Supply chain optimization chooses decision values that improve a stated objective while respecting constraints. Begin with a bounded operational question: for example, how to allocate available pallets to two destinations today. Define a feasible baseline before interpreting any saving.
Write the decision in operational units
Define each variable as a quantity on an allowed route within a stated period. Supply, demand and capacity must use compatible units. Include service restrictions such as delivery deadlines before comparing costs. If cases, pallets and vehicle positions coexist, make their conversion rules explicit.
Google OR-Tools' minimum-cost-flow documentation shows how supplies, demands, arc capacities and unit costs form a network problem. This is a useful model when costs and constraints fit that structure. Fixed vehicle charges, product mixing, time windows or discrete equipment choices can require a different formulation.
Work through a small allocation
The following fictional one-day example has one product, identical pallets, divisible routes in whole pallets and no route-capacity limit beyond available supply. Both destinations require 50 pallets. A supplies 60 and B supplies 40. All routes meet the same service deadline. Costs are invented transport units expressed as dollars per pallet; other costs are excluded.
On a narrow screen, scroll the table horizontally. Keyboard users can focus it and use the arrow keys.
| Route | Cost per pallet | Baseline pallets | Improved pallets |
|---|---|---|---|
| A to X | $2 | 30 | 50 |
| A to Y | $6 | 30 | 10 |
| B to X | $5 | 20 | 0 |
| B to Y | $1 | 20 | 40 |
Baseline cost is 30×2 + 30×6 + 20×5 + 20×1 = $360. The improved allocation costs 50×2 + 10×6 + 0×5 + 40×1 = $200. It saves $160, or 44.4%, within this simplified objective. Check both origins and destinations: A ships 60, B ships 40, and X and Y each receive 50.
Here the optimum can be checked without a solver. Let A-to-X be x. The balance equations make A-to-Y 60−x, B-to-X 50−x and B-to-Y x−10. Nonnegative flows require 10≤x≤50. Total cost is 600−8x, so the least-cost feasible choice is x=50. This reasoning depends on the assumptions above.
Test feasibility before celebrating a result
If B's supply falls to 25 pallets while A remains at 60, total supply becomes 85 against demand of 100. The 15-pallet shortfall makes the original requirements infeasible. A solver cannot repair missing supply by rearranging routes.
Escalate explicit alternatives: an authorized additional source, changed demand or a documented late-delivery decision. Price and approve those changes before solving the revised problem. If shortage is permitted in the model, show it as a visible variable and service outcome rather than quietly accepting an incomplete shipment.
Validate the model against the operation
Keep data vintage, units, assumptions, objective and solution status with the allocation. Recalculate cost independently and check every constraint. Confirm that the warehouse can identify, pick and dispatch the assigned stock within the actual labor, vehicle and timing limits. A cost optimum for a simplified model is a candidate for operational validation.
Repeat scenarios for credible demand and capacity changes. Connect approved plans to the transaction controls in supply chain execution, and keep broader uncertainty in the planning process.
Use the working record
Download the editable supply allocation scenario. Save a copy, complete the actual requirements and evidence fields, and review open issues with the responsible owner. The plain-text record includes instructions, units where needed, the fictional example, and source references.