UC San Diego

Characterizing the performance of retrofitted and unretrofitted single-family wood frame houses has become increasingly important in California due to the high seismicity of the state and the often-poor seismic resiliency of some portions of the housing stock. From field observations following earthquakes, inadequate lateral bracing of cripple walls and sill bolting are primary reasons for significant failures of residential homes, even in the event of moderate intensity earthquakes. Cripple walls are the short-elevated wall segments supporting the first-floor framing and upper stories of a dwelling. These important components were common in construction practices in older, particularly pre-1970s homes, as they provided a crawl space for ventilation and placement of utilities. Often, they lack adequate bracing and load transfer to carry seismic inertial loads generated during an earthquake from the superstructure to the foundation. While methods to retrofit weak cripple walls and improve sill anchorage have been developed, the improvement in performance with retrofit has not been adequately quantified. In addition, little knowledge is available to quantify the performance of houses with unretrofitted cripple walls and sill anchorages. To address the paucity of experimental data to characterize the seismic performance of cripple walls and sill anchorage, a series of 28 cripple wall-only assemblies were tested. Details representative of era-specific construction, specifically the most vulnerable pre-1970s construction, are of predominant focus in the present effort. Parameters examined include cripple wall height, exterior finish materials, retrofit condition, boundary conditions, anchorage type, gravity load, and loading protocol. Of the 28 assemblies, all but one was tested using a newly developed quasi-static reversed cyclically displacement-controlled protocol intended to simulate seismic demands from the upper story of a dwelling. The remaining assembly was monotonically loaded. Each specimen was nominally 12 feet in length and either 2 feet or 6 feet in height. Exterior finish materials were either wet (stucco) or dry (non-stucco) application types. The experimental program was designed to elicit the effect of a single parameter on the seismic behavior of the cripple wall by changing one parameter between each specimen and controlling the remaining. For each specimen, key measurements pertaining to the hysteretic response as well as documentation of the damage evolution were collected. Results from these experiments provided an experimental basis to validate a numerical modeling tool, and ultimately to develop loss models, which are intended to quantify the reduction of loss achieved by applying state-of-practice retrofit methods as identified in modern retrofit design guidance (FEMA P-1100, 2019). Findings from the experimental program revealed that the implementation of the FEMA P-1100 prescriptive retrofit dramatically improved the strength, stiffness, and energy dissipation of cripple walls, and in many cases, it also increased the drift capacity. In addition, it was found that horizontal siding was the weakest finish material tested regardless of the cripple wall height, while diagonal sheathing was found to be the strongest finish material. Following the experimental program, finite element connection-level numerical models were developed to predict the response of dry finished cripple walls tested in the experimental program. Subsequently, phenomenological numerical models were developed to capture the global response of wet finished cripple walls. A cross-comparison of the numerical models demonstrates a high level of accuracy of the connection-level modeling approach, while the phenomenological models did well in capturing the pre-peak response of wet finished cripple walls but lost accuracy when tasked with capturing well beyond post-peak response, due to the disengagement of the stucco finish. The validated connection-level models were used to conduct a parametric study on horizontal siding finished cripple walls in an effort to investigate equivalent, simple detailing methods as an alternative to modern retrofit guidelines, which prescribe installation of full or partial sheathing within the interior space. Detailing alternatives investigated included, siding board size, vertical nail spacing, cut-in diagonal blocking, and diagonal strap bracing. Parametric study results indicate that a strong linear correlation between the vertical nail pair spacing and lateral strength and hysteretic energy dissipation of horizontal siding finished cripple walls is realized. Furthermore, it was demonstrated that the implementation of diagonal cut-in blocking elements provided the largest increases in strength and stiffness for shorter, low aspect ratio 2-ft-tall cripple walls, while diagonal strap bracing provided significant increases for taller, high aspect ratio 6-ft-tall cripple walls. Ultimately, the findings of this dissertation fill prominent gaps in the testing database of cripple walls and sill anchorage, including both the hysteretic response as well as the damage evolution. Moreover, experimental findings demonstrate the substantial benefits of implementing modern retrofit design guidelines, while numerical simulations offer promising alternative details, each geared towards improving the seismic performance of cripple walls.