Benefit–Cost Analysis for Earthquake-Resilient Building Design and Retrofit: State of the Art and Future Research Needs

Discount rate ($r$) is the rate of return used to discount future cash flows back to the present value. A typical discount rate is between 2% and 10% (Gibson et al. 2014). Planning horizon ($T$) is the future time period in years over which benefits and costs are counted. Planning horizon is typically between 50 and 75 years for new buildings (NIBS 2019) and 30 years for existing buildings after seismic retrofit (FEMA 2009).

The process of economic discounting for future damage (e.g., discount rate) tends to prioritize the well-being of individuals living today over those who will exist in the future (Lind 2007). From an equity perspective, all individuals should be treated as equals, regardless of whether they are currently alive or yet to be born (Lenton et al. 2023). Therefore, when the strategy being evaluated has long-term impacts on life and health, such as climate change mitigation, it is recommended to use nonconstant discount rates for analysis periods that extend beyond the planning horizon (Lind 2007), utilize distinct discount rates to adjust long-term benefits and investment costs (Welsh-Huggins and Liel 2018), or refrain from translating benefits into monetary terms (Lenton et al. 2023). Because earthquake risk reduction measures are effective within the relatively short planning horizon of buildings, applying the same discount rate to life-saving benefits and investment costs is preferable as indicated by many studies (Pate-Cornell 1984; Liel and Deierlein 2013; NIBS 2019).

This step requires first identifying assets that are sensitive to earthquakes and then estimating the relationship between the severity of expected losses and the level of ground shaking hazard. Benefits are estimated from the avoided losses under action $i$ relative to the status quowhere $t$ = time starting from the year that a mitigation action is taken; and $\mathit{EA}{\mathit{L}}_j$ = expected annual losses under action j, for $j=0,\mathrm{.....},\mathit{\text{\hspace{0.17em}I,\hspace{0.17em}}}$ where I is the set of actions, and is calculated as follows:where $p(l)$ = annual rate of exceedance for the loss $l$, given as follows (Krawinkler et al. 2006):where $p(x|y)$ = exceedance probability of $x$ given $y$ (e.g., survival function, the complementary cumulative distribution function); $dm$ = damage measure (e.g., damage state); $edp$ = engineering demand parameters (e.g., maximum drift); $im$ = intensity measure (e.g., peak ground acceleration); and $p(im)$ = expected rate of return of the ground shaking hazard (e.g., hazard curve).

Direct benefits include avoided damage to buildings and contents, and avoided deaths and injuries. Indirect benefits may be economic or related to community resilience, social equity, and environmental sustainability, including avoided displacement and debris removal, loss of business or rental income, loss of life quality, loss of productivity, loss of customers, supply chain delays, reduction in employment, tax base, and affordable housing, among others (Fung et al. 2022b).

The cost for alternative design is estimated as the difference in initial construction cost or life-cycle cost relative to the baseline. Initial construction cost may include material, labor, equipment costs, and contractor overhead and profits. Life-cycle cost is the total cost associated with building design and construction, building operation and maintenance, and building disposal at the end of the life cycle. The cost for seismic retrofit is a combination of structural and nonstructural improvement costs and may also include changes in maintenance costs (FEMA 2009; Fung et al. 2022b).

Given estimates from Steps 1 and 2, one can distribute benefits and costs across stakeholders to obtain tiers of impacts (NIBS 2019; Fung et al. 2022a). Benefits and costs are compared using two metrics: benefit–cost ratio (BCR) and net present value (NPV)where a ${\mathbf{BCR}}_{\mathit{i}}>1$ or ${\mathbf{NPV}}_{\mathit{i}}>0$ implies that the benefit of the action outweighs the cost.

Sensitivity analysis can be applied to examine whether the BCR shifts dramatically when inputs vary due to uncertainties in a building’s useful life, inflation rate, benefit and cost assumptions, hazard level, and model simulations. It is often helpful to determine the sensitivity range for each input and identify the inputs most important to BCR estimation (Gibson et al. 2014; Fung et al. 2022a).

For new buildings, studies are regularly conducted to analyze the economic impacts of code changes. The economic impacts include reduced probabilities of property loss, death, and injury, population displacement, and business interruption in future earthquakes. Table 1 summarizes the literature on BCA for adopting new code requirements. At the national level, FEMA evaluated annual avoided losses for post-2000 buildings conforming to 2000 I-Codes (FEMA 2020a). The seismic requirements of 2000 I-Codes are equivalent to that of 1997 Uniform Building Code. The study combined damage functions from Hazus [a free geographic information system–based risk assessment tool developed by FEMA (2012)], parcel and building footprint data from multiple sources, and input from experts in building performance and building code history to develop detailed spatial loss estimates. The results show that annual avoided losses are significant for US states with high to moderate seismicity [Fig. 2(a)]. In highly seismic regions (Alaska, California, Hawaii, Oregon, Utah, and Washington states), an average 8% reduction in annual losses can be expected. The avoided losses are more pronounced in regions with higher seismic hazard (California), lower seismic design requirements (Hawaii), or both (Utah), as illustrated in Fig. 2(b). Similarly, the Multi-Hazard Mitigation Council (NIBS 2019) found that adopting the 2018 I-Codes for new construction in the 48 contiguous United States can result in a BCR of 12 compared to 1990s seismic codes. Specifically, implementing the 2018 I-Code requirements for earthquake can prevent annual property losses of $1,500 per building in 2018 US dollars, reduce annual deaths, injuries, and trauma-related losses by $800 per building, and lower annual business interruption losses by $2,000 per building (NIBS 2019). Other national-level studies focus on evaluating compliance costs. The goal of such studies is to demonstrate that the marginal cost of complying with the newer code is not very large relative to an earlier code (e.g., NAHB 2018).

Such studies naturally raise the question of the need to adopt new seismic standards in regions of moderate seismic risk. For instance, the middle Mississippi River Valley region experienced very large earthquakes in the past but no damaging earthquakes in recent decades (NEHRP 2013; Nordenson and Bell 2000). The National Earthquake Hazards Reduction Program (NEHRP 2013) assessed the benefits and costs of adopting the 2012 International Building Code (IBC) in Memphis, Tennessee, relative to the 1999 Standard Building Code (SBC). A major conclusion of their study is that the compliance costs are low, but the benefits associated with the improved design are significant. However, Stein et al. (2003) estimated that the total compliance cost for the 2000 IBC ($\$200\mathrm{million}/\mathrm{year}$ in 2001 US dollars) is an order of magnitude greater than the total benefit ($\$17\mathrm{million}/\mathrm{year}$) and argued that buildings in Memphis should not be designed to the same level as in California because of lower seismic hazard. Ryu et al. (2010) also showed that designing new commercial buildings in Memphis to the 2003 IBC, 2006 IBC, or 2009 NEHRP provisions has little effect on expected annual losses (EALs) relative to the 1999 SBC. The controversy between NEHRP (2013) and the two studies is due to different versions of seismic hazard maps used, building types analyzed, and benefit elements considered. Similar debates exist in Charleston, South Carolina, and Boston, where recent seismic activity is minor but magnitude 7 or larger earthquakes struck the regions in the past (Nordenson and Bell 2000; Nikellis et al. 2019; Joyner and Sasani 2018).

A potential gap in such studies is the absence of co-benefits, which accrue even in the absence of a hazard event during the planning period (Fung and Helgeson 2017). A few studies have evaluated the co-benefit of seismic codes on wind mitigation. Nikellis et al. (2019) analyzed steel moment frame (SMF) office buildings in two US cities and found that ignoring wind-induced losses in Los Angeles can lead to a 32%–62% underestimation of EAL for 40-story buildings. Ignoring earthquake-induced losses in Charleston, South Carolina, can result in a 33% and 29% underestimation of EAL for 30-story and 40-story buildings, respectively. However, Joyner and Sasani (2018) indicated that the co-benefit is negligible in earthquake or wind-controlled regions. Wind damage accounts for 5% of the total EAL for a 7-story concrete building in San Francisco (earthquake-controlled). Earthquake damage accounts for 1% of the total EAL for a 7-story concrete building in Boston (wind-controlled). The controversy between the two studies is mainly due to different building heights and structural types analyzed, and further research is needed.

Unlike many studies that assess benefits based on predicted building performance, a few studies have used historical insurance data to evaluate avoided losses due to the implementation of a building code (e.g., Simmons et al. 2020, 2018). This approach compares paid insured losses before and after code implementation, facilitating regional-level impact assessment. A caveat is that buildings built after the enactment of the code are assumed to comply with the code, whereas in practice, buildings may be built to lower or higher standards, depending on code enforcement, quality control, and owner requirements for safety and resilience. Moreover, this method is more suitable for frequent natural hazard events such as hurricanes because it requires comparable events in intensity or magnitude before and after code implementation.

Several studies have assessed the benefits and costs of above-code seismic design (Table 1) and found it to be a cost-effective option (NIBS 2019; Kutanis and Boru 2014). Specifically, the Multi-Hazard Mitigation Council (NIBS 2019) estimated that buildings above the 2015 IBC requirements could result in a national average BCR of 4, relative to 1990s seismic codes, meaning that $4 can be saved for every $1 spent to make new buildings stronger and stiffer. To achieve greater building strength than required by the 2015 IBC, the Multi-Hazard Mitigation Council assigned the buildings a higher importance factor than specified by the 2015 IBC. The vulnerability functions used in Hazus were also modified to reflect the increased strength and stiffness of the buildings. Likewise, Kutanis and Boru (2014) recommended adjusting the performance target of residential buildings to immediate occupancy in Turkey, where 71% of the land is located in high seismicity zones. Kutanis and Boru (2014) designed six benchmark residential buildings of three heights and two structural systems (infilled frame and dual system), and estimated that construction costs could increase by 4.2%–11.2% for 3-story buildings, 21.2%–28.8% for 6-story buildings, and 20.7%–27.4% for 10-story buildings built to the immediate occupancy level relative to the life safety level. The expected annual cost increase for new construction is comparable to the historical annual loss from earthquakes, meaning that the BCR is greater than 1, assuming no loss in the immediate occupancy scenario.

There is an extensive literature on evaluating the economic value of seismic retrofits for existing buildings. A major focus is on bringing existing residential and commercial buildings up to the life safety standard. Another focus is on improving the performance of hospitals and schools to the immediate occupancy level in the event of a major earthquake. Given that much of the variation is across structural systems and risk categories, we present our review by building type. Table 2 summarizes the methods employed in the literature for evaluating retrofit strategies applicable to hospitals, schools, and residential and commercial buildings.

Building-level analysis often requires information on site conditions, structural and nonstructural components, contents value, and number of occupants. Because detailed hazard, building, and occupancy information is supplied to the analysis, detailed estimates can be obtained for building damage, collapse probability, and casualties. Methods for quantifying these direct impacts are well developed and validated by many studies. In contrast, indirect economic impacts are less readily assessed in the literature due to lack of data or difficulty in quantifying the losses as well as gaps in estimating recovery time (Fung et al. 2022b). This can result in underestimated economic benefits of risk reduction measures, and thus discourage building owners from adopting such measures (Goettel 2016; Gibson et al. 2014).

Building-level analysis often begins with the development of vulnerability models that consider various levels of earthquake intensity. The models are developed based on empirical data, experimental tests, statistical analysis, or expert judgment. However, expert opinion and methods that are difficult to verify should be used with caution (Porter et al. 2006). Validating models with post-disaster building damage records can improve the accuracy of damage predictions (Cook et al. 2021; Baker et al. 2016). Earlier studies determined damage state probabilities based on building response characteristics (e.g., peak acceleration, peak drift ratio), while more recent studies have used the Performance-Based Earthquake Engineering (PBEE) approach to analyze damage states of single components and aggregate component-level results to the whole building. The PBEE approach enhances the analyst’s ability to predict damage state probabilities and probabilistic repair time and cost for individual buildings.

First, the data required to quantify various indirect benefits are scarce, which hinders the practice of incorporating indirect benefits into BCA (Goettel 2016; Gibson et al. 2014; Fung et al. 2022b). However, reducing indirect losses such as business interruption and displacement is part of the objective of resilient design. When possible, the co-benefits of risk reduction strategies should be accounted for; that is, the benefits accrued when there is no damage event during the analysis period. These include, but are not limited to, extending the life of existing structures (Keskin et al. 2021), improving environmental sustainability (Wei et al. 2016), reducing insurance premiums (Marshall 2018), increasing the market value of buildings, improving historic preservation, and maintaining the visual character of communities (Gibson et al. 2014). Second, there is a need to further improve the methods used to estimate recovery time. Recovery time is a crucial parameter for economic evaluation because it relates to business interruption, replacement and relocation, and critical services based on the building’s use category. The FEMA P-58 database is the most comprehensive resource currently available, providing extensive data on component damage and consequences to aid in recovery time estimation. However, the database does not cover all components used in US construction practice (Cook at al. 2022). Additionally, there is limited understanding of how external lifeline infrastructures may influence individual buildings, and this factor represents a significant source of uncertainty in estimating the time for buildings that rely on these external facilities to recover their services (Cook et al. 2022).

Regional-level analysis is crucial for policymakers to understand the large-scale social and economic impacts of the implemented standards and policies. In these analyses, benefit and cost outcomes are often aggregated for various building groups. Building groups are classified by building age, height, structural system, occupancy class, risk category, design level, performance objective, and/or seismic zone. Studies based in the United States have mostly relied on Hazus to estimate repair or replacement costs for structures and nonstructural components, contents losses, relocation costs, business and rental income losses. When using Hazus, it is important to supply detailed building information (FEMA 2020a), validated vulnerability models (Gibson et al. 2014), and the most recent hazard maps to the software (Ryu et al. 2010).

More recently, researchers have developed web-based software platforms to support advanced regional hazard and vulnerability analysis. The Interdependent Networked Community Resilience Modeling Environment (IN-CORE), created by the Center of Excellence for Risk-Based Community Resilience Planning (CoE), allows users to apply various hazards to infrastructure in selected areas, propagating the effect of physical infrastructure damage and loss of functionality to social and economic impacts (NIST 2020). The Natural Hazards Engineering Research Infrastructure’s Computational Modeling and Simulation Center (NHERI SimCenter) is creating a suite of computational workflows to simulate earthquake and hurricane effects on communities (Deierlein and Zsarnóczay 2021; Deierlein et al. 2020).

These platforms offer three key advantages compared with Hazus: (1) leveraging cloud computing techniques and high-performance computational resources to handle large data sets, (2) facilitating high-resolution simulation by combining the detailed assessment of individual facilities and comprehensive regional-scale simulation of natural hazard effects, and (3) using workflows to link numerous software applications, libraries, and databases, and allowing the integration of user-supplied workflow components, such as user-compiled building spatial data (Deierlein and Zsarnóczay 2021; Deierlein et al. 2020).

Using the NHERI SimCenter platform, Hulsey et al. (2022) studied the impact of safety cordons on the recovery of office space in downtown San Francisco. Xiong et al. (2019) also used the SimCenter platform to model the recovery process of nearly 70,000 residential buildings in Beijing, accounting for the impeding factor of labor constraints. Wang et al. (2021) extracted building information from street and satellite images using the SimCenter platform, and incorporated the data into the workflow to assess city-scale seismic risks.

The primary challenge in using Hazus to assess regional seismic risk is to modify or replace default vulnerability functions so as to correctly represent modern code-conforming buildings. The default seismic vulnerability functions in Hazus are based on the performance of the general US building stock through the 1990s (FEMA 2012). Modern code requirements and construction practices are not integrated into the risk analysis framework. However, it is expensive to create new vulnerability functions that reflect modern design practices for all US building types, which have nearly 700,000 categories (NIBS 2019). Therefore, some studies have used approximation methods that modify default vulnerability functions to characterize modern buildings or above-code designs, which is not a perfect solution but is efficient (e.g., NIBS 2019). In addition, while Hazus offers great flexibility for users to modify model parameters based on research needs, some fixed parameters can have a large impact on loss estimation, such as modifiers for occupancy class, damage state, and building downtime. These modifiers are determined by engineering judgment without rigorous validation, which may introduce large uncertainties into loss estimation.

Advanced regional simulation approaches such as the NHERI SimCenter and IN-CORE can overcome the constraints of Hazus, but the lack of an inventory of building characteristics presents a new challenge to the quantification of regional seismic risk (Deierlein et al. 2020). Moreover, recent improvements in these approaches are limited to building performance evaluation and loss estimation, while construction cost increases, which are important for BCA and decision-making, are assumed to be insignificant and rarely investigated. Another research opportunity is to develop computational methods for estimating costs and losses across a wide range of direct and indirect impact categories that are interoperable across systems, such as the NHERI SimCenter and IN-CORE.

In principle, BCA should take a holistic perspective that reflects the impact of a risk reduction decision on society as a whole. In practice, there are often implicit boundaries placed on the problem for tractability and measurement of losses. An alternative approach is to explicitly model the BCA perspective to properly bound the problem. An example is the owner-focused BCA method proposed by Cutfield and Ma (2015). This method entails distinguishing and identifying the cash flows between the building owner and external entities, with a specific focus on those cash flows that may vary depending on whether the building undergoes improvements. More broadly, a holistic BCA can distribute benefits and costs across relevant stakeholders, either by separating BCAs for each stakeholder (Fung et al. 2022a) or by using a benefit-transfer matrix on the community-level BCA (NIBS 2019). This step may be crucial for making a business case because it helps identify winners and losers from a risk reduction decision and thus can assist communities in prioritizing resources such as outreach and financial support.

To our knowledge, the best and only source of data for holistic, distributed BCA is the benefit-transfer matrix developed by the Multi-Hazard Mitigation Council (NIBS 2019), shown in Table 5. The benefit-transfer matrix can be used to allocate estimated costs and benefits to five closely involved stakeholder groups: developers, building owners (title holders), lenders, tenants, and communities (e.g., visitors, emergency service providers), as illustrated in Fung et al. (2022a, b). An important caveat, however, is that the benefit-transfer matrix was developed through an expert elicitation process and the underlying assumptions (summarized in Table 5) are not fully transparent or validated. While it is a valuable source of data, the lack of transparency and validation may limit its application in practice. We therefore note future research opportunities to validate the benefit-transfer matrix, provide a range of values to characterize uncertainties, or add other relevant stakeholders to this matrix.

It is a recommended practice to quantify uncertainties associated with loss prediction and cost estimation. Several methods are available:

Building codes provide minimum seismic design requirements that focus primarily on saving lives and reducing injuries, rather than ensuring that buildings are functional, habitable, or repairable after a seismic event (NIST 2021). Designing buildings beyond codes with higher performance objectives can prevent lengthy and costly repairs and decrease the likelihood of economic and social disruption. However, much of the literature to date has focused on bringing new construction up to code. The economic implications of above-code seismic design are not well documented and quantified.

Functional recovery is a new design paradigm that aims to prevent lengthy repairs of buildings after a natural hazard event (NIST 2021). Realizing functional recovery for individual buildings and infrastructure systems is a mechanism for achieving community-wide resilience (EERI 2019). Although some studies have evaluated the impact of above-code design on downtime using Hazus, the evaluations are based on regional average building characteristics, and building-specific attributes that are crucial for individual buildings to attain desired recovery performance are not considered (e.g., NIBS 2019). As new approaches are developed to estimate recovery time for single buildings (e.g., FEMA P-58, REDi, ATC-138), designing buildings to achieve desired functional recovery performance targets has become possible (Cook et al. 2022; Hutt et al. 2022; Terzic and Villanueva 2021).

In addition, Fung et al. (2022b) developed a BCA framework to support the economic evaluation of recovery-based design. The framework considers direct economic impacts and indirect economic impacts due to loss of building functions, such as population displacement, business interruption, supply chain disruption, and loss of life quality, while highlighting gaps in methods and data availability. Previous studies have suggested that indirect impacts are equally important as direct impacts from seismic events (Toyoda 2008; Petak and Elahi 2001). Table 6 presents an example application of the framework for three functional recovery objectives.

First, design standards for functional recovery are currently under development (NIST 2021). The benefits and costs of recovery-based design can vary significantly depending on the strategy employed to achieve the recovery time goal. However, this offers an opportunity for using BCA to explore the driving factors for the cost-effectiveness of resilient design. Second, functional recovery can affect stakeholders with conflicting goals if tenants rather than owners benefit from recovery of function through reductions in business interruption and displacement costs (Fung et al. 2022b). It is important to understand the mechanism by which benefits and costs are distributed among stakeholders and to utilize this mechanism to design buildings that are cost-effective to various stakeholders. Third, there is no public database that provides standard cost premiums for seismically rated nonstructural equipment, anchorage, bracing, and isolators (Tokas 2011), making cost estimates highly sensitive to market conditions and contractor profit.

Regardless of the building design criteria, there are practical challenges to implementing building codes. Adopting codes on a regular cycle incurs costs for staffing and public meetings, as well as the purchase of new code books for building departments and training for code officials (FEMA 2020a; NEEP 2021). Implementing new codes also requires extensive education and training for contractors, installers, and inspectors to ensure that code requirements are duly and properly followed (FEMA 2020a; NEEP 2021). Moreover, incentivizing compliance by offering design and construction grants, cost reimbursements, tax benefits, or fee waivers for plan review, building permits, and inspections can further increase the cost of code implementation (Multi-Hazard Mitigation Council 2020). Most of the costs are borne by local governments, except for plan review and site inspection, which are fully or partially funded by building owners through permit fees (FEMA 2020a, 1998).

Even though many states have adopted a modern building code, the effectiveness of code implementation varies widely among local jurisdictions, affecting the level of protection that building codes are able to provide in newly constructed assets (FEMA 2020a). Past natural disasters have demonstrated that poor code enforcement can jeopardize building integrity and occupant safety (ISO 2019; FEMA 1998). Porter et al. (2006) estimated that compared to typical and superior constructions, poor construction can increase the median lifetime repair cost of wood-frame small houses in California by $3,000 and $5,400, respectively, in 2002 US dollars, wood-frame townhouses by $1,400 and $1,700, and wood-frame apartments by $8,700 and $13,000, considering a useful life of 30 years and earthquake hazard only. Poor, typical, and superior constructions were classified by the strength of the structural member relative to the strength in laboratory conditions. Poor constructions have a strength between 60% and 85%; typical constructions have a strength between 85% and 100%; and superior constructions have a strength comparable to laboratory test results for high-quality specimens (Porter et al. 2006).

The costs and benefits of code enforcement are rarely assessed in BCA studies due to a lack of data and methods to quantify the monetary values. The insurance industry is using the Building Code Effectiveness Grading Schedule (BCEGS) to assess the effectiveness of communities in enforcing code requirements for earthquake and wind design (ISO 2019, 2018). The BCEGS program assigns a grade of 1 (best enforcement) to 10 (no enforcement) to municipalities every 5 years based on 27 indicators, including enforced building code edition, training and certification of code enforcers, contractor and builder licensing and bonding, number of inspection permits issued, and level of detail of plan review and inspection (ISO 2019, 2018). Policyholders may receive a premium discount in communities that actively enforce building codes, and the amount of discount could be a good predictor for code enforcement benefits. The BCEGS rating only applies to residential and commercial buildings in municipalities that participate in this program. In addition, the BCEGS database provides expenditure data collected from state and local building code enforcement departments (ISO 2019), which could be a good source for researchers to estimate code enforcement costs. However, this is an understudied area that requires further research.

The environmental benefits of seismic retrofits are rarely assessed or included in BCA. However, this does not mean that the benefits are negligible. Seismic retrofits can significantly reduce environmental impacts by reducing the likelihood of repairing, reconstructing, or demolishing a building after an earthquake (FEMA 2018; Ribakov et al. 2018; Comber et al. 2012). Comber et al. (2012) estimated that structural retrofitting of a 1960s research laboratory in California can reduce the carbon footprint by 48% (3,514 tons of carbon dioxide equivalent, or $\mathrm{t}{\mathrm{CO}}_2\mathrm{e}$), and structural and nonstructural retrofits collectively can reduce the carbon footprint by 77% ($\mathrm{5,636}\mathrm{t}{\mathrm{CO}}_2\mathrm{e}$). In highly seismic regions, seismic retrofits can prevent the same order of carbon emissions as energy upgrades (Comber et al. 2012).

On the other hand, building repair and demolition can have large environmental impacts (Gonzalez et al. 2022; Keskin et al. 2021). Following the 2010–2011 Canterbury earthquake, 7,500 dwellings and 1,400 commercial properties were demolished in Christchurch, New Zealand. Gonzalez et al. (2022) estimated that the demolition of 142 of the reinforced concrete buildings can lead to $\mathrm{159,966}\mathrm{t}{\mathrm{CO}}_2\mathrm{e}$ of carbon emissions. Demolishing newly constructed buildings can result in a greater environmental burden. Of the 142 buildings, 40% of the carbon emissions come from 30% of the buildings that were demolished less than halfway through their service life. This suggests that earlier retrofit intervention is beneficial for reducing environmental impacts.

There are three main challenges associated with quantifying the environmental benefits of seismic retrofits. First, currently available tools have a large discrepancy in environmental impact estimation due to different data collection methods and different inventories of carbon and energy for building materials, construction, transportation, maintenance, demolition, disposal, and recycling (Martinez-Rocamora et al. 2016). This implies large uncertainties in the quantification of environmental impacts and poses a challenge for incorporating environmental benefits into BCA. Second, the environmental impacts of seismic retrofits themselves are not negligible (Menna et al. 2022; Napolano et al. 2015). However, few studies investigated the trade-off between the environmental impacts of seismic retrofits relative to the avoided losses (Wei et al. 2016). The methodology for evaluating the trade-off for new building design has been discussed in the literature (e.g., Welsh-Huggins and Liel 2018), which could provide useful insights for existing buildings. Third, retrofits may extend the useful physical or economic life of buildings due to upgrades, alterations, or replacements to structural frames and nonstructural systems (Bonowitz et al. 2014). However, many studies ignore such effects and assume a useful life of 30 to 100 years based on statistical averages or initial building design (Table 2). Only a small number of studies consulted managers who are responsible for the retrofit project about the expected service life of the building (e.g., Keskin et al. 2021). In summary, with increasing attention on climate adaptation worldwide, there is a need to address these open questions on incorporating environmental benefits of seismic retrofits into BCA.

Recent studies also explored the use of combined seismic and energy retrofits for existing buildings. It is not surprising that the two forms of retrofits are often performed simultaneously for large commercial buildings in earthquake-prone regions because this can minimize displacement and business interruption compared to performing retrofits separately (Griffin 2017). In addition, combined retrofits can reduce payback periods and help cities meet decarbonization goals. Pohoryles et al. (2020) estimated that combined retrofits can reduce payback periods by up to 10 years compared to energy retrofits alone in moderate to high seismicity regions. Even in low seismicity regions, payback periods can be significantly reduced for the oldest buildings. Moreover, applying combined retrofits to 3% of the existing building stock in 20 European cities can reduce carbon emissions by 26.8%–37.7% in a decade, aligned with the European decarbonization goal of a 30% reduction by 2030 (Pohoryles et al. 2020). Mauro et al. (2017) also highlighted the importance of integrated design for seismic and energy retrofits because optimizing energy retrofits alone can diminish overall cost-efficiency.

Two methods have been proposed to optimize combined seismic and energy retrofits. One method focuses on minimizing life-cycle costs, which sums retrofit costs, seismic losses, and energy costs throughout a building’s service life. Caution should be exercised when seeking the lowest life-cycle cost because the lowest cost may correspond to the worst building performance. Gencturk et al. (2016) noted that high performance is often associated with high retrofit costs, which ultimately lead to high life-cycle costs. Another method is to maximize the benefit up to the point in which an additional investment does not result in increased performance. The building performance is measured by the green and resilient indicator (GRI), which is a ratio of expected annual seismic losses and energy costs to building replacement value (Calvi et al. 2016). A lower GRI value indicates better building performance.

There are two major challenges associated with retrofit optimization. First, energy consumption data (i.e., as energy use per square foot per year) for pre- and post-energy upgrades are rarely collected in the United States; instead, the outcomes of energy retrofits are always estimated through computer modeling (Griffin 2017). However, the outcomes of multiperformance retrofits cannot be fully modeled with existing tools designed for conventional energy retrofits (Griffin 2017). This may affect energy cost estimates and ultimately optimization results for combined retrofits. Second, retrofit costs are self-reported by building owners in the United States, which may or may not include retrofit grants, tax credits, and other financial incentives received. This causes large uncertainties in comparing data across buildings (Griffin 2017). While combined retrofits can increase cost-effectiveness, redesign and reconstruction could be more effective for buildings in very poor condition (Ferreira et al. 2015).

Third, research is needed to incorporate policy and other decision factors into retrofit design. This requires determining the appropriate weight for each item to enable condition-based optimization and prioritization (Tornaghi et al. 2018; Calvi et al. 2016). Nevertheless, some studies have argued that BCA is not a comprehensive method due to the challenges in monetizing some decision variables, such as business interruption during retrofits, the impact of structural upgrades on architectural aesthetics, and the need for specialized labor and technical design expertise. Clemett et al. (2023) proposed a multicriteria decision-making approach as an alternative method for optimizing combined retrofits, which may be a promising direction for a more comprehensive BCA that incorporates energy upgrades and other co-benefits into an economic evaluation.