Product mix variability with correlated demand in two-stage food manufacturing with intermediate storage
Introduction
The food industry is becoming a more and more competitive environment where manufacturers have to cope with short due dates imposed by the high market pressure, specifically from large retailers (Dobson et al., 2001, Rundh, 2005). In the food-processing industry, these due dates are especially important, as they are closely related to the best-before dates on the final consumer products. Other distinctive characteristics (see, e.g., Nakhla, 1995, Akkerman and Van Donk, 2006) are the perishability of products and the high quality demands. Also, a divergent product structure is common. Only a few raw materials (often agricultural) are processed and packaged to generate a multitude of end products (see, e.g., Fransoo and Rutten, 1994). Often, these end products are customer specific (e.g., mainly in package type, but sometimes also in product type).
Food production mostly consists of two stages: processing and packaging (Van Dam et al., 1993). Between these stages, intermediate storage is present (generally in the form of tanks or silos) to decouple the two stages. To deal with the short lead times and with customer-specific packages for end products, manufacturers often have make-to-order strategies on the packaging level. This puts additional pressure on the production system, as it becomes partly make-to-order (packaging stage), and partly make-to-stock (processing stage). This mixed MTO/MTS situation is related to the customer-order decoupling point (CODP) concept (see also Van Donk, 2001, Olhager, 2003, Soman et al., 2004, Wikner and Rudberg, 2005), which in this paper is located at the intermediate storage tanks between the processing and packaging stage.
The intermediate storage in these industries is normally constrained in capacity and time. Capacity is not only constrained by a limited number of tanks, but also because quality demands do not allow concurrent use of a tank (i.e., only one batch can be stored in a tank at a certain time). Time constraints result from perishability of the basic food product, which restricts the time until packaging (see also Akkerman et al., 2006). These storage constraints lead to dependency between the two stages, and often make scheduling a complicated matter in these industries (see also Van Dam et al., 1993).
In the literature, production scheduling of batch plants has been extensively studied in operations management and chemical engineering, mostly using mathematical modeling techniques such as MILP (see, e.g., Kondili et al., 1993, Pinto and Grossmann, 1998, Rajaram and Karmarkar, 2004). Most of these approaches become very difficult (i.e., computationally intensive) when considering limited intermediate storage. Furthermore, there is an important difference between batch plants and the type of food production system discussed in this paper. In most cases in the literature, the batches go through all production steps. Here, batches produced in the processing stage are used as inputs for the packaging stage. There are some papers that do treat such systems, like Méndez and Cerdá (2002), who study a make-and-pack facility and develop a MILP formulation, but unfortunately consider unlimited intermediate storage.
In food production, the processing stage commonly involves batch processes that produce various product types (recipes), while the packaging stage usually involves several lines to accommodate multiple package types (e.g., , , and 1 l). These different product and package types result in a product mix with two dimensions. Due to volatile market behavior in the food sector, the shares in the product mix change regularly—in both the product dimension and package dimension. For instance, new low-fat products are added to the mix, products can be on special offer, or customers (temporarily) buy more large family-sized packages. These changes cause shifts of the workload between packaging lines (package dimension), and cause changing storage tank usage (product dimension).
From the literature, we know that more variability in individual product demand results in a higher variance of the total demand (e.g., Ross, 1997), which in turn can have consequences like lost sales (Andreou, 1990), increasing flow times (Jensen et al., 1999), and increasing safety stocks (Vaughan, 2003). In the two-stage food production system studied in this paper, variability in the product mix causes short-term imbalance in the volumes for the product types and/or package types, which is likely to: (i) influence the blocking effects caused by occupied tanks and/or packaging lines, and (ii) affect the amount of waste due to perished product.
Concerning the variability in the product mix, the situation can be even more complicated due to dependency between demands for various product-package combinations. For example, it is well known that promotional activities within one retail chain affect the turnover of similar products in other chains, resulting in correlations between demands. Also, seasonal demands and new product introductions can result in products which have demand that is positively or negatively correlated with the demand for other products. In the literature, some papers address the issue of correlated demand. The main results are that the effects of variability in demand are stronger when demand is also correlated and that performance is negatively affected if correlations are ignored (see, e.g., Zhang, 1997, Vaughan, 2003, Ma et al., 2004).
For the two-stage food production system, the product mix variability can be correlated on two dimensions (products and packages), which has not been addressed before in the literature. Also, the interaction between an order-driven packaging stage, a forecast-driven processing stage, and limited capacity intermediate storage facilities is not at all clear from the literature. Although some papers discuss these types of production systems, they mostly concern mathematical optimization approaches (like MILP), which do not aim at understanding the basic behavior and interactions of such systems.
The aim of this paper is therefore to study the effects of product mix variability with correlated demand between product types and package types on the performance of a two-stage food production system with limited intermediate storage. We consider this to be explorative research, and we perform simulation studies to investigate the primary effects.
Section snippets
Production system
Fig. 1 illustrates the production environment studied in this paper. In the first stage, a batch process creates N basic food products from (agricultural) raw materials. In the intermediate storage stage, K storage tanks are available to store the basic food products (with ). Here, quality and traceability requirements restrict batches to concurrent storage. In the packaging stage, the basic food products are packaged in M different package sizes or types. More specifically, there are J
Product mix variability
End products are distinguished by product type () and package type (), which creates a total of possible end products. The demand is based on average order sizes . We assume that every arriving order can be for any of the possible end products (with equal probabilities). This means the average number or orders is the same for all end products. The average order sizes for these orders are not necessarily the same.
The product mix variability between different periods (we use
Experimental design and parameter settings
In the experiments, we initially focus on a situation with and . This means that we have three basic products that can be stored in three tanks. Furthermore, these products can each be packaged in three different package types (each on a separate packaging line). This creates a total of nine end products. We chose this configuration to have a minimum amount of interaction between different products and packages, while still having a reasonably simple system.
Customer orders arrive
Conclusions and further research
This paper studies the effects of product mix variability with dependency between product types and package types in a two-stage food production with intermediate storage. A simulation study is performed to investigate these effects in an explorative way. Dependency between product types and package types is modeled by defining a correlation coefficient for each of these dimensions. In this way, it easily translates to reality, and it also creates useful modeling possibilities.
The paper shows
References (22)
- et al.
A general algorithm for short-term scheduling of batch operations—I
MILP formulation, Computers & Chemical Engineering
(1993) Strategic positioning of the order penetration point
International Journal of Production Economics
(2003)- et al.
Combined make-to-order and make-to-stock in a food production system
International Journal of Production Economics
(2004) Make to stock or make to order: The decoupling point in the food processing industries
International Journal of Production Economics
(2001)- Akkerman, R., Van Donk, D.P., 2006. Analysing scheduling in the food-processing industry: structure and tasks....
- Akkerman, R., Van Donk, D.P., Gaalman, G.J.C., 2006. The influence of capacity- and time-constrained intermediate...
A capital budgeting model for product-mix flexibility
Journal of Manufacturing and Operations Management
(1990)- et al.
Buyer power and its impact on competition in the food retail distribution sector of the European Union
Journal of Industry, Competition and Trade
(2001) - et al.
A typology of production control situations in process industries
International Journal of Operations & Production Management
(1994) - et al.
Machine dedication and process flexibility in a group technology environment
Journal of Operations Management
(1999)
Statistical Theory
Cited by (14)
Product complexity and operational performance: A systematic literature review
2019, CIRP Journal of Manufacturing Science and TechnologyCitation Excerpt :Vilas and Vandaele [119] found that lead time increased as the variety within a manufacturing system became more differentiated or skewed in production times, setup times, and batch sizes such that one product form was notably higher in these dimensions than the others. Increased demand correlation between products was shown to increase lead times in a food processing company because of the resulting imbalance in the production system, which increased the machine blockage and starvation times [131]. A further moderating factor mentioned in the literature was the choice of the customer order decoupling point.
Planning and scheduling of the make-and-pack dairy production under lifetime uncertainty
2017, Applied Mathematical ModellingCitation Excerpt :Finally, the main conclusions are given in Section 6. The planning and scheduling problems of the make-and-pack production have been widely studied in the literature (e.g. [24,25,23]). Bongers and Bakker [27], Marinelli et al. [28], Kopanos et al. [29] have implemented planning and scheduling problems in yoghurt and/or dairy production, as yoghurt production is a two-stage semi-continuous make-and-pack process [26].
An off-price retailer with two ordering opportunities and demand updating
2017, International Journal of Production EconomicsCitation Excerpt :Also, as demand volatility increases, the advantage of advance buying over no advance buying increases. Since increased demand variability negatively affects performance (Akkerman and von Donk, 2009), it is important for managers to mitigate such a negative effect by using advance buying when faced with volatile demand. Finally, managers should utilize the advance buying option at the end of disappointing seasons for retailers.
Optimisation of product portfolio sales and their risk subject to product width and diversity
2020, Review of Managerial ScienceAn Approximate Dynamic Programming Approach for a Routing Problem with Simultaneous Pick-Ups and Deliveries in Urban Areas
2020, Food Supply Chains in Cities: Modern Tools for Circularity and Sustainability