15] evaluated the properties of recycled aggregate concretes containing

Effect of ultrafine fly ash on the properties of concretes containing construction and demolition wastes as coarse aggregates


Submitted for review: 05 March 2015; revision: 27 April 2015; accepted for publication: 08 June 2015. Discussion on this paper must be submitted within two months of the print publication. The discussion will then be published in print, along with the author’s closure, if any, approximately nine months after the print publication.

117

F. Shaikh · Effect of ultrafine fly ash on the properties of concretes containing construction and demolition wastes as coarse aggregates


high-volume slag as partial replacement for cement and reported improvements to the mechanical and durability properties. Corinaldesi and Moriconi [16] evaluated the effects of 30 % fly ash and 15 % silica fume as additions to cement in concretes containing recycled coarse aggre-gates originating from C&D waste. They, too, observed improvements to the mechanical properties of recycled aggregate concretes at later ages due to the addition of fly ash and silica fume. In recent studies, the effects of 30 and 40 % fly ash cement replacement on the properties of concretes containing recycled coarse aggregates [17] and recycled fine aggregates [18] originating from C&D waste have been evaluated and the results observed are very similar to those reported by Corinaldesi and Mori-coni [16]. Improvements to the mechanical properties of recycled aggregates concretes due to the addition of class C fly ash as partial replacement for cement have also been reported by Tangchirapat et al. [19]. In a recent study, the author also evaluated the feasibility of improv-ing the properties of concretes containing C&D waste as recycled coarse aggregates using nano silica and observed promising improvements to the mechanical and durability properties [20].

Ultrafine Fly Ash (UFFA) has recently been devel-oped as a type of fine pozzolanic material. It is produced by a proprietary separation system with a mean particle diameter of 1−5 microns and contains > 20 % amorphous silica than typical class F fly ash [21]. Based on previ-ous experimental research, a beneficial effect has been reported that UFFA was able to enhance the compressive strength when used as a partial replacement for cement in concrete with a low w/c ratio. Obla et al. [21] stud-ied the effect of UFFA on the compressive strength and durability properties of concretes. They concluded that UFFA concrete has a higher strength and a tendency to minimize the alkali-silica reaction expansion. They also found that the strength activity index at 7 and 28 days was 25–30 % higher than the unprocessed fly ash. Hos-sain et al. [22] observed the effect of UFFA in comparison to silica fume in the concrete. It was revealed that re-placing cement with 12 % UFFA improved the cracking resistance in comparison to conventional Portland ce-ment concrete and silica fume concrete. Subramaniam et al. [23] also carried out an experimental investigation of UFFA concrete with two percentages of cement replace-ment: 8 and 12 %. It was concluded that the compressive strength of 8 % UFFA was slightly low after 1 day but did not hamper the long-term strength development. On the other hand, increasing the UFFA content to 12 % resulted in an increase in the resistance to shrinkage cracking. Similarly, Choi et al. [24] also reported that the compressive strength of concrete increased with increas-ing fineness of fly ash content. The compressive strength was lower than the control mix prior to 7 days and higher after 14 days. From their investigations it was clear that the improvements to properties were mainly caused by the fly ash fineness and became a main factor contribut-ing to compressive strength. Supit et al. [25] also evalu-ated the effects of different amounts (5–15 %) of UFFA as partial cement replacement on the compressive strengths of cement mortars and found that 8 % UFFA cement replacement is the optimum content. However, although

extensive research has been carried out on the properties of concretes containing ultrafine fly ash as partial cement replacement, fewer studies exist which evaluate the effec-tiveness of ultrafine fly ash for improving the properties of concretes containing recycled coarse aggregates ob-tained from C&D waste. This paper presents the results of the effectiveness of ultrafine fly ash as partial cement replacement for improving the mechanical and durability properties of concretes containing 25 and 50 % recycled coarse aggregates sourced from C&D wastes as partial replacement for natural coarse aggregates.

2 Experimental programme 2.1 Materials

Ordinary Portland cement (OPC) was used in all mixes. Commercially available ultrafine fly ash (UFFA) was used for partial cement replacement. Table 1 summarizes the quantitative X-ray diffraction (XRD) analysis of the crystalline minerals of conventional class F fly ash and ultrafine fly ash such as hematite, maghemite-c, mullite and quartz. It also shows that ultrafine fly ash has an approx. 81 % amorphous content, which is about 22 % higher than the amorphous content of class F fly ash. The recycled coarse aggregate (RCA) was obtained from a local C&D waste recycling plant in Perth, Western Aus-tralia. Table 2 shows the analysis of the contents of a 5 kg sample of the C&D waste used for RCA in this study. It can be seen that approx. 78 % are concrete and the rest consisted of masonry, asphalt and other materials. The properties of recycled and natural coarse aggregates are also shown in Table 2. As expected, the RCA had higher water absorption, lower bulk density and more than ac-ceptable amounts regarding brick content according to Australian Standard HB-155:2002 [26]. However, it met the grading requirements for concrete aggregates specified

Table 1. Phase abundance (% by weight) of ultrafine fly ash samples

Phase % by weight

Class F fly ash Ultrafine fly ash

Hematite 1.7 −Maghemite-C 2.8 0.7Mullite 16.8 6.0Quartz 15.0 11.7Amorphous content 63 81

Table 2. Properties of aggregates

Properties measured RCA NCA

Water absorption (%) 4.88 0.45Uncompacted bulk density (kg/m3) 1700 2000

Compacted bulk density (kg/m3) 1900 2400

Constituents of RCA (% by wt.)

Concrete Brick Asphalt Other78.7 13 2.3 5.7

118



in Australian Standard 2758.1 [27] (Fig. 1). Both recycled and natural course aggregates were soaked in water for 48 h, washed thoroughly to remove impurities and dried afterwards to maintain the saturated surface dry (SSD) condition. Table 3 shows the chemical analysis and physi-cal properties of the OPC and UFFA used in this study.

2.2 Mix proportions

In total, six series of mixes were considered in this study. The first series was the control mix containing 100 % natural coarse aggregates and 100 % OPC. The second series was similar to the first series in every respect ex-cept for the OPC, where 10 % UFFA was used as partial cement replacement in order to evaluate its effect on the concrete properties. The third and fourth mixes were concretes containing 25 % RCA and 50 % RCA as partial replacement for natural coarse aggregates. The effects of 10 % UFFA as partial cement replacement on the proper-ties of the concretes in the third and fourth series were evaluated in the fifth and sixth series respectively. Details of all mixes are shown in Table 4. The 10 % UFFA content used in this study is a slightly conservative estimate based on the author’s recent study of the effect of UFFA on the compressive strength of ordinary cement mortars [25], where 8 % UFFA was found to be the optimum content.

2.3 Concrete casting and curing

All concretes were mixed in a pan mixer using a constant water-to-binder ratio of 0.45. The mixing sequence of the concretes was as follows: both fine and coarse aggregates were mixed dry for 2–3 min, OPC and UFFA and one-third of the water required were then added to the dry mix and mixed for another 3 min followed by the addition of the remaining two-thirds of the water and mixing of the whole mix for a further 3 min. The reason for adding one-third of the water at the second stage was to minimize the dust during dry mixing. Slump tests were conducted im-mediately after mixing the concrete to measure the work-ability of each mix. At least three specimens were cast and tested in each series. All specimens were water-cured until the day before the test dates.

The compressive strength, indirect tensile strength, water sorptivity and chloride ion permeability were meas-ured at 7, 28 and 56 days in each series. The compressive

Fig. 1. Sieve analysis of natural coarse and recycled coarse aggregates

Table 3. Chemical composition and physical properties of materials

Chemical analysis

Cement (%)

Ultrafine fly ash (%)

SiO2 20.2 73.4Al2O3 4.9 17.7Fe2O3 2.8 4.4CaO 63.9 0.9MgO 2.0 0.6MnO – < 0.1K2O – 1.03Na2O – 0.11P2O5 – 0.2TiO2 – 0.7SO3 2.4 0.2

Physical properties

Particle size 25–40 % ≤ 7 µm mean size 3.4 µmSpecific gravity 2.7–3.2 2.0–2.55Surface area (m2/g) – 2.51Loss on ignition (%) 2.4 0.6

Table 4. Concrete mix proportions

Proportion

Series 1 100 % OPC

2 90 % OPC+ 10 % UFFA

3 25 % RCA

4 50 % RCA

5 25 % RCA+ 10 % UFFA

6 50 % RCA+ 10 % UFFA

RCA replacement (%) − − 25 50 25 50UFFA (%) − 10 − − 10 10RCA (kg/m3) − − 304 607 304 607NCA (kg/m3) 1215 1215 911 607 911 607Fine aggregate (kg/m3) 654 654 654 654 654 654OPC (kg/m3) 430 387 430 430 387 387UFFA (kg/m3) − 43 − − 43 43Water/binder ratio 0.45 0.45 0.45 0.45 0.45 0.45Slump (mm) 150 170 165 150 185 175

119



strength test was carried out on ∅100 × 200 mm cylin-ders and the indirect tensile strength on ∅150 × 300 mm cylinders. The ∅100 × 200 mm cylinder was cut into three 50 mm thick slices and used for the water sorptivity and chloride permeability tests. The compression and tensile tests were conducted in accordance with the relevant Aus-tralian standards (AS1012.3.1 [28] and AS1012.10 [29]).

2.4 Water sorptivity

The rate of water absorption (sorptivity) of concrete sam-ples in the form of a 50 mm thick disk was determined at 7, 28 and 56 days according to ASTM C1585 [30]. The principle of the method is that a specimen had one surface in free contact with water (no more than 5 mm above the base of the specimen) while the other sides were sealed. This test determined the rate of absorption of water by hy-draulic cement concrete by measuring the increase in the mass of a specimen resulting from absorption of water as a function of time. In this study the mass of the concrete specimen was measured regularly to determine the initial absorption from 1 min to the first 6 h. The absorption I was the change in mass divided by the product of the cross-sectional area of the test specimen and the density of water. The initial rate of water absorption value (mm/sec1/2) was calculated as the slope of the line that is the best fit to I plotted against the square root of time (sec1/2).

2.5 Rapid chloride permeability

The chloride ion penetration resistance of concrete, popu-larly called the rapid chloride permeability test (RCPT), was conducted according to ASTM C1202 [31]. The speci-mens from each series were tested after 7, 28 and 56 days of water curing. Details of the procedure from preparation of specimens to test can be found in ASTM C1202 [31].

3 Results and discussion 3.1 Compressive and tensile strengths

The effects of 10 % UFFA on the compressive and ten-sile strengths of concretes containing natural coarse ag-gregates (NCA) and recycled coarse aggregates (RCA) measured at 7, 28 and 56 days are shown in Figs. 2 and 3 respectively. It can be seen that the improvement in

compressive strength is low between 3 and 5 % due to the addition of 10 % UFFA as partial cement replacement in the concrete containing 100 % NCA. The improvement at 7 days is higher than at other ages, i.e. 28 and 56 days, and can be attributed to the high fineness and high amorphous content of such UFFA. The addition of UFFA may also act as the nucleation and growth sites for the C-S-H and therefore the crystallization of portlandiate is more disori-ented [32]. Chindaprasirt et al. [33] also reported reduced pore sizes and higher numbers of nucleation sites for the hydration products in cement paste containing finer fly ash than those of original fly ash. In the case of tensile strength, no improvement is observed at 7 days; how-ever, concretes cured for 28 and 56 days showed some improvement. The small improvement can be attributed to the relatively low UFFA content used in this study. In a recent study, a 14 % increase in the 28-day compres-sive strength of cement mortar containing 10 % UFFA as partial cement replacement is also reported by Supit et al. [25]; however, it was in mortar without any NCA. The specific surface area of UFFA used in this study is 2.51 m2/g, which is higher than OPC but significantly low-er than conventional silica fume, which is about 20 m2/g [16]. Therefore, in order to obtain significant improvement in the properties of concrete, much higher UFFA contents are required and need a thorough experimental study to determine the effect of other higher contents on the properties. Nevertheless, the objective of this preliminary study was to evaluate the effect of UFFA on the improve-ment to the properties of concretes containing NCA and RCA through relative comparison. The effects of the same UFFA content on the compressive and tensile strengths of concretes containing 25 and 50 % RCA are also shown in Figs. 2 and 3. It can be seen that the concretes contain-ing 25 and 50 % RCA exhibited similar improvements in compressive strength over the concrete containing NCA. However, in the case of tensile strength, a slight reduction is observed and can be attributed to the presence of mi-crocracks in the RCA and/or between the aggregate and the old mortar/paste adhering to the RCA. The presence of microcracks has a greater adverse effect on the tensile properties of concrete than the compressive properties because under tensile load, the microcracks existing prior to the test tend to separate at a much lower load than for a compression load.

Fig. 2. Effects of UFFA on compressive strength development in concretes containing recycled coarse aggregate

Fig. 3. Effects of UFFA on splitting tensile strength development in con-cretes containing recycled coarse aggregate

120



3.2 Rate of water absorption and sorptivity

The rate of absorption of water by concrete is a func-tion of the penetrability of the pore system, where for unsaturated concrete the rate of ingress of water or other liquids is largely controlled by absorption due to capillary rise. This is dependent on many factors, including con-crete mix proportions, presence of chemical admixtures and supplementary cementitious materials, entrained air content, curing age, presence of microcracks, placement method (including compaction and finishing) and also the moisture condition of the concrete at the time of testing (ASTM C1585-13 [30]).

Water sorptivity describes the water ingress into pores of unsaturated concrete due to capillary suction. It is a function of porosity, including pore volumes and pore continuity within the concrete matrix, and can be related to permeability. The water absorption of the specimen up to 6 h was fitted using linear regression and the slope of the equation obtained was used to describe the sorp-tivity in the first 6 h. The rate of absorption (mm) of all types of concrete mix at 7, 28 and 90 days are presented in Figs. 4–6. The best-fit lines in those figures are based on R2 values > 0.98 for all mixes. It can be seen that the cumulative volume of water absorbed in the concrete specimens increased with the square root of time. It can be clearly seen that the water absorption rates of con-

crete containing 25 and 50 % RCA are higher than the concrete containing NCA. This increased absorption due to capillary rise was expected and is due to the inferior properties of the RCA, such as higher water absorption than NCA (see Table 2). The higher water absorption of the RCA is primarily linked to the mortars attached to its surface, which are very porous, and also the high percentage of masonry products. Furthermore, given the nature of the manufacturing process, the RCA tends to form cracks and fissures in the aggregate, which further contributes to increased sorptivity of the concrete. It can also be seen that as the curing age increases, so the water absorption rate of recycled aggregates concrete decreases due to the continuing hydration reaction and formation of calcium-silica-hydrate (CSH) hydration products, which generally fill the micropores in the matrix and densify the matrix. The effects of adding 10 % UFFA on the water absorption rate of concretes containing NCA and RCA at various curing ages are also shown in Figs. 4−6. It can be seen that the addition of 10 % UFFA significantly re-duced the water absorption rate of concretes containing NCA and RCA at all ages. The reduction in sorptivity is in the range 33−40 % in the case of concrete containing NCA and 38−54 % and 22−43 % in the case of concrete containing 25 and 50 % RCA respectively (see Fig. 7). This significant reduction can be attributed to fineness, high pozzolanic activity and the formation of secondary CSH

Fig. 4. Effect of UFFA on the water absorption rate of concretes containing recycled coarse aggregates and cured for 7 days


Fig. 7. Effects of UFFA on the sorptivity of concretes containing recycled coarse aggregates


121



amorphous compound due to the reaction of UFFA parti-cles with calcium hydroxide (CH) [25]. In a recent study, the author has also evaluated the effect of the addition of 8 % UFFA as partial cement replacement on the hydra-tion phases of the cement paste through X-ray diffraction (XRD) analysis and found a significant reduction in C2S, C3S and CH contents and an increase in the amorphous content (which is generally CSH) [25]. The addition of 8 % UFFA also significantly reduced the large capillary pores of the cement matrix [25]. As the UFFA content in this study was slightly higher (10 %), the amount of CSH and the reduction in large capillary pores should therefore be greater in the concretes in this study, which densified the microstructure of the matrix and contributed to the reduc-tion in water sorptivity. The above phenomenon in those concretes is also evident in the longer curing ages, where the reduction in water sorptivity of recycled aggregate concretes at 56 days is higher than those at 7 and 28 days.

3.3 Chloride ion permeability

Chloride migration through concrete is a very slow process, even in circumstances where the water/cement ratio is high. Due to limitations of time, the fastest ac-ceptable test method for measuring the chloride perme-ability of concrete is the rapid chloride permeability test proposed by ASTM C1202 [31]. Fig. 8 shows the effect of 10 % UFFA on the chloride ion permeability of concretes containing NCA and RCA with respect to curing ages measured according to ASTM C1202 [31]. It can be seen that the addition of 10 % UFFA as partial cement replace-ment in concrete containing NCA changed the ASTM chloride ion resistance level classification from ‘high’ to ‘medium’ for 7-day samples and from ‘medium’ to ‘low’ for 56-day samples. Similar behaviour is also observed in the concretes containing RCA, e.g. the chloride ion per-meability level of recycled aggregate concretes cured for 28 and 56 days is changed from ‘high’ to ‘medium’ due to the addition of UFFA. According to Otsuki et al. [34] the chloride ion resistance of the concrete depends largely on the porosity and interconnectivity of the pore system in the concrete, and to a lesser extent on the chemical binding capacity of the cement. Concrete’s resistivity to chloride also increases with curing age [20,35–37]. This

has been evident in most research projects where longer curing effectively densifies the microstructure of the concrete and prevents the permeability for fluids [38]. Hence, the continual hydration process of cement with UFFA narrows the capillary pores in the concrete, which is also evident in the work of Supit et al. [25].

4 Conclusions

Based on a limited study of the effect of ultrafine fly ash (UFFA) on the properties of concrete containing recycled coarse aggregates sourced from locally available C&D wastes, the following conclusions can be drawn:– The addition of UFFA increased the compressive

strength of concretes containing 25 and 50 % RCA at all ages up to 56 days. The improvement in compres-sive strength is higher in concrete containing 25 % RCA than in concrete containing 50 % RCA. Concrete containing 25 % RCA and 10 % UFFA achieved 94 % of the control concrete’s compressive strength at 56 days. With prolonged curing, this gap can be reduced further.

– The addition of UFFA, however, showed a slight re-duction in the tensile strength of recycled aggregate concretes. Both recycled aggregate concretes tested also achieved about 88 % of the control concrete’s ten-sile strength at 56 days and this gap can be reduced with prolonged curing.

– In both recycled aggregate concretes tested, the sorp-tivity is much lower at all ages due to the addition of 10 % UFFA. The concrete containing 25 % RCA and 10 % UFFA achieved lower sorptivity than the control concrete at all ages.

– It is suggested from the results that the addition of 10 % UFFA tends to increase the chloride ion resist-ance of recycled aggregates concretes, given the fact that it serves to promote hydration and block the cap-illary spaces within the concrete. The chloride ion pen-etration of concrete containing 25 % RCA and 10 % UFFA is lower than that of the control concrete.

Although further studies are necessary to improve our un-derstanding of the influence of UFFA on recycled aggregate concretes, it could be suggested from the results that using UFFA is a promising approach to improving both the me-chanical and the durability properties of concrete and hence its performance. Furthermore, it is suggested from the tests conducted and the results at this stage – and in accordance with the limitations of this research – that 10 % UFFA as par-tial cement replacement in concrete containing RCA from construction and demolition wastes can produce structural grade concrete that is able to perform without any draw-backs to its mechanical and durability properties.

Acknowledgements

The author gratefully acknowledges the assistance of final-year project students Luke Chadwick and Jawad Rajabi during the casting and testing of the concrete specimens de-scribed in this paper. The author would also like to thank Fly ash Australia for supplying the ultrafine fly ash and the All Earth Group for supplying the recycled coarse aggregates.

Fig. 8. Effects of UFFA on the chloride ion permeability of concretes con-taining recycled coarse aggregates

122



References

1. Hyder Consulting: Australian landfill capacities into the future. Department of the Environment, Water, Heritage and the Arts, 2009, pp. 1–48.

2. Neville, A. M.: Properties of concrete, 5th ed., Pearson Educati-on Ltd., 2011.

3. Australia’s Sustainable Aggregates Industry: Building our nati-on through smatter resource use. Australia’s Sustainable Aggre-gates Industry report, 2012, pp. 1–11.

4. Hyder Consulting: Management of construction and demoliti-on waste in Australia. Construction and demolition waste sta-tus report, Queensland, Australia, 2011.

5. Chen, H. J., Yen, T., Chen, K. H.: Use of building rubbles as re-cycled aggregates. Cem. Concr. Res., 2003, 33(1), pp. 125–132.

6. Yong, P. C., Teo, D. C. L.: Utilisation of recycled aggregate as coarse aggregate in concrete. UNIMAS e-J. Civ. Eng., 2009, 1(1), pp. 1–6.

7. Zhang, W., Ingham, J. M.: Using recycled concrete aggregates in New Zealand ready-mix concrete production. J. Mater. Civ. Eng., 2010, 10.1061/(ASCE)MT.1943-5533.0000044, pp. 443–450.

8. Xiao, J., Li, J., Zhang, C.: Mechanical properties of recycled aggregate concrete under uniaxial loading. Cem. Concr. Res., 2005, 35(6), pp. 1187–1194.

9. Yang, K. H., Chung, H. S., Ashour, A. F.: Influence of type and replacement level of recycled aggregates on concrete proper-ties. ACI Mater. J., 2008, 105(3), pp. 289–296.

10. Ahmed, S. F. U.: Existence of dividing strength in concretes containing recycled coarse aggregates. Journal of Materials in Civil Engineering, 2014, vol. 26(4), pp. 784–788.

11. Etxeberria, M., Vázquez, E., Marí, A., Barra, M.: Influence of amount of recycled coarse aggregates and production process on properties of recycled aggregate concrete. Cement and Con-crete Research, 2007, 37(5), pp. 735–742.

12. Kou, S. C., Poon, C. S.: Enhancing the durability properties of concrete prepared with coarse recycled aggregate. Construc-tion and Building Materials, 2012, 35, pp. 69–76.

13. Kou, S. C., Poon, C. S., Chan, D.: Influence of fly ash cement replacement on the properties of recycled aggregate concrete. Journal of Materials in Civil Engineering, 2007, 19(9), pp. 709−717.

14. Kou, S. C., Poon, C. S., Chan, D.: Influence of fly ash as a ce-ment addition on the hardened properties of recycled aggregate concrete. Materials and structures, 2008, 41, pp. 1191–1201.

15. Brendt, M. L.: Properties of sustainable concrete containing fly ash, slag and recycled concrete aggregate. Construction and building materials, 2009, 23, pp. 2606–2613.

16. Corinaldesi, V., Moriconi, G.: Influence of mineral additions on the performance of 100 % recycled aggregate concrete. Con-struction and building materials, 2009, 23, pp. 2869–2876.

17. Ahmed, S. F. U.: Properties of concrete containing construction and demolition wastes and fly ash. Journal of Materials in Civil Engineering, 2013, 25(12), pp. 1864−1870.

18. Ahmed, S. F. U.: Properties of concrete containing recycled fine aggregate and fly ash. Journal of Solid Waste Technology and Management, 2014, 40(1), pp. 70–78.

19. Tangchirapat, W., Buranasing, R., Jaturapitakkui, C.: Use of high fineness of fly ash to improve properties of recycled aggre-gate concrete. Journal of Materials in Civil Engineering, 2010, 22(6), pp. 565–571.

20. Shaikh, F. U. A., Than, A. B., Odoh, H.: Effect of nano silica on properties of concretes containing recycled coarse aggregates. Construction materials, 2014, DOI: 10.1680/coma.14.00009.

21. Obla, K. H., Hill, R. L., Shashiprakash, S. G., Perebatova, O.: Properties of concrete containing ultra-fine fly ash. ACI Mater J, 2003, 100-M49.

22. Hossain, A. B, Islam, S., Copeland, K. D.: Influence of ultrafine fly ash on the shrinkage and cracking tendency of concrete and the implications for bridge decks. Transportation Research Board annual meeting 2007, paper #07-0022. Transportation Research Board, Washington, 2007.

23. Subramaniam, K. V., Gromotka, R., Shah, S. P., Obla, K., Hill, R.: Influence of ultrafine fly ash on the early age response and the shrinkage cracking potential of concrete. J Mater Civ Eng, 2005, 17(1), pp. 45–53.

24. Choi, S., Lee, S. S., Monteiro, P. J. M.: Effect of Fly ash fineness on temperature rise, setting, and strength development of mortar. ASCE J Mater Civ Eng, 2012, 24(5), pp. 499–505.

25. Supit, W. M. S., Shaikh, F. U. A., Sarker, P. K.: Effect of Ultrafi-ne Fly Ash on Mechanical Properties of High Volume Fly Ash Mortar. Construction & Building Materials, 2014, vol. 51, pp. 278–286.

26. HB 155:2002: Guide to the use of recycled concrete and ma-sonry materials. Standards Australia, Sydney, 2002.

27. AS 2758.1: Aggregates and rocks for engineering purpose – Part 1 concrete aggregate. Australian Standard, 1998.

28. AS 1012.3.1:1999: Methods of testing concrete, Method 9: De-termination of compressive strength of concrete specimens. SAI Global, Sydney, NSW, Australia, 1999.

29. AS 1012.10: Methods of testing concrete, Method 10: Determi-nation of indirect tensile strength of concrete cylinders (“Bra-zil” or splitting test), 2000.

30. ASTM (2013) ASTM C1585-13: Standard test method for measurement of rate of absorption of water by hydraulic ce-ment concretes. ASTM International, West Conshohocken, PA, USA.

31. ASTM C1202: Standard test method for electrical indication of concrete’s ability to resist chloride ion penetration. American society for testing and materials, Philadelphia, USA, 2012.

32. Helmuth, R.: Fly ash in cement and concrete, Portland Cement Association, USA, 1987.

33. Chindaprasirt, P., Jaturapitakkul, C., Sinsiri, T.: Effect of fly ash fineness on microstructure of blended cement paste, Construc-tion and building materials, 2007, vol. 21, pp. 1534–1541.

34. Otsuki, N., Miyazato, S., Yodsudjai, W.: Influence of recycled aggregate on interfacial transition zone, strength, chloride pe-netration and carbonation of concrete. J. Mater. Civ. Eng. 2003, 15, pp. 443–451.

35. Shaikh, F. U. A. and Supit, W. M. S.: Mechanical and durabili-ty properties of high volume fly ash (HVFA) concrete contai-ning calcium carbonate (CaCO3) nanoparticles. Construction and building materials, 2014, vol. 70, pp. 309–321.

36. Shaikh, F. U. A., Supit, W. M. S.: Compressive strength and durability properties of high volume fly ash concretes contai-ning ultrafine fly ash. Construction and building materials, 2015, DOI:10.1016/j.conbuildmat.2015.02.068.

37. Supit, W. M. S., Shaikh, F. U. A.: Durability properties of high volume fly ash concrete containing nano silica. Materials and structures, 2014, DOI 10.1617/s11527-014-0329-0.

38. Olorunsogo, F. T., Padayachee, N.: Performance of recycled aggregate concrete monitored by durability indexes. Cem. Concr. Res., 2002, 32, pp. 179–185.

F.U.A. Shaikh, Ph.D.Senior LecturerDept. of Civil EngineeringCurtin University,GPO Box U1987Perth 8645, [email protected]

FIB_Advert_160112_210x297_3mmBleed_Paths.indd 1 1/13/16 9:28:04 AM

Composite Structures according to Eurocode 4

This book presents a large number of numerical examples with

detailed explanations of the provisions of Eurocode 4. It deals

with the most common structural components in building con-

struction: beams, columns and slabs. Furthermore, comprehen-

sive chapters provide insight into the topics of creep and shrin-

kage, as well as fatigue.

*€ Prices are valid in Germany, exclusively, and subject to alterations. Prices incl. VAT. excl. shipping. 1102106_dp

Order online: www.ernst-und-sohn.de


Boschstraße 12

D-69469 Weinheim

Tel. +49 (0)6201 606-400

Fax +49 (0)6201 606-184

[email protected]

Ernst & Sohn



Related titles:

Darko Dujmovic, Boris Androic,

Ivan Lukacevic

Composite Structures according

to Eurocode 4

Worked Examples

2014. 924 pages.

€ 89,–*

ISBN 978-3-433-03107-0

Also available as

Design of Composite

Structures

Fatigue Design of

Steel and Composite

Structures

1102106_dp_210x297mm.indd 1 20.02.15 14:17

Structural Concrete 17 (2016), No. 1 125

fib-news

fib-news is produced as an integral part of the fib Journal Structural Concrete.

Contents Issue 1 (2016)

fib MC2020: The story so far 125

New fib SG 126

fib-ACI MoC 127

Betontage in Shanghai 127

COM4: Materials-related priorities 128

4th IWCS 128

Ralejs Tepfers 1933-2015 129

Gabriel Tevec 1936-2016 129

Congresses and symposia 130

Acknowledgement 131

fib MC2020: The story so far

Discussions have been under way for some time in the fib about the advancement of the fib Model Code for Concrete Structures and in par-ticular about the need to extend the current code provisions for matters relating to existing concrete struc-tures.

The need to start on this journey be-came apparent to fib Special Activ-ity Group (SAG) 7 ‘Assessment and interventions upon existing struc-tures’, now fib Commission 3 ‘Exist-ing concrete structures’ (COM3).

other organizations, who brought perspectives from across the globe.

Their presentations stimulated vigorous discussion on many topics including:

– fundamental principles and reliability concepts for new and existing structures

The MC2020 Core Group are (from left to right) Joost Walraven, Frank Dehn, Aurelio Muttoni, Agnieszka Bigaj-van Vliet, Giuseppe Mancini, Stuart Matthews, Gerrie Dieteren, Tamon Ueda and Hugo Corres Peiretti Photo credit: Javier Leon

Initially, COM3 presented the idea of creating a model code for the assessment of existing structures to complement fib MC2010 during the informal commission chair meeting and the technical council meeting in Copenhagen in May 2015.

Following debate, it was decided that a model code for the assess-ment of existing structures would be the wrong path to take and the goal subsequently set was the more difficult and challenging task of de-veloping a new fib Model Code for Concrete Structures combining both new and existing structures.

A decision was made at the 2015 technical council meeting to hold a workshop to discuss the concept of such a combined code and its required attributes.

The workshop was hosted by the Netherlands Organization for Ap-plied Scientific Research (TNO) in The Hague on 30 June 2015 and over 40 people attended. Invited speakers included representatives from various fib commissions and


fib-news

– models and material characteri-zation for existing structures

– inspection, maintenance and retrofitting of existing structures

– evaluation of, decision-making about and management of exist-ing structures

The workshop summary set down ideas for possible ways forward and some aspirations for linking and addressing the requirements for new and existing structures.

The workshop findings were consid-ered by the presidium in their meet-ing in August 2015, enabling them to make the decision to form a core group to work on the development of a roadmap outlining a potential course forward in the MC2020 project.

The MC2020 Core Group met in Madrid on 11 and 12 December 2015. They identified numerous aspirational goals, including that MC2020 should ideally:

– be a single, merged structural code dealing with both new and existing concrete structures

– be an operational model code oriented towards practical needs

– include worldwide knowledge of materials and structural behav-iour

– recognize the needs of engineer-ing communities in different regions of the world

– take an integrated-life-cycle perspective

– provide holistic treatment of structural safety, serviceability, durability and sustainability

– define fundamental principles and a safety philosophy based on reliability concepts

– use the performance based concept to remove specific con-straints for novel types of con-crete and reinforcing materials

– provide fully integrated provi-sions based on generalized models and implement a level-of-approximation approach appli-

cable in both the design of new structures and all the activities associated with assessment (and so forth)

– allow full advantage to be taken of information that can be ac-quired by the testing and moni-toring of existing structures

– address robustness and redun-dancy for new and existing structures

– consider material degradation and / or the insufficient or defi-cient detailing of the provided material and behaviour models

– consider needs for model im-provement and the treatment of uncertainties in models and model parameters for existing structures and the (phased) con-struction of interventions

– give attention to aspects of through-life management

– give attention to new types of concrete / repair materials (and so forth)

– address end-of-service-life issues, such as demolition and disposal and recycling

There are profound challenges in various aspects of the proposed work. Important related considera-tions include involving all the com missions so as to draw upon the full range of expertise and knowl-edge in the fib, gathering a team of active authors and expert contribu-tors from all over the world, seeking improved ways of working and em ploying the most effective means of delivering MC2020 project outputs.

In the near future a road map pro-posal and terms of reference will be presented to the relevant fib bodies and meetings and workshops will be organized.

Regular progress reports will appear in fib-news to keep readers up-to-date.

Stuart MatthewsChair of fib Commission 3

New fib SG

The fib is pleased to announce the appointment of David Fernández-Ordóñez as our new secretary general.

Dr Fernández-Ordóñez commenced his executive duties in Lausanne on 4 January 2016.

At the time of his appointment Dr Fernández-Ordóñez was Deputy Director of the School of Civil Engi-neers at the Technical University of Madrid. Earlier in his career he was Technical Manager for two Spanish producers of precast concrete.

fib Secretary General David Fernández-Ordóñez Photo credit: Marianne Fourie

The fib’s new secretary general is not new to the fib as a member. Dr Fernández-Ordóñez has supported us since 1999 and was both a deputy of the Spanish delegation and Chair of fib Commission 6 ‘Prefabrication’ at the time of his appointment.

A Spanish native, he is also a member of ACHE and the Juanelo Turriano Foundation.

During the first quarter of this year Dr Fernández-Ordóñez will split his time between Madrid and Lausanne to finalize his move to Switzerland.


fib-news

construction, and materials, who share a commitment to pursuing the best use of concrete.’

ACI and the fib have cooperated on projects for many years, with the 2014 joint ACI-fib workshop on fibre-reinforced concrete held in Montreal, Canada, a recent example of collaboration. The proceedings of the workshop will be published in the near future as an fib Bulletin and an ACI publication.

The MoC, signed during the ACI Fall Convention in Denver, seals both associations’ commitment to developing and disseminating knowledge about concrete globally.

‘Globalization’ as fib President Harald S. Müller explained in his ‘Message from the president’, published in the March 2015 edition of Structural Concrete, means ‘developing closer official contacts’ with international associations that have comparable missions and publication strategies to those of the fib.

This new agreement follows the signing of a memorandum of coop-eration with RILEM on 2 Septem-ber 2015 in Melbourne, Australia, (announced in the fib-news section of Structural Concrete 16, No. 4).

From left to right: ACI Past President William E. Rushing, ACI Executive Vice President Ronald G. Burg, fib President Harald S. Müller and ACI President Sharon L. Wood Photo courtesy of ACI

fib President Harald S. Müller and ACI President Sharon L. Wood Photo courtesy of ACI

On 8 November 2015 fib President Harald S. Müller signed a memo-randum of cooperation (MoC) with ACI President Sharon L. Wood, ACI Past President William E. Rushing and ACI Executive Vice President Ronald G. Burg.

ACI (American Concrete Institute) is ‘a leading authority and resource worldwide for the development and distribution of consensus-based standards, technical resources, educational and training programs, certification programs, and proven expertise for individuals and organi-zations involved in concrete design,

fib-ACI MoCBetontage in ShanghaiAt the Betontage Asia 2015 confer-ence, held from 5 to 6 November during the 2015 Shanghai Interna-tional Building Industrialization Congress (BIC), Frank Dehn, fib Presidium Member and Deputy Chair of the Technical Council, gave a keynote speech about the fib, our mission and our activities in the advancement of European concrete technology and the development of the precast industry.

The main aim of the conference was to provide a local platform for those presenting trends in the internation-

Frank Dehn makes his keynote speech at Betontage Asia 2015 Photo courtesy of VNU Exhibitions Asia

al and Chinese concrete and precast industries and in technical innova-tions from the design to production stages.

A warm-up workshop, two plenary sessions and forums dealing with the following topics were held:

– Design and innovation – Manufacture and machinery – On site and off site

The event was attended by more than 500 participants.

The next BIC is scheduled for Octo-ber 2016.


fib-news

fib Commission 4 ‘Concrete and concrete technology’ (COM4), led by myself and Dr Tor Arne Martius-Hammer of Norway (Deputy Chair), used to be fib Commission 8 ‘Con-crete’ in the pre-2015 commission structure.

Today COM4’s task is to generate theoretical and practical devel-opments in the field of concrete technology, and to present these developments in an understandable and code-type form.

The commission positions itself at the forefront of new technolo-gies and techniques by taking both fundamental research and practical issues into consideration.

The aim of the commission is to collect and validate information on the properties and behaviour of concrete for structural applications subjected to various types of loading and environmental conditions. It has focused its attention on tradi-tional types of concrete, in particu-lar those exposed to severe condi-tions, as well as on new types of concrete and cementitious compos-ites subjected to all types of loading and exposure.

In its work on the properties of con-crete types, the commission wants to formulate things in such a way that it will be possible to generate physically, chemically and mechani-cally sound constitutive material and degradation models and estab-

fib COM4 Chair Frank Dehn and Deputy Chair Tor Arne Martius Hammer

lish recommendations for structural applications.

The commission’s areas of technical interest and investigation are:

– Durability and degradation of concretes and composites

– Traditional cements and blended binders, concrete admixtures and additions, and supplementary cementing materials

– Alkali-activated binders and concretes

– Geopolymer concretes – Constitutive laws and code-type

models for traditional and new types of concretes and compos-ites

– Time- and temperature-depend-ent properties of concrete and composites

– Concrete properties under multi-ple loadings and exposures

– Concrete with recycled materials – Properties and behavioural mod-

eling of concretes and compos-ites

Currently COM4 consists of six task groups (T) who cover diverse aspects of concrete as a material as well as the technological issues sur-rounding it:

– T4.1: Fibre-reinforced concrete – T4.2: Ultra high-performance

fibre-reinforced concrete – T4.3: Structural design with flow-

able concrete – T4.4: Aesthetics of concrete

surfaces – T4.5: Performance-based specifi-

cations for concrete – T4.6: Constitutive laws for

concrete with supplementary cementitious materials

One of COM4’s major tasks for the future is contributing to the initia-tive for the new fib Model Code for Concrete Structures 2020, which will include provisions for new as well as existing structures. In this context, beyond its original mate-rials-related priorities, COM4 will

COM4: Materials-related priorities

pay particular attention to outlining suitable in-situ and laboratory-test methods and to defining perfor-mance criteria for the evaluation of novel types of binders, constituents and concretes for structural applica-tions.

Frank DehnChair of fib Commission 4MFPA Leipzig GmbH / Leipzig University

4th IWCS

The 4th International Workshop on Concrete Spalling due to Fire Expo-sure took place from 8 to 9 October 2015 in Leipzig, Germany, with Frank Dehn, fib Presidium Member and Deputy Chair of the Technical Council, heading the organizing committee.

Thirty-nine presentations were made over the course of the two-day event attended by 80 people.

The aim was to offer an overview of current knowledge and to stimulate discussion between researchers and representatives from industry, the authorities and the code-making bodies.

SP in Sweden will organize the next IWCS, to be held in 2017.


fib-news

Gabriel Tevec 1936–2016

Technology in Bratislava in 1960. After graduating he joined Do-prastav, then the largest Slovak company in the field of design and bridge construction, as an independ-ent structural designer. From 1962 to 1993 he became Head Designer, followed by Head of Structural Design of Bridges and Roads and finally Director of Technical Project Management.

From 1993 to 2000 Gabriel co-founded and directed the structural design company Geoconsult. From 2000 to 2011 he again worked for Doprastav.

As a structural designer in the 60s, Gabriel was on hand for the con-struction of the first prestressed, balanced cantilever bridges and precast-beam bridges in Slovakia. In 1969 he contributed to the design of the pylon basement and the anchor block of the first cable-stayed bridge, whose 303-metre-long main span extends over the river Danube.

Between 1986 and 1990 Gabriel participated in the design and con-struction of another major bridge over the river Danube in Bratislava. The Lafranconi Bridge is the first bridge with external tendons in Slo-vakia. Among his contributions was the design of a new type of light-form traveller for its side spans.

After retiring Gabriel worked as an expert consultant on the design and construction of several major highway bridges in Slovakia as well as on a 10-span, extradosed viaduct with a total length of 941 metres in Považská Bystrica.

Gabriel passed away on 12 January 2016 at the age of 80. We honour his memory.

Milan ChandogaPresident of fib Slovakia

Peter PaulikSecretary fib Slovakia

Gabriel Tevec was an active member of the Czecho-slovak group of the FIP. From the FIP Congress in New York

in 1974 onwards he attended most of the other FIP and fib congresses. He played a large role in establish-ing the Slovak National Committee of the FIP and was its first president from 1994 to 1998.

Gabriel completed his Master’s degree at the Slovak University of

Ralejs Tepfers

1933–2015

the Nordic Concrete Federation, as well as to the CEB and later the fib. He never forgot the land of his birth and worked hard to generate opportunities for Latvian colleagues. The 1992 conference on Bond in Concrete, hosted by Riga Technical University in the face of consider-able logistical challenges, stands out in the memory of delegates.

As Professor Emeritus at Chalmers, Ralejs remained active in retire-ment, attending the university most days up to the week he died.

He is fondly remembered by his many international collaborators not just for his technical contribu-tions, but also for his quiet charm, humility and gentle personality.

Ralejs is survived by his wife, Ira, their two daughters and one son.

John CairnsHeriot-Watt University

Ralejs Tepfers, Professor Emeritus at Chalmers University, Gothenburg, and Honorary Member of the fib, passed away in October 2015.

Ralejs was born in Rezekne, Latvia, in 1933 but fled to Swe-den with his family in 1944. He completed a Master’s

degree in Civil Engineering at Chalmers University of Technol-ogy in 1958 and after a brief period on construction sites in Sweden and Switzerland returned to to the Department of Building Technology at Chalmers as an associate profes-sor. He was appointed Professor of Building Technology in 1995 and Professor Emeritus on his retirement in 2001.

His doctoral thesis ‘A Theory of Bond Applied to Overlapped Tensile Reinforcement Splices for Deformed Bars’, published in 1973, has been hugely influential.

In 1977 he was invited to establish a CEB task group on bond and an-chorage, which he led vigorously for 23 years and which resulted in three bulletins (the last under the auspices of the fib) as well as a significant contribution to the CEB-FIP Model Code 90.

Ralejs embraced new technologies in the field of structural concrete, in particular fibre-composite reinforc-ing materials, introduced for the first time in an fib publication in Bulletin 10. He was also a major contributor to fib Bulletin 40 through fib Task Group 9.3 (now 5.1).

Ralejs understood the importance of international cooperation and contributed to the Swedish and Latvian Concrete Associations and


fib-news

Date and location Event Main organiser Contact

13–15 June 2016 ICCS16 Second International Universidad Politecnica iccs16.org Madrid, Spain Conference on Concrete de Madrid Sustainability

18–20 July 2016 First International Interactive Iowa State University register.extension.iastate.edu/ Des Moines, Iowa, USA Symposium on UHPC University of Connecticut uhpc2016

29–31 August 2016 11th fib International Ph.D. Nihon University concrete.t.u-tokyo.ac.jp/fib Tokyo, Japan Symposium Tokyo University _PhD2016

5–7 September 2016 3rd International Conference on Gadjah Mada University conference.tsipil.ugm.ac.id/ Bali, Indonesia Sustainable Civil Engineering scescm/ Structures and Construction Materials (SCESCM)

12–14 September 2016 CONSEC2016 Politecnico di Milano consec16.com Lecco, Italy

19–21 September 2016 9th International Symposium on University of British befib2016.ca Vancouver, Canada Fiber Reinforced Concrete Columbia & RILEM (BEFIB 2016)

5–7 October 2016 8th International Conference on Wroclaw University of [email protected] Wroclaw, Poland Arch Bridges Technology

24–26 October 2016 3rd International Conference on Southeast Univ. of Tech., microdurability2016.com Nanjing, China Microstructure Related Durability Delft Univ. of Technology, of Cementitious Composites Jiangsu Research Inst. of Building Science

30 Oct – 2 Nov 2016 7th Intl Conference of the Asian Asian Concrete acf2016.vn Hanoi, Vietnam Concrete Federation – Sustainable Federation Concrete for Now and the Future

21–23 November 2016 fib Symposium University of Cape Town fibcapetown2016.com/ Cape Town, South Africa Performance-based approaches for concrete structures

6–8 March 2017 HPC/CIC Tromsø 2017 Norwegian Concrete tekna.no/en/hpccic Association

12–15 June 2017 fib Symposium fib National Member [email protected] Maastricht, Netherlands High tech concrete: Where Group Netherlands www.fibsymposium2017.com technology and engineering meet

6–12 October 2018 5th fib Congress and fib National Member fibcongress2018.com Melbourne, Australia Exhibition Group Australia

27–29 May 2019 fib Symposium fib National Member fibkrakow2019.pl Krakow, Poland CONCRETE – Innovations in Group Poland Materials, Design and Structures

Congresses and symposia

The calendar list with fib Congresses and Symposia, cosponsored events and, if space permits, events supported by the fib or organized by one of its na-tional member groups reflects the state of information available to the secretariat at the time of printing. The information given is subject to change.


fib-news

National member groups

AAHES – Asociación Argentina del Hormigón Estructural

CIA – Concrete Institute of Australia

ÖBV – Österreichische Bautech-technik Vereinigung, Austria

GBB – Groupement Belge du Bé-ton, Belgium

ABCIC – Associação Brasileira da Construção Industrializada de Concreto, Brazil

ABECE – Associação Brasileira de, Engenharia e Consultoria Estru-tural, Brazil

fib Group of CanadaCCES – China Civil Engineering

SocietyCyprus University of TechnologyCBS – Ceska Betonarska

Spolecnost, Czech RepublicDBF – Dansk Betonforening DBF,

DenmarkSuomen Betoniyhdistys R.Y., Fin-

landAFGC – Association Française de

Génie Civil, FranceDAfStb – Deutscher Ausschuss für

Stahlbeton e.V., GermanyTechnical Chamber of GreeceUniversity of Patras, GreeceHungarian Group of fibThe Institution of Engineers (India)Dept. of Technical Affairs, IranIACIE – Israeli Association of

Construction, and Infrastructure Engineers

AICAP – Associazione Italiana Calcestruzzo, Armato e Precom-presso, Italy

CTE – Collegio dei tecnici della industrializzazione edilizia, Italy

ITC – CNR, Istituto per le Tecnolo-gie della Costruzione, Italy

ReLUIS, ItalyJCI – Japan Concrete InstituteJPCI – Japan Prestressed Concrete

Institute Lebanese Concrete SocietyAdministration des Ponts et Chaus-

sées, Luxembourgfib NetherlandsNZCS – New Zealand Concrete

SocietyNorsk Betongforening, NorwayCommittee of Civil Engineering,

PolandGPBE – Grupo Portugês de Betão

Estrutural, PortugalFacultatea de Constructii, Transylva-

nia University of Brasov, RomaniaTechnical University of Civil Engi-

neering, RomaniaUPT – Universitatea Politehcnica

Timisoara, RomaniaAssociation for Structural Concrete,

RussiaAssociation of Structural Engineers,

SerbiaSNK fib, SlovakiaSlovenian Society of Structural

EngineersUniversity of Cape Town, South

AfricaKCI – Korean Concrete InstituteACHE – Asociación Cientifico-

Técnica del Hormigón Estructural, Spain

Svenska Betongföreningen, SwedenDélégation nationale suisse de la fib,

SwitzerlandTCA – Thailand Concrete Associa-

tionUniversité de Tunis El Manar,

TunisiaITU – Istanbul Technical University,

TurkeyNIISK – Research Institute of

Building Constructions, Ukraine

fib UK GroupASBI – American Segmental Bridge

Institute, USA PCI – Precast/Prestressed Concrete

Institute, USA PTI – Post Tensioning Institute,

USA

Sponsoring members

Liuzhou OVM Machinery Company Ltd, China

Consolis Oy Ab,FinlandECS – European Engineered

Construction Systems (formerly VBBF), Germany

FBF Betondienst GmbH, GermanyInstitut für Stahlbetonbewehrung

e.V., GermanyMKT Metall-Kunststoff-Technik

GmbH, GermanyLarsen & Toubro Ltd ECC Division,

IndiaATP s.r.l, ItalyFuji P. S. Corporation, JapanIHI Construction Service Company

Ltd, JapanObayashi Corporation, JapanOriental Shiraishi Corporation,

JapanP. S. Mitsubishi Construction Com-

pany Ltd, JapanSE Corporation, JapanSumitomo Mitsui Constructruction

Company Ltd, JapanHilti Corporation, LiechtensteinPatriot Engineering, RussiaBBR VT International Ltd, Switzer-

landSIKA Schweiz AG, SwitzerlandVSL International Ltd, SwitzerlandPBL Group Ltd, ThailandCCL Stressing Systems Ltd, United

Kingdom

Acknowledgementfib – Fédération internationale du béton – the International Federation for Structural Concrete – is grateful for the invaluable support of the following national member groups and sponsoring members, which contributes to the publication of fib Technical Bulletins, the Structural Concrete journal, and fib-news.

Leonardo Todisco, Karl-Heinz Reineck, Oguzhan BayrakEuropean design rules for point load near support evaluated with data from shear tests on non-slender beams with vertical stirrups

Giedrius Žirgulis, Oldrich Švec, Mette Geiker, Andrzej Cwirzen, Terj KanstadInfluence of reinforcement bar layout on fibre orientation and distribution in slabs cast from fibre-reinforced self-compacting concrete (FRSCC)

K. Balaji Rao, Ahsana Parammal Vatteri, Anoop BharathanTime-variant reliability analysis of RC bridge girders subjected to corrosion – shear limit state

Linyun Zhou, Zhao Liu, Zhiqi HeInvestigation of optimal layout of ties in STM developed by topology optimization

Yong Yuan, Qing Ai, Sankaran Mahadevan, Xiaomo JiangProbabilistic degradation modeling of segmental linings assembled circular tunnels

Frank Müller, Christian Kohlmeyer, Jürgen SchnellA new approach for the calculation of internal forces, moments and deflections of sandwich panels with reinforced concrete facings

Arash NajiModeling catenary effect in progressive collapse analysis of concrete structures

Wei ZhangExperimental study on fatigue behavior of CFRP plates externally bonded to concrete substrate

Faiz Shaikh, Sarvesh MaliProperties of stabilized recycled plastic concretes made with three types of cements

Martin Classen, Joerg Gallwoszus, Alexander StarkAnchorage of composite dowels in UHPC under fatigue loading

Hadi Ziaadiny, Reza AbbasniaUnified cyclic stress-strain model for FRP-confined concrete circular, square and rectangular prisms

Souzana Tastani, Stavroula Pantazopoulou, Georgia Thermou, Thanasis Triantafillou, Giorgio Monti, Dionysios Bournas, Maurizio GuadagniniBackground to European seismic design provisions for the retrofit of RC elements using FRP materials

Preview

Structural Concrete 2/2016

Order online:

Annual subscription print + online

Single issue order

Free sample copy ➡

Journal:

www.ernst-und-sohn.de/structural-concrete 1003146_pf

Structural Concrete

Abo-Coupon-2012-13_181x84.indd 9 07.07.2014 09:09:17

Ed.: Jean-Christophe Hadorn

Solar and Heat Pump

Systems for Residential Buildings

2014. approx. 252 pages.

approx. € 79,–

ISBN 978-3-433-03040-0

Eds.: Andreas Athienitis, William O‘Brien

Modelling, Design, and Optimization of

Net-Zero Energy Buildings

2014. approx. 352 pages.

approx. € 89,–

ISBN 978-3-433-03083-7

Most developed countries have adopted net-zero energy as a long term goal for new buildings. This book will aid designers in optimally using simulation tools for net-zero energy building design. It presents advanced modelling techniques as well as in-depth case studies. The strategies and technologies are also applicable for the design of energy-plus buildings.

The combination of heat pumps and solar components has a great potential for improving the energy efficiency of house and hot water heating systems. This book compares different sys-tems, analyses their performance, presents design methods and shows monitoring techniques.

"Solar Heating and Cooling” is a research programme initiated by the International Energy Agency. The programme's work is accomplished through the international collaborative effort of experts from Member countries and the European Union. The results are published in a series with Ernst & Sohn and Wiley.

pre

limin

ary

pre

limin

ary

* € Prices are valid in Germany, exclusively, and subject to alterations. Prices incl. VAT. excl. shipping. 1112106_dp

Order online: www.ernst-und-sohn.de


Boschstraße 12

D-69469 Weinheim

Tel. +49 (0)6201 606-400

Fax +49 (0)6201 606-184

[email protected]

Ernst & Sohn



1112106_dp_210x297mm.indd 1 18.06.15 11:15

ALLIANZ ARENA, MUNICH, GERMANYTask: Support of the roof construction. Project scope: 96 MAURER MSM® Spherical Bearings.

ARENA DA AMAZÔNIA, MANAUS, BRAZILTask: Unrestrained transmission of forces and rotations of the steel roof construction. Project scope: 36 fixed MAURER Spherical Bearings from 3,300 to 6,000 kN load.

GAZPROM-ARENA, ST. PETERSBURG, RUSSIATask: Support of the roof construction. Project scope: 96 MAURER Spherical Bearings, of which 60 Uplift-/ Load Bearings. Max. Load capacity: 17.500 kN, min. load capacity: – 3,500 kN (tensile load).

STADE PIERRE-MAUROY, LILLE, FRANCE Task: The system supports the movable soccer pitch.Project scope: 36 MAURER special recentring Spherical Bearings.

forces in motionMAURER AG | Frankfurter Ring 193 | 80807 Munich/GermanyPhone +49 89 32394–0 | Fax +49 89 32394–306 | www.maurer.eu

Structural Protection SystemsSTRUCTURAL BEARINGS | EXPANSION JOINTS | SEISMIC DEVICES | VIBRATION ABSORBERS | MONITORING

Documents

Structural Concrete 01/2016 free sample copy