&86720(5'5,9(1 0$18)$&785,1* ,1 7+(
)22'352&(66,1*,1'8675<
'U'LUN3LHWHUYDQ'RQN
6RPWKHPD$ 3ULPDU\SURFHVVHVZLWKLQILUPV
$EVWUDFW
Food processing industry copes with high logistical demands from its customers. This paper
studies a company changing to more customer (order) driven manufacturing. In order to help
decide which products should be made to order and which made to stock, a frame is developed
and applied to find and balance market and process characteristics. The frame is based on the
well-known Decoupling Point concept and adapted to the needs of the food processing
industry. The application in the company helped management in deciding and implementing
customer-driven manufacturing. The main results were lower inventories and less
obsolescence, while dependability remained the same. Further research should develop the
frame, along with general decision rules for locating the Decoupling Point.
Keywords: competitive response, decoupling point, make-to-order, production management,
food processing
,1752'8&7,21
The last decade has shown a remarkable amount of interest in a variety of aspects of
food processing and food processing industry. More specifically, the changes in
information technology and the power-balance of the supply of food (Meulenberg and
Viaene, 1998), have been catalysts for changes in supply chain management towards
new concepts such as Efficient Consumer Response (Kurt Salmon Associates, 1993)
or category management (Hutchins, 1997). As a result, these new themes have been
further explored and researched (e.g. Coopers & Lybrand, 1996). Cachon & Fisher
1
(1997) provide an illustrative example reflecting the actual implementation of these
new concepts. To date, it seems that most changes have dealt with supply chain
management issues like faster replenishment, the use of scanning, inventory planning
and Electronic Data Interchange, which are usually initiated by the (large) retail
chains. In fact, within the context of this renewed attention to supply and delivery, the
attention has been dominantly drawn towards the last part of the chain from producer
to consumer. The actual processing and production of foods and the consequences for
food processing industries of these changes have, to a large extent, been ignored.
Traditionally, food processing industries have organised their production systems in
such a way that they were able to produce in large quantities. In order to be as
inexpensive as possible, full use of capacity was necessary and as a result, production
was made to stock. Fast delivery from such stock is easy. However, given the changes
described and demands posed upon the industry, it is problematic to maintain this
production policy. Greater product variety, higher demands on quality and best-before
date, smaller quantities and new products can be barriers to proceeding with the
current way of producing.
Many of the above-mentioned developments are present in the recent
business situation we studied. The management of an industrial supplier in the food
processing industry found themselves in a situation of a growing variety of products,
fast delivery, and relatively high capacity utilisation. The immediate reason to start
this research in the company was twofold. On the one hand, the number of stock
keeping units and the amount of inventory grew faster than the total turnover, and on
the other hand, there was an occasion to obtain a number of silos for stocking
intermediate, semi-manufactured products. The aim of the study is to reduce the
amount of inventory of finished products and find which products should and could
be stored in the silos, from which the products can be further produced to order.
We found that there is little theory on guiding this type of decision. However,
the frame of Hoekstra & Romme (1992) could be transformed into one suitable for
the food processing industries. More importantly, the frame developed supported
actual decision-making in this case.
2
The paper is organised as follows: First the case is introduced. Next, the
theoretical background is developed. Subsequently, the frame is applied in the case-
study. Some conclusions and general remarks are made in the concluding section.
25*$1,6$7,21$/%$&.*5281'
The company under study is relatively small and employs around fifty people. Its
share of the Dutch market is about thirty percent, mainly in a price sensitive part of
the market. A growing share of turnover is realised in the higher price segments of the
market for which a European market exists. During the previous years total volume
has constantly risen.
The production process is relatively easy and consists of three main steps:
processing, granulating and packaging, as depicted in )LJXUH.
Processing Granulating Packaging
raw materials semi-manuf. finished
products products
)LJXUH3URGXFWLRQSURFHVV
The first stage (Processing) consists of several strongly related steps: raw
materials are mixed according to recipe and pre-processed; then the product is
processed in ovens, followed by an operation that breaks up the product as a
preparation for the second stage. Now, the product can either be stored in one of the
silos as semi-manufactured product or be directly granulated. Granulation means that
a product from one recipe is separated into several fractions of different sizes of
granule. Next, the product is packaged into different packaging forms, which are
stored or delivered to clients directly. Due to cleaning and set-up times, minimal
batch-sizes are present in all stages. Average throughput time for a batch is two hours
for processing and half an hour for granulating. The production speed (in tons/hour)
of granulation is twice as much as the production speed of the processing stage. The
3
product can be kept for about half a year, but customers demand a shelf-life of at least
four months. Although management aims at low inventory in all storage points, the
physical dimensions of the intermediate storage in silos are restricted: total storage is
less than a week’s volume.
The company has a large number of customers, and demand is stable on an
aggregate level. However, the forecasting on the level of the specific products is not
that simple. The variety is great: 200 different products, varying in recipe (40
recipes), granule (30 sizes) or packaging (type and/or amount). Five recipes (bases of
several finished products) account for about 70% of total demand. The largest
customer has a share in total volume of less than 10%. Lead times have a standard of
5 days, but tend to become shorter. Some customers actually demand a very short
delivery, in combination with the wish to maintain a dedicated amount of inventory
for immediate supply to them. On the other hand, some customers order in a
relatively regular way with a lead time of two weeks.
7+(25(7,&$/%$&.*5281'
The main concept used in analysing the above problem is the Decoupling Point,
which Hoekstra & Romme (1992) define as “the point that indicates how deeply the
customer order penetrates into the goods flow” (see also Wortmann et al., 1997).
Determining the location of the decoupling point aims at finding a balance between
the costs of procurement, production, distribution and storage, on the one hand, and
the customer service to be offered, on the other. In principle, each product-market
combination can have its distinct decoupling point. There are three important aspects
associated with the Decoupling Point (DP):
• The order-driven and forecast-driven activities are separated by the DP.
• The DP is the main point of inventory from which the customers are delivered.
• The forecast-driven activities can be optimised in some way, as they are not
directly influenced by the irregularities of the market.
4
)LJXUH illustrates the above in a simple production process consisting of two
stages, separated by a Decoupling Point.
Upstream Downstream
Decoupling
Point
Suppliers raw Customers
materials
Production process Production process
Forecast oriented Customer oriented
)LJXUH 7KH 'HFRXSOLQJ 3RLQW &RQFHSW DGDSWHG IURP +RHNVWUD 5RPPH
S
The main determinants of the DP are divided into two groups: product and
market characteristics, and process and stock characteristics. These determinants are
summarised in 7DEOH,.
Product and market characteristics Process and stock characteristics
• Required delivery reliability • Lead times and costs of steps in the
• Required delivery time (production) process
• Predictability of demand • Controllability of manufacturing and
• Specificity of demand procurement
• Costs of stock-holding and value
added between stock points
• Risk of obsolescence
7DEOH , 'HWHUPLQDQWV RI WKH 'HFRXSOLQJ 3RLQW DGDSWHG IURP +RHNVWUD 5RPPH
Each of the factors in 7DEOH, influences the location of the DP. For example,
the shorter the required delivery time, the more the DP will be located towards the
5
customer (downstream), while, e.g., a great risk of obsolescence will force the DP to
be located more towards the supplier (upstream). Both examples hold only if all other
factors remain stable. Hoekstra & Romme distinguish five basic logistical structures
(five basic locations for the Decoupling Point): make and ship to (local) stock, make
to stock, assemble to order, make to order, and purchase and make to order.
Usually, this frame is an excellent tool to describe and analyse an actual
situation, but the literature shows little evidence that this frame is also useful or
suitable for redesigning a situation. We will show that by adding some of the specific
characteristics of food processing industry to this frame, it is possible to use the frame
for that aim as well.
'(9(/23,1*7+(&21&(37)25)22'352&(66,1*,1'8675<
This section briefly develops a frame for the food processing industry. A detailed
description can be found in Van Donk (2000). Before the Decoupling Point concept is
elaborated upon for the food processing industry, it is necessary to describe this type
of industry from the point of view of production and logistical management. Because
of the type of industry we are interested in, given our case, we exclude from our
description the types of industry, that process agricultural raw materials into basic
food ingredients (such as mills, abattoirs, and sugar refineries).
7DEOH,, shows an enumeration of characteristics of food processing industry
compiled from the literature ( Bolander, 1980; Taylor et al., 1981; Van Dam et al.,
1993; Fransoo & Rutten, 1994)
6
Plant characteristics
• Expensive and single purpose capacity coupled with small product variety and
high volumes. Usually, the factory shows a flow shop oriented design.
• There are long (sequence dependent) set-up times between different product
types.
Product characteristics
• The nature and source of raw material in food processing industry often implies a
variable supply, quality and price, due to unstable yields from farmers.
• In contrast with discrete manufacturing, volume or weights are used.
• Raw material, semi-manufactured products, and end products are perishable.
Production process characteristics
• Processes have a variable yield and processing time.
• At least one of the processes deals with homogeneous products.
• The processing stages are not labour intensive.
• Production rate is determined mainly by capacity.
• Food industries have a divergent product structure, especially in the packaging
stage.
• Factories that produce consumer goods can have an extensive, labour intensive
packaging phase.
• Due to uncertainty in pricing, quality and supply of raw materials, several recipes
are available for a product.
7DEOH ,, &KDUDFWHULVWLFV RI IRRG SURFHVVLQJ LQGXVWU\ FI 9DQ :H]HO 9DQ 'RQN
In most cases, a limited number of these characteristics is present. An
example can be found in the case described by Van Donk & Van Dam (1998). The
list above is used for examining, describing and analysing real-life situations.
Whenever one of the characteristics is present, it has consequences for the lay-out of
the equipment, control and planning of activities and thus (as we will explore in the
7
next sections) on the location of the Decoupling Point. High set-ups and an
orientation to use capacity as much as possible, for example, cause planning of long
production runs and stocks of end products.
0DUNHWFKDUDFWHULVWLFV
Delivery times tend to be short and delivery reliability needs to be consistently high
in food processing industry. The effect on the Decoupling Point is downstream. While
the retail-chains demand short deliveries, they usually pass all the uncertainty onto the
producers who experience a kind of lumpy demand, which can hardly be forecasted.
Although scanning data and Point-Of-Sale data are available to the retailers, these are
seldom shared with the producer. Here, a better co-operation could be beneficial and
open up possibilities for longer lead-times. The last important point in relation to the
market is the specificity of demand. As noted, food processing companies have a
large diversity of products, originating from differences in packaging size, labels and
brands. The best-before-date adds to the specificity in two ways. Firstly, because the
best-before date is a technical, safety limit to sell the product. Secondly, because
retailers do not accept the same best-before date on subsequent deliveries. As such,
the commercial best-before is much shorter than the technical best-before suggests.
The above discussion is summarised in 7DEOH,,,.
3URFHVVDQGVWRFNFKDUDFWHULVWLFV
Food processing industries process natural materials, which vary in quality and
composition. Therefore, processes might be uncontrollable in their yield or the quality
of the output. The result is a downstream effect on the Decoupling Point: delivery is
buffered against the unpredictable production by placing a certain amount of
inventory after that process. Cleaning times and set-ups are another important factor
in the processing of food. Large (sequence-dependent) set-ups especially have a
downstream effect on the location of the Decoupling Point. We already mentioned the
perishability of food; products might easily become obsolete, due to the limited shelf-
lives of food products. As mentioned, products might become obsolete even if the
best-before date is not reached. The last point to consider is the value of stock to
8
which two factors are relevant: the value added in the process and the value of
purchased materials. A high value added stimulates an upstream movement of the
Decoupling Point, as it is more profitable to stock low-value materials instead of
high-value finished products. However, if the value of the raw materials is relatively
high (e.g. tobacco, coffee beans) and purchase of the raw materials can not be
postponed, then the location of the Decoupling Point is not relevant for this aspect.
We summarise this discussion in 7DEOH,,,.
Market characteristic Presence/Value in Food Effect on DP
Delivery reliability High Downstream
Delivery time Short Downstream
Predictability of demand (rather) Unpredictable Downstream
Specificity of demand High, through great Upstream
variety and best-before
Process characteristic Presence/Value in Food Effect on DP
Lead times and costs High set-ups/cleaning Downstream
times and costs
Controllability (sometimes)Low Downstream
Value added and costs of Unclear (in general) -
stock-keeping
Risk of obsolescence High Upstream
7DEOH ,,, 7KH LQIOXHQFH RI 0DUNHW DQG 3URFHVV FKDUDFWHULVWLFV RQ WKH 'HFRXSOLQJ
3RLQW
2WKHUFKDUDFWHULVWLFV
There are two points of interest for the location of the Decoupling Point that are not
directly covered in the above characteristics. The first is that in many food processing
industries there is a bottleneck capacity in either the packaging or the processing
stage. In general, it will be problematic to produce to order on a bottleneck capacity,
except for situations in which variability in orders is very limited. Locating the
Decoupling Point upstream from the location of the bottleneck will thus be virtually
9
impossible. The second point to pay attention to is the recent introduction of the
concept of ECR (Efficient Consumer Response). The concept of continuous
replenishment is especially relevant for the above discussion. There are potential
ways to improve communication and co-operation in a supply chain. Through
information sharing on actual sales, the Decoupling Point can move upstream and
more production to order seems possible.
/2&$7,1*7+('(&283/,1*32,17,17+(&$6(
The description of the case makes clear that there are three possible locations for the
Decoupling Point (DP): the stock of raw materials, the stock of semi-manufactured
products (in silos), and the stock of finished products. The silos, especially, have a
limited capacity.
With respect to the market characteristics, it is evident that the company is in
a situation in which delivery should be very reliable and delivery times are short for
most products, which has a downstream effect on the DP. The predictability of
demand is not experienced as very high, which leads to a more upstream location of
the DP. The variety in finished products is quite large. However, the commonality in
recipes is quite large as well, as five recipes are responsible for 70% of the total
amount produced. On the other hand, there are also a number of recipes used for only
one product/customer.
With respect to the process characteristics we notice that the throughput times
are not very large, but that set-up times are relevant. The process is reliable and the
storage costs are not very large in either of the possible locations. However, the risk
of obsolescence is not negligible, due to the variety in packaging forms and the shelf-
life guaranteed.
The starting point in deciding how each product will be stored (in the sense of
determining the Decoupling Point for each product) is the observation that the storage
space in the silos is limited. A natural point to start is there.
10
Firstly, if a product is produced from a unique, product-specific recipe, it will
not be stored in the silos. The idea is that such a product is client-specific from the
first step in processing; intermediate storage is not useful then. The decision to
produce either to order or to (end) stock depends on the amount and lead time of these
products. One third of the products is thus excluded from storage in the silos.
A second observation is that the processing stage has less capacity than the
subsequent stage. Capacity can be used most efficiently if the number of set-ups and
the amount of cleaning time is limited. Producing in large runs is advantageous, but
only necessary for products with a large aggregate demand (across all different
granules). These products (recipes) are stored in the silos to enable a fast reaction to
market demand without carrying excess inventory and running the risk of obsolete
products or having the right product in the wrong packaging. By applying this rule to
the products, a number of products could be selected that can be stored in the silos.
A last rule that was applied dealt with a number of products with irregular
demand and small quantities requested. Normally, one produces such products to
order (and the Decoupling Point is the stock of raw materials). However, because
demand for these products is smaller than minimum batch sizes, some stock of these
(finished) products will remain. This inventory has a relatively high risk of becoming
obsolete. It could be that reconsidering the market strategy and eliminating such
products yields better profits in some cases.
5(68/76$1'',6&866,21
In applying the above rules, it was possible to change the Decoupling Point for about
75% of the number of finished products, from production to end-stock to either the
production from raw materials or production from intermediate products. The main
reason for having end products on stock is the very short time asked for delivery,
usually in combination with the wish of customers to maintain an amount of
dedicated inventory. Other products are kept in this stock, as well, because demand is
very stable so no risk of obsolescence is present.
11
Changing the Decoupling Point in this way affected customer service only in
a positive sense: the number of end-items in stock is drastically reduced and inventory
control is made easier. Obsolescence and inventory costs are reduced. The efficiency
of the production process is not actually reduced, because aggregate demand is stable
and the run-length (minimal batch size) in the first stage is reasonably long.
Moreover, there are some products that have a longer lead-time or are made to stock,
to fill possible gaps in capacity usage.
A last result stems from the fact that in deciding the location of the
Decoupling Point, all factors related to market and process were analysed. This yields
a fresh look upon the position of the company in the market and discussions on the
product portfolio, market strategy, and the lack of commonality in recipes of
products. Discussion and interaction between sales and production departments are
stimulated.
There are a number of general conclusions to be drawn as well. The frame
developed for the location of the Decoupling Point in food processing industry is a
valuable tool for finding the important factors in market and process, and deciding the
actual location of the Decoupling Point. Moreover, in developing this frame we added
to the knowledge of both the characteristics of the food processing industry and the
Decoupling Point theory. The case study showed the theory’s usefulness, but further
validation across the range of food processing industry is needed.
Some issues need to be addressed. The theory of the Decoupling Point
focuses on describing the relevant factors for the location of the Decoupling Point,
without paying attention to the way these factors should be (or could be) balanced, in
a qualitative or quantitative sense. Further development of the frame as presented in
this paper could be helpful in developing such decision rules. Current developments
in the food processing industry make this industry an attractive area for further study.
This paper did not address the actual planning and organisational
consequences in the new situation. In general, combining make-to-order and make-to-
stock on one machine or process is a difficult planning task, which needs further
research. Organisational consequences apply to re-arranging tasks and responsibilities
12
between departments (production and sales), redesigning the process of order-
acceptance and more co-ordination around the flow of goods and information. The
above themes are also of interest in inter-organisational issues. In this case also, the
effect on transaction costs and supply chain development are relevant. More insight
into these issues is needed as well.
13
5()(5(1&(6
Bolander, S.F. (1980), “Materials Management in the Process Industries”, in:
American Production and Inventory Control Society 1980 Conference Proceedings,
pp. 273-275.
Cachon, G.P. & Fisher, M.L. (1997), "Campbell Soup's Continuous Replenishment
Program: evaluation and enhanced inventory decision rules", Vol. 6, pp. 266-276.
Coopers & Lybrand (1996), Efficient Consumer Response Europe, Value Chain
Analysis project overview.
Fransoo, J.C. & Rutten, W.G.M.M. (1994) “A Typology of Production Control
Situations in Process Industries”, International Journal of Operations & Production
Management, Vol. 14, No.12, pp. 57-72.
Hoekstra, S. & Romme, J.(Eds.) (1992), Integral Logistic Structures: Developing
Customer-oriented Goods Flow, McGraw-Hill, London.
Hutchins, R.(1997), “Category Management in the food industry; a research agenda”,
British Food Journal, Vol. 99, No. 5, pp. 177-180.
Kurt Salmon Associates (1993), Efficient Consumer Response: Enhancing Consumer
Value in the Grocery Chain, Food Marketing Institute, Washingston DC.
Meulenberg, M.T.G. & Viaene, J. (1998), “Changing food marketing systems in
western countries”, in: Jongen, W.M.F. & Meulenberg, M.T.G. (Eds.), Innovation of
Food Production Systems: Product Quality and Consumer Acceptance, Wageningen
Press, Wageningen (The Netherlands), pp. 8-36.
Taylor, S.G. Seward, S.M., Bolander, S.F. (1981), “Why the Process Industries Are
Different”, Production and Inventory Management Journal, Vol. 22, No. 4, pp. 9-24.
Van Dam, J.P., Gaalman G.J.C. & Sierksma, G. (1993), “Scheduling of packaging
lines in the process industry: An empirical investigation”, International Journal of
Production Economics, Vol. 30-31, pp. 579-589.
Van Donk, D.P. (2000) “Make to stock or make to order: the decoupling point in the
food processing industries”, International Journal of Production Economics, (to
appear).
14
Van Donk, D.P. & Van Dam, J.P. (1998) “Structuring complexity in scheduling: a
study in a food processing industry”, British Food Journal, Vol. 100, No. 1, pp. 18-24.
Van Wezel, W. &Van Donk, D.P. (1996), “Scheduling in Food Processing Industries:
Preliminary Findings of a Task Oriented Approach”, in: Fransoo, J.C. & Rutten,
W.G.M.M. (Eds.), Second Int. Conference on Comp. Integrated Manufacturing in the
Process Industries - Proceedings, BETA, Eindhoven (The Netherlands), pp. 545-557.
Wortmann, J.C., Muntslag, D.R. & Timmermans, P.J.M. (1997), Customer Driven
Manufacturing, Chapman & Hall, London.
15
