Frame, objectives, and scoring

State the choice, decision owner, portfolio boundary, planning horizon, currency or value unit, affected stakeholders, assumptions, and non-negotiable constraints. Make the question operational: “Which candidate capital projects should enter the 2027–2029 program under the approved labor, equipment, and budget envelopes?” is stronger than “Which projects are best?”

Objectives explain why the portfolio exists. Give each objective a clear name, meaning, and weight. Do not mix incomparable definitions across initiatives or double count the same benefit under several labels. The scoring weights for strategic alignment, value, risk, and confidence express a declared decision policy; they are not empirical facts. Record who agreed to them and test plausible alternatives.

Safety, legal, environmental, equity, contractual, and minimum-service requirements often belong as hard feasibility constraints or protected commitments, not as small weighted penalties that can be traded away. Keep non-monetized consequences visible even when the common value unit is currency.

The canonical constraints field contains human-readable notes only; it does not enforce feasibility and is never an alias for relationships. Encode a rule that the engine must enforce through resource capacity and demand, category bounds, typed relationships, mandatory status, or scenario locks, as appropriate.

Next: resources, periods, capacity, and demand.