Performance-Based Design of Wood Structures

FORUM

The Next Step for AF&PA/ASCE 16-95: Performance-Based Design of Wood Structures John W. van de Lindt, M.ASCE Associate Professor, Colorado State Univ., Civil and Environmental Engineering, Campus Mail Stop 1372, Fort Collins, CO 80523-1372. E-mail: [email protected]

Yue Li, M.ASCE Assistant Professor, Michigan Technological Univ., Dept. of Civil and Environmental Engineering, 1400 Townsend Dr., Houghton, MI 49931.

William M. Bulleit, M.ASCE Professor and Chair, Michigan Technological Univ., Dept. of Civil and Environmental Engineering, 1400 Townsend Dr., Houghton, MI 49931.

Rakesh Gupta, M.ASCE Associate Professor, Oregon State Univ., Dept. of Wood Science and Engineering, 114 Richardson Hall, Corvallis, OR 97331.

Paul I. Morris Durability and Protection Group Leader, FPInnovations-Forintek Division, 2665 East Mall, Vancouver BC, Canada V6T 1W5.

The concept of improving building performance has, for all practical purposes, always existed at what might be deemed the societal level. If engineered systems do not perform as society and users deem acceptable, these systems do not continue to be constructed and are eliminated over time. The inadequate systems are replaced by improved ones whose performance is expected to be more acceptable to the stakeholders and to society. In modern times this evolution manifests itself as revisions and updates to prescriptive design codes, which typically are anywhere from one to seven years apart. The impetus to formalize performance-based design 共PBD兲 has its roots in Housing and Urban Development’s “Operation Breakthrough” of the 1970s 共NBSIR 1977兲 whose goal was to enable product innovation via industry. However, the wood industry, particularly home building, did not fully embrace innovations for increased performance since they often resulted in higher costs. Subsequently, these higher costs were not tolerated by the marketplace and thus the innovations were not implemented. A recent Structural Engineering Institute 共SEI兲 special project entitled “Keeping Pace with ASCE 7” examined the reliability indexes inherent in the ASCE16-95 standard 共ASCE 1996; Bulleit 2006兲. The ASCE 16-95 committee of the SEI of the American Society of Civil Engineers 共ASCE兲 has the option to move beyond load and resistance factor design 共LRFD兲 and spearhead the next generation of standards for wood design, specifically a performance-based standard. Following the aforementioned SEI special project, the SEI/ASCE committee on Reliability-based Design of Wood Structures embarked on a two-year special project entitled “The Next Step for AF&PA/ASCE 16-95: Performance-Based Design of Wood Structures.” This paper presents the progress made by that project committee and is the first

in a series of general papers that will present a PBD framework for wood structures over the years to come. The performance levels, first articulated at the 1st Invitational Workshop on Performance-based Design of Wood Structures 共van de Lindt 2005兲 are 共1兲 occupant comfort; 共2兲 continued occupancy; 共3兲 injury/life safety; 共4兲 general structural integrity; and 共5兲 manageable loss/damage. Each discussion will address the hazard characterization, performance descriptors that align with the five performance levels, model complexity required for analysis, design procedure verification, and how these performance-based analyses can be extended to design. This discussion order is summarized in Fig. 1. Performance-based design has been defined in a variety of ways, including an engineering approach based on: 共1兲 specific performance objectives and safety goals of building occupants, owners, and the public; 共2兲 probabilistic or deterministic evaluation of hazards; and 共3兲 quantitative evaluation of design alternatives against performance objectives. However, PBD does not prescribe a specific technical solution 共Ellingwood 1998兲. Perhaps a more practical definition, related to seismic design, is that given by Nathan Gould of ABS Consulting: “PBSD 共PerformanceBased Seismic Design兲 is the seismic design methodology of the future. It allows the design team to work together to determine the appropriate levels of ground motion and performance objectives

Hazard Characterization

Performance Descriptors

Modeling Complexity

Verification Procedure

Extension to Design Fig. 1. Organization of each hazard within Part I of the SEI special project JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2009 / 611

Downloaded 06 Oct 2009 to 128.193.116.248. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright

for the building and nonstructural components in order to meet the owners expectations.” However, these definitions and visions for PBD are provided in a general building sense and not specific to wood frame construction. Wood structural systems, particularly light frame wood structures, require consideration of several issues that structures made of other materials such as steel and concrete do not face. Issue 1: Current Design Philosophy Much of current wood frame construction is not designed, but rather is conventional or prescriptive, using essentially “rules of thumb” and standard construction practices rather than engineering calculations for each structure. Thus, the application of PBD is, in a way, significantly beyond current wood construction practice in many 共but not all兲 cases. This is not true for all wood construction, however 共e.g., in seismic regions, particularly California, design calculations are explicitly required兲. This type of engineered design is also evident in unique custom homes 共i.e., high-end wood construction兲. Issue 2: Value of Wood Structures Most of wood frame buildings, particularly residential, are valued at less that $1 million and many are on the order of several hundred thousand dollars. The median price of a new house in the United States in September 2008 was $218,000. Note that this cost includes the price of the lot/land and varies significantly from region to region. According to the National Association of Home Builders 共NAHB兲 the average size of a U.S. home is approximately 2,300 square feet and throughout the 1990s the majority of U.S. homes were three bedrooms with 60% of homes having four to six rooms total 共i.e., living rooms, kitchen, dining room each counted as a room兲. The cost of performing a PBD is, at this point, significantly more costly than a standard prescriptive design, and therefore the impact on the engineering component of the building cost for wood buildings is very significant. This issue should be thought of as a challenge for the wood engineering research community, specifically how to develop prescriptivetype techniques to achieve specific performance objectives, all the while keeping in mind that this may violate the most general definition given by Ellingwood 共1998兲 above. On the other hand, if one considers that current housing mortgages in the United States currently total $12.4 trillion 共Freddie Mac 2008兲 and are projected to exceed $20 trillion by 2015, improving the performance of the housing stock will provide these structures and communities with resiliency and longevity. Issue 3: Wood Construction Culture and Quality Control The current culture of wood frame building construction, particularly residential construction, does not lend itself well to change. Much of the construction procedure is not well documented, particularly for details of stick-built construction. The use of prefabricated wall panels may help to reduce and change this culture, which is already being pursued by many homebuilders. In addition, consider the differences between the fasteners in a wood shear wall and the bolts in a steel moment connection. Both assemblies provide the lateral resistance for the building, but for the steel connection an exact number of bolts is used and tightened in a prescribed manner. For the wood shear wall built on site, fastener spacing is “eye-balled” rather than measured; further, fas-

teners are often overdriven, and can miss the wall stud entirely diminishing the capacity of the wall 共Rosowsky and Kim 2004兲. It should be noted that this is not a design code issue, but it must somehow be accounted for either through considerations of additional uncertainty in wood wall modeling or inspection practices. Left unaddressed, wood frame buildings may not satisfy performance objectives—not because the performance based design approach is insufficient, but because quality control issues make it impossible to tell what is satisfactory and what is not. Although there are obvious roadblocks to implementation, PBD of wood structures may have the greatest chance of success under certain conditions. Such conditions are 共1兲 modular construction; 共2兲 tract home communities 共e.g., offering performance upgrade packages much the same as current kitchen appliance upgrades兲; and 共3兲 highly customized homes where issue 2 共value of wood structures兲 does not have as significant an impact. The ability to apply PBD to a large portion of the wood structure stock can be realized if issue 2 is addressed. It is unlikely that exactly the same PBD methods used by the steel and concrete design communities can be ported directly for use in wood buildings. Issue 3 and specifically the level of inspection that is performed for wood structures is likely to be an issue as PBD develops for wood structures. As previously mentioned Fig. 1 shows the layout of this paper. Each hazard must be characterized in a meaningful enough way that it can be applied to analysis and design. Once a hazard is characterized, quantifiable performance descriptors must be defined, or at least identified to allow an analyst to determine if performance objectives are met. The values that will be compared to the performance descriptors will come from a numerical model whose complexity must be described for each hazard. Finally, the verification of the performance metrics as well as the extension to design for these concepts is presented. Table 1 presents example performance expectations for each performance level and natural hazard combination. Although the hazard type changes from one load to another, the performance expectation should be defined consistently, to the extent possible, to ensure a consistent level of risk. One tool to estimate costs based on failure is assembly-based vulnerability 共ABV兲, which has been used for seismic analysis 共Porter 2000; Pei and van de Lindt 2009兲 and flood 共van de Lindt and Taggart 2009兲. Although life safety is addressed herein, it is assumed within this project that ASCE’s standard “Minimum Design Loads for Buildings and Other Structures” 共ASCE 2006兲 life safety levels will be met or that level of life safety exceeded by the PBD. Fragility A concept that will be used throughout this paper is fragility. Fragility is essentially a conditional probability of failure 共loss兲 of a structural member or system over one or more input variables. For example, the probability of roof sheathing loss due to a demand exceeding the design capacity can be expressed as P关sheathing loss兴 =

兺 P关sheathing loss兩Demand = x兴P关Demand = x兴

共1兲

where demand is the force 共or other demand兲 on the component; and x is a given demand. The summation is over all the demands. The first term in the summation, i.e., the conditional probability, is defined as the fragility 共Ellingwood et al. 2004; Rosowsky and Ellingwood 2002兲.

612 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2009

Downloaded 06 Oct 2009 to 128.193.116.248. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright

Table 1. Performance Expectations and Corresponding Performance Descriptors for Different Hazards Proposed Occupant comfort

Continued occupancy

Life safety

Structural integrity

Hazard

Little or no reduction in living/ inhabitant comfort

Up to moderate reduction in comfort but no threat to safety or injury. Electrical, plumbing, egress still present

Seismic

1% Interstory drift 共ID兲

2% ID

3–4% ID

7% ID

Wind

No water entry 1 / 8⬙ uplift on panel edges.

Breach of windows or doors. Loss of a portion or full gable or sheathing panel. Strength exceeded analysis.

Roof truss-to-wall connection failure. Supporting column/post failure. Strength exceedance analysis.

Collapse of roof; loss of lateral capacity. Strength based analysis.

Snow

Roof leaks due to ice, etc.

Damage is sufficient enough to allow exposure to weather.

Local not collapse. Ice building up and falling from roof. Strength analysis.

Total roof system collapse. Strength analysis.

Durability

Little or no moisture load.

Low moisture load.

High moisture load.

Very high moisture load.

Flood

Not applicable.

Water recessive by pumping or natural means possible. Ventilation possible/ moldabatement. No sewage backup, no structural damage, no water entry to mechanics and elect.

No structural damage. Flooding may prevent egress. Flooding depth is significantly high.

Visible damage to structure.

Performance-Based Seismic Design This subset of PBD has received significant attention over the last 10 to 15 years, and particularly the past five years as a result of three damaging and deadly earthquakes: Loma Prieta in 1989, Northridge in 1994, and Kobe in 1995. In the latter two earthquakes, both fatalities and financial losses associated with wood structures were considered unacceptable by any standard. The CUREE-Caltech 共Consortium for Universities for Research in Earthquake Engineering—California Institute of Technology兲 Woodframe project, a FEMA-funded project to examine current state-of-the-art wood frame construction in California and make recommendations for improvement, set the groundwork for performance-based seismic design 共PBSD兲 of wood buildings in the United States. Rosowsky 共2002兲 developed mass charts for wood shear walls based on interstory drift requirements as part of that project. Filiatrault and Folz 共2002兲 proposed extending Priestly’s 共1998兲 original direct displacement design 共DDD兲 concept to wood frame buildings, which was recently extended to multistory wood buildings by Pang and Rosowsky 共2007兲. Van de Lindt and Walz 共2003兲 demonstrated the determination of wood shear wall performance based on reliability indices for several different cities around the United States. Nonstructural components in wood frame buildings, which have been shown to play a major role in the dynamic response and, thus, performance of wood frame buildings, were investigated experimentally 共Filiatrault et al. 2004; van de Lindt et al. 2007a兲. More recently, efforts are being made as part of the NEESWood project 共see 具http://www.engr.colostate.edu/NEESWood/典 for details兲 to develop a tiered PBSD philosophy with the focus on midrise buildings 共Pang et al. 2008兲. Also, van de Lindt et al. 共2008兲 developed a 共hysteretic兲 probabilistic system identification approach. Pei and van de Lindt 共2009兲 also have proposed a loss-based PBSD approach for wood frame buildings that allows the building owner 共and/or other stakeholders兲 to use expected loss with their loss management objectives. Li and Ellingwood 共2007兲 determined the reliability of wood frame residential construction under a spectrum of possible earthquakes by looking at drift limits.

Structural integrity is questionable; significant risk of serious injury; safety normally provided not present

Structure is visibly unsafe

Focusing the discussion on wood frame structures up to approximately three to four stories in height, which are common for multifamily buildings, consider again Table 1. The majority of these various performance expectations, with the exception of personal content damage, appear to be best correlated with interstory drifts. Collapse tests in Japan, although not strictly on western-style wood frame buildings, have indicated that soft-story collapse requires very large interstory drifts on the order of 15%. For two- and three-story wood frame buildings, a working interstory drift value of 7% is proposed in Table 1. The other values are relatively consistent with ASCE 41-06 共2007兲 for PBSD. Analysis methods for PBSD use nonlinear hysteretic models that account for degradation during reversing cycles and provide an accurate estimate of interstory drift 共ID兲. Models like this include full structure finite-element models built using software such as ABAQUS, as well as pancake-style models such as SAWS 共Folz and Filiatrault 2004a,b兲 and three-dimensional 共3D兲 models such as SAPWood v2.0 共Pei and van de Lindt 2009兲. Seismic designs based on performance expectations are being conducted currently within the research community but not yet within the practitioner community for wood buildings. Performance-Based Design for Wind Wind hazard is expressed by a wind speed probability distribution for a standard averaging time 共e.g., 3 s兲, exposure 共e.g., C - open兲, and elevation 共e.g., 10 m兲. Because of a lack of statistics, hurricane winds are simulated probabilistically from fundamental climate modeling principles and the resulting wind speeds at specific mileposts are derived from the wind field model. These wind speeds are postprocessed statistically and are used to develop design wind speed maps for the ASCE-7-05 “Minimum Design Loads for Buildings and Other Structures” 共2006兲. The statistics of parameters for wind load equations defined in ASCE-7 can be obtained from a Delphi study 共Ellingwood and Tekie 1999兲. The Weibull distribution was found to be an appropriate model for hurricane wind speed prediction 共Li and Ellingwood 2006兲. JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2009 / 613

Downloaded 06 Oct 2009 to 128.193.116.248. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright

The majority of studies to date have focused on fragility as a function of the design wind speed. This design wind speed is usually the ASCE 7-05 standard 共2006兲 3-s gust and fragilities can be characterized for various components and subassemblies based on these wind pressure models. Posthurricane disaster surveys have shown that the building envelope is the part of residential construction that is most vulnerable to hurricane-induced damage 共NAHB 1993兲. Once the envelope is breached, the building and its contents are increasingly likely to suffer severe damage from water or wind effects. Ellingwood et al. 共2004兲 investigated light-frame wood structures using fragility curves for both wind and earthquake hazard, demonstrating the development of fragilities for roof sheathing, truss spacing, and roof sheathing nail patterns. Lee and Rosowsky 共2005兲 stated that breach of the building envelope and resultant water penetration is the leading cause of financial losses in high winds. Li and Ellingwood 共2006兲 developed hurricane fragility curves for key components that are essential to maintain the light-frame wood building envelop. These components are roof panels, roof-to-wall connections, windows, and glass doors. The failure modes for windows and glass doors can be either excessive wind pressure or projectile impact. For PBD of wood frame structures to wind load it is envisioned that the standard wind loading 共e.g., ASCE 2006兲, could be used when site specific data are not available. Proposed performance descriptors associated with PBD for wind are component and assembly-based in that the most damage occurs to the roof and cladding materials. Again, referring to Table 1, some example damage descriptors associated with the various performance expectations are shown. If the building envelope is breached during a high wind or hurricane event the majority of the damage occurs from wind-driven rainwater intrusion. In fact, a moderate level of gable end damage 共i.e., loss of less than a full sheathing panel兲 to a structure during Hurricane Katrina in 2005 resulted in water damage equal to the purchase price of the house five years earlier 共van de Lindt et al. 2007b兲. Perhaps the closest parallel for PBD for wind can be drawn from seismic design, since losses are the major focus for PBD and not exclusively life safety, which, as discussed earlier, is addressed by ASCE’s standard 7-05 共2006兲. PBD is, by and large, felt by most researchers to be a philosophy that allows performance improvement by the incorporation of system-level behavior using system models. However, in wind engineering most failures occur at the component and subassembly level. A recent paper by Ellingwood et al. 共2006兲 highlights the current status and future challenges for PBD, including wind. In that paper it was stated that guidelines for PBD for wind do not currently exist in the United States. It was also stated that extreme winds 共with the exception of tornado winds兲 are not viewed as a life safety issue in force-based design primarily because of the opportunity for prior warning, which is not true for other hazards such as earthquakes. Finally it was articulated that analysis techniques are needed that model both load- and non-load-bearing walls as an integrated system. In PBWE, considering our performance expectations and damage descriptors in Table 1 共van de Lindt and Dao 2009兲, at a specified peak gust wind speed 共e.g. say 90 mph兲, an owner might expect no damage and no water intrusion. For a well-designed and constructed residential structure this may typically be the case provided wood is in a nondecayed state and fasteners are spaced appropriately. For another level or protection the owner might expect to be provided life safety at 170 mph, meaning 共for example兲 no total loss of the roof. The modeling of wood structures to wind load, especially

light-frame structures, can be done on an almost routine basis, as in the assumptions built around such design-related activities as assigning loads to diaphragms and shear walls. However, analysis also can be done on a more fundamental and specific basis, such as trying to predict connector forces and stresses in sheathing and framing. The required level of analysis for PBD for wind depends on what response parameter may be of interest. In the realm of PBD, a key limit state is the attachment of the sheathing to the framing, as well as the adequacy of windows and doors. As mentioned, loss of sheathing or breakage of windows or doors allows unpredictable air flow and water into the structure that may compromise the stability of the diaphragms and shear walls. Thus, a reliable model must be able to capture the localized effect of wind on specific portions of the roof and walls. In the ideal, this means that, having a full probabilistic time and spatial characterization of the wind, a model must be capable of including the timedependent spatial behavior of the sheathing/connector attachment to the framing over the external surfaces of the structure. In short, this approach mandates that the model has to recognize the continuity in the sheathing, the variability in the quality of the attachment of the sheathing spatially, and how, if at all, the resistance of the connectors to negative pressures is affected by their simultaneous assistance in making the shear walls and diaphragms work structurally. Performance-Based Design for Snow Life safety has been the primary performance objective for snow hazard. However, economic losses due to snow can be severe, even when the life safety objective has been met. For example, the March 1993 East Coast storm cost an estimated $1.75 billion 共O’Rourke and Wrenn 2007兲. In 1998, a major storm affected more than three million people in the eastern United States and Canada and caused major damage. Thus, immediate occupancy and collapse prevention performance objectives should be included in a PBD framework as well as life safety. Limit states, or conditions in which the structural system ceases to perform its intended functions in some way, associated with performance descriptors for snow load must be defined to carry out PBD. System limit states can be either deformation related or strength related. Examples of limit states for occupant comfort or immediate occupancy are excessive window or door header deflection, ice dams producing a breach of envelope or roof leaks, or excessive ice build-up causing ice slides that endanger people and property outside the structure or the structure itself. Partial failure of a roof member or supporting member 共e.g., wall stud兲 with or without building envelope breach would be a threat to life safety. Partial roof failure that compromises structural integrity or complete roof system collapse, whether or not that leads to whole building collapse, should be used as the limit states to provide collapse prevention and assure structural integrity. Currently, LRFD wood design is component-based. To truly examine the effect of snow on wood-frame construction, system models may be required. Simplified beam-spring models 共e.g., Bulleit and Liu 1995; Liu and Bulleit 1995兲 may be appropriate for system analysis. An analog beam models the roof sheathing as a continuous beam across the truss or rafter. Each spring models the sheathing, connectors, and truss or rafter interacting as a partial composite member or T-beam. Such system models will allow analysis of partial roof failure, although extensions of this type of model, based around work such as Cramer and Wolfe 共1989兲, likely will be required for truss systems. From a design stand-

614 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2009

Downloaded 06 Oct 2009 to 128.193.116.248. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright

point, a closed-form limit state approach for system performance is desirable. For instance, using a system factor for analysis of roof system based on analysis of a single roof truss is more feasible from a practitioner’s point of view. The percentage of roof area damaged due to partial roof collapse could be used as a quantitative measure. Modeling roof snow load and response is further complicated by the effects of roof geometry and type on drift loads, snow sliding, and system behavior. Exposure conditions and building orientation also affect drifting and snow removal by wind. Snowdrift modeling is needed to consider all the effects just cited. Another factor to consider is the effect of thermal conditions on snow load and ice dams. Furthermore, the effect of rain-on-snow events on total load and on ice dams cannot be ignored; weight of snow may increase dramatically when rain saturates the snow, and thus rain on snow models will be necessary. Extension of performance-based analysis to design for snow load is a challenge in terms of computational effort since subassembly or system level models may be required to check whether a life safety or collapse prevention performance objective can be achieved. For practical reasons, engineering design needs to be based on physical parameters that can be related to the performance limit states. Damage surveys and insurance data would be the best source of information to develop desired performance levels. Truss roof structural models suitable for PBD can be validated using data from sources such as Wolfe and McCarthy 共1989兲 or Wolfe and LaBissoniere 共1989兲. The acceptance criteria could be at the component level for the immediate occupancy objective, while the subassembly level 共ideally the system level兲 should be considered for life safety and collapse prevention. The proper definition of design snow loads for various performance levels is another issue. Snow loads associated with various return periods 共e.g., 30-year for occupancy comfort, 75-year for life safety, and 100-year for system collapse prevention兲 may need to be considered, which are similar to those used in performance-based earthquake engineering. For the maximum considered snow, the 475-year return period 共or 10% probability of exceedance in 50 years兲 could be considered to model a very rare snow event or represent repeated snow events that lead to significant snow accumulation. The maximum annual roof snow load can be determined as a product of ground snow load, which can be modeled as a lognormal distribution, and the ground-toroof conversion factor, which is also lognormally distributed. Thus, maximum annual roof snow load can be modeled by a lognormal distribution with site-dependent statistics. Other probability distribution functions that might be considered for annual maximum snow load are largest extreme value distribution type I 共Gumbel兲 and largest extreme value distribution type II 共Frechet兲. Ground and roof snow load statistics have been investigated by various researchers 共e.g., Ellingwood and Redfield 1983; O’Rourke and Stiefel 1983; and Taylor 1985兲. Performance-Based Design for Durability Since almost all durability issues in wood frame buildings are tied to moisture 共including most insects兲, as a first attempt, all performance descriptors should be linked to moisture level 共load兲 in the wall cavities, attic, crawl space, and inside the house. Referring to Table 1, one can see that the current performance descriptors are general in nature. It is clear at this point that these need to move from qualitative to quantitative in order for PBD for durability to proceed. Durability as discussed herein is limited to long-term and unnoticed or ignored moisture accumulation from the humid-

ity of exterior and interior air 共vapor diffusion, air leakage兲, plumbing leaks, precipitation 共rain, snow, ice dams, reverse vapor drive兲 or soil moisture. Air moisture loads can be determined by selecting a reference year having, for example, a 30-year return period from the documented range of exterior 共weather兲 and interior relative humidity conditions, and temperature differentials across the building envelope 共Dalgleish et al. 2005兲. Precipitation moisture loads can be determined by selecting a reference year having, for example, a 30-year return period from climate data 共e.g., NOAA兲, including direct rain and wind-blown rain or snow, plus temperature/precipitation conditions causing ice damming and reverse vapor drive. These loads may be offset by drying conditions after precipitation events, but drying should be accorded less weight than wetting. Soil moisture loads can be determined from climate data 共e.g., NOAA兲 for the same prescribed return period; however, these loads will also be influenced by local soil type and water table. For the purposes of developing a model for strength loss of wood components, the decay process under fluctuating conditions can be divided into three stages: 共1兲 establishment, 共2兲 growth and decay, and 共3兲 survival under adverse conditions 共Nofal and Kumaran 2000; Nofal and Morris 2003兲. Time to establishment is the rate-limiting step. Times to establishment and threshold moisture contents for growth and decay of Coniphora puteana 共the cellar fungus兲 on sapwood of European wood species have been determined through extensive work at VTT in Finland 共Viitanen 1997b兲. These data have been used to develop preliminary damage function for wood components 共Nofal and Kumaran 2000; Nofal and Morris 2003兲. However, the Finnish results need to be confirmed for Canadian and U.S. wood products. In addition, the work needs to be done using fungi prevalent in buildings in North America. Some progress has been made in this area 共Morris and Winandy 2002; Clark et al. 2006兲. Finally, the impact of small amounts of decay on the performance of complete systems needs to be evaluated. Again, some progress has been made in this area 共Melencion and Morrell 2007兲, but further work is required. As with strength loss, a model for the development of mold growth should contain three phases: 共1兲 establishment, 共2兲 growth and sporulation, and 共3兲 survival under adverse conditions. Time to establishment and threshold moisture contents for mold growth on sapwood of European wood species has been determined through extensive work at VTT in Finland 共Viitanen 1997a; Viitanen and Ojanen 2007兲. The Finnish results have been supplemented by data on Canadian and U.S. wood products 共Yang and Bisson 2004; Yang 2007兲. However, data need still to be developed for nonstill air conditions. Furthermore, the relationship between mold growth in wall cavities and mold spore concentrations in the living space needs to be determined 共Fazio et al. 2005兲. The second type of model should be a probabilistic model of insect attack events. Such a model has been developed by Australian scientists 共Leicester et al. 2003; Leicester and Wang 2004兲; however, further work needs to be done to develop similar model for the United States. Performance-Based Design for Flood Flood hazard differs from earthquakes, wind, and snow in that loss reduction is dealt with through zoning and insurance regulations. However, there are certain instances where it is advantageous to demonstrate the effect of relocation, elevating, or wet flood proofing. As such, flooding hazard needs to be characterized in at least the following three ways: 共1兲 flood depth, 共2兲 water velocity, and 共3兲 flood duration. Another factor that should evenJOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2009 / 615

Downloaded 06 Oct 2009 to 128.193.116.248. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright

tually be included is water contaminants since these may have an effect on the amount of loss for a wood frame 共or any兲 structure as the result of flooding. Flood depth has been selected as the primary hazard characterization here because it will determine which building components are affected by the flood and will be applicable in any flood scenario. The two secondary hazard characterizations are water velocity and flood duration and are more correlated with extent of damage to the various components and assemblies. Water contaminants have been excluded here due to the difficulty of predicting the probability of specific contaminants and their concentrations while accounting for the effects caused by different types of contaminants over a range of concentrations. Certain items and systems in a building 共e.g., electrical, appliances兲 can be considered to be nearly instantaneously damaged by flooding. Thus, the area of a certain building system that is in contact with the flood water, or location of an item causing it to be in contact with flood water, based on a specific flood depth, is the loading associated with flood depth. Certain building items or systems can withstand some water contact but will deteriorate with time. Therefore, the time the building system is in contact with water is the loading associated with flood duration. Finally, certain building systems may be damaged by water velocity leading to increased damage or collapse. The return period flood depth for flooding hazard can be taken from data already available through the Federal Emergency Management Agency 共FEMA兲 flood maps associated with the National Flood Insurance Program. These flood maps, associated with 100-year and 500-year flood depths, are then used to develop probabilities of various flood depths over the time period of ownership of a building. If available, data should be used to generate flood duration information; then local topography and distance from the flood source, as well as distance from the coastline and surge heights for hurricane events, can be used to estimate water velocity data. The load/response model for the flooding hazard will conclude with the development of a structure-specific location-specific fragility curve. The fragility, which might be, for example, the 90th percentile for annualized dollar loss as a function of flood return period, can be used for design. Generation of the fragilities should be based on an assembly-based vulnerability 共ABV兲 procedure that, using flood depth and duration, identifies the extent of the damage to each component and/or assembly 共Porter 2000; van de Lindt and Taggart 2009兲. ABV is a procedure that sums all dollar loses for each component in a building depending on the hazard intensity. In the present discussion the hazard level is flood depth and duration. When data exists or reasonable assumptions can be made, statistical distributions are assigned to costs. The ABV procedure is repeated multiple times, essentially using a Monte Carlo simulation 共MCS兲 framework. The result is a number of response results for the building, which can be rank ordered to form a cumulative distribution function 共CDF兲. The procedure is repeated to generate a CDF at each flood depth. Extraction from this ensemble of CDFs at a particular percentile provides the fragility curve, which is a vulnerability curve since it is damage as a function of flood depth. The monetary losses, at this stage, are calculated based on estimation of damage to the structure due to flood depth and duration and the associated cost of repair determined using data from the “RS Means Repair and Remodeling Cost Data” 共2003兲. To verify the proposed flood damage calculation methodology, one would need to compare the calculated damage with actual costs of repair. Some data may be available from FEMA based on

flood damage claims. A single flood could provide a significant amount of data from various building types as well as various flood depths. To perform a design for flood based on performance expectation, all the costs associated with a certain flood level and duration would need to be summed. A computer program for loss estimation from flooding was developed by Taggart 共2007兲. The software uses the same logic as ABV for earthquake loss estimation developed by Porter 共2000兲, but simply as a function of flood depth instead of spectral acceleration. Design would consist of a comparison between the expected cost savings versus cost of the improved construction 共or retrofit兲 approach. This might be as simple as raising grade level or as complex as building an elevated structure on columns 共e.g., wood piles in coastal areas兲 for a new building, and replacement of siding and raising appliances for an existing building.

Conclusion The current status of PBD for earthquakes, wind, snow, durability 共moisture and insects兲, and flood has been summarized herein. Some examples of research, completed, ongoing, or needed, have been provided. In order for AF&PA/ASCE 16-95 to move to a performance-based standard, two specific events should be put in motion. First, the current ASCE 16-95 committee should finalize the standard. Second, the committee should be reinstated/ reinvented to begin development of the performance-based standard for wood construction. The current ASCE/SEI Committee on the Reliability-based Design of Wood Structures served ASCE 16-95 in their development of the standard and can continue to serve during the development of the performance-based standard, either under a different or the same name. The expectations and choices within the document can be articulated by the committee, but a large body of research will be needed to provide numerical models and correlations between damage descriptors from those models and realistic building performance.

Acknowledgments This paper resulted from the Structural Engineering Institute 共SEI兲 of the American Society of Civil Engineers 共ASCE兲 special project entitled “The Next Step for ASCE 16-95: PerformanceBased Design of Wood Structures.” Their support of the Committee on the Reliability-Based Design of Wood Structures Special Project is sincerely appreciated. The 1st Invitational Workshop on Performance-Based Design of Wood Structures was held in Fort Collins, Colorado, in 2005. Those workshop participants 共list available on page 11 at 具http://www.engr.colostate.edu/⬃jwv/典兲 are acknowledged for their helpful discussion. Philip Line’s 共AF&PA兲 contributions to the durability section, and Dan Wheat’s 共UT Austin兲 contribution to the wind modeling section of this paper are both gratefully acknowledged.

References ASCE. 共1996兲. “Standard for load and resistance factor design 共LRFD兲 for engineered wood construction 共ASCE16-95兲.” ASCE, New York. ASCE. 共2006兲. “Minimum design loads for buildings and other structures, ASCE/SEI 7–05.” ASCE, Reston, Va.

616 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2009

Downloaded 06 Oct 2009 to 128.193.116.248. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright

ASCE. 共2007兲. “Seismic rehabilitation of existing buildings, ASCE/SEI 41-06.” ASCE, Reston, Va. Bulleit, W. M. 共2006兲. “Reliability of steel doweled wood connections designed to ASCE 16-95.” J. Struct. Eng., 132共3兲, 441–448. Bulleit, W. M., and Liu, W.-F. 共1995兲. “First-order reliability analysis of wood structural systems.” J. Struct. Eng., 121共3兲, 517–529. Clark, J. E., Symons, P., and Morris, P. I. 共2006兲. “Resistance of wood sheathing to decay.” Proc., Wood Protection 2006, Forest Products Society, Madison, Wis. Cramer, S. M., and Wolfe, R. W. 共1989兲. “Load-distribution model for light-frame wood roof assemblies.” J. Struct. Eng., 115共10兲, 2603– 2616. Dalgliesh, W. A., Cornick, S. M., Maref, W., and Mukhopadhyaya, P. 共2005兲. “Hygrothermal performance of building envelopes: Uses for 2D and 1D simulation.” Proc., 10th Conf. on Building Science and Technology, Building Envelope Council Ottawa Region, Ontario. Ellingwood, B. 共1998兲. “Reliability-based performance concept for building construction.” Structural engineering world wide 1998 共CDROM兲, Paper T178–4, Elsevier, New York. Ellingwood, B. R., and Redfield, R. 共1983兲. “Ground snow loads for structural design.” J. Struct. Eng., 109共4兲, 950–964. Ellingwood, B. R., Rosowsky, D. V., Li, Y., and Kim, J. H. 共2004兲. “Fragility assessment of light-frame wood construction subjected to wind and earthquake hazards.” J. Struct. Eng., 130共12兲, 1921–1930. Ellingwood, B. R., and Tekie, P. B. 共1999兲. “Wind load statistics for probability-based structural design.” J. Struct. Eng., 125共4兲, 453–463. Ellingwood, B. R., van de Lindt, J. W., Gromala, D., Rosowsky, D. V., Gupta, R., and Pryor, S. 共2006兲. “Performance-based engineering for light-frame wood construction in the United States: Status and challenges.” Proc., 2006 World Conference on Timber Engineering. Fazio, P., Bartlett, K., Yang, D.-Q., Rao, J., and Miao, G. 共2005兲. “Experimental evaluation of potential movement of mold spores from wall cavities to indoor environment.” Proc., 10th Canadian Conference on Building Science, Building Envelope Council Ottawa Region, Ontario. Filiatrault, A., Fischer, D., Folz, B., and Uang, C.-M. 共2004兲. “Seismic testing of a two-story woodframe house: Influence of wall finish materials.” J. Struct. Eng., 128共1兲, 1337–1345. Filiatrault, A., and Folz, B. 共2002兲. “Performance-based seismic design of wood framed buildings.” J. Struct. Eng., 128共1兲, 39–47. Folz, B., and Filiatrault, A. 共2004a兲. “Seismic analysis of woodframe structures buildings I: Model formulation.” J. Struct. Eng., 130共9兲, 1353–1360. Folz, B., and Filiatrault, A. 共2004b兲. “Seismic analysis of woodframe structures buildings. II: Model implementation and verification.” J. Struct. Eng., 130共9兲, 1361–1370. Freddie Mac. 共2008兲. Investor presentation, 具http://www.freddiemac.com/ investors/pdffiles/investor-presentation.pdf典 共Nov. 3, 2008兲. Lee, K. H., and Rosowsky, D. V. 共2005兲. “Fragility assessment for roof sheathing failure in high wind regions.” Eng. Struct., 27共6兲, 857–868. Leicester, R. H., and Wang, C.-H. 共2004兲. “A probabilistic model for termite attack on housing in Australia.” Wood frame housing, durability and disaster issues, Forest Products Society, Madison, Wis. Leicester, R. H., Wang, C. H., and Cookson, L. J. 共2003兲. “A risk model for termite attack in Australia.” IRG/WP 03–10468, The International Research Group on Wood Preservation, IRG Secretariat, Stockholm, Sweden. Li, Y., and Ellingwood, B. R. 共2006兲. “Hurricane damage to residential construction in the U.S.; Importance of uncertainty modeling in risk assessment.” Eng. Struct., 28共7兲, 1009–1018. Li, Y., and Ellingwood, B. R. 共2007兲. “Reliability of wood-frame residential construction subjected to earthquakes.” Struct. Safety, 29共4兲, 294–307. Liu, W.-F., and Bulleit, W. M. 共1995兲. “Approximate reliability analysis of wood structural systems.” Struct. Safety, 17共1兲, 59–78. Melencion, N., and Morrell, J. J. 共2007兲. “Effect of fungal attack on

maximum load capacity of simulated wall assemblies.” IRG/WP/07– 20363, International Research Group on Wood Preservation, IRG, Stockholm, Sweden. Morris, P. I., and Winandy, J. E. 共2002兲. “Limiting conditions for decay in wood systems.” IRG/WP/02–10421, International Research Group on Wood Preservation, IRG Seretariat, Stockholm, Sweden. National Association of Home Builders 共NAHB兲. 共1993兲. Assessment of damage to single-family homes caused by hurricane Andrew and Inik, NAHB Research Center Report, Upper Marlboro, Md. National Bureau of Standards Internal Report 共NBSIR兲. 共1977兲. “Performance criteria resource document for innovative construction.” Rep. No. NBSIR 77-1316, National Institute of Standards and Technology, Washington, D.C. 共available from NTIS兲. Nofal, M., and Kumaran, M. K. 共2000兲. “On implementing experimental biological damage-functions models in durability assessment systems.” Proc., Bugs, Mold, and Rot III. Nofal, M., and Morris, P. I. 共2003兲. “Criteria for unacceptable damage on wood systems.” Proc. Japan-Canada Conf. on Building Envelope, National Research Council Institute for Research in Construction, 1–14. O’Rourke, M. J., and Stiefel, U. 共1983兲. “Roof snow loads for structural design.” J. Struct. Eng., 109共7兲, 1527–1537. O’Rourke, M. J., and Wrenn, P. 共2007兲. Snow loads: A guide to the sse and understanding of the snow load provisions of ASCE 7–05, ASCE, Reston, Va. Pang, W., Pei, S., Liu, H., van de Lindt, J. W., and Rosowsky, D. V. 共2008兲. “Tiered approach to performance-based seismic design of wood frame buildings.” 2008 Structures Congress, Vancouver, Canada. Pang, W., and Rosowsky, D. V. 共2007兲. “Direct displacement procedure for performance-based seismic design of multistory woodframe structures.” NEESWood Project Rep. NW-02, 具http://www.engr.colostate. edu/NEESWood/publications.html典. Pei, S., and van de Lindt, J. W. 共2009兲. “Methodology for long-term seismic loss estimation in woodframe buildings.” Structural Safety, 31, 31–42. Porter, K. A. 共2000兲. “Assembly-based vulnerability of buildings and its uses in seismic performance evaluation and risk-management decision-making.” Ph.D. thesis, Stanford Univ., Stanford, Calif. Rosowsky, D. V. 共2002兲. “Reliability-based seismic design of wood shearwalls.” J. Struct. Eng., 128共11兲, 1439–1453. Rosowsky, D. V., and Ellingwood, B. R. 共2002兲. “Performance-based engineering of wood frame housing: A fragility analysis methodology.” J. Struct. Eng., 128共1兲, 32–38. Rosowsky, D. V., and Kim, J. H. 共2004兲. “Incorporating nonstructural finish effects and construction quality issues into a performance-based framework for wood shearwall welection.” Proc., Structure Congress, American Society of Civil Engineers, Washington, D.C. RS Means Co. 共2003兲. Repair & remodeling cost data: Square foot costs, 24th Ann. Ed., Kingston, Mass. Taggart, M. 共2007兲. “Performance-based design of wood buildings for flood.” MS thesis, Colorado State University, Fort Collins, Colo. Taylor, D. A. 共1985兲. “Snow loads on sloping roofs: Two pilot studies in the Ottawa area.” Can. J. Civ. Eng., 12, 334–343. van de Lindt, J. W. 共2005兲. The E-proceedings of the 1st invitational Workshop on Performance-Based Design of Woodframe Structures, Colorado State Univ., Fort Collins, Colo. van de Lindt, J. W., and Dao, T. N. 共2009兲. “Performance-based wind engineering for woodframe buildings.” J. Struct. Eng., 134共2兲, 169– 177. van de Lindt, J. W., Graettinger, A., Gupta, R., Skaggs, T., Pryor, S., and Fridley, K. 共2007a兲. “Performance of woodframe structures during Hurricane Katrina.” J. Perform. Constr. Facil., 21共2兲, 108–116. van de Lindt, J. W., Liu, H., and Pei, S. 共2007b兲. “Performance of a woodframe structure during full-scale shake table tests: Drift, damage, and effect of partition wall.” J. Perform. Constr. Facil., 21共1兲, 35–43. JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2009 / 617

Downloaded 06 Oct 2009 to 128.193.116.248. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright

van de Lindt, J. W., Pei, S., and Liu, H. 共2008兲. “Performance-based seismic design of woodframe buildings using a system identification concept.” J. Struct. Eng., 134共2兲, 240–247. van de Lindt, J. W., and Taggart, M. 共2009兲. “Performance-based design of residential structures for flood using fragility analysis methodology.” Nat. Hazards Rev., in press. van de Lindt, J. W., and Walz, M. A. 共2003兲. “Development and application of a wood shear wall reliability model.” J. Struct. Eng., 129共3兲, 405–413. Viitanen, H. 共1997a兲. “Modeling the time factor in the development of mold fungi—The effect of critical humidity and temperature conditions on pine and spruce sapwood.” Holzforschung, 51共1兲, 6–14. Viitanen, H. 共1997b兲. “Modeling the time factor in the development of brown-rot-decay in pine and spruce sapwood—The effect of critical humidity and temperature conditions.” Holzforschung, 51共2兲, 99–106. Viitanen, H., and Ojanen, T. 共2007兲. “Improved model to predict mold

growth in building materials.” 2007 ASHRAE/Proc., 10th Int. Conf. on Thermal Performance of the Exterior Envelopes of Whole Buildings X, American Society of Heating, Refrigerating, and Air Conditioning Engineers, Atlanta. Wolfe, R. W., and LaBissoniere, T. 共1991兲. “Structural performance of light-frame roof assemblies: II. Conventional truss assemblies.” FPLRP-499, USDA Forest Products Lab, Madison, Wis. Wolfe, R. W., and McCarthy, M. 共1989兲. “Structural performance of light-frame roof assemblies: I. Truss assemblies with high stiffness variability.” FPL-RP-492, USDA Forest Products Lab, Madison, Wis. Yang, D.-Q. 共2007兲. “Control of moisture content in wood materials for mold growth.” Report no. IRG/WP 07-10630, The International Research Group on Wood Preservation, IRG Secretariat, Stockholm, Sweden. Yang, D.-Q., and Bisson, M.-C. 共2004兲. Wood frame housing, durability and disaster issues, Forest Products Society, Madison, Wis.

618 / JOURNAL OF STRUCTURAL ENGINEERING © ASCE / JUNE 2009

Downloaded 06 Oct 2009 to 128.193.116.248. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright