Dynamic Analysis of Adobe Structures

Abstract

Adobe is an inexpensive material widely used in developing countries such as Peru. Its use as construction material does not demand expertise. Unfortunately, buildings constructed using adobe have a record of poor behavior when subject to strong ground motion. There are several ways to reinforce adobe buildings. For instance, black and nylon mesh can be introduced to improve the structural integrity. This research investigation studies the behavior of a one-story adobe building subjected to the 1970 Peruvian earthquake. Experiments were performed at the Catholic University of Peru (PUCP) and examined the undamaged and damaged conditions of a one-story adobe building. The experimental data from the phase 2 excitation case (60 mm maximum displacement) was analyzed and compared with ABAQUS models. A variety of statistical and transform methodologies are available to study problems in structural dynamics and this study explores their use in modeling and characterizing the excitation and response of this building. A brief introduction regarding applications of Fourier and Wavelet transforms, and statistical methodology used in structural dynamics is presented. This is followed by a discussion of the mechanical properties of adobe building materials and the development of the ABAQUS building models. The data and the corresponding numerical predictions area characterized in terms of the frequency content, cross-correlation, Root Mean Square Error (RMSE) and Statistical peak ground acceleration (SPGA). The analysis results included estimates of the structural natural frequencies and corresponding modal information. In addition, PDF and CDF plots of the response behavior were used to characterize time series to compare the experimental measurements and the numerical simulations. Finally, both Fourier and wavelet analyses were used to characterize frequency content and the latter method was used to investigate time as well as relation to natural frequencies in the numerical models.