Principles of a flexible logistics system
Service Controlled Agile Logistics |
Publications |
|
|
Introduction |
|
Service Controlled Agile Logistics (SCAL) is a new model for a logistics control system, in which the client is leading. Every service requested by any client is translated into an individualised plan of actions carried out by logistics operators, who make resource capacity available on agreement. The way this model came to life has been dominated by a search for real flexibility in logistics. Its history may be of interest for better understanding what SCAL is all about. |
|
TOP |
|
Flexibility of mail processing |
|
Why is the mail processing system so inflexible?
|
|
In the 1980's, almost everyone in industry was involved in studies of flexibility. Flexible Production Automation (FPA), robotics, production cells, avoidance of island automation, Just-In-Time (JIT), kanban. At KPN Research (that time still called Dr. Neher Laboratory), a researchgroup has been active also, searching for possibilities to introduce flexibility in mail processing. It failed, mainly due to a far too wide approach. But a follow up, restricted to only technical flexibility, did succeed. The main question to be answered has been: why is mail processing so rigid, so inflexible? Is this necessary? Is there any other way? For "mass-production" mail processing, it's perfectly fine, but for client controlled, flexible, dynamic logistics that fits into the e-commerce paradigm, it is not.
|
|
Basic principles of flexibility in logistics
|
|
This research succeeded and led to the thesis entitled: "Flexible Mail Processing, An application of Integral Postal Engineering". The main train of thought has been as follows (Systems Theory gives a lot of relevant background information). Every process, whatever flexible or variable it might be, always has a kernel that does not change, being fundamental and characteristic for that process. It makes sense to distinguish between these steady and fundamental parts (or aspects) of a system and all of the rest. Flexibility, in principle, must be searched for within these other parts (or aspects). If it is possible to find the kernel parts, flexibility can be studied right from the basics. For mail processing, 6 basic functions - only! - have been found, sufficient to describe the physical handling of any mail processing system. Later it became evident that this holds for any distribution or express delivery logistics system. There is a strong belief that it even is applicable to production logistics. The basic functions are: transporting and buffering, packing and unpacking, merging and splitting. The abstraction of all goods into one basic concept, the Transport Unit (TU), has been an important factor for being able to describe logistics systems with only 6 basic functions. All goods within the system, treated as passive elements, are covered by the TU, whether from outside, like parcels, letters, packets, or "made" within the system itself, like bundles, trays, bags, containers, up to vans, trucks, trains and airplanes.
|
|
Reason of inflexibility in mail processing
|
|
Accurate modelling and analysis of mail processing by means of the 6 basic functions has led to understand the reasons the system is so rigid. Each basic function has a number of fixed characteristics (closely related to the function itself) and a number of parameters - degrees of freedom - determining the way the function is carried out at any moment. These parameters establish the flexibility available in each function. In current logistics systems, all parameters important for flexibility have been set beforehand, at the time the system was designed. Goodsflows are defined inside logistics processing systems. Each flow corresponds to a single type of service delivered to clients (e.g. 24 hour or 48 hour delivery, with or without insurance, etc.). A separate chain of subprocesses is used for each service type. Subprocesses can be used for more flows, but these must be operated separately, in space or in time. Necessary capacities, technical alternatives, and organisational contexts, are important criteria for the way these chains are implemented. These criteria determine craftmanship and expertise needed to design such systems. And once designed and implemented, these systems hardly change any more. But what is more, a changing "outside world" inevitably necessitates the system to be redesigned and re-implemented. The last 25 years have seen three different postal processing systems in The Netherlands!
|
|
TOP |
|
Principles of a flexible logistics system |
|
Building blocks
|
|
If it is possible to translate basic functions into basic processes, a kind of construction kit for logistics systems can be realised. If, additionally, all degrees of freedom are kept available as long as possible, all ingredients for a flexible logistics system are there. Within the project "FlexPost", that led to the earlier mentioned thesis, precisely this happened. A simulation model and program have been developed for a just-in-time distribution logistics system. The main experiences are as follows. The basis of a flexible logistics system are the basic processes. The infrastructure is defined in a network, which is not rigid and is changeable at any moment. This physical network consists of nodes and connections. The latter are realised by transport processes of course. Here is an example of a small network consisting of 3 main centers each with two pick-up and delivery routes: |
 |
The same network, fully detailed with basic processes looks like this (N- and X-"processes" are entry- and exit-points, where goods-to-be-processed enter and leave the system): |
 |
Multimodality is established by choosing the type of TU's for the various transport processes - for each individual case if necessary. Within the nodes, processing of the TU's takes place, particularly packing, unpacking and splitting (choosing different routes). Important for flexibility is, that variables - parameters, degrees of freedom - are not set beforehand. Hence, the processes don't know ahead what to do, until a TU has arrived to be processed. Data necessary for processing (values of parameters) either have to travel with the TU itself, or have to reach the process by a separate route. If coupled to the TU itself, processes passed by sequentially are difficult to co-ordinate as to yield a good overall service quality for this TU. In this case, decisions tend to be local decisions only. Unless of course the co-ordination is set beforehand. In its most elementary form is this the case with letter mail processing: the address only travels with the TU. For a flexible logistics system, this is the wrong way!
|
|
Network logistics
|
|
Major conclusions from FlexPost for a flexible logistics system have been:
- information processing must be separated from goods handling processes;
- decisions necessary to realise requested services in the most flexible way must be taken in the information domain;
- from this information domain, parameters for individual physical processes are controlled;
- the greater the "granularity" of its processes, the more variably and flexibly the system can be.
This gives rise to a logistics system consisting of two networks: a physical network processing goods, and an information network taking control. Application of such a system is called network logistics. The FlexPost project resulted in a simulation of a just-in-time delivery system according to this principle, but with a dedicated control (information) layer. In SCAL, a generic control model has been defined, consisting of objects (object classes) analoguous to the basic functions of the physical network.
|
|
TOP |
|
Service Controlled Agile Logistics |
|
"Basic functions" in the information domain
|
|
In the year 2000, information network means ICT and Internet. Flexibility means client friendly. And network logistics means independent processes together responsible for delivering service products of high quality. These are the principles of SCAL. When high granularity, and hence independence of processes is a precondition for flexibility, who is going to guarantee quality of service? When flexibility means processes dynamically adapt to changing environments, who is going to stay stable and control this? The answer is simple: the very service itself, and hence, the client. From this idea, the control of a flexible logistics system has been conceived by completely turning upside down the old control concept. The "classical" system emphasises the different types of process chains responsible for the various types of services. Information about the progress of individual goods can be deduced from these chains; compare tracking and tracing. (But not without extra effort! Identity of individual goods are not known in classical process chains: everything in the flow is the same, and can not be distinguished).
SCAL focusses on the individual client and its requested service. The service as product, not as process. Each service can (and will) be defined by a time bound path through the physical network. The product is translated into a sequence of steps with time bounds dependent on the situation of the moment. The translation is done by an "agent", originated - in principle - by the client itself. This agent, the service planner, makes a plan for realising this service for this client. And it controls the progress when effectuated by the processes of the physical network. While building the plan, it visits the processes, resources in SCAL, and negotiates about actions to be carried out. To this end, each resource has a scheduler managing its operations. Accessible on Internet, it is this scheduler that welcomes service planners. While planning, service planners generate and evaluate possible relevant alternatives and choose the best they can find according to some criterion. If in reality things change, a new plan is made taking the actual situation as a starting point. This may happen, if interruptions out-of-control take place making the plan as-is incompatible, or when the planner discovers a better way for service fulfilment, for instance due to introduction of new, additional resources.
|
|
Object model
|
|
Basic to SCAL is an object model. It consists of 10 major classes, in the figure arranged in 3 columns by 4 rows; arrows denote the most important relationships. The right side column represents services, the middle column goods, and the left side column resources. |

|
Client and logistics operator (owner of logistics resources) are at the top level. On the second row, the service requested by the client concerns a TU (or more TU's) to be processed by resources. One row below, the service is effectuated according to a service plan holding all operations necessary for its completion. An operation is a logistic activity carried out on one TU by one resource. For each resource, all operations are dealt with in a resource schedule. Hence, every operation is referenced twice, once in a plan and once in a schedule. A plan is made by a service planner, the "agent" as mentioned above. In order to make a plan, the planner needs to have access to a "world model" containing up-to-date knowledge about the logistics domain, specifically about the physical network, what resources are present and where, and how these are accessible through the information network (Internet). This world model can be regarded as part of the service planner (and is not shown in the figure separately). At the bottom, the negotiation about operations to be carried out is shown between the service planner and the resource scheduler of all resources relevant for the plan to be made. Resource schedulers make the schedule, and maintain it.
|
|
INImini
|
|
INImini is a mini simulation system (INI stands for ILSP and IT: Integrated Logistics Service Providing and Information Technology), built to examine the feasibility of the model. Express parcel services are simulated with a small network consisting of 5 nodes, and later a larger one representing a European road network. Simulations have shown that the model forms a suitable basis for flexible logistics systems. Flexibility is tested by having road trucks delayed such that services can not be fulfilled as planned. Replanning is established for all parcels involved. Furthermore, available resources are better utilised than in a "classical" system, especially when the whole system is heavily loaded. Simulations also offered experience in how service planners have to make plans.
|
|
What next?
|
|
According to GLORI (Global Logistics Research Initiative, a world-wide co-operation between a number of universities and industry in a virtual organisation), an agile logistics system should have the following characteristics:
- it should be distributed;
- it should be scalable up or down in a linear fashion;
- it should consist of highly autonomous players;
- it should operate as a virtual organisation.
SCAL does have all these characteristics. A logistics system is distributed by nature. The object based model emphasises this again; control by objects is very well suited for distributed implementation. Mutual dependence between resources have been reduced to an absolute minimum. Therefore, they operate autonomously and that makes the system linearly scalable. Organisation of chains of processes has been reduced to activities of schedulers of independent resources and negotiations with "traveling agents" of services. Processes are loosely coupled by the red threads of services and in these volatile relationships, no one is principal nor main contractor. Physical (geographic) positions are of course important for resources, but not for the control elements (resource schedulers and service planners). This creates a highly virtual organisation, in which the only dependence arises between services making use of the same resource simultaneously. "Intelligent" behaviour of service planners will decisively determine the overall behaviour of the system. This will be the main topic of future research.
|
|
TOP |
|
Publications |
|
Service Controlled Agile Logistics
|
|
© April 2000, dr. ir. J.T.W.Damen
KPN Research Logistics, P.O. Box 421
2260 AK Leidschendam
|
|
Published in
Logistics Information Management,
MCB Press, UK, Vol. 14, nr. 3, 2001 |
|
Abstract |
Current logistics systems are unable to react fast enough to rapidly changing environments, mainly because they are heavily focused on the goods handling processes. To compensate for this lack of flexibility, logistics services should be carried out by independently controlled logistics resources. But these resources together do have to guarantee the overall quality of services. Service Controlled Agile Logistics solves the conflict that arises - independent resources working together - by strictly distinguishing between control and handling. It is based on control of logistics processes by the requested services themselves, which create their own "agents", made responsible for realizing the service in the best possible way under changing circumstances. In order to examine the feasibility of this approach, a simulation program has been developed, and some preliminary results are presented.
|
Keywords |
dynamic logistics control, agile logistics, object oriented model, logistics services, distribution logistics. |
|
TOP |
|
Customer controlled e-logistics
|
|
Dr J.T.W. (Zef) Damen
KPN Research
P.O. Box 421
2260 AK Leidschendam
The Netherlands
|
|
Presented at
E 2000 E-business and E-work Conference and
Exhibition,
18-20 October 2000,
Madrid, Spain
|
|
Abstract |
The lack of adequate flexibility in logistics is a major obstacle for e-commerce. Logistics service providers can not cope sufficiently with demands for services that fit within the e-commerce paradigm. E-logistics, the logistics part of e-commerce, has to deal with: individualized treatment; full customer control; virtual, decentralized and dynamic environment. In general, logistics systems are mainly based on fixed chains of processes of which only capacity can vary - within limits. Logistics services are realized by choosing from predefined chains and predefined time windows. This is the main reason logistics service providers are not prepared for e-logistics.
This paper presents a new model for logistics control: Service Controlled Agile Logistics (SCAL). It is aimed at: personalized control; dynamic planning of services; dynamic scheduling of logistics activities; flexible reaction upon unpredictable changes. It treats requested services as individual entities. The realization is controlled by a dedicated "agent" travelling over Internet in search for resources able to fulfil (parts of) the service requirements. This agent negotiates with relevant resources and plans the necessary activities. Replanning is established whenever activities deviate from what is planned, or when changing circumstances urge to do so. Thus, all resources available are used in the best possible way. In the object-oriented model on which SCAL is based, the demanders side of service requesting clients is perfectly balanced with the providers side of resource exploiting logistics operators. This mirrors precisely the interdependency of both more or less independent "worlds". The feasibility of this approach has been examined in a simulation of a European wide express network. The paper describes the SCAL model and the background from which it is derived. It briefly shows the simulation and some preliminary results, and puts the SCAL model into a broader perspective for future development.
|
|
TOP |
|
Customer controlled logistics
|
|
Zef Damen (KPN Research)
|
|
Presented at
IPA Fall Days 2000, 20-24 November 2000,
Millingen a/d Rijn, The Netherlands
|
|
Abstract |
E-commerce has called for a greater flexibility in logistics, especially in the delivery of ordered goods. The lack of adequate flexibility in logistics is a major obstacle; logistics service providers can not cope sufficiently with demands for services that fit within the e-commerce paradigm. Up to now, logistics systems were mainly based on fixed chains of processes of which only capacity could vary within limits. Logistics services were realized by choosing from predefined chains and predefined time windows. This is the main reason logistics service providers are not prepared for demands from e-commerce: individualized treatment; full customer control; virtual, decentralized and dynamic environment.
Customer controlled logistics aims at: personalized control; dynamic planning of services; dynamic scheduling of logistics activities; flexible reaction upon unpredictable changes. It is based on a new model for logistics control: Service Controlled Agile Logistics. SCAL treats requested services as individual entities. The realization is controlled by a dedicated agent travelling over Internet in search for resources able to fulfil (parts of) the service requirements. This agent negotiates with relevant resources and plans the necessary activities. When the requested service is actually carried out, it is the same planning agent that escorts the it. Replanning is established whenever activities deviate from what is planned, or when changing circumstances urge to do so. Thus, all available resources are used in the best possible way.
In the SCAL model, the demanders side of service requesting clients is perfectly balanced with the providers side of resource exploiting logistics operators. This mirrors precisely the interdependency of both more or less independent "worlds". The feasibility of this approach has been examined in a number of simulations. The presentation zooms in on the background of the SCAL approach and examines the characteristics of the planning agent in some more detail.
|
|
TOP |
|
Copyright © 2000-2002, Zef Damen, The Netherlands.
|