Stress testing involves evaluating how decision alternatives perform in different conditions.
When to use
Understanding how performance of decision alternatives can vary in future can help design better management options by understanding in what conditions an alternative is "robust" and performs well, or under what circumstances changes will need to made within an "adaptive" solution.
Measures of performance and one or more decision alternatives need to be specified. Stress testing can be performed in an optimisation context (e.g. to identify robust decisions), in which case the decision alternatives are defined within a model.
How
Stress testing methods differ in how they characterise scenarios and the form of output they provide. Scenarios can be pre-determined, defined by model parameters, or created within the stress testing process. Stress testing may report failures, simply show how performance varies, or aim to compare decision alternatives.
Specific classes of approaches therefore include:- Vulnerability analysis specifically identifies conditions in which management options can fail
- Mapping performance as a function of external drivers, e.g. in Decision scaling
- Evaluating robustness metrics, e.g. in Robust optimisation or "Robust decision making"