EXPLORING THE WAY A PUBLIC FOOD WAREHOUSE DEALS WITH SEASONALITY, PERISHABILITY, AND LIMITED CAPACITY

(1)

EXPLORING THE WAY A PUBLIC FOOD WAREHOUSE DEALS

WITH SEASONALITY, PERISHABILITY, AND LIMITED CAPACITY

MASTER THESIS

Abstract

Inventories represent a major cost source for food supply chains. The use of inventory management methods is often inhibited by the inherent variability that is present in food production. Two sources of variability are supply and demand seasonality and perishability. Seasonality causes amplifying cycles in the production and distribution of food. These cycles impose challenges for companies within food supply chains. Food products are perishable, which means that they deteriorate over time, which imposes certain requirements for food supply chain management. This thesis focuses on the role of public warehouses, which are independent providers of storage space and logistic services within food supply chains. By means of a single case study, this thesis explores how public warehouses deal with seasonality and perishability, while their capacity is limited. The findings of this study signify that there is a gap between the ways to deal with seasonality in literature and reality. We found that seasonality has a high impact on the supply side of a public warehouse. In the area of perishability, we found that requirements and norms concerning food safety impose lead-time constraints on transformational processes within public warehouses. However, the control of shelf life factor has proven to be of less influence. We present a number of parameters and variables that can be used as a basis for further research.

Keywords: Seasonality, perishability, public warehouse, cold chain, food supply chain management

Programme:

MSc Technology & Operations Management University of Groningen (NL)

Programme:

MSc Operations & Supply Chain Management Newcastle University (UK)

(2)

Preface

(4)

1. Introduction

With a growing urban population and rising food prices, the importance of a steady food supply is gaining attention. Ensuring availability of food throughout the world, at an acceptable price, calls for efficient production and worldwide distribution of safe and quality food products. However, there are still major opportunities for food supply chain management, both in theory and in practice (Van Donk et al., 2008). The level of uncertainty and variability is generally high in food chains. This is caused by seasonal supply and demand, variations in quality and yield, and limited shelf life of products before and during processing. Variability in production systems creates the need for buffering. The law of variability buffering states that variability in a production system will be buffered by some combination of time, capacity, or inventory (Hopp & Spearman, 2008). Since time is not flexible when harvesting and transforming perishable food products, it is not a suitable buffer method for this industry. Capacity buffering forms a problem as well, since food producers use capital-intensive production plants that have to be utilized to a high degree in order to be cost efficient. The most evident way to buffer variability in food production is by using food inventories, which serve as a cushion for disruptions in the constant flow of quality products. This paper researches that type of inventory.

The rapid deterioration rate of some food products requires quick processing and packaging, in order to extend the products’ shelf life. When demand is not immediate, storage under temperature-controlled circumstances is necessary. Therefore, food inventories are stored in temperature-controlled warehouses. These facilities store food products after production or processing. The supply and demand for these products is typically characterized by seasonality. Seasonal supply is present in the harvesting of crops such as corn. Seasonal demand occurs for products such as barbecue meat, which is mostly sold during summer. Seasonality causes known cycles in supply and demand patterns, which has an effect on the availability of food products (Slack et al., 2010). As a result, inventory levels of seasonal products follow a certain amplifying pattern.

(5)

logistic services. Public warehouses combine seasonal patterns of different food producers, thereby aiming to use storage capacity at a more optimal level than a private warehouse (Zhang et al., 2003). They are better able to take in upsurges and downturns of food production, which is often a major constraint for private warehouses. Furthermore, the location of public warehouses can have a positive effect on the responsiveness, as they allow food producers to store their end product inventory closer to their customers (Chen et al., 2001). The pursuit for a better capacity usage may have a positive effect on the cost efficiency; it does not intrinsically solve issues caused by seasonality. In this paper, we explore which trade-offs a public warehouse has to make when supply and demand are seasonal and capacity is limited.

The contribution to warehouse research has been extensive on the two-model, private-public warehouse system (Kar et al., 2001; Lee, 2006; Lee and Elsayed, 2005; Rao and Rao, 1998; Rong et al., 2011). However, the models proposed in these papers focus on optimization of the private warehouse. These models take the capacity of a private warehouse as a fixed parameter, and assume the capacity of a public warehouse to be infinite with stable costs. The public warehouse is used as a backup in case the capacity of a private warehouse becomes insufficient. The gap that this research aims to fill is to explore how a public warehouse operates under limited capacity. This paper explores the way a public warehouse deals with seasonality and perishability, using a case study research. Since seasonality is predictable but food production knows a high degree of variability, it is studied how a public warehouse manages its inventory under these constraints. The contribution to science that this paper aims for is to explore the role of a public warehouse within the food chain. Resulting from our research, we provide a number of variables and parameters that can be used as input for an inventory model that takes public warehouses into account.

1.1. Research Questions

We use a main research question with four sub questions.

RQ: How does a public warehouse with limited capacity deal with seasonal supply and demand of perishable food products?

SQ 1: How is the warehouse influenced by seasonality? SQ 2: How is the warehouse influenced by perishability? SQ 3: How does the warehouse deal with these forces?

(6)

1.2. Terminology

Within this paper, the role of a warehouse treated as a buffer node between producers and consumers of food products. Technically, a warehouse has customers that rent storage space, and suppliers that deliver packaging materials and energy. However, because we focus on supply and demand patterns, we view the warehouse as a supply chain node. Therefore, this paper refers to suppliers as companies that ship products to

the warehouse. We refer to customers as companies that receive goods from the warehouse. See figure 1.1.

Figure 1.1: Terminology

1.3. Structure of the paper

This paper is structured as follows. The next section presents a literature review that explains the basic concepts such as seasonality, perishability, and food chain characteristics. In the methodology section, it is described how this research was conducted. The findings that resulted from the research are presented in the results section, after which these are discussed and concluded in the last section.

2. Literature review

In this section, the current literature on private versus public warehouses, food chains, seasonality, and perishability are reviewed to create a basis for the explorative case study.

2.1. The main differences between private and public warehouses

In the introduction, it was already mentioned that a common distinction in the current body of science exists between private and public warehouses. The models that we have found in recent literature all adopt this two-warehouse structure (Kar et al., 2001; Lee, 2006; Lee and Elsayed, 2005; Rao and Rao, 1998; Rong et al., 2011). Based on these papers, we provide an overview of the main differences between private and public warehouses:

Suppliers Warehouse Customers

(7)

Description Private warehouse Public warehouse

Capacity Dedicated to one supplier Open for multiple suppliers Collaborative planning and

forecasting with supplier Yes No Opportunities for supply chain

integration High Low

Options for cost-efficiency Limited High

Table 2.1: Main differences between private and public warehouses

2.2. A food chain

Food chains encompass some properties that make them unique compared to other supply chains. The complexity of food chains is generally high, in both product and process characteristics (Akkerman et al., 2010; Van Donk and Van Dam, 1996; Van Donk et al., 2008). Perishability, food safety requirements and increased consumer attention for animal wellbeing are the main drivers for supply chain management within food chains (Boehlje et al, 1995).

Food chains often exist in the form of cold chains. This term describes a supply chain that operates under one maximum temperature or an array of multiple, maximum temperatures. Cold chains are not exclusive to the production of food; they are also typical for the production of photographic film, microchips, and pharmaceutics. For food products, quality degradation is slowed down by impeding the growth of potentially harmful bacteria, thereby stretching the product’s shelf life (Akkerman et al., 2010). Typical cold chain temperatures are < 7° C for chilled products and <-18° C for frozen products. Quality systems such as HACCP and BRC provide norms for temperature conditions of products within a cold chain.

On the one hand, food supply chains can be agile and adopt intermediate inventories, in order to be market responsive. On the other hand, the appliance of lean manufacturing is becoming increasingly common in food production, and companies aim to reduce their inventory costs. Also, current economic circumstances can inhibit the ability to finance large inventories. Gunasekaran et al., (2001) identify five sources of total costs associated with inventory in warehousing, being:

1. Opportunity cost of warehousing; 2. WIP;

3. Stock management;

(8)

This signifies that the services that a public warehouse offers can be interpreted as waste, considering point 1, 3 and 4. Therefore, the use of a public warehouse by a food producer is expected to be some kind of trade-off in costs versus agility or service level.

2.3. Seasonality

Seasonality in availability causes known cycles in the supply of food products, such as potatoes and beets, which are harvested in certain seasons (Dorfman and Havenner, 1991; Kar et al., 2001). The quality and lead-time in supply can fluctuate, because of changes in weather conditions or transport modes (Riezebos and Zhu, 2014). Supply cycles can be broken by outside events such as weather conditions, diseases, or political intervention. For meat products, seasonality and variability in supply can be caused by numerous factors, which are amplified by the fact that a large share of products is produced by a push system. This creates large interdependencies between different products. The availability of raw material such as calves for the production of veal meat is a result of the volume of milk production. The supply of this type of product is therefore influenced by additional factors such as milk quota and feed prices.

A general way of treating seasonality in availability is by sourcing inputs globally, which means importing goods from different parts of the world to ensure a steady supply (Riezebos and Zhu, 2014; Wang, 2013). However, this is not possible for fresh food products, since transportation time is often longer than their shelf life.

Seasonality is also apparent in demand, because of weather (the barbecue season) or to traditions, such as Christmas or Easter. Consumer purchases also occur in small cycles, as a share of the population likes to shop during weekends. Ehrenthal et al., (2014) state that “not accounting for demand seasonality can lead to very significant optimality gaps, yet incorporating only some form of demand seasonality does not always lead to cost savings.” Concluding, managing supply and demand seasonality seems a major element in the coordination of food chains.

(9)

prices. In a market where meat prices are declining, demand for storage capacity declines as well. The model that Tolley & Harrell developed, assumes that the choice of holding meat inventories will not depend solely on treating seasonal influence, but centrally on liquidation prices of inventory. The outcome of their paper suggests that “firms will undertake no storage as long as the average increase in price is less than storage cost, for expected profit will motivate none”. This signifies that meat prices can be a source of information for storage companies. The authors further state that meat inventory provides an incomplete cushion to variations in meat production, however no further information is given. Their description of meat storage is useful, but their paper is dated, because their market assumptions do not longer hold in modern times.

2.4. Perishability

(10)

companies shall result in additional gains”. They take vendor managed inventory (VMI) as an example to achieve this. Practices such as VMI are acknowledged as supply chain integration practices, and are broadly acknowledged as means to treat perishability and therefore reach better supply chain management (van Donk and van der Vaart, 2004; Van Donk et al., 2008).

In this chapter, we have identified the current positioning of the literature on cold chains, food warehouses, seasonality, and perishability. The next section describes the methodology that is used to answer the research questions.

2.5. Treating seasonality in supply and demand

Current warehouse literature provides a number of instruments that can be used to deal with seasonality in supply and demand. These are displayed in the figure below. We categorised the instruments corresponding to their area of influence within the supply chain, and whether the instrument has an anticipatory or reactive character. Anticipatory instruments are designed to regulate seasonality in advance. Reactive instruments treat seasonal effects after occurrence.

Figure 2.1: Instruments to treat seasonality

The instruments are further elaborated in the table below. We provide a general description as well as a comment on the applicability of the instrument. The second column lists the papers from which the description and applicability were derived.

Instrument Author Description Applicability

Supply chain integration (SCI) Amorim, Meyr, Almeder, & Almada-Lobo, 2011; Van Donk et al., 2008.

SCI practices facilitate the seamless flow of products between supply chain nodes. Electronic Data Interchange (EDI), Vendor Managed Inventory (VMI), and Collaborative Planning, Forecasting, and

According to Amorim et al. (2011), SCI is necessary to reduce quality issues and perishability risks that arise with seasonality. They argue that a strong integration between manufacturers and Third Party Logistic Providers (3PL) improves Supplier

- Supply chain integration - Order acceptance - Yield management - Alternative products Warehouse - Shared resources - Forecasting - Safety stock - Postponement - Alternative products - Overtime/idle time/ annualised hours - External space leasing

Customer - Supply chain integration

(11)

Replenishment (CPFR) increase transparency in forecasting and planning.

traceability and decreases costs associated with spoilage and deterioration. Van Donk et al. (2008) stress that food supply chains should adopt a high level of integration when the inherent uncertainty of the supply chain is high. They also state that the use of SCI is limited when resources are shared.

Order

acceptance Kilic, Van Donk, Wijngaard, & Tarim, 2010.

Orders are accepted when the available capacity is sufficient, such as in low seasonal times. Orders are rejected when the available capacity is insufficient, caused by seasonal peaks.

Kilic et al. (2010) stress that decision rules to coordinate the admission decisions for incoming orders are necessary. Known factors should be the revenue and costs associated with a particular order. They consider order acceptance as a major instrument to deal with seasonal variability in yield, and provide several examples of its use.

Yield

management Chiang et al., 2007; Lysons & Farrington, 2012; Slack et al., 2010; Smith et al., 2014.

Pricing strategies are used to attract or avert certain buyers. Yield management is common in the airline industry, where inventory reservations are managed in a way that increases or maximizes company profitability.

Yield management has proven to be a suitable way of managing seasonal supply in the consumer service industry. Chiang et al. (2007) provide some examples for yield management practices in cargo and freight, such as price determination based on cabin space and location. Overtime,

idle time, and annualised hours Kalantari, Rabbani, & Ebadian, 2011; Slack et al., 2010.

Overtime can be used to stretch the short-term capacity of the labour force. Idle time lets the labour force be unproductive during work. Annualised hours involve setting a number of hours per year rather than hours per week.

Applicable for short-term or daily seasonal variations, for organisations that have a significant labour intensity. Not applicable for long-term variation, because of increased costs of overtime and unproductivity, and issues in employee morale.

Shared

resources Van Donk et al., 2008. Use capacity to serve multiple suppliers. When equipment is flexible, the capacity a company’s storage can be used for a variety of products from different suppliers. This way, seasonal patterns can be complemented. According to Van Donk et al. (2008), shared resources inhibit supply chain integration, so the use of both these instruments can conflict and might impose the need for a trade-off.

Forecasting Slack et al., 2010; Dekker et al., 2004.

Forecasting is a method to predict changes in volume based on historical data. Tools such as Exponential Smoothing or a Moving Average can be used.

(12)

Safety stock Hopp and Spearman, 2008; Slack et al., 2010.

Safety stock is a common OM practice to be used as protection against uncertainty in demand quantities. In case of a warehouse, safety stock involves reservation of capacity (pallet spaces, slack capacity of equipment).

Safety stock is one of the most straightforward ways of hedging against uncertainty and seasonal variation.

External

space leasing Lee & Elsayed, 2005. Warehouse leasing, entails renting capacity outsourcing, i.e. from other warehouses, in the same way food producers do.

External capacity must be available within acceptable proximity, considering transportation costs. Seasonal variety can cause problems for the focal warehouse’s service level, as unforeseen orders may require goods that are stored externally.

Alternative

products Slack et al., 2010; Hackman and Bartholdi, 2011.

Empty storage space can be filled with other products than usual, or alternative services can be used to fill up the idle production capacity. This way, idle capacity can be utilized on the short term, when seasonal levels are low.

Filling idle storage capacity with alternative products is only possible when seasonal cycles are known and predictable to a certain degree. A high seasonal variability of core products can cause fluctuations in the availability of idle storage capacity.

Post-ponement Hackman and Bartholdi, 2011; Hopp and Spearman, 2008.

Postponement entails late customization of products, in order to respond quickly and improve the alignment of supply and demand.

Manufacturers can outsource activities such as packaging to warehouses, which can provide the warehouse with the opportunity to use a postponement strategy.

Table 2.2: Instruments to treat seasonality

In order to investigate the use of these instruments in reality, we conduct a single case study research further on.

3. Methodology

The literature review provided us with knowledge on the basic concepts that are central to this thesis, but did not provide a sufficient base to answer the main research question. The intrinsic behaviour of a public warehouse under seasonality and perishability constraints is therefore researched using a case study analysis. The exploration of the case study should provide us with new insights that we need in order to answer the research question.

3.1. Case selection

(13)

the case company has to be a public warehouse, in the sense that its capacity is not allocated to a single buyer. Second, its target market has to be within the food chain, and hence stores products that have a limited shelf life. Thirdly, the goods that the company stores should follow a seasonal pattern. This pattern can be apparent in supply, demand, or inventory level.

3.2. Introduction to the case study

The case company is a medium-sized, public, cold store warehouse. Amongst other services such as packaging and thawing, the competences of the warehouse are shock freezing and frozen and chilled storage of perishable food products. The majority of the suppliers operate in meat production and trade. Within this case company, we selected one product group that accounts for a major share of the used storage space in the warehouse. This particular product group is supplied by a meat producer and is characterised by a seasonal supply. The inventory can be divided in two types. Part is non-customer decoupled, make to stock (MTS), caused by divergent production (55%). The other part is customer-specific, make to order (MTO) production (45%), which is stored for accumulation and consolidation1_{. Considering the fairly equal MTS/MTO} ratio, the production process of the supplier can be characterised as both push and pull. The share of push/MTS production is exclusively stored in the warehouse in case. For that reason, the fluctuation in MTS production levels is apparent in the storage space usage of MTS products. The producer primarily supplies fresh, chilled products directly to retail customers. Products that cannot be sold immediately are sent to the warehouse, where they are shock frozen and placed in storage. For the export market, shock freezing is a necessity because the product’s shelf life needs to be extended for long distance, frozen transport. Additionally, export products need to be accumulated in the warehouse in order to consolidate container shipments. The inventory is completely vendor-managed; the producer controls the inflow and outflow of goods, while the warehouse keeps track of total inventory levels.

When goods enter the warehouse, they are inspected on factors such as quantity, weight, and product quality. After arrival, the goods are placed in the shock freezer as soon as possible. The shock freezer is a closed area of the cold storage, where low temperatures combined with high wind speeds freeze fresh products to their core in 2 days. Because the shock freezer requires a different layout of the boxes on the pallet, and uses different pallet racks than the cold storage itself, products need to be

(14)

palletized before and after shock freezing. Long-term storage takes place in the cold storage, where the goods are placed and registered on a single box SKU level. The shelf life of the frozen product is around 3 years, which is hence the maximum time that products are kept in storage, waiting for a customer order. The supplier has full control over the items that have to be shipped to customers. Customer orders arrive at the supplier, where the order is translated into an order for the warehouse. When this arises, the appropriate boxes are picked and shipped to the customer.

The process steps, within the supply chain, are illustrated in figure 3.1.

Figure 3.1: The production process within the supply chain

As can be deducted from figure 3.1, the different process steps have differences in the nature of their capacity. For that reason, we make a clear distinction between the different types of capacity at this point in the thesis. These capacities are divided in two time frames in which they are relevant, being short-term or long-term. The distinction between short-term and long-term is important when we review the influence of perishability, because the deterioration rate of products is altered after shock freezing. This is elaborated in the results section.

Capacity type Relevant

time frame Constraints

Inbound capacity Short term The inbound capacity is constrained by the floor space of the expedition rooms, personnel capacity, and availability of equipment such as weighing stations and forklifts.

Freezing capacity Short term The freezing capacity is constrained by the capacity of the shock freezer, which can process a specified tonnage per day.

Storage capacity Long term The available shelves in the warehouse determine the storage capacity of the warehouse, where goods a stored

Re-palletize palletizeRe- pickingOrder- Outflow

Inflow Shock freezing > 2 days Inbound buffer

< same day Storage

Outbound buffer < 1 hour * * * * = Frozen Else = Chilled Meat

producer Warehouse Retail

(15)

for long periods of time.

Outbound capacity Short term Entails order picking and shipping activities. Similar to the inbound capacity, the outbound capacity is constrained by the floor space of the expedition rooms, personnel capacity, and availability of equipment such as weighing stations and forklifts.

Table 3.1: Capacity types

3.3. Scope

Within the thesis, we view the warehouse as a supply chain node, and therefore examine its position within the supply chain. However, we limit the span of control to the warehouse itself. More specifically, we treat the inflow- and outflow volumes from a single supplier as a given. A meat chain is operating under a high amount of variability caused by the vast number of variables that influence the flow of goods.

3.4. Data collection

The data for the exploration of the case study was collected in two phases. In the first phase, we were looking for practical, qualitative data on how the company makes decisions regarding seasonality and perishability, and why these decisions are made. The findings then provided a direction for the collection of quantitative data from the company’s ERP system, which will be elaborated further on. The first step in phase one was to design an interview protocol, in order to add structure to the interviews. Because of the novelty of the interview topics, we chose for a semi-structured interview design, where questions were targeted in a specific direction. The semi-structured design gave us the possibility to ask further and let the respondent elaborate on the how’s and why’s of the answers initially given. The interviews were conducted with a member of the management board, a back office manager, an office employee and the production manager. Since the organisational structure of the company is fairly flat, we chose to ask our questions to a chosen ‘key informant’ who possessed the required knowledge for the particular question, after which the answers were validated by the other informants where applicable. This process projected multiple viewpoints on the questions asked.

3.5. Data analysis

(16)

2009-20132_{. The dataset that we collected consists of weekly inventory records. The} combination of the qualitative data from the interviews and the ERP inventory records enabled us to ensure validity and reliability of the data gathered. Data on perishability is registered on the individual SKU’s, in the form of a production- and expiry date. The warehouse itself does not register these dates. Perishability data was therefore gathered by a random sample of 20 SKU’s within the chosen product group.

4. Results

In this section, the results of the case study are presented, where we research the variables that were conceptualised in the literature review. This section is structured around the sub questions, which are answered in a discussion between the findings of our study and the available literature on the particular subject. As a start, we recall the main research question:

RQ: How does a public warehouse with limited capacity deal with seasonal supply and demand of perishable food products?

4.1. The influence of seasonality

The data that we gathered from the company’s ERP system gives insight in the inventory level of the chosen product group, over the years 2009-2013. This data is visualised in the graph below.

(17)

Figure 4.1: Total inventory levels per month

The chart shows that the total inventory levels follow a seasonal pattern, which peaks in the autumn and reaches its bottom point in the winter. The course of the patterns is fairly consistent, however the year 2011 yields an opposite, in this case exceptional pattern. The interviews signified that in this year, the prices for the type of meat in case were exceptionally high, therefore the inventory depleted with a much higher rate. This is consistent with the statement that we reviewed in literature, stating that firms will not undertake storage when the average increase in price is less than storage cost (Tolley and Harrell, 1955). Because the prices were expected to drop from that year on, the inventory was depleted at a much higher rate. The reason for this was due to a slaughter stock shortage in that year, caused by the announcement of a milk quota imposed by the Dutch government. This quota caused a rise in livestock for milk production, cutting livestock quantities meant for slaughter.

Considering the partial (55%) push production system of the chosen meat producer, the supply and demand for livestock is expected to provide some kind of indication for slaughter quantities and demand for storage space. Our expectation was that the livestock price would have a negative relationship to the demand for storage space, as rising livestock prices would mean falling slaughter quantities, whereas a fall in livestock (raw material) prices would give a producer the incentive to increase its production levels. 3,5 4 4,5 5 5,5 6 6,5 7 1 2 3 4 5 6 7 8 9 10 11 12 To ta l inv ent or y level Month

Total inventory levels per month

(18)

Figure 4.2: Inventory levels vs. market price3

Figure 4.2 shows the livestock price (per kg) combined with the storage levels. As can be seen in the chart, the price seasonality is less dramatic than the seasonality of the storage usage. Moreover, the relationship between the two patterns is not consistent at first glance. Figure 4.3 shows the correlations between the livestock price and storage level.

Correlation coefficient: 0,27 Correlation coefficient: -0,24

Figure 4.3: Correlation analysis

The scatterplot on the left shows the correlation over the years 2009-2013. We observe a flat regression line, which raises the suspicion of a weak positive relationship. This is further indicated by the correlation coefficient of 0,27. Earlier in this section, we argued

3_{Market price source: http://www.boerderij.nl}

€ 2,50 € 3,00 € 3,50 € 4,00 € 4,50 1 1,5 2 2,5 3 3,5 1 5 9 1 5 9 1 5 9 1 5 9 Ma rket p rice Inv ent or y level Months 2009-2013

Inventory levels vs. livestock price

Inventory levels Market price

(19)

that some exceptional inventory behaviour occurred around 2011. By looking at figure 4.2, it seems that there is some relationship in the last two years of observation. For that reason, we added a second scatterplot of the years 2012-2013, which shows a weak negative relationship, with a correlation coefficient of -0,24. Consequently, we are unable to derive any generalizable conclusions for a relationship between livestock prices and the demand for storage space.

Yearly disruptions in the inventory levels happen frequently and are often random and unforeseen, according to the warehouse. The recent loss of a large supplier is directly caused by a military coup in one of the exporting countries, a year before. A similar event occurred for the complete European market, when Russia closed its borders for meat products in 2014. The occurrence of such exceptions is not uncommon. As already argued by van der Vorst et al., (2009), food supply chains are subjected to random factors caused by weather, pests, and other biological hazards. Typical outside risks that can disrupt the supply chain, identified by the warehouse, are listed in table 4.1.

Scope Event Effect

International Foreign policy, trade blocs, wars Complete nations open up or are closed for export at once

International/domestic Diseases, food scandals Goods are blocked, quarantined, and stored for long periods of time. Domestic Government intervention, quota The government enforces quota or

removes goods from the market to influence prices

Table 4.1: Outside risks that disrupt the supply chain

According to Harland et al. (2003), supply chain risk can be assessed by one asking itself: How likely is it that an event will occur? And: What is the significance of the

consequence and losses? Public warehouses have a poor view of future events in the

supply chain, and the fact that many of its suppliers operate in different markets makes this more complex. Therefore, the probability of an outside risk is hard to predict in advance, if the risk is foreseen in the first place. At the least, the flow of goods is affected and disrupts the recurring seasonal pattern.

(20)

depicts a boxplot of the inventory levels in the warehouse, per month over the past five years.

Figure 4.4: Inventory levels from 2009-2013

Per month, the inventory level is displayed with its highest value of the past five years (top of the vertical line), as well as its lowest (bottom of line), together with the median value (diamonds). The first observation that can be made is that the median inventory levels follow a clear pattern for seasonal anticipation inventory. The inventory level is at its highest around August, and remains fairly constant from that point, until the end of December. The inventory level in January is considerably lower, and continues to decline until the lowest point around April. From this month on, the levels rise as the summer start, completing the cycle. According to an interview with an administration employee conducted on 16-7-2014, part of the inventory is accumulated for the Christmas holidays, and is supplied to European supermarkets around that time (this will be further verified by the inbound and outbound records, in the next section). The second observation that can be made is that the inventory fluctuation per month is rather high. The maximum inventory level is around 50% higher than the median for the majority of the months. However, the course of all three values is almost identical in seasonality. This signifies that the inventory levels fluctuate heavily, but their pattern is cyclical, and thus predictable.

Goods flow in and out on a daily basis. In figure 4.5, the left chart shows the inflow of goods, and the right chart shows the outflow.

0 0,5 1 1,5 2 2,5 3 3,5 1 2 3 4 5 6 7 8 9 10 11 12 To ta l inv ent or y level Months

Total inventory levels per month

High Median Low

(21)

Figure 4.5: Total inflow and outflow of goods, 2009-2013

The boxplots show that the seasonality factor is also apparent in the inflow of goods. The inflow fluctuates heavier per month than the total inventory levels, as can be seen from the high-low levels compared to the median. The seasonality of the inflow pattern seems to be consistent with the accumulation of total inventory. Clearly, the production levels of the supplier rise from its bottom level in February. This is consistent with the accumulation character of the inventory, as it builds up towards the summer. The outflows are clearly higher in the peak demand season, which is wintertime (interview with the back office manager conducted on 18-7-2014). The outflow levels have less seasonal variation, however some interesting patterns emerge. The outflow of goods experiences its bottom point in the summer, when the inflow level is at its highest. A simple subtraction learns that the inventory in storage should rise during this season, which is apparent in the total inventory graph in figure 4.4. Aside from the spike in November before Christmas; the outflow levels remain more constant than the inflow levels.

4.2. The influence of perishability

As we recall the literature review, perishability affects the shelf life of a product. The expectation from literature was that perishability causes major complexities within the food chain. However, for frozen storage, the complexities are limited. The general shelf life of the products is 3 years, while stored below -18 degrees Celsius. Considering the quality and safety of the product, the shelf life is one of the most important parameters for a food product. It falls, however, outside the warehouse’s scope. Since the food

0 2 4 6 8 10 1 2 3 4 5 6 7 8 9 10 11 12

Total inflow of goods per month

Median Low High

0 2 4 6 8 10 1 2 3 4 5 6 7 8 9 10 11 12

Total outflow of goods per month

(22)

producer owns the food products and the warehouse is a mere provider of temperature-controlled storage space, shelf lives are only administered by the food producer. This is remarkable, considering the role of the warehouse as a service provider, as well as the responsibility for food safety that lies on the complete supply chain. The shelf lives are considered to ensure that products can be sent into the market while they are still usable. Shipping products before they perish lies within the responsibility of the producer. For that reason, the complexities for storage operations are not high, nor are the risks.

However, perishability is a major constraint on the inflow and outflow operations, which are a significant share of warehousing activities. Before shock freezing, the inbound products are fresh, which means that they have just been slaughtered and chilled since. The shelf life of fresh products is around 1 month, but they need to be placed in frozen storage as quickly as possible after inflow. For outbound goods, the perishability constraint is even more critical. Products are transported from the freezer to a chilled expedition room, where the shipment is checked before transportation. This process has to be done within one hour, because the frozen products thaw in the expedition environment before they are loaded into a refrigerated truck.

4.3. The Internal process

The internal transformation process consists of various process steps, which not only differ in nature, but also in the minimum and maximum allowed lead-time. The effect of seasonality varies per process step. Incoming goods are processed in batches that are divided by the incoming shipments. One shipment consists of 12.5 tonnes on average4 (see table 4.2).

Sample Average High Low

24 12.5 18.2 1.4

Table 4.2: Incoming shipment quantity (tonnes)

The supply of fresh goods fluctuates between 1-4 full truck shipments per day. These shipments arrive evenly during working hours.

(23)

Figure 4.6: The internal transformation process

Figure 4.6 displays the internal transformation process. Below, we elaborate on the process steps in terms of function and capacity.

Process

type Name Description

Inbound

buffer The inbound buffer is formed by the expedition room and a chilled reception cell. This buffer is necessary because the fresh goods cannot be processed immediately. The fresh goods are allowed to stay in the inbound buffer until the end of the day, when they have to be placed in the shock freezer. The capacity of the inbound buffer is adapted to maximum supply levels.

Re-palletize Fresh goods need to be re-palletized for shock freezing. Spacers need to be placed between layers of boxes on the pallets, to increase the airflow that passes individual boxes. This is done by one worker, who operates a re-palletizing machine. The capacity of this process is mainly constrained by the operating speed of the machine, and lies on average around 10 tonnes per hour.

Shock

freezer The shock freezer is a closed area of the cold storage cell which is equipped with fans, which produce high wind speeds that cause the air temperature to drop to -30° C. The shock freezer can process 60 tons of fresh products per 24 hours. Products need to be stored in the shock freezer for a minimum of 48 hours to ensure complete freezing. During the freezing cycle, the shock freezer can be accessed in order to move batches in and out. The shock freezer is a shared resource, and is used for all incoming goods that have yet to be frozen.

Re-palletize This process removes the box spacers that have been added to the pallets before shock freezing. This process is conducted on a different machine than the first re-palletizing process, so the two are able to run in parallel. The capacity is similar, at around 10 tonnes per hour. The re-palletizing takes place in a chilled room, therefore a maximum lead-time of 1 hour applies.

Storage Products reside in the storage cell for a variable time. The exact time of storage is unknown beforehand, however products are always shipped before their shelf life ends, because perished products loose their value.

Order

picking The order picking process happens within the storage cell and is constrained by the speed of the forklift operator. Outbound

buffer The function of the outbound buffer is to facilitate rapid shipment of outbound products, as soon as an outbound truck is available. This buffer is the same expedition room that is used as inbound buffer; hence it is a shared resource. Outbound goods stay for a considerably shorter time in the expedition room, because they are constrained by a maximum storage time of 1 hour in the chilled environment.

Table 4.3: The internal process steps

Re-palletize Re-palletize Outflow Order-picking Inflow Shock freezing > 2 days Inbound buffer

< same day Storage

(24)

The length of internal lead-times can fluctuate depending on the WIP inventory caused by supply and demand upsurges. The company has fixed norm times for some of the process steps, in order to comply to cold chain requirements, set by HACCP and BRC food safety standards.

4.4. Evaluation of instruments to treat seasonality and perishability

The major instruments are elaborated in the section below. Per instrument, we aim to refer back to the literature review for a comparison between theory and practice.

4.4.1. External space leasing

(25)

warehouse. This jeopardises the responsiveness of not only the focal public warehouse, but inevitably of the whole supply chain. For this reason, goods are selected for internal leasing when they are expected to reside in storage for 6 months or longer. That way, the goods are normally moved back into internal storage, before they are coupled to a customer request. This eliminates the change of a rogue order, where goods are requested by a customer and have to be retrieved from an external warehouse. A share of the products cannot be stored externally because they are bound by traceability constraints. This means that the complete supply chain of the product has to be approved and documented, and for that reason, the public warehouse has no liberty to store products elsewhere.

An additional reason for external space leasing is because the case company wishes to validate the need for internal capacity expansion. The essence is that without leasing, capacity expansion creates empty storage space that needs to be filled as soon as possible. When extra storage space is leased, security of storage utilization is provided. When the capacity of a planned expansion equals the capacity leased at an external warehouse, the goods can be moved to the extended internal storage, which immediately yields the desired storage utilization level. Finally, there are legal concerns. Food products that reside in a warehouse are insured by the food producer itself. Once goods are moved to a different warehouse, risks in the insurance coverage arise. However, problems in this area have not occurred so far.

Concluding, external space leasing is the main way to treat a lack of storage space caused by seasonality. A risk that arises with external storage is decreased responsiveness. As a general rule, it is uncertain which specific SKU’s are requested by customers at a given time. For that reason, external storage is most appropriate for slow-moving items, because these reside in storage for the longest amount of time. Fast-movers have the highest probability of being requested shortly after external storage.

4.4.2. Overtime and idle time

(26)

overtime can cause a number of problems. First of all, overtime wages are higher than the normal wage, imposing additional labour costs. Secondly, research on overtime found that jobs with overtime schedules are linked to increased injury hazard rates (Dembe et al., 2005). Also, significant positive correlations were found between working hours and overall health symptoms, physiological and psychological health symptoms (Sparks et al., 1997). As opposed to overtime, idle time allows employees to be unproductive during their working hours. Idle time is a short-term measure as well, but opposed to overtime, does not aid perishability measures in any way.

4.4.3. Yield management

The acceptance of suppliers is regulated through yield management, which we recall as a pricing strategy to attract or avert certain buyers (Lysons & Farrington, 2012; Slack et al., 2010). This proves to be useful for influencing the demand for logistic services as well as storage. However, there is no clear strategy linked to the use of yield management. The company offers higher prices when the warehouse’s capacity is reaching its maximum, mainly by rules of thumb. Literature provides examples of excellent use of yield management, for instance in airline management (Slack et al., 2010). Within this application, prices vary as the availability of demand varies. It is however common in the meat industry to agree on fixed prices at the issue of a tender, i.e. prices do not fluctuate during a contract, or depending on the available warehouse capacity. The management of the warehouse argues that altering prices during a contract is not acceptable. We have trouble identifying the differences between an airliner or hotel, and a logistic service provider. We found evidence of capacity auctions in the UK energy and gas market (Yarrow, 2003). We believe that the difference appears in the number and customers and the mobility of the supplier of capacity. Further research is advised in this area.

4.4.4. Basic forecasting

(27)

Warehouse Supplier Planning Replenishment Forecast Physical flow Information flow

Figure 4.7: The current supplier-warehouse relationship

This way, a reasonable expectation can be made about the inventory levels in the next year. These levels are used for storage capacity decisions, which are essentially of a make-or-buy character: how much capacity to invest in, or outsource by leasing external space.

Surprisingly, forecasts for the company’s full operations are not used. One of the issues identified is that at any given time, the number of suppliers is somewhere in between 20 and 50. These supply different types of meat to the warehouse, and there are differences in how supply is affected by variables (for example veal dependence on milk production, pork belly production on bacon prices). Since manufacturers possess significantly more knowledge on their product type, traditional collaborative planning, and forecasting (CPFR) would seem to benefit this supply chain, whereby the supplier producer a forecast which is shared with the downstream supply chain nodes. Gunasekaran et al., (2001) stress that: “Measuring inventory at supply, production, distribution and scrap levels as well as accuracy of forecasting techniques, can provide an insight into the cost performance and reduce the lead-time in a supply chain”. The lack of supply side forecasts is remarkable, however it is not clear whether the warehouses’ suppliers do not use forecasts or merely do not share them.

4.4.5. Alternative products and services

(28)

allergens. We assume that bulk cold storage warehouses that focus on storage and have less shared resources, make more use of complementary seasonal products.

4.4.6. Order acceptance

In a supply chain without shared forecasts, capacity cannot be reserved in advance, whether it is storage or additional service capacity. Other measures have to be taken in order to regulate the supply quantities on a short to medium term. The apparent solution that we derived from literature, was by using an order acceptance policy (Kilic et al., 2010). Within such a policy, orders that can fill excess capacity are accepted, while orders that exceed the maximum capacity are rejected. However, order rejection is not allowed in the case, and the company has a policy that orders of existing suppliers are always accepted. When the internal services lack capacity, overtime or weekend work is scheduled. When there is a lack of storage capacity, space is leased elsewhere. The only lever the company has on this level, according to the general manager, is by using a supplier acceptance policy.

The interviews with the company have signified that the supplier does not accept order rejection. In other words, the volume of supply of a single supplier does not have a limit. The supplier’s attitude in this case holds a strong resemblance with the notion of the infinite-capacity public warehouse. The warehouse therefore has a policy where order rejection is not allowed, and is not willing to make exceptions, because the warehouse wants to retain its exclusive position. The back office manager confirms in an interview that the company’s policy is based on supplier acceptance rather than order acceptance, because the requirement of food producers to accept all orders is common in this industry. We have trouble understanding the policy of food producers to disallow order rejection. The two participants that commented on this subject, were interviewed separately, but gave similar views on the subject. The reason seems to be that the warehouse should provide full service regarding the suppliers’ access inventory, which in this case means taking responsibility over the full inventory of the producer.

(29)

The lack of an order acceptance policy inhibits the day-to-day control that the company has on its operations. The effect of seasonality is much present in the day-to-day operations. Especially the inflow of goods is a major factor. Typical orders are received a few hours in advance. The utilization of the production equipment is scheduled on a per-day basis. Gunasekaran et al., 2001 (following Wild, 1995 and Slack et al., 1995) state that the capacity utilization is a major factor influencing flexibility and lead-time. The influence in lead-times for the in- and outflow of goods depends largely on the utilization of production equipment. Since manufacturers use public warehouses as a means to treat sudden upsurges in production levels, the level of goods that flow into the warehouse can fluctuate heavily. The information on incoming shipments is usually produced and shared with the warehouse a few hours in advance. The situation in practice is that early in the morning, information on the quantity of incoming goods becomes available for the warehouse for that particular day. This seems to be coherent with findings in other cases of food producing companies in the food chain (Van Wezel et al., 2006), where it was found that smaller food processing companies struggle with meeting short term orders for large retailers. Considering the VMI system that the warehouse shares with its supplier, it surprised us that there was no shared forecast in any form. We believe that if the food producer could share a weekly or even monthly forecast with the warehouse, the latter would be able to schedule more effectively. A quick research on previous tenders that the company took part in, the longest shared forecast horizon that we found was one week, and this was in one case only. Also, the tender mentioned that the supply fluctuation could be 100%. A recent research by Taylor and Fearne (2009) on demand management in fresh food chains confirms our findings. They research 6 cases of fresh food companies, including fresh meat chains. In no case, a common forecast for the whole chain was found. In case of one meat chain, 8 different forecasts were used, at each single supply chain node, without any alignment between them. We did not manage to find hard evidence of factors that inhibit long forecasting horizons for food producers. However, Taylor and Fearne conclude what is assumed in this thesis as well, that “the synchronisation of demand and supply in agri-food supply chains is a complex task due to the variability at both ends of the chain”. We believe that this statement, in combination with poor supply chain integration, causes the lack of shared forecasting.

4.4.7. Postponement

(30)

expediting, freezing, and storage. For that reason, the ability to postpone is limited to logistical choices concerning mainly internal logistics. Because postponement entails late customization, we have reviewed sources of transformation that take place before long term storage. It became apparent that the products are bound to the sequence of the internal production process. Thawing of products is allowed once in the supply chain, and happens at the end of the chain.

4.5. Summary of the exploration phase

Besides the identification of sources for uncertainty, a major goal of the interviews was to identify the power the warehouses has to deal with uncertainties. The table below provides the instruments that are used or not used by the warehouse, in order to treat a seasonal supply. They are marked on a Likert-scale where ++ = strong use, + = normal use, +/- = incidental use, and - = no use. These are the result of the interviews that were conducted with the management of the warehouse.

Use Instrument Benefits Issues

++ External space leasing Order rejection can be

avoided. Potential quality issues, hence rework. ++ Overtime and idle time Overtime works well

for maintaining quality and maximum lead-time requirements.

Idle time is costly and not beneficial for employee morale. + Yield management Effective way of

demand management at supplier-levels.

Not possible at order-level.

+ Basic forecasting Enables the warehouse to make a rough estimation of future capacity needs.

+/- Alternative products and services Solves seasonal supply. Mostly rogue orders, and numerous products do not comply with the company’s food safety policy.

- Order acceptance Effective way to regulate short term capacity.

Not accepted by suppliers. Instead the company uses a supplier acceptance policy.

- Supply chain integration Collaborative planning, forecasting, and synergies in inventory control.

Quick turnover of warehouse’s supplier base inhibit ICT investment. - Shared resources Flexibility in storage

(31)

- Annualised hours Less idle time for

personnel. Agreement on terms needed from personnel.

- Safety stock Buffer against uncertainty, by reserving a small amount of capacity for existing customers.

Reserving empty warehouse shelves means lower utilization. - Postponement Ability to postpone

production when capacity is strained.

Raises issues concerning perishability. Table 4.4: Instruments used by the warehouse to treat seasonality

Derived from the insights we gained in the explorative phase, we present a number of variables an parameters that can be used as input for further research in the area of public warehouses.

Parameters Argumentation

Inflow and outflow volumes per supplier The volume of goods that are supplied to and shipped from the warehouse, are controlled by the supplier of the warehouse. Order rejection is not allowed.

Storage capacity Constrained by # of shelves.

Freezing capacity Constrained by shock freezer capacity. Lead-time norms Imposed by food safety standards.

Seasonality pattern of goods Seasonal cycles are fixed. External disruptions can be ignored, because of scoping reasons. Deterioration rate of goods The deterioration rates are fixed, but vary

between fresh and frozen.

Variables Argumentation

Inbound and outbound capacity Personnel and equipment can be scaled up if necessary.

# of suppliers Suppliers can be accepted or rejected based on the availability of capacity.

Prices Prices are divided in a fixed- and a variable component.

Fixed:

- Inbound costs, per order (Reception, re-palletizing 2x, freezing)

- Outbound costs, per order (order picking, shipping)

Variable:

(32)

Assumptions Argumentation

Order rejection is not allowed We maintain this notion since it falls outside the warehouse’s control.

Table 4.5: Parameters and variables

5. Discussion

The aim of this paper was to explore how a public warehouse with limited capacity deals with seasonal supply and demand of perishable food products. We conducted the explorative study by using a single case study.

We found that seasonality has a large impact on a public warehouse’s operations, despite the cyclical character of inventory levels. The first and foremost finding is that seasonality has a strong influence on the supply side of a warehouse, in regulating the capacity requirements of inflow, freezing, and storage processes. A large inflow fluctuation at a central supply chain node such as a public warehouse raises questions on the influence of possible bullwhip effects, despite the fact that the warehouse is not a buyer or seller of goods in storage. However, we do not know how a meat supply chain is coordinated, and what the degree of centralisation in purchasing might be. Some directions for further research will be given in the conclusion.

The perishability of food products brings major complexities in the food supply chain. The shelf life of products in storage did not turn out to be a constraint, since it is coordinated by an external supply chain node, which is the supplier of the products in this case. Different but not independent from the product’s shelf life, is maintaining the cold chain. The maximum lead-times imposed by the cold chain requirements have a considerable effect on the warehouse operations. Especially at the outflow level, the maximum lead-time imposes a bottleneck at the buffer of order-picked goods. The level of routings caused by external sourcing, compared with the safety requirements of the cold chain, gives rise to some questions. Storing products at different warehouses increases complexity in coordination, reduces traceability, and increases transportation, which inevitably imposes additional risks on maintaining the cold chain.

(33)

seasonal supply patterns cause a source of predictability in capacity requirements. Treating alternative products should be a suitable way of treating seasonality. The main reason for the lack of policy on these areas could be a lack of information from first tier suppliers and customers.

In the literature review, we identified instruments to treat seasonality. These were compared with the current state at the warehouse. The main control instruments that we found from the case study are the inbound/outbound capacity, the number of suppliers, and prices. Their structure is conceptualised in the figure below.

Figure 5.1: Conceptual model of the variables

The number of suppliers can be regulated through alteration in the prices calculated for storage and logistic services. The inbound and outbound capacity serves as a moderator for the number of suppliers that can be accepted. Pricing is a direct way to manage supplier requests. Pricing occurs on a fixed price, contract basis. We believe the warehouse could adopt a pricing strategy that is linked to the actual availability of storage capacity at a given moment.

The fact that we explored a relatively unknown phenomenon in literature, this research is subject to several limitations. First of all was the use of a single case study. Whilst it enabled us to do an in-depth research, the findings may be unique to the case company. However, it is assumed that the company in case operates in a similar way compared to other public warehouses. It is nonetheless a necessity to be careful in the generalisation of our results. We broadly used visual representations of data, which can be interpreted in different ways. A second limitation is the methodological focus on the warehouse as a supply chain node. We did not conduct interviews up- or downstream in the supply chain at meat producers or processors, who remain largely unexplored in literature.

6. Conclusion

This thesis aimed to contribute to science by identifying how seasonality affects a public warehouse, and which instruments a public warehouse has in order to deal with this

Inbound/outbound capacity

# customers

Prices -/+

(34)

source of variability. We have provided insight in the way a public warehouse makes decisions regarding its capacity, and listed a number of parameters and variables that can be used as input for a public warehouse model.

We identified that seasonality has a major effect on perishable food inventories, despite its cyclical and thus predictable pattern. Combining complementary seasonal forces is perceived as an ideal way of inventory management, and public warehouses initially seemed a suitable accommodator of this method. We found however that in practice, a public warehouse struggles in achieving this. An additional barrier we found was that seasonal patterns in the meat supply chain can shift and can be disrupted easily, caused by external forces that are essentially not controllable. The influence of perishability on the role of a warehouse within the food supply chain is especially apparent in the form of lead-time norms, which in their part constrain a seasonal supply.

(35)

is advised. We believe that SCI research could move farther upstream, since the supply complexity is high in goods such as fresh meat.

7. References

Akkerman, R., Farahani, P., Grunow, M., 2010. Quality, safety and sustainability in food distribution: a review of quantitative operations management approaches and challenges, OR Spectrum. doi:10.1007/s00291-010-0223-2

Amorim, P., Meyr, H., Almeder, C., Almada-Lobo, B., 2011. Managing perishability in production-distribution planning: a discussion and review. Flex. Serv. Manuf. J. 25, 389–413. doi:10.1007/s10696-011-9122-3

Chen, F.Y., Hum, S.H., Sun, J., 2001. Analysis of third-party warehousing contracts with commitments. Eur. J. Oper. Res. 131, 603–610. doi:10.1016/S0377-2217(00)00102-8

Chiang, W., Chen, J.C.H., Xu, X., 2007. An overview of research on revenue management: current issues and future research. Int. J. Revenue Manag. 1, 97–128.

Dekker, M., van Donselaar, K., Ouwehand, P., 2004. How to use aggregation and combined forecasting to improve seasonal demand forecasts. Int. J. Prod. Econ. 90, 151–167. doi:10.1016/j.ijpe.2004.02.004

Dembe, a E., Erickson, J.B., Delbos, R.G., Banks, S.M., 2005. The impact of overtime and long work hours on occupational injuries and illnesses: new evidence from the United States. Occup. Environ. Med. 62, 588–97. doi:10.1136/oem.2004.016667 Dorfman, J.H., Havenner, A., 1991. State-Space Modeling of Cyclical Supply, Seasonal

Demand, and Agricultural Inventories. Am. Agric. Econ. Assoc. 829–840.

Ehrenthal, J.C.F., Honhon, D., Van Woensel, T., 2014. Demand Seasonality in Retail Inventory Management. Eur. J. Oper. Res. doi:10.1016/j.ejor.2014.03.030

Gunasekaran, A., Patel, C., Tirtiroglu, E., 2001. Article information : Int. J. Oper. Prod. Manag. 21, 71–87.

Hackman, S.T., Bartholdi, J.J., 2011. Warehouse & Distribution Science. Atlanta, GA. Harland, C., Brenchley, R., Walker, H., 2003. Risk in supply networks. J. Purch. Supply

Manag. 9, 51–62. doi:10.1016/S1478-4092(03)00004-9

Hopp, W.J., Spearman, M.L., 2008. Factory Physics. New York: McGraw-Hill.

(36)

Kar, S., Bhunia, A.K., Maiti, M., 2001. Deterministic inventory model with two levels of storage, a linear trend in demand and a fixed time horizon. Comput. Oper. Res. 28, 1315–1331.

Karlsson, C., 2009. Researching Operations Management. New York: Taylor & Francis, Inc.

Kilic, O. a., Van Donk, D.P., Wijngaard, J., Tarim, S.A., 2010. Order acceptance in food processing systems with random raw material requirements. OR Spectr. 32, 905– 925. doi:10.1007/s00291-010-0213-4

Lee, C.C., 2006. Two-warehouse inventory model with deterioration under FIFO dispatching policy. Eur. J. Oper. Res. 174, 861–873. doi:10.1016/j.ejor.2005.03.027 Lee, M.K., Elsayed, E.A., 2005. Optimization of warehouse storage capacity under a

dedicated storage policy. Int. J. Prod. Res. 43, 1785–1805. doi:10.1080/13528160412331326496

Lysons, K., Farrington, B., 2012. Purchasing and Supply Chain Management. Harlow: Pearson Education Limited.

Lütke Entrup, M., Günther, H.O., Van Beek, P., Grunow, M., Seiler, T., 2005. Mixed-Integer Linear Programming approaches to shelf-life-integrated planning and scheduling in yoghurt production. Int. J. Prod. Res. 43, 5071–5100. doi:10.1080/00207540500161068

Rao, a. K., Rao, M.R., 1998. Solution procedures for sizing of warehouses. Eur. J. Oper. Res. 108, 16–25. doi:10.1016/S0377-2217(97)00159-8

Riezebos, J., Zhu, S.X., 2014. Seasonal lead times in inventory control. Groningen.

Rong, A., Akkerman, R., Grunow, M., 2011. An optimization approach for managing fresh food quality throughout the supply chain. Int. J. Prod. Econ. 131, 421–429. doi:10.1016/j.ijpe.2009.11.026

Slack, N., Chambers, S., Johnston, R., 2010. Operations Management. Harlow: Pearson Education Limited.

Smith, B.C., Leimkuhler, J.F., Darrow, R.M., Interfaces, S., Edelman, F., Papers, A., Feb, J., Smith, B.C., Fort, D.I., Airport, W., Leimkuhler, J.F., Darrow, R.M., 2014. Yield Management at American Airlines 22, 8–31.

Sparks, K., Cooper, C., Fried, Y., 1997. The effects of hours of work on health : A meta-analytic review. J. Occup. Organ. Psychol. 70, 391–408.

Taylor, D.H., Fearne, A., 2009. Demand management in fresh food value chains: a framework for analysis and improvement. Supply Chain Manag. An Int. J. 14, 379– 392. doi:10.1108/13598540910980297

EXPLORING THE WAY A PUBLIC FOOD WAREHOUSE DEALS WITH SEASONALITY, PERISHABILITY, AND LIMITED CAPACITY