Capacity planning and bottleneck analysis
Every operation has a bottleneck. Our model finds yours before it costs you missed orders and overtime.
- Free 30-minute call
- No commitment
- You own the code
In short: A bottleneck analysis finds the one resource that limits how much your whole operation can produce, often a machine, a team or a process step. Capacity planning then decides how you make better use of that bottleneck, keep it from standing still or expand it.
The problem
On Friday afternoon, three orders are late. Overtime is booked for the weekend, the second time this month. In production, one machine runs flat out while the next station waits for its parts. Everyone agrees there’s a bottleneck. Nobody agrees which one.
The bottleneck is the slowest resource, for example one machine. It caps what the whole operation can produce. If that machine handles 50 parts an hour, the plant ships at most 50 an hour. In a plant it might be one machine that almost every product has to pass through. When a power grid is being expanded, it can be the wait for permits. Every operation has one. When demand shifts to products that need a different machine, that machine becomes the bottleneck.
Signs you’re managing the wrong bottleneck
Overtime keeps growing, but output doesn’t. Queues pile up in front of different stations from week to week. New equipment didn’t raise throughput. You and your colleagues argue about which resource limits the plant, and nobody can settle it with numbers.
Why finding it during production is too late
Most operations find the bottleneck only once orders are late and overtime is booked. The tools are a whiteboard, a stopwatch and a manager who walks through production and looks for where work piles up. Relieve one resource and another becomes the bottleneck. Add capacity at the wrong station and you’ve bought a faster machine that still waits for parts from the bottleneck.
What changes
Today, you walk the floor with a stopwatch after the orders are already late. Our model names the bottleneck before you fix the plan. It works from your demand and from how much each station handles per hour. You see which resource limits output this month, and which one becomes the bottleneck in peak season. Then you decide where one more machine, shift or crew pays for itself.
Capacity planning optimisation: what you get
Bottleneck analysis looks at every capacity limit in your operation: how much each machine, crew or process step can handle. You learn which limit holds everything else back. And you learn it before you fix the plan. We build the capacity model on three things: how much each resource handles per hour, when it’s available and what your customers order. Then we test it against the demand scenarios you name. A scenario is one possible future, such as peak season or a new large customer. For each scenario, you see which resource is the bottleneck and which single investment would remove that bottleneck.
- A model written around your own capacity planning and bottleneck analysis rules.
- Results measured in the numbers you already track.
- It slots into how you already work, and you can check every answer it gives.
What should the bottleneck make first?
Most bottleneck analysis in manufacturing stops once the bottleneck is found. The harder question comes next: the bottleneck is full, so what should it work on? Cost accounting says push the product with the best margin, the most profit per unit. Throughput accounting says look at what each product earns per minute on the bottleneck. It belongs to the theory of constraints, which we explain below. There, throughput means the sales price minus the costs that come with every unit made, usually the raw materials.
Take two products. Product A earns 100 francs of throughput and needs 20 minutes on the bottleneck. Product B earns 60 francs and needs 6 minutes. Per unit, A looks better. Per bottleneck minute, A earns 5 francs and B earns 10.
So while the bottleneck is full, every hour you give to B earns twice what an hour on A does. Most plants have dozens of products, minimum quantities for key customers and more than one tight resource. That’s too much to work out by hand. Throughput optimisation software built on a model sets the product mix: how much of each product you make. It picks the mix that earns the most from the hours your bottleneck has.
How it works
We write your resources, the work they have to cover and your rules into one model. You get a plan that says which resource does what and when. It’s the plan that fits your goals best, such as the lowest cost or the shortest waiting time. Every rule is kept at the same time, and that’s the part a spreadsheet can’t do.
Theory of constraints, made computational
The core is the Theory of Constraints (TOC), a management method Eliyahu Goldratt set out in his 1984 book The Goal. In TOC language the bottleneck is called the constraint, and it sets the throughput of the whole system. An hour lost there is lost for everyone. An hour saved anywhere else only builds stock nobody can ship.
TOC works in five steps. Find the constraint. Use it fully. Make everything else follow its pace. Add capacity there. When the bottleneck moves elsewhere, start again. A capacity model does the first step with numbers. It records how fast each resource works, when it’s available and how much work waits in front of it. Then it calculates how the work must flow so that the plant ships as much as possible and no resource gets more than it can handle. The result shows which resource is the bottleneck. That’s theory of constraints software built on your own data.
Drum-buffer-rope is the TOC method for pacing work. The constraint is the drum, which sets the pace. The buffer is a small stock of work in front of the constraint. It keeps the constraint from running out of work. The rope is the rule for releasing new material. Material goes in only as fast as the constraint can process it, so no work piles up in front of it.
From bottleneck to investment decision
Utilisation is the share of its time a resource is busy. A machine that runs 6 of 8 hours is at 75 per cent. Utilisation by resource shows this for every machine, crew or permit office. You see overload and spare time side by side, where an average would hide both. Run several scenarios, and you see which resource becomes the bottleneck in which future. A cell, a group of machines that work together, may be the bottleneck at peak and fine in quiet weeks. A crew may only become the bottleneck when the second project starts.
Investment modelling then answers what a stopwatch can’t. Add one machine, one shift or one crew at the constraint. How much more does the whole system produce, and does that beat the cost? One model decides how much of each resource you need, the order of the process steps and the buffer sizes.
One model for the whole capacity system
Every resource’s rate and availability become hard limits. Throughput becomes the objective, the goal the model pushes as high as it can. Mixed-integer linear programming (MILP) is maths for yes-or-no choices mixed with quantities. One such choice is whether to add a machine or not. The model decides two things together: where to add capacity and how work flows through the process steps.
The same method drives production scheduling. It plans every step at the speed of the bottleneck. So the bottleneck never runs out of work and never gets more than it can handle. Machine capacity planning makes sure capacity exists before the schedule promises it. For every option you check, you see which resource binds and which single investment would release it.
Where it matters most
Manufacturing is the classic case for throughput optimisation. Often a single machine decides how much the whole plant can ship. If almost every part runs through the same milling machine and that machine is full, free hours on the other machines don’t help. The plant ships no more than that one machine can handle. The same goes for a welding line or a test bench. Energy and utility companies work with the same maths, only with larger assets. An expansion programme moves only as fast as its slowest step. That can be the authority that reviews every building permit one by one. Or a power line that may be switched off for the work on only a few weekends a year. You want to know that bottleneck before you schedule the construction work.
Common questions.
Something else? Ask us directly.
Not necessarily. A machine can look fully loaded because it gets large amounts of work at once, or because it’s busy with work nobody needs yet. The test is this: would one more hour on that machine raise what you ship? A capacity model answers that directly. It shows which machine limits the amount you ship.
Because you don’t make the same products in the same amounts every month. A group of machines that limits output in one month can have spare hours in the next. And when you improve one station, the next station becomes the bottleneck. The fifth step of the theory of constraints says exactly this: when the bottleneck moves, start again. Running the model per scenario shows in advance where it will go.
In front of the bottleneck. Measure its size in time: how many hours of work are waiting there. That should be enough to keep the bottleneck running through the usual disruptions, such as a short machine stop or missing material at the station before it. Too small and the bottleneck runs out of work. Too large and half-finished parts pile up in front of it, and stops upstream go unnoticed. A model can test buffer sizes against your own record of disruptions.
Either one only pays off at the bottleneck. A second shift gives the bottleneck more hours, and you pay wages for them. A new machine costs a lot of money once. After that, the next station may simply become the bottleneck. The model calculates both cases and shows how much more you would ship with each.
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”.
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