Network Optimisation Modelling Basics
Before optimisation, a company sees chaos: trucks moving half-full, one warehouse overflowing, another starved, and customers waiting because stock is in the wrong city. After optimisation, the same network looks almost calm - each plant, warehouse and route has a role, and every shipment answers one question: what is the best feasible move?
- Network optimisation modelling chooses the best facility locations, capacities and flows under cost, service and capacity constraints.
- The model has four building blocks: nodes, arcs, decision variables and constraints.
- The objective is usually to minimize total landed cost, but good models also protect service, resilience and working capital.
- Common model types include transportation models, facility location models, vehicle routing models and multi-echelon inventory models.
- A strong answer never says “open the cheapest warehouse”; it compares network-wide cost, lead time, utilization and risk.
- The interview trick is to translate business language into model language: demand becomes requirements, plants become capacity, lanes become costs.
Big Picture: Network Optimisation Turns a Messy Supply Chain into a Feasible Choice
At its core, network optimisation is not “drawing a better map.” It is converting a supply chain into a decision model where every possible flow has a cost, every facility has a limit, and the final plan must satisfy demand without breaking constraints.
Core Explanation: The Model, the Logic and the Trade-Offs
Network optimisation modelling is the mathematical selection of facilities, flows and capacities that minimizes cost or maximizes service under constraints.
The easiest way to understand it is to imagine a map with dots and lines. The dots are nodes - plants, vendors, warehouses, stores and customers. The lines are arcs - transport lanes between nodes. The model decides how much should move on each arc, which nodes should operate, and what capacity each node should carry.
The Four Building Blocks of a Network Model
A clean network model has four parts. If you can explain these clearly, you can handle most interview questions on the topic.
Common Types of Network Optimisation Models
Not every network problem needs the same model. The model type depends on what decision you are making.
If inventory placement is central to the problem, revise setting inventory policy for a multi-product business because network design and inventory policy often decide service together.
The Basic Network Optimisation Process
In practice, good teams do not jump straight into software. They first structure the business problem, then model it.
Worked Example: A Simple Transportation Model
Suppose a company has two plants and two regional warehouses. Demand must be met, plant capacity cannot be exceeded, and the objective is to minimize transport cost.
Plant A can supply 700 units. Plant B can supply 500 units. North Warehouse needs 600 units and South Warehouse needs 400 units.
The low-cost logic is clear: serve North from Plant A and South from Plant B.
This is the simplest form of optimisation: choose flows that meet demand at the lowest feasible cost. In real projects, you add more plants, warehouses, routes, fixed costs, lead times, minimum order quantities and disruption scenarios.
Metrics to Track in Network Optimisation
A model is only useful if its recommendation is judged using the right metrics. Track cost, service, capacity, speed, inventory and resilience together.
For a deeper AI-led view of replenishment once the network is designed, revise using AI for inventory optimisation and replenishment.
Definitions You Should Be Able to Say Clearly
- Network optimisation modelling: Selecting the best facilities, flows and capacities under cost, service and capacity constraints.
- Node: A physical or logical point where goods are sourced, stored, processed, transferred or consumed.
- Arc: A possible movement path between two nodes, usually carrying cost, time and capacity assumptions.
- Decision variable: A model choice whose value is solved, such as units shipped or facilities opened.
- Constraint: A rule that limits feasible solutions, such as demand, capacity, lead time or regulatory limits.
- Objective function: The mathematical expression the model tries to minimize or maximize.
Case Study: Domino's India and the Power of Dense Local Nodes
Domino's India shows why network optimisation is not just about transport cost - it is about placing capacity close enough to demand to protect speed, quality and economics.
Hot pizza has a brutal supply chain problem: the product loses value quickly after preparation. A low-cost central kitchen may look efficient on paper, but if it is too far from customers, delivery time rises and the customer experience weakens.
Domino's India, operated by Jubilant FoodWorks, demonstrates the opposite design logic. The network is built around many local outlets that act as production and delivery nodes. Each node serves a limited delivery radius, while common processes, standardized ingredients and demand visibility help the system stay consistent.
The primary driver is proximity to demand: smaller service territories reduce delivery distance and protect freshness. The supporting drivers are equally important: standardized store processes, centralized supply planning, menu discipline, digital ordering data and local capacity planning during peak meal windows.
The strategic lesson is powerful: in time-sensitive categories, network optimisation may recommend more nodes, not fewer, because customer promise is part of the objective function.
How AI Changes Network Optimisation Modelling Basics
AI does not replace the optimisation logic. It improves the inputs, speeds up scenario generation and helps managers interpret trade-offs faster.
Student workflow: Put a simple network table into ChatGPT or Claude with columns for nodes, lanes, costs, capacities and demand. Ask it to identify decision variables, objective function, constraints and three interview-worthy scenarios. Then validate the arithmetic yourself - AI is useful for structuring, not for blindly trusting the final recommendation.
Interview Relevance
“A consumer goods company serves all of South India from one warehouse near Mumbai. Delivery cost and lead time are rising. How would you decide whether to add a warehouse in Bengaluru or Hyderabad?”
Use the phrase “total landed cost plus service constraint.” It signals that you understand both finance and operations, which is exactly what interviewers look for in network design answers.
Common Mistake
The biggest mistake is optimizing one lane or one warehouse in isolation. That fails because a cheaper lane can overload a node, increase inventory, worsen lead time or create single-point failure risk. The fix: always answer at the network level - objective, constraints, scenarios and trade-offs.