2007/06/18 Wongwanida Kuhirunyarat (AE)
Managing demand uncertainty in supply chain planning
Anshuman Gupta, Costas D. Maranas * To highlight the comparison between deterministic model and two-stage stochastic model under demand uncertainty in the decision making problem of tactical planning of supply chain network. To focus on the significance of uncertainty for planning process in terms of risk management, customer satisfaction, market share, etc.
1. Deterministic production planning model: Allocation problem
The considered supply chain network: multi-product, multiple production site and multi-period. To aim at determining the optimal sourcing and allocation of limited resources to its manufacturing assets to satisfy the market demands. Each production site is characterized by one or more single stage semi-continuous processing units having limited capacity. The various products, which are grouped into product families, compete for the limited capacity of these processing units. The decision making process at the tactical level can be decomposed into two phases: 1. Manufacturing phase: allocation of the production capacity at the various production sites 2. Logistics phase: the post-production activities such as demand satisfaction and inventory management
-1-
2007/06/18 Wongwanida Kuhirunyarat (AE)
Midterm Deterministic Planning Model (MP)
Objective function: combined cost in Fixed Cost from any site Manufacturing phase Variable production cost Cost of raw material purchase Transport charges for intersite shipment of work in process
Transport charges for shipping the final product to customer Logistics Phase Inventory cost Safety stock violate penalty Penalties for shortage
-2-
2007/06/18 Wongwanida Kuhirunyarat (AE)
Manufacturing Constraints:
Amount of product = (rate of production)*(run length) Input-output relationship Products shipped to one site in one period will be consumed in the same period. The allocation of products to products families
Capacity restriction Upper and lower bounds for the family run lengths The amount available for supply in logistics phase following the manufacturing phase
Logistics Constraints: Subject to Inventory level links between Manufacturing and Logistics phase. Products are not supplied more than demand. The customer shortages are carried over time.
The violation of the safety stock levels
-3-
2007/06/18 Wongwanida Kuhirunyarat (AE)
2. Production planning with uncertain demand
The demand uncertainty is assumed as normal distribution. The manufacturing decisions are made before demand realization. Thus the manufacturing variables are considered as the first-stage, here-and-now decisions. On the other hands, the logistics decisions are postponed in the second-stage wait and-see mode to optimize in the face of uncertainty. Two-stage Stochastic Program with recourse(2SMP)
Model (2SMP), thus, aims at determining the optimal production settings in the supply chain that minimize the manufacturing cost, which are deterministic, and the expected logistics costs. The uncertainty in the product demand is translated into the uncertainty in the logistics decisions through the second stage inventory management problem.
Note The formulation of two-stage stochastic program with recourse is
-4-
2007/06/18 Wongwanida Kuhirunyarat (AE)
Supply Chain Network Planning Case Study
Six production sites, 30 products: Prod.1-10 are produced at site1 and 2. Prod.11-20 are produced at site3 and 4. These 20products are grouped into 10products families as follows: F1_ (1, 2); F2_ (3, 4); F3_
/ / /
(5, 6);. . .; F10_ (19, 20).
/
The demand for prod.1-20 exists at market1 and internally at site5 and 6. Prod.21-25 are produced at site5. Prod.26-30 are produced at site6. The demand for prod.21-30 exists at market2. The demand of all products is assumed to be normal distribution with a coefficient of variation 20%
Solution Comparisons between MP and 2SMP 1. Capacity Allocation
The overall capacity utilization for each site is also predicted differently by the two models. As shown in table1, the deterministic solutions consistently underestimate compared with stochastic solution.
-5-
2007/06/18 Wongwanida Kuhirunyarat (AE)
2. Standard deviation of cost distributions
The standard deviations for the two distributions are 77.5 (2SMP) and 103.4 units (MP) implying that a wider spread of values is obtained by neglecting uncertainty.
3. Probabilities of exceeding a particular cost level
The cumulative probability distributions can be used to determine the probability that the cost will exceed a specified level. For any given cost level, model MP results in a higher probability compared to model 2SMP. For instance, the probability that the total cost exceeds 1650 units is predicted to be as high as 21% by the MP model as compared to just 3% by the 2SMP model.
4. Worst-case cost
The worst-case costs obtained for the MP and 2SMP models are 1802 and 1689, respectively. Thus, the inclusion of demand uncertainty results in better risk management across the supply chain.
In conclusion, deterministic planning models, which do not recognize the uncertainty, can be expected to result in inferior planning decisions as compared to models that explicitly account for the uncertainty. Thus, underestimating uncertainty and its impact can lead to planning decisions that neither safeguard a company against the threats nor take advantage of the opportunities.
-6-
