Machine capacity planning
Many plants find out a machine is short of hours only when it falls behind. Capacity planning shows how much you can build before you promise it.
- Free 30-minute call
- No commitment
- You own the code
In short: A manufacturing capacity plan checks whether your machines, shifts and people can produce the orders you plan to accept, week by week or month by month. It shows early which machine reaches its limit first. Then you can add shifts, move work or buy in parts before promising dates.
The problem
Sales has a big order on the table and needs a delivery date by tonight. You open the schedule. Next month the plant as a whole still has enough free hours, so you say yes. But the big order has to go through the milling machine, and next month is already almost fully booked there. The free hours are on other machines. Three weeks later the parts reach the milling centre, and there isn’t enough time for every order planned on it. Two other orders have to wait and miss their delivery dates. You could have seen this on day one by checking free hours per machine instead of for the whole plant.
Signs you schedule before you check capacity
Delivery dates slip, and it’s the same machine every time. Sales promises dates the plant only hears about afterwards. You book overtime at the last minute. It goes to the order sales pushes hardest for. On another order, the same overtime would keep more delivery dates. Once the plant is full, you buy parts from suppliers straight away. You don’t check whether a different plan would let you make them yourself. Ask whether next quarter’s plan fits, and you’ll get a gut feeling.
Why scheduling first fails
Most plants build the schedule first. You only notice a capacity problem when a machine falls behind. By then the order of jobs is already fixed. Utilisation is the share of a machine’s hours that is booked. Many plants count every hour the machine could theoretically run, without taking off maintenance and changeovers. So a machine looks well used even though its orders don’t fit in the hours it has. The due dates built on that schedule were never possible. And you don’t work out what an extra shift would cost and what it would bring.
Checking this by hand is hard. Each part has a routing, the list of machine steps it goes through. Hundreds of parts need the same machines, and each machine can do only some of the steps. A spreadsheet can add up hours per month. It can’t tell you which order to move when two orders need the same machine in the same week.
What changes
Today, sales promises a date and the plant learns weeks later whether it fits. With our model, you check every order against the machine hours you actually have, before the promise. You see the week and the machine where hours run short, while there’s still time to add a shift or buy in parts.
Manufacturing capacity planning: what you get
You see how much the plant can build before you commit to a sequence, a shift plan or a delivery date. You see the bottleneck while you can still act. The bottleneck is the machine that limits how much the whole plant builds. You see the cost of overtime next to the cost of a missed delivery, so you choose shifts from numbers. Make-or-buy, whether you make a part yourself or buy it in, becomes a planned choice. You make it before the plant is full.
We build the model on your own machine calendars, routings and demand, and test it on your peak weeks first. You promise only what you can build, even when demand shifts or a machine goes down. In effect it becomes your capacity requirement planning software, built on your own data.
- A model written around your own machine capacity 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
Your orders and shift patterns go into one model. So do the hours each machine can run per week and each part’s routing, the machine steps it goes through. It checks every order against the hours you actually have before you promise a date. You see which machine runs short in which week. And you see the cheapest fix: an extra shift, overtime or parts you buy from a supplier instead of making them.
Rough-cut and detailed capacity planning
Machine capacity planning comes before the schedule. It asks one question before any sequence exists: does the plant have the machine hours for the demand? It works on two levels. A common mistake is to mix them. Rough-cut capacity planning (RCCP) starts from planned sales, not from single orders. It groups similar products into product families. Then it asks whether each machine group has roughly enough hours.
Capacity requirements planning (CRP) starts from each single order. It takes the routing of every order and turns it into hours per machine per week. Then it asks a sharper question: is there enough time on the right machines in the weeks when these orders arrive? Rough-cut planning shows early when a machine group lacks hours for months. Detailed planning confirms the parts fit each machine’s calendar. Merge the two, and you either check every single part at once or see only totals for the whole plant.
Measuring machine utilisation
The first number to get right is utilisation. People quote it often and rarely calculate it properly. It’s booked hours divided by available hours: 32 booked out of 40 available is 80%. Available hours are what remains after planned maintenance and changeovers. A changeover is the time you need to set up a machine for the next job. Then you take off a little more, because a machine doesn’t always run at full speed. Many plants divide by theoretical hours instead: every shift hour, as if the machine were never maintained or set up. Say a machine could run 50 hours, 40 are available and 44 are booked. Against 50 hours that’s 88%, which looks fine. Against the 40 available hours it’s 110%: four hours of work don’t fit.
Why a machine at 90% is already full
Capacity spreadsheets treat a machine at 95% as having 5% to spare. Queues don’t work that way. In 1961 John Kingman published a formula for the wait in front of a single machine. The wait grows with utilisation divided by one minus utilisation. At 80% that factor is 4, that is 0.8 divided by 0.2. At 90% it’s 9, and at 95% it’s 19.
So going from 80% to 95% load makes the queue nearly five times longer, with the same jobs. If orders arrive unevenly or some jobs take longer than others, it gets worse still. That’s why a machine can look fine on the monthly total and still deliver many orders late. We plan the bottleneck with headroom on purpose, and the model shows you what that headroom costs.
Bottleneck identification
The bottleneck is the machine group that reaches its limit first. Until you take load off it, the whole plant can’t produce more than the bottleneck does. That’s what bottleneck identification in manufacturing is for. Spot it in the capacity plan, and you still have time to add a shift or buy in parts. Meet it first in the schedule, and you’ve already promised a customer a date you can’t meet.
Shift patterns and make or buy
The shift pattern changes most how much your plant can build. It sets how many shifts run per day. Demand that one shift a day can’t cover may fit into two. So for each shift pattern, the model works out what it costs and how many machine hours it gives you. It shows the point where overtime becomes cheaper than paying for a missed delivery.
When the plant is full, which demand do you make and which do you buy? The internal cost includes the machine hours you take from other work. It also includes the opportunity cost at the bottleneck: what those hours would have earned on other orders. Say an hour on the bottleneck earns 200 francs on another order. If you make the part yourself, that order loses the hour. So add 200 francs per bottleneck hour to the cost of making the part. Leave that out, and making a part always looks cheaper than it is. The model treats parts you buy in as extra capacity you chose to buy.
One model, solved with MILP
We use mixed-integer linear programming (MILP). It writes the plan as equations with yes-or-no and whole-number decisions, such as adding a second shift. So the model checks capacity and assigns demand to machines in one calculation. It keeps to your machine calendars and to the hours each routing needs. It also meets your service level, the share of orders that must ship on time, for example 95 out of 100. You get the buildable plan with the lowest total cost. It shows which machine group is the bottleneck, which shift pattern to run and which parts to buy in.
Common questions.
Something else? Ask us directly.
Rough-cut capacity planning (RCCP) compares your planned sales with a few scarce resources, such as the bottleneck machines. It usually counts by product family, meaning groups of similar products. It’s quick and coarse. Capacity requirements planning takes every order’s routing, the list of machine steps, and turns it into hours per machine per week. It’s detailed but takes more work. Rough-cut shows you months ahead where capacity will run short. Use the detailed check for the weeks in which you’re about to promise delivery dates.
Fewer than you’d think. Include the machines that hit their limit in the past year. Add any crew with a skill only a few people have. Then add any supplier that limits your output. Everything else adds work without changing the answer. If a new bottleneck shows up in the detailed plan, add it to the rough-cut list.
Plan on what the machine actually delivers. OEE multiplies three shares: how much of the time the machine runs, how fast it runs and how many good parts it makes. For example, 90% × 95% × 98% gives about 84%. So stops and scrap are already taken out. Theoretical hours assume a machine that never stops. That’s why so many plans look doable and then aren’t. And leave headroom on the bottleneck, because queues grow sharply as load nears full.
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.
Similar pages.
Terms on this page
Every abbreviation explained in plain words, in our article “Metrics and methods”.
- Cycle service level and fill rate
- RCCPRough-cut capacity planning
- CRPCapacity requirements planning
- OEEOverall equipment effectiveness
- MILPMixed-integer linear programming
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