Skip to content

Multi-site production planning

When each plant plans alone, volume lands at costly sites and stock is held twice. We plan all your plants together, as one network.

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

In short: A multi-site production plan decides which plant makes which products, and in what volume, across all your plants. It weighs each plant’s costs and capacity, transport between sites and shared stock, so all plants together produce at the lowest total cost.

The problem

At Monday’s sales meeting, a large order comes in from the north. The northern plant takes it, because that’s where the customer called. It runs overtime to make it, while a cheaper plant two hours away has a quiet week. Both plants hold their own safety stock of the same product. Each site made a sensible choice for itself. Together, your plants cost you more than they had to.

From head office, multi-plant production looks like a capacity question. Do we have enough plants to make everything we sell? Once two or more plants can make the same product, the question changes: which plant makes how much? Decide well, and all plants together produce at the lowest total cost. Decide badly, and you pay too much at several sites.

Signs each plant is planning alone

Orders go to the plant that has free hours, however expensive it is. Orders stay at a costly plant, because the cost of moving goods between plants is booked in the logistics budget. Every site holds its own safety stock of the same products. Each plant fills its own order book, and no plant manager sees the total cost.

Why planning each plant alone fails

In the spreadsheet, each site’s costs are counted on their own. Every plant plans its own orders, and everyone hopes the total works out. But move an order to the cheaper plant, and freight between sites comes on top. And every site that makes a product needs its own buffer for it. Planning the sites separately misses those links, so the network pays more than it needs to.

What changes

Today, each plant fills its own order book, and transfers and duplicate stock land in other budgets. Our model sends each order to the plant where it costs your whole company least, wherever the order came in. It counts production cost, freight between sites and stock held, all at once. You see why each plant got the volume it did.

What you get

We build your multi-site production planning model on your own plant costs and capacities, your transports between plants and your demand. One model decides which plant makes how much. It keeps the delivery times and stock rules you set. Before anything goes live, we test it on your past orders and on an outage of your most important plant. You see where you pay too much today, and where to act first.

  • A model written around your own multi-site production planning 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.

Production allocation across plants

Each plant has its own cost curve. It shows how the plant’s cost per unit changes with volume. Some plants make a unit more cheaply, thanks to wages, energy or size. Others have more expensive capacity but are closer to an important market. Network allocation sends every unit of demand to the plant where it costs the network least.

Plant cost and capacity profiles

First, each plant’s numbers have to be right. Each site has its own cost per unit and its own finite capacity. It also has its own qualified product families, the products it’s approved to make. Two plants that can make the same product often differ widely in cost once you count energy, wages and setup. Capacity is a hard limit, so the model treats it as a constraint, a rule no plan may violate.

Inter-site transfer logistics

Then comes the cost the single-plant view never sees: moving goods between sites. When demand goes to the cheapest plant instead of the nearest one, goods have to travel, with a cost, a lead time and a carbon footprint. A plan that cuts production cost but ignores the freight it creates hands the savings to the carrier. Our model counts transport and production together, from the plant to the customer.

Inventory pooling

When each plant holds its own safety stock, buffer is duplicated across sites, each sized for that plant’s demand alone. Pooling brings these buffers together at one or a few shared sites. One pool, sized against the combined swings in demand, replaces several separate buffers. If a plant goes down, the pool also lets the network ship from elsewhere.

How much pooling saves, and when it doesn’t

The saving follows from simple statistics. When demand at the sites swings independently, the swings partly cancel out. The spread of the combined demand grows with the square root of the number of sites, not with the number itself.

Take four plants, each holding safety stock against a lead-time demand swing of 100 units. Kept separately, that’s four buffers, each sized for a swing of 100. Pooled, the combined swing is 200, so the buffer halves at the same service level. With nine sites, it falls to a third.

The catch is correlation. If your regions rise and fall together, the swings don’t cancel and the saving shrinks. Pooled stock is also stored further from some customers, which adds freight and lead time. Inventory pooling optimisation weighs both, so you see per-plant stock against pooled stock and choose.

Joint optimisation with MILP

Joint optimisation solves the allocation as one problem. The method is mixed-integer linear programming (MILP). It mixes yes-or-no choices, such as whether a plant makes a product at all, with quantities such as volumes. It minimises total network cost: production, plus transfers, plus inventory held. Network production optimisation weighs all three at once, where separate plants each see only their own piece.

Where it fits in the supply chain

Which plant makes how much is part of planning your network. This decision sets which plants run at full capacity and which have hours to spare. It also sets which transport routes the goods take. So it’s closely tied to supply chain network design. It comes before each plant’s own production schedule. The volume a site receives becomes the input to its scheduling.

Common questions.

Something else? Ask us directly.

Compare the full cost to the customer, not the cost at the plant gate. That’s production cost at each plant, plus freight to the customer or between sites, plus the stock each option needs. Then respect what each plant can make and how much. If you decide product by product, you easily miss that two products need the same machine at the same time. One model places all products at once and respects how much each plant can make.

No. Pooling means you hold the safety stock for several sites together in one place. That saves stock because swings at different sites partly cancel out. If demand at your sites rises and falls together, they don’t cancel, and the saving shrinks. Pooled stock is also stored further from some customers, which adds freight and lead time. So pooling pays most for products whose demand swings a lot and moves differently in each region.

Yes, or the plan will move volume to a cheap plant and hand the saving to the carrier. Transfers cost money, add lead time and use truck capacity. They’re often booked in a logistics budget that production planning never sees. Putting them in the same model as production cost is the only way to find the cheapest plan end to end.

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