Network Design Optimization

When this approach is needed

Typical business situations

Case 1

After growth, an acquisition or a change in demand pattern, the facility network no longer fits.

DecisionWhich sites to keep, close or open, and with what capacity?

Case 2

Concentrated sourcing is cheap until it fails.

DecisionWhat is the cost of diversification, and what resilience does it buy?

Case 3

Commercial targets and cost targets pull the network in opposite directions.

DecisionWhat does each additional point of service level cost, and where is the efficient frontier?

Case 4

Before a peak season, a contract renewal or an investment, the current network needs a robustness check.

DecisionWhich single site or lane hurts most if lost, and what combination of failures is worst?

What you receive

Recommended network configurations that perform well across the scenario set
The cost–resilience trade-off, expressed in the same currency as cost
An N-1 contingency ranking and the worst plausible combination of failures
The distribution of outcomes under realistic failure rates, not only the expected case

How an engagement runs

DataBuild the network model: sites, lanes, capacities, costs, demand and failure rates.
OptimiseSolve the design problem across demand and disruption scenarios.
StressTest the candidate designs by N-1 analysis, Monte Carlo sampling and worst-case search.
DecideCompare the shortlisted designs on cost, service and resilience, and agree the plan.

Reference

Approach: Facility, sourcing and capacity decisions are usually optimised for landed cost against a single forecast, with resilience assessed afterwards. Optimising across demand and disruption scenarios from the outset puts the cost of failure into the same decision as the cost of operation.
Core question: Which network configuration is best once the cost of things going wrong is priced in, not only the cost when everything goes right?
Method: Mixed-integer and two-stage stochastic programming for the design; min-cost flow, Monte Carlo sampling and adversarial worst-case search to characterise performance under failure.
Evidence: The Network Stress Test solves a production–warehouse–consumer network by min-cost flow and runs N-1 contingency scans, Monte Carlo failure sampling and an adversarial worst-case search; Network Design Visuals is a library of seventeen network-design visualisations.
Related offer: Network Flexibility Study.

Get in touch