nodes are retained while retuning the replenishment rule parameters through sim-ulation based optimization technique. The conflicting nodes are optimized by changing replenishment rules to Proportional-Integral policy. In all improvement stages, bullwhip is considered as a dominant constraint to facilitate equal advan-tage to all the nodes by dampening adverse effects caused to the network. The ultimate performance obtained from this framework is closer to the performance benchmark which is the optimum performance obtained using the similar type of replenishment rule in all the nodes of the network while respecting the bull-whip constraint. Industrial heuristics is restricted to proportional-integral policy, SOP1 and SOP2. The proposed framework has little implementation difficulty in achieving enhanced performance closer to the performance benchmark.
4.5 Problem Description
In this section, a decentralized distribution network with fixed architecture, con-nectivity, and location among the distribution nodes is considered for performance improvement. To be realistic, a multi-product multi-echelon decentralized supply chain studied by Perea-Lopez et al [26] (excluding plant details) will be used to illustrate the ideas. The distribution network (shown in figure 4.2) consists of ten retailers (i ∈ R1 to R10), four distribution centers (g ∈ DC1 to DC4) and man-ages nine different products with warehouse (W)-manufacturing facility (P) for each product. We seek to enhance the performance of this product multi-echelon distribution network by analyzing the network data followed by multistage optimization that is implemented in stages. This demand driven system is fully
4.5 Problem Description
Figure 4.2: Schematic representation of the Decentralized Distribution System
decentralized in which all distribution nodes belong to different companies. Each distribution unit prefers to adopt its own internal strategy to optimize local per-formance without considering the adverse bullwhip effects caused to the other parts of the network or the overall network performance. The internal strategy practiced by the distribution nodes of the existing network are given in Table: 4.1.
Explanations for the terms used under “Internal Strategy” and “Replenishment Policy” in Table: 4.1 will be provided in detail in the following section.
The internal strategies of the distribution nodes differ depending upon the indi-vidual decisions made to manage the inventory level at a constant target value or made responsive to the uncertain demand. Dejonckheere et al [56] and Lin et al
4.5 Problem Description Table 4.1: Internal Strategies of the Distribution Nodes
Distributor Node Internal Strategy Replenishment Policy R1, R2, R3 Responsive PI
R4,R5 Non-Responsive PI
R6, R7, R8 Non-Responsive Order-upto-policy
R9,R10 Responsive Order-upto-policy (OUP)
DC1 Responsive PI
DC2 Responsive Order-upto-policy
DC3 Responsive SOP1
DC4 Responsive SOP2
[18] described the responsive inventory target to manage inventory in accordance with the uncertain demand pattern to provide reliable customer satisfaction with less backorder and minimal excess inventory. The well-balanced relation between flow entities of the distribution node is described using information and material balance relations on a discrete-time basis by Lin et al [18]. The model equations of the distribution network are described elaborately in Chapter 3.
The flexibility in inventory position is obtained by setting desired inventory po-sition target SIP(t) in response to the forecasted demand (responsive strategy).
Such a policy is practiced by several retailer nodes and distribution centers in our example. In contrast, retailers R4 to R8 adopted constant inventory position as the desired target. The exponential forecaster with α = 0.111 was used in all dis-tribution nodes to forecast the downstream demand as suggested in the literature [18].
The rate at which downstream orders satisfied depends on availability of inventory level at-hand. We consider two cases of downstream order processing methods.
4.5 Problem Description
Case (1): supplier maintains high inventory (IHi,p) and is capable of satisfying all downstream customer orders (dj,p). This situation can be modeled by Equation 4.1. Case (2) supplier maintains limited inventory, and therefore equal proportion
(0 ≤ mi,p ≤ 1) of all downstream orders are satisfied with respect to the inventory at-hand (equation 4.2).
Yij,p = z−1X
dj,p when IHi,p ≥X
p
dj,p (4.1)
Yij,p = z−1X
mi,p× dj,p when IHi,p <X
p
dj,p (4.2)
4.5.1 Market Demand
As expressed in section 3.2 of chapter 3, we will consider two demand patterns namely stationary stochastic demand and nonstationary stochastic demand as represented by equations 3.12 and 3.13 respectively.
4.5.2 Performance Indicators
A performance indicator is a measurable entity that quantifies the performance of the supply chain. Choosing the right performance indicator or a combination of performance indicators depends on the characteristics of the supply chain system.
The performance measure chosen must reflect the behavior of the distribution node in relation to the business goals. Business goals may be customer focused,
4.5 Problem Description
company focused or a combination of the two [75]. In customer focused strat-egy, the downstream customer satisfaction is the dominant concern as compared to other objectives like supply chain cost (excess inventory and backorder). This kind of approach is often practiced for newly developed products so as to establish them in the market. Company focused (cost effective) strategy is practiced in supply chains facing insignificant competition, where cost minimization has more pronounced effect than order fill rate. This approach is suitable for well estab-lished products and for products having high depreciation and inventory holding cost. Tradeoff strategies may also be practiced in supply chains. We choose to minimize the distribution system cost such as excess inventory and backorder in our illustrative case study.
4.5.3 Performance index of the Distribution node and network
The performance index of the distribution unit ‘i’ for product ‘p’ (ψi,p) (equation 4.3) is represented as the weighted combination of excess inventory and backorder
with the bullwhip constraint. Minimizing the performance index is the ultimate goal to minimize the distribution system cost (or maximizing the revenue). The weight parameter (φi,p) depends on the relative importance of the performance indicators and the objective of the distribution network. The objective varies with the establishment of product at the market. For a newly developed product, im-portance is given (to minimize the backorder) to increase the sales by establishing the product at the expense of maintaining more inventory. For a well established product, importance is given to minimize the excess inventory at the expense of few backorders.