Citation

Kumar V, Ramamritham K, Jana A. Transp. Res. Proc. 2020; 48: 345-362.

Copyright

(Copyright © 2020, Elsevier Publications)

DOI

10.1016/j.trpro.2020.08.030

PMID

unavailable

Abstract

Emergency Response Services (ERS) in the developing countries often face the challenge of distributing the resources in a manner to provide optimal service. Moreover, the exclusion of heterogeneous urban fabric and considerable variations of travel time and coverage demand throughout the day often leads to inadequate solutions. Hence, we propose an approach to incorporate dynamic aspects like demand, travel time, and coverage area in developing an asset location model. We illustrate how the travel time distribution produces more reliable coverage results when compared to the model considering fixed travel times over the periods. We also incorporate the influence of urban settlement elements like built-up compactness etc. in the resource allocation. A machine learning based approach is proposed to estimate the varying demand. The prediction model predicts the firefighting vehicles demand with high accuracy. We formulate a mixed-integer program to maximize the empirical demand coverage by firefighting vehicles. The proposed model is applied to the southern region of Mumbai, India. We use percentage coverage and coverage variation difference as the metrics to test the performance of the model. When compared to the static travel time model considering fixed average travel times the dynamic model showed more variability in the coverage. This supports our hypothesis that demand and the travel time variations influence the coverage. For the dynamic case, the maximum coverage variability for the vehicles was found to be in the range of 9%, whereas for the static case it was found to be in the range of 6%.

Language: en

Keywords

Data-driven optimization; Dynamic adaption; Emergency response services; Resource allocation; resource planning; Sustainable cities; Travel time