Scenario design, stress testing and uncertainty

Now combine that price rise with 20% more demand in the affected interval. The 09:20 quantity becomes 12 MWh and the new interval cost is A$3,600. Replace its original A$2,000 in the base total to obtain A$6,350. The joint delta is A$1,600, not A$1,400.

The difference is an interaction: price-only adds A$1,000; volume-only at the original price adds A$400; the additional 2 MWh also experiences the A$100 price increase, adding another A$200. A one-factor-at-a-time chart can miss that term. Use coherent combinations when high demand and high prices can occur together.

Describe which quantities stay fixed and why. A demand-response option may alter both consumption timing and price exposure; a fixed profile cannot demonstrate its optimised value. Keep the simple scenario transparent, then identify the operational, contractual and behavioural data needed for a richer model.

Suppose an analyst proposes shifting 2 MWh away from 09:20. First state where and when the energy goes, the service obligation and the asset's power, energy and recovery limits. If those 2 MWh move to 09:25 at its stated zero price, the simplified base-case energy saving is A$400. That arithmetic ignores whether the customer can actually move the task and whether the extra demand fits its connection.

If a battery supplies the interval instead, account for charging energy, conversion losses, state of charge, degradation and any competing service. Do not count the same capacity as simultaneously available for backup and unrestricted trading without an operating policy. Compare like-for-like costs and service quality, not a constrained base case against an unconstrained alternative.

State a break-even condition. For an option that saves A$400 in the example, all incremental costs and lost service value must remain below A$400 for a positive net benefit at that boundary. This is not an annual value estimate. Use many representative intervals and realistic availability before assessing project economics.

Lead with the decision and its owner, followed by the reconciled base case, the material downside and the action you recommend. Keep facts, model outputs and judgement distinguishable. A manager should be able to locate the data source, reproduce the calculation and see what would change the recommendation.

For this lab, a defensible conclusion is that the illustrated exposure is concentrated enough to justify examining interval-level procurement or flexibility, not that a particular asset must be bought. Propose a next step with evidence: verify demand profiles, obtain product terms or test a service constraint. Name what cannot be concluded from six invented observations.

For a fuller analyst portfolio, pair this memo with the battery and PPA cases and a new dataset not used during instruction. Require an independent reviewer to check the calculations, assumptions, alternatives and communication. A quiz pass or a self-checked spreadsheet is useful practice but does not confer an intermediate professional grade.