By William S. Ginell

The Andres Pico Adobe (originally constructed in the mid-19th century) after the 1994 Northridge earthquake. Photo: E. Leroy Tolles.

Adobe buildings—the vernacular earthen architecture of the Spanish colonial past in the Americas—are a vanishing feature of the western United States. In California, for example, fewer than 5 percent of the estimated 900 adobes originally constructed in the San Francisco Bay area still survive. Statewide, only about 350 historic adobes are left.

Many of the buildings now gone were destroyed by earthquakes, and those still standing remain highly vulnerable to future quakes. But because current methods of strengthening adobes are both invasive and expensive, most building owners have been reluctant to undertake retrofitting measures. Using materials such as steel and concrete, these methods often produce results that are too intrusive and are physically and aesthetically incompatible with adobe. In addition, they frequently involve architectural alterations inconsistent with preserving the historic fabric of the buildings, ignoring the fact that every part of an adobe—down to the individual, handmade bricks of dried mud—is an artifact whose modification or loss diminishes the historic record that the building represents.

In 1990 the Getty Conservation Institute initiated the Getty Seismic Adobe Project (GSAP) to investigate alternatives to existing methods of retrofitting. Six years later, after studying dozens of historic adobe buildings, analyzing recent earthquake damage to adobes, and developing and evaluating new retrofitting techniques through numerous tests, the project's team has come up with ways to provide seismic protection at a reasonable cost while substantially preserving the authenticity of historic adobes. A departure from current retrofitting practice, the methods developed by the project are, for the most part, simple and inexpensive enough to be implemented by unskilled labor in areas with limited resources. They can therefore also be used in the many communities around the world that still rely on earth as a basic building material.

At the project's beginning, a number of historic California adobes were studied so that their structural conditions could be correlated with their architectural features and any prior retrofits. The study suggested (as might be expected) that tall, thin-walled adobe buildings are very susceptible to collapse once the walls crack. In contrast, many unretrofitted buildings with thick walls and a low, wall height/thickness ratio had survived earthquakes, even though the walls were cracked. Cracking in these structures results in the formation of large segments of wall that rub against each other during quakes and dissipate energy by friction. Only when the relative displacement of the blocks becomes very large do walls fall and roofs collapse.

The challenge for the GSAP team was not how to prevent cracking - an inevitable consequence of an earthquake - but, rather, how to minimize the movement of those large wall segments during a quake. For that reason, the team proposed retrofitting methods aimed at achieving stability rather than at increasing strength. The principal techniques investigated included the installation of nylon straps that encircle the walls horizontally, vertically, or both. These straps absorb energy and can be easily hidden beneath a coat of plaster. A second method involved the use of center cores - thin, flexible steel rods placed in holes drilled vertically into the wall and grouted in place.

A one-fifth-size scale adobe model, unretrofitted, after being subjected to a series of simulated earthquakes.

A retrofitted model still standing after undergoing simulated earthquakes of substantially greater intensity. Photos: Louise Walker.

Both methods were designed to restrain the movement of adobe blocks and prevent walls from overturning. They were tested on one-fifth-size model buildings that were subjected to simulated earthquakes on a computer-controlled shaking table at the Stanford University Blume Earthquake Engineering Test Center. The tests studied the effects of various combinations of retrofitting techniques on both the in-plane and out-of-plane behavior of walls and the impact of varying wall height/thickness ratios. The model buildings consisted of four walls, 1.5 meters long and 0.6 meters high (5 by 2 feet). Each model had door and window openings. Also tested were models based on a tapanco-style building, a typical southwestern American design that includes floor and roof systems and highly vulnerable gable end walls. Two of the three tapanco models were retrofitted with different combinations of straps and center core rods.

The results of the roofless model tests demonstrated that stability-based retrofits do indeed increase a model's seismic resistance to collapse, preventing walls from overturning and minimizing permanent displacements. The retrofitted tapanco models also displayed dramatic improvement in stability over the one unmodified model. Especially significant were the stability and damage control afforded by the thin center core rods.

The one-half-size scale model of a tapanco-style adobe with no retrofitting after the collapse of a gable end wall during testing on the shaking table in Skopje.

The gable end wall at the moment of collapse (the thin cables visible on the wall were part of the testing's instrumentation and not retrofitting).

The gable end wall of a retrofitted tapanco adobe model after a test of equal intensity. Photos: William S. Ginell.

One factor that could not be assessed in the small models was the effect of gravity on the behavior of more massive adobe walls. To address this, two tests were conducted on a large shaking table in Skopje, capital of the Former Yugoslav Republic of Macedonia, where the GCI is doing seismic retrofitting research on Byzantine churches (see Conservation, vol. 9, no. 3). Retrofitted and unretrofitted tapanco models, identical in design to those tested earlier but larger (being one-half size), were built with walls 3.6 meters long and 3 meters high at the gable end walls (12 by 10 feet, excluding the roof). Both models were instrumented to provide quantitative information. In these tests, the gable end wall of the unretrofitted building collapsed at about the same level of shaking intensity as the smaller scale models. The crack patterns, too, were very similar. In the retrofitted building, the straps and especially the center cores proved very effective in preventing collapse. Since an increase in the size of the scale models did not change the test results, gravity does not appear to be a significant factor.

The January 1994 Northridge earthquake provided affirmation of the approach taken by GSAP. In the months following the quake, the GSAP team surveyed 19 historic adobes in the Los Angeles area, documenting 8 in detail. It was found that many of the adobes suffered damage similar to that seen in the tests of unretrofitted models. This finding supports GSAP's experimental methods and results, offering additional evidence that the project's retrofitting techniques would prove effective in an earthquake.

The inexpensive and minimally invasive techniques tested in the project have application beyond the preservation of historic California adobes. In seismically active areas of the world, such as Latin America and China, where earthen architecture is widely used, the groundbreaking work of GSAP can also be employed to increase the stability of buildings and limit loss of life. The project's final annual report and a summary report entitled Guidelines for Seismic Stabilization of Historic Adobe Structures (both to be published later this year) will encourage the use of the methods developed to retrofit and protect not only the western United States's historic architectural heritage but vernacular earthen architecture around the world.

William S. Ginell is Head of Monuments and Sites in the GCI's Scientific Program; he coordinates the work of GSAP for the Institute.