سهمیه بندی موجودی و به اشتراک گذاری در توزیع قبل از فروش با فن آوری ارتباطات تلفن همراه
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|20557||2009||17 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : International Journal of Production Economics, Volume 121, Issue 2, October 2009, Pages 584–600
In pre-sell distribution, the uncertain customer demands are revealed by the company's sales representatives who visit the customers and arrange delivery quantities on the spot, prior to physical execution of deliveries. Given a periodic base-stock of a distributed product, we consider allocation of the product to the customers in two different settings: with and without utilization of mobile communication technologies. There are two performance measures considered: the customer-average fill rate, and the sales profit under service level constraints. The mobile setting is shown to enable a generally better system-wide performance, featuring the capability of inventory pooling. To observe the magnitude of this advantage we determine the optimal allocation policies by means of stochastic dynamic programming. Computational examples for selected configurations and demand distributions are presented.
Distribution of goods is realized in many industries by means of field representatives who travel through assigned territories, discover the customer demand, and execute sales. A predominant role in such distribution practice has been traditionally played by the so called route-sell method, which assumes that a number of vehicles loaded at the depot are sent on tour to meet uncertain customer demands: each vehicle carries the goods on board and distributes (sells) them to the geographically dispersed customers as it follows its route. This distribution mode has been extensively studied in the past decades by the research community in the context of the vehicle routing problem and its stochastic extensions ( Baldacci et al., 2004, Bertsimas and Simchi-Levi, 1996, Golden and Assad, 1988 and Laporte et al., 2002). The presented research deals with an alternative distribution method designated commonly as pre-sell. In pre-sell distribution the sales are arranged with the customers by salespersons in advance, prior to the physical execution of deliveries. These advance salesmen are typically travelling through assigned territories on a regular basis and visit pre-determined customers—like manufacturers, wholesalers and retailers—in order to assess their requirements and “to sell any quantity and every item that might be appropriate” ( Makadok, 1993, Anily and Federgruen, 1990, Golden and Wasil, 1987 and U.S. Bureau of Labor Statistics, USA, 2006). They are often assigned to promote products to the customers, check stock on-site, negotiate the sales, and estimate delivery dates ( California Employment Development Department, USA, 1995 and U.S. Bureau of Labor Statistics, USA, 2006). Comparing to route-sales, the advance fixing of delivery quantities provides pre-sell with several potential advantages (Makadok, 1993 and Puric and Schreib, 2002): (a) delivery trucks are stocked to accurately match the orders; (b) no stock-outs will occur on the trucks; (c) the trucks carry no excess stock, saving fuel costs; (d) they return empty, therefore no product unloading and handling back at the depot, what saves labor costs and reduces breakage; (e) knowing zero demands in advance means less stops; (f) the routes can be scheduled with more certainty and customers’ time windows better met; (g) fewer trucks might be needed. The well-known industries that employ pre-sell are, e.g. the beverage, soft drink, and consumer goods. Though the route-sell mode had previously been dominating over pre-sell (Anily and Federgruen, 1990 and Golden and Wasil, 1987), the evidence of the recent years shows pre-sell to win more popularity than before (Goldberg, 2003), and not least due to novel opportunities enabled by mobile communication technologies. They deliver new attractive capabilities: remote access to the CRM and ERP systems; taking customer orders on-site and transmitting them immediately to the CRM or ERP back-end; tracking roaming employees and assets; communicating customer orders, instructions and data to the right employees in the field (Makadok, 2003). These capabilities result in several new advantages for the enterprise (MEI Computer Technology Group Inc., 2007, Puric and Schreib, 2002 and TechRepublic, Inc., 2006): (a) a greater speed and accuracy of data collection and transmission; (b) more efficient resource allocation by utilizing real-time data from the field; (c) performance improvements due to a better sourcing of mobile workers with the up-to-date corporate data; (d) a better job assignment. Convinced by ever growing adoption of mobile solutions (Krebs, 2004) and their potential impact on pre-sell distribution, we take in this paper a closer look at the above advantages (b) and (c). For that we consider a company which pre-sells a single good to a number of customers. We show that mobile technologies may act as a means of inventory pooling in pre-sell. Besides that, we let the company use an inventory rationing policy for matching its limited stock against customer demands. Both pooling and rationing represent important dimensions in inventory management. The concept of inventory pooling has received a remarkable attention in the literature since the publication of the work by Eppen (1979), who has shown how consolidation of inventories and aggregation of stochastic demands can reduce the expected holding and penalty costs. Since then many authors have studied pooling strategies in various settings. Alfaro and Corbett (2003) give a recent overview for uncapacitated inventory systems. As they point out, much of the literature on such systems assume normally distributed demands. They make then a detailed analysis of the impact of demand correlation on the value of pooling in an uncapacitated setting with a computational study for some selected non-normal demands. Corbett and Rajaram (2006) generalize Eppen's (1979) results to non-normal dependent demands. Research on pooling in capacitated systems has been recently advanced by Benjaafar et al. (2005). Typically the analysis of such systems is based on considering Poisson arrivals of unit demands. Whereas pooling is likely to be a strategic decision, inventory rationing deals with the following, rather operational, problem ( de Véricourt et al., 2002). If one distinguishes between customer demands of different priority, then one must decide how to allocate the stock to the incoming demands when it runs low: it can sometimes be more rational to stop filling the demands of low priority classes in order to save the stock for meeting possible demands of higher priority—i.e. to ration the inventory. Much of the studies focus namely on determining stationary inventory levels r1,…,rnr1,…,rn such that filling the iith class demand stops when the on-hand inventory drops to or below riri. Recent reviews of rationing research can be found in Deshpande et al. (2003) and Arslan et al. (2005). There also exists research that considers both dimensions of pooling and rationing together. However, this body of research is essentially smaller than that on pooling or rationing only. de Véricourt et al. (2002) consider rationing in a production–inventory system with backorders and characterize the optimal policy. They then enable inventory pooling in a system with two demand streams and show that ignoring the stock rationing dimension can lead to wrong managerial decisions. Deshpande et al. (2003) run numeric tests to figure out how well their rationing policy does perform in a pooled system with two demand streams compared to a non-pooled system, as well as to a pooled system without rationing. Zhao et al. (2005) deal with a decentralized system where independent stocking locations can, besides procuring the items from the manufacturer, also share stock with each other as well as ration it against each other's requests. Our setting possesses some similarities as well as some distinctions from the existing approaches. Similar to Topkis (1968) and Frank et al. (2003), we consider an uncapacitated periodic-review base-stock system. Uncertain demands are filled at predetermined customer locations sequentially once in a review period. Similar to the assumption made in Frank et al. (2003), our policy allows partial filling of uncertain customer demands; the distinction is that we assume all demands are uncertain. The rationing policy that we employ is non-stationary and of a similar nature as in Topkis (1968). Utilizing mobile communication technologies, we pool separate stocks and fill customer demands with a rationing policy. We confirm the advantage of pooling in our setting and express it as a function of the base-stock level by means of stochastic dynamic programming. As opposed to most of the studies on pooling or rationing or both, we do not stick to any particular form of demand distribution. The computational method requires the demands to be represented by a finite distribution though, what may assume a discretization of continuously distributed demands. The rest of the paper is organized as follows. Sections 1.1 and 1.2 introduce a company which pre-sells a single good to a number of customers in two different environments: with and without utilization of mobile communication technologies. In Section 2 we model the company's distribution operations using a stochastic programming framework. Section 3 presents the computational method. In Section 4 we observe the advantage of the mobile setting in a number of numeric examples. In Section 5 we optimize the base-stock policies and make some further comparisons. Section 6 concludes the paper.
نتیجه گیری انگلیسی
We have considered a periodic-review base-stock system where a stock rationing strategy was used for filling customer demands. The mobile communication technologies were utilized in this system to implement inventory pooling. We have shown how to obtain optimal allocation policies and express the resulting performance for pooled and non-pooled systems. This gave us a possibility to capture the advantage of utilizing mobile facilities as a function of the base-stock level explicitly. We could observe in some numeric examples that utilizing mobile facilities is most beneficial under moderate levels of product supply. We then determined optimality conditions for the base-stock levels and have seen on some examples, how do pooled systems perform compared to their non-pooled counterparts, if their base-stock levels are set up optimally. The presented model allows to define different customer priorities by choosing customer–individual service levels and sales profit rates without increasing the complexity of the model. Another feature of the presented approach is its capability to efficiently handle every finite probability distribution of customer demands with an option of obtaining approximate solutions in the case of continuous and infinite distributions, too. There are several directions in which this research can be extended. We emphasize the following ones: studying the impact of demand correlation on the optimal performance; studying how much pooling and rationing each contribute to the performance improvement; incorporating the strategic behavior of customers and agents into the framework.