Failure envelopes for asphaltic concrete

D. B Y N U M , Jr (1), R.N. T R A X L E R (2)

The purpose of the investigation was to obtain the variation in trends of the thermoviscoelastic properties of an asphaltic concrete subjected to normal temperature and to high temper- ature during pre-heating, mixing, and compaction. The speci- mens were tested in uniaxial tension and uniaxial compression at constant strain rates varied over 5 decades and for temper- ature between - - 5 0 and 150~ The ultimate stress varied 2 orders of magnitude while the ultimate strain varied by a factor of about 3 for the given extremes of test temperature and strain rates investigated. The four fundamental thermovisco- elastic properties for each set of tests were summarized on a single graph. Smith failure envelopes showed how the maximum ultimate stress increased and the maximum ultimate strain decreased with increased mixing and compaction temperatures. Results of the material characterizations in this study compare favorably with previous investigations.

N O M E N C L A T U R E

Symbol Var iable Un i t

A Cross sectional area, W H in. 2

CH Cross-head rate in. /min.

(1) Ph.D., M.S., B.S., P.E., member of the Graduate Faculty and Assistant Research Engineer in the Texas Engineering Experiment Station in the College of Engineering at Texas A & M University at the time of this work, and now Manager of the Research and Development Department at The Offshore Company, P.O. Box 2765, Houston, Texas 77001, U.S.A.

(2) Ph.D., M.S., B.S., Research Chemist in the Texas Trans- portation Institute, and Professor in Chemistry and Civil Engineering Departments, Texas A & M University, College Station, Texas 77843, USA.

CS Chart speed in. /min.

Es Secant modulus at u l t imate force, ~u/lO eu ksi

Et Init ial tangent modulus , FtCS/ (1000 ARGt) ksi

Ft Arb i t r a ry force on initial slope of force-displacement trace lb.

Fu Ult imate force lb.

Gt G r a p h length f rom start o f test to Ft in.

Gu G r a p h length f rom start of test to Fu in.

H Height (average o f 4 edges) in.

L Length (average o f 4 edges) in.

Pi Percentage o f componen t mixture percent

R Strain rate, C H / L in./( in.-min.)

SGe Exper imenta l ly de termined speci- fic gravi ty numeric

SG~ Specific gravi ty o f mix componen t numeric

SGt Theoret ic specific gravity 100/ ~(Pi/SGO numeric

T Tempera tu re ~

v Air voids, 100 (1-SGe/SGt) percent

W Wid th (average o f 4 edges) in.

eu Ult imate strain, 100 R Gu/CS percent

au Ult imate stress, Fu/A psi

441

V O L . 6 - 14" 36 - 1973 - M A T I ~ R I A U X E T C O N S T R U C T I O N S

Fig. 1 - - Machine Set-up for Tension Test Calibration. Fig. 2 - - Compression Test of an Asphaltic Concrete Specimen.

I N T R O D U C T I O N (*)

The scope of this reported work covers thermo- viscoelastic characterization of two types of asphaltic concrete tested in uniaxia[ compression and tension at various strain rates and temperatures [1]. The regular mixture was preheated, mixed, and compacted at 300 ~ The second mixture was overheated at 450 ~ during the preheating, mixing, and compacting procedure, which caused oxidation of the asphalt. The data reduction and results for 55 tests are given in this report. The reason for determining the thermo- viscoelastic nature of asphaltic concrete is that the present margin of failure must be known in order that practical performance requirements can be established from the economical standpoint.

This work was necessitated by a lack of comprehen- sive fundamental thermoviscoelastic properties for flexible pavement over the range of temperatures used, as shown by a literature review. A summary of exist- ing materials data on asphaltic concrete has been given by Finn [2]. The most comprehensive study on linear viscoelasticity found in the literature was the work by Alexander [3]. Fundamental properties need- ed for purposes of design or for setting performance requirements [4] are the time-temperature dependent initial tangent moduli, ultimate secant moduli, and the Smith envelope[5]. Monismith, Secor, and Secor [6] included the Smith failure envelope for the materials tested in their study on temperature distress in flexible pavement.

(*) The data reported here was obtained in a research study sponsored jointly by the Texas Highway Department and the Federal Highway Administration. The overall objectives of the project were to obtain performance requirements for high- quality, flexible pavement, develop suitable control tests, and to find materials that will satisfy these requirements using the control tests. The opinions, findings, and conclusions expressed in this publication are those of the authors and not necessarily those of the Federal Highway Administration.

Use of these properties has been discussed else- where [7]. In general, for conservative design to preclude pavement cracking, the initial tangent modulus can be used to calculate stress when given the strain or displacement, and the ultimate secant modulus can be used to calculate strain when given the load. This procedure is not appropriate when the initial tangent modulus is greatly different from the ultimate secant modulus since the results would then be too conservative. Further, large differences would indicate stability problems (shoving, rutting) and not pavement cracking.

While the tests reported here follow the procedures used in standardized tests [8], and are probably the most appropriate for any given time, feasibility studies indicate that the use of GPC to assess long term che- mical degradation of the binder may be more appro- priate. It has been experimentally determined that the shifts in molecular weight distribution follow the same trend for recovered asphalts from the highway as obtained from the accelerated environmental laboratory tests [9-11].

Fig. 3 - - Typical Failure of Specimen After Compression Loads.

442

ULTIMATE STRESS,

%.

ULTIMATE SECANT

MODULUS, Es~

k s i

104

I0 3

IO';

I01

I0;

iO ~

iO ~

' I i I ' I ' I

y / /

~ ~ __e___---r-

/

I SYMBOL T I o -50

IO I

I0 ~

103

ULTTMATE STRAIN E u ,

per cent

102 INITIAL TANGENT MODULUSj

~o z kEti '

= I i I I I I I = 10 ~ 10 -4 I0 - 3 IC) 2 I0' IO ~ I0'

S T R A I N RATE , R s in. in.- rain.

Fig. 4 - - Properties of a-Regular Concrete in Uniaxiai Com- pression.

D . B Y N U M - R . N . T R A X L E R

ULTIMATE STRESS,

p~,

ULTIMATE SECANT MODULUS,

I03 �9 I ' I ' I ' I ' SYMBOL .T_ / 0 V l o 125 O 150

I0 I

i0 o lO I

ULTIMATE cl ~ iO o STRAIN,

per ce~t

iO 2 tO i

I01 @

I0 r

I3

I() ~ i I02 I01

INITIAL TANGENT MODULUS,

I I 0 o E t , k s i

I , I ~ I , I , I , I I ~ I i0"4 id3 id z io n iO o io ~

IN STRAIN RATE, R , I N . - MIN.

Fig. 5 - - Properties of a Regular Concrete in Uniaxial Tension.

ULTIMATE ST!~ESS.

O'u," p s i

ULTIMATE SECANT

MODULUS, ES,

k s i

104 ' ' " I ' I " I

I0 ~ ~

102 ~,

I01

IO2

i0 ~ /,/

SY ~ 4aoaX ,so"S 1

I01

ULTIMATE I0 o STRAIN,

Eu per ~ent

@

10 3

N T A L TANGENT MODULUS,

E t , k s i

i0 =I" I0 "~ I0 16 I0 I0 IN.

STRAIN PATE,R, IN.-MIN.

Fig. 6 - - Properties of an Overheated Concrete in Uniaxial Compression.

tO 4 �9 I ' I ' I ' I

IO 3 ULTIMATE

STRESS,

p s i 102

IO I

,o3

SYMC..GL "I" I i - 50 0 25 75

IO I

ULTIMATE i0 o STRAIN,

EU, per cent

iO I

1O 2 L ^ ULTIMATE SECANT MODULUS,

i0 ~ IO 3

TANGENT MODULUS~

IO I kE':i J

10-4 163 16 z Id ~ I0 ~ IO I fN

STRAIN RATE,R, IN.-MIN.

Fig. 7 - - Properties of an Overheated Concrete in Uniaxial Tension.

443

V O L . 6 - N * 36 - 1973 - M A T ~ R I A U X ET C O N S T R U C T I O N S

EXPERIMENTAL PROCEDURES

The mixtures consisted of AC-IO asphalt and Georgetown limestone with Asphalt Institute Gra- dation IV-B where the maximum size was 3/4 in. Asphalt content was 6 % of the aggregate weight. The components of the regular mixture were preheated, mixed, and compacted at 300 ~ The components of the overheated mixture were preheated, mixed, and compacted at 450 ~ The experimentally determ- ined SGe of the asphalt and aggregate were 1.002 and 2.68, respectively, which results in a theoretical SGt of 2.45, neglecting the aggregate absorption.

Samples of mixtures measuring 17-1/2 in. diameter by 2 in. thick were made using the kneading com- pactor developed by Jimenez [12] and mentioned briefly in his paper [13]. Details concerning operation of this compactor were also given by Layman [14]. Briefly stated, the mixture components and the mold were preheated to the selected compaction temperature. The components were mixed for approximately two minutes, tilt compacted for two minutes, and then level compacted for thirty seconds.

The samples were sawed with a diamond bit blade into test specimens having nominal dimensions of 1.5 in. • 1.5 in. x 5 in. The air void content of the test specimens ranged from 2.9 to 12,3 % for regular mixture, averaging 8.8 for 23 samples, and ranged from 0 to 6.6 Vo for the overheated mixture, averaging 3.0 for 32 samples. The percent air voids were calculat- ed from the indicated relations given in the nomencla- ture, where the specimen volume was calculated from an average of the measurements of the four edges for each dimension.

An Instron Testing Machine with an environmental control chamber was used to perform the constant strain rate tests. The compression specimens were placed between platens fixed on the crosshead and on the compression bench. Using a 4 in. length sample in a compression control test, a calibrated LVDT having a one inch gauge length was placed on each side of the specimen, and the average of the LVDT readings showed that the indicated strain from the Instron Machine amounted to 85 % of the strain mea- sured at mid-length of the specimens using the L VDT.

The test set-up for calibrating the Instron Machine to determine the machine deformations with the tension tests is shown in figure 1. The true specimen deformation was taken as the indicated deformation from the Instron chart less the machine deformation. A compression specimen being tested is shown in figure 2. A typical compression specimen after testing to failure is shown in figure 3.

The tension configuration was prepared by bonding each end of the specimen to a 2 in. diameter, 2-1/2 in aluminum cylinder using 1 part hardener(t) per seven parts epoxy resin(2). Tape was wrapped around the end of the cylinder to prevent run-off of the epoxy. The cure time was accelerated by placing

(z) Diethylene Triamine, Ring Chemical Company, Houston, Texas.

(2) Epon Resin 828, Shell Chemical Company, New York, New York.

ULTIMATE STRESS

O'u, p s ,

101

I04 ' ' | ' I

[ SYM 13t- ''r FLAG NOMINAL

I g ,9, ~ ~x,9:! o ?~ o- 5xlU..~ o 125 9 5:<[0.'

l 0 150 ~ 5x10 [ COMPESSION-OPEN SYMBOL

TENSION-SHADED SYMBOL

�9 - - ~

�9 - 0

i -

I0 ~ , I 0 3.0

I a I a h0 2.0

ULTIMATE STI:~AIN, ~U, pet" cent

Fig. 8 - - Smith Failure Envelopes for a Regular Concrete.

the samples in an oven for 30 minutes at 200 ~ with specimen stability achieved using a wooden fixture during curing. To minimize bending moments during the tests, a universal joint was attached to each end of the test configuration.

The test temperatures were - -50 (well below the glass point of the binder), 0, 75, and 150 ~ for uni- axial compression, and 25, 75, and 125 ~ for the uniaxial tension tests. The upper temperature extreme for the tension tests was limited due to the low strength and fragility of the specimens. An attempt was made to obtain 150~ tension data, and over 20 specimens failed during placement of the samples in the fixtures or before the test could be initiated. The nominal strain rates used at each temperature were 0.0004, 0.04, and 4 in./(in.-min.), which are the slowest, medium, and fastest strain rates that can be obtained with the given machine using a test specimen having a 5 in. gauge length.

The test data were reduced using the indicated relations included in the nomenclature, and the tabu- lated results were reported elsewhere [I]. The results for the uniaxial compression and tension properties of both mixes considered here are graphically por- trayed in figures 4 through 7, where the properties are given in terms of temperature and strain rate.

DISCUSSION OF RESULTS

As shown in figures 4-7, the ultimate strain varied in a relatively minor fashion while the ultimate stress varied drastically as a function of rate of loading and temperature. These trends as well as the numerical

444

I0 4

10 5

ULTIMATE STRESS, 10 2

O-u, p s i

IO I

' i ' I ' I SYMBOL T FLA'- NOMINAL

o -~o ~~ 5~'~ - X 2~ ~ s~,~:~-

" ~ ~" 0 T5 O- 5xlO , - ~ a ~z5 0 5xtO- o'

0 - ~ 0 150 -0 5xlO COMPrESSiON "OPEN SYMBOL

-- ~ ~ TENSION SHADED SYMBOL

\

I0" , I , I , I 0 L0 2.0 ~.0

ULTIMATE STRAIN, EU, per cent

Fig. 9 - - Smith Failure Envelopes for an Overheated Concrete.

values compare favorably with the previous work by Tons and Krokosky [15] even though different pro- cedures for mixture compaction and specimen prepar- ation were used here.

The Smith failure envelopes for uniaxial tension and compression are shown in figures 8 and 9 for the regular mixture and the overheated mixture, respectively, and a composite of these envelopes is given by figure 10. The net result of overheating the mix was that the ultimate stress increased, the ulti- mate strain decreased, and the air voids decreased. In particular the maximum ultimate compression strain decreased from 3 to 1.8 ~o while the maximum ultimate compression stress increased from 1400 to about 3900 psi, and the average air voids decreased from 8.8 to 3.0 7oo due to increasing the mixing and compaction temperature from 300 to 450 ~ The shift in failure envelopes shows how overheating of the mixture is very undesirable for thin flexible pave- ment, since thin pavement essentially follows the deflection of the base for imposed wheel loads, and the resultant strains are usually large.

A few ten-inch length specimens were tested. For instance, of the two data points for T = 75, R ---- 0.02, (fig. 5) one is for a ten-inch length and the other is for a five-inch length, and the difference in au (and eu) is only about 10 ~. The error due to end effects is then negligible relative to other sources of error. The major source of error is believed to be due to the angularity and orientation of the largest size aggregate in the mix since it was noted that all of the tensile specimens failed in a plane at a large stone. The orien- tation of the large stones also affects the mode of rock locking in compression, which was observed

D . B Y N U M - R . N . T R A X L E R

from the unusual nature of a few of the force-deform- ation traces.

The experimental variation of the results is consi- derably less if the maximum aggregate size is small compared to the thickness of the specimen cross section. However, the given aggregate gradation and the specimen thickness corresponds to the gradation and pavement thickness often used on Texas High- ways.

CONCLUSIONS

For an asphaltic concrete mixed at 300 ~ and tested in uniaxial compression, the ultimate stress varied between about 14 and 1400 psi, and the average ultimate strain varied between 1 and 3 ~o, depending upon strain rate and test temperature. For specimens from the regular mix tested in uniaxial tension, the ultimate stress varied between about 16 and 490 psi, and the ultimate strain varied between 0.22 and 1.2 ~o. For an asphaltic concrete mixed at 450 ~ and tested in uniaxial compression, the ultimate stress varied between about 32 and 3500 psi, and the ultimate strain varied between 1.1 and 4 %. For specimens from the temperature-abused mixture tested in uniaxial tension, the ultimate stress varied between about 6 and 480 psi, and the ultimate strain varied between 0.15 and 1.4 %. The Smith failure envelopes showed how increased mixing and compaction temperature increases the maximum ultimate stress and decreases the maximum ultimate strain. The average percent air voids in the specimens mixed and compacted at 300~ and 450 ~ were 8.8 and 3.0, respectively. Since the ultimate

I0 4 | I

COMPRESSDN

.' I

1 0 3 COMPRESSION

450 ~ F MIXj TENSION

ULTIMATE STI~ESS,

O'U, psl

10 2

I01

i0 ~ I I I I 1.0 2.0

ULTIMATE STI~AIN, Eu, per cent

Fig. 10 - - Composite Failure Envelopes.

I 3.0

445

V O L . 6 - N * 36 - 1973 - M A T t ~ R I A I d X E T C O N S T R U C T I O N S

stress varied 2 orders of magnitude while the ultimate strain varied by a factor of about 3 for the extremes of test temperature and strain rates, ultimate normal strain could be used, but ultimate normal stress should not be used as a failure criteria.

ACKNOWLEDGMENT

The meticulous attention to detail exercised by Mrs. Carolyn E. Sanders while typing this manu- script was appreciated.

RI~SUMI~

Les enveloppes de rupture du b6ton bitumineux. On s'est propose, par cette etude, d'obtenir des

donnees sur les variations des propridtes thermo-visco- $lastiques d'un b~ton bitumineux ?z temperature nor- male et d haute temperature au cours du prdchauffage, du malaxage et du compactage. Les essais en traction uniaxiale et en compression uniaxiale ont ~te effectuds sur 5 groupes de 10 dprouvettes ?l diffdrentes vitesses de d~formation constantes et d des temperatures entre - - 4 5 , 5 oC et 65,5 ~ On a obtenu des variations de contrainte ultime de deux ordres de grandeurs alors que la d$formation ultime variait d'un facteur 3 pour

les valeurs extremes de temperatures d'essai et de vitesses de deformation. On a pr#sentd d'une fagon simplifi#e sur un diagramme les 4 propridt#s visco- dlastiques fondamentales pour chaque s#rie d'essais. Les enveloppes de rupture de Smith montrent comment la contrainte ultime maximale augmente et la d#for- mation ultime maximale diminue lorsque les tempk- ratures de malaxage et de compactage augmentent. Les r#sultats fournis par cette dtude soutiennent favora- blement la comparaison avec ceux fournis par les recherches prdc#dentes.

REFERENCES

[1] BYNUM, D., Jr. - - A thermoviscoelastic characteriz- ation of an asphaltic concrete, Res. Rep. 127-2, Aug. 1970, TTI (Texas Transportation Inst., Texas A & M Univ., College Station, Texas 77843).

[2] FINN, F . N . - Factors involved in the design of asphalt- ic pavement surfaces, NCHRP Rep. 39, 1967, HRB.

[3] ALEXANDER, R.L. - - Limits of linear viscoelastic behavior of an asphalt concrete in tension and com- pression. Ph.D. Dissertation, June 1964, Univ. of Calif., Berkeley, California.

[4] BYNUM, D., Jr., TRAXLER, R.N. - - Thermovisco- elastic performance requirements for flexible pave- ment, Res. Rep. 127-1, Aug., 1969, TTI.

[5] SMITH, T.L. - - Stress-strain-time-temperature rela- tionships for polymers. Spec. Tech. Pub1. No. 325, 1962, ASTM.

[61 MONISMITH, V.L., SECOR, G.A., SECOR, K.E. - - Temperature induced stresses and deformations in asphalt concrete, AAPT, V34, 1965, p. 248-285.

[7] BYNUM, D., Jr., TRAXLER, R.N., LEDBETTER, W.B. - - Application of solid propellant technology to flexible pavement design, a paper presented to HRB, Jan. 1970.

[8] ICRPG Solid Propellant Mechanical Behavior Manual, Publ. No. 21, 1968, Army-Navy-Air Force- NASA, Working Group on Mechanical Behavior, Physics Lab., 8621 Georgia Ave., Silver Springs, Maryland.

[9]

[10]

[11]

[12]

[131

It4]

[151

BYNUM, D., Jr., TRAXLER, R.N. - - Gel permeation chromatography data on asphalts before and after service in pavements. AAPT, V39, 1970, p. 683-702.

BYNUM, D., Jr., TRAXLER, R.N., PARKER, H.L., HAM J.S. - - Correlation o f GPC data with hardening of asphalt by oxidation and ultraviolet radiation. J. Inst. Pet. (London), V56, N549, May 1970, p. 147-154.

TRAXLER, R.N., BYNUM, D., Jr. - - Gel permeation chromatography applied to asphalts. Texas Trans- portation Researcher (Texas A & M Univ.), Jan. 1970, p. 3-6.

J[MENEZ, R.A. - - An apparatus for laboratory invest- igations of asphaltic concrete under repeated flexural deformations. Ph.D. Dissertation, Jan. 1962, Texas A & M Univ.

JrMENEZ, R.A. - - Aspects on the binding strength of asphaltic concrete, AAPT, V36, 1967, p. 703-730.

LAYMAN, A.D. - - A study of the flexural properties of a black base. Ph. D. Dissertation, Jan. 1968, Texas A & M Univ.

TONS, E, KROKOSKY, E.M. - - Tensile properties of dense graded bituminous concrete, AAPT, V32, 1963, p. 497-529.

446