Iterated local search for the placement of wildland fire suppression resources
•Combating forest wildfires requires deciding on where to place suppression resources.•We model the effect of positioning resources on the fire spread behaviour.•We present a mixed-integer programming model and an iterated local search heuristic.•The heuristic is effective both in terms of time and...
Uloženo v:
| Vydáno v: | European journal of operational research Ročník 304; číslo 3; s. 887 - 900 |
|---|---|
| Hlavní autoři: | , |
| Médium: | Journal Article |
| Jazyk: | angličtina |
| Vydáno: |
Elsevier B.V
01.02.2023
|
| Témata: | |
| ISSN: | 0377-2217, 1872-6860 |
| On-line přístup: | Získat plný text |
| Tagy: |
Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
|
| Shrnutí: | •Combating forest wildfires requires deciding on where to place suppression resources.•We model the effect of positioning resources on the fire spread behaviour.•We present a mixed-integer programming model and an iterated local search heuristic.•The heuristic is effective both in terms of time and objective function values.•Tests were made for rasterised landscapes with up to 900 nodes.
We consider the problem of, given a landscape represented by a gridded network and a fire ignition location, deciding where to locate the available fire suppression resources to minimise the burned area and the number of deployed resources as a secondary objective. We assume an estimate of the fire propagation times between adjacent nodes and use the minimum travel time principle to model the fire propagation at a landscape-level. The effect of locating a resource in a node is that it becomes protected and the fire propagation to its unburned adjacent nodes is delayed. Therefore, the problem is to identify the most promising nodes to locate the resources, which is solved by a novel iterated local search (ILS) metaheuristic. A mixed integer programming (MIP) model from the literature is used to validate the proposed method in 32 grid networks with sizes 6x6, 10x10, 20x20 and 30x30, with two different number of fire suppression resources (64 problems). Our ILS produced optimal solutions in 40 cases out of 41 known optimal lower bounds. The proposed method’s effectiveness is also due to its short computing times and small coefficients of variation of the objective function values.
We also provide a categorised literature review on fire suppression deterministic optimisation models, from which we conclude that approximate collaborative approaches seldom have been applied in the past and, according to the results obtained, can successfully address the complexity of fire suppression, reaching good quality solutions even for large scale instances. |
|---|---|
| ISSN: | 0377-2217 1872-6860 |
| DOI: | 10.1016/j.ejor.2022.04.037 |