Modeling housing recovery after the 2018 Lombok earthquakes using a stochastic queuing model

Irene Alisjahbana and Anne Kiremidjian

Post-earthquake housing recovery monitoring is necessary, especially since the housing sector usually represents 50 percent of the total monetary disaster loss. However, very scarce recovery data, in addition to the complexities of the recovery process, make modeling housing recovery very difficult. Time-based stochastic models, which are commonly used in well-known frameworks such as the U.S. Federal Emergency Management Agency’s HAZARD-US (HAZUS), do not explicitly capture how the recovery process occurs in real life. In this article, we introduce a stochastic queuing model that considers the total number of damaged buildings, the damage distribution, resource constraints, and government-led reconstruction prioritization strategies. We applied our model to seven regions affected by the 2018 Lombok earthquakes, which destroyed over 226,000 residential buildings. In this study, we use publicly available daily data of the reconstruction progress obtained by local authorities for all damaged buildings. Using that dataset, we present recovery parameters for the Lombok region, including delay and reconstruction times. These parameters are an improvement over parameters currently available, which only apply to the U.S. region. Furthermore, we show that our model captures the observed recovery trajectory disaggregated by damage states, which provides insights into how the different building damage states recover individually. Results show that our queuing model reduces the root mean squared error (RMSE) of the recovery trajectory by 31.58% and is better able to represent the observed overall recovery trajectory compared to a time-based stochastic model.