360 likes | 685 Views
Project #3: Optimizing Fire Station Locations for the Istanbul Metropolitan Municipality. Ceren Uzun Abdullah Yılmaz IE 479. Content. General Info What Istanbul Metropolitan Municipality( IMM) wants,why needed? Current Situation
E N D
Project #3: Optimizing Fire Station Locations for the Istanbul Metropolitan Municipality Ceren Uzun Abdullah Yılmaz IE 479
Content • General Info • What Istanbul Metropolitan Municipality(IMM) wants,why needed? • Current Situation • Project: Objective, Proposed models, Additional Examples, Data Acquisition(GIS & ArcGIS) • Scenarios • Model 1&2 • Results • Web Search:Questioning, Comparison in China, Refrences
General Info • Emergency facility location is critical • Deals with human lives • Appropriate locations crucial for megacities, e.g. Istanbul • Istanbul • Megacity • Cultural and financial center • Population: 13.5 million (2009) • 2.5 million immigrants since 2003 • Continous growth & expansion • Traffic congestion, infrastructure deficiencies
What IMM Wants • IMM responsible for station location • 790 subdistricts • Istanbul, My Project – encourage universities for research projects on public services • Locations for additional fire stations • Aim: Fire incidents responded within 5 mins • Restrictions: • At first, no restricitions • Later on, a budget limit = $38,392,768
Why? • Earthquake • Cultural Heritage • Continous growth & expansion • Traffic congestion, infrastructure deficiencies • Consequence: Fire stations can’t meet demand!
Current Situation • 60 fire stations, 37 on European, 23 on Asian side • Four size categories: A, B, C, D
Current Situation cont. • 58.6% of subdistricts (463 subdistricts) reached within 5 mins
Project • Start of the project: 2009 • Objective: • Cover max number of possible locations by opening as many fire stations as possible, given budget constraints. • Determine the locations for the stations
Proposed Models • Exact methods used: • Two models (just in case) • Set-covering Model • Find the optimal locations to minimize the cost of facility location while covering all subdistricts within time limit (5 mins) • Maximal-covering Model • Maximize the coverage level given a budget limit ($38,392,768)
Additional Examples • The set and maximal covering models used also for police department and bus station locations.
Data Acquisition for the Models • GIS • Proximity Matrix • IMM • Historical fire incident data for 1994-2006 • Locations of available fire stations
Geographic Information System (GIS) & ArcGIS • GIS enables: • Storage, retrieval, manipulation, analysis, and visualization of geographical content • In this project, ArcGIS is used. • Determine the coverage areas of fire stations • Calculates travel times btw subdistricts • If <5 mins, then considered as covered • Takes into account type of roads, travel speed, etc. • Each subdistrict represented as a single point, the coordination of center of gravity of every building in a subdistrict, for ease of distance calculations
Proximity Matrix • ArcGIS results:
10 Scenarios • Scenario 1: • The initial situation • Scenario 2: • Scenario 1 w/ set-covering model, 100 percent coverage • Scenario 3: • Scenario 2 w/ forecasted fire incidents for 2015
10 Scenarios (cont.) • Scenario 4: • No existing fire station, set-covering model, 100 percent coverage • Scenario 5: • Scenario 4 w/ forecasted fire incidents for 2015 • Scenario 6: • Budget restriction & maximal-covering model
10 Scenarios (cont.) • Scenario 7: • Scenario 6 w/ forecasted fire incidents for 2015 • Scenario 8: • Budget restriction & maximal-covering model (Scenario 6) & weighted heritage size • Scenario 9: • Scenario 8 w/ forecasted fire incidents for 2015 • Scenario 10: • Past situation (2005) used for comparison
Model 1 • (A1) min cost of opening stations • (A2) right type of station to respond to SRs • (A3) only one type for each subdistrict
Model 2 • (B1) max’s coverage of service requests • (B2) right type of station to respond to SRs • (B3) tot. # of stations < 64 • (B4) only one type for each subdistrict • for Scenario 6 • for Scenario 8 • +1)
Models Solved • Gams is used to code the IP model and CPLEX to solve. Largest model: 3,208 binary variables, 6,416 constraints, 0.708 seconds
Results • Scenarios: unlimited budget, building from scratch, placing additional weight • Mathematical model , visual GIS serve as decision support system • IMM selected scenario 6 : 64 new type D station; coverage percentage increases from 58.6 to 85.9
Questioning • Question: Why not consider where to locate within the subdistrict? • Question: Coverage time=6 min?
Questioning cont. • Mehmet Savsar from Kuwait University • In metropolitan areas, cross of highways or other traffic obstables through obstacle-overcoming points • E.g. Overpass, underpass with traffic lights, etc. • Focusing on the location of one station and its coverage • A model developed based on a function called Achievability Strength
Comparison in China • Shanghai Fire Research Institute of Ministry of Public Security • (0,4min],(4,8min],(8,12min],(12,15min],(15,45min] 5 scenarios • Traditional circle-drawing method • Planning method based on travel time of fire vehicles • Priority: important buildings, companies, factories are with high fire risk • Genetic algorithm (heuristic) could be used • %140 increase on fire protection response
Refrences • Yan-FeiYu, "Optimization for Fire Station Location Planning Based on Travel Time of Fire Vehicles," Computational Intelligence and Security (CIS), 2012 Eighth International Conference on , vol., no., pp.328,331, 17-18 Nov. 2012 • Savsar, Mehmet. "Modeling of Fire Station Locations Under Traffic Obstacles." Wseaes. N.p., n.d. Web.