Safe Enough? A Building Code to Protect Our Cities and Our Lives

Safe Enough? A Building Code to Protect Our Cities and Our Lives Keith A. Porter,a) M.EERI Seminal works on earthquake engineering hold that greater seismic resistance of building stock is impractical; that the public is unwilling to pay for it; that the public has no proper role in setting code philosophy; and that current seismic provisions encode the proper performance goals. Recent projects undermine these conventionalities. In light of performance expectations for new buildings, the code seems to almost guarantee that a future large but not very rare earthquake will damage enough buildings to displace millions of people and hundreds of thousands of businesses from a major metropolitan area, producing a catastrophe more severe than Hurricane Katrina. A discussion with the public should take place in which we reconsider how to measure risk and how to balance risk and construction cost in code objectives. [DOI: 10.1193/112213EQS286M] INTRODUCTION U.S. building codes aim to protect life safety and limit property damage. The explicit intent of the 2009 International Building Code, for example, is to “establish minimum requirements to safeguard the public health, safety, and general welfare” and includes “safety to life and property” among its goals (ICC 2009). The NEHRP provisions (BSSC 2009) aim “to avoid structural collapse in very rare, extreme ground shaking” and “to provide reasonable control of damage to structural and nonstructural systems that could lead to injury and economic or functionality losses for more moderate and frequent ground shaking.” Note well the inclusion of protecting the general welfare and avoiding property loss along with protecting life safety. Building codes have historically recognized that it is impossible to achieve perfect safety and avoid property loss from overloading by dead and live loads, wind, earthquakes, and so forth. FEMA P-695 (ATC 2009) and the NEHRP Consultants Joint Venture (2012) estimate that the provisions result in at least partial collapse of as much as 10% of code-compliant buildings in risk-targeted maximum considered earthquake motion, (MCER ). Luco et al. (2007) explain how this maximum is generally established so that, considering all possible levels of shaking, achieving an upper bound 10% collapse probability in MCER shaking produces an overall 1% chance of collapse in 50 years due to any level of ground shaking, with exceptions close to faults. Here “MCER shaking” means, for example, 0.2-s, 5%-damped spectral acceleration response equal to ASCE 7’s (2010) SMS , or 1.0-s, 5%-damped spectral acceleration response equal to ASCE 7’s SM1 . For earthquake loads in particular, authors of modern codes have assumed it is impractical or uneconomical to achieve seismic resistance much greater than what is implicit in prior codes. (Here seismic resistance refers generally to the capacity of buildings to resist damage a) Research Professor, University of Colorado Boulder, 428 UCB, Boulder, CO 80309-0428 677 Earthquake Spectra, Volume 32, No. 2, pages 677–695, May 2016; © 2016, Earthquake Engineering Research Institute 678 K. A. PORTER or loss of functionality in earthquakes.) This means that these authors believe that the public would be unwilling to pay for safer buildings—perhaps buildings that would be functional after very strong shaking. ASCE 7 (ASCE 2010) and, by adoption, the International Building Code (IBC) (ICC 2012a) measure acceptability on a per-building basis, which implies that per-building or per-person is the best measure of risk. Let us first review these assumptions more closely and then consider recent projects that call them into question. Concerning whether it was practical to design water tanks and other structures to remain elastic under earthquake loading, Housner (1956) wrote, “It would be quite costly to design for lateral forces of this magnitude, and it would probably be considered desirable to make a less strong structure and accept permanent deformations in the event of a severe earthquake.” Housner and Jennings (1982) wrote, “It is not economical to design every structure to resist the strongest possible earthquake without damage,” and therefore codes “permit yielding and structural damage in the event of very strong shaking.” The authors of ATC-3-06 (ATC 1978) tried but were unable to provide figures on the probable costs to keep buildings functional after a rare earthquake. Still, they codified the assumption that it is economically infeasible to do so and prefaced the document with a philosophy of allowing structural damage in major earthquakes. To say that it is uneconomical to provide greater seismic resistance is roughly equivalent to saying that people would be unwilling to pay for it (“economical” being a subjective judgment, not necessarily measured in terms of, say, benefit-cost ratio). Is that true? In these sources and in others that underlie current code requirements for new buildings, there is no examination of the public’s willingness to pay. Owners and tenants have generally not been asked to express their preferences and are typically absent from the committees that establish codes. (“code” is used here as shorthand for earthquake design provisions for new buildings in ASCE 7 and the IBC.) The authors of SP 577 (Ellingwood et al. 1980) expressed the notion that the A58 standards committee (the precursor to ASCE 7) represented “those substantially concerned with its [the standard’s] scope and provisions.” Though committee members included “a broad-spectrum group of professionals from the research community, building code groups, industry, professional organizations and trade associations,” it did not include representatives of owners or tenants. The authors of a FEMA-sponsored workshop on communicating earthquake risk interviewed a small group of primarily commercial real-estate interests about their preferred measures of seismic performance (ATC 2002). The authors expressed the belief that “this workshop represented one of the first significant attempts to obtain input on issues of acceptable levels of seismic risk used as a basis for design, from other than the technical community.” However, they acknowledged that “several important stakeholder groups, notably residential and institutional building owners, and retailers were not represented at all.” The point is that the public is generally not consulted about their preferences for the seismic performance of the buildings they rely on. It would be easier, perhaps necessary, for engineers to support judgments about what is economical if the public were asked what it would be willing to pay for greater seismic resistance. One could argue that engineers’ clients could opt to build above code, but most new commercial buildings are short-term investments for owners who are subject to competitive pressure and it would be disadvantageous for them to opt for above-code design. One could also SAFE ENOUGH? A BUILDING CODE TO PROTECT OUR CITIES AND OUR LIVES 679 argue that construction industry representatives, having been involved in code development for generations, strongly influence any proposed code change. However, the construction industry is not the public. It is a commercial sector with its own interests that may diverge from those of the public. For example, one might compare the construction industry with the auto industry, which has resisted mandatory seatbelts, airbags, and greater fuel efficiency. One could argue that code objectives have been presented to the public with little or no pushback. If so, the implications of building stock performance in large earthquakes remain unclear to the public. In the opinion of a U.S. Geological Survey (USGS) scientist who deals regularly with local governments, city councils and mayors “absolutely do not know” how a code-compliant building stock designed to meet the life-safety objective will perform in the aggregate, and are unsatisfied when they do learn of it (L. Jones, pers. comm. 2013). Conversely, one could argue that FEMA P-695 was merely a description of how codecompliant buildings perform rather than an expression of its authors’ belief that a 10% collapse probability is acceptable. But in the section Acceptable Probability of Collapse, the authors state, “The fundamental premise of the performance evaluation process is that an acceptably low, yet reasonable, probability of collapse can be established as a criterion for assessing the collapse performance of a proposed system. In this Methodology, it is suggested that the probability of collapse due to Maximum Considered Earthquake (MCE) ground motions be limited to 10%.” Judging by an April 2014 draft of the NEHRP provisions, the next edition of that document may reject the implication that a 10% failure rate is considered acceptable, but if 10% is unacceptable, why does a design that meets that objective satisfy the provisions? What does “acceptable” mean if it is not synonymous with satisfying the provisions’ requirements? GREATER SEISMIC RESISTANCE CAN BE COST-EFFECTIVE Let us consider a few projects that suggest that greater seismic resistance might not be so costly or impractical. As part of the CUREE-Caltech Woodframe Project, the present author and colleagues (Porter et al. 2002, 2006) employed second-generation performance-based earthquake engineering, which was recently standardized in FEMA P-58 (ATC 2012), and found that seismic retrofit of several of the project’s so-called index buildings could be cost-effective in the sense that the retrofit cost would be exceeded by the expected present value of the future reduction in earthquake-related repair costs, across much of California. In some cases and locations, the benefit-cost ratio (BCR) reached 8∶1. The BCRs included only reduced future building repair costs; they would have been higher if they had included casualties, living expenses, and other costs of dislocation. Another project, a cost-benefit study, this one of FEMA-funded seismic risk mitigation (Multihazard Mitigation Council 2005) performed for the U.S. Congress, showed that a broad portfolio of retrofit projects can be cost-effective. The overall portfolio exhibited a BCR of 1.4 comprising a variety of benefits, including the acceptable cost per statistical casualty avoided. These were all retrofits, for which the BCR is generally lower than similar enhancements to new design (the “ounce of prevention” principle). More to the point for new seismic design, Reitherman and Cobeen (2003), who created the CUREE-Caltech Woodframe Project’s index buildings, also presented variants with above-code performance. One building was designed to remain immediately occupiable 680 K. A. PORTER (IO) after design-level shaking and had a marginal cost of 3% over that of the conventional variant ($229,000 versus $221,000 in 2002 USD). See Porter et al. (2002) and Isoda et al. (2001) for details about index-building variants. Another example comes from the use of buckling-restrained braced (BRB) frames. The Broad Center for the Biological Sciences on the campus of the California Institute of Technology in Pasadena, a 120,000-sf, $47 million science building, was the second new U.S. building to include BRB frames in its lateral force– resisting system. According to an Arup engineer involved in the design (Zekioglu, pers. comm. 2014), the braces allow the building to meet something near the immediate occupancy performance level at design-level shaking and added approximately 2% to the construction cost over conventional alternatives. It is estimated that the facades, clad in travertine and stainless steel, added 10% to the construction cost, which suggests that the marginal cost for a remarkable seismic enhancement can be modest compared with acceptable costs for premium finishes. While it may be too expensive to make every structural system achieve IO performance at design-level shaking, it does appear to be practical for some systems. THE PUBLIC IS SOMETIMES WILLING TO PAY FOR SEISMIC RESISTANCE As part of the San Francisco Community Action Plan for Seismic Safety (CAPSS), a public advisory committee was formed comprising self-selected volunteers representing neighborhood groups, landlords, tenants, affordable housing advocates, and others. One of the committee’s functions was to consider the risk to high-occupancy woodframe residential dwellings with soft-story conditions. When CAPSS engineers (including the present author) provided the committee with risk estimates in terms that they had requested—number of red, yellow, and green tags—along with costs to reduce the risk, the committee strongly recommended a mandatory retrofit program (ATC 2010). A red tag is a placard placed on a building by local authorities indicating that the building has been seriously damaged and is unsafe to enter or occupy, making entry unlawful. A yellow tag restricts entry, occupancy, and use of a building, sometimes to parts of the building and sometimes to brief occupancy to remove possessions. A green tag indicates that that lawful occupancy is permitted (ATC 2005). The CAPSS landlords and tenants agreed to share the burden of paying for evaluation and mitigation. The committee recommended the strongest, not the least expensive, retrofit. This recommendation evolved into a mandatory retrofit ordinance enacted into law in 2013 (City and County of San Francisco 2013). The CAPSS Soft-Story program was about remediating the worst existing buildings in a city, not about design of new buildings, but it shows that under some circumstances the public is willing to pay more for a more earthquake-resistant building stock. Those circumstances may relate as much to the process by which the decision is made as to the wealth of the community making the investment. IMPLIED PERFORMANCE OF ENGINEERED BUILDINGS What if we continue to design buildings to a life-safety objective—a 10% maximum collapse probability in MCER shaking—and not to be usable? Let us focus on engineered buildings, as opposed to conventional construction, and on buildings designed to meet a current code—say the 2012 IBC. To decouple the present discussion of the code from SAFE ENOUGH? A BUILDING CODE TO PROTECT OUR CITIES AND OUR LIVES 681 the separate issue of existing buildings, let us consider what happens when a large earthquake—the Big One—occurs after most of the older buildings have been replaced. Let us consider the Big One to be something like the Mw 7.8 ShakeOut scenario or an Mw 7.9 repeat of the 1906 San Francisco earthquake. These are not very rare events, at least compared with MCER shaking. An earthquake like the ShakeOut on the southern San Andreas Fault has a mean recurrence interval on the order of 150 years. It has been 300 years since the last one. What will happen in such an event? Rather than relying on computer models of building vulnerability, let us assume that the outcome is exactly what ASCE 7-10 aims for: an upper bound of 10% collapse rate in MCER shaking. In the ShakeOut, or a repeat of the 1906 earthquake, some areas will be shaken very strongly, with shaking close to the fault reaching or exceeding MCER . Other areas farther from the fault will experience less shaking. Note that the likelihood of a particular fault rupture is not the same as the likelihood of shaking at a particular location in that earthquake. A rupture with a 200-year mean recurrence interval can produce shaking in some places with much rarer occurrence rate, in part because of variability in ground motion. A 200-year earthquake can produce 2,500-year shaking in small areas. According to FEMA P-695, in places with MCER shaking “the probability of collapse due to Maximum Considered Earthquake (MCE) ground motions [is] limited to 10%.” The NEHRP Consultants Joint Venture (2012) estimated that this goal was achieved by current code requirements. Its authors plotted the collapse probability of 179 buildings given MCER shaking as a function of design period; the result is shown in Figure 1. Lead author Figure 1. Collapse probability in MCER shaking in 179 buildings examined in FEMA P-695 and NIST GCR 10-917-8 (NEHRP Consultants Joint Venture 2012). 682 K. A. PORTER C. A. Kircher (pers. comm. 2014) believes that collapse probabilities in buildings with period T ≤ 0.5 s are overestimated. Assuming that one should completely ignore those buildings and omit the three circled data points, the average collapse probability of the remaining data points in Figure 1 is approximately 6%. Let us assume for the moment that this is how the new building stock performs: in areas shaken at MCER , 6% of buildings collapse. Many other buildings would probably be otherwise impaired. What is the broader impairment rate, by which is meant here the fraction of buildings that collapse, are red-tagged, or are yellow-tagged? The authors of FEMA P-695 did not address red or yellow tagging. The present author is unaware of studies on new code-compliant buildings that do so. One could use HAZUS-MH (e.g., Kircher et al. 2006) or another analytical methodology (as in the CAPSS Soft Story Loss Study, Porter 2009). Both efforts were validated by hindcasting losses in past earthquakes. Kircher et al. (2006) hindcasted a number of damaged buildings, casualties, and property repair costs within a factor of 3. In the case of CAPSS, the present author hindcasted tag colors in the Marina District in the 1989 Loma Prieta earthquake within a factor of 1.6. However, both required judgment, and a reader might doubt the tags’ application to new buildings, claiming that the model parameters can be tweaked to achieve any desired results. As another option, one could assume that ratios of yellow tags to red tags and red tags to collapse observed in past earthquakes would apply in a future one. Balanced against this shortcoming (the assumption just mentioned) is the advantage that this approach avoids recourse to models and offers simplicity and transparency. Let us use history to estimate red and yellow tags. Table 1 lists the ratios of red tags to collapses, where both are known (an admittedly limited data set). It suggests approximately 13 red tags per collapse. Table 2 suggests 3.8 yellow tags per red tag. Let r denote the ratio of impaired buildings to collapsed buildings. With 13 red tags per collapse and Table 1. Ratio of red tags to collapses Earthquake 1989 Loma Prieta SF Marina Dist 1989 Loma Prieta Santa Cruz City1 1994 Northridge2 Total Red Collapse Reference 110 100 2,157 2,367 7 40 133 180 NIST (1990), Harris et al. (1990) SEAONC (1990), Fradkin (1998) EQE/OES (1995) Ratio ¼ 13∶1 1 100 is an estimate: 300 countywide factored by number of structures in city versus county and reduced by number of collapses to avoid double-counting 2 133 is taken from the ATC-20 form data in an unpublished database referenced by EQE/OES (1995); red tags reduced by number of collapses to avoid double-counting Table 2. Ratio of yellow tags to red tags Earthquake Yellow Red Reference 1989 Loma Prieta Bay Area 1994 Northridge Total 3,330 9,445 12,775 1,114 2,290 3,404 SEAONC (1990) EQE/OES (1995) Ratio ¼ 3.8∶1 SAFE ENOUGH? A BUILDING CODE TO PROTECT OUR CITIES AND OUR LIVES 683 3.8 yellow tags per red tag, r ¼ 63 impaired buildings for each collapsed building (1 þ 13 þ 13 · 3.8 ¼ 63.4), which includes the collapsed building. The data in Tables 1 and 2 do not reflect a code-compliant stock, and they do not account for the fact that the ratios might actually vary with shaking intensity. However, using this empirical evidence seems preferable to relying entirely on HAZUS-MH or another model. Table 3 summarizes implications for engineered buildings in MCER shaking. The table reflects the code’s explicit objective, reduced to provide a best estimate rather than an upper limit, plus the assumption that ratios of red and yellow tags to collapses in new buildings are the same as those observed in existing buildings in two California earthquakes. The Big One produces MCER shaking over a small fraction of the strongly shaken area, so any given location experiences MCER shaking very rarely, on the order of once in 2,500 years. Weaker shaking is more common and therefore perhaps of greater interest. What happens to buildings shaken at one-half MCER ? Collapse capacity is commonly modeled as lognormally distributed in terms of spectral acceleration response. Let x denote shaking (e.g., in terms of 0.2-s, 5% spectral acceleration response), let xMCE denote MCER shaking in the same terms, and let DDR denote the demand-to-design ratio, x∕xMCE . Let β denote the standard deviation of the natural logarithm of collapse capacity. Luco et al. (2007) considered a range of β values between 0.6 and 1.0 and settled on 0.8. One can now express collapse probability Pc as a function of DDR, as in Equation 1. Equating MCER shaking (DDR ¼ 1) with a 6% collapse probability, one can evaluate the median collapse capacity as a multiple of xMCE by rearranging Equation 1 as in Equation 2 (Figure 2a). One can use Equation 1 with θ ¼ 3.47 and β ¼ 0.8 to evaluate collapse probability at lower levels of DDR and Equation 3 to estimate the impairment rate, denoted Pi .   lnðDDR∕θÞ (1) Pc ðDDRÞ ¼ Φ β EQ-TARGET;temp:intralink-;e1;62;340 EQ-TARGET;temp:intralink-;e2;62;294 θ ¼ expð Φ 1 ðPc ðDDR ¼ 1ÞÞ · βÞ ¼ expð Φ 1 ð0.06Þ · 0.8Þ ¼ 3.47 ð2Þ EQ-TARGET;temp:intralink-;e3;62;230 Pi ðDDRÞ ¼ r · Pc ðDDRÞ. ð3Þ ≤1 Table 3. Performance of new buildings in a small area with MCER shaking Condition Collapse Red and not collapsed Yellow Total Ratio Fraction of stock 6% of stock 13 red tags per collapse 3.8 yellow tags per red tag 6% 78% Most of the rest Virtually all 684 K. A. PORTER Figure 2. (a) Collapse probability as a function of demand-to-design ratio; (b) approximately half of the building stock is impaired at shaking of DDR ¼ 0.5 (i.e., one-half MCER ). With these parameter values, at DDR ¼ 0.5 the collapse probability is approximately 0.8% and the impairment rate is approximately 49%, as shown in Figure 2. One can map DDR and the impairment rate in an earthquake with known shaking. Figure 3a shows a map of shaking in one hypothetical Big One: a Mw 7.8 earthquake on the southern San Andreas fault, produced used as the basis of the Southern California ShakeOut scenario (Jones et al. 2008). Figure 3b shows an extract of ASCE 7-10’s map of SS, the mapped MCER, 5%-damped spectral response acceleration parameter at short periods, which for the most part is equivalent to SMS, the MCER spectral response acceleration parameter for short periods accounting for site amplification. The ratio of the shaking in Figure 3a to that in Figure 3b at any location is therefore an estimate of DDR for that particular scenario earthquake, that particular Big One. Figure 4a shows DDR for the ShakeOut scenario. For reference, in the mapped area approximately 120 km2 has DDR ≥ 1.0; 820 km2 has DDR ≥ 0.75; 4;400 km2 has DDR ≥ 0.5; and 8;000 km2 has DDR ≥ 0.4. For reference, the City of Los Angeles covers 1;215 km2 , so an area roughly 4 times its area experiences at least one-half MCER shaking in this hypothetical earthquake. The foregoing analysis uses the Allen and Wald (2007) V S30 model as reported by OpenSHA’s Site Data Viewer/Plotter (OpenSHA 2016) to estimate NEHRP site classes at 0.02-degree gridpoints across the map, and ASCE 7 (2010, Table 11.4-1) to estimate the site coefficient F a and thus the MCER shaking SMS at each gridpoint. Applying the DDR values shown in Figure 4a to the collapse fragility function shown in Figure 2a produces the map of impairment rate shown in Figure 4b. In the 8;000-km2 area of Figure 4b with a DDR ≥ 0.4, an area seven times that of Los Angeles has an average collapse probability of 1.3% and an average impairment rate of 60%. (The upper bound in Equation 3 forces the area-average ratio of impairment to collapse below r.) If we assume that buildings built to conventional design requirements (e.g., ICC 2012b) perform the same as engineered buildings, then a modern building stock in an area 7 times that of Los Angeles would have 60% of its buildings impaired in a ShakeOut earthquake, an event whose mean recurrence SAFE ENOUGH? A BUILDING CODE TO PROTECT OUR CITIES AND OUR LIVES 685 Figure 3. (a) ShakeOut Sa (0.3-s, 5%) compared with (b) SS from ASCE 7-10. Shaking exceeds 0.5 MCER shaking over roughly 10;000 km2 . interval has been estimated to be 150 years. Much of this area was relatively lightly developed in 2014, but population growth may change that in the next several decades. The population density in the Greater Los Angeles area was approximately 203 per km2 in 2011, so 8;000 km2 is home to approximately 1.6 million Californians (Wikipedia 2016a). 686 K. A. PORTER Figure 4. (a) Demand-to-design ratio DDR in the ShakeOut and (b) fraction impaired. SAFE ENOUGH? A BUILDING CODE TO PROTECT OUR CITIES AND OUR LIVES 687 Nor is ShakeOut an outlier. Aagaard et al. (2010) estimated broadband shaking from 39 hypothetical earthquakes in the San Francisco Bay Area, including an Mw 7.05 earthquake on the Hayward Fault that ruptures the southern and northern segments and nucleates under Oakland. The mean recurrence interval for a Hayward Fault earthquake of this magnitude is approximately 200 years, according to Field et al. (2013). Figure 5 maps DDR and impairment for a modern building stock in this earthquake, which causes shaking of at least 0.4 MCER across 3;300 km2 of the San Francisco Bay Area. This area has a population of approximately 1.4 million people and a population density of 411 per km2 (Wikipedia 2016b). Here DDR ≥ 0.75 in 260 km2 and DDR ≥ 1.0 in 20 km2 .) These conclusions do not depend on the physics-based modeling of ShakeOut or Aagaard et al. (2010). Kircher et al. (2006) examined a repeat of the 1906 San Francisco earthquake using more conventional modeling and estimated shaking in the range of 0.5 times to 1.0 times MCE over an area comparable to that of the ShakeOut (Figure 6). According to UCERF3, the mean recurrence interval for an earthquake of Mw ≥ 7.8 on the San Andreas Fault near San Francisco is on the order of 200 years. These Big Ones are not extremely rare events. One like them is more likely than not to occur by the time most of the building stock complies with current code objectives. Where will people live and work after any of these events? Most buildings with red and yellow tags are eventually returned to functionality, but the process can take Figure 5. (a) Demand-to-design ratio for M 7.0 Hayward Fault scenario and (b) fraction of lowrise buildings impaired (collapsed, red- or yellow-tagged) under this scenario. 688 K. A. PORTER Figure 6. Comparison of (a) Sa (1.0 s, 5%) in a repeat of the 1906 San Francisco earthquake (Kircher et al. 2006) compared with (b) risk-targeted S1 in %g. SAFE ENOUGH? A BUILDING CODE TO PROTECT OUR CITIES AND OUR LIVES 689 months or more. Comerio (2006) writes of experience after the 1994 Northridge earthquake that: [I]nitial inspections… listed 7,000 single-family homes, 2,000 mobile homes, and 49,000 multifamily units red- and yellow-tagged. Three years after the event, when insurance claims were tallied, it became clear that moderate damage to single-family homes was under-counted in the post-event inspections, as more than 195,000 homeowners made insurance claims, for an average of $30,000 to $40,000…. About 40% of homeowners began repairs within one year…. For the remainder, it took two to three years to resolve the insurance claim, and repairs were likely to be delayed until the insurance funding was available. The time needed for repairs of large apartment and condominium buildings was often longer. SEAONC (1990) provides some insight into how long yellow-tagged buildings remained unusable after Loma Prieta: “The majority of areas used yellow tags to qualitatively describe the amount of damage with the expectation that a report by a structural engineer, after a detailed analysis and/or some remedial repairs, would restore the structure to a green-tag status.” Consistent with Comerio’s (2006) observations from Northridge, in most cases either action to address a yellow tag would have taken weeks, months, or more. Note also that Comerio (2006) points out that 195,000 homes had damage costing an average of $30,000 to $40,000 to repair, versus only 58,000 with yellow tags, suggesting that 3.4 buildings suffered costly damage for every yellow tag. This means that the number of impaired buildings as calculated here may comprise only a minority of buildings with significant, costly damage that would challenge owners’ financial resources to repair. In 2012, Los Angeles residential vacancy rates were 2–5%, 11% for commercial, and 5% for industrial, with vacant space impaired the same as occupied buildings. There would be insufficient space to accommodate the displaced population. Even if some yellow-tagged space were usable and many of the temporarily displaced population were able to remain in the metropolitan area, perhaps in shelters or commuting from great distance, something on the order of 25% or more of households and businesses in an 8;000-km2 area would have to move out of the metropolitan area. The SPUR Shelter-in-Place Task Force (2012) cites seven natural disasters since 1989 with significant outmigration (people moving away) that the authors link to disaster-induced loss of housing. The outmigration of 25% or more of the population represents a catastrophic change to a region’s economy and character and a profound shock to the populace. It would seriously affect the state and national economy as well. All of this follows solely from current code objectives, well-vetted models of shaking and collapse fragility, and recent experience with the ratio of red tags to collapses and yellow tags to red tags among the general building stock. Such catastrophes are the logical consequences of the performance allowed in new buildings. One could argue that some classes of existing older buildings are the real problem, if by “the real problem” we mean the near-term threat to life safety and the economy. But today’s new buildings are tomorrow’s existing buildings. As shown here, if the future building stock performs only as well as the code intends and no better, the code will produce catastrophes that society probably does not know it is getting. Perhaps the next Big One will occur before much of the building stock is replaced and older buildings will predominate in the damage. But the next Big One will not be the last Big One. Big Ones will occur after the building stock meets current requirements. The current code will drive the outcomes of the next-plus-one 690 K. A. PORTER Big One, which should be important to code writers and communities that adopt the code. If the consequences estimated here are plausible and intolerable, then the current code is also the real problem. There can be more than one real problem. Society can probably afford greater seismic resistance. The public is sometimes willing to pay increased construction costs, even for seismic retrofit. Remember the examples showing that stronger design requirements can be practical. In light of the implications of the current design philosophy for our cities, perhaps it is time to review what we want from the seismic provisions in our building codes. A SOCIETAL CONVERSATION Code provisions for seismic safety have generally evolved by back-calibrating new requirements to match the safety provided by older ones. In drafting SP 577, Ellingwood et al. (1980) wrote, “The new probability-based load criterion should lead to designs which are essentially the same [level of safety]… as those obtained using current acceptable practice.” The 2009 International Building Code (IBC) aims to be “consistent with the expected performance expressed in the Commentary of the 2003 NEHRP provisions— namely, that ‘if a structure experiences a level of ground motion 1.5 times the design level [i.e., if it experiences the 2,500-year ground motion level], the structure should have a likelihood of collapse… [of] 10%.’” The 2012 IBC employs new risk-targeted ground motion maps that aim to ensure an upper limit of 1% collapse probability in 50 years, considering all levels of shaking that could happen and their various likelihoods. However, the adjustment factor (called the risk coefficient) relative to 2%/50-year shaking is on average 0.9 (for S1 ) and has a standard deviation of 0.06. The new map is similar to the old one and slightly lower on an average geographic basis. The point is that each update involved calibration, not reconsideration. Some studies have questioned the adequacy of seismic design criteria. Ellingwood et al. (1980) were concerned that seismic and wind reliability in SP 577 was “relatively low when compared to that for gravity loads,” and called for “a profession-wide debate” over whether wind and seismic loads ought to have reliability similar to that inherent in gravity loads. Overload from gravity is different from earthquake in the way that automobile accidents are different from nuclear accidents: the former affect a few people at once; the latter, millions. Therefore, lower seismic reliability is even more of a problem when one considers societal impacts. In discussions in 2008 over setting the goal for new design to be 10% collapse probability in 2,500-year shaking, one discussant reported that “there was literally no debate” over whether the goal was reasonable or the right measure. In discussions in BSSC Project 07 (reassessment of seismic design procedures), there “may have been a little discussion” about measuring societal impacts but no formal deliberation of the topic (Luco, pers. comm. 2012). The SPUR Shelter in Place Task Force (2012) called for greater building stock seismic resilience, with a target that 95% of San Francisco’s housing units be in buildings that are strong enough for occupants to shelter in place after a Mw 7.2 earthquake on the San Andreas Fault. Suppose that conversations about seismic performance objectives took place in communities throughout the United States and that some communities expressed a desire for a code that would both protect life safety and prevent urban catastrophes. Such a code could SAFE ENOUGH? A BUILDING CODE TO PROTECT OUR CITIES AND OUR LIVES 691 uniformly increase per-building design requirements to provide post-earthquake operability for most buildings, which would have the indirect consequence of creating a catastropheresistant building stock for urban and rural communities. Recently, the City of Moore, Oklahoma (2014) adopted code revisions to require a basic wind speed of 135 mph, 50% greater than their then-current 90 mph and 17% higher than ASCE 7-10’s 115 mph, equivalent to a wind importance factor of 2.25 and 1.38, respectively. The case of Moore demonstrates that some communities are willing to consider above-code-minimum design, at least when modifications to the design requirements are simple and uniformly and fairly applied. To determine what is equitable and what the public prefers will require a discussion beyond the engineering community, such as the CAPSS Public Advisory Committee but on a larger scale and involving geographically diverse communities. How should the discussion be framed? Bonstrom et al. (2012) examined a variety of successful risk-mitigation efforts to determine how their leaders dealt with public risk perception (what the public thinks about risk), public values (what the public cares about), risk communication (how engineers talk with the public about risk), public involvement in risk-mitigation decisions, and policymaker’s political constraints. The researchers showed that successful efforts to bridge the gap between engineers and the public are commonly structured to express probabilities over a long time period, to employ lessons from historical catastrophes, to present risk in terms of particular scenario outcomes, to educate the public on the specific issue at hand, to incorporate public values into alternatives and solutions, and to increase the importance and credibility of public influence on decision making. They warned about a conflict between long-term optimal policy and short-term political accountability, which might hamper policymaker support for hazard mitigation, suggesting that public education can counteract this effect, influencing political accountability for long-term planning and promoting action by elected officials. Recent news about Los Angeles’ concrete buildings supports this argument (Lin et al. 2013 and Lin 2013). Policymakers and stakeholders widely accept that the public should be involved early and often in decisions involving environmental risks (National Research Council 1996). Such a discussion requires more exhaustive study of the costs and benefits of a resilient building stock, where benefits are measured in terms of public values. Such a study could involve designing a variety of buildings for current code compliance and again for better performance. The difference in construction cost and the seismic fragility or vulnerability of each would be estimated and a hypothetical building stock would be created and analyzed for both futures: one with current code objectives and one assuming a catastrophe-resistant building stock. The consequences could be presented to the public, perhaps through larger or geographically diverse versions of the CAPSS Public Advisory Committee, which could then express its preferences and make recommendations to code-writing authorities and local jurisdictions that adopt or modify model building codes. The public may express its preferences in fluid and imperfect ways. It may often express those preferences without fully grasping the issues. But engineers do not necessarily do a better job determining what is best for the public, nor do the imperfections in public decision making justify the profession in declining to elicit those preferences. As the CAPSS Public Advisory Committee and the examples cited by Bonstrom et al. (2012) show, the public is capable of expressing its preferences sufficiently to direct policy about seismic safety. 692 K. A. PORTER CONCLUSIONS U.S. seismic design philosophy since 1980 reflects a life-safety performance objective with a lower reliability index for earthquakes than for gravity loads, despite the fact that earthquakes, unlike gravity loads, cause large numbers of buildings to experience extreme loading simultaneously. Furthermore, even though buildings are designed for very rare shaking, new buildings can collapse at lower levels of excitation, so if buildings just achieve the code’s performance objective, a not very rare earthquake like the Big Ones discussed here could realistically impair half of a fully modern building stock in an area on the order of 10;000 km2 that is occupied by more than a million people. Local vacancies will not accommodate the displaced population. People will move away, as much of New Orleans’ population did after Hurricane Katrina in 2005. These figures follow from explicitly stated seismic performance objectives; ground motion maps produced and vetted by dozens of leading seismologists; and historically observed ratios among collapses, red tags, and yellow tags. This is not a once-in-2,500-year outcome, but Big Ones whose mean recurrence intervals are on the order of 200 years, and there are many possible Big Ones. The catastrophe discussed here could happen during the career of many readers of this paper. These estimates might be overly pessimistic. Maybe engineered buildings will perform much better than the FEMA P-695 authors calculated. Maybe conventional construction will perform much better than engineered buildings. At issue here, however, are the societal implications of seismic code provisions that aim for life safety rather than operability, not how society might fortunately escape the catastrophe that seems baked into the code. The code and its objectives are not immutable. Civil engineers could revisit the seismic performance objectives they assumed when developing design standards and set them deliberately rather than back-calibrating to prior codes or expecting the ICC to undertake that policy discussion. With the advent of second-generation performance-based earthquake engineering as exemplified by FEMA P-58 (ATC 2012) and using modern earthquake scenarios like ShakeOut, we can estimate earthquake risk in terms of dollars, deaths, and downtime. We can evaluate the costs of producing a catastrophe-resistant building stock rather than assuming that it is uneconomical to do so. We can compare the cost of stronger buildings and the benefits in terms of reduced future losses. The present author does not presume to know what such studies will eventually show, but the point is that they can be performed. As CAPSS showed, the public can express its preferences for balancing risk and cost. The engineering and building professions do not need to make those decisions in isolation. Institutional constraints within ICC and ASCE may prevent adoption of a catastropheresistant building code unless an actual catastrophe occurs and public reaction compels a change. An alternative would be to create catastrophe-resistant design standards and a substantial educational program to inform city councils and the general public about an option to adopt stricter design requirements. Civil engineers could reinterpret their trusteeship of the public’s safety, health, and welfare as requiring an effort to involve the public in deciding what its interests are, how to measure its risk, and what it is willing to pay for a seismically resilient society. That dialogue can be part of a review of what we want our building codes to provide, how to achieve those ends, how cost-effectively to enhance society’s safety, and how to avoid catastrophe. SAFE ENOUGH? A BUILDING CODE TO PROTECT OUR CITIES AND OUR LIVES 693 ACKNOWLEDGMENT This work was funded by the U.S. Geological Survey under Cooperative Agreement G14AC00095. The support is gratefully acknowledged. REFERENCES Aagaard, B. T., Graves, R. W., Rodgers, A., Brocher, T. M., Simpson, R. W., Dreger, D., Petersson, N. A., Larsen, S. C., Ma, S., and Jachens, R. C., 2010. Ground-motion modeling of Hayward Fault scenario earthquakes—Part II: Simulation of long-period and broadband ground motions, Bulletin of the Seismological Society of America, 100(6), 2945–2977. Allen, T. I., and Wald, D. J., 2007. Topographic Slope as a Proxy for Seismic Site-Conditions (V S30 ) and Amplification around the Globe, USGS Open-File Report 2007-1357, Golden, CO. American Society of Civil Engineers (ASCE), 2010. Minimum Design Loads for Buildings and Other Structures, SEI/ASCE 7-10, Reston, VA. Applied Technology Council (ATC), 1978. ATC-3-06, Tentative Provisions for the Development of Seismic Regulations for Buildings, Redwood City, CA. Applied Technology Council (ATC), 2002. ATC-58-1: Proceedings of FEMA-sponsored Workshop on Communicating Earthquake Risk, Redwood City, CA. Applied Technology Council (ATC), 2005. ATC-20: Procedures for Postearthquake Safety Evaluation of Buildings, Redwood City, CA. Applied Technology Council (ATC), 2009. Quantification of Building Seismic Performance Factors, FEMA P-695, ATC-63 project, Redwood City CA. Applied Technology Council (ATC), 2010. Here Today–Here Tomorrow: The Road to Earthquake Resilience in San Francisco. A Community Action Plan for Seismic Safety, Redwood City CA. Applied Technology Council (ATC), 2012. Seismic Performance Assessment of Buildings Volume 1–Methodology FEMA P-58-1, Federal Emergency Management Agency, Washington DC, 2012. Bonstrom, H., Corotis, R., and Porter, K., 2012. Overcoming public and political challenges for natural hazard risk investment decisions, IDRiM Journal 2(1) 1–23. Building Seismic Safety Council (BSSC), 2009. NEHRP Recommended Seismic Provisions for New Buildings and Other Structures, FEMA P-750, Federal Emergency Management Agency, Washington, D.C., 406 pp. City and County of San Francisco, 2013. Ordinance No 66-13: Building Code—Mandatory Seismic Retrofit Program—Woodframe Buildings; Optional Evaluation Form Fee. March 25, 2013, San Francisco. City of Moore, 2014. City Adopts New Building Codes, First in the Nation, available at http:// www.cityofmoore.com/node/2111 (last accessed 25 November 2014). Comerio, M. C., 2006. Estimating downtime in loss modeling, Earthquake Spectra 22(2), 349–365. Ellingwood, B., Galambos, T. V., MacGregor, J. G., and Cornell, C. A., 1980. Development of a Probability-Based Load Criterion for American National Standard A58, SP 577, National Bureau of Standards, Washington, D.C. EQE International (EQE) and Geographic Information Systems Group of the Governor’s Office of Emergency Services (OES), 1995. The Northridge Earthquake of January 17, 1994: Report 694 K. A. PORTER of Data Collection and Analysis, Part A: Damage and Inventory Data, EQE International, Irvine CA. Field, E. H., Biasi, G. P., Bird, P., Dawson, T. E., Felzer, K. R., Jackson, D. D., Johnson, K. M., Jordan, T. H., Madden, C., Michael, A. J., Milner, K. R., Page, M. T., Parsons, T., Powers, P., Shaw, B. E., Thatcher, W. R., Weldon, R. J., II, and Zeng, Y., 2013. Uniform California Earthquake Rupture Forecast, Version 3 (UCERF3)—The Time-Independent Model, USGS Open-File Report 2013–1165/California Geological Survey Special Report 228/SCEC Publication 1792, Menlo Park, CA. Fradkin, P. L., 1998. Magnitude 8: Earthquakes and Life along the San Andreas Fault, New York, NY: Henry Holt. Harris, S. K., Scawthorn, C., and Egan, J. A., 1990. Damage in the Marina District of San Francisco in the October 17, 1989 Loma Prieta earthquake, in 8th Japan Earthquake Eng Symposium, Tokyo Dec 1990. vol. 1, 19–24. Housner, G. W., and Jennings, P. C., 1982. Earthquake Design Criteria, Earthquake Engineering Research Institute, Oakland CA. Housner, G. W., 1956. Limit design of structures to resist earthquakes, in Proceedings of the 1st World Conference on Earthquake Engineering. Berkeley CA, June 1956. 5-1 through 5-13. International Code Council, (ICC), 2009. International Building Code, Country Club Hills IL. International Code Council, (ICC), 2012a. International Building Code, Country Club Hills IL. International Code Council (ICC), 2012b. International Residential Code, Country Club Hills IL. Isoda, H., Folz, B., and Filiatrault, A., 2001. W-12 Seismic Modeling of Index Woodframe Buildings, Consortium of Universities for Research in Earthquake Engineering, Richmond CA, 144 pp. Kircher, C. A., Seligson, H. A., Bouabid, J., and Morrow, G. C., 2006. When the big one strikes again—estimated losses due to a repeat of the 1906 San Francisco earthquake, Earthquake Spectra 22(S2) S297–S339. Lin, R. G., II, 2013. Scientists visit L.A. to discuss list of buildings at risk in quake, Los Angeles Times, 19 November 2013, available at http://articles.latimes.com/2013/nov/19/ local/la-me-ln-earthquake-concrete-20131119. Lin, R. G., II, Xia, R., and Smith, D., 2013. Concrete risks, Los Angeles Times, 13 October 2013, available at http://graphics.latimes.com/me-earthquake-concrete/. Luco, N., Ellingwood, B. R., Hamburger, R. O., Hooper, J. D., Kimball, J. K., and Kircher, C. A., 2007. Risk-targeted versus current seismic design maps for the conterminous United States, in Proceedings of the Structural Engineers Association of California 2007 Convention, Structural Engineers Association of California, Sacramento, 163–175. Multihazard Mitigation Council, 2005. Natural Hazard Mitigation Saves, National Institute of Building Sciences, Washington, DC National Institute of Standards and Technology (NIST), 1990. Performance of Structures during the Loma Prieta Earthquake of October 17, 1989, NIST SP-778, Gaithersburg MD, 212 pp. National Research Council, 1996. Understanding Risk: Informing Decisions in a Democratic Society. National Academy Press, Washington, D.C. NEHRP Consultants Joint Venture, 2012. Tentative Framework for Development of Advanced Seismic Design Criteria for New Buildings, NIST GCR 12-917-20, National Institute of Standards and Technology, Gaithersburg MD, 302 pp. SAFE ENOUGH? A BUILDING CODE TO PROTECT OUR CITIES AND OUR LIVES 695 OpenSHA, 2016. OpenSHA Site Data Viewer/Plotter Web page, available at http://opensha.org/ apps-SiteData (last accessed 5 January 2016). Porter, K. A., Beck, J. L., Seligson, H. A., Scawthorn, C. R., Tobin, L. T., and Boyd, T., 2002. Improving Loss Estimation for Woodframe Buildings, Consortium of Universities for Research in Earthquake Engineering, Richmond, CA, 136 pp. Porter, K. A., Scawthorn, C. R., and Beck, J. L., 2006. Cost-effectiveness of stronger woodframe buildings, Earthquake Spectra 22(1) 239–266. Porter, K. A., 2009. CAPSS Soft-Story Loss Study: Scenario Losses to Large Soft-Story Woodframe Buildings in San Francisco for ATC 52-2, Community Action Plan for Seismic Safety (CAPSS), SPA Risk, Denver CO, 28 pp. Reitherman, R., and Cobeen, K., 2003. Design Documentation of Woodframe Project Index Buildings, Pub. W-29, Consortium of Universities for Research in Earthquake Engineering, Richmond CA. Structural Engineers Association of Northern California (SEAONC), 1990. Posting of Buildings after the Loma Prieta Earthquake. Tagging Subcommittee of the Professional Practices Committee. SPUR Shelter-in-Place Task Force, 2012. Safe Enough to Stay, San Francisco, 44 pp. Wikipedia, 2016a. Web page for Greater Los Angeles Area, available at http://en.wikipedia.org/ wiki/Greater_Los_Angeles_Area (last accessed 5 January 2016). Wikipedia, 2016b. Web page for San Francisco Bay Area, available at http://en.wikipedia.org/ wiki/San_Francisco_Bay_Area (last accessed 5 January 2016). (Received 22 November 2013; accepted 30 December 2014)