Skip to content

Network simulation and what-if

Most network decisions rest on a spreadsheet comparing two cost totals. Simulate the new facility against past and possible demand before you commit millions.

  • Free 30-minute call
  • No commitment
  • You own the code
Typical industries
Supply chain, Logistics and transport
Problem type
Supply chain
You receive
A documented model, run on your data

In short: A logistics network simulation is a model of your warehouses, transport links and demand that tests a change before you make it. It runs a new site or route plan against many possible demand scenarios and shows the cost and the share of orders delivered on time.

The problem

The board wants a decision on the new distribution centre (DC) by the end of the quarter. You build a spreadsheet with two columns. One shows today’s network: your warehouses, your distribution centres and the transport between them. The other shows the same network with the new site. You calculate both columns with last year’s volumes, and the new network comes out cheaper. The site gets approved. Nobody has asked what happens if one region orders 20% more next year.

Signs your network decisions rest on one scenario

You argue about network changes with cost totals from a single year. Two other numbers come up only after the site is open. Service level is the share of orders delivered on time, say 97 out of 100. Utilisation is how full a site runs, say at 85% of its capacity. Nobody can say how your sites cope with a demand spike, demand moving from one region to another or a large new account. Each new what-if question means rebuilding the spreadsheet. By the time the answer arrives, the lease is often signed.

Why a one-shot comparison fails

A network decision is a bet you can’t easily take back. Maybe you open a DC, move a transfer warehouse or close a site. Or you change which region a DC serves. Each of these decisions costs millions for buildings and equipment. It also fixes for years which site serves which customers. Yet most companies make this decision with a spreadsheet that compares two totals.

That comparison runs last year’s volumes through the current and the proposed network. Whichever is cheaper gets the money. It tests neither a demand spike nor demand moving from one region to another. It measures cost, but not service or utilisation. A network variant is one set of sites and transport routes. It can look cheaper and still deliver late to customers at the edge of its delivery area. Or it runs one facility flat out while another is half empty.

What changes

Today, the new site is approved on one spreadsheet and one year of volumes. With our model, the same site is tested twice before you sign the lease. It runs first against last year’s demand, then against thousands of plausible future years. For each of these years, you see cost, service and utilisation. And you see which site runs out of capacity first. The board decides and knows which risks it takes.

Logistics network simulation: what you get

You test the new facility on a model before you commit to the site. You see whether it works on the demand you had, and how often it fails on demand you might get. The risk gets a name, such as a demand jump in one region, before it becomes a loss.

We build your DC network simulation on your own data: where your goods went in the past, what sites and transport cost, and the delivery times you promise. The same model then answers new what-if questions as they come up. What if you lose a large customer, a motorway closes or you add a site? You see the new case next to today’s network and compare cost, service and utilisation. You get transport network scenario analysis you can reuse.

  • A model written around your own network simulation and what-if rules.
  • Results measured in the numbers you already track.
  • It slots into how you already work, and you can check every answer it gives.

How it works

Forecasts, stock and decisions about your warehouse network go into one model instead of three spreadsheets. By network we mean all your warehouses, distribution centres and branches, and which site supplies which. The model plans forecasts, stock and network decisions together in a single run. So a fix in one warehouse no longer creates a new problem at another site.

Static network simulation

Static simulation takes your historical volumes and reruns them through a proposed configuration. The demand stays as it was. What changes are the sites, the routes between them and which site serves which customer. For the demand that happened, you get the new network’s cost, service and utilisation. You see which site becomes the bottleneck and on which route you need more trucks than you planned.

Static simulation is the minimum check for every network decision. It tests whether a network variant copes with the demand you had. That alone rules out a surprising number of attractive-looking ideas. Take a new warehouse that saves transport cost on paper but would have been too small for three weeks last December.

Monte Carlo demand scenarios

A rerun shows the network on the year that was. It says nothing about a demand spike, about demand moving to another region or about a large new customer. Monte Carlo simulation covers those cases. This method plays through thousands of possible years, chosen at random. It’s named after the casino in Monaco, because it works with chance. The computer draws many plausible demand years at random. Each one varies as much as your demand has varied in the past. Then it runs the candidate network through every one of them.

A static rerun tells you whether the network works for the year you had. A Monte Carlo run tells you how often it fails, how badly and in which situation. “The new DC looks fine” becomes a sharper statement. For example: “it works in 92% of scenarios but fails on a northern demand jump.” With demand scenario logistics modelling, your board gets a statement it can decide on.

How many scenarios are enough?

Any board member may fairly ask why a thousand runs and not ten. The error of a Monte Carlo estimate is how far the result may be from the true value. It shrinks with the square root of the number of runs: four times as many runs halve the error. That’s true however many sites you have.

Say the new network misses a service target in 5% of simulated years. With 1,000 runs, the true value lies between about 3.6% and 6.4%, so 1.4 points either side. You can be 95% sure of that range. Statisticians call it a 95% confidence interval. With 4,000 runs it’s about 0.7 points either side. If the decision turns on 5% versus 8%, a thousand runs is plenty. If it turns on 5% versus 6%, you need more runs.

Static and Monte Carlo, together

You see most when both test the same configuration. Static simulation checks the decision against what happened. Monte Carlo tests it against what could happen. Drop either one, and you lose half the picture. Supply chain network design uses optimisation to propose a network variant. Simulation shows whether you can trust that variant before you invest.

In a network with many sites, thousands of scenarios make a large computation. Two techniques keep it manageable. Decomposition splits the problem into smaller parts. Parallel computing runs many scenarios at the same time. The answers arrive while the decision is still open. Later, you plan stock, restocking and backup plans for exactly the sites you choose. A well-tested site makes each of those later decisions easier.

Common questions.

Something else? Ask us directly.

Optimisation searches for the best network design for the data you give it. Simulation takes one design and tests how it behaves when demand varies. You need both. Optimisation proposes a few candidates. Simulation shows which one still works in a bad year. A design that’s best for last year’s volumes can fail when demand moves.

From your own history first. We measure how much your volumes have varied by region, customer and season. Then we generate many possible years at random that swing just as much. On top come scenarios you name, such as a big account won or lost, a region that grows or a site that closes. You see which results come from history and which from assumptions.

Yes, and that’s where you get a good part of its value. Once your flows, costs and service rules are in the model, a new what-if takes one model run. You don’t start a new spreadsheet project. You can use it for tenders, a new customer’s volumes or a change of carrier. Keep the data up to date, and the model stays up to date.

Sometimes it can, and if that’s the case, we’ll tell you. Standard software is built for an average operation, though. Yours runs on its own rules: your contracts, your exceptions, the things you know by heart. A custom model writes those rules in, and that’s where the better plan comes from.

You don’t need either to start. We build the model from the data you’ve already got, even if it lives in three spreadsheets and one person’s head. You get a system you run yourself. We train your team until they no longer need us.

You see a first result with the prototype, before you commit to anything bigger. We start with a free 30-minute call about your operation and the model that would fit it. Then we build a working prototype and run it on your numbers.

Tell us what you’re planning.

In a free 30-minute call with a specialist, you find out whether optimisation can help.

  • Free, no commitment
  • You talk to a specialist directly
  • A clear go or no-go