Prestandard for Seismic Assessment and Retrofit of One- and Two-Family Dwellings (ATC-110 Project)

Transcription

1 Prestandard for Seismic Assessment and Retrofit of One- and Two-Family Dwellings (ATC-110 Project) Colin Blaney, Structural Engineer, Buehler & Buehler, San Carlos, California Kelly Cobeen, Structural Engineer Wiss Janney Elstner Assoc., Emeryville, California Andre Filiatrault, Professor State University of New York, Buffalo, New York David P. Welch, Post-Doctoral Researcher Stanford University, Stanford, California Michael Stoner, Graduate Student Clemson University, Clemson, South Carolina Weichiang Pang, Professor Clemson University, Clemson, South Carolina Taylor Vincent, Graduate Student Washington State University, Pullman, Washington Abstract The California Earthquake Authority (CEA) and the Federal Emergency Management Agency (FEMA), through the Applied Technology Council s ATC-110 Project series, have sponsored development of a prestandard for seismic assessment and retrofit of cripple wall, house-over-garage, and hillside vulnerabilities in residential dwellings. Although simplified for the purposes of implementation, methodologies are being developed using best available numerical tools and performance objectives consistent with the philosophies of current seismic codes and standards. After more than two years of work, the project has generated retrofit concepts, preliminary retrofit criteria, and an extensive set of nonlinear analysis results for wood light-frame dwellings. The project is currently developing prestandard provisions and prescriptive plan sets for assessment and retrofit. This paper discusses the project methodology, illustrates retrofit approaches and interim numerical study results, and shares insights gained. Introduction Wood light-frame buildings are the most common type of dwelling in the United States. Although generally providing good performance in past earthquakes, there are well-known vulnerabilities in wood light-frame dwellings that have led to a notable number being rendered uninhabitable or even unrepairable following earthquakes. Improved seismic design and seismic retrofitting of vulnerable configurations will increase the probability that a high percentage of homes are available to provide shelter immediately following moderate to large seismic events. Current model building codes and available seismic retrofit standards do not adequately address the specifics of assessing and retrofitting light-frame one- and two-family wood dwellings. The ATC-110 project has undertaken development of a prestandard addressing practical assessment and retrofit of common seismic vulnerabilities in wood light-frame dwellings, with significant emphasis put on practicality and the intent to encourage wide spread implementation. Overall Objectives and Methodology The prestandard being developed by the ATC-110 Project is intended to provide a single stand-alone resource for addressing assessment and retrofit of selected structural and nonstructural seismic vulnerabilities. Both engineered and prescriptive retrofit design methodologies are being developed for all vulnerabilities being addressed. The prescriptive methods will address dwellings that fall within defined limits of applicability; the engineered methods will be available for dwellings falling outside of those limits. It is anticipated that the prestandard will be subject to further development in an ANSI-approved consensus standard process, after which it will be made available as a standard. 1

2 In addition, plan sets containing prescriptive retrofit design methodologies are being developed where practical. The plan set designs are derived from and complying with the prestandard. The plan sets are being developed based on experience suggesting that prescriptive retrofit designs, presented in plan set form and not requiring the involvement of an engineer, effectively encourage wide-spread implementation of dwelling seismic retrofit. The prestandard will not include triggers for use; policy directing voluntary or mandatory use is outside of the scope of this project. The scope of the project is wood light-frame dwellings, with a primary emphasis on one- and two-family dwellings, and consideration given to buildings with three or more dwelling units, where vulnerabilities and retrofits are similar. The scope includes dwellings located in moderate to high seismic hazard areas, including Seismic Design Categories C and higher. The approach taken by the project focuses on identification and retrofit of specific vulnerabilities, with the objective of risk reduction. This is believed to provide a notably higher benefitto-cost ratio and less invasive than with systematic assessment and retrofit of the entire dwelling. The prestandard is intended to permit addressing an individual vulnerability, multiple vulnerabilities or all identified vulnerabilities. The vulnerabilities addressed include cripple walls and anchorage to the foundation for dwellings with a cripple wall configuration (Figure 1), weak stories and open fronts in dwellings with a house-over-garage configuration (Figure 2) or room-over-garage configuration (Figure 3), and vulnerable anchorage to the foundation in dwellings with a hillside configuration (Figure 4). Also included in the project are assessment and retrofit of masonry chimneys. Figure 2. House-over-garage dwelling configuration. Figure 3. Room-over-garage dwelling configuration. Figure 1. Cripple wall dwelling configuration. Figure 4. Hillside dwelling configuration. 2

3 Previously available guidance on assessment and retrofit of these vulnerabilities has had little or no rigorous study of the improvement in performance resulting from retrofit. While still using previously available retrofit guidance as a starting point, this project has taken the approach of using numerical studies to quantify performance improvements resulting from retrofit. Figure 5 provides a flowchart indicating the overall approach to development. Consistent with the analytically-informed approach, Step 1 is the definition of performance criteria used to guide numerical studies. Step 2 involves development of representative index buildings and initial retrofit designs, and Step 3 involves numerical studies to determine if performance criteria are met. Ideally Step 4 would allow derivation of assessment and engineered retrofit criteria directly from numerical results, but it has been found that some cycling back to the performance criteria is required. As the performance criteria and retrofit criteria are resolved, the end products of development are the assessment method of Step 5, the engineered retrofit methodology, captured in prestandard language in Step 6, and the prescriptive plan sets of Step 7, derived from the Step 6 engineered methodology. project numerical studies might be somewhat high. This understanding is included in judgements made by the project regarding acceptable retrofit performance. In addition, the numerical results are emphasized more as a relative measure of improved performance rather than an absolute measure of performance. The ten percent probability of exceedance was chosen realizing that it would serve as a starting point, to be adjusted as necessary as information from the numerical studies became available; this adjustment is currently being determined. Two secondary criteria have been considered in evaluating the analysis results. These are intended to inform the choice of retrofit criteria, while not necessarily being a deciding factor. The first secondary criterion uses drift as an indicator of the level of repair. This involves tracking transient drift in both upper occupied stories and stories with retrofit. These are compared to drifts identified by research to create increments in repair type and repair cost. The purpose is to understand possible increased damage as a function of level of retrofit. The bases of the selected drift are FEMA P-58 (FEMA, 2012a) fragility functions and CUREE EDA-02 (CUREE, 2010). The criterion uses a transient drift ratio of 0.75% at a seismic demand level corresponding to 30% probability of exceedance in 50 years (140-year mean return period). The other secondary criterion uses drift as an approximate indicator of structural safety for continued occupancy. This relates to possible post-earthquake safety assessment tagging of the building, based on expert judgement. Bases for the selected drift include CUREE EDA-02 and FEMA P-807 Appendix D.9 (FEMA, 2012b). The criterion uses a transient drift ratio of 1.5 % at a seismic demand level corresponding to 10% probability of exceedance in 50 years (475-year mean return period). Figure 5. Flowchart of project approach to development of retrofit solutions. In Step 1, the initial target for the primary performance criterion was chosen as ten percent probability of collapse in the risk-based maximum considered earthquake (MCE R), as determined using the FEMA P-695 (FEMA, 2009) methodology with explicit numerical modeling of collapse. This criterion was chosen to be consistent with current building code performance targets for new buildings, and to make use of the best available numerical modeling tools. The probabilities of collapse being numerically predicted by projects currently using this methodology are higher than is generally anticipated based on observed performance in past earthquakes. Based on this there is a general acknowledgement by the project that probabilities of collapse reported by the The extent to which either of the secondary criteria will affect the prestandard is being decided for each vulnerability as numerical study information becomes available. In recent discussions, it has been suggested that the retrofit criteria be chosen first based on the primary performance criterion. Where possible, the retrofit will be modified to improve the secondary performance criteria, provided it does not notably reduce performance at the primary criterion or notably increase retrofit cost. In Step 4, the retrofit design methodology selected is based on International Building Code (IBC) (ICC, 2015) and ASCE 7 (ASCE, 2010) equivalent lateral force seismic design methods. The use of this methodology is felt to best serve the segment of the engineering community thought to be the target for use, and will keep the cost and complication of retrofit design from becoming too burdensome. Further, the use of performance based studies by the project to develop the retrofit means that 3

4 the retrofits will have the benefit of performance-based design principles without requiring that performance-based design be performed for each dwelling. As part of the IBC/ASCE 7 approach, R-factors are being developed for retrofit of each vulnerability and are expected to differ between vulnerabilities due to significant differences in dwelling seismic response. R-factors are being determined based on numerical studies. It has not yet been decided whether it is necessary to develop overstrength factors, bounding displacements, or stiffness criteria as part of the engineered methodology for retrofit design. Load path connections for retrofit elements are being developed using capacity methods as part of the engineered design methodology. It is necessary that the load path develop the peak capacity of the retrofit elements in order to make valid the probability of collapse studies. If the load path did not develop the retrofit element capacity, probability of collapse could be expected to increase above that reported by the analyses. As the project has progressed, it has found that the numerical studies develop a wealth of information relative to both primary and secondary criteria. The information has been found to vary widely enough between the three vulnerabilities that it has become necessary to revisit and revise the performance criteria separately for each vulnerability. At the time of writing, performance criteria have not been finalized. The discussions that follow capture interim performance considerations for the vulnerabilities studied. Numerical Methodology program is an extension of detailed 2D programs developed earlier for the collapse analysis of light-frame wood shear walls (Pang and Shirazi 2012; Christovasilis and Filiatrault 2010, 2013). The Timber3D program operates on the Matlab platform using a co-rotational formulation and large displacement theory. The horizontal floor and roof diaphragms are modeled using co-rotational 3D, two-node, 12-DOF elastic beam elements, which account for geometric non-linearity. Using a corotational formulation allows proper consideration of the inplane and out-of-plane motions of the diaphragms under large deformations. The elastic flexural and axial stiffness of vertical wall studs are modeled using 3D, two-node, 12-degrees-of-freedom (DOF) elastic frame elements. The vertical wall panel-to-framing assemblies are modeled using 6-DOF, Frame-to-Frame (F2F) link elements. Only one (lateral) DOF of the F2F link element is activated to model the lateral non-linear cyclic response of vertical walls sheathed with wood panels and other (nonstructural) materials. The non-linear lateral cyclic response of vertical walls is captured by the CUREE hysteretic rule (Folz and Filiatrault 2001), as illustrated in Figure 6. The loading forcedeformation paths OA and CD follow a non-linear exponential monotonic envelope curve, while all other unloading and reloading paths exhibit a linear relationship between force and deformation. This hysteretic rule allows for stiffness and strength degradation as well as post-capping reducing strength. The CUREE hysteretic rule is completely determined by ten physically identifiable parameters, as illustrated in Figure 6. A primary task of the numerical studies is to generate data used to measure the performance of dwellings before and after retrofit, thereby allowing the team to compare results to performance criteria and judge the improvement in performance with retrofit. Equally important to project team members developing the assessment and retrofit methodologies is the determination of both global seismic demand and variation in the distribution of seismic demand in the dwellings and their load path. In order to serve these several purposes, three analysis teams (one team studying each vulnerability) generated a range of numerical analysis results including backbone curves and IDAs, and where needed extracted detailed information on load path forces and displacement histories. The numerical studies have used the Timber3D analysis program, a three-dimensional (3D) program originally developed as part of the NEES-Soft project (van de Lindt et al. 2012) to capture the non-linear dynamic response and seismic collapse mechanisms of light-frame wood buildings. This 3D Figure 6. CUREE hysteretic rule for modeling force-displacement response of wood shear walls under cyclic loading. 4

5 A modified version of the CUREE hysteretic rule is also available within Timber3D in order to introduce a user-defined residual strength of vertical walls. The post-capping strength stiffness (r 2K 0) is replaced by a reversed S-shaped curve anchored at a displacement D x and converging to predetermined residual strength level at large displacements, as shown in Figure 7. This modification was used in all of the project numerical studies. properties and modeling assumptions are accounted for using pre-defined dispersion factors according to FEMA P-695. Further, a set of consistent assumptions are used specifically for the collapse performance of the wood light-frame dwellings within ATC-110. This includes using a constant intensity measure of the spectral acceleration at a period of 0.25s; the lowest period allowed by FEMA P-695. Adjustments in median collapse intensity are made to account for using 3D analysis and the absence or presence of large ductility capacity (i.e. large period elongation) according to FEMA P-695. The adjusted median collapse intensity is used in combination with the pre-determined dispersion factors to obtain the final probability of collapse at an intensity of interest (e.g. the MCE R intensity level). Diaphragm (Beam Element with linear in-plane stiffness) Stud(Beam Element + Rigid F2F Connection to Diaphragm + Pinconnection to ground) Figure 7. Modification of CUREE hysteretic rule for modeling residual strength. Figure 8 illustrates a single-story light-frame wood building modeled using the Timber3D analysis program with two different levels of modeling details, namely simplified and intermediate models. In the simplified model, the horizontal diaphragms are modeled using co-rotational 3D, two-node, 12- DOF elastic beam elements with high stiffness resulting in rigid behavior. The intermediate models incorporate more detailed representation of horizontal diaphragms, an explanation of which follows. Both models include vertical wall elements composed of standard non-linear wall building blocks. Using the Timber 3D tool, analytical studies included both initial push-over analysis to provide understanding of peak capacities, and incremental dynamic analyses (IDAs) using non-linear response history analysis. Consistent with the FEMA P-695 methodology, IDAs are conducted in order to obtain reliable estimates of median collapse intensity. Intensities are selected to give feedback for secondary (drift) criteria and the MCE R level. Additional intensities are then conducted in order to better estimate the median collapse intensity if necessary. Currently, the FEMA P-695 far-field set (22 pairs of horizontal ground motions) is used to represent seismic input across the ATC-110 project. Uncertainties due to record-to-record variability, material Diaphragm (Beam Element with linear in-plane stiffness) Strut (Rigid Beam Element Pin-connected at corners) Wall (F2F Element with calibrated hysteretic behavior) Stud(Beam Element + Rigid F2F Connection to Diaphragm + Pinconnection to ground) Wall (F2F Element with calibrated hysteretic behavior) Figure 8. Schematic illustration of one-story, light-frame wood building model in Timber3D (top) simplified model, and (bottom) intermediate model. The level of complexity of numerical models used within ATC-110 is governed by balancing the ability to capture pertinent physical behavior while minimizing the computational onus wherever possible. The models used for cripple wall dwellings represent the most simplified of the different structural types considered. An illustration of a single story cripple wall dwelling and the equivalent Timber3D model is shown in Figure 9. The figure shows that the model is comprised of two stiff (essentially rigid) diaphragms representing the floor of the occupied space and the roof. These are comprised of a series of rigid beam elements with diaphragm masses applied. The diaphragms are separated by pinned stud elements that allow for the vertical geometry between diaphragms to be represented (i.e. cripple wall height 5

6 and first story height). Stud elements also allow for the mass of the vertical wall materials to be accounted for appropriately. The horizontal non-linear force-displacement behavior of different sections and materials of wall elements (between stud elements) are included with 1D frame-to-frame (F2F) elements exhibiting the modified CUREE hysteretic rule (Figure 7), with each element given properties calibrated to available material test data considered within ATC-110. rigid pin-connected boundary members that could be used to determine boundary member (chord) forces. This configuration allowed for the same transfer of vertical load to pinned stud elements and utilized the same 1D frame-to-frame (F2F) elements exhibiting the modified CUREE hysteretic rule as the cripple wall dwelling. An illustration of a house-overgarage configuration and the equivalent Timber3D model is shown in Figure 10. The rather simple modeling assumptions considered for cripple wall dwellings focuses on gaining a better understanding of how differences in global strength and ductility capacities between the cripple wall and the superstructure affect seismic performance; both for existing conditions and dwellings incorporating structural retrofit. By minimizing the complexity of these models, numerous archetype models were able to be studied in order to better define appropriate design considerations that will be implemented in retrofit plan sets for cripple wall dwellings (i.e. R-factors, general limitations of applicability, etc.). Figure 10. Illustration of intermediate Timber3D model used for analysis of house-over-garage dwellings. Cripple Wall and Anchorage Vulnerability Figure 9. Illustration of simplified Timber3D model used for analysis of cripple wall dwellings. In the cases where house-over-garage, room-over-garage, and hillside dwelling configurations were studied, additional information, including chord forces and calibrated diaphragm stiffness, was desirable for a full understanding of the behavior of these configurations. To achieve the increased level of detail in the diaphragm, beam elements calibrated to the desired in-plane elastic shear stiffness were modeling with The cripple wall and anchorage vulnerability is found in wood light-frame dwellings with a crawlspace or basement below the first occupied level, including crawlspaces enclosed by woodframe cripple walls, concrete or masonry stem walls, basement walls, or combinations thereof, on flat to low slope sites. Included in the scope of assessment and retrofit methods addressing this vulnerability are dwellings with cripple walls with heights from 0 (wood floor framing sits directly on foundation or foundation stem wall) to 6-0. The scope of cripple wall dwelling studies is limited to dwellings in which 6

7 the difference in height between its tallest and shortest cripple walls does not exceed 4-0 (Figure 11). Dwellings with a difference in height greater than this are addressed in the hillside dwelling studies. The cripple wall working group is developing assessment and retrofit methods for the cripple wall vulnerability. Retrofit includes plywood sheathing of existing cripple walls studs, connection of the cripple wall to the structure above, and anchorage of the cripple wall to the foundation system. Preliminary provisions have also been included to address replacement of foundation systems where existing foundations are not present or not continuous. Figure 11. Limits of applicability of cripple wall vulnerability assessment and retrofit methods. The primary approaches for flat and low slope sites remain very much the same as that included in the current International Existing Building Code (IEBC) (ICC, 2015b) Appendix Chapter A3 provisions, and the similar provisions adopted into various plan sets, including the FEMA P- 1024RA2 plan set (FEMA, 2015) (Figure 10). Figure 12. Cripple wall vulnerability retrofit concept. Although code prescriptive provisions and plan sets addressing cripple walls and anchorage to foundations are available, outstanding questions regarding this retrofit type have remained. One question is the seismic force level appropriate for retrofit design. When thought of from the standpoint of ASCE 7 R-factors, the appropriate R-factor could be implied to be 2, based on ASCE 7 treatment of vertical combinations of systems, or 6-1/2 based on the materials such as plywood typically used for retrofitting. These different design parameters would result in significantly different solutions in terms of extent, cost, and practicality. As previously mentioned, an initial target of 10% probability of collapse in the risk-targeted maximum considered earthquake (MCE R) as defined in ASCE 7-10 was chosen as the primary performance criterion. While the cripple wall studies originally planned to focus on the cripple wall level, it quickly became evident that the performance of the combined superstructure and cripple wall needed to be considered. Project team consensus also favored that collapses reported by the numerical studies should largely occur within the cripple wall level rather that the occupied stories for a better safeguard to life-safety. In addition the project team was concerned that over-strengthening the cripple wall level could potentially lead to propagation of damage to weaker occupied stories, even under earthquakes with low intensities. Due to the large variation in the configuration of existing cripple wall dwellings, the project team conducted an extensive study with the goal of characterizing a median superstructure in terms of strength and weight, for use in the numerical studies. Representative one and two-story home plans were studied and grouped into six different decades (1900 through 1960) and subsequently categorized in terms of peak lateral strength to weight (V/W) Avg, peak lateral strength to area, (V/A) Avg, baseline weight to area (WBL/A), and strong to weak direction strength ratios. The subscript Avg in these performance parameters relate to the average values considering both principal directions of the building. Three one-story and two two-story plans were selected from each decade and evaluated with combinations of existing finishes including exterior stucco and wood siding, and interior plaster on wood lath and gypsum board. In total, 140 unique combinations from available home plans were analyzed to establish trends of median properties between 1900 and The analysis developed push-over curves to determine the peak story shear capacity, V. A sample of resources used for this study are shown below in Figure 13. The main results for the one-story median study are shown in Figure 14. These results were used to calibrate the numerical models for all analysis runs. 7

8 Figure 13. Sample of Collected Resources Based upon the results of the study, shown in Figure 14, it became evident that both a stronger median and a weaker median minus beta one and two-story superstructure were required to appropriately capture the population of target dwellings. Current thinking is to use the results from the median superstructures to establish the primary performance objective. However, the weaker median minus beta superstructures are being evaluated to investigate level of damage to the superstructure. In general, the median superstructure is anticipated to largely occur in pre-1950 dwelling and correspond with the presence of interior walls with plaster over wood lath or plaster over gypsum lath (button board), and the weaker median minus beta superstructures and largely influenced by the presence of interior walls consisting of gypsum board mainly found in post-1950 dwellings. Figure 15 provides a snap-shot of interim results from the numerical studies that are being used to establish the final performance criteria and retrofit design criteria. This chart plots the probability of exceedance of primary and secondary criteria for a 2-0 cripple wall below a one-story median superstructure. Similar plots have been developed for 4-0 and 6-0 tall cripple walls for both median and median minus beta superstructures. For the configuration shown, it is anticipated that the predicted probability of collapse under MCE R ground motions will be reduced up to 80% relative to the unretrofitted dwelling. Preliminary results also suggest that seismic performance improves modestly as the cripple wall height increases from 2-0 up to 6-0, due to added displacement ductility. For taller cripple wall heights, it is anticipated that P- effects will start to control collapse probabilities. (a) (b) (c) Figure 14. Results from the one-story median study (a) average strength to seismic weight ratios of era specific materials, (b) average strength to weight ratios of era specific materials, and (c) total weight to area ratios of era specific materials. 8

9 anticipated to include excessive drift in the lower story relative to the upper story resulting in significant damage, and possibly a partial or complete story collapse. Included in the scope of assessment and retrofit methods addressing this vulnerability are dwellings with up to 9-0 story clear height in the ground story (Figure 16). Figure 15. Interim results for the one-story 2-0 high median cripple wall While a reasonable estimate of existing cripple wall finishes has been embedded within the numerical models and will influence the choice of an overall R factor, both assessment and retrofit methods will ignore the contribution of existing cripple wall bracing materials (other than existing wood structural panel sheathing) for purposes of retrofit design. This is primarily because the condition of finishes can be widely varying, will be unknown, and will not be practical to determine short of destructive testing. Numerical studies are leading the project team to recommend an R-factor, or R factors that are lower than those used for wood structural panel retrofits in recent retrofit standards and plan sets. This is based on the numerical study predicted probabilities of collapse under MCE R ground motions, and brings predicted probabilities more in line with expectations for hazard reduction or collapse prevention performance objectives in current standards. The final selection of the R factor will include a reasonable balance of anticipated improvement of collapse probability under MCE R ground motions over a wide range of cripple wall dwellings, with the economics and practicality of the strengthening solution. House- or Room-over-Garage Vulnerability The house-over-garage and room-over-garage working group is developing assessment and retrofit methods for vulnerabilities found in wood light-frame dwellings with living space over the garage, where the garage front is unbraced or has minimal lateral bracing. Included are single or multi-level dwellings over a first story consisting of a garage or a combination of a garage and living spaces (Figure 2). Also included are two-story ranch-style configurations, which include bedrooms or other occupancies directly above or partially above a garage (Figure 3). Typical damage modes are Figure 16. Limits of applicability of house- or roomover-garage vulnerability assessment and retrofit methods. The primary approaches to retrofit at the front of house-overgarage configurations include solutions with wood structural panel shear walls where there is enough wall length at the front to allow this retrofit (Figure 17a), and a cantilevered steel column just inboard of the front wall otherwise (Figure 17b). In addition to bracing at the front, these retrofit approaches include transverse wood structural panel bracing at the back wall of the ground story, and wood structural panel bracing on the longitudinal walls. Where house-over-garage configurations have offices or in-law units built into the back of the ground story, alternate designs have been developed to locate retrofit work outside of the built-out spaces. Rather than looking at a range of superstructure capacities, as was done in the cripple wall analytical studies, the house-overgarage working group numerical studies have primarily focused on study of a representative dwelling, consistent with the cripple wall working group s median home. Figure 18 provides a snap-shot of interim results from the numerical studies that are being considered in developing final performance criteria and design criteria for retrofit. This chart plots the probability of collapse under MCE R ground motions and the probability of exceedance of the two secondary criteria previously discussed for a representative house-over-garage with one occupied story. This chart directly illustrates the 9

10 significant reduction in probability of collapse that can occur if any of the studied retrofits are provided. The chart also shows a beneficial decrease in the probability of exceeding the secondary criteria. The wood structural panel retrofit with R=4 is shown to be a desirable retrofit solution. For the cantilevered steel column solutions, the R=4 retrofit satisfies the primary performance criterion. The R=3 retrofit might provide a somewhat better balance between primary and secondary criteria. The difference in cost between these retrofit solutions is believed to be nominal. FEMA P-807 methodology (FEMA, 2012b), which identified an optimum range of retrofitting above which the benefits of retrofit started to decrease. While this behavior is being reflected in a general way in this project through selection of an R-factor, further optimization is not practical due to the variability inherent in the strength and stiffness of the building stock, and the simplified engineering and prescriptive design methodologies to be used for retrofit design. (a) (b) Figure 17. House- or room-over-garage vulnerability retrofit concepts at building front (a) wood structural panel retrofit and (b) cantilevered steel column retrofit. Of interest in comparing the results for the R=2, R=3 and R=4 steel cantilevered column retrofit solutions is that the probability of collapse increases slightly with decreased R- factor. While this seems counter-intuitive, as the R-factor is reduced more inelastic response is being pushed into the occupied story, slightly increasing reported collapses in that story. This is consistent with the sweet-spot concept of the Figure 18. Interim results from house-over-garage working group analysis. Hillside Dwelling Vulnerability The hillside dwelling working group is developing assessment and retrofit methods for vulnerabilities found in wood lightframe dwellings sited on low to steep sloped hillsides with unoccupied space below the lowest framed floor. The unoccupied space in hillside dwellings might be enclosed with crawlspace walls, be open with wood light-frame post and beam systems that have no bracing, wood or steel diagonal bracing, or have skirt walls. Side walls may occur on stepped or sloped continuous foundations. Foundation systems may include shallow continuous foundations, shallow isolated foundations, or deep foundations (such as drilled piers) with or without connecting grade beams. Included in the scope of assessment and retrofit methods addressing this vulnerability are dwellings with cripple walls between zero-height (wood floor framing sits directly on foundation or foundation stem wall) and Use is limited to dwellings in which the difference in height between tallest and shortest cripple walls is 4-1 or greater (Figure 19). This is meant to dovetail with the cripple wall retrofit provisions, which apply when the difference in wall height is 4-0 or less. The approach to retrofit of dwellings with a hillside vulnerability is conceptually very different than for cripple wall dwellings. The primary retrofit approach builds from an 10

11 approach developed by the City of Los Angeles Hillside Task Group following the 1994 Northridge Earthquake, and included in City of Los Angeles Building Code Division 94. This method recognizes that seismic forces will be attracted to the stiffer load path of the uphill foundation (Figure 20). As a result it is necessary to make the strength of the anchorage to the uphill foundation high enough to resist forces that cannot be reduced due to ductility, and to provide a load path stiff enough that shear anchorage to the uphill foundation is not damaged. The project has arrived at a retrofit approach that includes substantial primary anchors at each end of the uphill foundation, as well as secondary anchors, uniformly distributed between primary anchors. (a) Figure 19. Limits of applicability of hillside dwelling vulnerability assessment and retrofit methods. Figure 21 shows an isometric of the retrofit concept and Figures 22 shows an example detail of a secondary anchor, attaching the floor diaphragm to the uphill foundation. Figure 23 provides a plan view of the dwelling floor, showing the concept of primary and secondary anchor placement at the dwelling uphill foundation. An unexpected finding of the numeric studies is that even with primary and secondary anchors to the uphill foundation, significant performance benefits occur with wood structural panel sheathing on the downhill crawlspace wall. The seismic response of the first occupied story is highly torsional, with the forces concentrating in the occupied story walls immediately on top of the uphill foundation, due to this wall line providing a stiffer load path. Numerical studies show high drifts associated with significant damage in this occupied story wall line when crawlspace wall sheathing is not provided, and drifts notably reduced when crawlspace wall sheathing is provided. For this reason, prescriptive requirements for retrofit will include both anchorage to the uphill foundation and sheathing of crawlspace walls. (b) Figure 20. Hillside dwelling seismic demands (a) loading away from the hill pulls diaphragm away from uphill foundation (b) cross-hill loading pulls corner of diaphragm away from uphill foundation. Figure credit FEMA 547. Figure 21. Hillside dwelling retrofit concept with anchorage to the uphill foundation. Figure credit FEMA

12 methods, with the exception of R1, second from the left. The high probability of collapse of the R1 retrofit is attributed to the retrofit solution with only secondary anchors not being able to effectively resist torsion. Additional analyses are being run prior to selection of the R-factor for retrofit design. It is currently envisioned that the R-factor will be close to one. Figure 24. Interim results from hillside dwelling working group analysis. Figure 22. Hillside dwelling vulnerability retrofit concept - secondary anchor. Anticipated Next Steps The ATC-110 Project will complete its work in June It is anticipated that the developed prestandard and plan sets will move forward into an ANSI Standard process. At this time it is not known whether the prestandard and plan sets will be made publically available while the ANSI standard process is ongoing. Information on intended publication will be made available through ATC. Acknowledgements Figure 23. Hillside dwelling vulnerability retrofit concept - plan of dwelling floor showing placement of primary and secondary anchors at the uphill foundation. Figure 24 provides a snap-shot of interim results from the numerical studies that are being considered in developing final performance criteria and retrofit design criteria. Significant reductions in probability of collapse and probability of exceeding secondary criteria can be seen for all of the retrofit The authors would like to thank project sponsors the California Earthquake Authority and the Federal Emergency Management Agency, the Applied Technology Council project managers, and the project steering committee. The work discussed in this paper has been developed by a number of project team members, who we thank for their contributions. Included are Project Technical Committee Members Vikki Bourcier, Michael Cochran, Dan Dolan, Brian McDonald, John Osteraas and Tom Anderson, and the many members of the project working groups. The work forming the basis of this publication was conducted pursuant to a contract with the California Earthquake Authority and the Federal Emergency Management Agency. 12

13 Work on the ATC-110 Project is ongoing, and the numerical study results and retrofit design concepts presented in this paper are interim, and not final conclusions or recommendations of the project. Users of information contained in this publication assume all liability arising from such use. References ASCE, Minimum Design Load for Buildings and Other Structures (ASCE 7-10) American Society of Civil Engineers, Reston, Virginia. Christovasilis, I.P. and Filiatrault, A Two- Dimensional Seismic Analysis of Multi-Story Light-Frame Wood Buildings, 9th US National & 10th Canadian Conference on Earthquake Engineering: Reaching Beyond Borders, Toronto, Canada, Paper No. 69, 10 p. ICC, 2015b. International Existing Building Code (IEBC), 2015 edition, pp to 2-163, International Code Council, Country Club Hills, Illinois. Pang and Shirazi 2012; Pang, W., and Shirazi, S.M. (2012) A Co-rotational Model for Cyclic Analysis of Light-frame Wood Shear Walls and Diaphragms, ASCE J. of Structural Engineering. van de Lindt, J., Symans, M.D., Pang, W., Shao, X., and Gershfeld, M Seismic Risk Reduction for Soft-story Woodframe Building: The NEES-Soft Project, 121 th World Conference on Timber Engineering, Auckland, New Zealand. Christovasilis, I.P. and Filiatrault, A Numerical Framework for Nonlinear Analysis of Two-Dimensional Light-Frame Wood Structures, Ingegneria Sismica: International Journal of Earthquake Engineering. CUREE, General Guidelines for the Assessment and Repair of Earthquake Damage in Residential Woodframe Buildings, (CUREE EDA-02), Consortium of Universities for Research in Earthquake Engineering, Richmond, California. FEMA, Quantification of Building Seismic Performance Factors (FEMA P695), Federal Emergency Management Agency (FEMA P695, Washington, D.C. FEMA, 2012a. Seismic Performance Assessment of Buildings (FEMA P-58), Federal Emergency Management Agency, Washington, D.C. FEMA, 2012b. Seismic Evaluation and Retrofit of Multi-Unit Wood-Frame Buildings with Weak First Stories (FEMA P- 807), Federal Emergency Management Agency, Washington, D.C. FEMA, Earthquake Strengthening of Cripple Walls in Wood-Frame Dwellings (FEMA P-1024RA2), Federal Emergency Management Agency, Washington, D.C. Folz, B., and Filiatrault, A Cyclic Analysis of Wood Shear Walls, ASCE Journal of Structural Engineering, 127(4), ICC, 2015a. International Building Code (IBC), 2015 edition, pp to 2-163, International Code Council, Country Club Hills, Illinois. 13