INDEX BUILDINGS IN EARTHQUAKE ENGINEERING RESEARCH

Robert Reitherman1

ABSTRACT In the CUREE-Caltech Woodframe Project, conducted 1998-2003, the concept of

the index building was developed to assess the costs and benefits of various seismic enhancements to new designs or retrofits of existing buildings. In that project, four basic woodframe residential index buildings were developed, which with variants resulted in nineteen different designs. The concept, however, is applicable to any type of construction.

An index building is a very specific although hypothetical building, completely

described with drawings such as floor plans and sections; structural notes; weights of materials; nonstructural features; and construction cost breakdown. The design is translated into a structural model suitable for analysis, so that engineering parameters such as drift and acceleration at various locations within the building are calculated, and then these engineering outputs are translated into repair costs. One can keep the building model constant and vary the ground motions or vice versa to develop a statistical basis for dealing with uncertainties. Besides providing the geometry of the structural model, the computer-aided drafting drawings are used to build 3D visualizations (rendered drawings and animations) of the building, which are very useful in communicating the results.

The term “index” building is derived from the concept of the Consumer Price

Index, which is an extremely specific basket of consumer goods used by the US Bureau of Labor Statistics to track inflation. By very specifically defining the consumer items, two researchers can discuss inflation with reference to a common baseline, just as defining an index building provides a common framework for evaluating earthquake performance.

Introduction This paper summarizes the concept of the index building, gives an example of its use in the CUREE-Caltech Woodframe Project, places the development of the index building into the historical context of predecessors, and concludes with recommendations as to several uses of the index building in earthquake engineering research. The CUREE-Caltech Woodframe Project, the full name of which was “Earthquake

1Executive Director, Consortium of Universities for Research in Earthquake Engineering, 1301 S. 46th Street, Richmond, CA 94804, USA, reitherman@curee.org

Hazard Mitigation of Woodframe Construction,” was motivated by the scale of the losses to this type of “two by four” construction in the January 17, 1994 Northridge Earthquake. The References section here (see CUREE, 2004) lists the reports in that project’s publication series. A number of the references are related to the following discussion, such as experimental work that developed data used to model the index buildings. The W-29 report (Reitherman and Cobeen, 2003) provides under one cover an overview of index buildings, explains how they were used in other research reports in the report series, and contains a compact disc (CD) of the drafting, drawing, animation, and cost estimation files. While that project was devoted to one construction type, its development of the index building is an accomplishment that researchers can apply to any type of construction. It is discussed here in the more general context. Economists often rely on the ceteris paribus approach, changing one variable and holding all else constant, to sort out cause and effect. One can only state that a decrease in demand will generate a decrease in price in a market economy if all other things are kept fixed in the analysis. Similarly, one can only say that a decrease in the size of an anchor bolt washer or the decrease in strength or stiffness of a wall will decrease seismic performance if we keep the other variables constant—if we keep the overall geometry of the building the same, keep the state of deterioration and material properties the same, and so on. This is elementary logic, but too often engineering discussions tend to become murky as one engineer offers the opinion that a particular kind of damage usually occurs in a general type of construction, a type defined verbally such as “single family houses of the 1970s” or “concrete frames built in the 1960s,” only to be contradicted by another who has seen different experience. It can be unclear whether different engineers are discussing the same set of facts when using these general verbal descriptions. As discussed later in comparison to more general verbal descriptions of types of construction, an index building must be a very specifically designed building, though one that will usually only be analyzed, not built. Rather than assuming what variables are important and not bothering to include others in the design, a complete reference point is provided. This allows other researchers to come later, using different theories or with different strategies to test, and have a complete building with which to work. This implies more effort to have design professionals on the research team, producing drawings and notes in a realistic manner almost as if for a real building project. Consideration of software used to hand off the design at each step must also be given. It is cost-beneficial to have the computer-aided drafting files (e.g., AutoCad) be importable into structural models and also into graphic programs for illustration purposes. As variables are explored via changes in the building design, the cost estimation files must easily track those changes. Lead designers of the index buildings in the Woodframe Project were Kelly Cobeen, SE; Ray Young & Associates, cost estimators; Goetz Schierle, architect; John Coil, SE. The overall concept of the index building and its relation to the research tasks in the Woodframe Project was developed by the author; Tom Tobin coined the “index” term. For each building, a broader committee provided input on what details should be included to make the building representative of a segment of the inventory that was built in a given era. This requires more than simply

retrospectively producing a design to meet the requirements of a then-current building code. For example, based on input from knowledgeable engineers and code officials, the Woodframe Project developed an index building nominally governed by a 1960s edition of the Uniform Building Code. A literal reading of that UBC edition would lead one to infer that the entire seismic load path was considered. In practice, for many buildings such as apartment houses, the shear force calculation in a wall was not carried through into a calculation of overturning, and hold-downs were not specified. Thus, the index building intended to realistically represent that kind of construction did not include hold-downs even though the design lateral forces implied their necessity to today’s engineer.

An Example of An Index Building Shown in Figures 1 and 2 are examples of the “blueprint” (Figure 1) and rendering (Figure 2) descriptions of an index building, in this case a three-story apartment building developed in the Woodframe Project. As explained above, there are numerous drawings, spreadsheets for the construction cost breakdown, and other details going along with the examples in the figures, but these illustrations convey the idea that the building must be realistic and specifically designed. Note that because nonstructural damage often causes a large amount of loss, the modeling must include those pertinent features. The product is not an “index structure” but an “index building.”

Figure 1. Elevation views of the three-story apartment index building.

An index building is related in name and concept to the Consumer Price Index, which is calculated by the Bureau of Labor Statistics of the US Department of Labor. To track price changes, a consistent basket of goods must be compared. (The consumer items very literally form a shopping list, a list that is priced out in detail by government surveyors who actually walk the aisles of retail stores.) One also needs standardized ways of adjusting for region, recurring seasonal changes, whether the consumer drives a car or not, and other variables. Different researchers (e.g. university economists), government programs (e.g., social security) or financial industries (e.g. stock market analysts) can then adapt these benchmark costs for their own purposes, applying different assumptions concerning future inflation or isolating particular aspects such as food or energy costs. Similarly, an index building is a specific basket of goods in the construction sense, providing a similar type of benchmark.

Figure 2. Illustration of selected features of the three-story apartment index building. The structural engineering analysis of the index buildings in the Woodframe Project is described in reports W-8 (Folz and Filiatrault, 2001), W-12 (Isoda, Folz, and Filiatrault, 2002), and W-21 (Folz and Filiatrault, 2002). Again, one virtue of the index building approach is that one can use the same given building and yet compare different analysis methods, later refine a particular analysis method, or update input data and re-run calculations using the same analysis method. As an example from the Woodframe Project, one level of quality of construction for the lateral capacity of stucco was set at a particular percentage of the strength obtained for precisely

made lab samples, with the percentage reduction accounting for weathering deterioration and less-than-lab-quality workmanship in actual buildings. A researcher can select a higher or lower percentage, tied to the same laboratory test benchmark, and then perform a comparative analysis. Similarly, values obtained for tests of samples of sheathing where the nails were driven precisely on centerlines of the framing can be used to quantify high quality, as compared to values based on the tests where nails were driven a specific amount off-center. Numerous tests of this type were conducted in the project to provide this ability to calibrate. (Fonseca et al., 2002). To put it simply, engineers calculate with numbers, not with adjectives. To advance the ability of theory to predict performance, quantification of a variable such as “high quality” or “low quality” is necessary. If a physical model is constructed and tested, a comparison of the analytical index building results and the experimental results can be made. Besides the work in the Woodframe Project, a current project (van de Lindt et al., 2005) in the NSF NEES program is providing testing results for a physical model of a woodframe building that in full-scale matches an index building. The work in the Woodframe Project that put the index building to use as a tool for loss estimation was by Porter et al. (2002). Note that the loss estimation process requires a complete description of the building, including cost estimation files of the building in its pre-earthquake state as well as algorithms for converting damage states into losses. The Porter et al. (2002) study was designed to mesh smoothly with the structural engineering and cost estimating calculations, and that coordination is essential regardless of the precise methods used.

Credit Where Credit is Due: Index Building Predecessors The index building, because of its specificity and its completeness in describing an archetypical building, is different than predecessor modeling of “typical,” “model,” or “representative” buildings, although it is part of the same historic progression. The insurance industry developed descriptions of building types (“construction classes”) for earthquake performance prediction purposes early in the twentieth century. Freeman (1932) provides a thorough summary of ways of categorizing buildings, breaking down the typical inventory of buildings in an American city into eleven construction classes. A more recent treatment was provided by Steinbrugge (1982), with the construction class list remaining similar in number. Far from being limited to insurance usage, construction classes were used in the first large-scale earthquake loss estimates (Algermissen et al., 1972). In the ATC-13 loss estimation study (ATC, 1985) a dozen basic building types were verbally defined, which, with height ranges, added up to 40 model buildings. A convenient comparative discussion of loss estimation building categorization schemes is provided in the National Research Council Panel on Earthquake Loss Estimation report (1989). The HAZUS earthquake loss estimation method (first version, NIBS, 1999) uses a similar typology based on about a dozen basic building types, not counting height ranges. It extends the specificity of the usual model building type description to allow engineering calculations, for example by assigning exact heights to each model building and defining some of its structural parameters. The HAZUS method provides enough specificity for each building type that structural calculations can at least approximately

mimic those that an engineer might employ for an actual building. A seismic study of 2,007 unreinforced masonry buildings in San Francisco took advantage of floor plan, occupancy, and other information to divide that inventory into subtypes for the purpose of estimating costs of various retrofits. (Holmes et al., 1990) Often general statements are made about “unreinforced masonry buildings,” but it is useful to divide up such an inventory into subtypes to make specific analyses. For example, strengthening of diaphragms in residential occupancies requires dislocation of the renters, a very significant impact as compared to a similar retrofit in a warehouse. Buildings with taller story heights pose greater out-of-plane strengthening challenges. Illustrations of the sub-types were prepared, though corresponding whole-building models of a structural nature were not developed. See Figure 3.

Figure 3. Sub-types of San Francisco unreinforced masonry buildings (Holmes et al., 1990)

These previous applications of what can be generically called the model building approach did not define a building type so specifically that two different researchers could independently apply their own analyses to the given model building to predict and compare losses. For example, one researcher might sketch what they assume to be the detailing of a joint or the dimension of a story height, with resulting calculation implications on response or earthquake resistance, while another might make different assumptions. The calculations of the two analysts would differ, yet they would call their model buildings the same name, which introduces unwanted ambiguity.

Perhaps the predecessor model building approach that came closest to the complete index

building concept was the work of Gauchat and Schodek (1984). As shown in Figure 4, they very specifically defined several archetypes of housing, then those specific descriptions became the common point of reference for “analysis” by engineers. The analysis was judgmental, estimating global damage levels for various Modified Mercalli Intensities. With the index buildings in the Woodframe Project, which included construction cost models as well as quantitative structural models, the analysis was via the inelastic time history method using multiple ground motion

records, and specific dollar losses were produced. Nonetheless, Gauchat and Schodek showed the value in holding one element of the engineering conversation constant and demonstrated the importance of clear and specific graphic depictions of what one means by a “typical” building.

Figure 4. Illustration of a three-story townhouse model building (Gauchat and Schodek, 1984).

Conclusions: Recommended Applications One disadvantage of the index building approach is the added degree of difficulty in elevating research to the index building level from the model building level. Especially with non-residential construction, realistically including a full array of nonstructural components is difficult. Because damage to ceilings, partitions, piping, HVAC equipment, and other nonstructural components is a major contributor to loss, excluding them from a model greatly reduces the utility of the index building approach. A general proviso is that an index building is a generalization, an archetype that is intended to represent some significant population of actual buildings—from dozens of buildings of one type and vintage in one application on up to millions of more generically known buildings in a region in another study. The index provides a benchmark to allow extrapolation from the particular to the general, but that process, like any form of generalizing, must be wisely carried out for the results to be meaningful. Ideally, numerous index buildings, with variants to account for different design criteria or quality factors, would be developed within any particular construction class or model building type. While experimental verification of analytical modeling is facilitated by use of an index building, the whole-building scale of the analysis cannot always be feasibly matched in tests, though current large reaction wall apparatus and shake tables offer new capabilities, as do in-situ experiments.

One of the best experiments is actual earthquake shaking of a building, so long as the data on the ground motion, building response including soil-structure interaction effects where significant, and construction details are available. (Unfortunately, those three data sets intersect for only a small number of buildings). If an index building closely approximates an actual building that undergoes an earthquake, and if the analytical results match well with the observed performance, this provides a valuable means of validation. The index building approach is specifically recommended for the following applications.

• Providing a coordinating mechanism for collaborative research. Researchers need not be co-located, and a division of labor is facilitated, yet their work is tied to a common basis.

• Keeping research from becoming obsolete. Large portions of an index building research product can be re-used in the future, even if components are later thought to be in need of revision. For example, if different earthquake records become available, or more test information or actual earthquake performance data is obtained, the analyses can be re-run using the same building design.

• Comparing analytical with experimental results or with earthquake performance observations.

• Testing and benchmarking changes in building codes over time. It would be very instructive to see how various editions of a code evolve over the years, as applied to the same basic building, both in terms of initial construction cost and earthquake performance. The Consumer Price Index aspect of the index building concept allows this tracking over time.

• Illustrating research results to clearly communicate to the public as well as design professionals how the numbers produced by earthquake engineering calculations relate to practical examples.

• Applying the index building concept to subjects other than earthquake engineering, such as in studies of energy conservation, vulnerability to wind or other non-seismic hazards, or long-term life cycle costs.

Acknowledgments

The Woodframe Project was funded by the Federal Emergency Management Agency through a grant administered by the California Office of Emergency Services. The project manager was Professor John Hall, California Institute of Technology.

References Algermissen, S.T., et al., 1972. A Study of Earthquake Losses in the San Francisco Bay Area.

Washington, DC: National Oceanic and Atmospheric Administration. CUREE, (2004). The following Woodframe Project reports are available from CUREE, Consortium of

Universities for Research in Earthquake Engineering, Richmond, CA, USA; http://www.curee.org. Free graphics, video clips, and other project information are also available under “Woodframe Project” at that website.

W-01: Seible, F., A. Filiatrault, and C.-M. Uang (1999), Proceedings of the Invitational Workshop on Seismic Testing, Analysis and Design of Woodframe Construction.

W-02: Krawinkler, H., F. Parisi, L. Ibarra, A. Ayoub, and R. Medina (2001), Development of a Testing Protocol for Woodframe Structures.

W-03: Filiatrault, A., editor, (2001), Woodframe Project Testing and Analysis Literature Reviews.

W-04: Schierle, G. G., editor, (2001), Woodframe Project: Case Studies. W-05: Fischer, D. and A. Filiatrault (2001), Two-Story Single Family House Shake Table

Test Data. W-06: Fischer, D., A. Filiatrault, B. Folz, C.-M. Uang, and F. Seible (2001), Shake Table

Tests of a Two-Story Woodframe House. W-07: Andrews, J. (2000), Video Update: Fall/Winter 1999/Summer 2000, Hazard

Mitigation of Woodframe Construction. W-08: Folz, B. and A. Filiatrault (2001), CASHEW: Version 1.1, A Computer Program for

Cyclic Analysis of Wood Shear Walls. W-09: Schierle, G. G. (2003), Northridge Earthquake Field Investigations: Statistical

Analysis of Woodframe Damage. W-10: Rosowsky, D. and J. H. Kim (2002), Reliability Studies. W-11: Camelo, V., J. Beck, and J. Hall (2002), Dynamic Characteristics of Woodframe

Structures. W-12: Isoda, H., B. Folz, and A. Filiatrault (2002), Seismic Modeling of Index Woodframe

Buildings. W-13: Gatto, K. and C.-M. Uang (2002), Cyclic Response of Woodframe Shearwalls:

Loading Protocol and Rate of Loading Effects. W-14: Mahaney, J. A., and B. E. Kehoe (2002), Anchorage of Woodframe Buildings:

Laboratory Testing Report. W-15: McMullin, K. M., and D. Merrick (2002), Seismic Performance of Gypsum Walls:

Experimental Test Program. W-16: Fonseca, F. S., S. K. Rose, and S. H. Campbell (2002), Nail, Wood Screw, and

Staple Fastener Connections. W-17: Chai, Y. H., T. C. Hutchinson, and S. M. Vukazich (2002), Seismic Behavior of

Level and Stepped Cripple Walls. W-18: Porter, K. A., J. L. Beck, H. A. Seligson, C. R. Scawthorn, L. T. Tobin, R. Young,

and T. Boyd (2002), Improving Loss Estimation for Woodframe Buildings. W-19: K. Mosalam, C. Machado, K.-U. Gliniorz, C. Naito, E. Kunkel, and S. A. Mahin

(2002), Seismic Evaluation of an Asymmetric Three-Story Woodframe Building. W-20: Symans, M. D., W. F. Cofer, Y. Du, and K. J. Fridley (2002), Evaluation of Fluid

Dampers for Seismic Energy Dissipation of Woodframe Structures. W-21: Folz, B. and A. Filiatrault (2002), A Computer Program for Seismic Analysis of

Woodframe Structures (SAWS). W-22: Ryan, T. J., K. J. Fridley, D. G. Pollock, and R. Y. Itani (2003), Inter-story Shear

Transfer in Woodframe Buildings. W-23: Deierlein, G. G. and A. M. Kanvinde (2003), Seismic Performance of Gypsum

Walls – Analytical Investigation. W-24: Xiao, Y. and L. Xie (2003), Seismic Behavior of Base-Level Diaphragm Anchorage

of Hillside Woodframe Buildings. W-25: Pardoen G. C., A. Waltman, R. P. Kazanjy, E. Freund, and C. H. Hamilton (2003),

Testing and Analysis of One-Story and Two-Story Shear Walls Under Cyclic Loading.

W-26: Krawinkler, H., F. Zareian, L. Ibarra, R. Medina, and S.-J. Lee (2003), Seismic Demands for Single- and Multi-Story Wood Buildings.

W-27: Dolan, J. D., D. M. Carradine, J. W. Bott, and W. S. Easterling (2003), Design Methodology of Diaphragms.

W-28: Ficcadenti, A. J., E. J. Freund, G. C. Pardoen, and R. P. Kazanjy (2004), Cyclic Response of Shear Transfer Connections Between Shearwalls and Diaphragms in Woodframe Construction.

W-29: Reitherman, R. and K. Cobeen (2003), Design Documentation of Woodframe Project Index Buildings.

W-30: Cobeen, K., J. Russell, and J. D. Dolan (2004), Recommendations for Earthquake Resistance in the Design and Construction of Woodframe Buildings.

Freeman, John R., 1932. Earthquake Damage and Earthquake Insurance. New York: McGraw-Hill. Gauchat, U. P, and D. L. Schodek, 1984. Patterns of Housing Type and Density: A Basis for Analyzing

Earthquake Resistance. Department of Architecture, Harvard University. Holmes, William, et al., 1990. Seismic Retrofitting Alternatives for Unreinforced Masonry Buildings.

Rutherford & Chekene, a study prepared for the San Francisco Planning Department. National Research Council, 1989. Estimating Losses From Future Earthquakes. Washington, DC:

National Academy Press. NIBS, National Institute of Building Sciences, 1999. HAZUS 99. Technical manual and associated

software. Washington, DC: NIBS. Steinbrugge, Karl, 1982. Earthquakes, Tsunamis, and Volcanoes. New York: Skandia. van de Lindt, John, et al., 2005. “Development of a Performance-Based Seismic Design Philosophy for

Mid-Rise Woodframe Construction.” NSF grant 0529903.