A DECISION SUPPORT FRAMEWORK FOR REDESIGNING THE SUPPLY CHAIN AGAINST UNCERTAINTIES

(1)

A DECISION SUPPORT FRAMEWORK FOR

REDESIGNING THE SUPPLY CHAIN AGAINST

UNCERTAINTIES

Master thesis, Supply Chain Management

University of Groningen

June 26, 2017

Edzard Koopal

Studentnumber: 2051443

e-mail: e.koopal.1@student.rug.nl

Supervisor

N.D. van Foreest

Co-assessor

O. Kilic

(2)

Abstract

Unreliable suppliers are a common cause for imbalanced supply chains. Firms struggle dealing with uncertainties stemming from such suppliers. Redesigning the supply chain to benefit from inventory pooling can be a worthy alternative when ‘classic’ supply chain improvement strategies cannot be applied. This study provides managers with a decision support framework for redesigning the supply chain. The framework is tested with data from a company which redesigned its own supply chain. Adding an inventory hub near the suppliers decreases transport cost and increases service levels. The framework was further improved after these tests. Whether or not to redesign the supply chain depends largely upon potential risks stemming from resources and effort invested in the redesign.

Key words

(3)

1. INTRODUCTION

This paper investigates if pooling inventory closer to suppliers can be a viable remedy against unreliable suppliers in global supply chains. Especially smaller sized firms serving local markets often have trouble managing their overseas supply stream. Their suppliers are incapable guaranteeing in time processing and deliveries. Finding ways to improve supply chain performance is problematic for these firms. Most methods to reduce or avoid supply chain uncertainties proposed by literature involve suppliers directly (Davis, 1993; Li, Ragu-Nathan, Ragu-Nathan, & Subba Rao, 2006). Furthermore, altering sourcing strategies is widely known as an effective supply chain strategy (Treleven & Schweikhart, 1988), but only if better supplier options are available. However, in some type of supply chains these approaches are not feasible. Organizations characterized by possessing little buyer power or limited resources caused by their relative size are not capable to demand change from, or increase cooperation with suppliers. Thus, firms will have to search for alternative approaches making their supply chain more stable.

The observed firm in this paper is purchasing their products from large suppliers which supply numerous firms around the world. These suppliers make their products in a typical batch-production manner. The suppliers send no information about when a batch-production run is started or finished, or how big the batch will be. Once a batch is finished, buyers with the highest outstanding order of that product are served first. It is not uncommon batches are shared between multiple buyers because the size of the batch is not sufficient to fulfil all orders. Variable yield is therefore another problem for this firm. These issues make it difficult for managers to develop standardized ordering policies, because they have to deal with 1) erratic and stochastic lead-times and 2) variable yield of received orders.

(6)

Decreasing the risk of having insufficient stock levels is an important aspect in preventing lost sales when every sale counts.

Supply chain uncertainty is still an important issue within Supply Chain Inventory Management (SCIM) research. Managers still struggle making optimal decisions when dealing with uncertainty (Gumus, Guneri, & Ulengin, 2010). Demand and lead time variability are the two common compounds of this uncertainty. Where demand uncertainty has the emphasis of most literature concerned with the risks of randomness within inventory management, supply uncertainty is considered less. Research with a focus on lead time uncertainty is an area which lacks richness (Dolgui, Louly, & Prodhon, 2007). In addition, Iakovou, Vlachos, & Xanthopoulos (2010) point towards variability of the orders’ yield and inclusion of service level concerns as promising avenues for future research.

The ‘classic’ improvement theories dealing with unreliable suppliers give unsatisfying solutions for the investigated problem in this paper. Variability pooling remains to be tested as an appropriate strategy. Here, variability pooling is referred to as the centralization of inventories is used as a strategy against geographical variety (Véricourt, Karaesmen, & Dallery, 2002). Stocking products at a centralized warehouse closer to the suppliers will require significant changes in the supply chain structure. Therefore, a decision has to be made if adding a centralized stock point, closely located near suppliers, is a convenient option when trying to improve the supply stream. This leads to the following research question;

How to decide whether or not the redesign effort of the supply chain network is compensated by the potential benefits of the variability pooling strategy?

(7)

the strategic importance of considering the operational decisions concerned with supply chain design.

(8)

2. THEORETICAL BACKGROUND

2.1 Fashion industry resemblance

Apart from the e-commerce peculiarity, the case company bears great resemblance to fashion markets. Products sold become obsolete following a product life cycle (PLC) trend. In this case, salvage value plays an important role when determining a suitable inventory policy. By adding a warehouse overseas, opportunities arise to sell superfluous stock to 3rd party suppliers more easily.

Fashion industry markets often demonstrate high levels of ‘chaos’, because of its complex open nature (Christopher, Lowson, & Peck, 2004). Often, fashion retailers order their products one single time per season as a result of long lead times and limited life of the products (Jacobs & Chase, 2013). Therefore, underestimating demand is the lost profit due to sales not made. On the other hand, overestimating demand causes costs (order, transport, holding) to be higher in hindsight, in addition to the costs of discounting or salvaging products for lower prices (Jacobs & Chase, 2013). This makes coming up with sound inventory decisions increasingly complicated. The case company in this paper is characterized by a relative high lost sales cost ratio. When increasing service levels as a counter measure to temper the total lost sales factor, it is important to weigh trade-offs between service level, holding cost and salvage revenue. Like fashion retailers, under- or over –estimating parameters can have serious consequences. Therefore, a cost analysis will be produced to get more insight in the possible financial repercussions.

2.2 Variability/risk pooling

(9)

Spearman (2001: 279) acknowledge the benefits from variability pooling. Variability pooling (or risk pooling) is recognized as a useful tool within Inventory Management research.

de Véricourt et al. (2002) mention that stock allocation strategies are not popular in design research. This is mainly because such decisions are made at the operational level of the organization. Undervaluing stock allocation strategies can lead to incorrect design decisions (Véricourt et al., 2002). Therefore, the effects of stock allocation approaches, such as centralizing inventories, should not be underestimated at the strategic decision making level.

Bretthauer, Mahar, & Venakataramanan (2010) asked some relevant questions in their research which concerns the impact of ‘e-tail’ on supply chain network design. More specifically, they questioned the impact of demand variability on supply chain network design. The results of this research suggest there are potential cost savings by pooling inventories when demand variability is high (Bretthauer et al., 2010). Additionally, Bretthauer et al. (2010) results show that locating fulfilment at the store with the highest demand variability seems to minimize costs. It can be promising to research if these assumptions are valid when dealing with lead-time variability. Thus, pooling inventory closest to the source of lead-time uncertainty, the suppliers.

2.3 Facility location

Kuehn & Hamburger (1963) pointed out the potential cost advantages by adding regional warehouses, these are;

1. Economies of scale in transportation costs between warehouses and/or factories by permitting bulk shipments.

2. Reducing delivery costs by combining products from different suppliers into single shipments to individual customers, and in this case to the single internal customer in NL. 3. Increasing service levels through improved delivery times which stem from increased

proximity to customer locations.

(10)

Merely left over stock can be sold to 3rd_{party buyers from the additional warehouse in this paper.} Nonetheless, the mentioned cost advantages can be an important reason to add a warehouse closely to the suppliers in a situation with variable lead-times.

2.4 Recent frameworks

Frameworks in recent research literature, mainly form the network design topic, are concerned with either minimizing transport costs, or finding an optimal number and location for serving customer demand in some region. The framework of Creazza, Dallari, & Melacini (2010) is analysing the suitability of five pre-specified global logistic network configurations. Their taxonomy suggest to adopt a consolidation hub near suppliers when supplier dispersion values are low and handling cost improvements can be realized in sourcing countries. Location and dispersion of suppliers is not a focus in this research. It is interesting to see if such a taxonomy for selection of the network configurations can be based upon different factors, such as an effort rating.

(11)

The focus of operations research literature in the last two decades has been primarily on the disadvantages of holding inventory, particularly supply chain responsiveness would be compromised. Theories, like the well-known Lean philosophy, aim towards the reduction of inventory as it is seen as waste. Contrastingly, the study of Baker (2007) highlighted the importance of buffer inventory and the role of warehouses. This was caused mainly due to another important trend in the past decades, which is globalization and its effects on firms. Lead-times and lead-times variability tend to increase in global supply chains, therefore it is difficult to eliminate the need for inventory completely (Baker, 2007). Baker (2007: 23) stated:

“There has been little research into the recognition of the need for inventory in many situations and how companies should identify the ‘word-class best-practice’ of inventory.”

(12)

3. THE FRAMEWORK FOR SUPPLY CHAIN REDESIGN

A framework is presented which guides the reader through the decision making process when redesigning the supply chain. The framework looks at the decisions that stem from this supply chain structure change in a two-fold manner. First of all, decisions have to be made on a strategic level, which concerns the ‘bigger picture’. Here, it should be decided if a supply chain structure change is suitable for the firm in the first place. The second part of the framework is designed to test the proposed structure changes. This part looks at the operational level of such a change. Decisions have to be made regarding inventory and distribution policies. A total cost analysis is conducted to find if a structure change is viable. Figure 1 presents the decision support framework for redesigning the supply chain based on current literature.

(13)

Main objective of the framework is to provide managers a decision support tool. With this tool they can determine and evaluate if a possible redesign is worth the effort. All steps of the framework are illustrated by the case company.

3.1 Step 1 - Strategy

First of all, assumed is that inventory is needed. Global supply chains are characterized by explicitly long lead-times which means that, unless a firm operates in a make-to-order fashion, demand needs to be fulfilled from inventory.

Most firms already pursue a company-wide or supply chain strategy when planning for changes or improvements in their supply chain. Other firms change their strategy in an attempt to realize improvements. Inventory and distribution should be an important factor in this strategy, unless they possess little cost or performance impact. In the following section the development of a supply chain strategy is illustrated with inventory and distribution considerations. Most strategies evolve from a products’ business model, therefore product type implications with regard to strategy are discussed after this section.

Stock, Greis, & Kasarda (1998) defined three main elements to determine a firm’s business strategy with close consideration of logistics and network structure. The first element is setting

competitive priorities. There are several areas in which a company can choose to excel in fulfilling

customer demand. The most traditional are; cost, quality, flexibility, and delivery. The second element of setting a strategy is competitive scope. Nowadays, strategic management literature beliefs firms can choose more than one competitive priority as its strategy. Choosing two or more priorities can have conflicting implications for a supply chain. First of all, choosing more than one priority could be beneficial when they complement each other. However, it could also lead to making strategic trade-off decisions, see table 1. The last element of determining a strategy is

geographic scope. Geographic scope relates to the dispersion of a firm’s market area (Stock et al.,

(14)

Table 1. Three main elements of a business strategy illustrated.

Case company Competitive

priorities

The most important priorities are cost and delivery. These translate into the objectives of lowering inventory cost and increasing service levels.

Competitive scope

The base case firm has two important priorities set. These priorities cannot be handled separately as they influence each other. When searching for methods to decrease inventory cost, the firm simultaneously tries improving inventory control so that it can increase service levels. Would it only have set one priority (e.g. cost), it would try lowering cost of delivery too, and thereby possibly harm service levels. However, focusing too much on increasing service levels could lead to exponential cost increases. Thus, a trade-off has to be made between decreasing inventory cost and increasing service levels.

Geographic scope

The base case firm is a small sized enterprise. Its geographical dispersion is low as their market is located in only one small area of a continent. Would the firm have set priority to increase its market reach, it should rethink if the current network structure is capable of serving this potential market. Nonetheless, suppliers are located overseas which makes managing this global supply chain increasingly complex.

3.1.1 Strategic implications of product type

Additionally, strategic choices depend heavily on which type of products are sold. Here, a product taxonomy, like the one described by Christopher, Peck, & Towill (2006), can be useful in search for a strategy. For example, high value products, like monitors or tv-screens, can be relatively easily stored. Nevertheless, producers tend to store and transport these products as little as needed, due to risks related with theft and fragility. Here, a strategy which includes adding a hub gets increasingly complex. Investing in a self-owned warehouse is often the only option.

Some products are associated with different risks or opportunities. For example, if products become obsolete very fast, it is inadvisable to hold inventory for a longer period of time. Other products have special requirements, such as food products which need to be stored at a certain temperature. A strategy which involves opening an additional consolidation hub or warehouse can be inappropriate for certain product types.

(15)

demand patterns (Stock et al., 1998). Level of stability and relative impact of these factors also have to be considered. These factors are hard to measure in general. Additionally, it can help to gain insight in a firm’s competitive environment. See table 2 for further illustrations.

Table 2. Strategic implications illustrated.

Strategic implications Case company implications Product type Determine if a product or product range is

suitable for a proposed strategy and structure.

Products sold by base case company are commodity goods. This means they are easily stored and transported.

Risks or

opportunities associated with product type

Search for the possibility of any risk occurring when pursuing a certain strategy. Additional opportunities could also arise with certain strategies.

Inventory visibility decreases because the new warehouse is placed and managed overseas, far away from headquarters.

Salvage value of the products sold by base case firm increases when pursuing the proposed strategy.

Competitive environment

Know market trends, listen to customer requirements and specify demand patterns.

Demand patterns are hard to determine, due to the high volatility market characteristics. Over the entire product range, market is growing. Products are complementary to the smartphone industry. Business prospects look positive.

3.2 Step 2 – Setting tasks and objectives

(16)

Table 3. Illustration of task and objective setting.

Competitive priorities Areas of improvement Case company strategic tasks Reduce overall inventory

cost (cost)

No, ore inadequate, inventory control methods, like order policies applied.

Set-up and test appropriate inventory control methods. Compare inventory control cost in current and proposed network structure.

Improve service levels (delivery)

The availability of products must increase. Chances of stock-outs must decrease.

Expand the network structure with a consolidation hub close to suppliers

Noticeably, in table 3 there are several tasks possible in order to achieve the proposed strategic objectives. For example, where one company chooses for a different sourcing strategy, another option can be to improve and increase engagement with suppliers. Both options were not applicable for the case company, therefore a novel approach to improve lead-times had to be searched for. Now, some strategic tasks do not need a structure change. The framework provides the possibility to stop hereafter.

3.3 Step 3 – Choosing alternative network structure(s)

Now that a strategy is set, a convenient structure is to be developed next. Stock et al. (1998) suggest that a fit between environment, strategy, structure and logistic capabilities will enhance overall organizational performance. Therefore, network structure should align with priorities and objectives of the strategy, suitable for a designated environment. This paper suggests that not only logistics capabilities are of importance for increased performance. In addition, stock allocation plays a vital role. A total cost analysis will determine if the redesign of the supply chain structure will impact the supply chain positively.

3.3.3 Current structure

(17)

somewhere in-between 9 and 35 days. The firms’ product catalogue exists for about 90% out of these less common products, one can imagine the problems stemming from ordering these. The current structure of case company can be visualized in figure 2.

Figure 2. The current supply chain network structure of the case company.

3.3.4 Proposed structure

(18)

Figure 3. Case company proposed supply chain structure.

Table 4 shows all the advantages and disadvantages of the proposed supply chain structure. All aspects are mentioned and explained elsewhere in this paper.

Table 4. (Dis)Advantages of the proposed supply chain structure.

Advantages Disadvantages

More stable supply stream towards the Netherlands Decreased inventory visibility Bundling of orders to save transport cost Extra handling cost

Easier to sell superfluous stock to 3rd_{party buyers} _{Increase of sunk and hidden cost}

Damaged or incomplete orders can be returned more easily Possibility of assembly postponement

3.3.5 Evaluate relevant cost variables

An in depth cost evaluation can be found in Appendix A.

There are several Key Performance Indicators (KPI) taken from the cost evaluation and investigated further. These KPIs will be the main performance variables to test and see if cost improvements occur. These KPIs are;

- Lost sales cost

- Total inventory and transport cost

(19)

If these KPIs turn out to be positive, a ‘fit’ between structure and strategy can be identified. The last two steps of the framework are explained in the following sections.

3.4 Step 4 – Setting appropriate policies

The main goal of the last two steps of the framework is to check if the most important criteria set at the strategy phase, the competitive objectives, are a ‘fit’ with the network structure. Structure is determined in the previous section and operationalized with appropriate inventory and transportation policies. If a strategy and structure ‘fit’, then they are expected to have a reasonable positive effect in terms of performance outcomes (Miles & Snow, 1984). Figure 4 shows this phenomenon conceptualized.

Figure 4. The strategy and structure fit conceptualized.

(20)

Table 5. The expected performance outcomes per type of fit.

Type of fit Expected performance outcomes (KPI)

No fit - almost no considerable cost improvements, ≤ 5%. Cost improvements are not a considerable part of total cost, ≤ 5%.

- too much hassle to implement structure change, sunk cost are equal or more compared to cost improvements. It takes ≥ 5 years to compensate sunk cost. - business environment or product type not suitable for structure change Moderate fit - some cost improvements, ≥ 5% ≤ 20 %. The cost improvements are a

considerable part of total cost, ≥ 5% ≤ 10%.

- a few aspects give considerable difficulties when changing structure. It takes ≥ 3 to ≤ 5 years to compensate sunk cost.

- business environment and/or product type are somewhat suitable

Full fit - major cost improvements, ≥ 20%. The cost improvements are a big part of total cost, ≥ 10%.

- almost no difficulties when implementing structure change. It takes ≤ 3 years to compensate sunk cost.

(21)

4. METHODOLOGY

The purpose of this section is to set-up a method to test several inventory and transportation policies and to find out what impact they have on total cost (the last two phases of the framework, figure 1). A representation of the real system is turned into an appropriate model. The aim of any model should be to give researcher and managers a sense of how the real-life system behaves (Robinson, Brooks, Kotiadis, & van der Zee, 2011). Therefore, the model in this paper tries to represent reality as much as possible, with the amount of resources it has available. Additionally, the researcher can take in multiple factors in the model, which would be too costly and time consuming if it were to be tested in the real system (Neetu, 2011).

Robinson (2008) mentioned several requirements for a ‘good’ model. First of all, the model should produce sufficiently accurate results for the purpose (validity). In this research, this is understanding how the different cost components act and react, and giving assurance they can be used confidently to base decisions on. Secondly, the results should be believed by the readers and the clients (credibility). The representation of the results should be in such a way that the reader can understand it immediately. Thirdly, the model should be feasible to build within the constraints of the available data and time. This study has a limited amount of time because it is performed within one semester. It is also a ‘one-man job’ and thereby constrained by its resources. Lastly, the model must be useful, which means it has to be; easy to use, flexible, visual and quick to run.

The dataset used is provided by the case company. Data of all products from the past four year are recorded. Lead-times were not recorded, indication of lead-times are made based on managerial experiences and indications. In order to simplify the model, all products which had ≥1 demand in the past four years are included in the model. All products are ‘bundled’, as no further specific product type research is conducted.

4.1 The model

(22)

bound cost performances are showed. Additionally, alternative scenarios are sketched. These scenarios are based on data and information gathered from the base case firm. This firm is a small-sized e-commerce retail business without any physical stores operating in the Netherlands.

Assumptions

 Pipeline inventory is not counted towards the total holding cost function. Pipeline inventory makes no considerable impact due to the relative low holding cost factor.

 Transport from the warehouse in the Netherlands to the customer is not affected by the structure change.

 Customers are only served from the warehouse in the Netherlands.  All shortages count as lost sales, thus no backordering allowed.

All expressions which calculate units are rounded up to the nearest integer. The following notation is used to develop expressions:

𝐷 = 𝑒𝑥𝑝𝑒𝑥𝑡𝑒𝑑 𝑑𝑒𝑚𝑎𝑛𝑑 𝑝𝑒𝑟 𝑦𝑒𝑎𝑟 (𝑖𝑛 𝑢𝑛𝑖𝑡𝑠) 𝜇 = 𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑑𝑒𝑚𝑎𝑛𝑑 𝑝𝑒𝑟 𝑡𝑖𝑚𝑒 𝑢𝑛𝑖𝑡 𝑐 = 𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑝𝑢𝑟𝑐ℎ𝑎𝑠𝑖𝑛𝑔 𝑐𝑜𝑠𝑡 𝑝𝑒𝑟 𝑢𝑛𝑖𝑡 (𝑎𝑙𝑠𝑜 𝑐𝑎𝑙𝑙𝑒𝑑 𝑢𝑛𝑖𝑡 𝑐𝑜𝑠𝑡) 𝑝 = 𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑟𝑒𝑡𝑎𝑖𝑙 𝑝𝑟𝑖𝑐𝑒 𝑝𝑒𝑟 𝑢𝑛𝑖𝑡 (𝑎𝑙𝑠𝑜 𝑐𝑎𝑙𝑙𝑒𝑑 𝑠𝑎𝑙𝑒𝑠 𝑝𝑟𝑖𝑐𝑒) ℎ = 𝑎𝑛𝑛𝑢𝑎𝑙 𝑢𝑛𝑖𝑡 ℎ𝑜𝑙𝑑𝑖𝑛𝑔 𝑐𝑜𝑠𝑡 (𝑖𝑛 𝑒𝑢𝑟𝑜𝑠 𝑝𝑒𝑟 𝑢𝑛𝑖𝑡 𝑝𝑒𝑟 𝑦𝑒𝑎𝑟) 𝑜 = 𝑓𝑖𝑥𝑒𝑑 𝑐𝑜𝑠𝑡 𝑜𝑓 𝑝𝑙𝑎𝑐𝑖𝑛𝑔 𝑎𝑛 𝑜𝑟𝑑𝑒𝑟 (𝑖𝑛 𝑒𝑢𝑟𝑜𝑠) 𝑟 = 𝑟𝑒𝑜𝑟𝑑𝑒𝑟 𝑝𝑜𝑖𝑛𝑡 (𝑖𝑛 𝑢𝑛𝑖𝑡𝑠), 𝑡ℎ𝑖𝑠 𝑖𝑠 𝑎 𝑑𝑒𝑐𝑖𝑠𝑖𝑜𝑛 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 ℓ = 𝑟𝑒𝑝𝑙𝑒𝑛𝑖𝑠ℎ𝑚𝑒𝑛𝑡 𝑙𝑒𝑎𝑑 𝑡𝑖𝑚𝑒 (𝑖𝑛 𝑤𝑒𝑒𝑘𝑠) 𝑠 = 𝑠𝑎𝑓𝑒𝑡𝑦 𝑠𝑡𝑜𝑐𝑘 𝑄 = 𝑟𝑒𝑝𝑙𝑒𝑛𝑖𝑠ℎ𝑚𝑒𝑛𝑡 𝑞𝑢𝑎𝑛𝑡𝑖𝑡𝑦 (𝑖𝑛 𝑢𝑛𝑖𝑡𝑠), 𝑡ℎ𝑖𝑠 𝑖𝑠 𝑎 𝑑𝑒𝑐𝑖𝑠𝑖𝑜𝑛 𝑣𝑎𝑟𝑖𝑎𝑏𝑙𝑒 𝑘 = 𝑐𝑜𝑠𝑡 𝑜𝑓 𝑎 𝑙𝑜𝑠𝑡 𝑠𝑎𝑙𝑒 (𝑖𝑛 𝑒𝑢𝑟𝑜𝑠)

The order quantity is calculated with the well-known EOQ formula

(23)

The reorder point is calculated by

r = μℓ + 𝑠

where, both µ and ℓ are expressed in the same time-units. The demand during lead time, ℓ, is assumed to be continuous and normally distributed, then safety stock is calculated by

𝑠 = 𝑧_𝛼𝜎√ℓ

where,

 𝛼 is the service level percentage, and 𝑧𝛼 is the inverse distribution function of a standard normal distribution with cumulative probability; for example, 𝑧_𝛼 = 1,65 for a 95% service level.

(24)

5. EXPERIMENTS

In this section, the model from the previous section is used to calculate inventory, transportation cost and salvage revenue functions. An EOQ scenario and a ‘realistic’ scenario are sketched to indicate lower and upper bounds of where total cost can lie in-between. This is done to give indications and insights in the cost of adding an inventory hub (IH) in the supply chain network. The results section hereafter will present the cost improvements, if present.

5.1 Step 5 – Testing performance variables

The last step of the framework is conducted in the following paragraphs.

5.1.1 Current structure - EOQ scenario

Inventory cost

The firm has normally distributed demand for smartphone cases with an annual demand of 𝐷 = 75.000 units. This means an average weekly demand of μ = 1443 units. Assumed is that the standard deviation is 𝜎 = 100 units. A lead time of five weeks is used to calculate the total cost function. Five weeks is the maximum it takes for an order to be produced, shipped and delivered to the DC in Europe. This seems reasonable because the holding cost factor has a relatively small impact in this case, compared to the other types of cost.

A smartphone case, on average, costs 𝑐 = 3 €. Annual holding cost per unit in the Netherlands is ℎ = 2,08 €, and ℎ = 0,04 € are the weekly holding cost. Order cost are 𝑜 = 60 € per order. A reasonable order quantity can now be determined:

𝐸𝑂𝑄 = √2(60)(75.000)

2,08 = 2081

Here, safety stock is calculated based on a 98% service level:

(25)

where, 𝑧𝛼 = 2,05 for a 98% service level. Now, the reorder point can be calculated by

𝑟 = 1443(5) + 460 = 7675 𝑢𝑛𝑖𝑡𝑠

A total cost function for the inventory part can be calculated by the sum of the annual order cost, holding cost, and the cost of all lost sales.

𝑇𝑜𝑡𝑎𝑙 𝑖𝑛𝑣𝑒𝑛𝑡𝑜𝑟𝑦 𝑐𝑜𝑠𝑡 = (𝑜𝐷

𝑄) + ℎ ( 𝑄

2+ 𝑠) + 𝑘𝐷(1 − 𝛼)

where,

 the total purchasing cost is not included because it is not relevant for now. Only results with the same total purchasing cost are being compared.

 𝑘 is the cost of a lost sale in euro’s. This value is determined by decreasing the average retail price by the average purchase cost per unit. This function indicates the cost due to lost profit, loss of future sales, loss of reputation, and possibly other intangible losses. These are all difficult to measure and because the case company sets high strategic priority towards service levels, this value is relatively high. Here, 𝑝 is €17,- and 𝑐 is €3,- thus,

𝑘 = 𝑝 − 𝑐 = 17 − 3 = €14, − 𝑝𝑒𝑟 𝑙𝑜𝑠𝑡 𝑠𝑎𝑙𝑒

If, 𝑜 is €60,- per order, ℎ is €2,08 per unit per year, and a service level of 98% is maintained. This makes the total inventory cost, consisting of order, holding, and the cost of lost sales:

𝑇𝑜𝑡𝑎𝑙 𝑖𝑛𝑣𝑒𝑛𝑡𝑜𝑟𝑦 𝑐𝑜𝑠𝑡 = (60)75000 2081 + (2,08) ( 2081 2 + 460) + (14)(75000)(0,02) = €26.283,46 Transportation cost

(26)

is equally divided over all five suppliers. With an air freight delivery cost ratio of €200,- per shipment (relatively small-sized, low-weight, and low valued products), the total transportation cost per year are,

𝑇𝑜𝑡𝑎𝑙 𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡𝑎𝑡𝑖𝑜𝑛 𝑐𝑜𝑠𝑡 = (5)(37)(200) = €37.000, −

Salvage revenue

In the EOQ scenario, all demand is fulfilled. This means there are no items left to be salvaged and no salvage revenue is received.

5.1.2 Current structure - Realistic scenario

The previous paragraph sketched the optimal version of the current situation. The actual inventory control situation looks quite differently. Currently, inventory levels are checked every week. Additionally, emergency orders are made every month. Demand is assumed to be the same, 𝐷 = 75.000. Now, 20.000 items are salvaged throughout the year. These products have to be ordered, held and transported to the Netherlands. Orders are placed,

(52 𝑤𝑒𝑒𝑘𝑠 ∗ 1 𝑜𝑟𝑑𝑒𝑟) + (12 𝑚𝑜𝑛𝑡ℎ𝑠 ∗ 1 𝑜𝑟𝑑𝑒𝑟) = 64 𝑜𝑟𝑑𝑒𝑟𝑠 𝑝𝑒𝑟 𝑦𝑒𝑎𝑟

Salvage revenue

Currently, salvage revenue is almost non-existent. When a product gets labelled as unsellable, at the end of its product life cycle, it has almost no salvage value left. Because of the multi-product nature of the inventory system, inventory levels rose steadily. If a large TPB is willing to buy a large amount of superfluous inventory in the Netherlands, then salvage values lie between 2 and 5 percent of the unit cost. It is assumed this TPB buys 20.000 superfluous units throughout the year.

(27)

This gives total inventory and transportation cost of,

𝑇𝑜𝑡𝑎𝑙 𝑐𝑜𝑠𝑡 = 3.840 + 21.144,40 + 73.500 + 64.000 = €160.684,40

where,

 𝑄 =75.000+20.000

64 = 1485 𝑢𝑛𝑖𝑡𝑠, order cost are,

𝑡𝑜𝑡𝑎𝑙 𝑜𝑟𝑑𝑒𝑟 𝑐𝑜𝑠𝑡 = (60)(64) = €3840, −

 throughout the year, 20.000 superfluous units are sold at the highest salvage value possible. Then, the maximum inventory levels of 20.000 items, gives a yearly holding cost of,

𝑡𝑜𝑡𝑎𝑙 ℎ𝑜𝑙𝑑𝑖𝑛𝑔 𝑐𝑜𝑠𝑡 = (2,08 (20.000 + 1485

2 )) − €1.200, − = €21.144,40

 the current service levels are 93%. This gives a lost sales cost function of

𝑡𝑜𝑡𝑎𝑙 𝑐𝑜𝑠𝑡 𝑜𝑓 𝑙𝑜𝑠𝑡 𝑠𝑎𝑙𝑒𝑠 = (14)(75000)(0,07) = €73.500, −

 in both the optimal and actual situation, the amount of pallet space in the airplane never exceeds one. Then, transportation cost only grow because the amount of orders increases. Total transport cost are,

𝑇𝑜𝑡𝑎𝑙 𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡𝑎𝑡𝑖𝑜𝑛 𝑐𝑜𝑠𝑡 = (5)(64)(200) = €64.000, −

5.1.3 Proposed structure – EOQ scenario

Inventory cost IH

(28)

Here, order quantity and safety stock are,

𝐸𝑂𝑄 = √2(60)(75.000)

1,04 = 2942 𝑢𝑛𝑖𝑡𝑠

𝑠 = 2,05(100)(√5) = 460 𝑢𝑛𝑖𝑡𝑠

Customers are not served from the overseas warehouse, thus lost sales are counted at the warehouse closest to the customers. This is where the ultimate performance of the complete system is determined. 𝑇𝑜𝑡𝑎𝑙 𝑖𝑛𝑣𝑒𝑛𝑡𝑜𝑟𝑦 𝑐𝑜𝑠𝑡 = (60)75000 2942 + (1,04) ( 2942 2 + 460) = €3537,81 Inventory cost DC

A pallet with items is send towards the Netherlands two times per week. Only items available at the overseas stock point can be ordered. This means safety stock in the Netherlands can be reduced and inventory is replenished with shorter intervals. Assumed is that there still will be a small amount of order cost. This number accounts for both the extra time spent ordering and the increased handling cost because of the existence of two warehouses, 𝑜 = €10, −. Here, the average order size is,

𝑄 = 75.000

(52)(2)= 722 𝑢𝑛𝑖𝑡𝑠

Safety stock are now drastically decreased because of the shorter lead times

(29)

This leads to the following total cost function, 𝑇𝑜𝑡𝑎𝑙 𝑖𝑛𝑣𝑒𝑛𝑡𝑜𝑟𝑦 𝑐𝑜𝑠𝑡 = (10)75000 722 + (2,08) ( 722 2 + 145) + (14)(75000)(0,02) = €23.091,26 Transportation cost

In the proposed situation, orders are still derived from five different suppliers. Now, suppliers ship their orders to the IH, which is relatively close. At the IH, orders are either consolidated and directly shipped to the DC in the Netherlands, or stocked for a longer period of time. Thus, suppliers will still charge a small amount of transportation cost. In addition, transportation cost are incurred from shipping items from the IH towards the DC. Delivery cost charged by suppliers are now €5,- per order, and shipping cost of overseas transportation stays the same (€200,- per shipment). Overseas transport is planned two times per week. Then, transportation cost are,

𝑇𝑜𝑡𝑎𝑙 𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡𝑎𝑡𝑖𝑜𝑛 𝑐𝑜𝑠𝑡 = ((5)(26)(5)) + ((104)(200)) = €21.450, −

Salvage revenue

At the EOQ scenario there is no salvage revenue received, because all demand is assumed to be fulfilled.

5.1.4 Proposed structured – Realistic scenario

In this scenario, there are 20.000 items ordered more, for the same reasons as mentioned in the current structure. Orders are send to the suppliers one time per week, and every month one extra emergency replenishment is made.

Salvage revenue

(30)

If, throughout the year, a large TPB is willing to buy a large amount of superfluous inventory at the IH, then salvage values lie between 20 and 10 percent of the unit cost. It is assumed this TPB buys 20.000 superfluous units per year.

𝑀𝑖𝑛𝑖𝑚𝑢𝑚 𝑠𝑎𝑙𝑣𝑎𝑔𝑒 𝑟𝑒𝑣𝑒𝑛𝑢𝑒 = (0,1)(3)(20.000) = €6.000, − 𝑀𝑎𝑥𝑖𝑚𝑢𝑚 𝑠𝑎𝑙𝑣𝑎𝑔𝑒 𝑟𝑒𝑣𝑒𝑛𝑢𝑒 = (0,2)(3)(20.000) = €12.000, −

This gives total inventory and transportation cost of

𝑇𝑜𝑡𝑎𝑙 𝑐𝑜𝑠𝑡 = 3.840 + 5172,20 + 73.500 + 22.400 + 2.091,26 = €107.003,46

where,

 𝑄 =75.000+20.000

64 = 1485 𝑢𝑛𝑖𝑡𝑠, order cost are,

𝑡𝑜𝑡𝑎𝑙 𝑜𝑟𝑑𝑒𝑟 𝑐𝑜𝑠𝑡 = (60)(64) = €3840, −

 throughout the year, 20.000 superfluous units are sold at the minimum salvage value possible. Then, the maximum inventory level of 20.000 items, gives a yearly holding cost of,

𝑡𝑜𝑡𝑎𝑙 ℎ𝑜𝑙𝑑𝑖𝑛𝑔 𝑐𝑜𝑠𝑡 = (1,04 (20.000 + 1485

2 )) − €6.000, − = €5.172,20

 the current service levels are 93%. This gives a lost sales cost function of

𝑡𝑜𝑡𝑎𝑙 𝑐𝑜𝑠𝑡 𝑜𝑓 𝑙𝑜𝑠𝑡 𝑠𝑎𝑙𝑒𝑠 = (14)(75000)(0,07) = €73.500, −

(31)

𝑇𝑜𝑡𝑎𝑙 𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡𝑎𝑡𝑖𝑜𝑛 𝑐𝑜𝑠𝑡 = ((5)(64)(5)) + ((104)(200)) = €22.400, −

(32)

6. RESULTS

This section presents the main results gathered from the model in the previous section.

In every table, 1 represents the maximum cost level identified. All the other results are compared to that base number. Only the total improvement row represent the improvement calculated by comparing the numbers from the corresponding column. For example, at the EOQ scenario in the current structure, calculations result in a total cost of €100,-. At the EOQ scenario in the proposed structure, calculations result in a total cost of €50,-. This implies a total cost improvement of 50%. The reader can look up the actual cost numbers in Appendix B, table 9. For every scenario, calculations of two different service levels are made (93 and 98 percent).

Table 6 shows the cost improvements at a 93% service level. These numbers do not include total unit cost. Improvements between 13 and 34 percent can be realized changing from the current to proposed structure with a 93% service level.

Table 6. Cost improvements at the 93% service level.

Total cost EOQ scenario Average Realistic scenario

Current structure 37 orders/75.000 units 64 orders/95.000 units

Service level 93% 0.71 0.86 1

Proposed structure 26 orders/75.000 units 64 orders/95.000 units

Service level 93% 0.62 0.64 0.66

(33)

Table 7 shows the cost improvements at a 98% service level. These numbers do not include total unit cost. Improvements between 24 and 50 percent can be realized changing from the current to proposed structure with a 98% service level. The results with a 98% service level show bigger improvements compared to the 93% service level improvements.

Table 7. Cost improvements at the 98% service level.

Service level 98% 0.58 0.79 1

Service level 98% 0.44 0.47 0.5

Total improvement sl 98% 24% 41% 50%

Table 8 presents the total cost improvements at a 98% service level. These number include total unit cost. Total cost improvements between 5.3% at an EOQ scenario, and 14.1% at a realistic scenario can be realized changing from the current to proposed structure.

Table 8. Total cost improvements at the 98% service level including total purchasing cost.

0.73 0.87 1

0.69 0.78 0.86

Total improvement 5.3% 10.4% 14.1%

All the scenarios result in inventory and transport cost improvements when changing to the proposed structure. The results of the scenarios at the proposed structure show a smaller variation between each other, compared to the scenarios at the current structure.

(34)

Appendix B, figure 6 and 7 show the distribution of total cost at each structure type and service level. The lost sales and transport costs are the main contributors to total cost. When changing to the proposed structure, shipment size increases, on average, from 406 to 722 items per overseas shipment.

6.1 Potential alternatives

Next to the presented scenarios in the previous section, there are some other options worthwhile considering. These solutions include somewhat more conservative options in contrast to the previous ‘all-in’ scenarios. Most of the implications stemming from the cost calculations are applicable to these alternatives.

Pure consolidation hub

One alternative scenario can be to keep no inventory at the IH. Orders from suppliers are cross-docked directly for shipping to the DC. Orders are moved through warehouses without stocking them. All orders gathered at the cross-dock are consolidated and shipped at a fixed time interval and thereby realizing economies of transport (Apte & Viswanathan, 2000). However, lead-times can increase if time intervals are set too wide. This problem could be tackled by investing transport cost reductions made by cross-docking into more and bigger orders, at the cost of some holding cost in the Netherlands.

Selective stocking

(35)

can be decided to be more conservative in transporting these product to the Netherlands. Demand of certain item groups can be checked after every, 𝑋_𝑡, period of time. For example, taking a product life cycle of four years, every four to six months.

There is a wide range of criteria whereby a firm can select products to be stocked overseas. For example;

Lead-time variability – When lead-times vary between products or product groups, firms can

choose to select those products with increased variety to be stocked at the overseas warehouse. Increasing safety stocks at IH can aid the availability of these products

Holding cost variety – Firms can choose to stock their items according to their holding cost

characteristics. Holding cost can also vary between warehouses.

Product value – If there are products with exceptional high product values, a firm can select these

to be stored at the warehouse with the best inventory visibility, the warehouse managed by the most trustful 3PL, the warehouse with the best insurances regarding theft or damaged goods, or with the most advanced security equipment and protocols.

Deterioration/perishability rate – When products deteriorate fast, certain conservation procedures

have to be executed which will increase holding cost. Management should carefully take such product characteristic into consideration when making decisions regarding storage.

Fraction of purchasing expenditure – Firms can select items based on the fraction of purchasing

(36)

7. CHANGING THE SUPPLY CHAIN

The next step is translating these results into a decision regarding the earlier mentioned ‘fit’ between strategy and structure. All indicators point towards a ‘full fit’. The proposed structure, together with the targeted strategy, positively influence performance indicators. Service levels, inventory and transport cost performance all benefit (some marginally) from this structure change. The product type is suitable. Strategy and structure can at least considered to be a moderate fit and therefore the actual structure change can be started.

7.1 Insights from practice

First of all, a firm should check if it possesses sufficient resources available to invest in the structure change. Sufficient funds are needed for a transition period, wherein expenses will increase drastically. A conservative investment of €50.000,- takes at least four years to be returned considering average cost improvements. Next to the investment, which can be calculated beforehand, a manager at the case company indicated that the firm probably incurs extra costs at every level of the firm. These ‘hidden’ cost are very problematic to measure and thereby being nearly impossible to calculate in advance. For example, management implemented a ‘pre-order’ option at their website in an attempt to get more insight in future demand. Suppliers were not able to send the correct products in-time. Costly backorders had to be made and the service desk had to inform customers about the delays. Events like these will presumably happen more often.

(37)

The location of the 3PLs’ warehouses can be an important factor when choosing the right 3PL. A relevant consideration can be deciding between warehousing nearby suppliers or closer to the main port or airfield. Local transport cost from the suppliers to the IH are relatively low for the case company. For these reasons they chose for a 3PL located nearby an airfield.

Alternatively, firms can try a form of co-opetition. Activities performed at a greater distance from customers could be interesting for cooperation and activities closer to the customers remain competitive (Bengtsson & Kock, 2000). For example, industry peers from Europe could share warehouse space overseas and bundle transport towards Europa. If firms share storage space, it would be relatively easy to set-up a web-based platform where industry peers can offer their superfluous products and thereby creating a mini-marketplace.

(38)

8. DISCUSSION

This section discusses several interesting points from the results section.

Pooling inventory closer to the source of lead-time variability, the suppliers, gives a considerable amount of cost improvements (≥ 13% in this case). The model used here is supposed to give rough estimations as the ultimate decision, to add the IH or not, depends on additional factors.

First of all, a cutback of 10.4% in total cost (unit cost included) on average, can be considered worthwhile. Transport cost is one of the main contributors to cost. As the amount of suppliers and orders increase, consolidating orders to benefit from economies of transport becomes more profitable. That said, transport cost is calculated in a simplistic manner. In reality, transport cost are calculated on a weight or size based scale, where discounts occur at certain rates. However, it can be argued that the cost improvements from transport cost calculated by this paper can be retraced with other factors. While transport cost increase with increased order sizes, consolidation of more orders decreases the frequency of overseas transports. This argument is generally underlined by network design literature, like in the study of Graves et al. (2007).

Additionally, managers from the case company indicate that the biggest improvements they observe are stemming from the postponement of assembly. Products from suppliers are stored unboxed and without labels overseas. The firm does not incur transport cost and import tariffs as long as products are stored overseas. They can even share the risk of overstocking with their overseas 3PL which, additionally, works incentivizing for the 3PL.

(39)

Thirdly, the holding cost rate, being relatively low here, gives additional reason to increase inventory to improve demand fulfilment. Especially when the lost sale cost proportion is this big (65% in the proposed structure at a 93% service level). However, it is generally well known that holding too much inventory can harm overall organization performance.

At first, salvage revenue was expected to be a major advantage when changing to the proposed structure. Salvage revenues will arguably never make a substantial impact on total cost. The relatively low average purchase cost and fast depreciation of products make salvage revenues not outstandingly lucrative.

In addition, where backorders were impossible at the current structure, it can be argued that they now become feasible. When a product is unavailable in the Netherlands, it can be backordered and delivered within 2.5 days from the overseas warehouse at increased cost. Backorder cost are relatively smaller compared to the cost of a lost sale. If customers are willing to wait for their orders, backordering becomes a possible option. Firms could decrease the time between transports to the Netherlands to make backordering a more viable option.

(40)

9. CONCLUSION

Main objective of the research was finding out if redesigning for improvement is worth the effort. A framework was developed to support managers with this decision process. An exploratory study with data and information provided by a relatively small e-tail firm operating in a highly volatile environment was used to test the framework. This exploratory study helped finding the missing link mentioned at the end of the discussion. Thus, an additional last step is added in the decision support framework for redesigning the supply chain (see figure 5).

Figure 5. The updated decision support framework of supply chain redesign.

(41)

change, firms should preferably start small. Simple testing and simulations, like mentioned in section 7.1, can provide valuable insights about how a supply chain functions. Noticeably, the coordination and cooperation with a 3PL determines, for the greater part, the overall logistics and inventory performance. The relationship with a party from outside the organization is largely based on mutual trust and previous involvements.

Schwarz, (1989); Bertrand & Fransoo, (2016: 301) mention it is often hard to gain insight from exploratory and partially empirical model-based research. This is because of the amount of structure configurations possibilities, differences between the systems modelled, employment of various improvement approaches or analysis not being precise enough. All these reasons are applicable for this research and therefore makes its generalizability hard to determine.

Opportunities for future research are testing the framework with; 1) more sophisticated models, 2) other types of products, and 3) more differentiated types of industries. For example, pipeline inventory could be added to models to see what impact the cost of pipeline inventory has. In addition, specific methods to perform risk assessments in a redesign situation could be valuable assets to practice.

This research contributes to literature in stressing the strategic importance of redesign decisions. Considerable cost improvements can be made and service levels are presumably highly affected by such decisions. Designing a supply chain, capable of proper operating inventory and logistical control, strengthens the competitiveness of firms operating in global supply chains. The case company made a significant growth transformation during the course of this research, more specifically; 8% sales growth. A lot of different factors could contribute to this number, but it can be argued that the redesign efforts were definitely not harming these numbers.

9.1 Managerial recommendations

(42)

(43)

Appendix A – Cost evaluation

The cost considered in this framework are based on certain standard, well known, assumptions. See Axsäter, chapter 3 (2015: 37) or Rushton et al. (2014: 120) for a more in depth analysis of the various costs parameters.

Holding cost

Holding cost are the costs tied up in ‘holding’ inventory. The capital tied up in inventory is usually considered to be the dominant part of the holding cost. Material handling, storage, damage and obsolescence, insurance and taxes can be the other parts (Axsäter, 2015). This cost function increases significantly as the number of item types (SKUs) and the duration of holding an item increase. Often, duration of holding items increases when demand and/or lead times are stochastic. The holding cost factor can vary, differently for every item type.

At the case company, holding costs are relatively low because of the nature of the product. The item sold by this company can be easily stored and therefore initial holding costs are low. Nevertheless, holding costs play an important role in this case, mainly because of the number of SKUs (currently +/- 8500). Logistics and warehousing are outsourced to a third party logistics provider. This 3PL charges a fixed price per unit and per time, for holding (and handling) an item in their warehouse.

Order cost

(44)

Unit cost

Usually, suppliers charge a price for each unit ordered. In addition to discounts associated with order sizes (the sum of all products ordered), buying a certain amount of one specific product can give quantity discounts too.

Lost sales/back orders/shortage cost

Various costs can occur if a product is demanded but the inventory level is not sufficient to fulfil the order. For whatever reason these shortages happen, the consequences can be significant. As can be recalled from the introduction, for this papers’ particular situation, shortages almost always lead to lost sales. Lost sales do not only mean a loss of the contribution of that particular sale but, additionally, have a negative impact on customer good will (Axsäter, 2015). A customer is more likely to come back to your store after a positive experience and, moreover, your firms’ reputation is at stake. Shortage costs are somewhat more difficult to estimate in a production system, where this can lead to a chain of negative consequences if shortages occur early in the production system. In some industries, backorders or partial backorders are traditional. This is when a customer agrees to wait when his order is backlogged and fulfilled some time later. Backlogging raises extra costs, such as administrative cost and price discounts. This is why service levels are considered as an important factor for a lot of managers and researchers, determining its value is therefore crucial.

Sunk/hidden cost

Investing in a warehouse can be a big investment. It could possibly turn out in losses if the whole project fails. Therefore, whether to invest in a warehouse as a firm, or to contract a 3PL provider temporarily is an important decision. A firm should take into account that every level of the firm is probably going to incur extra cost. Main problem with these cost is that they are hard to keep track of and no real value can be accounted. The sunk and hidden costs in this paper stems from the decision to change the supply chain structure.

Salvage value

(45)

deteriorates or not can influence this value. Fast deteriorating items will decrease the salvage value drastically over time. Postponement is another method to have some form of control over the salvage value. Production postponement means that a firm prefers having inventory of common components over holding finished goods, as components can become finished goods, but finished goods usually cannot become components (Hoek, 1998). Contrastingly, there are products where firms incur cost to get rid of inventory.

For example, by postponing the labelling of products until the products are received at the stock point closest to the customer, all products at stock points before this point can be sold with higher salvage values to third party buyers (TPB). In this case, TPB are potentially interested in this firms products because; a) products are not produced by the original suppliers any more, b) products are produced by the original suppliers, but TPB favour directly available products as opposed to the suppliers longer lead times, or c) the firm can sell superfluous products at reduced prices compared to the original suppliers.

Location associated cost

There are several other types of costs relevant to consider, like currency and tax rates. These cost types often depend on the location of operations. This is beyond the scope of this paper and for more insight in these costs it is recommended to look into the ‘facility location problem’ literature, which focus on finding an optimal number and location for distribution centres or warehouses to serve certain customer areas.

Logistic cost

(46)

Appendix B – Additional results

Table 9 shows the actual cost numbers used to calculate the results.

Table 9. Results showed with actual cost numbers.

Total cost in € EOQ scenario Average Realistic scenario

Service levels 93% 115783 139133.5 162484

Service levels 98% 63283 86633.5 109984

service levels 93% 100579 103791 107003

Service levels 98% 48079 51291 54503

Total improvement 93% sl 13.1% 25.4% 34.1%

Total improvement 98% sl 24% 40.8% 50.4%

Figure 6 shows the distribution of total cost of the current structure at the 93 and 98 percent service level. Unit cost are not included. The labels of each bar are in the same order of the graph.

Figure 6. Distribution of total cost of the current structure at the 93% and 98% service levels.

2% 14% 45% 39%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Distribution of total cost - sl 93%

Order cost Holding cost Lost sales cost Transport cost

3% 20% 19% 58%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Distribution of total cost - sl 98%

(47)

Figure 7 shows the distribution of total cost of the proposed structure at the 93 and 98 percent service level.

Figure 7. Distribution of total cost of the proposed structure at the 93% and 98% service levels.

4% 11% 65% 20%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Distribution of total cost - sl 93%

Order cost Holding cost Lost sales cost Transport cost

8% 20% 35% 37%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Distribution of total cost - sl 98%

(48)

References

Apte, U., & Viswanathan, S. (2000). Effective cross docking for improving distribution

efficiencies. International Journal of Logistics Research and Applications, 3(3), 291–302. Axsäter, S. (2015). Inventory control (Vol. 225). Springer.

Baker, P. (2007). An exploratory framework of the role of inventory and warehousing in

international supply chains. The International Journal of Logistics Management, 18(1), 64– 80.

Bengtsson, M., & Kock, S. (2000). ”Coopetition” in Business Networks—to Cooperate and Compete Simultaneously. Industrial Marketing Management, 29(5), 411–426.

Bertrand, J. W. M., & Fransoo, J. C. (2016). Modelling and simulation. In Research methods for

Operations Management (p. 301). New York: Routledge.

Bretthauer, K. M., Mahar, S., & Venakataramanan, M. A. (2010). Inventory and distribution strategies for retail/e-tail organizations. Computers and Industrial Engineering, 58(1), 119– 132.

Christopher, M., Lowson, R., & Peck, H. (2004). Creating agile supply chains in the fashion industry, 32(8), 367–376.

Christopher, M., Peck, H., & Towill, D. (2006). A taxonomy for selecting global supply chain strategies. The International Journal of Logistics Management, 17(2), 277–287.

Creazza, A., Dallari, F., & Melacini, M. (2010). Evaluating logistics network configurations for a global supply chain. Supply Chain Management: An International Journal, 15(2), 154–164. Davis, T. (1993). Effective supply chain management. Sloan Management Review, 34(4), 35–46. Dolgui, A., Louly, A. O., & Prodhon, C. (2007). Supply planning under uncertainties in MRP

environments: A state of the art. Annual Reviews in Control, Elsevier, 31, 269–279.

Eppen, G., & Schrage, L. (1981). Centralized ordering policies in a multi-warehouse system with lead times and random demand. Multi-Level Production/Inventory Control Systems: Theory

and Practice, 16, 51–67.

Graves, S. C., Bhatnagar, R., & Cheong, M. L. F. (2007). Logistics network design with supplier consolidation hubs and multiple shipment options. Journal of Industrial and Management

(49)

Gumus, A. T., Guneri, A. F., & Ulengin, F. (2010). A new methodology for multi-echelon

inventory management in stochastic and neuro-fuzzy environments. International Journal of

Production Economics, 128(1), 248–260.

Hoek, van R. I. (1998). Reconfiguring the Supply Chain to implement postponed manufacturing.

The International Journal of Logistics Management, 9(1), 95–110.

Hopp, W., & Spearman, M. (2001). Factory Physics: Foundations of manufacturing

management. Cycle.

Iakovou, E., Vlachos, D., & Xanthopoulos, A. (2010). A stochastic inventory management model for a dual sourcing supply chain with disruptions. International Journal of Systems Science,

41(3), 315–324.

Jacobs, R. F., & Chase, R. B. (2013). Chapter 11: Inventory management. In Operations and

Supply Chain Management: The Core. McGraw-Hill.

Kuehn, A. A., & Hamburger, M. J. (1963). A heuristic program for locating warehouses.

Management Science, 9(4), 643–666.

Li, S., Ragu-Nathan, B., Ragu-Nathan, T. S., & Subba Rao, S. (2006). The impact of supply chain management practices on competitive advantage and organizational performance.

Omega, 34(2), 107–124.

Miles, R. E., & Snow, C. C. (1984). Fit, failure and the hall of fame. California Management

Review, 26(3), 10–28.

Neetu. (2011). Simulation and its Applications in Inventory Control. International Journal of

Computer Applications.

Robinson, S. (2008). Conceptual modelling for simulation Part I: definition and requirements.

Journal of the Operational Research Society, 59(3), 278–290.

Robinson, S., Brooks, R., Kotiadis, K., & van der Zee, D. J. (2011). Conceptual modeling for

discrete-event simulation. CRC Press.

Rushton, A., Croucher, P., & Baker, P. (2014). The handbook of logistics and distribution

management: Understanding the supply chain (4th edition). London: Kogan Page

Publishers.

(50)

Srinivasan, S. S., Anderson, R., & Ponnavolu, K. (2002). Customer loyalty in e-commerce: An exploration of its antecedents and consequences. Journal of Retailing, 78(1), 41–50. Stock, G. N., Greis, N. P., & Kasarda, J. D. (1998). Logistics, strategy and structure.

International Journal of Operations & Production Management, 18(1), 37–52.

Treleven, M., & Schweikhart, S. B. (1988). A risk/benefit analysis of sourcing strategies: single vs. multiple sourcing. Journal of Operations Management, 7(3–4), 93–114.

A DECISION SUPPORT FRAMEWORK FOR REDESIGNING THE SUPPLY CHAIN AGAINST UNCERTAINTIES