FLEXIBLE MAIL PROCESSING|
An Application of Integral Postal Engineering
Technical University Delft
ter verkrijging van de graad van doctor|
aan de Technische Universiteit Delft,
op gezag van de Rector Magnificus Prof. ir. K.F. Wakker,
in het openbaar te verdedigen ten overstaan van een commissie,
door het College van Dekanen aangewezen,
op dinsdag 26 april 1994 te 13.30 uur
Jozef Theodorus Wilhelmus DAMEN|
geboren te Apeldoorn.
|Dit proefschrift is goedgekeurd door de promotor:|
Prof. dr. ir. J.J.M. Evers.
|The Postal Service has been processing mail for centuries, using a large range of techniques. In the last few decades in particular, advanced technology has enabled the Postal Service to cope with the rapidly growing volume of mail. However, despite the sophisticated new systems, the principles underlying the postal process have hardly changed at all. As a mass production process, today's mail processing system has difficulty coping with the rapidly changing market. Flexibility, at present not a characteristic of the postal system, is seen as its major challenge.|
The Postal Service has tackled the problem by creating business units. These largely self-supporting sub-organizations divide the postal organization into smaller parts that concentrate on different market segments. There may be some drawbacks from an operational point of view, however, especially regarding the overlap that may occur, resulting in reduced efficiency.
This study has adopted a different approach to flexibility, namely inherent flexibility, which refers to the large range of possibilities for putting the system into practice. The study can be summarized in three words: invention, innovation, and engineering. Invention is the utilization of information throughout the postal system (assumed to cover a country or other large area) to control all the processing directly. Innovation is the objective of the thesis: to define the functional architecture of a flexible postal system based on inherent flexibility. This leads to integral postal engineering.
The development of the functional architecture follows a three-layered approach: the service layer, network layer and process layer (a lot of relevant background information can be found in Systems Theory). Flexible delivery of a broad range of postal services is the ultimate goal of flexible mail processing. A postal network is defined as being responsible for the overall flexible behaviour of the postal system. Information exchange is crucial. The postal network is viewed as the control layer for the basic processes carrying out the basic functions, which relate directly to the basic principles of mail processing. Six basic functions are defined: transporting, buffering, merging, splitting, packing and unpacking. We believe these functions to be essential, complete and disjunct.
The postal system is essentially a goods-transporting process. We have abstracted "goods" into the concept of the Transport Unit (TU), the passive element that is exchanged by the active elements, the basic processes. An important feature of the TU is its hierarchy. TUs are contained in TUs, making up levels of TUs.
Sorting is important for processing mail but has not been included in the list of basic functions. An analysis of sorting revealed crucial aspects of flexibility, with sorting emerging as the equivalent of the packing function. This led to a new approach called buffer sorting, which contrasts sharply with present-day sorting systems that are inherently inflexible. The new approach made it necessary to define sorting structure explicitly.
Next, we constructed a conceptual model of the flexible postal system. We translated basic functions into basic processes which, to gether with an implementation of the Transport Unit, were included in a discrete computer simulation model that acted as a platform for carrying out experiments. In addition, we set up a data model. Various levels of data were distinguished: global data covering the entire system, process data originating at the processes, and local data relating to item information that tells us the processing the mail items have undergone and the kind of services they require. Performance indicators were defined in conjunction with the data model.
At the process layer, the model consists of chains of basic processes of the various types and TU levels, making up the infrastructure. There is minimal interdependence; interactions are purely bilateral, mainly through the exchange of TUs. For global cooperation, the basic processes rely on the network services provided by the postal network: routing to guide splitting processes dynamically, sorting to deliver the necessary information to the packing processes so that optimal "parent TUs" can be formed, and regulation for merging processes to regulate TU traffic. The network layer is responsible for controlling mail processing in such a way as to meet the service requirements of all mail items.
An experiment compared two different mail processing systems: a model of present-day "classical" systems and a model of a flexible mail processing system. Simulation was used to compare required flexibility with realized flexibility. We selected "just-in-time" as the test of flexibility. Every mail item has to be delivered on time, but the time varies considerably from "super fast" to "very slow". The model of the classical system had no means of responding to the required variability. It simply delivered the mail at its own pace. The model of the flexible system, however, displayed a significant correlation between the required and realized delivery times.
The present aim of postal engineering is to switch from manpower to automation in order to reduce costs. The emphasis is on developing machinery. Operations research techniques are used to optimize system configurations statistically. Configuration and machine-level development are completely separate. The resulting system set up for mass production performs well provided service demands do not vary too greatly.
Achieving good quality in a changing environment with a flexible system requires a different approach to systems development: integral postal engineering. It aims at the coordinated, synchronized development of system and machine levels. System-level integration is achieved by an information network. Machinery has to be developed that "listens" to the information. The use of buffer sorting requires an explicitly selected sorting structure. The system configuration can remain relatively simple but using it dynamically requires more complex and flexible decision-making. The business units concentrate on the commercial side of services, leaving the operations side to the flexible system. Services are translated into information to control the operations.
The results of the experiments show that it is possible to define the functional architecture of a flexible mail processing system that responds sensitively to changing environments. Based on buffer sorting, it integrates in two ways. First, it ensures that information can be used effectively, thus avoiding the inefficiency of island automation through overall system cooperation. Second, it enables integration of physically different mail streams, since buffer sorting applies equally to all TU levels.
The use of relevant and up-to-date information is vital. This requires the availability of wide area communication networks. Information needs to be generated and maintained on a daily basis. Optimizing mail processing will become more complex as an increasing number of variables are introduced.
We have adopted a technical approach. The cost aspect has not yet been addressed: it will form an important part of future research.
- Proofs are subjective.
- As a technical achievement, (tele)communication can be
viewed as data transfer to support information exchange.
It is wrong not to distinguish between data and
- It is a misconception to assume we live in the
information age; we do live in the data
- The economical importance of the speed of communication
(or, the other way around, the disadvantage of delay) can be
found by distinguishing between data and
information in the right way: when data comes
too late, it may have lost its value as a source of
information. Data can be assigned
timedependent information value; data having rapidly
aging information value is transfered profitably by
telecommunication, data that has longer lasting information
value is better transfered by postal mail.
- Suppliers of data should be enforced to include an
expiration date to delivered data.
- The six basic processes of a mail processing system
(transporting, buffering, merging, splitting, packing, and
unpacking) are essential, complete and disjunct (this
- The postal transport problem, in the sense of computing
the optimal route of distributing mail over a network,
is NP-complete. (E.L.M. van Unen, Complexiteit van het
postvervoerprobleem. NT-IR-1083, PTT Research,
Netwerktechnologie, oktober 1991. In Dutch).
- Tracking and tracing as a service to the
public will loose importance as logistics service provisioning
becomes more reliable.
- Being captivated by any subject must be viewed as a form
of deprivation of freedom.
- Compared to answering an affirmative question, answering
a disaffirmative question in the Dutch language is
inconsequent. (Explanation: the question: "didn't he walk?"
is commonly answered by "no" (not "yes"), if he indeed did not
- The functional design of a control system containing a
"central control" is not correct.
Copyright © 2000-2002, Zef Damen, The Netherlands.