(614b) Sizing Safety Stock for Supply Chain Risk Management: The Sole Source Disruption Risk

Relying on a sole source supplier for raw material used at process facilities (such as petrochemical facilities, integrated coke ovens, blast furnaces at steel production facilities, or coal-fired power plants) may be strategically chosen for economic or other reasons, but such a strategy is not without significant risks if safety stock / inventory is not properly determined and maintained.  Consider the following from current research in this area:  “Single sourcing, where one supplier is responsible for the supply of a specific item or service may be advantageous from a cost and quality management perspective, but is dangerous in terms of resilience.  While it may be desirable to have a lead supplier, wherever possible alternative sources should be available.”[1]  However, the availability of alternative sources may not mean that alternative or multiple supply transport routes are available, leading to a “sole source supplier” problem even if multiple suppliers are available.

Properly determining the size of an inventory stockpile presents numerous challenges even without the additional constraint of sole sourcing.[2],[3],[4]  In addition, the “decision maker must balance the cost of holding inventory against the risk of running out of fuel,”[5] that is, the classic trade-off between inventory holding cost and potential shortage cost.  The option of continually increasing stockpile size without knowledge of what size is sufficient given anticipated risks might protect against certain supply disruptions and is easy to implement,[6] but it can be cost prohibitive.  Reducing the number of suppliers for a significant raw material (such as coal for a utility) can itself “breed vulnerability in the supply chain.”[7]

We define “sole source supplier” in a general sense to not only include single commodity suppliers but also the situation that might involve multiple suppliers, all of whom rely on the same individual transport route for shipping the commodity to its point of use.  Examples of this situation are numerous.  The Joint Line in the Powder River Basin (Wyoming) is a “single-path-to-destination” transport route that is relied upon by a number of utilities across the United States for a steady and regular supply of low-sulfur coal.  The Ohio River with its multiple transloading facilities (from rail to barge) is another primary example.  Commodity stockpiles have been seen in certain circumstances as a “backstop”[8] against unreliable rail or waterway service that might be caused by a significant disruption (e.g., derailment or flooding, respectively) in these transportation routes.

When relying on a sole source supplier or multiple suppliers that rely on a single transportation route (such as a single captive rail line or waterway), how much raw material should be held as safety stock?  The answer depends on many factors, including the manufacturer’s awareness of impending disruptions[9] (“warning time”) and its contingency plan to procure emergency inventory from other sources.  In this paper we focus on protecting a continuous manufacturing operation from supply disruptions due to transportation failures involving single-path-to-destination routes.  Specifically, we consider how to determine the optimal size of a coal stockpile for a coal-burning power plant subject to disruptions in single-path rail and/or barge transport.  The frequency distributions of transportation disruptions involving single-path-to-destination or captive routing are drawn from historical data collected in the United States from 1990 to 2010.  The data reveal a broad statistical distribution for frequency and duration of disruption events.

It is assumed that demand is constant and that plant operation responses such as reduced burn are not viable options.  The size of the safety stock is determined using discrete event simulation guided by ideas drawn from queueing theory and performed in the context of existing research in this area.  The economic factors contributing to the objective function are drawn from published data.  The study objective is to determine the safety stock level required to cope with the median (50%), 90^th percentile, and 95^th percentile disruptions based on the selected data sets.