University of Twente Student Theses

Login

Positioning Police Emergency Vehicles : Determining Facility Locations and Routing Techniques for Fast Emergency Response

Muller, P. (2014) Positioning Police Emergency Vehicles : Determining Facility Locations and Routing Techniques for Fast Emergency Response.

[img] PDF
3MB
Abstract:Last year, the Dutch police adopted a major change, resulting in the need for closing a significant number of police stations in the nearby future. For the base team IJsselstreek, where this research is conducted, this means that at least one, probably two and maybe three out of four police stations need to close. One of the major impacts of closing a police station in the area is the assumed higher average response times for high-priority incidents, since not every location is covered anymore. This means that a different approach is needed to be able to cover the total area. During this research, the organization does not know which and how many police stations will close (a decision where this research can help), so different scenarios need to be analysed. Currently, three emergency vehicles are available for the base team IJsselstreek, but the organization is also interested in the consequences when less and more vehicles are used. This resulted in the following research goal: ‘Give the base team IJsselstreek insight in the consequences of (i) using different police station locations and (ii) the number of emergency vehicles, with respect to the response times of high-priority incidents.’ We developed an emergency vehicle positioning model, where we maximize the expected coverage fraction, given a number of available vehicles and locations of police stations. This model is able to be solved to optimality within a reasonable amount of time when we generate a 48 hour plan. Moreover, when this model is used real-time, it can be solved again after an incident happens. This means that, when an incident happens and the nearest vehicle responds to it, the positioning model is able to reallocate the remaining vehicles in an optimal way. For the expected demand, which is the input of the positioning model, we used historical data that includes all incidents from the years 2011 – 2013, registered with date/time groups, priority and coordinates. This enables us to create a forecast for the area of IJsselstreek, where we included weekly, daily and hourly patterns. We decided to divide the total area of IJsselstreek into 85 regular hexagons, resulting in a forecast for each hexagon. Furthermore we used a smart heuristic to get good estimates of the travel times between each pair of hexagons. To test the developed positioning model in combination with the created forecast, we set up five sets of experiments to simulate for one year. Our key performance indicator is the on-time percentage of high-priority incidents, i.e., the percentage of prio 1 incidents that has a response time less than 15 minutes. It appeared that, when at least one police station needs to be closed, the best option is to close Eerbeek. Furthermore, there is no significant difference between the police stations in Lochem and Twello. Another conclusion is that, on average, an improvement of 5.4% in the on-time percentage can be achieved when applying the optimal positioning method, in comparison with an alternative way of positioning where all vehicles are standby at the police stations. Furthermore, the length of the shift changing time has a significant impact on the on-time percentage; going from 1 hour to 4 hours, results in an average decrease of 4.4%. Also the use of the proposed forecast is tested versus a simple forecasting method, where the performance of the extended forecast scores on average 3% better than the simple one. Finally we concluded that the addition of fairness constraints, i.e., prevent having some areas never be covered within 15 minutes, results in a loss in the average on-time percentage of 2.2%. Based on this research, we recommend applying (i) the proposed forecasting method and (ii) the developed mathematical positioning model. Implementing the positioning model requires a smooth cooperation with existing Geographical Information Systems that the organization uses, so we recommend developing an integrated support tool. The forecasting method has room for improvement, since we aggregated all prio 1 incidents and we did not distinguish, for example, robberies from traffic incidents. Finally, options for further research include the behaviour of other vehicles, like motors, and the cooperation with neighbouring areas to get an overall optimal result.
Item Type:Essay (Master)
Clients:
Politie, Zutphen, Netherlands
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:50 technical science in general, 55 traffic technology, transport technology
Programme:Industrial Engineering and Management MSc (60029)
Link to this item:https://purl.utwente.nl/essays/65207
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page