The value of production schedule integration in supply chains

INTRODUCTION

Intercompany collaboration has long been recognized as an opportunity for improving operational efficiency in supply chains. Through better coordinated business processes across organizational boundaries and increased sharing of more timely and accurate information, intercompany collaboration can help streamline information and material flows and cut wastes. In recent years, internet-enabled technologies have opened many new avenues for intercompany collaborations. It is exemplified by the fast adoption of various cutting-edge supply chain management software and solutions, such as Web-capable Enterprise Resource Planning systems and Advanced Planning Systems. Although the general belief is that inter-- company collaboration does improve supply chain performance, identifying the specific format of collaboration and quantifying its costs and benefits remain topics worthy of continual research.

Recent research has shed light on the benefits of intercompany collaboration. One proposition is to share real-time demand data collected at the point-of-sales with upstream suppliers (Lee, So, & Tang, 2000; Cachon & Fisher, 2000; and Raghunathan, 2001). Another notion is to use a centralized forecasting mechanism that accesses the final demand, develops forecasts, and shares them with all members in the supply chain (Chen, Drezner, Ryan, & Simchi-Levi, 2000). An alternative form of collaboration involves coordinating the price incentives that encourage buyers to place advanced order commitments with the appropriate level of buffer capacity at the supplier plants (Gilbert & Ballou, 1999; Donahue, 2000). In this paper, we explore yet another form of intercompany collaboration-the integration of production schedules. Departing from the setting of supply chains that involve retailers purchasing products from suppliers that is most common in the supply chain management literature, this paper specifically addresses the integration between the production schedule at a final product assembler (the buyer firm) and the production schedules at its components supplier firms. We propose a stochastic cost model to study the cost implications of schedule integration. We design numerical experiments to test different environmental parameters and examine what conditions are more conducive to the success of such integration.

Schedule integration will not be beneficial in all circumstances. We found that schedule integration can bring cost savings to the entire supply chain, but the savings come at the expense of the buyer firm. While in some situations the savings can be significant, there are also situations where it is better for all the firms to develop their production schedules independently. We found that the savings are more significant when (1) inventory holding cost is low, (2) supplier lead times are short, (3) forecasts are very poor, and (4) the suppliers' flexibility of accommodating schedule changes is low and forecasts are poor.

INTEGRATED PRODUCTION SCHEDULING POLICIES

In this paper, we explore the value of integrated production schedules for reducing the negative effects of schedule revisions in a supply chain. To reduce the impact of schedule changes, a customer (or buyer) can share forecasts with suppliers and make firm material commitments for a specified time horizon so as to reduce the uncertainty associated with supplying materials. Integrating production schedules in a supply chain involves commitments that take the form of purchasing and scheduling policies specified by the buyer, which take into account the operational flexibility of the supply chain members, and sharing them with suppliers. It also requires suppliers to coordinate their production and delivery schedules with those of the buyer.

Purchasing Policy

The purchasing policy specifies the nature of the material commitments to the suppliers in the form of a materials contract. While the buyer provides forecasts to the suppliers, the actual quantities purchased can deviate from the forecasts. Often, the forecasts are future production schedules that are changed as the delivery date approaches. In practice, purchasing contracts under these conditions are known as quantity flexible contracts (Tsay, Nahmias, & Agrawal, 1999) and typically are negotiated for a given price with limits placed on the amount of deviation from the forecasts. These contracts provide for the purchase of the material requirements of a certain item for a specified time interval, given production and delivery schedules. However, we are concerned about the "hidden costs" of schedule changes that exist in the supply chain. Normally, the suppliers absorb these costs in the short term, but they will ultimately be reflected in future purchasing costs for the buyer.

With integrated production scheduling, buyers can mitigate the possibility of higher prices by providing a schedule of requirements that are fixed for a certain time interval. The longer the time interval, the more certainty there is for the suppliers, and the better the future negotiated prices to the buyer. However, a policy of longer fixed commitments to suppliers means that the buyer's company must use buffer inventories to minimize the effects of uncertain demands from its customers. We define F as the fixed time interval covering committed material requirements to suppliers consisting of announced, but varying, order quantities per time period. If the integration of production schedules is to be effective, once the buyer commits to the fixed schedule of requirements, the timing and size of the order quantities cannot be changed.

Scheduling Policy

The scheduling policy specifies the planned revision interval of the production schedule at the buyer's assembly plant. We define R as the number of time periods between successive updates of the assembly schedule. The scheduling policy of the buyer is of interest to the supplier because it determines the frequency at which the buyer's production requirements might change. Frequent schedule revisions enable the buyer's plant to be more responsive to changes in its environment, but require more flexibility in the supply chain because the rates of material usage are changing. Infrequent schedule revisions foster stability in manufacturing planning and execution at the suppliers, but may be unresponsive to changes in demands at the buyer's plant.

An obvious approach to reduce the costs of schedule revisions at the buyer's plant is to freeze the master production schedule, a practice commonly used in manufacturing industries. A considerable literature exists on freezing master production schedules and choosing schedule revision intervals (Kern & Wei, 1996; Sridharan & LaForge, 1994; Lin, Krajewski, Leong, & Benton, 1994; Lin & Krajewski, 1992; Zhao & Lee, 1993; Sridharan & Berry, 1990a, 1990b; Sridharan, Berry, & Udayabhanu, 1988, 1987). Manufacturing firms commonly face these freezing and rescheduling decisions (Hahn, Watts, & Kim, 1991; Monden, 1993; Hall, 1982). However, that literature is directed at the single firm rather than the supply chain. In general, the literature has not addressed the problem of managing schedule revisions in a supply chain.

Flexibility Capability in Supply Chains

There are many reasons for schedule revisions in a supply chain, and the impact on suppliers is costly (Inman & Gonsalvez, 1997). A survey of automotive suppliers indicated that 76% agreed that there are cost penalties associated with expediting parts, and 62% said that the cost could best be reduced by consistent forecasts from their customers (Plumb, 1992). Ideally, a supply chain should respond to changes in the timing and volume of requirements immediately without loss of quality, delivery dependability, or efficiency. However, cost and capacity limitations in various members of a supply chain often preclude such a capability. The purchasing and scheduling policies associated with production schedule integration define the stability of the component and material requirements passed upstream in the supply chain. To be cost effective, these policies should reflect the flexibilities of the suppliers to make schedule changes (Wei & Krajewski, 2000).

It is important to clarify the nature of the flexibility we will address in this study. First, the need for the flexibility arises from volume changes to production schedules induced by forecast revisions on the part of the buyer. These schedule revisions are passed to suppliers upstream through the buyer's authorized delivery schedules. Second, the reaction time interval for the suppliers to meet schedule changes is very short, often measured in hours or days. We call this time interval the advanced warning. Finally, production schedule integration requires that we become specific as to products, components, and processes. For example, we identify a specific product, the process that produces the product, the components required for that product, and the processes at the suppliers that produce the components. Consequently, our interest is in flexibility at the process level as opposed to the plant level.

In this paper we define the flexibility capability of a supply chain member as the ability of its process (for the specific product or a component produced by the member) to efficiently respond to schedule revisions within the advanced warning. A number of dimensions of flexibility can be associated with the flexibility capability of a process (D'Souza & Williams, 2000; Koste & Malhotra, 1999; and Vokurka & O'Leary-Kelly, 2000). The flexibility dimensions most related to the focus of this study include expansion, routing, machine, labor, mix, material handling, and the transition penalties of going from one state to another in a given dimension within the advanced warning time period. The transition penalties could include the time and cost of changeovers and disrupted schedules, the scheduling efforts (managerial time) required to affect the transition, short-term capacity adjustments, or the scrap and rework generated because of the transition. The ramifications of the various dimensions of flexibility for a given process manifest themselves in its lead time and the cost to make schedule changes on the process within the advanced warning, which we use to measure flexibility capability. In this regard, a firm's process for producing a component or product is more flexible if it has a short lead time or it can efficiently change volumes in its production schedules with a short advanced warning. This definition of flexibility capability adheres to Upton's (1994) definition of "operational flexibility."

To further clarify the concept of flexibility capability, consider Figure 1, which is the process flow map for a metal screw produced by one of 10 plants in a recent field study on schedule changes (Krajewski, Wei, & Tang, 2001). The manufacturing lead time is eight weeks and the bottleneck operation is the 5/8" Header, which can take up to 16 hours to set up. The flexibility of changing production schedules on short notice is very low; it is most efficient to produce large lots to stock. Realizing that this supplier has low flexibility capability, the customer freezes eight weeks of purchase requirements for the metal screw and typically provides four weeks of advanced warning for schedule revisions.

In this paper we develop a stochastic cost model that will enable us to analyze the value of schedule integration in a supply chain. While founded in our empirical work on the implications of schedule changes in supply chains, the model is intended to provide motivation for further empirical and analytical research by identifying fruitful avenues for exploration. We perform experimental analysis with the model to determine some conditions under which production schedule integration would have value. The design of the experiments was motivated by our study of two supply chains involving 10 plants. We conclude the paper with suggestions for future research in the integration of production schedules in a supply chain.

SUPPLY CHAIN SCENARIO

We consider a supply chain consisting of a customer (which could be a group of consumers placing a stream of demands for a single product, or a central warehouse that stores and distributes a single product), an assembly plant (herein referred to as the Assembler) producing a functional product (Fisher, 1997), and a network of tier 1 suppliers to the assembly plant. We focus on the Assembler as the interface between the customer demands for the product and the suppliers of materials and components needed in the assembly of that product. The Assembler engages in repetitive manufacturing, which we define to be the production of discrete units in a high-volume concentration of available capacity using fixed routings (Spencer & Cox, 1995).

Environment

Daily customer demands for the product, d^sub t^, follow a probability distribution with a mean of d(overscored) and a standard deviation of (sigma)^sub d^, with no seasonality or trends. See Table 1 for a complete description of the variables and parameters used in this paper. These demands could represent a collection of independent demands from numerous customers of a final product or an aggregation of demands representing the production schedules of industrial customers. The Assembler is unaware of the universe parameters governing the demand distribution and, consequently, must use standard statistical forecasting methods to estimate future demands. Forecasts of demands become more precise as the day for which the forecast is made approaches because of managerial judgment and other available information.

The Assembler maintains a distribution inventory that consists of a cycle inventory and a safety stock. The safety stock provides a given level of customer service. No backorders are allowed. The cycle inventory is minimal because the Assembler moves finished product to the inventory daily and makes daily shipments to its customers. Consequently, the Assembler plans production to meet forecasted demands and to maintain its safety stock, adjusting assembly rates for production overages or shortages.

The Assembler provides assembly schedules to its suppliers, freezing them for the next F days, where F >= R. The assembly schedule is updated every R days. Schedule changes are driven by errors in the Assembler's forecasts of demand. These changes are transmitted to the suppliers every R days. The suppliers are committed to accommodating schedule changes when the Assembler needs them. Suppliers incur a cost for schedule changes, which is a function of the advanced warning provided by the Assembler.

Production Decisions

Production decisions at the Assembler are executed with a rolling horizon. Figure 2 shows the relevant time intervals for two planning cycles. Planning Cycle 1 shows the fixed material commitment interval F. With integrated production schedules, the material commitment corresponds to the Assembler's production schedule for that interval; it is passed along to the suppliers for planning purposes. In planning Cycle 2, we do not revisit the assembly schedule until day T^sub 2^, when we can adjust the assembly production rate for the time interval [T^sub 3^,T^sub 4^], the newly authorized assembly schedule. In this context, "authorization" means the inclusion of the new schedule in the firm material commitments to the suppliers. Since the revision interval equals R days, the interval [T^sub 3^,T^sub 4^] also equals R days. At T^sub 2^, the newly authorized assembly schedule is passed along to the suppliers as an update to the material commitments already in place. In this dynamic planning environment, it will always be the case that F >= R.

To apply integrated production scheduling in a supply chain, the Assembler must choose an (F,R) couplet. The (F,R) decisions are considered fixed for the foreseeable future. The following repetitive activities take place:

Every day

* The assembly plant produces r units, places them in the distribution inventory, and ships d^sub t^ units to its customers.

* The suppliers ship materials sufficient to cover production of r units at the assembly plant. IMAGE FORMULA 32IMAGE FORMULA 33IMAGE FORMULA 34IMAGE FORMULA 35IMAGE FORMULA 36IMAGE FORMULA 37IMAGE FORMULA 38IMAGE FORMULA 39IMAGE FORMULA 40IMAGE FORMULA 41IMAGE FORMULA 42IMAGE FORMULA 43IMAGE FORMULA 44IMAGE FORMULA 45IMAGE FORMULA 46IMAGE FORMULA 47IMAGE FORMULA 48IMAGE FORMULA 49IMAGE FORMULA 50IMAGE FORMULA 51IMAGE FORMULA 52IMAGE FORMULA 53IMAGE FORMULA 54IMAGE FORMULA 55