Evaluating Inventory Management Performance: a Preliminary Desk-Simulation Study Based on IOC Model

Transcription

1 Evaluating Inventory Manageent Perforance: a Preliinary Desk-Siulation Study Based on IOC Model Flora Bernardel, Roberto Panizzolo, and Davide Martinazzo Abstract The focus of this study is on preliinary use and ipleentation of an innovative stock perforance evaluation procedure, based on the Inventory Operating Curves odel. The evaluation of Inventory Syste perforances is becoing a priary concern for senior anagers in any sectors, due to the global financial crisis and the budget-cutting policies. However, traditional odels require to siultaneously assess and onitor any interrelated paraeters, in order to describe the logistic behavior of a huge quantity of ites. In that sense, traditional tools are not very useful or effective for decision-aking and anagerial practice. The Inventory Operating Curves (IOC) odel, based on an aggregate Inventory Level approach, can be successfully eployed to perfor an evaluation of the current Inventory Syste. In addition, it can support the re-design of the logistic paraeters consistently with the anifold requireents of the downstrea channel. The case study provided in this paper refers to a selected Italian pharaceutical anufacturer. The evaluation of the real inventory data available, is developed around the seasonality issues which are expected to affect the deand pattern of a particular product faily. Index Ters Inventory Manageent, Desk Siulation, Inventory Control. I. INTRODUCTION Today anufacturing organizations are dealing with ore coplex copetitive dynaics than those experiented in the recent years, because of the persistence of the financial crisis and the raising of soe well known challenges, such as globalization, sustainability and a ore sensitive consuer arket. Soe new general clais for an increased productivity, i.e. a high efficiency use of resources, are eerging. Since inventories represent a significant investent by any firs, and with annual carrying costs typically ranging fro 20 to 40 percent of inventory value, anaging the well is a top-anageent priority. In practice, while a proinent focus on the Inventory Manageent policies (i.e. Stock Replenishent Policies) is the priary result of the literature analysis [1], two general conceptual streas can also be recognized. On the one hand, the Inventory Optiization strea includes studies aied to identify the optial operating point, i.e. the specific value set for the inventory paraeters, which lead a fir to achieve, at the sae tie, a fixed service level and a certain perforance of efficiency. On the other hand, the Inventory Manageent strea provides several ethods and studies based on a Manuscript received July 31, 2012; revised Septeber 24, F. Bernardel and R. Panizzolo are with the Manageent and Engineering Departent, University of Padua, Italy (e-ail: flora.bernardel@unipd.it; roberto.panizzolo@unipd.it). D. Martinazzo is with the Manufacturing Departent, Carel Industries, Padua, Italy (e-ail: davide.artinazzo@carel.co). odeling approach. In these works, the basic inventory paraeters, such as the consuer deand (i.e. stochastic or deterinistic), the tie basis (i.e. continuous or periodic) and the perforances of the inventory syste (i.e. the logistic or financial perspective), are specified in analytical forulas, in order to describe the integration between the logistic activities and the traditional inventory control systes. In this context, the evaluation of the perforance for an Inventory Syste currently in use, cannot be uniquely aied to the traditional need to cut costs, by reducing the quantities of stocked aterials. Therefore, it rather evolved in the requisite to properly regulate and adjust the inventory levels in their aggregate for, in order to achieve soe strategic objectives. As counterpart, it s of priary iportance that the design of the logistic paraeters is realized consistently with the anifold requireents of the downstrea channel. This paper illustrates an original approach to the inventory anageent perforance evaluation, by providing a IOC-odel-based procedure, which can be developed fro readily available copany data. Specifically, the purpose of this research is to introduce preliinary results of how the IOC odel can be foralized [2] and eployed in the actual business environent as an evaluation tool, in order to support the inventory and stocks control process. The research is applicative in nature with the ai to contribute to the knowledge and diffusion of a well docuented theory, i.e. the IOC odel. Despite the descriptive validation provided by the author [2], practical iplications and potential of the IOC odel are still unexplored and highly unrealized. Therefore, a ethodological approach is proposed, which clarifies how collect, treat and analyzed data fro a real fir inventory syste through ultiple desk-siulation sessions. II. THEORETICAL BACKGROUND The theoretical odel of Logistics Curves, originally provided by Wiendahl et al. [3] and, in recent ties, applied to the Inventory Manageent area [2] (i.e. the IOC odel), is a fraework which allows to derive the sufficient stock level to eet a certain custoer deand in any given point in tie. This ethod is proven to be an effective support to the inventory sizing decision. Therefore, after a categorization of the stocked ites, it allows to analyze for each of the identified groups the relative relationship between the variable delivery capability (here deployed through two perforance indicators such as service level and average delay of delivery ) and the average aount of aterial held in stock. DOI: /IJEEEE.2012.V

2 In particular, two types of curves are foralized in the IOC odel [2]: The Service Level Operating Curve (SLOC), that investigates values potentially achievable for the service level indicator, while considering different configurations of the input-output stock control syste; The Delivery Delay Operating Curve (DDOC) that calculates the average delivery delay due to stock-outs, in the sae input-output control configurations. For soe coprehensive considerations about the theoretical odel, the reader is pleased to refer to the work of Nyhuis et al. [4], in which any relevant indications are provided as regard to the two operative fors taken by the odel: the graphical ethod through the representation of the logistic variables above, based on the siulation of the logistic process for individual groups of stocked aterials, and the approxiate equations ethod, based on a deductive-experiental odeling of the logistics process. The advantage in applying the IOC odel to the Inventory Manageent evaluation proble, copared to the traditional ethods (e.g. the throughput diagra), results in the effective and siultaneous representation of the logistic variables and the perforance objectives and their utual interrelations [3]. As in next sessions we are going to refer only to the SLOC curve, soe essential notes on the forulas are cited, in order to facilitate the reading of this docuent. The approxiate forula for the Service Level Operating Curves (SLOC) can be derived fro the subsequent position: SL where SL is the Service Level, which is at 100% in the liit condition of IL = IL1. Thus, the foregoing paraeters can be calculated by the forula below: IL Qin Qout 2 Sybols eployed in the above equations and their eanings are shown in the Table I. TABLE I: SUMMARY OF THE APPROXIMATE IOC MODEL TERMS Sybol Measure Unit Description IL IL 1 2 L ax * RD Q ax D D LTR 2 1 ax * L+ax days Maxiu delay in the delivery date as input in the inventory stock L-ax days Maxiu advance in the delivery date as input in the inventory stock Q-ax units Maxiu negative deviation in the output of finished goods Q+ax units Maxiu positive deviation in the input volue of finished goods Qout units Average lot size as output fro the inventory stock Qin units Average lot size as input to the inventory stock Dax units per day Maxiu Deand LTR days Replenishent Lead tie (i.e. the anufacturing lead tie) IN units Average input lot size (i.e. the production lot size) R units per day Average Deand Din units per day Miniu Deand III. METHODOLOGY The practical application of the IOC odel, in the for introduced in the previous section, ay be based on historical data. Alternatively, if historical data is not available for the Inventory Syste of a copany, the logistic paraeters ay be derived through a odeling analysis. The latter approach can be ipleented with reference to the huge quantity of theoretical odels provided by specific literature, such as Shapiro et al. [5]. We discuss here only the forer approach, because it was the one applied in the case study developed below. This approach, which is general and repeatable, involved five ajor steps. Step 1: Identification of the unit of analysis. All the aterials stocked or Stock Keeping Units (SKUs), included and anaged in the inventory syste, should be classified through a logistic segentation ethod. The ost popular is the ABC analysis, which divides products based on soe logistic characteristics. In this context, the selected variables should be relevant for the subsequent inventory evaluation. Soe general exaples ay include operative easures such as the product service goals, the product unit cost, the unit handling costs, the flow-through volue, the lead tie, and others. Consistently with a recoended procedure [6], due to the IOC odel applies to the individual product level, the desk-siulation below is perfored on a single ite or SKU, which has been identified as a representative saple. Step 2: Inference of the current perforance. Since a SKU or a product faily have been selected, according to the IOC odel, the service level and average delivery delay paraeters have to be extracted for inclusion in the segentation process. This preliinary analysis can be derived fro general stock status reports, which usually include inventory levels and shipping volues. According to the stocking rules applied to the, SKUs which are siilar, ay be classified together and plotted as a group in the relative graphs. At this stage is iportant to define the tie basis for the collection of data. If stock reports are available for periods of tie less than a year, inventory levels should be averaged and the throughput should be sued, in order to refer the to the sae tie interval. The IOC curves should be prepared for the ites or SKUs failies selected, on the tie basis of interest. Then, the target levels for the expected or the future state should be specified and the corresponding operating points plotted on the IOC figure. Step 3: Preliinary odel setting. The actual operating points are then placed on the IOC curves, and the relative stock levels as currently set in the Inventory Control syste, are identified. As reported above, the stock levels can be drawn fro the IOC figures through a graphical approach or derived fro the approxiated equations. Generally, the siulation reflects the logistic activity realized at one specific instant in tie, which can be assued as now, or can be placed in the past if the evaluation of the Inventory Syste is undertaken in order to give new insights over an experiented situation. Step 4: Analysis of deviations fro the expected values. After the paraeters have been identified, the distance between the present and the future state is evaluated and analyzed. A copany ay eploy foral inventory anageent rules or inforal judgents-based rules, or a 405

3 cobination of these. Inventory levels ay deviate fro those expected or planned based on the rules being applied. Even not grounded in a ite-by-ite analysis, the extent of the deviation, in ters of a benchark against which current stocking levels ay be copared, should be deducted fro the IOC curves plotted at the previous step. This could lead to an Inventory sizing proble. Therefore, because the IOC odel apply at an aggregate level and eploys the average aount of ites stocked, the siulation ay result in a reduction or an increase of the current stocked quantity for the selected SKU, according to the objectives established on logistic perforances. Step 5: Creation of an action plan. Then, the appropriate actions to achieve the levels planned for perforance are identified. The easures can range fro the deand leveling to the iproveent of forecasting techniques currently applied, fro the reducing of lead ties to the re-sizing of the input and output batches. Notice that, while considering and redefining this types of paraeters or indicators, the siulative analysis reains at the operational level. Strategic issues about the logistic behavior (actual or planned) of the investigated SKU, ust be factored in, during the qualitative judgent phase of objectives setting. IV. THE CASE STUDY PERFORMED The procedure introduced above, and in particular the desk-siulation analysis, is perfored with reference to the inventory data provided by a anufacturer in the pharaceutical and biotech industry, which actually produces and distributes in the Northern Italy. In recent years, this copany experiented the aggressive requireents posed by the distribution channel, whose logistic perforance ust eet specific regulations and deal with new arket dynaics. Moreover, in this industry the inherent coplexity of the downstrea channel is exacerbated by instances of a very short fulfillent tie in face of an extreely liited nuber of ites per order, and by a regulation on the axiu delivery delay acceptable by the custoer. As a consequence, the copetition results significantly affected by cost reduction progras, because of a current phase of arket saturation and price pressure. Indeed, ost available evidence has found the Italian deand for pharaceuticals to be quite insensitive or even unresponsive to price policies. Those factors led the top-anageent to redesign the current inventory stock syste of the anufacturing copany, in light of a ore systeic efficiency achievable as regard to the logistic perforance objectives. In addition, soe key product, even belonging to the A-class of an ABC classification, are characterized by a seasonality trend in the deand pattern (Fig. 1). The ipact of seasonal variations and unrepresentative data shifts, which can occur in the short tie periods, ay be reduced by plotting annual data. Therefore, plotting the IOC logistic paraeters on a onthly base ay provide better insights, as the nature of deviations fro the expected\planned perforance. The analysis reported below tried to gather all the suggestions here entioned. V. PRELIMINARY MODEL SETTING The product portfolio stocked by the selected copany, consists of 102 ites. The relative data base was first reviewed through an ABC cross-analysis, and secondly divided in three classes, while eploying the single ite turnover and the inventory stock as segentation variables. In this step, the indicators are both assued as onetary valued in Euros. Then, for the purpose of this work, a product belonging to the AA class was extracted. This ite, hereinafter referred as, is selected as representative of a product faily within the AA class. The onthly averaged profiles of turnover and inventory stock for the selected product are reported in Table II and Table III, for the peak season of sales and for the low one, respectively. This phase of the real case application protocol required the definition and setting of the input paraeters for the products. TABLE II: INVENTORY STOCK AND SALES FOR FAMA ITEMS (HS) Month February March April May June July August Average Turnover Inv_Stock 436, , , , , , , ,896 1,908,635 1,867,253 1,533,552 1,629,102 1,542,406 1,550,840 1,453,930 1,640,817 TABLE III: INVENTORY STOCK AND SALES FOR FAMA ITEMS (LS) Month Septeber October Noveber Deceber January February Average Turnover Inv_Stock 339, , , , , , ,372 1,813,440 1,908,862 1,884,580 1,791,520 1,905,551 1,908,635 1,868,765 The second colun in the Table IV reports the preliinary assuptions ade in order to initialize the desk-siulation for the peak seasonality data, while the third colun reports the values assued for the low seasonality period. In the next sessions, findings of IOC odel application are provided, in which the paraeter c is assued equal to 0.3. Notice that, this value has been assigned after trying any 406

4 Shipents/Sales housand Euros International Journal of e-education, e-business, e-manageent and e-learning, Vol. 2, No. 5, October 2012 different values through iterative testing, anything else being the sae. Peak seasonality an average stock level equal to ( )/2 = units. Such a value is lower than the average quantity which can be calculated through the application of the IOC odel on the year basis for the ites, without separate the evaluation aong different seasonality trends. Consequently, the first finding is that accounting for seasonality effects with respect to the analysis on the coplete period, ay lead to an inventory stock iniization approach, in order to draw soe cost efficiency issues. Month in the year Fig. 1. Deand pattern on a onthly base TABLE Ⅳ: INITIAL PARAMETER ASSUMPTIONS Model Paraeters (HS) (LS) L - ax 6,6 6,6 Q - ax D ax LTR IN D L - ax 0 0 Q + ax 0 0 D in OUT Fig. 2. SLOC curve for peak seasonality period (HS) Further setting values, such as Dax, D, Din, LTR, IN, were drawn fro data relating to deand, production and picking lot size, production lead tie. For products labeled as, the lead tie is 3 onths (i.e. 66 days); the production batch, which corresponds to the econoic lot size, is fixed in 50,000 units. Moreover, further assuptions are ade: Q-ax=2,500 units, because a negative deviation of 5% on the replenishent lot size is accepted for the volue as input to the inventory stock; L+ax=6.6 days, because a axiu delay in delivery date of 10% on replenishent lead tie is assued; OUT=1,000 units, which is calculated for the volue of finished products as output fro the inventory stock; L-ax e Q+ax are considered as negligible in the real case under investigation. VI. MODEL DESK SIMULATION The analysis of the current situation resulting fro the Inventory Syste truly adopted by the anufacturing organization, is provided in this session, which also reports the figures relating to SLOC functions for the product (see Fig. 2 and Fig. 3), as introduced whit the approxiate odel. The target or future state is fixed corresponding to a Service Level equal to 100%, because of the requisites of the downside channel. In this case, the IOC odel provides an average inventory stock equal to 395,000 units in the first period (HS), while in the second period the stock provided is 315,000 units (LS). If we consider a coplete period (i.e. the su of the two partial tie horizons for HS and LS), the IOC odel provides Fig. 3. SLOC curve for low seasonality period (LS) Secondly, in order to proper understand the output of iterative siulations, it is iportant to separate inventories according to their function in the syste. In particular, the ain difference between the two periods, as eerging fro the SLOC curve analysis, is on the coposition of the Safety Stock (SS). The value of the safety stock (SS) aintained in the syste is designed to copensate for deviations fro the ideal situation, due to: Delays in delivery date (L + ax); Negative variations in the quantity delivered fro the production to inventory stock (Q-ax) Variations in the downside deand (Dax) and replenishent lead tie (LTR). The safety stock, as usual, is defined in order to guarantee the expected perforance levels toward custoers, even if soe unexpected deviations occur fro the ideal condition (i.e. the paraeters above are not null in the approxiate odel). By assuing the three deviation effects to be statistically independent, the subsequent expression for the Safety Stock (SS) variable can be derived: = For instance, in the HS case, the fraction of the stock assigned to deal with delays in delivery date (SS1) ay be higher than in the LS case, and the IOC odel application can 407

5 support this intuitive assuption (see Table V). TABLE V: COMPONENTS ANALYSIS FOR THE SAFETY STOCK SS coponents SS1 (Delay in the delivery date) SS2 (Deviation in the quantity delivered) SS3 (Deviation in the deand for LTR) (HS) (LS) 29.01% 27.51% 0.51% 0.66% 70.48% 71.83% Therefore, in the LS period, the coponent related to deviation in deand (SS3) is increasing as percentage. This trend shows a further reduction potential of the individual additional quantity, through the adoption of appropriate delivery strategy in first case and by iproving forecast reliability in the second one. Moreover, by analyzing the coponents of the SS variable and their respective behavior, the IOC odel allows to identify all paraeters to be controlled, in order to liit the stock volue proliferation. VII. CONCLUSIONS The objective of this paper was to introduce a foral procedure based on the Inventory Operating Curves (IOC) odel, in order to discuss its potential application to a real Inventory Syste. Through the case study conduct, soe previous conclusions are given further credibility; specifically: that the IOC odel can lead to a reasonable representation of the logistic perforances, as alternative to the coon inventory control policies; that the optial c value included in the IOC equations can be adaptively set, aybe drawing upon any iterative desk-siulations; that the IOC odel can be a reasonable predictor of the stock reduction potential, which will perit taking advantage of cost savings and providing rapid liquidity increases. The ost iportant and original contribution of this study is on the 5-step analytical procedure, drawn fro the IOC theory. Even if the setting phase is based on the availability of a considerable aount of input data relating to the inventory control syste currently in use, the IOC odel can lead to a suitable and effective ethod for the inventory perforance evaluation. The real case application has soe points of originality as well. The product faily under exaination was affected by a seasonality trend. Thus, the IOC odel was iteratively applied to two saple of data in parallel, the first relating to the peak seasonality (HS) and the second relating to the low seasonality one (LS). The desk-siulations perfored, allow identifying soe typical proble in the Inventory Manageent area, which can be afforded through the IOC curves, such as the sizing of the Safety Stock held in inventory corresponding to a target Service Level (i.e. the SLOC curve). Although the procedure provided stes for a preliinary study, the analytical approach has been deonstrate as general as it ay be used with benefit in any other circustances. REFERENCES [1] R. J. Tersine, Principles of Inventory and Materials Manageent, Prentice Hall PTR, [2] S. Lutz, H. Loedding, and H. P. Wiendahl, Logistic-Oriented Inventory Analysis, Int. J. of Prod. Ecs., vol. 85, no. 2, pp , August [3] H. P. Wiendahl and P. Nyhuis, Perforance Measureent of the Procureent Process on the Basis of a New Operating Curve of Stock-on-Hand, Prod. Eng., vol. 4, no. 1, pp , January [4] P. Nyhuis, G. von Cieinski, and A. Fischer, Applying Siulation and Analytical Models for Logistic Perforance Prediction, Annals of CIRP Manufacturing Technology, vol. 54, no. 1, pp , January [5] J. F. Shapiro and S. N. Wagner, Strategic Inventory Optiization, J Business Logs, vol. 30, no. 2, pp. Autun [6] J. D. Steran, Business Dynaics, New York: McGraw-Hill Higher Education, Dr. Flora Bernardel holds a Master of Science degree (Electronics Engineering) fro the University of Padua in Italy and a PhD degree (Manageent and Engineering) fro the sae institution. In her doctoral project, she has defined and epirically validated a tool-kit for Supply Chain Risk Assessent. She currently works as Post Doctoral Research Fellow at the Manageent and Engineering Departent. She has published several scholarly articles and gave presentations on Supply Chain Risk issues on international acadeic conferences. She is a eber of the North Aerican Institute of Science and Inforation Technology (NAISIT) Prof. Roberto Panizzolo holds a Master of Science degree (Electronics Engineering) fro the University of Padua in Italy and a PhD degree in Industrial Engineering. He is currently the Director of the Postgraduate course in Lean Manufacturing of the School of Engineering of the University of Padua. Prof. Panizzolo has ore than 20 years of experience in the field of Operations Manageent and he is a senior consultant and qualified teacher and instructor for business organizations. He has worked on any international projects and his research has appeared in a nuber of books and international journals. He is a eber of the European Operations Manageent Association (EurOMA) and the Production and Operations Manageent Society (POMs). Dr. Davide Martinazzo has a BSc degree (Mechanical Engineering) fro the University of Padua in Italy, a MSc degree (Mechanical Engineering) and a PhD degree (Manageent and Engineering) fro the sae institution). He holds a Master in Business and Adinistration (MBA) fro the University of Iowa (US). He has three years of experience as consultant in the Operations and Manufacturing Manageent field, with a particular focus on probles of the Italian SMEs. He is currently eployed at Carel Industries S.p.A as Lean Change Agent. 408