Inventory Optimization

What it covers

It starts with formulation, which is where everything is decided, moves on to the cases a rule cannot handle, and ends with how to read and put into production what the solver returns.

1. From rules to modelsWhat a per-SKU replenishment rule solves and what it does not: budgets, capacity, trucks and products competing for the same thing.
2. Linear programming for plannersVariables, constraints and an objective, with a purchasing example. What a shadow price is and why it is the most useful number a solver returns.
3. Formulating a purchaseFrom the statement to the model: what is a variable, what is data and what is a constraint. The formulation mistakes that return the optimal answer to a problem that was not yours.
4. Integer programmingOrder minimums, case multiples, fixed ordering cost and full trucks. Why adding a single integer variable can multiply solving time.
5. Products that competeBudget, warehouse space and supplier capacity shared across thousands of SKUs. How to allocate what is scarce without the same item always winning.
6. Multi-echelonDistribution centre and branches. Risk pooling, how much stock belongs upstream and how much downstream, and when a transfer between warehouses pays off.
7. Optimization under uncertaintyScenarios, two-stage stochastic optimization and chance constraints. How the forecast distribution gets inside the model.
8. Robust optimizationWhen you do not trust the distribution: solutions that hold up under the reasonable worst case, and what that peace of mind costs.
9. Multiple objectivesService level, tied-up capital and logistics cost together. Pareto fronts, weights and constraints, and who picks the point.
10. Solvers in practiceOpen-source and commercial solvers, optimality gaps, time limits and heuristics. When a solution within 1% of optimal is the one to take.
11. Reading the solutionSensitivity, shadow prices and binding constraints: explaining to a buyer why the model ordered what it ordered.
12. In productionRe-optimizing every week without the plan jumping around. Stability, frozen horizons and exception handling.