Real-life vehicle routing problems usually cannot be precisely represented by an analytical model which is required for computer-based decision support. Even if the analytical model is of high quality, discrepancies between model representation and real-life problem arise as a result of the sheer cost of getting information into the computer. Telematics systems can be used to improve the timely availability of information regarding the actual transportation processes. Electronic Data Interchange (EDI) can be used to integrate information systems of shippers, e.g. to obtain all relevant data regarding transportation requests and customer locations. Despite of the improved possibilities of getting the data into the model, the information is generally not only incomplete but also imprecise. A shipper, for example, may ask that a shipment be picked up in the morning before noon, when his dock is not as busy. In the model such restrictions are usually treated as time window constraints. The computer system has no way of interpreting whether such a request for early pickup is a hard constraint or whether the shipper was only trying to express a preference.
The impreciseness of the model representation results in two fundamental problems:
Although these problems occur in static as well as in dynamic planning, the impact is quite different. In static planning there is usually more time for the collection of data, resulting in a more accurate representation of the real-life problem. Furthermore, there is more time to manually verify and validate a solution recommended by the planning system.