Conf_Structural Design of Glass_2008

www.gpd.fi

GLASS PERFORMANCE DAYS INDIA 2008

Structural Design of Glass Bruce Wymond & Mahesh Arumugam Meinhardt Faade Technology Pty Ltd L2, 400 Kent Street, Sydney, NSW 2000, Australia Keywords 1=Glass 2=Wind Loading 3= Strength 4=Deflection 5=Structural Glazing

ABSTRACT

This paper investigates the different approaches taken internationally in the design and engineering of glazing for buildings. Primarily it focuses on ASTM, BS/EN, AS/NZ, PRC and India Standards, for wind loading and the design of glass.

Wind load calculations, limit state factors, and structural properties of glass and adhesives vary between countries, and it is therefore important to understand these differences, their impact globally and on developing countries in particular. Mixing design criteria between countries is a very common problem which needs to be addressed.

It is therefore intended that a series of clear recommendations be provided that clarify various approaches to glass engineering calculations.

INTRODUCTION

There are two issues that have an overwhelming influence on the structural design of facades, namely the derivation of wind loading and the calculation of material strengths. Both issues are handled differently around

the globe as major countries USA, UK/Europe, PRC and Australia have inconsistent wind loading and material codes for glass, sealants and other materials. Difference have grown out of a national approach to engineering standards, and yet glass, aluminium curtain walls, and structural silicone are global commodities, often designed in one country, manufactured in a second country and shipped to a third country. Adopting a consistent design approach is essential in order to get more efficient design. There is an urgent need to look at conflicting engineering methods and suggest ways that consistent and efficient design can be achieved.

WIND LOADING

Location

Wind loading is primarily determined based on critical wind conditions for a region. This varies substantially for areas that are cyclonic, have strong prevailing winds, are subjected to severe winds generated by storms, or locations that are relatively stable. To assess and compare winds at different locations there is an attempt

to isolate the wind from local ground effects and this is determined through establishing a gradient wind speed for a given location. Then an allowance is made for variations in surface roughness based on the size and density of surrounding buildings.

Return Period

A return period is a measure of the highest wind velocity that will not be exceeded over a specified number of years. The inverse of this is the probability of exceedance which is the wind velocity that will not be exceeded for a given recurrence interval. In AS/NZ1170 Part 2 wind speeds are listed for intervals from V5 to V1000 for a 3 second gust. V500 and V1000 are ultimate limit state winds for normal and high importance buildings, respectively.

Gust Duration

Wind velocities are not constant. They have a background component that changes as a function of the air flow, and a resonant component that changes as a result of the structure or element vibrating. There are short sharp peaks

Figure 1.1

Wind Rose used to determine directional wind velocity

Figure 1.2

Google Earth image used to assess shielding from adjacent buildings

Architectural & Faade Session

2www.gpd.fi


in the resonant component of the wind that are highest for short recurrence intervals. Hence AS/NZ1170 Part 2 refers to a 3 second gust wind speed. That is the peak wind speed that occurs for a 3 second interval. A one minute duration wind speed as specified in some US codes is substantially lower.

Directional Factors

The wind velocity is a function of direction, as prevailing winds are generally stronger from a given direction. Storm fronts are directional and to some extent cyclones have a particular direction of approach. Some codes allow for directional effects by including a wind direction multiplier. Wind roses are provided in some countries which illustrate wind direction, strength and frequency (Refer to Figure 1.1).

Height Factors

As a general rule wind velocity and pressure increases with height and some codes treat pressure as varying

with height but pressure coefficients are constant. In other codes down drafts on the building are considered and corner suction for instance is taken as the same over the full height of the building, based on the peak suction at the top of the building. This results in a substantially different profile of wind pressures over the facade.

Shielding Factors, Topographical Factors and Funnelling Effects

Surrounding buildings often block or partially block high winds and provide a pressure reduction condition (Refer to Figure 1.2). On the other hand wind speeds increase as air flows up a hill. Buildings in close proximity can force wind to be funnelled between them and the air flow speeds up. This can cause the cladding pressures to increase.

Local Pressure Factors

Glass is subject to external as well as internal wind pressures. Local external pressures vary according to

the building shape and proximity to corners as pressures increase due to wind flows around corners (refer to Figure 1.3). Internal pressures also need to be considered and are a function of faade porosity and internal compartmentalization.

Importance Factor

In the event of a natural diaster it is essential that buildings such as hospitals and police stations provide a post disaster function so they cannot fail. To help ensure that essential services are maintained, facades are designed for higher wind loads than normal buildings through the application of an importance factor, an increased limit state factor, or an increase of the return period for a wind event.

OTHER LOADING CONSIDERATIONS

Temperature

The capacity of laminated glass is affected by temperature, as an increase in temperature causes a softening of the interlayer. This effect varies with the type of interlayer that is used. Manufacturers technical information should be assessed.

Internal Pressure - Insulating Glass

The variation in pressure between the inside of an insulating unit (double glazed) unit and the external environment imparts a load on the glass that needs to be considered for buildings constructed at a high altitude or on very tall buildings. Breaker tubes can be utilized to balance internal and external pressure to mitigate this effect.

Concentrated Loading and Impact Loading

Impact loading needs to be considered in safety glass applications and it is normally assessed using soft body impact testing. In cyclonic areas glass

Figure 1.3

Wind pressure distribution allowing for height, edge zones, directionality, and shielding

Internal Pressure - Insulating Glass

The variation in pressure between the inside of an insulating unit (double glazed) unit and the external environment imparts a load on the glass that needs to be considered for buildings constructed at a high altitude or on very tall buildings. Breaker tubes can be utilized to balance internal and external pressure to mitigate this effect.

Concentrated Loading and Impact Loading

Impact loading needs to be considered in safety glass applications and it is normally assessed using soft body impact testing. In cyclonic areas glass need to be designed to prevent the penetration of flying debris. This is usually achieved through the use of laminated glass incorporating a high strength interlayer that retains the glass in its frame through structural silicone glazing or special retention details.

Table 1: Table Comparing Wind Codes

Country United States UK/EU PRC India Australia Wind Loading Code ASCE7-95 BS6399 GB5009-2001 IS875.3 2004* AS1170.2 2002 Permissible/Limit State Limit State Limit State Limit State Permissible Limit State Return Period 50yrs 50yrs 50yrs 50 yrs 500 to 1000yrs Gust Duration 3se c 60m n i 10m n i 3 s c e 3secDirectional Multiplier X x X x 9Shielding Multiplier X x X x 9Topographical Multiplier 9 9 9 9 9Funnelling Effects X 9 X x XImportance Factor 9 x 9 x 9Serviceability Limit State 9 x 9 x 9Other Issues Conversion of 500yr

wind to 50yr wind 1.5. Requires wind tunnel test to assess

shielding and funnelling. Limit state factor for wind is 1.6.

Code not intended to address high rise buildings.

Limit state is derived through multiplying

loads x1.4, and importance is

addressed through a 100 year return period rather than 50 years.

* Denotes draft code, the previous version of IS875.3 was reaffirmed

in 1997. Limit states and importance are not

addressed.

Addresses cyclonic and non-cyclonic areas,

limit states and importance factors are

addressed through selecting the return period of the wind.

LOAD FACTORS

Permissible Design

Traditionally engineering analysis was based on trying to determine the performance of a structure as accurately as possible in terms of stresses and deflections, and then applying factors of safety to reduce the yield or failure strength of the material to a permissible strength limit. The reduction depended on the material type and its uniformity. The advantage of this method was that it gave an accurate picture of structural behaviour of the material. It however did not provide material scientists with separate methods of addressing statistical variations of loads and materials and was therefore phased out.

Limit State Factors

Over the past 20 years capacity design using limit state factors, has taken over from permissible design as the preferred method of analysing structural elements. In limit state design the load multiplication factors are applied to increase the applied loads (dead, live, wind and seismic loads) to allow for load variations, and material factors are applied to reduce material strengths to allow for material variations. The factors that are used vary from one country to the next are highlighted in Table 2. Also the factored loads the capacity design provides unrealistically high deflections, and cannot be used. Hence a separate analysis is required to assess serviceability limit states for deflection.

Serviceability Factors

A problem with limit state factors for strength is that when they are applied they cause the deflection of an element to be over stated. To overcome this problem requires a separate serviceability analysis to be carried out in addition to the limit state analysis. In order to prevent deflections being over-estimated, serviceability limit state factors are used, or serviceability wind speeds are considered.

Conversion of Wind Loads

Conversions of wind loads are often overlooked but it has a substantial impact on design. Consider the following example of changes in North America:

Table 1

Table Comparing Wind Codes

Arch

itect

ural

& F

aad

e Se

ssio

n

www.gpd.fi


need to be designed to prevent the penetration of flying debris. This is usually achieved through the use of laminated glass incorporating a high strength interlayer that retains the glass in its frame through structural silicone glazing or special retention details.

LOAD FACTORS

Permissible Design

Traditionally engineering analysis was based on trying to determine the performance of a structure as accurately as possible in terms of stresses and deflections, and then applying factors of safety to reduce the yield or failure strength of the material to a permissible strength limit. The reduction depended on the material type and its uniformity. The advantage of this method was that it gave an accurate picture of structural behaviour of the material. It however did not provide material scientists with separate methods of addressing statistical variations of loads and materials and it was therefore phased out.

Limit State Factors

Over the past 20 years capacity design using limit state factors, has taken over from permissible design as the preferred method of analysing structural elements. In limit state design the load multiplication factors are applied to increase the applied loads (dead, live, wind and seismic loads) to allow for load variations, and material factors are applied to reduce material strengths to allow for material variations. The factors that are used vary from one country to the next and are highlighted in Table 2.

Serviceability Factors

A problem with limit state factors for strength is that when they are applied they cause the deflection of an element to be over stated. To overcome this problem requires a separate serviceability analysis to be carried out in addition to the limit state analysis.

In order to prevent deflections being over-estimated, serviceability limit state factors are used, or serviceability wind speeds are considered.

Conversion of Wind Loads

Conversions of wind loads are often overlooked but they have a substantial impact on design. Consider the following example of changes in North America:

GANA-2004 Converts 1 minute duration wind to a 3 second gust by multiplying by 1.21 to provide 50 year return 3 second gust data. This approach is approximately only since the relationship between mean pressures and gust pressures varies.Limit state codes then convert 3 second gusts with a 50 year return period to limit state loads by multiplying by 1.4 for instance, or by increasing the return period to 500 to 1,000 years.

The combined effect of this is that if a 3 sec gust limit state wind is used, and

the glass and sealant are checked using old glass codes, or procedures developed for 1 minute mean wind, the loading may have been over-estimated by 1.21 x 1.4 = 1.694 (an increase of 69%). This results in substantial over-design.

GLASS STRENGTH

Glass Surface Compression

Heat treatment of glass to lock in surface compression is the most common method of increasing glass strength. The strength is a function of the glass thickness, surface compression and location on the glass surface (mid panel or edge condition). ASTM E1480 and AS/NZ 1288 provide stress limits for surface compression of heat strengthened and fully toughened glass. In some instances the surface compression range can be controlled to a tighter tolerance than code limits to achieve improved performance characteristics. What is surprising is that heat treated glass

Figure 2

Faade failure in funnelled facades at corner zones and around the glazed link bridge corresponding reasonably closely to points of maximum local pressure according to wind codes

Table 2

Table Comparing Glass Codes

Table 2: Table Comparing Glass Codes

Country USA-Code USA-Industry UK PRC Australia Glass Design Code ASTM E1300 GANA-2004 BS6262.2 2005 JGJ 102-2003 AS1288Permissible/Limit State Permissible Permissible Limit State Limit State Limit State Wind Load Duration 3 sec 60 sec 60min 10 min 3 sec Stress Limits Provided No Yes No Yes Yes Laminated - Load Share Eq. Thickness* No No Chart Only Yes Yes Insulating - Load Share No No No Chart Only Yes Yes Edge Stress Limits Yes No No No YesAnnealed Stress Limit No 2800psi(19.3MPa) No 2828psi(19.5MPa) 4786psi(33MPa) HS Stress Limit No 5600psi(38.6MPa) No - 8412psi(58MPa) FT Stress Limit No 11200psi(77.2MPa) No 8528psi(58.8MPa) 11893psi(82MPa) Other Issues Laminated glass

assessment based on equivalent

thickness being introduced.

GANA-2008 is due to be released in Nov 2008.

Uses a limit state factor of 1.4, does

not address HS glass or provide

guidance on stress limits.

Stress Limits on 5 to 12mm glass, uses a

limit state factor of 1.4 for wind.

Switch from limit state factors to higher return period is

inconsistent with other countries.

Laminated Glass Interlayer Load Sharing

Laminated glass provides a safety function for impact resistance, blast resistance, acoustic performance and solar control. It comprises of two or more sheets of glass that are bonded together using a resin, polyvinyl butyl sheets (PVB) or other special interlayer. Traditionally the interlayer allow some partial composite action as the interlayer is semi-rigid. Recent developments in materials have seen the development of interlayer such as Sentry Glass Interlayer by Du Pont which is more rigid and achieves full composite action. An effective thickness method of converting laminated glass into an equivalent monolithic glass is now being introduced into glass codes.

Probability of breakage

The occurrence of flaws in glass is subject to statistical variation. In order to provide a reasonable degree of safety the probability of failure according to ASTM E1300 is 8/1000 for vertical glass and 1/1000 for overhead glazing.

Thickness tolerance

Glass thickness varies as a consequence of the float glass manufacturing process. ASTM E1300 provides thickness limits that are generally recognized as an industry for glass production in most countries.

Stress limits Centre and Edges

Some codes provide stress limits for glass edges and the centre of glass. Figures are provided for float glass, heat strengthened and fully toughened glass. The limits vary based on short term or long term loading, as well as glass thickness and edge treatment.

GLASS DEFLECTION

Mid span deflection limits

Glass deflection limits are largely subjective as moving glass provides a sense of anxiety for building occupants in stormy weather. The limit of Span/60 or 20mm has been widely adopted for serviceability wind in Asia and Australia over the past 20 years.

Edge Deflection Limits

Glass edges deflect either as free edges for point fixed and patch fixed glass, or they deflect with their supporting frame for curtain wall facades. The deflection limits are specified to prevent metal contact which can lead to premature failure. For insulating glass units the deflection is further limited to prevent shear transfer through the insulating glass unit spacers that could lead to premature unit failure.

Architectural & Faade Session

www.gpd.fi


strength, rather than increase in strength as a result of improved heat treatment processes has been downgraded over the past 20 years according to the FGMA -1986 and GANA-2004.

The strength of Heat Strengthened glass to FGMA in 1986 was 7,000psi and GANA in 2004 allow 5,600psi a fall of 25%, for toughened glass FGMA in 1986 allowed 17,200psi and GANA now allow 12,000psi, a fall of 43% for an 8/1000 probability of failure.

Load Duration

The strength of glass is a function of both the duration of loading and loading history. This is due to growth of Griffith flaws in the glass which create an origin of failure. The fracture mechanics of glass due to the propagation of Griffith flaws is described in numerous texts including Button, D.; Pye, B.: Glass In Buildings, Oxford 1993.

Size of Glass

The capacity of glass is also a function of its size. The larger the glass, the greater the likelihood of a significant Griffith flaw. Design charts often take this effect into account, whereas it is rarely considered in computer modelling based on non-linear elastic analysis.

Insulating Glass Load Sharing

Insulating glass is able to load share as the sealed gas or air inside the insulating glass unit transfers forces from the outer glass to the inner glass in proportion to their relative stiffness. A modification factor is applied to the load sharing to account for minor compression of the air or gas within the unit.

Laminated Glass Interlayer Load Sharing

Laminated glass provides a safety function for impact resistance, blast resistance, acoustic performance and solar control. It comprises of two or more sheets of glass that are bonded together using a resin, polyvinyl butyl sheets (PVB) or other special interlayer. Traditionally the interlayer allow some partial composite action as the interlayer is semi-rigid. Recently the development of interlayer such as Sentry Glass Interlayer by Du Pont which is more rigid and achieves full composite action. An effective thickness method converts laminated glass into an equivalent monolithic glass and is now being introduced into glass codes.

Probability of breakage

The occurrence of flaws in glass is subject to statistical variation. In order to provide a reasonable degree of safety

the probability of failure according to ASTM E1300 is 8/1000 for vertical glass and 1/1000 for overhead glazing.

Thickness tolerance

Glass thickness varies as a consequence of the float glass manufacturing process. ASTM E1300 provides thickness limits that are generally recognized as an industry standard for glass production in most countries.

Stress limits Centre and Edges

Some codes provide stress limits for glass edges and the centre of glass. Figures are provided for float glass, heat strengthened and fully toughened glass. The limits vary based on short term or long term loading, as well as glass thickness and edge treatment.

GLASS DEFLECTION

Mid span deflection limits

Glass deflection limits are largely subjective as moving glass provides a sense of anxiety for building occupants in stormy weather. The limit of Span/60 or 20mm has been widely adopted for serviceability wind in Asia and Australia over the past 20 years.

Edge Deflection Limits

Glass edges deflect either as free edges for point fixed and patch fixed glass, or they deflect with their supporting frame for curtain wall facades. The deflection limits are specified to prevent metal contact which can lead to premature failure. For insulating glass units the deflection is further limited to prevent shear transfer through the insulating glass unit spacers that could lead to premature unit failure.

COMMON ERRORS IN GLAZING DESIGN

Using short duration wind loading 3 second gusts with glass codes or glass tables that are based on 1 minute duration wind: This occurs through mixing wind codes and loading codes and leads to over estimating the glass thickness and wasting material. To overcome this problem an adjustment factor is needed to covert a 3 second gust wind to a 1 minute duration wind speed.

Ultimate limit state load factors used for permissible stress design of glass and structural silicone: This occurs through mixing limit state load tables with permissible stress design codes. It is important to verify whether the stress limit is permissible or ultimate, and make sure that the loading matches.

Over estimating the size of structural

silicone bites for glazing: The allowable stress on structural silicone joints was limited to 138kPa (20 psi) for a 1 minute duration wind loading. This limit needs to be adjusted for short duration gusts (e.g. 3 sec) and needs to be adjusted further to account for ultimate limit state factors. AS1288-2006 recognises this and the limit is 210kPa for ultimate limit state wind loading.

Ultimate limit state load factors used for checking glass deflection: This limit results in an over estimate of the glass deflection. Instead an assessment needs to be made on the basis of serviceability limit state factors, which provide a more realistic assessment of glass deflections.

RECOMMENDATIONS

Code Committees: co-ordinate laterally between disciplines, in particular between loading codes and materials codes to ensure that they are consistent, co-ordinate internationally to ensure assumptions such as wind return periods, gust durations, and limit state factors are treated correctly (not over conservative). Consider the wider implications of their codes beyond their own countries, particularly the use of BS and ASTM codes from a global perspective. ASCE7-2005 and AS1170.2-2002 are the most comprehensive wind codes, with GANA-2004 and AS1288-2006 being the most suitable glass code for international projects.

Glass, Interlayer and Sealant Manufacturers: reassess all materials based on consistent loading assumptions, particularly where the 1 minute duration mean wind speed has been used to assess material properties. Improve material properties by tightening controls over manufacture processes such as tightening limits for surface compression of heat treated glass using modern tempering ovens. Raise the bar on material properties such as sealant tensile capacity above the current 138kPa (20psi) limit.

Curtain Wall and Glass Specialist Engineers: consider all design parameters carefully, and address problems caused by inconsistent codes, rather than making conservative assumptions and over designing. Refine designs as far as possible and utilize wind engineering solutions in order to save cost, materials and to reduce embodied energy in building facades.

Architects & Designers: The most important design issue is to ensure that a building is appropriate for its climate. European style buildings with large areas of very clear glass are not appropriate for much of Asia and the Middle East. When designing a faade address the cost of architectural embellishments, areas of vision glass, treatment of spandrels and use of shading devices to minimize the energy consumption and embodied energy of a building.

Arch

itect

ural

& F

aad

e Se

ssio

n

www.gpd.fi


CONCLUSIONS

Over the past twenty years it is estimated that hundreds of millions of dollars have been wasted as a consequence of over-designing glass and sealant by being over conservative, using a 3 second gust to determine the wind loading when 1 minute wind speed was used to derive glass and structural silicone design data, and by failing to optimize designs

based on wind engineering solutions. Whilst this over-design may have

been overlooked as it provided an added safety margin, it now needs to be recognized as an unnecessary source of CO2 that adds to the embodied energy of buildings, with associated environmental costs.

It is incumbent on leading Architects, Designers, Faade and

Glass Specialist Engineers, Wind Engineers, Code Committees and Material Suppliers to consider the global realities of glass manufacturing and ensure that climate is considered carefully, and that loading code and material code design assumptions are consistent. This is essential in order for us to produce more economical and environmentally efficient buildings. Architectural &

Faade Session

Documents

Conf_Structural Design of Glass_2008