Delft University of Technology
Faculty Mechanical, Maritime and Materials Engineering
Transport Technology

P. de. Jong Simulating city distribution with eM-plant and Mappoint, an evaluation
Computer program, Report 2005.TL.6966, Transport Engineering and Logistics.

City distribution is defined as the transportation of consumer goods from, to and in urban regions. City distribution is dedicated on deliveries of consumer goods to stores, department stores, hotel and catering industry, offices and direct to the customers homes. Due to changing conditions, the distribution of goods in urban regions must also change. The section Transport Engineering and Logistics of Delft University of Technology has done research on city distribution which resulted in the CiDiS model.

CiDiS model
The CiDiS model (City Distribution) investigates the city-box concept which represents a logistic chain. City boxes are transported by different vehicles from distribution centres via transhipment locations to stores as shown in the next picture.

The model works as follows: The model consists of different modules to control the simulation of infrastructure, transportation and planning.

eM-plant model
The eM-plant model is developed after the CiDiS model. This models works globally the same as the CiDiS model. An important difference is the animation of the distribution chain with help of Microsoft MapPoint. The eM-plant model has three input modes for the simulation of more logistic distribution concepts which is an extra possibility in comparison with the CiDiS model.

The eM-plant model also consists of different modules for controlling the different objects (vehicles, locations, rides and orders) during the simulation. Important modules are that for data input, route construction, animation and trip planning.

Validation is necessary if the working of the eM-plant model must be compared to working of the CiDiS model. Two cases (Amsterdam city centre, province Groningen) are used for the validation test. The results show that travelling times and driving distances differ because both models use different route data. This difference in route information can be eliminated by correction of vehicle speeds in Microsoft MapPoint which levels travelling time differences.

The validation shows that he eM-plant model has a slightly better order planning. This is due to the possibility of the eM-plant model to deliver city boxes to different stores in one ride. The CiDiS model has not this possibility.

An animation gives a clear overview and insight in the working of the model and performance of the distribution concept. The simulation objects are projected on a map in eM-plant. eM-plant uses Microsoft MapPoint to construct routes and to receive maps as background layer for the animation. An interface between MapPoint and eM-plant takes care that route information is send from MapPoint to eM-plant and vice versa. Unfortunately, the routes used by the vehicles are not realistic at all. They drive through cannels for example. This is very unrealistic and not convincing. The interface returns only one point per route part which is too inaccurate for a good and convincing animation. Some visual programming tricks are proposed to improve the animation.

Simulation of congestion
To make the simulation more realistic, traffic congestion is taken into account. In contrast with the CiDiS model, the eM-plant model cannot use congestion information in the planning of rides. Congestion is simulated with help of areas projected on the map used in the animation. Vehicles drive slower in these areas during the simulation. The creation of congestion area creation is time-consuming job. Only the areas shown by the map can get congestion properties. This type of congestion simulation in combination with the earlier mentioned bad animation makes the congestion simulation unreliable.

Order planning
Order information from the stores is read from an Excel file. The eM-plant simulation model uses heuristics to construct a ride planning with this data. After the trips are constructed, they must be assigned to vehicles during the simulation. A heuristic constructs the rides in the following steps: During the simulation, rides are assigned to the vehicles on time basis. The delivery with the shortest time to delivery is loaded first. This applies both for distribution centres and for transhipment locations.

Three differences can be distinguished between the eM-plant model and CiDiS model: The eM-plant model is also compared to an improved CiDiS planning model made by Van Rijswijk [H.J. van Rijswijk "CiDiS: City Distribution Simulation", report 2002.LT.5680]. This model anticipates on time window delivery and congestion information and can be described as the Vehicle Routing and Scheduling Problem with Time Windows Forecasted model (VRP-TW-F). Its improved heuristic gives a better efficiency than the eM-plant model. More than two stores can be delivered in one ride and congestion information is taken into account during the ride planning. Implementing this order planning heuristic will improve the performance of the eM-plant model.

Reports on Transport Engineering and Logistics (in Dutch)
Modified: 2006.08.13; , TU Delft / 3mE / TT / LT.