Widespan Membrane Roof Structures:Design Assisted by Experimental Analysis M.Majowieckil IUAV Facolta di Architectura University of Venice,ITALY majo@mail.asianet.it Summary.Wide span structures are today widely applied for sport,social,indus- trial,ecological and other activities.The erperience collected in last decades identified structural typologies as space structures,cable structures,membrane structures and new-under tension-efficient materials which combination deals with lightweight structural systems,as the state of art on long span structural design.In order to increase the reliability assessment of wide span structural systems a knowledge based synthetical conceptual design approach is recommended.Theoretical and erperimen- tal in scale analysis,combined with a monitoring control of the subsequent perfor- mance of the structural system,can calibrate mathematical modelling and evaluate long term sufficiency of design. Key words:Wide span structures,snow and wind loading,experimental analysis, reliability 1 Introduction Considering the statistical results of [1],the unusual typologies,new materials and the "scale effect"of long span structures,several special design aspects arise as: the snow distribution and accumulations on large covering areas in function of statistically correlated wind direction and intensity; the wind pressure distribution on large areas considering theoretical and exper- imental correlated power spectral densities or time histories; rigid and aeroelastic response of large structures under the action of cross- correlated random wind action considering static,quasi-static and resonant con- tributions; the time dependent effect of coactive indirect actions as pre-stressing,short and long term creeping and temperature effects; the local and global structural instability; the non linear geometric and material behaviour; reliability and safety factors of new hi-tech composite materials; 173 E.Onate and B.Kroplin (eds.).Textile Composites and Inflatable Structures,173-194. 2005 Springer.Printed in the Netherlands
Widespan Membrane Roof Structures: Design Assisted by Experimental Analysis M. Majowiecki 1 IUAV Facolt`a di Architectura ` University of Venice, ITALY majo@mail.asianet.it Summary. Wide span structures are today widely applied for sport, social, industrial, ecological and other activities. The experience collected in last decades identified structural typologies as space structures, cable structures, membrane structures and new - under tension - efficient materials which combination deals with lightweight structural systems, as the state of art on long span structural design. In order to increase the reliability assessment of wide span structural systems a knowledge based synthetical conceptual design approach is recommended. Theoretical and experimental in scale analysis, combined with a monitoring control of the subsequent performance of the structural system, can calibrate mathematical modelling and evaluate long term sufficiency of design. Key words: Wide span structures , snow and wind loading, experimental analysis, reliability 1 Introduction Considering the statistical results of [1], the unusual typologies, new materials and the “scale effect” of long span structures, several special design aspects arise as: - the snow distribution and accumulations on large covering areas in function of statistically correlated wind direction and intensity; - the wind pressure distribution on large areas considering theoretical and experimental correlated power spectral densities or time histories; - rigid and aeroelastic response of large structures under the action of crosscorrelated random wind action considering static, quasi-static and resonant contributions; - the time dependent effect of coactive indirect actions as pre-stressing, short and long term creeping and temperature effects; - the local and global structural instability; - the non linear geometric and material behaviour; - reliability and safety factors of new hi-tech composite materials; 173 E. Oñate and B. Kröplin (eds.), Textile Composites and Inflatable Structures, 173–194. © 2005 Springer. Printed in the Netherlands
174 M.Majowiecki the necessity to avoid and short-circuit progressive collapse of the structural system due to local secondary structural element and detail accidental failure; the compatibility of internal and external restrains and detail design,with the modeling hypothesis and real structural system response; the parametric sensibility of the structural system depending on the type and degree of static indeterminacy and hybrid collaboration between hardening and softening behaviour of substructures. In the case of movable structures,the knowledge base concerns mainly the mov- ing cranes and the related conceptual design process have to consider existing observations,tests and specifications regarding the behaviour of similar struc- tural systems.In order to fill the gap,the IASS working group no.16 prepared a state of the art report on retractable roof structures [2]including recommen- dations for structural design based on observations of malfunction and failures. From the observations of the in service performance,damages and collapses of all or part of structural systems,we have received many informations and teachings regarding the design and verification under the action of ultimate and serviceability limit states.Limit state violation for engineered structures have lead to spectacular collapses as the Tay (1879)and Tacoma bridges (1940).Sometimes an apparently "unimaginable"phenomenon occurs to cause structural failure.The Tacoma Nar- rows Bridge previously cited was apparently one such a case.It was also a design which departed considerably from earlier suspension bridge design. Long span coverings were subjected to partial and global failures as that of the Hartford Colisseum (1978),the Pontiac Stadium (1982)and the Milan Sport Hall (1985)due to snow storms,the Montreal Olympic Stadium due to wind excitations of the membrane roof(1988),the Minnesota Metrodome(1983)air supported structure that deflated under water ponding,etc.Those cases are lessons to be learned from the structural failure mechanism in order to identify the design and construction uncertainties in reliability assessment.Many novel projects of long span structures attempt to extend the "state of the art".New forms of construction and design techniques generate phenomenological uncertainties about any aspect of the possible behavior of the structure under construction service and extreme conditions. Fortunately.structures rarely fail in a serious manner,but when they do it is often due to causes not directly related to the predicted nominal loading or strength probability distributions.Other factors as human error,negligence,poor workman- ship or neglected loadings are most often involved (Ref 1).Uncertainties related to the design process are also identified in structural modelling which represents the ratio between the actual and the foreseen model's response. According to Pugsley (1973),the main factors which may affect "proneness to structural accidents"are: new or unusual materials; new or unusual methods of construction; new or unusual types of structure; experience and organization of design and construction teams; research and development background; financial climate: industrial climate: political climate
174 M. Majowiecki - the necessity to avoid and short-circuit progressive collapse of the structural system due to local secondary structural element and detail accidental failure; - the compatibility of internal and external restrains and detail design, with the modeling hypothesis and real structural system response; - the parametric sensibility of the structural system depending on the type and degree of static indeterminacy and hybrid collaboration between hardening and softening behaviour of substructures. - In the case of movable structures, the knowledge base concerns mainly the moving cranes and the related conceptual design process have to consider existing observations, tests and specifications regarding the behaviour of similar structural systems. In order to fill the gap, the IASS working group no. 16 prepared a state of the art report on retractable roof structures [2] including recommendations for structural design based on observations of malfunction and failures. From the observations of the in service performance, damages and collapses of all or part of structural systems, we have received many informations and teachings regarding the design and verification under the action of ultimate and serviceability limit states. Limit state violation for engineered structures have lead to spectacular collapses as the Tay (1879) and Tacoma bridges (1940). Sometimes an apparently “unimaginable” phenomenon occurs to cause structural failure. The Tacoma Narrows Bridge previously cited was apparently one such a case. It was also a design which departed considerably from earlier suspension bridge design. Long span coverings were subjected to partial and global failures as that of the Hartford Colisseum (1978), the Pontiac Stadium (1982) and the Milan Sport Hall (1985) due to snow storms, the Montreal Olympic Stadium due to wind excitations of the membrane roof (1988), the Minnesota Metrodome (1983) air supported structure that deflated under water ponding, etc. Those cases are lessons to be learned from the structural failure mechanism in order to identify the design and construction uncertainties in reliability assessment. Many novel projects of long span structures attempt to extend the “state of the art”. New forms of construction and design techniques generate phenomenological uncertainties about any aspect of the possible behavior of the structure under construction service and extreme conditions. Fortunately, structures rarely fail in a serious manner, but when they do it is often due to causes not directly related to the predicted nominal loading or strength probability distributions. Other factors as human error, negligence, poor workmanship or neglected loadings are most often involved (Ref 1). Uncertainties related to the design process are also identified in structural modelling which represents the ratio between the actual and the foreseen model’s response. According to Pugsley (1973), the main factors which may affect “proneness to structural accidents” are: - new or unusual materials; - new or unusual methods of construction; - new or unusual types of structure; - experience and organization of design and construction teams; - research and development background; - financial climate; - industrial climate; - political climate
Widespan Membrane Roof Structures 175 Cause % Inadequate appreciation of loading conditions 43 or structural behaviour Mistakes in drawings or calculations 7 Inadequate information in contract documents 4 or instructions Contravention of requirements in contract 9 documents or instructions Inadequate execution of erection procedure 13 Unforeseeable misuse,abuse and/or sabotage,catastrophe, 7 deteriora tion(partly“unimaginable'”?) Random variations in loading,structure, 10 materials,workmanship,etc. Others 7 Table 1.Prime causes of failure.Adapted from Walker(1981) All these factors fit very well in the field of long span structures involving oftenly something "unusual"and clearly have an influence affecting human interaction. In Table 1,the prime cause of failure gives 43%probability (Walker,1981)to inadequate appreciation of loading conditions or structural behaviour.Apart from ignorance and negligence,it is possible to observe that the underestimation of in- fluence and insufficient knowledge are the most probable factors in observed failure cases (Matousek Schneider,1976). Performance and serviceability limit states violation are also directly related to structural reliability.Expertise in structural detail design,which is normally consid- ered as a micro task in conventional design,have an important role in special long span structures:reducing the model and physical uncertainties and avoiding chain failures of the structual system. According to the author,knowledge and experience are the main human inter- vention factors to filter gross and statistical errors in the normal processes of design, documentation.construction and use of structures. The reliability of the design process in the field of special structures must be checked in the following three principal phases:the conceptual design,analysis,and working design phases. Fig.1.Montreal Olympic Stadium-A cable stayed roof solution
Widespan Membrane Roof Structures 175 Cause % Inadequate appreciation of loading conditions 43 or structural behaviour Mistakes in drawings or calculations 7 Inadequate information in contract documents 4 or instructions Contravention of requirements in contract 9 documents or instructions Inadequate execution of erection procedure 13 Unforeseeable misuse, abuse and/or sabotage, catastrophe, 7 deteriora tion (partly “unimaginable”?) Random variations in loading, structure, 10 materials, workmanship, etc. Others 7 Table 1. Prime causes of failure. Adapted from Walker (1981) All these factors fit very well in the field of long span structures involving oftenly something “unusual” and clearly have an influence affecting human interaction. In Table 1, the prime cause of failure gives 43% probability (Walker, 1981) to inadequate appreciation of loading conditions or structural behaviour. Apart from ignorance and negligence, it is possible to observe that the underestimation of in- fluence and insufficient knowledge are the most probable factors in observed failure cases (Matousek & Schneider, 1976). Performance and serviceability limit states violation are also directly related to structural reliability. Expertise in structural detail design, which is normally considered as a micro task in conventional design, have an important role in special long span structures: reducing the model and physical uncertainties and avoiding chain failures of the structual system. According to the author, knowledge and experience are the main human intervention factors to filter gross and statistical errors in the normal processes of design, documentation, construction and use of structures. The reliability of the design process in the field of special structures must be checked in the following three principal phases: the conceptual design, analysis, and working design phases. Fig. 1. Montreal Olympic Stadium - A cable stayed roof solution
176 M.Majowiecki 1.1 Some Wide Span Enclosures Long span structures needs special investigations concerning the actual live load distribution and intensity on large covering surfaces.Building codes normally are addressed only to small-medium scale projects.The uncertainties relate to the ran- dom distribution of live loads on long span structures imply very careful loading analysis using special experimental analysis. Due to the lack of space,only some design analysis illustrations of wide span enclosures,where the author was directly involved,will be included in the present paper with the intention to transmit some experiences that today may be part of the knowledge base. From the direct author's experience in designing large coverings,the most impor- tant experimental investigation regarding live load distribution concerns the snow drift and accumulation factors and the dynamic action of wind loading. 2 Design Assisted by Experimental Analysis 2.1 Snow Loading Experimental Analysis on Scale Models Olympic Stadium in Montreal.During the design of the new roof for the Montreal Olympic Stadium(Fig.1)a special analysis of snow loading was made considering three roof geometries varying the sag of the roof from 10 m,11.5 m and 13 m.,in order to find a minimization of snow accumulation. The experimental investigation was carried out by RWDI [3]to provide design snow according to FAE (Finite Area Element)method,representing up to day a state of the art on the matter. The FAE method uses a combination of wind tunnel tests on a scale model and computer simulation to provide the most accurate assessment possible to estimate 30 year snow loads. Snow loads depend on many cumulative factors such as,snowfall intensity,redis- tribution of snow by the wind (speed and direction),geometry of the building and all surroundings affecting wind flow patterns,absorption of rain in the snowpack, and depletion of snow due to melting and subsequent runoff.The current NBCC (National Building Code of Canada)provides minimum design loads for roofs which are based primarily on field observations made on a variety of roofs and on a sta- tistical analysis of ground snow load data.There are,however,numerous situations where the geometry of the roof being studied and the particulars of the site are not well covered by the general provisions of the code.In these situations,a special study,using analytical,computational and model test methods,can be very benefi- cial since it allows the specific building geometry,site particulars and local climatic factors to all be taken into account.The National Building Code allows these types of studies through its "equivalency"clause and various references to special studies in its commentary. The model of the three new roof shapes were each constructed at 1:400 scale for the wind tunnel tests.The three model roof designs were each instrumented with 90 directional surface wind velocity vector sensors covering the surface.On the console roof,an additional 90 sensors were installed.Measurements of the local wind speed and direction,at an equivalent full-scale height of 1 m above the roof surface,were
176 M. Majowiecki 1.1 Some Wide Span Enclosures Long span structures needs special investigations concerning the actual live load distribution and intensity on large covering surfaces. Building codes normally are addressed only to small-medium scale projects. The uncertainties relate to the random distribution of live loads on long span structures imply very careful loading analysis using special experimental analysis. Due to the lack of space, only some design & analysis illustrations of wide span enclosures, where the author was directly involved, will be included in the present paper with the intention to transmit some experiences that today may be part of the knowledge base. From the direct author’s experience in designing large coverings, the most important experimental investigation regarding live load distribution concerns the snow drift and accumulation factors and the dynamic action of wind loading. 2 Design Assisted by Experimental Analysis 2.1 Snow Loading Experimental Analysis on Scale Models Olympic Stadium in Montreal. During the design of the new roof for the Montreal Olympic Stadium (Fig. 1) a special analysis of snow loading was made considering three roof geometries varying the sag of the roof from 10 m, 11.5 m and 13 m., in order to find a minimization of snow accumulation. The experimental investigation was carried out by RWDI [3] to provide design snow according to FAE (Finite Area Element) method, representing up to day a state of the art on the matter. The FAE method uses a combination of wind tunnel tests on a scale model and computer simulation to provide the most accurate assessment possible to estimate 30 year snow loads. Snow loads depend on many cumulative factors such as, snowfall intensity, redistribution of snow by the wind (speed and direction), geometry of the building and all surroundings affecting wind flow patterns, absorption of rain in the snowpack, and depletion of snow due to melting and subsequent runoff. The current NBCC (National Building Code of Canada) provides minimum design loads for roofs which are based primarily on field observations made on a variety of roofs and on a statistical analysis of ground snow load data. There are, however, numerous situations where the geometry of the roof being studied and the particulars of the site are not well covered by the general provisions of the code. In these situations, a special study, using analytical, computational and model test methods, can be very benefi- cial since it allows the specific building geometry, site particulars and local climatic factors to all be taken into account. The National Building Code allows these types of studies through its “equivalency” clause and various references to special studies in its commentary. The model of the three new roof shapes were each constructed at 1:400 scale for the wind tunnel tests. The three model roof designs were each instrumented with 90◦ directional surface wind velocity vector sensors covering the surface. On the console roof, an additional 90 sensors were installed. Measurements of the local wind speed and direction, at an equivalent full-scale height of 1 m above the roof surface, were
Widespan Membrane Roof Structures 177 taken for 16 wind directions.The wind speed measurements were then converted to ratios of wind speed at the roof surface to the reference wind speed measured at a height equivalent at full scale to 600 m. The 30 year ground snow prediction is obtained by interpolation of the data using the Fisher-Typett type I extreme value distribution method (Fig.2),including both snow and rain (S.+S,),to be 2.8 kPa,which is in agreement with the code value. 3.5 2.5 1.5 0.5 LN(-LN(P)) Fig.2.Fisher-Typett Type 1 extreme values plot ground snow load prediction Fig.3.Comparative analysis of snow loading distribution in function of roof shape (10-13m) Results of structural load cases and local peak loading,not to be considered as acting over the roof simultaneously are shown in Figs.3-4.The shape of the roof with a sag of more than 12m.gives separation of the air flow and turbulence in the wake increasing considerably the possibility of snow accumulations.The order of
Widespan Membrane Roof Structures 177 taken for 16 wind directions. The wind speed measurements were then converted to ratios of wind speed at the roof surface to the reference wind speed measured at a height equivalent at full scale to 600 m. The 30 year ground snow prediction is obtained by interpolation of the data using the Fisher-Typett type I extreme value distribution method (Fig.2), including both snow and rain (Ss + Sr), to be 2.8 kPa, which is in agreement with the code value. Fig. 2. Fisher-Typett Type 1 extreme values plot ground snow load prediction Fig. 3. Comparative analysis of snow loading distribution in function of roof shape (10-13m) Results of structural load cases and local peak loading, not to be considered as acting over the roof simultaneously are shown in Figs. 3-4. The shape of the roof with a sag of more than 12m. gives separation of the air flow and turbulence in the wake increasing considerably the possibility of snow accumulations. The order of