FEAT - Fair and Efficient Allocatin of Transportation

The main aim of this project is to improve the efficiency of the allocation of transport activities to carriers by re-allocating transportation orders between carriers. This would allow carriers to develop more efficient plans, which in particular avoid empty trips, thus improving efficiency of the overall system. Such a re-allocation requires collaboration among carriers. Existing research on carrier collaboration has proposed various mechanisms for collaborative planning, which have been studied in the context of different transportation problems. The present project will systematically compare a spectrum of different approaches to carrier collaboration, in particular a fully centralized planning approach, a centralized auction mechanism, decentralized auctions and bilateral exchange between carriers. These different collaboration mechanisms will be studied in the unified setting of a hub-and-spoke transportation scenario, in which less than truckload transportation orders are collected via local tours to a hub, transported to another hub on a long-distance leg, and from there are distributed to customers via local tours. This problem setting on the one hand represents a realistic scenario of the transportation industry. On the other hand, it involves less-than- truckload tours, which so far have only rarely been considered in the context of carrier collaboration, thereby providing another innovative aspect of the current project. Furthermore, the project will take into account that carriers (as well as possible centralized institutions) have only limited information about the cost structure and other relevant information of other carriers by utilizing concepts from decision making under incomplete information. The present project will require fundamental research at two levels: Firstly, collaboration mechanisms need to be designed and the optimal behavior of carriers within each collaboration mechanism needs to be modeled. This will lead to specific problems in the context of centralized mechanisms, where also aspects of fairness need to be taken into account. Secondly, these models require particularly fast and efficient solution algorithms for the underlying transportation problem (both the local vehicle routing problem and the planning of long-range transportation). Development of efficient algorithms for the repeated solution of these problems will therefore be another focus of the project.

A considerable part of the road transport capacity in countries such as Austria is wasted by empty trips that are e.g. necessary for trucks to return to their home base after performing a transport task. Rather than driving empty, these trucks could possibly execute transport requests that are now performed by other carriers (who then might have to bring back their trucks empty from a location close to the depot of another carrier). The main aim of this project was to increase the overall efficiency of the transport systems, and thus decrease negative effects such as pollution and road congestion, by developing mechanisms that allow carriers to exchange transport requests among each other, leading to a more efficient allocation of requests to carriers. We studied this problem in a specific, realistic setting in which goods to be shipped (such as parcels) are collected within one region, shipments are consolidated at a depot in one region, then goods are transported via some long-haul transport to another region, where they are distributed to their final destinations from another depot. Such multi-modal scenarios are quite common in the transportation industry, but also create additional difficulties. Since tours must be planned in two different regions, it is no longer so clear which carrier is best suited to execute a given task. This multimodal perspective already makes the planning for one carrier difficult. In the project, we developed specific methods to solve the multi-region problem efficiently. Our computational results indicate that the new algorithms are significantly faster than previous approaches. These algorithms were initially developed for the planning problem of one single carrier. In a second part of the project, they were extended to the even larger and more complex problem of a central institution that allocates requests across multiple carriers. Although we do not view such a centralized planning as a viable system in practice, knowing the optimal solution from a system-wide perspective is important to evaluate the quality of solutions found via decentralized approaches. In the last part of the project we studied decentralized approaches such as auctions, performed either jointly by an association of carriers or by each carrier individually, and a posted price mechanism. Our results show that even without creating any centralized institution and the overhead involved in it, already decentralized auctions undertaken by each carrier to redistribute transportation requests that do not fit well with other requests handled by this carrier, can increase the efficiency of the overall system considerably.