AN INTELLIGENT APPROACH FOR MANAGING MATERIALS PURCHASING IN THE RMC INDUSTRY

Transcription

1 AN INTLLIGNT APPROACH FOR MANAGING MATRIALS PURCHASING IN TH RMC INDUSTRY Min Wu Dept. of Real state and Construction, University of Hong Kong, Pokfula Road, Hong Kong Steve Rowlinson Dept. of Real state and Construction, University of Hong Kong, Pokfula Road, Hong Kong ABSTRACT A odel for coparing the inventory costs of purchasing under the econoic order quantity (OQ) syste and the just-intie (IT) order purchasing syste in existing literature concluded that IT purchasing was virtually always the preferable inventory ordering syste especially at high level of annual deand. By expanding the classical OQ odel, this paper shows that it is possible for the OQ syste to be ore cost effective than the IT syste once the inventory deand approaches the OQ-IT cost indifference point. The intelligent agent paradig, a natural fit to solve dynaic aterials control proble, however has seldo been deployed in construction aterial anageent. Based on the OQ-IT cost indifference point function, an alternative agent based inventory anageent syste is thus developed. A case study deonstrates how the equation affects the purchasing approach of a aterial. KYWORDS OQ, IT, Multi-agent syste, control 1. BACKGROUND The intelligent agent paradig is a natural fit to certain classes of dynaic aterials control probles because the paradig focuses on coordinating the activities of loosely coupled distributed entities, e.g., various raw aterials, aterials suppliers, shippers, anufacturers, distribution centers, and retails [1]. One goal of the paradig is to enable agents to decide when and how uch a raw aterial should be ordered, based on the resources constraints. The existing agent based aterial control systes are heavily skewed towards just-in-tie (IT) purchasing []. This is ainly driven by the success achieved by those IT copanies. By developing a series of atheatical odels that directly coparing the cost difference between the OQ and the IT purchasing systes, two researchers, i.e. Schniederjans and Cao [3] concluded that in situations where plants adopting the IT operations could capitalize 515

2 upon physical plant space square eter reduction, a IT syste would virtually always be preferable to an OQ syste. However, IT purchasing is not always successful even if the plants adopting IT operations can capitalize upon physical plant space reduction [4]. In addition, these two researchers [3] had difficulties to either scientifically or epirically ascertain the capability of an inventory facility to hold the OQ-IT cost indifference point s aount of inventory. Their odels cannot clearly explain the success achieved by the OQ copanies. Hence this study has two objectives. The first objective is to revise Schniederjans and Cao s OQ-IT cost indifference point function, and to develop a new OQ-IT cost indifference point function. The new function will deonstrate that it is still possible for an OQ syste to be ore cost effective than a IT syste, even if IT purchasing ay capitalize upon physical plant space reduction. The second objective is, based on the new function, to develop a new agent based aterial control syste. The odels of Schniederjans and Cao [3] were developed fro the classical OQ odel. Hence, the classical OQ odel should be revisited.. CLASSICAL OQ MODL Harris s [5] OQ odel, naely, the classical OQ odel, ais to iniize the total of ordering and holding costs, while assuing that soe inventory operating costs such as rental, utilities and personnel salary, etc are fixed costs. The total annual cost of the classical OQ syste, TC, is the su of the inventory ordering cost, inventory holding cost, and the cost of the actual purchased units, or: TC Qh = kd+ Q + P D (1) where Q is the fixed order quantity, h is the annual cost of holding one unit of inventory in stock, k is the cost of placing an order, D is the annual deand for the ite, P is the purchase price per unit, D/Q is the annual ordering frequency, Q/ is the annual average inventory level in the inventory facility. It should be noted that although the ter the total annual cost of an inventory ite under an OQ syste is widely used to refer to TC in q. 1, TC is not the actual total annual cost of an inventory ite under an OQ syste. The actual total annual cost of an inventory ite under an OQ syste should be the su of TC and the fixed costs. Fixed costs, including rental, utilities, and personnel salary were excluded fro the inventory holding cost ite in q. 1. This was also an iportant assuption ade by Schniederjans and Cao [3] when they derived their OQ-IT cost indifference points. However, since (a) It is agreed that the so called fixed costs were left out fro the so called total annual cost of the OQ syste, and (b) Gaither [6] suggested that the annual inventory holding cost should include the opportunity cost of the working capital tied up in purchased goods, taxes and insurance paid on inventory ites, inventory spoilage cost and inventory obsolescence cost, together with the cost of physical storage, and (c) Wantuck [7] proved that the so called fixed costs would no longer be constant during IT operations, and (d) Schniederjans and Cao [, 3] observed that the saved inventory facilities can be rented out when the annual average inventory level dropped, then there is a reason to include all coponents of inventory holding costs into the holding cost ite, when coparing an OQ syste with a IT syste. To su up, one of the assuptions of the classical OQ odel, naely, the so called fixed costs are excluded fro the holding cost ite, needs to be revised. Consequently the traditional OQ odel needs to be expanded when coparing an OQ purchasing syste with a IT purchasing syste. To include the so called fixed cost into the holding cost ite, this study assues that the inventory physical storage costs under the OQ syste, for exaple, rental, utilities and personnel salaries are linearly related to the average inventory level. However, it should be noted that the inventory physical storage cost under the OQ syste is not necessarily a linear function with the average inventory level. For instance, let us put ourselves in the position of a warehouse an storing cubic boxes. All the boxes are identical. The area of each surface of each box is. Let us look at a siple scenario of 3 boxes 516

3 and then extend it to a ore coplicated scenario of n boxes. For 3 boxes, the total inventory space occupied ay be, or 3. This is shown in Figure 1. For n boxes, the total inventory space occupied ay be,, n 1, 3 ( ) or n. This is shown in Figure. Hence, the total inventory space taken up by n boxes does not have to be n, which is the logical extension of the assuption ade in this study. Figures shows that this assuption is biased against the OQ odel, as the n boxes ay be stored in an inventory space with an area that is saller than n. 3 Figure 1. Inventory space occupied by 3 cubic boxes n n-1 n n Figure. Inventory space occupied by n cubic boxes 517

4 In addition, when average inventory level increases, a larger and ore copact inventory facility is usually used to save physical plant space. For instance, in places where real estate is costly, it ay be prohibitively expensive if not ipossible to acquire a large aount of space. This is pertinent to RMC facilities as they necessarily have to be located within travel distance fro urban construction sites. In those cases, a larger and ore copacted inventory facility is usually used to save physical plant space. It is also iportant to note that the inventory physical storage costs per unt in a larger sized inventory facilty are usually lower than that in the saller sized inventory facility, provided that the other conditions, such as rental rate, utility rate and labor rate reain unchanged. The production anager interveiewed by the researchers in a RMC batching plant suggested that if the inventory physical storage costs per unit in the larger sized inventory facility were even higher than that in the saller inventory facility, then there was no reason to have a larger inventory facility. The copany can siply install a few saller sized inventory facilities together to hold the sae aount of inventoy in the RMC industry. Hence, the assuption ade by the authors tended to be quite conservative for the OQ syste when copared with a IT purchasing syste, provided that the annual cost of holding one unit of inventory in stock is deterined based on the average value of inventory facilities. This study shows that it is still possible for the OQ syste to be ore cost effective than the IT purchasing syste even when a) the IT operation can experience physical plant space reduction; and b) an unfavorable assuption is ade against the OQ syste. 3. RVISD OQ-IT COST INDIFFRNC POINT The total cost under the revised OQ odel is thus: TC HQ = kd+ Q r + P D () TC is the su of the inventory ordering cost, the r expanded inventory holding cost, and the cost of the actual purchased units. TC is the actual total cost r of the OQ ordering syste and is greater than TC in q. 1, as H is significantly greater than r h in q. 1. The optiu order quantity of the revised OQ odel, Q, derived fro q. is: r Q kd r = (3) H q. 4 results in a total annual optial cost under the OQ purchasing approach of: TC = kdh P D (4) r + Under the IT syste, the ordering cost and holding cost, including the so called fix costs are ainly transferred to the supplier. The total annual cost under the IT syste, TC, thus, is the annual purchase cost [3], given by: TC = P D (5) where P is the unit price under the IT syste. P is greater than P. This is to partially reflect the holding costs and ordering costs that have been transferred to the aterials suppliers [3]. The cost difference between an OQ purchasing syste and a IT purchasing syste, Z r thus is: Z r = kdh + P D P D (6) Setting Z r equal to zero, the root of q. 7 is the revised OQ-IT cost indifference point, D : D ( P ) P = kh (7) 4. A NW AGNT BASD MATRIAL CONTROL SYSTM In this study, an agent syste, which based on the newly developed OQ-IT cost indifference point function, is developed to decide whether a aterial should be ordered by using OQ approach or IT approach. Fig. 3, Fig. 4 shows the scenario of the proposed ulti-agent syste for aterial control. The syste proposed in Fig. 3 consists of three agents, naely aterial agent, logistics anager agent, and production anager. The aterial agent records the historical data of each of the aterial 518

5 regarding its ordering cost, holding cost, ordering frequency, its supplier inforation etc. The logistics anager agent, based on the OQ-IT cost indifference point, calculates the OQ-IT cost indifference point and decides whether a aterial should be purchased via an OQ or IT Figure 3 Scenario of the odel 5. A CAS STUDY Figure 4. RFID platfor This case study deonstrates how the OQ-IT cost indifference point function ay affect the purchasing approach of a aterial. Ready ixed concrete (RMC) is a product that is widely used in the construction of building and civil works in the construction industry. The production of RMC is a highly repetitive anufacturing process [3]. Ceent, one of the raw aterials of RMC ay be purchased either using the OQ syste or the IT syste [4]. In Singapore, ost of the ceent consued was iported ainly fro apan by 40,000- ton ceent carriers. The cost of placing an order for 40,000 ceent carrier fro apan was k = S$ 43, 000 / order for transportation alone. The annual cost of holding one ton of ceent was H = S$ 344 /year per ton. Purchasing ceent according to the OQ odel costs P = S$ 40 /ton. If ceent was purchased under a IT syste, the cost was P = S$ 65 /ton. The data for this case study were collected by interviewing the overseas investent anager, the financial anager, the production anager and the custoer service supervisor of the ceent division of a supplier. Based on q. 3, the econoic order quantity was Qr = 36,139 ton / order, which was close to the routine order quantity 40,000 ton / order. Hence, q. 7 can be used to derive the OQ-IT cost indifference point. According to q. 7, D the OQ-IT cost indifference point was 475,545 ton. A survey revealed that the RMC suppliers in Singapore, whose annual ceent deand were greater than D purchased their ceent in an OQ fashion [4]. 519

6 6. CONCLUSIONS The existing agent based inventory anageent approaches are heavily skewed toward just-in-tie (IT) purchasing. This paper develops an OQ-IT cost indifference point function, and shows that it is possible for the OQ syste to be ore cost effective than the IT syste. Based on the function, an alternative agent based inventory anageent syste is developed. A case study deonstrates how the equation affects the purchasing approach of a aterial. RFRNC [1] Wagner T., Guralnik V., and Phelps., proceedings of Software agents: nabling dynaic supply chain anageent for a build to order product line, Agents in Business Autoation, pp.1-9, 00. [] Yung S.K., Yang C.C., Lau A.S.M., and Yen., Applying ulti-agent technology to supply chain anageent, ournal of lectronic Coerce Research, 1(4), pp , 000. [3] Schniederjans M.. and Cao Q., An alternative analysis of inventory costs of IT and OQ purchasing, International ournal of Physical Distribution & Logistics Manageent, 31 (), pp , 001. [4] Wu M., Modeling IT purchasing in the ready ixed concrete industry. Unpublished PhD thesis, the National University of Singapore, 004. [5] Harris F.W., Operations and Cost Factory Manageent Series, A.W. Shaw Co, Chicago, [6] Gaither N., Production and Operations Manageent, Duxbury Press, Belont, CA [7] Wantuck K.A., ust-in-tie for Aerica: a Coon Sense Production Strategy, The Foru Ltd., Milwaukee, WI,