Editor's Note: The 2017 update to the AIR Earthquake Model for the United States includes significant improvements for assessing the vulnerability of non-engineered wood frame construction, which constitutes the overwhelming majority of residential buildings in the United States.
This year marks the 23rd anniversary of the deadly and devastating M6.7 Northridge earthquake, which struck the San Fernando Valley near Los Angeles on January 17, 1994. Northridge resulted in more than USD 12.5 billion in insured losses.1 In Los Angeles County, more than 79,000 wood frame buildings were damaged—95% of all damaged buildings. Of the 79,000, approximately 70% were single-family dwellings, such as the house shown in Figure 1.
Given the prevalence of wood frame construction throughout the United States, the Northridge earthquake underscores the importance of accurately modeling this construction type.
The Significance of Wood Frame Buildings
Wood is the most widely used construction material in the United States, particularly for residential buildings. Census data show that, nationwide, out of 14,000 multi-family buildings completed in 2015, more than 87% were wood frame. Of 648,000 single-family houses completed that year, more than 93% were wood frame. And in the earthquake-prone Western United States, roughly 98% of all existing homes—from modest dwellings to luxurious mansions—are wood frame structures, such as the row of houses on a hilly street in San Francisco shown in Figure 2.
Significant changes in U.S. housing stock during the last 50 to 60 years have increased the focus on various primary and secondary characteristics of wood frame dwellings and their impact on vulnerability. For example, overall square footage has increased due to both the proliferation of multi-story wood frame structures and a general trend of increased room sizes in buildings.
The aftermath of historical earthquakes in the United States has revealed time and again that wood frame buildings generally perform well during earthquakes. Minor or moderate damage usually translates to low monetary losses; however, due to the prevalence of wood frame structures, the accumulated total losses could be significant for an insurer with a large number of wood frame buildings in its portfolio.
Key Updates to Wood Frame Vulnerability Assessment
The 2017 update to the AIR Earthquake Model for the United States includes several important enhancements for modeling the vulnerability of wood frame structures, which we discuss in the following sections.
Zonation: The Impact of Region on Vulnerability
Traditionally, wood frame structures have rarely benefited from engineering design mainly because, until the 1990s, wood construction practices had not been codified uniformly across the country. Unlike the seismic design provisions for engineered concrete and steel buildings—for which a precise calculation of seismic force and resisting capacity is required—the seismic design provisions for one- and two-family dwellings and townhouses (mostly wood frame construction) have been less specific.
The first code pertaining to the construction of wood frame dwellings was published in the early 1970s. It was neither mandatory nor were its requirements universally enforced by local building departments. In the absence of a mandated code, the conventional wood frame construction practice roughly followed the provisions of the One- and Two-Family Dwelling Code and prescriptions of other building codes (e.g., Uniform Building Code, UBC). However, with the elevated awareness of wood frame vulnerability prompted by the 1994 Northridge earthquake, subsequent editions of building codes and construction standards for wood frame construction became mandatory in a growing number of jurisdictions throughout the country.
In 2000, the first International Residential Code (IRC) was published by the International Code Council (ICC), the organization that publishes the International Building Code (IBC). The IRC was developed with the specific intention of overseeing the design and construction of detached one- and two-family dwellings and townhouses, as well as other residential occupancies (including dormitories and apartments of fewer than three stories). The code was adopted very quickly by many state and local governments, and by the end of 2002 the 2000 version of the IRC was in use or adopted by most states. As of February 2017, the IRC was in use or had been adopted by 49 states and the District of Columbia.
In the AIR earthquake modeling framework, seismic vulnerability zonation defines geographical regions that have followed similar trends in construction practices and code adoption over time. In the updated U.S. earthquake model, vulnerability zonation for wood frame construction is derived from the regions of moderate- to high-seismicity where prescriptive seismic design is required as defined in the IRC. Figure 3 illustrates the seven zones with distinct seismicity and construction practices implemented in the model: California, the Pacific Northwest, Nevada, the Intermountain Seismic Belt, the New Madrid Seismic Zone, South Carolina, and the rest of the contiguous United States.
In addition to code adoption and construction practices, code enforcement is a significant factor affecting the vulnerability of wood frame structures. To account for variations in building code enforcement, which can result in variations in construction quality, AIR has incorporated the Building Code Effectiveness Grading Schedule (BCEGS®) scores to determine a jurisdiction’s adherence to the adopted building code and the quality of enforcement. The BCEGS program was established by AIR sister company ISO® to evaluate how the building codes are enforced in a particular community.2
In coastal areas of the Atlantic Seaboard and the Gulf of Mexico, modern wood frame houses are designed and built to resist the lateral loads imposed by hurricane-force winds. As a result, the earthquake vulnerability of these structures is also lower. The AIR model accounts for these side benefits by including an additional vulnerability zone along theses coastal areas, which is overlain with the zonation shown in Figure 3.
Age Bands: Impact of Year of Construction on Vulnerability
In the updated U.S. earthquake model, the periods of time in which certain construction practices and design provisions fundamentally remain unchanged are termed “age bands.” The age bands are defined by identifying milestones in the evolution of design and construction practices. Buildings located within the same vulnerability zone and built during the same age band are assumed to exhibit similar seismic performance. Accurate modeling of the complex evolution of spatial and temporal vulnerability variation across the entire country is achieved through a combination of appropriately defined seismic zones and age bands.
Figure 4 shows the relative vulnerability of two-story wood frame houses built in different years in Los Angeles. With the exception of a temporary increase in the mid-1960s, which is addressed bit further on, a fairly consistent reduction in vulnerability (i.e., better seismic performance) has occurred over time.
Better performance of more recently constructed buildings can be attributed to improved standards and building codes (requiring plan review and more frequent inspection by building officials), resulting in better design, higher-quality construction materials, more sophisticated hold-down systems to resist overturning, specially manufactured hardware to enable more robust load paths, better workmanship, and more stringent code enforcement throughout the design and construction process. In addition, the better seismic performance of newer construction over old is partly attributable to the fact that wood is a natural material that can deteriorate as structures age, resulting in increased vulnerability over time. The higher vulnerability of pre-1940 single-family dwellings was evident in the damage survey after the 1971 M6.6 San Fernando earthquake in Southern California.3
The relative vulnerability shown in Figure 4 is consistent with damage observations after the 1994 Northridge earthquake, which indicated that wood frame buildings constructed in the 1940s and 1950s performed, on average, better than those built in the 1960s. Why was the newer construction more vulnerable? Although houses were built with conventional practices in the 1940s all the way to 1970s, houses built during and immediately after WWII generally had simpler geometries in both plan and elevation, were smaller in size, had smaller rooms, and had smaller and fewer window and door openings in the exterior walls, which provide the majority of the lateral support for wood frame buildings. Architectural features introduced in the 1960s led to homes with fewer solid exterior and interior partition walls, resulting in a reduction in the lateral load-carrying capacity of these homes. In addition, during the 1960s it was assumed that the performance of stucco and gypsum wallboard was higher than it in fact was.4
In addition to changes in vulnerability as codes and architectural styles evolve within a state, the relative vulnerability of wood frame construction varies from region to region of the country, as codes and construction practices have not been adopted and enforced at the same time across the United States.
Additional Vulnerability Differentiators
While the vulnerability of wood frame houses has generally decreased over time (as depicted in Figure 4 for two-story single-family dwellings), substantial differences have been observed between the seismic performance of one- and two-story homes. For example, during the 1971 San Fernando earthquake it was observed that of all contemporary homes, one-story dwellings performed better than those of other heights.5 Also, detailed evaluation of building damage data after the 1994 Northridge earthquake revealed that two-story and taller wood buildings were 2.5 times more likely than one-story wood buildings to have been assigned a “Red Tag” (a “Red Tag” indicates that a building is unsafe for habitation).6
Taking into account similar observations for several historical earthquakes, the AIR model distinguishes between the vulnerability of one-story and multiple-story wood frame buildings, including modern wood frame, masonry veneer, and heavy timber. Figure 5 illustrates the relative vulnerability between one- and two-story wood frame single-family homes built in 1980 in Los Angeles.
There are several reasons for the difference in performance. First, many multi-story wood frame buildings are affected by the “soft story problem,” in which the structural stiffness or strength is much lower on the first floor than other floors because of large openings (such as garage doors or open floor plans), while upper floors have more partitioned rooms. In addition, seismic demand is driven by the combined effects of the frequency content of ground motion and the natural period of buildings. The natural period of a single-story building is typically shorter than that of a two-story building, which means two wood frame buildings of different heights will likely experience different spectral accelerations. The spectral acceleration experienced by a one-story wood frame building is typically lower than that of a two-story building during an earthquake, resulting in lower seismic demand and typically reduced damage and loss for one-story buildings.
It has also been observed that larger, higher value homes tend to be more susceptible to earthquake damage. While it may seem counterintuitive for more expensive homes to perform more poorly, there are several engineering explanations that support this observation. First, surveys following the 1971 San Fernando and 1994 Northridge earthquakes indicated that the vast majority of damage to residential structures was to the interior and exterior finishes. Analysis of data from Verisk’s 360Value® show that, for more expensive houses, more of the home’s value is concentrated in the most vulnerable interior and exterior finishes. Therefore, expected losses are higher for larger, high-value homes, which have a larger proportion of their value in the interior and exterior finishes. Additionally, rooms in larger houses are typically larger as well, with longer, unsupported spans. These longer spans are more susceptible to damage during an earthquake, causing these homes to be more vulnerable than their standard-sized counterparts. The updated AIR Earthquake Model for the United States explicitly models the distinct vulnerability of high-value homes (also referred to as high net-worth properties).
Managing U.S. Earthquake Risk through Accurate Modeling of Wood Frame Structures
The accurate modeling of wood frame structures is a very important component of any model that seeks to estimate earthquake losses in the United States. This is particularly the case for the seismically active regions of California, Oregon, and Washington, where wood framing is the predominant construction method for residential dwellings.
In general, wood frame buildings perform well during earthquake events, and the level of damage per building is low. However, total losses could still be high because of the substantial number of wood frame structures in the United States. In addition, even strict adherence to building codes is not a guarantee against substantial losses because building codes are designed to prevent fatalities and injuries resulting from building collapse but are not necessarily intended to prevent monetary loss due to building damage. In the 2017 update, the AIR model’s temporal and spatial resolution is increased to account for the evolution of construction practices of non-engineered wood frame buildings across the United States. The seismic vulnerability of one-story and two-story dwellings is differentiated as is the vulnerability of large, high-value homes.