Production scheduling with uncertain demand and no
stockkeeping
Citation for published version (APA):
Dellaert, N. P., & Wessels, J. (1985). Production scheduling with uncertain demand and no stockkeeping. (Memorandum COSOR; Vol. 8522). Technische Hogeschool Eindhoven.
Document status and date: Published: 01/01/1985
Document Version:
Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)
Please check the document version of this publication:
• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page numbers.
Link to publication
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal.
If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:
www.tue.nl/taverne
Take down policy
If you believe that this document breaches copyright please contact us at:
openaccess@tue.nl
providing details and we will investigate your claim.
EINDHOVEN UNIVERSITY OF TECHNOLOGY
Department of Mathematics and Computing Science
r
Memorandum COS OR 85-22
Production scheduling with uncertain
demand and no stockkeeping
by
Nico Dellaert and Jaap Wessels
Eindhoven, The Netherlands
November 1985
PRODUCTION SCHEDULING WITH UNCERTAIN DEMAND AND NO STOCKKEEPING
Nico Dellaert, Eindhoven Jaap \'lessels, Eindhoven
Abstract. We consider a production process, in which one part forms the bottle-neck. For this situation, in which there is great uncertainty about future demands and no products are kept in stock, we describe a Inethod for production scheduling and pre-sent some results. The most important feature of the method is that some orders are scheduled on arriving and other orders, with a greater agreed due-data unreliability, are scheduled at a later moment.
Zusammenfassung. Wir betrachten ein Produktionsprozess mit einem Engpass. Fur diese Situation, wo Nachfrage sehr unsicher ist und Produktion nur im Auf trag statt findet, beschreiben wir eine Methode zur Planung der Produktion und wir prasentieren einige Resultate. Das wichtigste Aspekt der Methode ist dass ein Teil der Auftrage gleich beim Eintreffen eingeplant wird und dass Auftrage, wofur die Liefergenauigkeit ge-ringer ist, erst spater eingeplant werden.
1. Introduction
In this paper we will consider the problem of scheduling production for a production process with large set-up times, a very uncertain demand process, no possibility of
stockkeeping, whereas i t is requested to deliver on short notice. This type of situ-ation occurs in several instances and is not covered by the common planning methods. It became problematic in the last few years because of the trend of manufacturers to require very short delivery times from their suppliers. Using an example of such a situation, we will demonstrate an approach for such problems.
In a factory one produces steel pipes to order; for several reasons there are no products kept in stock. When an order arrives, a delivery date has to be agreed upon. In order to realize this delivery date as good as pOSSible, a good
production-schedule is needed.
The production process can be s'plit up into three separate parts: a preliminary treatment, the welding-process and a follow-up treatment. The welding-process is forming the bottle-neck of the production. There are several welding-machines, each with a partially different range of sizes, i t can treat. If we want to make a pro-duct of another size, parts of the welding-machine have to be rebuilt. These
rebuild 2 rebuild
-times depend on the change of size.
In order to keep the total amount of time, needed for rebuilding, small, we divide the products into several groups -of si:<:es, between which the rebuild-times are small. These groups play an important role in the production scheduling. Other important aspects of a good production schedule are the necessity of keeping the delivery dates small and rather constant.
According to the wishes of the factory management we divide the orders into 4 cate-gories, varying from very urgent orders to orders with a requested delivery-date after a few months. For every category of orders the management has given constraints on the average delivery time and the allowed uncertainty in realizing this time. In principle one would like to collect a large set of orders before scheduling them. However, this approach would lead to. long delivery times. Therefore, we designed a method in which orders with high priority are scheduled immediately, whereas the orders with low priority are put into a waiting room. The problem then remains to decide which orders are scheduled immediately and for which orders scheduling is postponed. It will appear in our example that orders of Category 1 and most of the orders of Category 2 can be scheduled upon arrival.
This paper is organized as follows. In Section 2 we describe the structure of the production process, the available welding-capacity and the products that are made. In Section 3 the current production-scheduling method is described, which uses a fixed-cycle schedule. The demands upon the planning method are presented in Section 4. The new production-schedule, with its most important elements, is described in Section 5. Some numerical results are given in Section 6. Conclusions are drawn in Section 7.
2. The production process
In one part of the factory steel coils of different quality and size are stocked. These coils are slitted into rings. After the slitting the rings are transported to the welding-machines, where they are rolled, welded, calibrated and sawed. After-wards several follow-up treatlnents can be done, such as sawing to little pieces, coating, heating and collecting.
For all treatments, except on the welding-n~chines, there is sufficient capacity, so most of the time orders have no substantial waiting-time due to these treatments. The waiting-time caused by the welding-machines can be several days or even several weeks. For this reason we only consider the scheduling on the welding-machines.
In the factory we studied, there were 6 welding-machines, each with a partially different range of sizes, depending on the available sets of rolls on the machine. With every set a group of products, with certain shapes, can be produced. If we want to make a product of another group we have to take another set, if we want to make a product with another shape belonging to the same group, we only have to change a
3
-part of the set. It is understandable that we want to produce products of the same group on a machine as long as possible in order to avoid rebuilding-times.
On the 6 welding-machines products of about 150 different shapes, length not consi-dered, can be made. These products can be grouped into 30 groups. Most of the groups can be produced on two different machines, some groups on only one machine and a few groups on three machines.
3. The present scheduling
In the present situation, it is decided every year how often and on which machine(s) orders from a certain group are produced. Some groups are made every 3 weeks, other groups, with smaller demand, every 6 or 12 weeks. The groups are made in a fixed or-der and during a fixed time, except in case of emergencies.
The main advantage of this way of scheduling is its simplicity from the point of view of the production department, the customers as well as the sales department. The main disadvantage is its inflexibility. This inflexibility becomes particularly clear if demands deviate from their forecasts and if urgent order~ come in. As a consequence one uses ad-hoc policies for urgent orders and delivery dates become instable and large.
4. Demands upon the scheduling
We divide the orders into 4 categories in order of priority: 1. very urgent orders;
2. urgent orders from industrial users; 3. urgent orders from stocktraders; 4. non-urgent orders.
The company stated the following requirements for the average delivery times and the accuracy in realizing the delivery date:
category delivery time (in weeks) accuracy (in days)
1 1-1,5 ± 1
2 3-4 ± 3
3 6-7 ±8
4 as requested (> 7) ±8
Table 1. Requirements for the production planning.
In the present situation these requirements can only be satisfied for the last two categories.
4
-5. Proposed scheduling method
The base of the proposed scheduling method is formed by capacity reservations for groups of sizes. A new capacity reservation is made if one wants to schedule an or-der but there is not sufficient capacity available in a reservation for its group. The amount of reserved capacity dependB on the forecasted amount of production time, needed for orders belonging to the considered group, that will arive before the ex-pected production moment of the group and that would best be scheduled in that servation. So, for every machine we have a sequence with scheduled orders and re-served capacity (Fig. 1).
Machine 1
scheduled orders from group i
reserved capacity for orders from group i
Fig. 1. Example of a machine-sequence.
Furthermore for every group of sizes we have a stand-by sequence. Orders which are not immediately scheduled are put into the stand-by sequence for their group. In this company this leads to the following situation in which we have 6 machine-se-quences and 30 stand-by semachine-se-quences (Fig. 2):
Machine Machine Group 1
.
•
Group 30 1~
•
6~~
~2 1~
2~
22~/~
;,0b
26Fig. 2. Example of a possible situation.
3
Now we have to decide on the precise use of these sequences:
- when do we open a new reservation and on which machine do we open it?
I I
5
-sequence?when are orders of a stand-by sequence scheduled, i.e. transported to a machine sequence?
Suppose an order arrives. If the order is of Category 1 or the order is of Category 2 and has a current size, i t is placed in a machine-sequence, either in a reserved part or after the last reserved part on a machine, in which case new capacity is re-served. The delivery-date is based upon the place in the sequence. If the order is of Category 3 or the order is of Category 2 and has not a current size, i t is put into a stand-by sequence and its delivery date is based upon the total amount of production time and reservations in the Inachine-sequences and the currency of the order. If the order is of Category 4 i t is also put in a stand-by sequence, but now the delivery-date is the same as the delivery-date required by the customer.
Orders of a stand-by sequence are scheduled if: - they are nearly too late;
- if in the neighbourhood of their last possible production moment an order of (nearly) the same size is planned.
6. Numerical results
We did several simulations with real orders. In Table 2 and Table 3 the results of a representative simulation with real orders during a period of 4 months are given.
Category %
<
2 weeks %<
3 %<
4
%<
71 89 98 100 100
2 14 62 80 98
3
a
5 16 96Table 2. Delivery-dates in a simulation study.
Category %
<
1 day %<
2.5 %<
5 %<
7.5 1 98 100 100 100 2 70 93 99 100 3 25 48 76 91 4 22 43 I 71 87Table 3. Accuracy of delivery-dates in a simulation study.
In this case the occupation of the lllachine was 76% for net production and 15% for rebuilding. Compared with the real situation during the period which generated the orderdate, there was a considerable improvement in delivery time for orders of