Sunteți pe pagina 1din 10

Capacity Investments in Supply Chains: Sharing the Gain Rather Than Sharing the Pain Brian Tomlin Manufacturing

& Service Operations Management Vol. 5, No. 4, Fall 2003, pp. 317333 The supply chain risk investigated in this paper is demand uncertainty. The problem setting is one supplier, one manufacturer, one time period. His modeling has similarity to the newsvendor problem. The supplier and the manufacturer decides on capacity at the beginning of the period, the manufacturer orders from supplier and produces after observing the demand. There is no extra cost of lost sales, only loss of possible profit. The article is interested in the capacity of the supplier and trying to use price contract to coordinate the supply chain. The supplier's capacity is restricting because the supplier would not install capacity more than the manufacturer's capacity, as there is zero probability of the supplier selling to manufacturer more than the manufacturer's capacity. The result is, the manufacturer's optimal contract is a quantity-premium price only schedule, that is, the average wholesale price per unit increases in the order size. This is opposite of trying to get discount at ordering larger amounts. The manufacturer promises higher margin for larger order so that the supplier would install more capacity. The higher average payment is dubbed as sharing the gain. Simple piecewise linear (one or two breakpoints) quantity-premium schedules are shown to be highly efficient. There is an extension for comparing firm-commitment (if order is less than a level, the difference is compensated) to single-break-point quantity-premium contracts. If wholesale price is higher than a limit than quantity-premium is always better for higher capacity, if price is lower than the desired capacity plays a role. If higher capacity is desired, firm-commitment is better, there is a threshold value of capacity.

Process Flexibility in Supply Chains Stephen C. Graves Brian T. Tomlin Management Science Vol. 49, No. 7, July 2003, pp. 907919 The supply chain risk investigated in this paper is demand uncertainty. There is a multistage, multiproduct supply-chain. The problem is to satisfy the uncertain demand with available capacity. There is no failure, the capacity of nodes in the supply chain are fixed and known. Nodes in the supply chain can have flexibility in meeting the demand for more than one of the end products. They want to compare different supply chain designs with different production flexibilities in terms of how much of the uncertain demand of end products can be met with the given design. They define inefficiencies for such flexible supply-chains as floating bottlenecks and stage-spanning bottlenecks. These bottlenecks cause unused capacity and unmet demand. They develop a metric g and show in simulations that higher value of g protect against inefficiencies.

On the Value of Mix Flexibility and Dual Sourcing in Unreliable Newsvendor Networks Brian Tomlin, Yimin Wang Manufacturing & Service Operations Management Vol. 7, No. 1, Winter 2005, pp. 3757 There is one company, one or many resources, many (two) products and one period. The problem is to meet the unknown demand with resources that has some probability of failure. The company invests in resource capacity at the beginning of the period. Production planning is made after demand and resource failures are observed. The result is flexible networks are good for demand uncertainty but dedicated resources are better if there is a higher failure rate or the company is risk averse. Four configurations are compared: 1) SD: single-source dedicated network, two products each are produced at separate source. 2) SF: single-source flexible network, two products both produced in the same source 3) DD: dual-source dedicated network. Each product has two facilities 4) DF dual source flexible network. Each product has a dedicated facility and there is third flexible facility that can produce both products. Flexibility is good for meeting uncertain demand, dual sourcing is good for meeting demand when there is source failure probability. The firms investment problem is formulated as a two-stage stochastic program. In the second stage, after demands and investments have been realized, the firm allocates production to maximize the contribution. Capacity investment has a linear cost. The risk neutral firm maximizes profit which is the expected value of sales minus investment and production costs. The loss averse firm uses a weight higher than one for loss instances in calculating expected profit. The firm interested in the downside uses mean of a percentage of the left tail of the wealth distribution to calculate the expected profit. The flexibility premium is defined as the relative difference in the marginal total costs at which the firm is indifferent between SD and SF networks. Higher than zero implies firm is willing to pay a higher price for flexibility and lower than zero implies the firm requires a lower price for flexibility to be preferred. They show that the SF network is preferred to the SD network for all reliabilities and for all demand distributions in the case of a risk-neutral firm facing equal resource reliabilities and costs. (indifferent if there is perfect correlation between demands.) When firm is not risk-neutral the flexibility premium value goes down and can be negative for higher failure risks. The downside of the single facility breaking down can be higher than the upside of flexible production in meeting uncertain demand. Numerical study is performed to compare all four (SD, SF, DD, DF) networks. Experiment design factors are correlation of demand signals, relative contribution margin of products, source reliability probabilities and degree of risk aversion including risk-neutral. Source failures are independent and identical. A full factorial design of 7599=2,835 instances were solved for each of the four networks. For each instance 200 demand couples are generated to calculate expected profit for each instance. The result of the numerical study, as expected, finds that the desirability of a dual-sourcing network (DD versus SD or DF versus SF) increases as supply-chain reliability decreases. They further report that flexible sourcing is more preferable as demands become more negatively correlated, or the spread in contribution margins increases, or the number of products increases. The dedicated sourcing is more preferable when resource investments become less reliable or firm becomes more concerned about the downside risk.

On the Value of Mitigation and Contingency Strategies for Managing Supply Chain Disruption Risks Brian Tomlin Management Science Vol. 52, No. 5, May 2006, pp. 639657 In this article there is a single product and a single firm with two suppliers, one reliable and one unreliable. Unreliable supplier fails from time to time such that it supplies nothing. There is infinitehorizon and full backlogging of unmet demand. The aim is to show the value of having a plan for the possibility of a supplier's failure. In general companies do not plan for supplier failure so by default choose the acceptance strategy, which is optimal only if the failure happens with a very low probability and the supplier's time to recover is short. In the real life examples given, companies that do something after a supplier disruption, ended up being more successful than other companies which were affected by the same calamity and just waited for the supplier to recover. Actions taken by companies for a supplier failure can be categorized as mitigation or contingency tactic. Mitigation tactics are those in which the firm takes some action in advance of a disruption (and so incurs the cost of the action regardless of whether a disruption occurs). Contingency tactics are those in which a firm takes an action only in the event a disruption occurs. The contingency tactic in the article is paying the reliable supplier to increase capacity in case of failure. An example of contingency tactics, which is not modeled but mentioned, is demand management. Demand management is switching customer demand to a substitute. Another strategy is financial mitigation. Financial mitigation is being compensated by insurance or something similar for losses due to the disruption. Contingency tactic of using another source needs a readily available reliable source with volume flexibility. The firm balances the replenishment cost of the backlogged demand to carrying inventory and rerouting orders. Capacity of the unreliable supplier, the flexibility of the reliable supplier, the risk averse or neutral nature of the firm, the probability of the supplier being up and the expected length of down time of the supplier are elements that impacts the selection of the best tactic. In conclusion firms has to be ready for downtime of their suppliers. Using inventory mitigation is not an attractive strategy in an environment of rare but long disruptions, as significant quantities of inventory need to be carried for extended periods without a disruption. So it may best for the firm to plan for an alternative supplier in case the current supplier goes down for some reason and is not likely to resume production anytime soon. Advance notice of an upcoming disruption, to be discussed in later articles, would change the strategies and decrease the cost of tackling the problem.

On the Value of a Threat Advisory System for Managing Supply Chain Disruptions Brian T. Tomlin, Lawrence V. Snyder Working paper Current version: April 26, 2007 This article is about the value of information about the failure risk of the supplier. The article assumes that the firm can observe and calculate the failure risk fully and in the risk is regularly updated. The benefit of the risk system is the cost reduction that comes from changing inventory level and use of an additional supplier when necessary. They conclude that there is a value to a system that analyzes and updates risk in suppliers. The value is calculated by comparing the cost if the firm employs a single strategy disregarding the changes in risk and the adaptive inventory level and supplier policy that changes with changing risk and recovery probabilities. They assume that the supplier has multiple threat levels with different failure probabilities that can be observed and identified. The change between threat levels is a Markov process independent of time passed since last recovery. After a failure, the recovery probability in the next period depends on the time passed since the failure, the state from which failure occurred and the threat level of the supplier after the recovery. There is full backlogging of unmet demand. When model has finite number of periods the ending inventory is salvaged. The value of the threat advisory is significantly influenced by the structure of the disruption risk process and the supplier capacity. The structure of the disruption risk process consists of the relative disruption risk in different threat levels and the nature of transitions between threat levels. When there is an alternative supplier available, in the finite horizon, the product life-cycle may affect the strategy chosen.

The power of flexibility for mitigating supply chain risks Christopher Tang, Brian Tomlin Int. J. Production Economics 116 (2008) 1227 The article models five different instances to show that investing in flexibility gives back in significant savings. They report that firms are reluctant to invest in flexibility because reliable data and accurate cost-and-benefit analysis are hard to obtain. They conclude that with the examples they have shown even when data is inaccurate firms are better off by investing in some flexibility rather than doing nothing to prepare for a possible disruption. The cost of investment for flexibility is not discussed. They do not consider rare events that may disturb a supply chain. They model a different strategy to deal with each risk, so 5 in total. For each strategy, they plot the various degrees of the flexibilities versus the percentage gain over no flexibility. They show in every model that gain from flexibility is an increasing concave function. So they conclude even a low degree of flexibility would bring significant benefit and be better than not planning for a possible disturbance. Here are the models considered in the rest of the paper: Model 1. The risk is supplier's cost is random. The strategy is to use have more than one supplier and order all from the cheapest. The number of suppliers ranges from one to five. The price of each supplier is IID. The percentage cost reduction of multiple suppliers compared to one supplier is a concave increasing function. Model 2. The risk is the demand is random. The strategy is to have a flexible amount supply contract rather than fixed amount contract. The firm places the order before the demand happens. After the demand I realized, the firm has percentage flexibility in changing the amount of the order up or down. For different values of flexibilities, the percentage profit gain over fixed order profit is a concave increasing function. Model 3. The risk is the capacity of the plants. The flexibility is the ability of plants to produce more than one product. Demand for each product is equal and constant. The model has four products and four plants. The numerical example shows margin improvement in sales, ability to meet the demands, for 2 flexible system. Expected profit for 2,3 and 4 flexible plant system is same. So they say it is a concave increasing function in flexibility. Model 4. The risk is the uncertain demand. The flexibility is postponement of product differentiation. There are 2 products. Production takes T periods for both. Both products begin as a generic product and differentiates at time 'tau'. It takes T-'tau' periods to transform the generic product into 2 different end products. The model has infinite-horizon and full backlogging of unmet demand. Percentage savings in safety stock over no postponement is plotted for two different demand models. Savings in IID (independent and identically distributed) demand is linear and RW (random walk) demand is concave and increasing with increase in postponement 'tau'. Model 5. The risk is the uncertain demand. There are two substitutable products. Flexibility is the timing of the pricing decision on the products. As more time passes demand is less uncertain. Demand shock terms are additive and follows auto-regressive process of order one (AR(1)). The plot of percentage increase in expected profit via t postponement over zero postponement is increasing and concave in t.

Disruption-Management Strategies for Short Life-Cycle Products Brian Tomlin Naval Research Logistics, June 2009, 56 (4), 318-347. There are two risks, supplier may fail and the demand is uncertain. There are three methods considered for mitigating these risks, supplier diversification, contingent sourcing and demand switching. These three methods are investigated in all possible combinations resulting in twelve strategies. There is one period, a newsvendor type setting. Cases of a single product and two products are investigated. There is a cost of ordering per unit, so the firm pays something even when the supplier fails to ship goods. There is no cost of having a supplier diversification. Both suppliers are same in every aspect, like cost and risk of failure. The contingent source has a higher marginal cost and perfect reliability. There is no fixed cost to retain a contingent source. Switching demand has an extra cost and there is a limited fraction of customers who are willing to switch. The firm is modeled both as risk-neutral and risk-averse. For the risk averse firm the utility is equal to profit when there is a profit and when there is a loss the firm's utility is a multiple of the loss, where multiplier is bigger than one. For the single product, obviously, there is no demand switching. So there is a total of four possible strategy combinations. When the firm uses only a contingent supplier, the initial order amount is same as when the firm uses no strategy. The contingent source is only used in case of the supplier's failure. In case of supplier diversification, the firm uses two identical suppliers. The optimal ordering amount is same for both suppliers. The total initial order is either equal to when there is a single supplier or more. The firm orders more and makes inventory hedging when the contingent source is unavailable or very expensive compared to regular suppliers. If the firm is risk-averse, the firm may choose a strategy where the contingent source is used even when the regular supplier delivers. This strategy happens when the firm is very risk-averse and the supplier reliability is low. The article concludes single product case with some numerical results to compare single-tactic strategies, use of dual-sources and contingent source. For the risk neutral firm, the benefit of using a single tactic is less than 1% when supplier reliability is higher than 97%, (the base case reliability is 98.5%). For the two product firm, if there is no demand switching, the risk-neutral firm acts equivalent to one product case. If there is no demand switching, the risk-averse firm prefers dedicated sourcing to single common sourcing, where risk-neutral firm is indifferent between them. When there is demand switching some results about strategies is given by assuming deterministic equal demand for both products. Optimal single-tactic under different conditions (base, risk averse, negative correlation, high product substitutability) is investigated (demand covariance versus supplier reliability) by numerical studies. Another numerical study is made to compare two-tactic strategies for risk-neutral firm. If either supply or demand risk is negligible, two-tactic strategy does not add value over single tactic. Best twotactic strategy delivered on average about 90% value of the three-tactic strategy. Three extensions to the model are: alternative dual sourcing strategies, contingent capacity reservation and non-identical suppliers. The alternative dual sourcing compares sourcing for two products and more than one suppliers. The regular model uses two suppliers which supply both products. The extension defined as shared dual source is three suppliers where one supplies both products, the others are dedicated suppliers. The extension defined as dedicated dual source has four suppliers where each product has two dedicated suppliers. The ability to switch demand makes these extensions more profitable than base case. If the firm has to reserve capacity at the contingent supplier for a cost, using contingent supplier becomes less desirable. If the firm uses contingent supplier, for the risk-neutral firm the initial order may differ and the contingent source may be used even if the supplier does not fail.

The article concludes that depending on the conditions a firm is in, the strategy that is best for a firm differs. Thus same strategy may not be best for every firm. If the firm is risk-averse, implementation of some form of strategy is more important.

Impact of Supply Learning When Suppliers Are Unreliable Brian Tomlin Manufacturing & Service Operations Management, Spring 2009, Vol. 11, No. 2, pp. 192-209 There are two risks, supplier failure and unknown demand. The supplier's failure rate is not known for sure and the firm updates the probability over time. Supplier uncertainty is modeled as both all or nothing and percentage yield. There is a cost of ordering that is linear to the order amount that is paid even if the supplier fails. There is a minimum order amount. Supplier lead times are assumed to be zero. Demand no filled in a period is lost and incurs a cost of \pi per unit. In the first model there are two suppliers. First supplier's failure rate is known. The failure rate of the second supplier is assumed to follow a Beta family distribution. A Beta distribution easily updates after a Bernoulli trial. There are T time periods, inventory cannot be carried from one period to the next and unfilled sales are lost. The model minimizes cost. Because of equivalence of expectation, after the decision to whether or not to source from supplier two, the amount of order is independent of uncertainty on the second supplier's failure rate. If it is not optimal to order from second supplier in a period then for the rest of the periods it will not optimal to order from the second supplier. Closed form solution do not exist in general to which estimated values of the distribution that the second supplier becomes unprofitable. Second supplier can be more attractive or stay attractive by having a lower ordering cost and or lower minimum order amount.

To Wait or Not to Wait: Optimal Ordering Under Lead Time Uncertainty and Forecast Updating Yimin Wang, Brian Tomlin Naval Research Logistics, 2009, Vol 56, pp. 766-779 Managing Disruption Risk: The Interplay Between Operations and Insurance Lingxiu Dong, Brian Tomlin Working Paper Mitigating Supply Risk: Dual Sourcing or Process Improvement Yimin Wang, Wendell Gilland, Brian Tomlin Manufacturing & Service Operations Management, Summer 2010, Vol. 12, No. 3, pp. 489-510 Regulatory Trade Risk and Supply Chain Strategy Yimin Wang, Wendell Gilland, Brian Tomlin Production and Operations Management, Forthcoming.

S-ar putea să vă placă și