Optimizing Supply Chains With Pyomo and HiGHS

About This Book

Supply chains are not optimized by spreadsheets alone. They are optimized by translating real business decisions—what to produce, where to source, how much to ship, which facilities to open, what capacity to install, and how competing products should share constrained resources—into mathematical models that can be solved, validated, and trusted. Optimizing Supply Chains with Pyomo and HiGHS is a practical technical textbook that takes the reader from the fundamentals of mathematical optimization to production-oriented supply-chain decision systems using Python, Pyomo, pandas, NumPy, and the HiGHS optimization solver. The central philosophy of the book is: Data → Model → Optimize → Validate → Decide Rather than presenting optimization as a collection of mathematical formulas, the book develops problems as complete decision systems. Business assumptions are converted into sets, parameters, variables, objectives, and constraints; data is prepared through pandas pipelines; models are solved using Pyomo and HiGHS; and solutions are independently validated before they are interpreted as business decisions. A recurring case study, Kestrel, provides a common supply-chain environment through which increasingly sophisticated optimization problems are developed. The network evolves from basic production and distribution decisions into capacity constraints, facility location, multi-capacity facility design, and finally multi-commodity network flow. The book also emphasizes something frequently missing from optimization training: how models fail. Readers encounter deliberately broken models in which the solver still reports optimal—but the business answer is wrong. These examples demonstrate why model validation, data validation, constraint verification, independent reconciliation, Big-M discipline, sparse modeling, and interpretation are essential in real-world optimization. The result is not simply a guide to writing Pyomo code. It is a framework for learning how to think like a supply-chain optimization analyst. What Makes This Book Different? 1. It connects mathematics to business decisions The book continuously answers: What does this equation mean operationally? Instead of learning: \(x_{ij} \geq 0\) in isolation, the reader learns what x[i,j] represents in an actual supply-chain network and why its value matters to procurement, transportation, inventory, production, or facility decisions. 2. It teaches modeling, not just software Pyomo is the implementation language—not the subject itself. Readers learn to recognize: decision variables objective functions equality and inequality constraints capacity constraints fixed-charge decisions binary decisions Big-M formulations network balance facility-sizing decisions shared-resource constraints linear programming mixed-integer linear programming integrality and LP relaxation model validation 3. It uses real data-pipeline thinking The book incorporates pandas as part of the optimization workflow. Data is organized into structures such as: node tables edge tables balance tables facility tables capacity-tier tables commodity/balance tables This creates a bridge between the messy data an analyst receives and the mathematical structure an optimizer requires. 4. It teaches readers to distrust an unexplained optimal One of the strongest themes of the book is: A solver saying "optimal" does not mean the business model is correct. The book deliberately demonstrates silent failures where: a facility is reported closed but still carries flow; a small facility carries more volume than its physical capacity; multiple commodities effectively receive private access to shared capacity; a loose Big-M weakens the formulation; an incomplete model produces an optimal answer to the wrong decision space. The reader therefore learns to validate the model and reconcile the result independently of the objective that produced it.