Planning models for long-haul operations of postal and express shipment companies

https://doi.org/10.1016/S0377-2217(99)00234-9Get rights and content

Abstract

Postal and express shipment companies face the task of providing service between a number of fixed locations. The increasing competition and service orientation of this market sets new standards for planning. One must meet the required service level without letting costs get out of control. This paper identifies important planning tasks which have to be addressed at these companies and defines corresponding optimization models. The relationship between these models and their use within a decision support system are highlighted as well as future research challenges stemming from this application area.

Introduction

The increasing market competition and service focus of the customers forces transportation providers such as postal organizations and express shipment companies to re-evaluate and constantly improve their business processes. In the European postal markets this tendency is further amplified by the current deregulation and privatization efforts of the European Union and governments in the middle and east European countries. The core service provided by postal companies is package and mail delivery. In this market segment it can be observed that one of the key efforts in the last few years has been a reduction in delivery time. This gives postal companies a better strategic position in comparison to express shipment providers who traditionally have guaranteed short-time delivery. One of the main consequences of these efforts is that the available transportation time has been strongly reduced. On the other hand, the number of transportation requests is usually large while the quantity associated with each request tends to be small. The main problem, therefore, is to balance the requirements of short service times with the needs of low cost operations. These criteria are indeed conflicting and make the planning of the companies' logistics operations difficult. Both postal and express shipment companies must manage a large fleet of trucks and (sometimes) aircraft to provide such services. The problems encountered are of the less-than-truckload (LTL) type, which means that it is usually possible and – for economic reasons – necessary to transport a large number of transportation requests in one truck or aircraft simultaneously. One must, therefore, consolidate requests at various levels of the system. Indeed, consolidation is the main instrument of cost reduction in the transportation process. The potential to do this is, however, strongly reduced by the tight constraints on transportation time.

The growing complexity of these processes suggests the use of modern computer-based software applications in order to assist the planner. Such systems are well known in the Operations Research domain and usually referred to as decision support systems (DSSs). A DSS called ISLT (information system for letter transportation) has been developed in the past few years for the German post, the Deutsche Post AG. It supports manual planning by providing answers to `what if' type of queries, has modules which support replanning, and implements model-based optimization algorithms, which generate solutions with the help of optimization techniques. This paper focuses on the decision models which are used in the system. A detailed description of the architecture, interfaces, and functionality will be given in a forthcoming paper.

In this paper we consider letter mail as an example for postal services. We now shortly introduce how letter mail flows through the different stages of the Deutsche Post AG's transportation chain. The postal transportation process basically consists of the following operations:

  • 1.

    Mail collection is the process of collecting mail from customers and mail boxes and bringing it to a depot (letter mail center, LMC).

  • 2.

    During input sorting all mail is automatically sorted and coded with a barcode according to its destination LMC. In Germany the LMC code is identical to the first two digits of the zip code.

  • 3.

    Global area transportation organizes the movement of mail from the origin LMC to the destination LMC. Transportation is by car, truck, aircraft, or railway (although the German post currently does not use railway).

  • 4.

    Output sorting is concerned with the assignment of the mail at the destination LMC to the corresponding local area. In Germany the local area corresponds to the last three digits of the zip code.

  • 5.

    Mail distribution assures that every letter eventually is delivered to the given address by the postman. The letters are usually carried to intermediate delivery bases (DBs) where they are sorted by the postman according to the route before delivery.

The Deutsche Post AG guarantees next day delivery of all letter mail. This implies that the entire process described above must be performed in less than 24 hours on a regular basis. It can easily be recognized that many well-known logistics problems occur in the planning of the transportation network. Vehicle routing problems [31], location routing problems [32], [33], network design problems [2], arc routing problems [19], [20], crew assignment problems [21], etc. have to be solved at the different stages of this process. Fig. 1 shows the organization of the processes and their relationship. The network drawn at one of the distribution bases (DBs) corresponds to the street network which has to be served by one or several postmen. The reader should note that the operations in the global area network are much more complex than indicated in the figure. A more detailed description is given below.

The focus of this paper is on tactical problems in the global area transportation network (GATN). This comprises the transportation between different depots (LMCs) and corresponds to the third operation mentioned above. We will not address strategic and operational aspects. Strategical aspects are, for example, the type of service which should be offered to the customers, the configuration of the transportation network, the location of facilities, etc. Operational issues concentrate on the short-term issues such as assignment of crews to operations and vehicles to tours. We do take into account operational constraints in the tactical plans. These include the availability of resources (such as aircraft and sorting capacities) and particularly the temporal constraints involving driving, flying, and sorting times as well as the routing of drivers.

The aim of this paper is to introduce models which cover different aspects of GATN design and to discuss their relationship to models which have been presented in the literature. This enables us to highlight similarities and differences between postal/package delivery applications and the related fields of railway and motor carrier applications. We also hope to encourage research on some of these problems, which are important generalizations of well-studied problems and should have applications in a wide range of industries.

This paper is organized as follows. The next section introduces the GATN of the Deutsche Post AG. It is similar to the long-haul network used by other postal organizations and package delivery companies. Different aspects of the network and constraints on the operations of the network are discussed. Section 3 introduces models of important subproblems which can be identified in the network. Their relationship to well-known models from the area of transportation research is discussed. The integration of the different models in a DSS is outlined in Section 4. Model requirements which stem from the practical use of the DSS are also discussed. Section 5 examines possible model extensions, especially the use of soft modeling in order to increase the flexibility with respect to data variations. Some final conclusions are given in Section 6.

Section snippets

Global area transportation network

In the GATN one faces the problem of transporting a large number of requests (mail) from origins to destinations by a fleet of heterogeneous vehicles (including aircraft). We assume that the configuration of the transportation network is given (by a prior strategic decision) and concentrate on the transportation between terminals (LMCs). The collected mail with a given quantity and a earliest availability at its origin and latest arrival time at its destination defines the transportation

Planning models

This section introduces the major GATN planning problems which have to be addressed by postal organizations. This means the organization of all necessary activities of transportation and handling. It was shown above that planning comprises the definition of vehicle, driver and request routes as well as the allocation of capacities for sorting, loading and unloading activities. It is clearly beyond the scope of current optimization techniques to handle integrated models for the entire GATN of

Model integration and DSS requirements

We now address the issue of the relationship between the different types of models introduced above and how their solutions can be used during the process of constructing a feasible plan. The approach we have chosen is essentially a decomposition of the problems based on different criteria. The first criterion is the mode of transportation, i.e. air and ground. Air and ground problems are rather different both from a planning and implementation side. First, ground transportation is provided by

Model extensions

The models we have discussed so far are deterministic. We have, for example, assumed that demands, capacities, costs, etc. are known perfectly well. This is indeed a very unrealistic assumption. We may, therefore, introduce new flights or vehicle routes in order to ship a few requests with a small quantity. Whether this is necessary in practice or not is not known at the current stage of planning. The same applies for time windows and sorting capacities. A convenient way of relaxing the strict

Conclusions

The focus of this paper has been on models for long-haul transportation in postal and package delivery systems. We have shown how such models can be constructed and how they relate to frequently addressed models in the literature. It can be concluded that new algorithmic developments are necessary in order to solve most of the models we have presented. We have developed algorithms for some of the models above but a significant body of work remains to be done. Further developments are necessary

Acknowledgements

The authors wish to thank K. Büdenbender, C. Kranz, and S. Irnich who read earlier versions of this paper. Their helpful suggestions improved the presentation significantly. We are also indebted to the Deutsche Post AG, and especially to the director of the transportation department, Mr. J. Weith, for the opportunity to engage in this challenging project.

References (41)

  • C. Barnhart et al.

    Air network design for express shipment service

    Operations Research

    (1996)
  • L.J.J. van der Bruggen et al.

    Variable-depth search for the single-vehicle pickup and delivery problem with time windows

    Transportation Science

    (1993)
  • K. Büdenbender, T. Grünert, H.-J. Sebastian, A tabu search algorithm for the direct flight network design problem,...
  • CPLEX, Using the CPLEX Callable Library, version 5.0, Technical Report, CPLEX, Division of ILOG,...
  • T. Crainic, A. Frangioni, B. Gendron, Bundle-based relaxation methods for multicommodity capacitated network design,...
  • T.G. Crainic, Service network design in freight transportation, European Journal of Operaions Research 122 (2000) this...
  • T.G. Crainic et al.

    Design of regular intercity driver routes for the LTL motor carrier industry

    Transportation Science

    (1992)
  • G.B. Dantzig et al.

    Decomposition principle for linear programs

    Operations Research

    (1960)
  • P.J. Dejax et al.

    A review of empty flow and fleet management models in freight transportation

    Transportation Science

    (1987)
  • J. Desrosiers, Y. Dumas, M.M. Solomon, F. Soumis, Time constrained routing and scheduling, in: M.O. Ball, T.L....
  • Cited by (77)

    • Parcel delivery using the hyperconnected service network

      2022, Transportation Research Part E: Logistics and Transportation Review
      Citation Excerpt :

      In the classic PDP literature, each parcel is carried by the same vehicle from its origin to its destination, and therefore, consolidation of shipment is hard to achieve. Grunet and Sebastian (2000) extended the general pickup and delivery framework by allowing intermediate storage of items at the locations of customers. Kerivin et al. (2008) studied a simplified version of this model and proposed a mixed-integer linear programming (MILP) formulation and a branch-and-cut solution method.

    • Hub network design problem with profit optimization for time-definite LTL freight transportation

      2018, Transportation Research Part E: Logistics and Transportation Review
    • Optimization of postal express line network under mixed driving pattern of trucks

      2015, Transportation Research Part E: Logistics and Transportation Review
    View all citing articles on Scopus
    View full text