Production, maintenance, materials coordination and transportation is concerned with the design, analysis and control of production units and the coordination of operational units across supply chains (networks). Operational units consist of an organized set of resources in a workspace. These resources carry out processes to realize work orders for products or services. Examples of such production units are assembly lines, work cells, internal warehouse operations and maintenance departments. Production management concerns the management of operational processes such that objectives with respect to flow time and due date reliability of work orders and with respect to resource utilization are achieved. Maintenance management concerns the management of maintenance processes such that objectives with respect to availability of technical systems are achieved under economic constraints.
In a chain or network setting, we consider the actual transformation activities within a unit in an abstract form, reflecting the fact that coordination of such activities is typically the responsibility of some hierarchically higher authority, or such coordination is only implicitly enforced through e.g. contracts on delivery performance from one unit to the other. Main issues in such networks are the coordinated management and control of procurement, inventories and transportation activities at multiple levels from external suppliers to external customers that have to be served conform a set of delivery requirements (time windows, service levels). There is of course a strong link to decisions with respect to resource usage in operational units within the supply chain, since release of work orders is only possible if materials as required by the bill of materials of the product (to be delivered by suppliers) are indeed available.
As for the control of work order flows within units, various resource planning levels can be distinguished:
In (3) the service level in the sense of due date performance of the production unit towards its clients is realized as a non-trivial result of the planning and scheduling system. As an aspect of (3) saving set-up time by adequate sequencing of jobs on a resource is an important issue. Another characterization of production units is the uncertainty about the activities involved in the execution of the work orders at acceptance. In discrete manufacturing one distinguishes between (a) make-to-order, if design of the product and the structure of the required resources are known, and (b) engineer-to-order, if the (partial) design is part of the order. In the case of maintenance management, preventive maintenance typically leads to orders with a pre-given structure as in make-to-order and corrective maintenance is structured upon diagnosis during execution, which is an extreme analogue of engineer-to-order.
Regarding supply chain coordination across units, typical research questions are: Where should we position decoupling stocks and how should these stocks be composed? Which frequency of information exchange is required and at which frequency should materials be shipped from one stock point to another? Which allocation rules can be applied in conjunction to echelon-order-up-to policies to yield near-optimal results? How can shipments of materials be organized efficiently in terms of utilization of vehicles, transport kilometers and effectively in terms of delivery within agreed time windows? How can reverse logistics and the management of recovered components, sub-assemblies be organized in relation to new ones? How can public sector procurement be best organized?
Maintenance management has an intriguing relationship to production management. The maintenance-production interface requires chain coordination with its own characteristics. Key objective is to determine preventive maintenance policies so that the system availability in the long run is optimized at low costs. Furthermore, maintenance gives rise to its own typical inventory problems concerning spare parts and service tools, but this issue will be addressed under the program 'Materials coordination and transportation'. A new topic within maintenance management is the effect of decisions in the design phase of capital goods on Total Cost of Ownership (TCO) and availability.
prof.dr. A.G. de Kok, dr. M.C. van der Heijden
Senior researchers involved
Dr.ir. I.J.B.F. Adan, prof.dr.ir. J.W.M. Bertrand, prof.dr. R.J. Boucherie, dr.ir. R.A.C.M. Broekmeulen, dr. M. Cloodt, dr.ir. N.P. Dellaert, dr. K.H. van Donselaar, ir.dr. S.D.P. Flapper, prof.dr.ir. J.C. Fransoo, dr.ir. E.W. Hans, prof. dr. J van Hillegersberg, prof. dr. ir. G.J.J.A.N. van Houtum, dr. G.P. Kiesmüller, dr.ir. M.R.K. Mes, prof.dr. M.J. Newby, dr.ir. H.P.G. van Ooijen, dr.ir. F. Schotanus, dr.ir. J.M.J. Schutten, dr. P.C. Schuur, dr. M. Slikker, dr. T. Tan, prof.dr. J. Telgen, dr.ir. W. van der Valk, prof.dr.ir. J. van der Wal, prof.dr. A.J. van Weele, dr. T. van Woensel.
For production planning in job shop-like production systems (rough cut capacity planning, scheduling) we analyzed the DSS for multi-level, multi-project planning under multi-resource constraints under the uncertainties inherent to engineer-to-order and project environments. We developed various capacity planning techniques that take advantage of special characteristics of a planning situation. For project environments, we have developed a project scheduling approach with deadlines including costs for outsourcing and overtime in the optimization process.
Furthermore, we have developed new semi-Markov decision models for Order Acceptance (OA) under uncertainty, a methodology for tuning the training parameters to the associated Reinforcement Learning (RL) process and a general framework for generating new heuristic rules for decision problems under uncertainty by using RL.
For batch-process industries we have developed regression based models that can reliably predict the makespan of a set of work orders from job and job-set characteristics. For multi-stage production systems (two consecutive production units) we have investigated the effect of dynamic lead time setting on the delivery reliability and found that under stationary conditions, lead times should be static and lead time updating is detrimental to performance; for the non-stationary cases we have developed lead time updating procedures using clearing functions.
For contract manufacturing in the pharmaceutical industry, we have developed order release mechanisms which can handle the limited availability of information. We have investigated the use of contingent labor for make-to-stock, and make-to-order production, and for service industries.
For integrated production-inventory systems with job shop type production, we have developed order batching and ordering policies that minimize system wide costs under a service constraint, and have shown that under strict conditions regarding the shop load, and product set, cyclical production policies can outperform non-cyclical policies. The interest in cyclical policies led to quantitative modeling research in collaboration with researchers from the QMAOP research program. The results of this research are reported there.
Within warehouse management, we worked on Systematic Warehouse Planning. We have developed a new strategy for the assignment of SKUs to warehouse locations, called Order Oriented Slotting (OOS). Our approach distinguishes itself from standard slotting strategies which store the items based on how frequently the item is picked without regard to the picking sequence. We developed several effective heuristics for this.
Within maintenance management, we focused on influencing TCO and availability in the design phase of capital goods. We developed new models and methods for Level-Of-Repair Analysis (choice of reparability of modules and components and the location of maintenance, taking into account the costs of the resources needed).
In the period 2002-2007 research in this area focused on the topics Service Supply Chains for Capital Goods and Supply Chain Management. Research on Service Supply Chains for Capital Goods was led by prof.dr.ir. G.J. van Houtum at TU/e and dr. M.C. van der Heijden at UT. The research focused on optimization of spare parts networks with the objective of minimizing capital investments subject to system uptime constraints. Examples of such systems are wafer steppers and medical equipment.
Various aspects of such systems and the associated spare parts networks have been taken into account, such as component commonality, lateral transshipments and multiple demand classes. A fundamental contribution in this area consists of modeling and analyzing the interaction of spare part inventory management, repair capacity and preventive maintenance policies. The research is also strongly related to the program 'production and maintenance management'. We developed models and tools to make trade-offs between the use of additional capacity in repair shops and additional spare part inventories. An interesting result is that major reductions in spare part inventories can be achieved if we use appropriate priority settings for repair of certain spare part types (particularly with high value-to-repair time ratios) in the repair shops. Furthermore, research has started on the logistics optimization of service contracts.
Research on Supply Chain Management focused on the development of operational control concepts for various so-called Bill Of Material (BOM) structures in the supply chain. Existing results from literature on serial structures were extended to situations with periodic batching and situations with discrete demand. For divergent structures it was shown that optimal policies are computationally intractable and existing heuristics may result in inferior solutions, thereby identifying the need for research on effective heuristics for controlling divergent and general supply chain structures. Most of this work is contained in Dogru (2006).
A handbook on Supply Chain Management was edited with contributions by leading researchers. Based on earlier work on the comparison of base stock control policies and LP-based rolling scheduling heuristics hybrid methods, i.e. models and methods combining both approaches, have been developed and shown to be very effective (Spitter (2006)). Furthermore we have done research on the interaction between control of supply chains by shippers and control of transportation networks by logistics service providers (Smits (2005)). Finally extensive research is executed on multi-level lot sizing problems, where the focus is on determining the lot sizes of multiple items that are interacting in a supply chain. New modeling frameworks incorporating forecast uncertainty led to new insights on the impact of demand uncertainty on lot-sizing. All of this research led to numerous papers in ISI journals.
Research on retail operations led to empirical insights into demand substitution, differences between management of perishable and non-perishable items, causes of errors in inventory records and potential directions for improvement. An important finding was the fact that in most supermarket stores the focus in a frequent replenishment, leading to excessive handling costs and unnecessary stockouts. New control policies have been proposed that increase product availability on the shelf, while overall costs are substantially decreased. Based on a blueprint for the control of the retail chain, it can be shown that with this new approach supply chain-wide inventory capital investments are reduced.
Research in the area of transportation has focused multi-agent systems for decentralized control of real-time, multi-actor transportation networks using auctioning mechanisms. We found that such multi-agent systems are very flexible in dynamic environments with significant uncertainties and that their performance under these conditions is at least as good as traditional OR heuristics. New research has started on vehicle routing under traffic congestion. We developed new formulations for problems with time-dependent and stochastic traveling times between two locations, which in turn led to formulations of new vehicle routing problems. In particular, formulations where the characteristics of the travel times are derived from queuing theory have led to robust solutions that can be efficiently computed.
Furthermore, we have made progress in the application of Operations Research techniques in purchasing decision-making with a focus on tendering procedures. Major results concern the optimal number of bidders to invite into a tender, the choice between the open and the restricted procedure in public tendering, the distribution of gains among the members of a purchasing consortium, the optimal composition of a purchasing consortium and the analysis of legally acceptable ways to let a preferred supplier win.
National and international cooperation
The universities of Eindhoven and Twente collaborate within the joint IOP-IPCR project 'life cycle oriented design of capital goods'. Furthermore, a structural collaboration in the area of service logistics between the universities in Twente, Eindhoven and Rotterdam has been set-up via the Service Logistics Forum Research program (see also 'Application of research and collaboration with industry'). Dr. M.C. van der Heijden and prof.dr. A. van Harten have collaborated with the research institute TNO-FEL in the area of service logistics.
In the area of transportation, dr. M.C. van der Heijden, dr. P.C. Schuur, prof.dr. J. Van Hillegersberg and prof.dr. A. van Harten have been strongly involved in a collaboration with TU Delft and Erasmus University Rotterdam in the context of the TRANSUMO-DIPLOMA and TRANSUMO-PILOT projects. This has resulted in the PhD thesis of dr. Mes and a number of publications. Also, the universities of Eindhoven, Twente and Rotterdam collaborate in the project TRANSUMO-Chain synchronization in the area of vehicle routing under congestion.
In the area of purchasing, dr.ir. F. Schotanus and prof.dr. J. Telgen participated in a project led by the Maastricht School of Management on capacity building in public procurement in Uganda. As part of that project five PhD theses are currently being prepared.
Research in this domain on retail operations is partially based on empirical data gathered from the stores of companies like the Van Eerd Groep, Schuitema and Laurus. We participate in the Consortium for Operational Excellence in Retailing
Application of research and collaboration with industry
Within the IOP-IPCR project on 'life cycle oriented design of capital goods', we cooperate with Philips Healthcare, Thales Nederland, PANalytical, and Vanderlande Industries. This project is funded by SenterNovem. For a manufacturer of kitchens (De Beek), a dedicated scheduling approach has been developed. This project has been sponsored by SenterNovem.
Our methods for multi-project planning and scheduling under uncertainty have been applied at several ship repair yards in the Netherlands, within the externally funded REPRO project.
The potential of the order oriented slotting strategy in warehousing was shown by application to a case at a leading bookseller’s warehouse (Noordhoff).
In a project on outsourced production, prof.dr.ir. J.C. Fransoo closely cooperates with ScheringPlough, who funds a substantial part of the project.
In 1996 the European Supply Chain Forum (ESCF) was established. ESCF collaborates closely with Stanford University and Hong Kong University of Science and Technology. Currently about 25 global companies are members of the forum. There are extensive research contacts with a number of these companies, allowing for empirical research that enables validation of model-based research and development of new models.
In 2005, the Service Logistics Forum Research program was established. Currently, 14 companies are involved: ASML, Ceva Logistics, DAF, Fokker Services, IBM, IHC Holland, Nedtrain, Océ Technologies, ORTEC, Philips Medical Systems, Railpro, Stork Food Systems, Thales, and Vanderlande Industries. These companies sponsor application-oriented PhD projects at the universities in Eindhoven, Twente and Rotterdam.
Prof.dr. G.J. van Houtum has been collaborating with ASML since 2001, via PhD, design and master thesis projects.
Within the TRANSUMO research program on sustainable transportation, we closely collaborated with INITI8, Industrial bakery Merba and ORTEC.
PhD Research on Purchasing was sponsored by NEVI (Dutch Purchasing Society).
For the period ahead we plan to further extend our cooperation with industry. We plan to continue our research into the relationship between lead time control and capacity control in operational units. In the processing industry, we will continue the research into outsourced production started up in 2005.
We will extend our research on the effect of design decisions on TCO and availability of capital goods. Further, we aim to study the interaction between maintenance management and the service parts/tools supply chain in more detail.
In the area of warehousing, future research aims at designing the building blocks of Systematic Warehouse Planning, so that it will eventually become a valuable tool to roll-out a warehouse. One of the future research areas will be the application of the Reinforcement Learning meta-heuristic that have been derived for order acceptance in a production environment to fleet management problems, e.g. the allocation of full truckloads under uncertainty.
The future research on supply chain coordination is based on taking a total life cycle costs perspective. For some research questions, it is appropriate to distinguish between the initial supply chain that covers design, manufacturing, and installation of new systems, and the service supply chain that covers maintenance and availability management during the exploitation phase.
For the initial supply chain, the main research questions are as follows:
The research on vehicle routing problems in a stochastic and dynamic setting will be focused on adding the travel time component to vehicle routing problems with time windows, finding robust solutions in a dynamic environment, including stochastic demand functions (e.g. based on the store operations model), etc. Also the interaction and coordination between inventory and routing decisions in order to gain cost savings is a promising research line which will lead to more insights in the retail supply chain. The latter also fits into the general research area of replenishment policies, where resource restrictions (e.g. truck capacity, handling capacity, backrooms, etc.) are considered. Also the use of real-time information (e.g. based on GPS and traffic information), to update schedules when realized is an interesting new venue. We will expand the research on a multi-agent for real-time scheduling of multi-actor transportation systems. These are only a few of the issues enhancing agent learning behavior using e.g. reinforcement learning (see above) and less-than-truck-load scheduling.
In the area of procurement, we will expand on the results obtained so far on public tendering. In a new PhD project we focus on the bundling (or splitting) decision: buying in more or fewer separate lots. In future projects we plan to research buying social services, safeguarding quality delivery after contract closure and optimal contract length. With respect to supply management, i.e. from a purchasing and supply perspective, value-based sourcing, outsourcing and off-shoring are important here. Specific research questions include how to develop a suitable supply base and how to manage these different business partners within an NPD Project, as well as the outsourcing of development and R&D activities. This research establishes a link between the research in this program and the research in the New Product Development Processes program.