Master Thesis - University of Tilburg - Tilburg University

Loading...

Master Thesis Evaluating Inventory Collaboration in Healthcare Industry: A System Dynamics Study

Zhongning Li [166875] [BSc. DongBei University of Finance and Economics 2009]

A thesis submitted in Partial fulfillment of the requirements for the degree of Master of Science in Operation Research & Management Science

Faculty of Economics and Business Administration Tilburg University

Supervisor: Prof. dr. ir. J. Ashayeri

Date: Aug 27, 2010

Table of Contents Acknowledgements ........................................................................................................ 3 Introduction .................................................................................................................... 4 Chapter 1 Healthcare Sector .......................................................................................... 7 1.1 Healthcare in Netherland ......................................................................................... 7 1.2 Healthcare Supply Chain ......................................................................................... 8 1.3 Characteristic of Healthcare Products ...................................................................... 9 Chapter 2 Literature Review ........................................................................................ 11 2.1 Collaboration Techniques ...................................................................................... 11 2.1.1 Causes of inefficiency in traditional supply chain .............................................. 11 2.1.2 Inventory collaboration techniques ..................................................................... 12 2.1.3 Literature review: collaboration techniques applied in healthcare industry ....... 15 2.2 System Dynamic Simulation.................................................................................. 17 2.2.1 System Dynamics................................................................................................ 17 2.2.2 Literature review: Application of SD method in supply chain ........................... 17 Chapter 3 Inventory Management in System Dynamic Model ---- Stock Flow Diagram........................................................................................................................ 20 3.1 Traditional Inventory Models ................................................................................ 21 3.1.1 Customer ............................................................................................................. 21 3.1.2 Hospital Department ........................................................................................... 22 3.1.3 Central Warehouse .............................................................................................. 23 3.1.4 Supplier ............................................................................................................... 24 3.2 VMI Model ............................................................................................................ 25 3.2.1 Customer ............................................................................................................. 26 3.2.2 Hospital Department ........................................................................................... 26 3.2.3 Central Warehouse .............................................................................................. 27 3.2.4 Supplier ............................................................................................................... 28 3.3 CPFR Model .......................................................................................................... 29 3.3.1 Customer ............................................................................................................. 29 3.3.2 Hospital Department ........................................................................................... 30 1

3.3.3 Central warehouse ............................................................................................... 31 3.3.4 Supplier ............................................................................................................... 32 3.4 Model Validation ................................................................................................... 32 3.4.1 Equation Checks ................................................................................................. 33 3.4.2 Data Inspection ................................................................................................... 35 Chapter 4 Case Study & Results Analysis ................................................................... 36 4.1 Case Introduction & Data Setting .......................................................................... 36 4.2 Model Configuration .............................................................................................. 39 4.3 Experimental Design .............................................................................................. 40 4.4 Results Illustration ................................................................................................. 42 4.4.1 Base case ............................................................................................................. 44 4.4.2 VMI ..................................................................................................................... 46 4.4.3 CPFR ................................................................................................................... 48 4.5 Comprehensive Analysis ....................................................................................... 50 4.5.1 Base Case ............................................................................................................ 50 4.5.2 VMI ..................................................................................................................... 51 4.5.3 CPFR ................................................................................................................... 52 Chapter 5 Concluding Recommendations ................................................................... 54 5.1 A Better Inventory Management ---- VMI ............................................................ 54 5.2 A Better Inventory Management ---- CPFR .......................................................... 55 5.3 Outsource the Central warehouse .......................................................................... 57 Conclusion and further research .................................................................................. 58 Conclusion ................................................................................................................... 58 Further Improvement ................................................................................................... 59 Reference ..................................................................................................................... 61 Appendix ...................................................................................................................... 65 Appendix 1 Experimental Design Simulation Result .................................................. 65 Appendix 2:Model Program ........................................................................................ 68

2

Acknowledgements Here is the final result of the one year master study at Tilburg University: an applied research master thesis based on the case study data of Pantein Hospital, a health provider in the Netherland. Throughout the master thesis research period, I gradually understood that models are not meant for copying exactly the real world system but rather to symbolize them with many assumptions. This case study gave me an opportunity to see how the academic theories and methods can be applied in the practical problem solving. The person I would like to thank is my supervisor Jalal Ashayeri. Professor Ashayeri has given me great help from selecting the topic, building the model until the revision of the thesis context. I had a hard time with the modeling which took me rather a long time; Professor Ashayeri was very patient and encouraging and helped me go through this period. I would also like to thank my parents and friends. They have given me a lot of supports and cares during the whole 5-month research period. Tilburg, August 27

3

Introduction In the past, supply chain management has hardly ever been the priority to the hospitals. Due to the friendly financial legislation of the government, the inventories of medicals and medicines in many hospitals are quite large which although guarantees a high service level yet not efficient for the whole operation within hospitals (KPMG 2009). As time passes by, healthcare sector is changing fundamentally, with the severe competitions and the narrowing down market, along with the less government interference, hospitals are seeking ways to reduce their costs so as to drive higher revenue or maintain the present competitive level. Adams (1995) highlighted that healthcare costs are under attack by the public because healthcare information management systems are merging and consolidating. Rundle (1997) also argues that bar codes must be implemented on all medical supply packaging to control the healthcare industry’s increasing costs. As the substantial investment in healthcare industry, supply chain especially inventory investments are estimated to be 10% to 18% of net revenues (Nicholson et al. 2004) and 25%-30% of the operational costs (Mustaffa and Potter 2009) for hospital. Thus, any cost savings which can be generated through a more efficient management of supply chain can lead to direct increases in profitability. Researchers have been focusing on the operational issues in healthcare industry during the past twenty years such as inventory management, just-in-time deliveries, maximization of facilities layouts and material management, etc, which are aiming to stem the increase of the cost through integrating each party within supply chain. Inventory management obtains a significant weight in cost reduction which can be proved by many research results and the global trend of healthcare networking and collaboration strategies. In Dutch healthcare performance report 2008 (Westert GP et al. 2008), it is pointed out that better logistics in healthcare sector can result in efficiency saving around 20% to 25% and solely by improving purchase and inventory procedures, around 150 million euro can be saved. Thus, more and more health care providers turn their focuses on the collaborative methods to improve their supply chain performances so as to saving more budgets and attention 4

for their core business—health care. Healthcare sector has a higher requirement than other industry on the demand, service level and supply chain management since this industry relates to people’s health and calls for adequate and in-time medical supplies to patients’ need (Mustaffa & Potter 2009). Applying collaboration techniques within supply chain can reduce misinformation and mistrust due to time delays, distorted demand signals and poor visibility of exceptional conditions from traditional vision of supply chain (Holweg et al. 2005). And nowadays, taking advantage of the new technologies, it is accelerating the race toward the development of new business models and solutions which can improve or even guarantee the highly required service in this industry. As long as each player in the chain is working seamlessly, such cooperation should help the operational system efficient and viable. In this master thesis, inventory collaboration within healthcare supply chain is studied. The thesis posits that collaborative techniques can effectively reduce the large stock level and the variations due to the inefficiency and distorted orders of the traditional supply chain. Using a case study data of Pantein which is a healthcare provider in the Netherland, the thesis applies the system dynamic method to analyze the changes of inventory level and service satisfactions resulted from collaboration techniques. The models are designed in the stock flow diagram by the software of ITHINK version 9.0.2 which is one of the world leading systems thinking software. The main body of thesis includes five chapters. Chapter 1 mainly gives a general idea of the research area and objectives of this master thesis. The healthcare situation in the Netherland is first briefly presented. This then continues with the introduction of healthcare supply chain and the characteristics of healthcare products. In Chapter 2, literatures on healthcare inventory and the system dynamics methods applied in the related industry for supply chain management are reviewed. Then in Chapter 3, system dynamics models are built for traditional supply chain case, vendor management inventory case and the collaborative, planning, forecasting and 5

replenishment case. The model differences are discussed as collaborative techniques advances. The model validation and verification are also stated in this chapter. The case study of Pantein is discussed in Chapter 4. With the data setting, an experimental design is first performed. Then the simulation results both for separate and cross cases are analyzed. In the end, concluding recommendations are made in Chapter 5.

6

Chapter 1 Healthcare Sector This chapter mainly presents the products and logistics characteristics of healthcare industry. Section 1.1 consists of a brief introduction of the healthcare industry situation in the Netherland. Section 1.2 shows the general scheme and the involved parties of the healthcare supply chain. Section 1.3 addresses the characteristics of healthcare products and its similarity with other perishable product in inventory aspects.

1.1 Healthcare in Netherland In the Netherland, the government is not in charge of the day-to-day management of the healthcare system. Private health suppliers are responsible for the provision of services in this area whereas government is only responsible for the accessibility and quality of the healthcare. In the Dutch healthcare Performance Report 2008 (Westert GP et al. 2008), it is pointed out that Dutch health care system is accessible, but further improvement in quality can be made. Dutch legislation once (Borst-Eilers, 2000) states that healthcare organizations should deliver ―top-class care, delivered effective, efficient and patient centered and geared to realistic patient expectations‖ which means professionals and providers are under constant pressure to deliver evidence-based, high-quality treatment (Meijboom 2007). Although radically restructuring Dutch healthcare system led to a decline in growth of cost in 2003 (Meijboom 2007), since 2004 (Wester 2008), health care expenditure has risen annually by 5%, the rate of growth is comparable to that of neighboring countries. Although for many aspects of healthcare like the quality is quite high, as the Dutch Healthcare Performance Report stated, the Netherlands does not excel at an international level. Coordination and cooperation in healthcare industry scores relatively low and the efficiency of healthcare in the Netherlands is not optimal. Because of the growing interest in financial position of healthcare organizations and the gradually withdrawn of government financial support, the healthcare institutions 7

are nowadays required to bear more liabilities for capital costs as the competitions increase severely within the healthcare industry (McDougall et al. 2003). Consequently, healthcare organizations are taking more financial risks which make them urgently search and perform the effective cost reduction methods.

1.2 Healthcare Supply Chain In the healthcare market, the relevant players can be quite easily discriminated. Typically, four kinds of players are involved in the supply chain of healthcare sector. A schematic view of the supply chain can be found in Figure 1.

Supplier

Wholesaler

Central warehouse

Hospital department

Figure 1: Schematic view of the health care supply chain

According to Figure 1, the products which can be categorized as medicines (specialties, generics and druggist medicines) and medical articles (hand gloves, injection needles etc.) are shipped from numbers of manufactories to the wholesaler in large volume. The wholesaler then separates the production batch based on the order of central warehouse and ships them. The central warehouse receives the products and then distributes the order to each hospital department where patients are treated and prescribed. Facing the end user, the hospital department can obtain the most accurate demand information, with which, the forecast of demand is made, together with its own inventory level and adjustment factor, orders are sent back to central warehouse. Similarly, the central warehouse and the wholesaler do the same forecast work and send back orders. For the traditional supply chain, each echelon makes its own 8

forecast simply based on next entity’s history data, thus, the classical bull’s effect and vast waste in the inventory and order cost can be caused. Particularly for the healthcare sector, in order to sustain the high service level as well as coping with the emergency demand in case of massive accident etc, a large amount of inventory are kept in each party, and this multi-echelon inventory management within healthcare supply chain mostly would cost 35% of the total budgets which are spent on holding, supplies and labor (Nathan and Trinkaus 1996). The situation needs improvement so that information within each party can be transmitted efficiently and cost reduction can be made through the collaborative management in the supply chain.

1.3 Characteristic of Healthcare Products The healthcare products mainly include two type products, medical articles and medicines. Medical articles are the necessities for daily treatments and surgeries. As the fast consuming and disposable products, medical articles need to be sufficient to support daily operations. Despite the occasional accidents which can be covered by the safety stock, the regular consumptions of medical articles can be estimated most accurately and no concerns are needed for the inventory quality since they have a relatively long shelf life. This thesis however will be mainly focused on the distribution of medicines which have short shelf life and fluctuant demand. The consumption of medicines is less predictable, thus keeping large inventories is probably the best and simple way for each party to meet the requirements from its downstream party. On one hand, due to the differences of season, disease type, area, habit of doctors and patients requirements etc, the needs for medicines vary, besides, the preparation to cope with the emergency situation is also crucial. On the other hand, because of the specificity of medical products, the consumptions cannot be decided by the end customers’ willingness but rather stays at the doctors’ prescription level which can also be influenced by the local living standard and health conditions of people. For both reasons, it is necessary for healthcare providers to store a large amount of 9

medicines to satisfy the clinic demands yet eventually leads to a longer replenishment cycle, severe capital overstock, complicated medicine flow management, and high management fee. Medicine in principle can be categorized as perishable product which has several particular characteristics other than normal products. For most inventory models, it is assumed that the stock items can be stored indefinitely to meet future demands. However, for perishable product, for instance, red blood cells are unacceptable for transfusion 21 days after it is drawn (Higuchi 2004) , fresh products, meats and other foodstuffs become unusable after a certain time elapse (Minegishi and Thiel 2000), furthermore, for medicines and certain medical articles, due to their limited shelf time of their medical effects, they might be partially or entirely unsuitable for treatment and should be discarded. Therefore, reinforcing effective inventory management and furthermore finishing the circulation from the producer to the end customer within the perishable products lifetime has a significant meaning for each party’s profitability. As the consequences of the efficiency needs, logistic collaboration, particularly the collaborative techniques for inventory as the only static part within supply chain can realize the seamless cooperation and information sharing between both the upstream and downstream players. Accordingly the goal of inventory reduction and quick response to customer demand can be achieved.

10

Chapter 2 Literature Review In this Chapter, literature reviews from two major aspects are presented. In Section 2.1, the initial problem which causes inventory inefficiency is first addressed. This inefficiency derives the different collaboration techniques which are then introduced in this section. These collaborative methods have been applied and researched in many fields, however the last part within this section only reviews the literatures with regards to the implementation in healthcare industry. Section 2.2 first describes the evolution of system dynamic with brief words. Then the review of the researches using system dynamic methods in general and particularly in perishable products supply chain is made.

2.1 Collaboration Techniques 2.1.1 Causes of inefficiency in traditional supply chain It is widely admitted that the traditional supply chain has the shortcomings of inefficiency and cost waste. As a system consisting of producer, supplier, distributor and end customer who are linked via the downstream feed-forward flow of products and upstream feedback flow of orders, each player issues production orders and replenishment without considering the practical situation at other tiers of the supply chain (Schouten 2006). Thus, each level controls its own inventory level and hardly shares information which leads to the results that demands are only estimated based on the purchased orders from downstream, therefore, amplification of orders is made by each player within the supply chain (Towill and Disney 2003). This famous bullwhip effect was first discovered by Proctor & Gamble and can be concluded as the results of low level of information sharing (O’Donnel 2005). The effect is especially obvious when sudden demand occurs. There are various cost saving aspects that need to be considered before place an order such as batch orders, discount, seasonal price variation etc, so when unexpected demand comes, the situation of high material and transportation costs, quality problems and later the problems of excessive inventory as well as the low service level may occur with high probability (Lee and 11

Whang 1997). Due to the special characteristics of the healthcare products as introduced in the previous chapter, bullwhip effect may cause even more negative impacts on the healthcare industry than on the other industries. Almost all researches with regard to this inefficiency suggest that applying collaborative techniques are effective ways to improve each player’s supply chain management. 2.1.2 Inventory collaboration techniques The implementation of collaboration between players within supply chain is based on the depth of their relationships. As the relationship matures, the level of collaboration strengthens. For inventory management, the techniques with the maturity degree between players can be concluded by the following box (Anthony 2004, Holweg et.al 2005).

High

Data Exchange

CPFR

Low

Planning

Inventory Management Techniques

Traditional Supply Chain

Vendor Managed Inventory

Low

High Replenishment

Data exchange Data exchange means that each party still orders independently, yet they exchange demand information and action plans for each others’ forecasting and planning. When information sharing is poor along with the time lag, the inefficiency is caused. The forecast at the supplier level can gain major improvements, under or over estimation of the demand can be removed which leads to the reduction or even elimination of the unnecessary uncertainty (Anthony 2004). Information sharing not only helps players to create a more visible and predictable demand in system, but it also can be regarded as the foundation of the other advanced collaboration techniques. Traditionally, data exchange can be made via telephone, fax and email which are used to set initial communications and negotiations among parties. As the relationship 12

progresses along with the IT development, implementing an information platform can be much more efficient and economic (Holmstrom 2005). Vendor Management Inventory VMI developed during the nineties is regarded as a new way to reach the competitive advantage by organizations. There are substantial and thorough studies on this collaboration technique. The brief idea behind VMI is that the upstream partner takes over all replenishment decisions of the downstream partner (Turhan and Vayvay, 2009). In this case, tradeoffs between either large shipments with less transportation cost or small but efficient inventory policy can be decided by the upstream partner. Since there is a predetermined service level of upstream player, the inventory level of the downstream partner can be managed either by using the traditional Min/Max economic order quantity or the forecasting replenishment models (Towill and Disney 2003). The disadvantage of this technique is that there should be a full commitment between partners to share information especially for the downstream player. Also the situation of small trade volumes or long distance between partners can cause the inefficient replenishment (Schouten 2006). After covering major shortcomings of VMI, the merit of this technique should be pointed out. Since the demand pattern is control by upstream partner, the safety stock at both parties can be reduced and service level can be increased which not only decrease the inventory space and costs but also enhancing the customers’ satisfaction. For the upstream player, clearly knowing the inventory and demand situation of its partners can help to better plan and benefit from their own business with the cost reduction from transportation and inventory which are considered as the two main parts within the supply chain (Holmstrom et al. 2003). As to the downstream players, more benefit can be gained. Since the administration for planning and ordering is taking over by the upstream partner, not only the order as well as the inventory cost can be reduced, but also the administration reduction can be obtained. Furthermore, more focuses will be put on the downstream players’ core businesses and this point is especially important for healthcare sectors since they are not literally deemed as industry or business (Disney and Towil 2003).

13

CPFR As the relationship goes more strategic, collaborative planning, forecasting and replenishment can be applied to strengthen the overall supply chain visibility and management. Different from VMI, the responsibility of replenishment decisions is shared between the upstream and downstream partners when fully commitment of downstream player is not available (Sari 2008). With the joint forecast and agree to parameters of acceptable variances, the downstream demand information is tracked against the joint forecast and when variances are not within the agreed upon tolerance levels, both parties are notified so that the appropriate action can be taken. This proactive approach to the inventory replenishment process calls for close interaction between two parties to help reduce the inventory level (Danese 2005). For the downstream players, this technique can avoid fully exposure towards its upstream partners, whereas for the whole supply chain, adding the procedures of more decision discussing and negotiating is nothing but reducing the efficiency (Chen et.al 2007). This probably is the major flaw of this collaborative technique. In the long-run, however, since more communications have been made between two parties on demand forecasts with the growing mutual trust and deeper acquaintance, mid-long term forecast can be made rather than single deal basis, whilst it leads to a smaller variation of forecast errors together with benefit and efficiency (Holweg 2005). Third Party Logistic (3PL) Third party logistic can be incorporate with any of the techniques mentioned above. The activities within supply chain can be outsourced to a specialized party which has far more experience with divisions of logistics. It is considered to be the most valuable aspect of the third party on its IT information integration and relationship building since requirement of its clients should be accurately and swiftly transferred with security and reliability. Besides, a good IT system can save costs by its automatic process and raise service level by its accuracy and professional process (Jayaram and Tan 2010). For healthcare sector, the trend of outsourcing is pretty obvious in recent years from washing and preparing meals extended to part of division management within supply chain (Nicholson 2002). There are inherent reasons driving this outsourcing trend in 14

healthcare sector. Patients nowadays are getting increasingly informed and demanding and requesting healthcare earlier and more frequently (Sinha and Kohnke 2009). Therefore, healthcare departments need much more attention on their core business than before. Outsourcing certain parts within the hospital supply chain can generate a more efficient inventory management which can directly lead to the cost saving. With the saved capital, hospital can invest more on researches and clinic trails which are related to their core business to meet the patients various demands. Also, it has been studied that the move towards outsourcing of inventory can lead to improved internal performances and furthermore, it has a positive impact on customer satisfaction and customer perceptions of service quality (Nicholson et al. 2002). Moreover, there has been an increase in the expertise and in the number of third-party providers which offer inventory management services in healthcare. These providers have successfully lowered the inventory cost with their professional experiences in the healthcare sector (Roberts 2001). 2.1.3 Literature review: collaboration techniques applied in healthcare industry Over years, attention has been given to the coordination of supply chain, particularly inventory management. This scrutiny in healthcare sector has not escaped with its historically high cost of these organizational managements. Qualitatively and quantitatively methods have been studied with respect to the improvements with healthcare department. In most hospitals and medical establishments, about 35 percent of their budgets are spent on suppliers and the labor to manage the inventories (Nathan 1996). Healthcare industry like any other industries, customers demand lower costs and tailored distribution service, and players all want cost containment through prudent management. Barnes (1987) pointed out that players within supply chain cooperation are going through the fire of providing ―quality service‖ at the ―affordable price‖, however, with the close working partnership, this liaison should help to ensure that the health care delivery system remains viable. Nicholson (2002) suggested ways in which a multi-echelon inventory approach can be adopted by the healthcare industry based on the case study in Jamaica. Results deducted from the non-liner inventory model shows that savings can be gained by using a third party to manage the inventory and distribution of hospital supplies. 15

However, the analogies between healthcare inventory and other inventory though are comparable yet not perfect as Javad (1980) stated, several limitations are still to be recognized. It is important for making inventory policy to respond to both the realities of a multi-echelon inventory system and the urgency requirements of delivery system in healthcare sector. Nicholson et al. (2004) analyzed the trend of outsourcing as the third party logistic method to distribute non-critical medical supplies directly to the hospital departments by comparing the inventory costs and service levels of an in-house three-echelon distribution network vs. an outsourced two-echelon distribution network. Results show that outsourced two-echelon network not only in inventory cost savings but also not compromise the quality of care as reflected in service levels. In the case analysis of Mount Sinai Hospital Medical Center in Chicago (Saphir 1994), the successful inventory management results from the partnership with vendor and VMI techniques which increased their service levels and shifted several costs to vendor without incurring additional costs to the product’s purchase prices or hidden service fees. The first and most important step taken was to make sure the vendor’s data on the inventory items were in synchronization with the hospital’s data. This tedious step is the key to the success of the program and also gives a clue to the vendor’s information system capabilities and to the smoothness of the implementation. Saphir (1994) also suggested that holding the attitude of managing the inventory rather than handling it can enable hospitals to drive down their operation costs. Schouten (2006) who studied on a real case of Dutch healthcare supply chain improvements recommended that outsourcing the central warehouse to an external specialized partner which led to more focus on the core activities and a reduction in costs due to scale and efficiency advantages and a better inventory management strategy. Also it is indicated that the case in Netherland might be a good idea if VMI is incorporated when switches to a wholesaler for the supply process. In the book of Health Care Operations Management, the author Langanbeer uses the quantitative approach to analyze the business and logistics. In the supply chain collaboration chapter, he stated that implementing CPFR method in healthcare industry can maximize the effectiveness of the entire chain not just components within the chain. Although most hospitals and health providers do not today have 16

much of a role in this process, CPFR as a supply chain optimization tool would be of great use when the healthcare industry is ready.

2.2 System Dynamic Simulation 2.2.1 System Dynamics The researches on supply chain management have applied mostly by heuristics and mathematical models. Javad (1980) pointed out that although the mathematical model can produce a close-to-optimal solution, care must be exercised because the compounded effects of errors could produce far-from–optimal results. This possible side effect of mathematical model can be minimized by using the simulation method instead. Simulation can be done in several ways and system dynamic as the recently re-emerged method is among the most effective of them. System dynamic was created during the mid-1950 to simulate the stock flow feedback structure of GE plants. As the years pass, SD method has been applied to a wide range of problem domain. Two major tool of this system thinking are casual loop diagram and the stock flow diagram which can be implemented in the software such as STELLA, VENSIM, ITHINK etc. This thesis adopts stock flow diagram to simulate the inventory changes under the collaboration techniques and use ITHINK v9.0.2 to build and analyze the inventory collaboration models. 2.2.2 Literature review: Application of SD method in supply chain System dynamic method has been applied in the supply chain study and improvement in various industries such as software engineering, economic behavior, biological and medical modeling, etc. One of example can be found is by Ovalle (2003) who showed the application of SD method in improving the financial flows and working capital. The application particularly in inventory management was first by Barlas (1997) who studied a case of typical retail supply chain of three-echelon. From then on, the researches of inventory management with SD method for productions with various characters gradually enlarged. With the dynamic simulation, a far more straightforward improvement by adopting the collaborative method is presented. Zhong and Dilts (2004) modeled a dynamic supply chain network system with two 17

identical suppliers. The different interactions were evaluated to explore the effects of echelon interactions by the casual loop diagram, the conclusion of which shows that the collaboration has a crucial position for supply chain management in this era. More specifically, the well known collaboration techniques of VMI and CPFR are studied by Chen (2005) and Kohli (2005), both using the stock flow diagram to present the better inventory level and cost reduction compared to the traditional supply chain. Vlachos (2006) and Poles (2008) used casual loop diagram for the planning of supply chain and inventory control of remanufacturing industry, and the study of SD method have been extended to a new study field. Due to the nature and policy influence of healthcare industry, supply chain management for cost reduction purposes within healthcare industry has only been focused in recent year. Thus, limited literatures can be found that are purely using the SD methods in Healthcare industries. There is one case study based on Malaysia healthcare supply chain management (Mustaffa 2009), the author uses Data Flow Diagram to simulate the inventory replenishment. With the data flow analysis, the author was able to examine the performances of the urgent order as well as the stock availability at the wholesaler side. The major statement the author made is that VMI technique appears as the best solution for improving the inventory and distribution. There are however a number of supply chain researches about products which have short life cycle or shelf life can be referred to this thesis since medicine share the same characteristic as these products, from the order pattern, possible of demand surge and the gap to life cycle. For all these short life products, such as food, medical related, high-tech, innovation etc, the desire of collaboration between partners is not only driven by the severe marketing competition but more due to the product nature, the huge loss resulted from longer inventory and expiration of shelf life. The SD method is regarded as a superior methodology and can excel the other methods in study short life cycled products, stated by Georgiadi, Vlachos and Iakovou (2005) who have analyzed a major Greek fast food chain and shows that collaboration techniques are desirable for improving the efficiency of the food supply chain. Likewise, the appreciation of SD method is also proposed by Minegishi and Thiel (2000) who showed how system dynamics could contribute to improving the 18

knowledge of the complex logistic behavior of an integrated food industry and had taken the particular poultry production and process as examples, the dynamic behavior of food industry can be clearly captured even the input variables evolving with time. Moreover, the use of system dynamics can properly be adapted in to food chain analyze especially dealing with sudden demand as well as other oscillations so that the proper suggestions and order policy on shortening the delivery time and lessening inventory can be made to guarantee the freshness of the products. The advantage of system dynamic method is also stated in high-tech industry to management the demand gap by using collaboration strategies. Yuan et al. Ashayeri (2009) compare the VMI, JMI and CPFR with traditional supply chain by using the stock flow diagram. From the experimental analysis, the essence of the collaboration techniques is pointed out that is the establishment of trust, information and risk/profit sharing between players within supply chain. They indicate that CPFR is not necessarily the best alternative solution for collaboration as is commonly known in practice.

19

Chapter 3 Inventory Management in System Dynamic Model ---- Stock Flow Diagram This chapter gives the descriptions of the system dynamic inventory models on the traditional supply chain and the ones implemented with collaboration techniques. Section 3.1 gives a thorough elaboration of the traditional inventory model structure. The first evolvement which is vendor management inventory (VMI) from the traditional model is introduced in Section 3.2. In Section 3.3, a more advanced inventory collaboration technique CPFR model is presented. The last Section 3.4, model verification methods are discussed before all models built are applied to the case study in Chapter 4. All three models include four echelons, customers, who are the patients, generating daily demand; hospital department, that provide healthcare to its customers; central warehouse, that is backing up the hospital department inventory and supplier who produce the healthcare products. In reality, the number of each echelon participants should be more than one, however, for the modeling purposes, only one player is modeled for each echelon to represent the overall effect. This setting makes sense since for each participant, together with its upstream and downstream player can form a simply supply chain and the whole supply chain can be viewed as the combination of many single supply chains. Thus even there are economic of scale or any other collective effect, this setting is reasonable enough to reflect the optimized effect brought by collaborative techniques. As previous chapter has stated, the full pack of players involved in the supply chain should include the wholesaler whose function is to centralize all the supplies from suppliers and then dispatch the products to the central warehouse. However, wholesaler does not necessarily exist in all cases. In this thesis, under the setting of one representative for each echelon, the role of supplier can be extended to be the wholesaler if its function needs to be examined thus there is no need to add another 20

echelon to show the wholesaler functions in current study. The thesis mainly evaluates the inventory management within the supply chain, and therefore, no considerations are given with regards to the transportation efficiency. The major functions that construct the model are demand forecasting, inventory in and out flows as well as replenishment.

3.1 Traditional Inventory Models Customer

Hospital Department

Central Warehouse

Information flow

Supplier

Logistic flow

Traditional inventory flow chart

In the traditional inventory model, each echelon receives its downstream player’s orders. According to their own inventory level and demand forecasting, they calculate the inventory adjustment amount and place the order to the upstream player. After receiving the order, upstream player sends the products with either the order amount or the whole inventory in case inventory cannot cover the order amount, then products are delivered after the transit lead time to the downstream player. As the graph shows, there are no interactions cross echelons. Each party simply makes its own orders irrespective of the situation of other players. 3.1.1 Customer

Customer level stock flow diagram 21

The customer level mainly generates the daily demand of the patients. To be practical, this daily demand is simulated stochastically. The model uses exponential function in the software to generate a series of exponential random numbers to create the exponential inter-arrival time with the mean of one day. Thus given the average demand, the stochastic daily demand can be obtained. The customer demands are kept in the record for the purpose of the forecast demand at hospital department level. 3.1.2 Hospital Department

Hospital Department level stock flow diagram When receiving customer orders, hospital department directly fulfills the customer order from hospital inventory. In case the inventory cannot cover the order amount, backlog is kept in record and waits for the new inventory replenishment to be fulfilled with. On the other hand, according to the customer daily orders, hospital department makes its demand forecast on the possible future orders from customer. The forecast demand is determined on the inventory review day and supposed to cover all the possible demand during the inventory replenishment lead time which is defined as the review period plus the transit lead time. With the forecast demand, the desired inventory level 22

for the length of replenishment time can be calculated, that is the total forecast demand during the replenishment lead time plus the safety stock. Safety stock is used to cover the customer demand deviation during the hospital department replenishment time. The work that also needs to be done on the review day is to place the order to the central warehouse. The order amount should be the mismatch between the desired inventory and the current real inventory which includes the amount in transit, it should also include the backlog amount hospital owes to the customer as part of the order. By receiving the order from hospital department, central warehouse immediately send the products to hospital inventory. After the transit lead time which includes order observation delay and transport delay, hospital inventory is replenished and able to fulfill the customer demand again. 3.1.3 Central Warehouse

Central Warehouse level stock flow diagram Same as hospital department, after receiving the order, central warehouse immediately 23

fulfills the demand from its inventory. If there is a shortage in inventory, backlog is recorded and waits for the new replenishment. The order of hospital department is kept in record each time for the purpose of forecasting. Normally, as the upper echelon, central warehouse has a relatively longer review period. With the same logic, on the review day, central warehouse first gives forecast for the hospital future demand which can cover the time span of central warehouse replenishment time. Then the calculation of the desired inventory level is made for adjusting actual inventory by placing the order to supplier. The safety stock at central warehouse level is used to cover the demand deviation of hospital order during the central warehouse replenishment time. The order amount to the supplier includes the inventory adjustment and the backlog amount for hospital department. Supplier sends the demand amount to central warehouse inventory the time receiving the order. Central warehouse inventory is replenished after the transit lead time and then fills the backlog to the hospital department, and the rest inventory is kept for the next order from hospital. 3.1.4 Supplier

Supplier level stock flow diagram 24

Supplier fulfills the central warehouse demand when receiving the order. A backlog is kept in record and waits for the new production to fulfill. Based on the central warehouse order, combining the supplier own review period and production lead time, the forecast demand for central warehouse is determined to cover the review period and the production lead time which are the replenishment time for the supplier. Then the desired inventory is calculated to adjust the inventory level by sending the production order. Central warehouse order standard deviation is used for calculating the safety stock.

3.2 VMI Model Hospital Department

Central Warehouse

Supplier

Customer

Information flow

Logistic flow

VMI model flow chart

Vendor Management Inventory is an important inventory management technique and has been proposed by many researchers and applied in a large number of practical cases. There are two kind of VMI—Upstream VMI which is implemented between central warehouse and wholesaler and Downstream VMI which is implemented between central warehouse and hospital department (Schouten 2006). In this thesis, since no wholesaler is involved, thus Downstream VMI technique is modeled. Compare to the traditional case, central warehouse now can monitor the inventory changes of hospital department and negotiate a reorder rule with hospital in advance. When customer sends the order to hospital department, central warehouse can receive the order information at the same time via the internet. Hospital department no long needs to place order to central warehouse, with the pre-determined reorder point, central warehouse is able to make the replenishment whenever hospital department 25

inventory is below the reorder point. The order placement and inventory replenishment between central warehouse and supplier are the same as traditional case. 3.2.1 Customer The customer order placement is the same as the traditional case.

Customer level stock flow diagram 3.2.2 Hospital Department

Hospital Department level stock flow diagram When receiving customer order, hospital department directly fulfill the demand from its inventory. The backlog is kept in record until new replenishment is made. In VMI model, demand fulfillment for customer is the only work for hospital 26

department. By accessing the customer demand information, central warehouse makes the demand forecast for hospital department to adjust the reorder point, which can also be called as the desired inventory. This forecast is done on daily bases by monitoring the daily order changes from customer, and thus, the reorder point is calculated everyday to compare to the actual inventory for the adjustment. With this continuous replenishment from central warehouse, there is no actual backlog between central warehouse and hospital department, and the backlog at hospital department can only occur when both levels are out of stock and no in-time replenishment considering the transit lead time from supplier to central warehouse and then to the hospital department. As soon as the inventory is replenished, backlog is fulfilled for customer. 3.2.3 Central Warehouse

Central Warehouse level stock flow diagram In the VMI model, central warehouse plays the role as the vendor to hospital department, besides the function described above about the forecast and the in-time replenishment for the hospital department, central warehouse also needs to make its own demand forecast to adjust inventory. Different from the traditional model, in the VMI model, central warehouse is able to access the customer demand information, thus the forecast can be made on the customer order bases. This is the major improvement from the traditional model that central warehouse is able to give a more 27

accurate forecast to the demand and place the order with much less distorted information to supplier. Therefore central warehouse can reduce its inventory level without causing many backlogs. Safety stock can also be reduced since the calculation is based on customer demand standard deviation, the less uncertainty on demand information enables central warehouse to decrease its safety stock. Same as traditional model, the forecast and desired inventory level for central warehouse itself is determined on the review day. Then comparing to its inventory level, order is placed to supplier with the amount of the mismatch between desired and actual inventory and also the backlog amount hospital department owes to customer, which is the exact amount that central warehouse is not able to fulfill the hospital department inventory in time. Central warehouse inventory is replenished after the lead time since supplier sends the products. After the replenishment, central warehouse is able to continue adjusting the hospital department inventory. 3.2.4 Supplier

Supplier level stock flow diagram

28

At the supplier level, the fulfillment for central warehouse, forecast, inventory adjustments as well as production are the same as traditional case. When receiving the order, supplier directly sends the amount from its inventory. When the inventory cannot fully cover the demand, a backlog is recorded for new production to fulfill with. The forecast and desired inventory is determined to cover the production lead time and the review period. Safety stock is still based on the central warehouse order standard deviation. Then, after the production is finished, the supplier inventory is fulfilled again to serve the central warehouse order.

3.3 CPFR Model Hospital Department

Central Warehouse

Supplier

Customer

Information flow

Logistic flow

CPFR model flow chart By applying the collaborative planning, forecasting and replenishment (CPFR) technique, each echelon can obtain a more accurate demand forecast since the electronic platform enables each party access to the customer demand. Also, with the pre-agreed objective of collaboration and confidentiality agreements, each player is able know the other player’s business related information such as business plans, inventory, sales-forecast, order-forecast, etc (Caridi 2005). This highly transparency enables each upper echelon replenishes its down play inventory in time and avoid the backlog, consequently, higher service level can be reached with the minimum desired inventory level for each party. 3.3.1 Customer Customer order comes stochastically the same as traditional and VMI case. 29

Customer level stock flow diagram 3.3.2 Hospital Department

Hospital Department level stock flow diagram When customer demand comes, demand is fulfilled directly from the inventory, and backlog is kept in record in case of inventory shortage. The demand forecast and reorder point (desired inventory) is determined on daily bases, at the end of the day, the inventory adjustment can be reviewed by central warehouse through the internet. Whenever the hospital department actual inventory is below the desired inventory level, central warehouse will deliver the adjustment amount to the hospital department. Then inventory is replenished after the transit lead time from central warehouse to hospital department.

30

3.3.3 Central warehouse

Central Warehouse level stock flow diagram Through the internet, central warehouse is able to access to the customer daily demand and the hospital department inventory level, thus daily inventory adjustment for hospital department can be obtained and the adjustment amount can be sent in time. After fulfilling the hospital department demand, the daily inventory adjustment for central warehouse itself can be gained by the forecast and reorder point (desired inventory). The safety stock involved in the reorder point is calculated based on the customer demand standard deviation. Since supplier can also review the daily adjustment amount for central warehouse inventory, thus whenever inventory adjustment is needed, central warehouse inventory will be replenished after the transit lead time from supplier to central warehouse.

31

3.3.4 Supplier

Supplier level stock flow diagram With the internet, supplier is able to replenish central warehouse inventory in time by reviewing the daily inventory adjustment information through internet and then the adjustment amount is sent from supplier inventory to the central warehouse. In CPFR model, supplier has also access to the customer demand information on daily basis. Thus, the safety stock at supplier level can also be reduced since it only needs to cover the demand variation from customer order. The demand forecast and the production adjustment for supplier are done every day after supplier fulfills the central warehouse inventory adjustment. And then production starts with the amount to adjust the supplier inventory to the desired level.

3.4 Model Validation Model validation is an important aspect of any model-based methodology in general, and system dynamics in particular. Although model validation takes place in every stage of modeling methodology, it is safe to state that a majority of ―formal‖ validation activities take place right after the model construction has been completed 32

and before policy design simulations (Barlas, 1994). The general logical order of validation is, first to text the validity of the structure, and then start testing the behavior accuracy (Barlas 1994). There are normally three main tests used for the model validation. Direct structure tests, Structur-oriented behavior tests and behavior pattern tests (Barlas 1996). Several specific tools within each test methods can be applied both from theoretical and empirical point of view. Comparing the equations of the simulated model with the relationships existing in the real system or with the generalized knowledge in the literature (Forrester and Senge 1980) is proposed as the basic methods to test the model. Some other semi-formal tools that can also be applied such as formal inspections, reviews, walkthroughs and data flow analysis, which are typically used in the verification and validation of computer model (Balci 1994). To test the system dynamic model built above, both theoretical equation checks and the data inspections are taken into account to see if the model are simulated as close to the real system as possible. 3.4.1 Equation Checks The essence of the three proposed inventory models relies on three aspects which are the determinations of demand forecast, desired inventory and inventory adjustment. Demand Forecast In reality, the demand forecast is crucial to inventory management. It is normally modeled quite complex with considerations on as many influencing elements as possible. In most literatures such as Yuan et al. (2009) and Minegishi et al. (2000), the forecast demand is simulated as the smoothing exponential which can obtain a forecast on daily basis. However, in the case study of Pantein, demand forecast is quite different especially in the traditional case. The demand forecast is only done on the review day which is not daily; the forecast is simply done by using the history of 33

demand data. It is modeled in such a way, whenever forecast is needed; the total sum of the latest order amounts that would cover the length of replenishment time is used as the forecasted demand for the next replenishment period. In later cases when apply the collaborative techniques, forecast needs to be done on daily basis, then it is modeled as the moving average as stated in the book Inventory Control by Sven Axsater (2006). Desired inventory The desired inventory is also regarded as the reorder point. It is the major indicator for inventory management. The standard suggested function in many inventory management books such as Inventory Management by John W. Toomey (2000) or Inventory Control by Sven Axsater (2006) is suggested as: Desired inventory = Average lead time * Average demand + Safety Stock,

where Safety stock= Z*sqrt (average lead time* Demand STD^2 + Avg demand^2*Lead time STD ^2),

Z is the service level, Lead time = review period + transit lead time. The first part of the desired inventory is just what has been stated above the forecast demand during the lead time. Inventory Adjustment Inventory adjustment is calculated as the order amount to the upper level parties. The adjustment is expressed as: Inventory Adjustment = Desired Inventory – Actual Inventory + Backlog

where the actual inventory also includes the products in transit. It is a quite straight 34

forward expression for adjusting the inventory to the desired level and a simple way to simulate the reorder scheme. 3.4.2 Data Inspection The models are built under the general principles and equation relations discussed above; however, data inspection is still needed to gain a more clear confirmation of the model logic when incorporating principles in the software environment. The test is done by giving a fixed daily demand at customer level. Without the stochastic influence, the data flow is relatively easy to trace at each echelon. Major aspects that need inspections are the delays in the model especially given the natural system delay of one day between each system level. One other major aspects that needs attention is the backlog amount for the inventory adjustment since backlog should be accumulated before fulfilled again. By comparing the data which is generated by the model against the manual calculation, the logic and the equations incorporated within the model can be confirmed and applied for the case study.

35

Chapter 4 Case Study & Results Analysis Pantein is a typical healthcare provider in the Netherland. It is in charge of the inventory cost caused by hospital department and central warehouse as referred in the model. By reducing the inventory level through examining the forecast, order and replenishment scheme of each party involved and comparing the inventory efficiency at each echelon through different inventory collaboration techniques, cost saving inventory management approaches can be suggested to Pantein without jeopardizing its service level to its customer. In this chapter, the case background is first introduced in Section 4.1 including the data setting for the model simulation. In Section 4.2, given the data information at each echelon, model configuration should be done to correctly incorporate these figures in the model. Before generating the simulation results, an experimental set of scenario are considered to test the impact of the key factors to the system. This experiment design is discussed in the Section 4.3. Section 4.4 presents the graph simulation results from base case, VMI and CPFR models, mainly focusing the inventory level and backlogs at each echelon. Thereafter, the comparison of inventory management improvements between the base case and the VMI & CPFR collaborative cases are analyzed in Section 4.5.

4.1 Case Introduction & Data Setting The number of locations of Pantein analyzed is 20. Those locations among all the Pantein local hospital departments have relatively large demand. The rest of the Pantein locations are not taken into account since they have very small demand throughout the year which makes it hardly worthwhile to analyze them. The inventory models simulate the situations considering the products that are analyzed as one mixed type which includes all necessary products for fulfilling the customer demand. Although this is for simplicity, since the inventory models are studied for the 36

management improvement by collaboration, the product type is not a major issue for the analysis. In the model, the daily customer demand is simulated as stochastic demand with exponential inter-arrival time. According the Pantein history data, the daily average demand is about 10 mixed medical goods to each department location which is assumed to have no strong correlations over the 20 department locations. Thus when combining the 20 departments’ demand as one hospital department’s, the total customer demand can also be treated as 200 mixed medical goods per day. Hospital departments order two times a week (Monday and Thursday) to central warehouse which makes the review period either three or four days. For simplicity, the review period for hospital department is 3.5 days. Central warehouse orders once a week on Friday to supplier, thus its review period is 7 days. After placeing the order, it takes one day to deliver the goods from central warehouse by truck which makes the transit lead time between hospital and central warehouse one day. On average, it takes seven days for supplier to deliver the goods to central warehouse, and thus the transit lead time from supplier to central warehouse is seven days. Since the data provided by Pantein is only at hospital department and central warehouse level, no information with regard to the supplier can be accessed. In order to reveal the inventory management improvements that collaboration techniques bring, the settings at supplier level are assumed. As stated above, product classification does not influence the analysis, thus one supplier is assumed to provide the mix of all product type. When fulfilling the central warehouse order, supplier immediately reviews his inventory level and determines the production for adjusting the inventory level, which leads to the supplier review period also 7 days as central warehouse. The production lead time at supplier is assumed to be 2 days considering the automatic production lines. The demand standard deviation (STD) is used to calculate the safety stock for each 37

party. The standard deviation of customer demand is unknown, however, for the research purposes; it is assumed that the total standard deviation is 20 goods that is 1 good at each location. For simplicity, the standard deviations of hospital department order and central warehouse order are assumed to be the fixed number of 140 goods and 280 goods by considering the possible demand variation due to the lead time from the lowest echelon to the upper levels. For each party, initial inventory must be set in such a way that they do not distort the ordering process in the first period of the simulation. The initial inventories are set to cover the demand during the review period of each level, which are average customer daily demand multiply the review period. This is a simple approximation for the mean inventory level during the review period and can be regarded as a preparation for the simulation warm up. The desired service level is 95% which is widely used in inventory models and it is also the criterion for Pantein in 2005-2009. Thus all three echelons have the service level factor 1.64 for safety stock calculation. The simulation length should be a time span which a problem might reasonably present and which its analysis, through simulation, will be conducted. (MaLucas, 2005) The time horizon must also be at least long enough to accommodate the longest expected delays and all the effects those delays might produce. Therefore, the simulation is determined to be a hundred days which should be sufficient to make a thorough observation of each echelon’s performances over time. To sum the above setting information, the major figures for the models are in the table below:

38

Setting Information Summary Customer Average Demand

200

goods/day

Demand STD

20

goods

Service level

95%

Hospital Department Initial inventory

700

goods

Inventory review period

3.5

days

Transit time from central warehouse

1

day

Hospital order STD

140

goods

Service level

95%

Central warehouse Initial inventory

1400

goods

Inventory review period

7

days

Transit time from supplier

7

days

Central warehouse order STD

280

goods

Service level

95%

Supplier Initial inventory

1400

goods

Inventory review period

7

days

Production lead time

2

days

Service level

95%

4.2 Model Configuration The ITHINK software has an artificial one day system delay between each stock level which should be dealt with extra attention, otherwise would cause a big distortion to the simulation results. Since the system delay cannot be avoided, the model in this thesis incorporates this artificial delay into the figures set for the case study, that is to say treating the system delay as part of the transit lead time or the review time. The model validation of data inspection can assist to guarantee the correctness of the time and delays set for the model. The review period of hospital inventory is 3.5 days while all the other model related times are integers. To keep the time unit consistence and to avoid the unnecessary 39

trouble caused by time setting, the model is adjusted that all the review period, transit & production lead time and the simulation length are doubled while the daily customer demand is halved. With such system adjustment, the system time unit is changed into 0.5 day however the simulation result remains the same as this thesis desires to study. The customer demand is simulated stochastically. It is modeled with the function which generates a series of exponentially distributed random number with a mean of average customer daily demand. In order to compare the results over the three models, it is needed to replicate the stream of random numbers by selecting the specific seed to replicate the simulation. One of the major differences between healthcare and other industries is that patients can come to the hospital department for treatments every day, and medicines as the major products discussed in this thesis are needed all the time, so there are no holidays and breaks at the hospital department and the central warehouse. As to the supplier level, nowadays, automatic production line enables the products to be produced 24/7 only with the shift between workers, thus, supplier also needs no breaks. As a result, models are simulated 24/7 with no artificial breaks for Saturday and Sunday. This setting makes the simulation consistence and no need to exclude weekends from the lead time setting which can save a lot of trouble for integrating the model.

4.3 Experimental Design In this section, the experimental design with regard to several influential factors is discussed. The design is based on the fractional factorial orthogonal arrays which is a method of setting up experiments that only requires a fraction of the full factorial combinations. This method examines the various combinations of parameters which are given equal opportunity to express its effect on the response. To be specific, in this study, 6 factors are investigated namely the review period and the transit or 40

production lead time for hospital department, central warehouse, and the supplier since they are the major parameters that influences the forecast and replenishment of each party. By changing these factors, the relationship between the factors and the severity of their effects on the inventory and backlog can be observed. The experimental results however are not meant to deeper investigate the exact impact of these parameters but to shed light on the performances and check the effect consistency that the collaborative techniques can bring in various scenarios. In the factorial arrangement with 6 factors, three levels for each factor are chosen as shown in the following table. Factorial Arrangement Hospital Department

Central Warehouse

Supplier

P1

P2

P3

P4

P5

P6

review

transit lead

review

transit lead

review

production

period

time

period

time

period

lead time

1

2,5

0,5

6

6

6

1

2

3,5

1

7

7

7

2

3

4,5

1,5

8

8

8

3

level

A complete factorial experiment requires 36 =729 runs which is quite a large number, thus the principle of orthogonal design is applied by choosing the form of L27 (36) as follows.

41

Orthogonal Design Table Runs

Parameters P1

P2

P3

P4

P5

P6

1

2,5

0,5

7

6

8

3

2

2,5

0,5

7

6

7

2

3

2,5

0,5

7

6

6

1

4

2,5

1

6

7

8

3

5

2,5

1

6

7

7

2

6

2,5

1

6

7

6

1

7

2,5

1,5

8

8

8

3

8

2,5

1,5

8

8

7

2

9

2,5

1,5

8

8

6

1

10

3,5

0,5

6

8

8

2

11

3,5

0,5

6

8

7

1

12

3,5

0,5

6

8

6

3

13

3,5

1

8

6

8

2

14

3,5

1

8

6

7

1

15

3,5

1

8

6

6

3

16

3,5

1,5

7

7

8

2

17

3,5

1,5

7

7

7

1

18

3,5

1,5

7

7

6

3

19

4,5

0,5

8

7

8

1

20

4,5

0,5

8

7

7

3

21

4,5

0,5

8

7

6

2

22

4,5

1

7

8

8

1

23

4,5

1

7

8

7

3

24

4,5

1

7

8

6

2

25

4,5

1,5

6

6

8

1

26

4,5

1,5

6

6

7

3

27

4,5

1,5

6

6

6

2

4.4 Results Illustration After performing the experimental simulation with different parameter settings in base case, VMI and CPFR models, detail information of all the scenarios are skipped but highlights of all the simulation runs are concerned. In general, the longer the review period and the transit lead time are, the more variance and uncertainty will be created to each party. Therefore the positive effects by adopting the collaboration techniques are more obvious. The results show the consistency of the collaborative effect under 42

different scenarios that the supply chain total inventory on average and inventory STD of each party has been decreased through the techniques by creating more information transparency among each echelon. The overall simulation results of total average inventory within the supply chain are presented in the graph as following. The total average inventory indicates that collaboration techniques bring the welfare to all parties in general, and it can be pointed out that the different review period and transit lead time for each party can influence their forecast mechanism which will then lead to some uncertainty on the inventory level. Thus, the 27 experimental simulations not only gives an overall intuition on the possible impact of the main influencing factors, but also comes to a conclusion that the order system including inventory review and goods transit of each party should be compatible so that more information can be gained to create more certainty on the demand forecasting of each party.

The results overview of 27 simulation runs In the following are the general results that reveal the inventory and service satisfaction of each party in each model. The graphs of the inventory level compare to the backlog frequency and amount are presented to give a brief idea of the deficits and 43

improvements in each case. The focus putting on the inventory levels and the backlogs relies on the reasons that, as the healthcare provider, searching the effective inventory management method is the major purpose of conducting this research, in the meanwhile the service satisfaction (backlog) should also be guaranteed, otherwise patients might not get treated in time and their health might be in danger which is not only the matter of profit loss but also the jeopardy of its fame and trust in the long run. No cross echelon order and inventory comparisons are presented graphically in this section is because that besides the customer demand which is on daily basis, the order placement for all the upper stream parties has time interval. Since the order and inventory changes of hospital department, central warehouse and supplier are not in the same time unit, thus there is no straight forward comparison can be shown in graphs of the inventory oscillations with three parties at the same time. The analysis of the inventory variation will be conducted numerically in Section 4.5. 4.4.1 Base case In the base case, no collaboration techniques have been applied, which is the approximation of the current situation. After a length of 200 days simulation, the inventory and the backlog situation are obtained as following:

Base case Hospital department Inventory vs. Backlog 44

Central warehouse Inventory vs. Backlog

Supplier Inventory vs. Backlog

First of all, the trend of inventory changes shows the order and replenishment during the lead time of each echelon. Due to the review period difference between each party, it is noticeable that at central warehouse and supplier level, the inventory has periodical stable stage. Then, when focusing on the hospital department level, although the average inventory level is quite low, it is quite obvious that the frequency of backlog is quite high which means that with the current forecast, reorder and replenishment scheme, the inventory at hospital department cannot satisfy the patients in time especially facing their stochastic demand. On the contrary, central warehouse maintains a relatively high 45

inventory level, yet since central warehouse can only replenish the hospital inventory when receives the order, thus resulting in a most undesirable situation which is the low service level at hospital department while a high idle inventory level at central warehouse. Two major backlogs at central warehouse and supplier level are all resulted from being unable to anticipate the large order from the downstream player. In the meanwhile, the average inventory level and the idle inventory are still quite large. To conclude the base case situation, the information delay as well as the inventory management (including order & replenishment) policy inefficiency can be regarded as the major deficit of the current system which has a large room for improvement. Therefore, collaborative approaches are needed to optimize the system efficiency. 4.4.2 VMI When VMI technique is incorporated between hospital department and central warehouse in the system, the simulation results are as follows:

Hospital department inventory vs. Backlog

46

Central warehouse inventory

Supplier Inventory vs. Backlog

Because of the information transparency between the hospital department and the central warehouse due to incorporating the VMI technique, it can be observed that both the backlog frequency and the amount at hospital department level are much less compared to the base case. Since central warehouse now can fulfill the hospital department inventory without the formal order placement made by hospital, thus, there is no backlog within the two levels. Compare to the base case, in the VMI model, the average inventory level at central warehouse has been reduced.

47

Supplier on the other hand still has the same situation as the base case. When central warehouse does not place the order, then the idle inventory are stored in the supplier inventory, especially with a high level since supplier still needs to make sure to cover the possible order of central warehouse. 4.4.3 CPFR When incorporating the CPFR technique to the whole system which is adding the transparency between central warehouse and supplier as an addition to the VMI model, the results are as follows:

Hospital department inventory vs. backlog

Central warehouse inventory 48

Supplier inventory

By incorporating the CPFR model, there is no backlog at hospital department level at all. The oscillation of the inventory also reduced. Since each echelon of the whole system has the access to the demand information and the downstream party’s inventory situation, replenishment can be made in time to maintain each party’s desired inventory level. Thus the backlog at both the central warehouse and supplier level are eliminated.

49

4.5 Comprehensive Analysis Section 4 mainly presents the simulation results at each echelon. In this section, a further analysis of the inventory level by putting the inventory of the three parties together is made to reveal the inventory improvement particularly at hospital department and central warehouse level created by collaboration techniques. Results Summary Table

Base case

VMI

CPFR

Hospital Department

Central Warehouse

Supplier

Total

Average inventory

492

1445

2860

4797

Inventory VAR

97155

672912

1253061

-

Inventory STD

312

820

1119

-

Backlog

2800

534

2045

5379

Average inventory

687

509

2707

3902

Inventory VAR

89344

294393

715577

-

Inventory STD

299

543

846

-

Backlog

1000

0

973

1973

Average inventory

912

1419

1537

3869

Inventory VAR

42747

293991

181896

-

Inventory STD

207

542

426

-

Backlog

0

0

0

0

The table above is a summary of the inventory situation as well as the backlog of each party in each model. 4.5.1 Base Case It is apparent that the base case model generates the largest inventory, backlog as well as the inventory standard deviation from both the individual party and the whole system. Although in the base case, the average inventory at the hospital department is smaller than the VMI and CPFR case, however, when observing the history data, the inventory excesses the level in VMI and CPFR in the same time period several times. Together with such large backlog, it indicates that this low average inventory level is 50

abnormal which is caused by lack of in time fulfillment rather than an efficient inventory management, in other words, the inventory level at hospital department is either too high or low. The large standard deviation at central warehouse and supplier level in the base case bring the ―Bullwhip Effect‖ to light, due to the lack of information transparency, upper stream player cannot make accurate forecast for the future demand and the fixed reorder day makes the situation even worse, eventually, this amplified effect spreads through the supply chain. Therefore, as the results shows, central warehouse and the supplier has to keep a large inventory but still generates many backlogs when unexpected large orders come. For the parties that the research concerns, the cost that hospital department and central warehouse inventory generates is quite high considering the inventory amount, which is a large burden for Pantein especially with such low service satisfactory due to the large backlog. 4.5.2 VMI The results of VMI model shows that by applying the VMI techniques, the backlog at hospital department has reduced about 64%, although inventory at hospital department has a slight increase yet the inventory level at central warehouse has reduced dramatically. These two major changes compared to the base case shows that VMI technique has successfully improved the inventory management of Pantein by balancing the inventory between hospital department and central warehouse while increasing the customer satisfaction. The inner logic is that central warehouse continuously adjusts the hospital department inventory in order to maintain the inventory at the desired level, whereas central warehouse has to place its own inventory adjustment on a fixed reorder date. Therefore, the hospital department can almost keep closely to its desired inventory level while the central warehouse inventory for most of the time is far under the desired level or even zero inventory. Such situation might result in the backlog when large order comes yet both hospital 51

department and central warehouse are out of stock. The VMI model results also shows that because of the ability to access the customer demand information, the inventory STD at central warehouse decreases a lot which is also a good indicator showing that VMI techniques reduces the ― bullwhip effects‖ at central warehouse level. However, since this technique is only incorporated between central warehouse and hospital department, thus leads to not much improvement of inventory level of the supplier. The slight decrease on inventory STD and backlog of supplier on the other hand is because that now the central warehouse can make a more accurate forecast, the order to the supplier thus involves less false information and unnecessary order amount. This positive effect enables supplier to adjustment its inventory with much less oscillation over time which eventually improves the supplier service situation. 4.5.3 CPFR The CPFR technique brings each party of the system into a new and desired situation. No backlog has occurred at any parties which is a big improvement on the service satisfaction. With the internet information, each party has the ability to access the customer demand as well as knowing each other’s situation. The pre-determined desired inventory level of each party can be maintained in such scheme by the in-time replenishment from the upper echelon whenever the actual inventory is below the desired level. The information transparency leads an overall welfare to all the parties involved in the supply chain, comparing to the base case, not only the total average inventory has been reduced about 20%, but also the backlogs are eliminated. The inventory STDs at all level also come out much smaller than the base case. By further observing the results, in CPFR model, since supplier is also able to fulfill the central warehouse in time, the inventory is balanced between central warehouse and supplier. The in time fulfillment at central warehouse guarantees the maintenance of hospital department inventory level to its desired level. Therefore, both hospital 52

department and the central warehouse inventory are larger than the VMI case whereas the supplier inventory is almost halved. The same transparency at central warehouse level makes its inventory STD no difference between VMI and CPFR model, but the transparency at supplier level reduces the inventory oscillation a lot. The comparison between VMI and CPFR can be concluded that CPFR technique has a superior power to improve the entire supply chain inventory whereas VMI technique can be more focused on specific parties that adopt this technique to improve the inventory management.

53

Chapter 5 Concluding Recommendations In this Chapter, the recommendations with regard to the inventory management for Pantein will be made based on its current situation and the possible improvements that the proposed collaborative techniques can bring. It is for sure that in practical, a large number of factors should be considered before changing the policy or implementing any facilities to upgrade the current situation, thus it should be pointed out that this research and the obtained results are not the definitive answers but rather to provide some insights on the possible improvements only from the inventory management point of view. In the model analysis, the hospital department and central warehouse are under the charge of Pantein. The current situation (base case) clearly shows the inefficiency of the inventory management of Pantein. The large backlog at hospital department level and the large idle inventory at central warehouse suggest the reform of order and replenishment scheme within Pantein should be conducted immediately. The major deficit of the current situation, to a large extends results from the lack of information transparency. The following sections propose the possible methods based on the research results aiming to optimize the current situation of Pantein. The advantages as well as the possible obstacles and issues are concluded for VMI and CPFR techniques in Section 1 and 2. In Section 3, a outsourcing approach is briefly discussed to qualitively present another options for Pantein to improve their inventory management.

5.1 A Better Inventory Management ---- VMI The VMI technique implemented between central warehouse and hospital department indicates a better inventory situation for Pantein. With the central warehouse personal regularly checking the status of all local department inventories and managing the order and replenishment, the hospital department inventory will remain within a 54

normal and relatively low range while raise the service satisfaction to the patients. Or it can be expressed that VMI technique balances the inventory between hospital department and central warehouse and makes better use of the inventory to achieve its total reduction on inventory. VMI also successfully reduces the safety stock at central warehouse, as a neglected part for most of the time in inventory; safety stock reduction can also generates a substantial amount of cost over time. Considering the related costs by incorporating the VMI technique, the tradeoffs between saving and additional spending are needed to be taken care of. On one hand, reducing the total inventory level of Pantein can save the inventory cost, on the other hand however, by maintaining the hospital department inventory to the desired level, constant products transit are needed which leads to raising the internal transportation cost. Besides, the model only simulates the local hospital department with large demand yet there are many other locations with small demand that are impossible and inefficient to do the replenishment centrally, and the possible electronic facilities implementation which is used to update inventory status would also costs a lot of money. Therefore, it can be suggested that VMI technique would be an options for the large demand hospital departments while for those small demand locations, they can remain the status quo of executing the order and replenishment by the local personnel.

5.2 A Better Inventory Management ---- CPFR As the study shows, VMI technique can efficiently reduce the total inventory of Pantein (both central warehouse and hospital department), however if highly irregular demand occurs, then there is possibility that both central warehouse and hospital department are out of stock. CPFR technique compared to VMI excels at the forecast and the in time replenishment function between each echelon throughout the whole supply chain. By applying this technique, Pantein is able to provide an even higher service satisfaction to the patients and stay robust when facing the irregular demand. The simulation results suggest that CPFR technique brings the welfare to all the 55

parties involved in the supply chain. The higher service satisfaction at hospital department is at the cost of more Pantein inventories compared to the VMI situation. As long as the central warehouse is still under the supervision of Pantein, the inventory balancing between central warehouse and supplier does not seem desirable to Pantein. Moreover, CPFR brings more responsibility to the central warehouse personnel, now, they not only have to take the replenishment decisions of the hospital department but also have to negotiate and compromise with the supplier on the desired inventory level. This would lead to even slower decision making, the inefficiency of communication can also generate loss or even worse, jeopardize the cooperation with supplier. If the CPFR technique can be implemented, the cost side should be always be considered first. Both internal and external transportation cost would increase since the in time replenishment is performed more frequently between parties. The possible electronic facility costs might be even more considering the additional connections with supplier. The possible negotiation and compromise would also become the expenses of Pantein as a pre-stage investment to reach the desired management scheme. As the rewards of these possible expenses of implementing the CPFR, the efficient information exchange and transparency would enable Pantein to negotiate a lower reorder point with supplier externally and for hospital department internally, thus Pantein can keep up its high service satisfaction while further reduce its the inventory cost. In the long run, the stable relationship within the supply chain will bring benefit that not only covers the necessary facility investment but also generates extra profit for Pantein. In conclusion, CPFR technique is a long term investment, if the difficulties in the initial period can be overcome, then CPFR is a worthwhile option for Pantein for an even better future development.

56

5.3 Outsource the Central warehouse The third party logistic possibility is outsourcing the central warehouse to a specialized external partner. The hospital department can focus more on providing the health care and Pantein can get rid of the responsibility of managing the central warehouse. A professional third party may has far more experience with proper inventory management and handling the incoming receipts, moreover, the good and mature relations with various suppliers will help Pantein to reduce a lot of efforts in communicating with old and new suppliers and the purchase cost considering the possible group purchasing. A professional third party may even obtain a mature electronic system and the tailor made integration to the system in hospital department and the supplier for implementing the collaborative methods. Since there are a lot concerns with regard to the electronic facility implementation as well as efficiency in VMI and CPFR method, outsourcing the central warehouse to professions would be an ideal way to manage the inventory for Pantein. The major disadvantages of outsourcing is that Pantein has to give out a lot of data and loss control and flexibility of central warehouse which however might not be a major issue if a well rounded contract is signed between Pantein and this external partner. The outsourcing fee for the third party should also be taken into account, different from VMI and CPFR techniques; outsourcing method has an immediate effect on the cost reduction. However, if a long term development is planned, Pantein has better to do an investment project to compare the benefits between investing collaboration technique and the outsourcing the central warehouse.

57

Conclusion and further research Conclusion The aim of this thesis is to evaluating the inventory collaborations in healthcare industry. By studying the Pantein case, the current situation is analyzed and used as the base case to reveal the inefficiency of the traditional order and replenishment scheme of each party within the supply chain. Then the two collaboration techniques, vendor inventory management and collaborative, planning forecasting and replenishment are introduced to improve the inventory management by reducing the inventory level while raising the service satisfaction to the patients. The research is performed in the system dynamic models and the simulation results indicate that collaborative techniques can bring the reforms to the whole supply chain inventory management situation. Numerically, the total average inventory reduces almost 20% by implementing the techniques, the amount of backlogs decreases over 60% by VMI and fully eliminates by CPFR. The inventory oscillations which reflect the ―bullwhip effect‖ is also changed to be more concentrated, about 50% reduction compared to the base case. The general conclusion with regard to the Pantein case can be stated as follows. The integration of the supply chain by incorporating the VMI technique helps Pantein to balance its internal inventory between its central warehouse and hospital departments. The information transparency enables Pantein to transfer the central warehouse idle inventory into the in-time replenishment for hospital departments and thus realizes both the total Pantein inventory reduction and the service satisfaction promotion. However, considering the cost of implementing the electronic facilities, it might only be worthwhile for those hospital departments with large demand. CPFR technique although improves the overall inventory management especially at the supplier level, it takes more time and effort for Pantein to gain the benefit. The 58

advantage of CPFR relies in the following. The fully information access by each party ensures the in time replenishment not only for hospital department but also for the central warehouse provided by the supplier, in the long term, Pantein central warehouse is able to negotiate an even lower desired inventory level with supplier and set a lower reorder point for hospital department to achieve the purpose of inventory reduction. However, this re-negotiation normally takes more time and effort or even compromises with suppliers and the facilities needed to integrate this technique may also cost a lot. Thus CPFR is recommended as a long term investment that can be realized by steps but rather a short term effect that helps Pantein to save the costs. Outsourcing the central warehouse would also be an option. This would lead to the more focus on the core activities of Pantein and a reduction in costs due to the economics of scale, the efficiency and the better inventory management strategies. The third party with more experiences can easily implement the collaborative techniques. It is an ideal choice for Pantein to change its inventory situation and save efforts. However, if Pantein seeks for the self development, then the cost of outsourcing and the technique investment should be taken into account.

Further Improvement Due to the limited data access, the research conducted actually only reflects the influences of the review period and the transit & production lead time on each party’s inventory management. The system dynamic model can be designed more accurate by incorporate more influencing factors. From the Pantein cost point of view, the analysis can be extend from the holding cost to more others such as order costs, transportation costs and personnel costs which could give a better perspective for Pantein to restructure their supply chain tailored for health care. For the simulation models, multi-product types and product lifecycles can be introduced into the model thus to give better indications of how the collaborative techniques perform under the more complex situations provided that more data can be 59

collected. There are other methods that can be incorporated in the supply chain to achieve the inventory optimization purpose such as joint management inventory (JMI), fourth party logistics (4PL), group purchasing organizations (GPO), sales and operation planning (S&OP) etc (Ross, 2009). Therefore, a hybrid collaborative situation can also be taken into account since not necessarily the entire chain should follow one unique collaboration strategy.

60

Reference Adam, C. and W. Cavendish (1995). Early Privatizations. In Privatizing Malaysia: Rents, Rhetoric, Realities. Boulder: Westview Press. Ann Vereecke (2006). Performance improvement through supply chain collaboration in Europe. International Journal of Operations & Production Management; Vol.26, Pg. 1176-1198. Anthony D. Ross, Vaidyanathan Jayaraman (2009). Strategic purchases of bundled products in a health care supply chain environment. Decision Science Vol.40, Pg. 269-293. Balci O. (1994). Validation, Verification and Testing Techniques Throughout the Life Cycle of A Simulation Study. Paper presented to Proceedings of the 26th conference on Winter Simulation, Orlando, Florida, United States. Bernhard J. Angerhofer and Marios C. Angelides (2000). System Dynamics Modelling in Supply Chain Management : Research Review. Paper presented to Proceedings of the 26th conference on Winter Simulation, Orlando, Florida, United States. Bert Meijboom Carolien de Blok and Katrien Luijkx (2007). Modular care and service packages for independently living elderly. International Journal of Operations & Production Management. Vol.30, Pg.75-97. Hu Chen, Yuqi Han and Bin Wang (2005). A Study of Inventory Management Based on System Dynamics. Engineering Management, Vol.19 No.3, 132-140. Dimitrios Vlachos, Patroklos Georgiadis, Eleftherios Iakovou (2007). A system dynamics model for dynamic capacity planning of remanufacturing in closed-loop supply chains. Computer & Operations Research 34, Pg. 367-394. E.Borst Eilers (2000) Amending Pharmaceutical Assistance Scheme. Dutch Health, Welfare and Sport Report. Forrester J.W. and P.M. Senge (1980). Tests for building confidence in system dynamics model. TIMS Studies in the Management Science,14 Pg.209-228. Gerald L. Kilty (2000). Inventory Management within the Supply Chain. Hospital Materiel Management Quarterly 21(4), Pg. 18-24. Glen Stuart Peace (1993). Taguchi Methods. 2nd edition. Massachusetts : 61

Addision-Wesley Jan Holmstrom, Johanna Smaros, Stephen M. Disney, Denis R. Towill (2003). Collaborative supply chain configurations: the implications for supplier performance in production and inventory control. Paper presented to 8th International Symposium on Logistics, Seville, Spain. Jay Nathan, John Trinkaus (1996). Improving healthcare means spending more time with patients and less time with inventory. Hospital Materiel Management Quarterly, 18(2), Pg.66-68. Jayanth Jayaram, Keah-Choon Tan (2010). Supply chain integration with third-party logistics providers. Production Economics, 125, Pg. 262-271. Javad Shahriar (1980). Multi-echelon inventory system in healthcare delivery organizations. Ph.D Thesis. University Microfilms International. John Sterman (2006). Business Dynamics: systems thinking and modeling for a complex world. Massachusetts: McGraw-Hill College. Kazim Sari (2008). On the benefits of CPFR and VMI: a comparative simulation study. Production Economics, 113, Pg.575-586. Kanvuz E. Maani, Robert Y. Cavana (2002). System thinking and modeling : understanding change and complexity. 2nd edition. Auckland : Pearson Education. Kingshuk K. Sinha, Emily J. Kohnke (2009). Health care supply chain design: toward linking the development and delivery of care globally. Decision Science Vol. 40, Pg.197-212. Lawrence Nicholson, Asoo. Vakharia, and S, Selcuk Erenguc (2004). Outsourcing inventory managmenet decisions in healthcare: models and application. European Journal of Operational Research 154, Pg.271-290. Lawrence A. Nicholson (2002). Improving inventory management in healthcare systems. Social and Economic Studies 51(3), Pg.66-103. H.L. Lee, V. Padmanabhan and S.Whang (1997). Information distortion in s supply chain: the bullwhip effect. Management Science 43, Pg. 546-558 Http://anthony.ASCET.com (2006). Supply chain collaboration: success in the new internet economy, March, 2006. Matthias Holweg, Stephen Disney, Jan Holmstrom, Jonhanna Smaros (2005). Supply 62

chain collaboration: Making Sense of the Strategy Continuum. European Management Journal; Vol. 23, No. 2, Pg. 170-181. M.A.Darwish, O.M. Odah (2010). Vendor managed inventory mnodel for single-vendor multi-retailer supply chains. European Journal of Operation Research 204, Pg. 473-484. M.A.Schouten (2006). Improving the supply chain in the health care sector. Tilburg University. Noorfa Haszlinna Mustaffa, Andrew Potter (2009). Healthcare supply chain management in Malaysia: a case study. Supply Chain Management: An International Journal, 14/3, Pg 234-243. Oscar Rubiano Ovalle, Adolfo Crespo Marquez (2003). The effectiveness of using e-collaboration tools in the supply chain: an assessment study with system dynamics. Journal of Purchasing & Supply Management 9, Pg. 151-163. Patroklos Georgiadis, Dimitrios Vlachos and Eleftherios Iakovou (2005). A system dynamics modeling framework for the strategic supply chain management of food chains. Journal of Food Engineering 70, Pg. 351-364. P.Danese (2010). Collaboration forms, information and communication technologies, and coordination mechanisms in CPFR. Production Research, 44(16) Pg.3207-3226. Robert E. Barnes (1987). Reshaping the supply chain: A cooperative effort. Hospital Materiel Management Quarterly, 9(2) ,Pg. 90-92. Roberto Poles, France Cheong (2001). Inventory control in closed loop supply chain using system dynamics. RMIT University, Australia. Rundle, Rhonda, L.(1997). Doctors’ Orders Hospital Cost cutters push use of scanners to track inventories – Pioneers promote bar codes in money saving efforts, urge supplier to help –taking cues from Wal-Mart. The Wall Street Journal, 1997.Pg.49-66. Shotaro Minegishi, Daniel Thiel (2000). System dynamics modeling and simulation of a particular food supply chain. Simulation Practice and Theory 8, Pg.321-339. S.M.Disney, D.R.Towill (2003). The effect of vendor managed inventory (VMI) dynamics on the bullwhip effect in supply chain. Production Economics 85 Pg. 199-215. S.M.Disney, D.R.Towill (2003). Vendor-managed inventory and bullwhip reduction in 63

a two-level supply chain. Operations & Production Managements. Vol 23 No. 6 Pg. 625-651. Sultan N.Turhan, Ozalp Vayvay (2009). Modeling of VMI implementation via SOA in a healthcare supply chain. Paper presented to European and Mediterranean Conference on information system, Izmir. Sven Axsater (2006). Invetnory Control. 2nd edition.

New York: Springer Science.

Toru Higuchi, Marvin D.Troutt (2004). Dynamic simulation of the supply chain for a short life cycle product. Computer & Operations Research. 31 Pg.1097-1114. Vijay Wadhwa and A.Ravi Ravindran (2007). Vendor selection in outsourcing. Computers & Operations Research, 34,3725-3737. Westert GP, Berg ML van den, Koolman X, Verkleij H (2008). Dutch Health Care Performance Report 2008. Bilthoven: National Institution for Public Health and Environment. William H. Saphir (1994). Implementing a new inventory management system. Hospital Materiel Management Quarterly, 15(3), Pg. 42-45. Xumei Yuan, Ling Shen, Jalal Ashayeri (2009). Dynamic simulation assessment of collaboration strategies to management demand gap in high-tech product diffusion. Hebei University of Science & Technology, Qinhuangdao, China. Yaman Barlas (1994). Model validation in system dynamics. Paper presented to the 1994 International System Dynamics Conference. Yaman Barlas (1996). Formal aspects of model validity and validation in system dynamics. System Dynamics Review Vol. 12, Pg.183-210. Zhang Yong and David Dilts (2004). System dynamics of supply chain network organization structure. Information Systems and e-Business Management, 2,187-206.

64

Appendix Appendix 1 Experimental Design Simulation Result

Simulation

Traditional Model Hospital Department

Central Warehouse

Supplier

Total Average Inventory

1

417.79

1289.31

2424.58

4131.67

2

417.79

1325.16

2683.87

4426.82

3

416.76

1269.33

1730.37

3416.45

4

404.21

1108.79

3050.30

4563.30

5

404.18

1170.63

2561.40

4136.21

6

408.44

1196.45

2569.35

4174.24

7

509.22

1658.85

4284.33

6452.39

8

548.78

1829.27

3650.78

6028.83

9

497.37

1613.39

2454.75

4565.52

10

476.13

1073.29

2415.82

3965.24

11

476.13

1077.51

1948.72

3502.37

12

473.05

1096.96

2431.34

4001.35

13

483.18

1542.74

2915.56

4941.47

14

483.18

1466.44

1837.34

3786.96

15

483.18

1465.43

2171.46

4120.06

16

492.94

1048.28

2024.60

3565.82

17

492.94

1062.26

2443.44

3998.64

18

498.91

1081.51

1871.57

3451.99

19

567.06

1423.94

3073.89

5064.90

20

552.24

1373.13

2740.98

4666.34

21

566.34

1386.92

2568.51

4521.76

22

577.42

1320.74

2372.58

4270.74

23

577.42

1322.05

2958.81

4858.28

24

575.26

1303.66

2204.97

4083.90

25

543.82

1415.86

2535.56

4495.24

26

546.04

1425.59

2244.38

4216.01

27

543.82

1404.32

2804.94

4753.08

65

Simulation

VMI Hospital Department

Central Warehouse

Supplier

Total Average Inventory

1

551.79

584.00

2315.77

3451.56

2

551.79

624.89

2634.99

3811.67

3

546.15

534.13

1718.74

2799.02

4

524.56

513.09

2482.72

3520.37

5

524.56

523.77

2261.38

3309.71

6

524.56

536.53

2407.56

3468.65

7

567.52

644.09

3199.67

4411.28

8

603.73

764.39

2026.60

3394.72

9

557.14

587.94

1953.08

3098.17

10

698.33

502.36

2469.90

3670.59

11

698.33

517.86

2112.43

3328.62

12

698.33

526.80

2431.43

3656.56

13

839.21

876.66

3033.87

4749.74

14

832.65

796.86

1914.27

3543.78

15

833.73

793.06

2085.50

3712.29

16

686.93

484.27

2281.63

3452.83

17

686.93

513.39

2552.77

3753.10

18

703.92

481.85

2056.23

3242.00

19

1025.66

862.01

2928.14

4815.81

20

1036.49

867.76

2066.59

3970.84

21

1020.99

780.20

2077.96

3879.15

22

864.69

467.18

2180.89

3512.76

23

864.69

467.47

2810.71

4142.86

24

850.64

376.76

1999.64

3227.04

25

819.63

387.53

2246.45

3453.62

26

822.76

377.52

2310.87

3511.15

27

822.76

390.57

2553.04

3766.37

66

Simulation

CPFR Hospital Department

Central Warehouse

Supplier

Total Average Inventory

1

712.37

1565.63

1879.73

4157.73

2

712.37

1569.64

1644.70

3926.70

3

712.37

1571.91

1331.89

3616.17

4

709.85

1339.45

1866.48

3915.78

5

709.85

1345.46

1635.65

3690.96

6

709.85

1348.32

1328.76

3386.92

7

712.20

1713.62

1864.97

4290.79

8

712.20

1722.23

1634.43

4068.86

9

712.20

1723.83

1318.78

3754.81

10

906.86

1312.42

1839.17

4058.45

11

906.86

1315.94

1535.01

3757.82

12

906.86

1287.82

1397.98

3592.67

13

918.58

1758.22

1868.84

4545.64

14

918.58

1760.11

1547.62

4226.31

15

918.58

1737.38

1426.42

4082.38

16

913.61

1534.35

1853.26

4301.23

17

913.61

1537.24

1537.71

3988.56

18

913.61

1511.38

1415.61

3840.61

19

1119.98

1732.97

1743.22

4596.17

20

1119.98

1710.39

1629.37

4459.73

21

1119.98

1720.60

1406.96

4247.54

22

1113.55

1509.84

1727.65

4351.04

23

1113.55

1483.97

1614.17

4211.69

24

1113.55

1494.91

1401.37

4009.83

25

1108.47

1382.97

1743.13

4234.57

26

1108.47

1363.71

1618.43

4090.61

27

1108.47

1372.18

1409.49

3890.14

67

Appendix 2:Model Program Traditional Inventory Model central warehouse central_warehouse_goods_in_transit(t) = central_warehouse_goods_in_transit(t - dt) + (supplier_fullfillment - central_warehouse_goods_received) * dtINIT central_warehouse_goods_in_transit = 0 TRANSIT TIME = 12 INFLOW LIMIT = INF CAPACITY = INF INFLOWS: supplier_fullfillment = fullfill_central__warehouse_order OUTFLOWS: central_warehouse_goods_received = CONVEYOR OUTFLOW central_warehouse__inventory(t) = central_warehouse__inventory(t - dt) + (central_warehouse_goods_received - fullfill_hospital_department_order) * dtINIT central_warehouse__inventory = average_demand*central_warehouse__review_period INFLOWS: central_warehouse_goods_received = CONVEYOR OUTFLOW OUTFLOWS: fullfill_hospital_department_order = Min(central_warehouse__inventory,hospital_order_and_central_warehosue_backlog) hospital_order_and_central_warehosue_backlog(t) = hospital_order_and_central_warehosue_backlog(t - dt) + (hospital_department_order - central_warehouse_backlog_release) * dtINIT hospital_order_and_central_warehosue_backlog = 0 INFLOWS: hospital_department_order = if(hospital_adjustment>0)Then(hospital_adjustment)Else(0) OUTFLOWS: central_warehouse_backlog_release = central_warehouse__fullfillment hospital_order__accumulate(t) = hospital_order__accumulate(t - dt) + (hospital_demand) * dtINIT hospital_order__accumulate = 0 INFLOWS: hospital_demand = hospital_department_order 68

central_warehosue_inventory_vs_backlog_for_adjustment = hospital_order_and_central_warehosue_backlog-central_warehouse__inventory central_warehouse_adjustment = if(central_warehouse__desired_inventory=0)Then(0)Else(central_warehouse__desire d_inventory+central_warehosue_inventory_vs_backlog_for_adjustment-central_ware house_goods_in_transit) central_warehouse_backlog = if(central_warehouse__inventory>hospital_order_and_central_warehosue_backlog)T hen(0)Else(hospital_order_and_central_warehosue_backlog-central_warehouse__inve ntory) central_warehouse_forecast_demand = if(TIME=central_warehouse__review_period)Then(hospital_order__accumulate/TIM E*central_warehouse_replenishment_time)Else(if(MOD(TIME,central_warehouse__r eview_period)=0)Then(hospital_order__accumulate-History(hospital_order__accumu late,TIME-central_warehouse_replenishment_time))Else(0)) central_warehouse_replenishment_time = 28 central_warehouse__desired_inventory = if(central_warehouse_forecast_demand=0)Then(0)Else(central_warehouse_forecast_d emand+1.64*sqrt(central_warehouse_replenishment_time)*hospital_demand_sd) central_warehouse__review_period = 14 hospital_demand_sd = 70 customer_demand__accumulate(t) = customer_demand__accumulate(t - dt) + (customer_demand) * dtINIT customer_demand__accumulate = -actural_demand INFLOWS: customer_demand = actural_demand hold(t) = hold(t - dt) + (exponential__interarrival_time) * dtINIT hold = 0 INFLOWS: exponential__interarrival_time = EXPRND(1,84563) actural_demand = average_demand*exponential__interarrival_time average_demand = 100 customer_order_and_hospital_department_backlog(t) = customer_order_and_hospital_department_backlog(t - dt) + (customer_order hospital_backlog__release) * dtINIT customer_order_and_hospital_department_backlog = 0 INFLOWS: customer_order = History(customer_demand,TIME-1) OUTFLOWS: hospital_backlog__release = fullfill_customer__demand 69

hospital_inventory(t) = hospital_inventory(t - dt) + (central_warehouse__fullfillment fullfill_customer__demand) * dtINIT hospital_inventory = average_demand*hospital_deparment__review_period INFLOWS: central_warehouse__fullfillment = fullfill_hospital_department_order OUTFLOWS: fullfill_customer__demand = Min(hospital_inventory,customer_order_and_hospital_department_backlog) customer__demand_sd = 10 hospital_adjustment = if(hospital_department_desired_inventory=0)Then(0)Else(hospital_department_desire d_inventory-hospital_inventory+History(hospital_backlog,TIME-1)) hospital_backlog = if(hospital_inventory>customer_order_and_hospital_department_backlog)Then(0)Els e(customer_order_and_hospital_department_backlog-hospital_inventory) hospital_deparment__review_period = 7 hospital_department_desired_inventory = IF(hospital_department_forecast_demand=0)Then(0)Else(hospital_department_foreca st_demand+1.64*sqrt(hospital_department_replenishment_time)*customer__demand _sd) hospital_department_forecast_demand = If(TIME=hospital_deparment__review_period-1)Then(customer_demand__accumula te/TIME*hospital_department_replenishment_time)Else(if(MOD(TIME,hospital_dep arment__review_period)=hospital_deparment__review_period-1)Then(customer_dem and__accumulate-History(customer_demand__accumulate,TIME-hospital_departmen t_replenishment_time))Else(0)) hospital_department_replenishment_time = 9 central_warehouse_order_accumulate(t) = central_warehouse_order_accumulate(t - dt) + (central_warehouse_demand) * dtINIT central_warehouse_order_accumulate = 0 INFLOWS: central_warehouse_demand = central_warehouse__order central_warehouse_order_and_supplier_backlog(t) = central_warehouse_order_and_supplier_backlog(t - dt) + (central_warehouse__order supplier_backlog__release) * dtINIT central_warehouse_order_and_supplier_backlog =0 INFLOWS: central_warehouse__order = if(central_warehouse_adjustment>0)Then(central_warehouse_adjustment)Else(0) OUTFLOWS: 70

supplier_backlog__release = supplier_fullfillment production(t) = production(t - dt) + (productivity - goods_sent_to_inventory) * dtINIT production = 0 TRANSIT TIME = 2 INFLOW LIMIT = INF CAPACITY = INF INFLOWS: productivity = supplier__adjustment OUTFLOWS: goods_sent_to_inventory = CONVEYOR OUTFLOW supplier_inventory(t) = supplier_inventory(t - dt) + (goods_sent_to_inventory fullfill_central__warehouse_order) * dtINIT supplier_inventory = average_demand*supplier_review_period INFLOWS: goods_sent_to_inventory = CONVEYOR OUTFLOW OUTFLOWS: fullfill_central__warehouse_order = Min(supplier_inventory,central_warehouse_order_and_supplier_backlog) central_warehouse__demand_sd = 140 supplier_backlog = if(supplier_inventory>central_warehouse_order_and_supplier_backlog)Then(0)Else(c entral_warehouse_order_and_supplier_backlog-supplier_inventory) supplier_desired__inventory = if(supplier__forecast_demand=0)Then(0)Else(supplier__forecast_demand+1.64*sqrt( supplier__replenishment_time)*central_warehouse__demand_sd) supplier_inventory_vs_backlog_for_adjustment = central_warehouse_order_and_supplier_backlog-supplier_inventory supplier_review_period = 14 supplier__adjustment = if(supplier_desired__inventory=0)Then(0)Else(supplier_desired__inventory+supplier _inventory_vs_backlog_for_adjustment-production) supplier__forecast_demand = if(TIME=supplier_review_period+1)Then(central_warehouse_order_accumulate/supp lier_review_period*supplier__replenishment_time)Else(if(MOD(TIME,supplier_revi ew_period)=1)Then((central_warehouse_order_accumulate-History(central_warehous e_order_accumulate,TIME-supplier_review_period-supplier_review_period))/supplie r_review_period/2*supplier__replenishment_time)Else(0)) supplier__replenishment_time = 18 71

Not in a sector time1 = TIME

72

VMI Model central warehouse central_warehouse_goods_in_transit(t) = central_warehouse_goods_in_transit(t - dt) + (supplier_fullfillment - central_warehouse__goods_received) * dtINIT central_warehouse_goods_in_transit = 0 TRANSIT TIME = 12 INFLOW LIMIT = INF CAPACITY = INF INFLOWS: supplier_fullfillment = fullfill_central__warehouse_order OUTFLOWS: central_warehouse__goods_received = CONVEYOR OUTFLOW central_warehouse__inventory(t) = central_warehouse__inventory(t - dt) + (central_warehouse__goods_received - fullfill_hospital__demand) * dtINIT central_warehouse__inventory = average_demand*central_warehouse__review_period INFLOWS: central_warehouse__goods_received = CONVEYOR OUTFLOW OUTFLOWS: fullfill_hospital__demand = Min(hospital_department__adjustment,central_warehouse__inventory) central_warehouse_desired_inventory = if(central_warehouse_forecast_demand_for_itself=0)Then(0)Else(central_warehouse_ forecast_demand_for_itself+1.64*sqrt(central_warehouse_replenishment_time)*custo mer_demand_sd) central_warehouse_forecast_demand_for_itself = if(TIME<1)Then(0)Else(If(TIME=central_warehouse__review_period)Then(custome r_demand__accumulate/central_warehouse__review_period*central_warehouse_reple nishment_time)Else(if(MOD(TIME,central_warehouse__review_period)=0)Then(cus tomer_demand__accumulate-History(customer_demand__accumulate,TIME-central_ warehouse_replenishment_time))Else(0))) central_warehouse_replenishment_time = 28 central_warehouse__adjustment = if(central_warehouse_desired_inventory=0)Then(0)Else(central_warehouse_desired_i nventory+hospital_department_backlog-central_warehouse_goods_in_transit-central_ warehouse__inventory) central_warehouse__review_period = 14 customer_order_and_hospital_department_backlog(t) = customer_order_and_hospital_department_backlog(t - dt) + (customer_order 73

hospital_backlog__release) * dtINIT customer_order_and_hospital_department_backlog = 0 INFLOWS: customer_order = History(customer_demand,TIME-1) OUTFLOWS: hospital_backlog__release = fullfill_customer__demand hospital_inventory(t) = hospital_inventory(t - dt) + (central_warehouse__fullfillment fullfill_customer__demand) * dtINIT hospital_inventory = average_demand*hospital_department_review_period INFLOWS: central_warehouse__fullfillment = fullfill_hospital__demand OUTFLOWS: fullfill_customer__demand = Min(hospital_inventory,customer_order_and_hospital_department_backlog) central_warehouse_forecast_for_hospital_department = if(TIME<1)Then(0)Else(if(TIMEcustomer_order_and_hospital_department_backlog)Then(0)Els e(customer_order_and_hospital_department_backlog-hospital_inventory) hospital_department_replenishement_time = 9 hospital_department_review_period = 7 hospital_department__adjustment = if(hospital_inventory
INFLOWS: central_warehouse__order = if(central_warehouse__adjustment>0)Then(central_warehouse__adjustment)Else(0) OUTFLOWS: supplier_backlog__release = supplier_fullfillment production(t) = production(t - dt) + (productivity_2 - goods_sent_to_inventory) * dtINIT production = 0 TRANSIT TIME = 2 INFLOW LIMIT = INF CAPACITY = INF INFLOWS: productivity_2 = supplier__adjustment OUTFLOWS: goods_sent_to_inventory = CONVEYOR OUTFLOW supplier_inventory(t) = supplier_inventory(t - dt) + (goods_sent_to_inventory fullfill_central__warehouse_order) * dtINIT supplier_inventory = average_demand*supplier_review_period INFLOWS: goods_sent_to_inventory = CONVEYOR OUTFLOW OUTFLOWS: fullfill_central__warehouse_order = Min(supplier_inventory,central_warehouse_order_and_supplier_backlog) central_warehouse__demand_sd = 140 supplier_backlog = if(supplier_inventory>central_warehouse_order_and_supplier_backlog)Then(0)Else(c entral_warehouse_order_and_supplier_backlog-supplier_inventory) supplier_desired__inventory = if(supplier__forecast_demand=0)Then(0)Else(supplier__forecast_demand+1.64*sqrt( supplier__replenishment_time)*central_warehouse__demand_sd) supplier_inventory_vs_backlog_for_adjustment = central_warehouse_order_and_supplier_backlog-supplier_inventory supplier_review_period = 14 supplier__adjustment = if(supplier_desired__inventory=0)Then(0)Else(supplier_desired__inventory+supplier _inventory_vs_backlog_for_adjustment-production) supplier__forecast_demand = if(TIME=supplier_review_period+1)Then(central_warehouse_order_accumulate/supp 75

lier_review_period*supplier__replenishment_time)Else(if(MOD(TIME,supplier_revi ew_period)=1)Then((central_warehouse_order_accumulate-History(central_warehous e_order_accumulate,TIME-supplier_review_period-supplier_review_period))/supplie r_review_period/2*supplier__replenishment_time)Else(0)) supplier__replenishment_time = 18 Not in a sector customer_demand__accumulate(t) = customer_demand__accumulate(t - dt) + (customer_demand) * dtINIT customer_demand__accumulate = -demand INFLOWS: customer_demand = demand Noname_2(t) = Noname_2(t - dt) + (Noname_1) * dtINIT Noname_2 = 0 INFLOWS: Noname_1 = EXPRND(1,84563) average_demand = 100 demand = average_demand*Noname_1 time1 = TIME

76

CPFR Model central warehouse central_warehouse_goods_in_transit(t) = central_warehouse_goods_in_transit(t - dt) + (supplier_fullfillment - central_warehouse__goods_received) * dtINIT central_warehouse_goods_in_transit = 0 TRANSIT TIME = 13 INFLOW LIMIT = INF CAPACITY = INF INFLOWS: supplier_fullfillment = fullfill_central__warehouse_order OUTFLOWS: central_warehouse__goods_received = CONVEYOR OUTFLOW central_warehouse__inventory(t) = central_warehouse__inventory(t - dt) + (central_warehouse__goods_received - fullfill_hospital__demand) * dtINIT central_warehouse__inventory = average_demand*central_warehouse__review_period INFLOWS: central_warehouse__goods_received = CONVEYOR OUTFLOW OUTFLOWS: fullfill_hospital__demand = Min(hospital_department__adjustment,central_warehouse__inventory) central_warehouse_desired_inventory = central_warehouse_forecast_demand+1.64*sqrt(central_warehouse_replenishment_ti me)*customer_demand_sd central_warehouse_forecast_demand = if(TIME<1)Then(0)Else(if(TIME
INFLOWS: customer_demand = demand Noname_2(t) = Noname_2(t - dt) + (Noname_1) * dtINIT Noname_2 = 0 INFLOWS: Noname_1 = EXPRND(1,84563) average_demand = 100 demand = average_demand*Noname_1 customer_order_and_hospital_department_backlog(t) = customer_order_and_hospital_department_backlog(t - dt) + (customer_order hospital_backlog__release) * dtINIT customer_order_and_hospital_department_backlog = 0 INFLOWS: customer_order = History(customer_demand,TIME-1) OUTFLOWS: hospital_backlog__release = fullfill_customer__demand hospital_inventory(t) = hospital_inventory(t - dt) + (central_warehouse__fullfillment fullfill_customer__demand) * dtINIT hospital_inventory = average_demand*hospital_department_review_period INFLOWS: central_warehouse__fullfillment = fullfill_hospital__demand OUTFLOWS: fullfill_customer__demand = Min(hospital_inventory,customer_order_and_hospital_department_backlog) customer_demand_sd = 10 hospital_department_backlog = if(hospital_inventory>customer_order_and_hospital_department_backlog)Then(0)Els e(customer_order_and_hospital_department_backlog-hospital_inventory) hospital_department_forecastd_demand = if(TIME<1)Then(0)Else(if(TIME
hosptial_department_desired_inventory = hospital_department_forecastd_demand+1.64*sqrt(hospital_department_replenishem ent_time)*customer_demand_sd production(t) = production(t - dt) + (productivity_2 - goods_sent_to_inventory) * dtINIT production = 0 TRANSIT TIME = 3 INFLOW LIMIT = INF CAPACITY = INF INFLOWS: productivity_2 = supplier__adjustment OUTFLOWS: goods_sent_to_inventory = CONVEYOR OUTFLOW supplier_inventory(t) = supplier_inventory(t - dt) + (goods_sent_to_inventory fullfill_central__warehouse_order) * dtINIT supplier_inventory = average_demand*supplier_review_period INFLOWS: goods_sent_to_inventory = CONVEYOR OUTFLOW OUTFLOWS: fullfill_central__warehouse_order = Min(supplier_inventory,central_warehouse__adjustment) supplier_desired__inventory = if(supplier__forecast_demand=0)Then(0)Else(supplier__forecast_demand+1.64*sqrt( supplier__replenishment_time)*customer_demand_sd) supplier_review_period = 14 supplier__adjustment = if(supplier_inventory+production
79

Loading...

Master Thesis - University of Tilburg - Tilburg University

Master Thesis Evaluating Inventory Collaboration in Healthcare Industry: A System Dynamics Study Zhongning Li [166875] [BSc. DongBei University of Fi...

858KB Sizes 7 Downloads 12 Views

Recommend Documents

Master Thesis - University of Tilburg - Tilburg University
Master Thesis. The effect of online coaching interventions on employee's self-efficacy and resilience at work. A quantit

master thesis - University of Tilburg - Tilburg University
Aug 17, 2010 - MASTER THESIS. Fashion Cities and the Urban Regime Theory. Tilburg University. Faculty of Social and Beha

MSc Thesis - University of Tilburg - Tilburg University
Dec 13, 2010 - A-F Rutkowski. J. Pluyter MSc. Exam committee dr. A-F Rutkowski dr. B.A. van de Walle. Study. : Tilburg U

master thesis international business law - University of Tilburg
Jun 12, 2010 - in Europe in 2010. 26. Table 6 – Nature of specific public regulations on sports agents. 34. List of Fi

Master Thesis_R.A.Sabrina Stamboel - University of Tilburg
4 About BRISyariah. Retrieved May 9, 2012, from http://www.brisyariah.co.id/?q=sejarah. 5 Ibid. 6 Kementerian Badan Usah

Digital ethnography - Tilburg University
increasing 'de-computerisation' (i.e. the increased use of smart phones and other mobile ... e.g. Georgakopoulou 2013; L

IVA, university of Tilburg, the Netherlands | CONTABILIDADE
The CONTABILIDADE Department at IVA, university of Tilburg, the Netherlands on Academia.edu.

Tilburg University - Program International Business Tax Law
The sub-track Globalization focuses on international economic integration, such as the European Union and the WTO. Numer

Tilburg University Macroeconomic policy - Research portal
Macroeconomic policy coordination during the various phases of economic and monetary integration in Europe van der Ploeg

Citing Sources in APA Style - Tilburg University
Cite an author as 'Anonymous' only if that author is specifically designated as such in your source. In the reference ..