Part of the WDSA / CCWI 2018 Joint International Conference in Ontario was the Battle of Post-Disaster Response and Rehabilitation. Unfortunately no winner was declared, but I would like to share the solution Claudia Quintiliani and Simone Santopietro and I developed, using A Many Objective Optimization and a Greedy algorithm. You can read about or watch the eight minute video with step-by-step explanations.
In this year’s problem five different Damage Scenarios are considered in which a set of pipes in a drinking water network are either broken (require isolation and replacement), or have leakages (require only reparation). For this battle, six (6) criteria are considered in the problem formulation:
- Time that the hospitals (2 per scenario) and the hydrants (2 per scenario) are without supply,
- Rapidity of recovery (time until the system recovers permanently 95% of its functionality),
- Resilience Loss (accumulated loss of functionality = area under functionality),
- Average time each consumer (node) is without service,
- Number of consumers (nodes) without service for more than 8 consecutive hours, and
- Volume of water lost during the next 7 days after the event
Here functionality is estimated as the sum of all demands of all nodes for a particular solution of a damage scenario in relation to the sum of all demands of the system when no damages have occurred, for each time step.
Solution TEAM Castro-Gama, Quintiliani & Santopietro
The solution proposes the order in which the pipes need to be repaired by 3 different crews, for each of the Damage Scenarios.
Results for 2 different algorithms (A Many Objective Optimization and a Greedy algorithm) and 4 different preventive engineering interventions considered (increasing or decreasing pumping capacity and storage capacity) are presented in the following video together with the Methodology and proposed solution.