Models for Supply Chain Risk Management
Helene Tokou and Dennis L.
Bricker
Many
strategies to improve the performance of supply chains (e.g., outsourcing and
JIT) simultaneously increase their vulnerability to the risk of various disruptions.
Exposure to risks can be reduced if robust decisions are made, i.e., decisions
which allow for corrective actions (recourses) in order to recover from
disruptions to the supply chain. This paper addresses the management of
external risk by means of stochastic programming models which incorporate
uncertainty, robustness, and reliability.
We
take uncertainty into account through the use of a finite number of possible
future scenarios. As in any scenario-based models, the decision-makers first identify
a number of future scenarios, and associate a probability of occurrence to each
of them. Decisions in these models are of two types: those which must be made
before the random scenario is realized, and recourses made to mitigate the
impact of the disruptions implied by each scenario.
Of
course, in the case of any scenario that might occur, the earlier decision will
hardly ever be ideal in hindsight, that is, the decision-maker will experience
some regret. Unlike typical stochastic
programming models which attempt to optimize the expected performance of the
supply chain, our model attempts to minimize the maximum regret of the
decision-maker. Because it may be
unrealistic to insist on 100% reliability of the supply chain, the model allows
for the exclusion of a set of scenarios with acceptably low total
probability. A new decomposition scheme
is described for the optimization of our model.