Skip to content

Production scheduling

A bad production schedule costs overtime, wasted changeovers and missed delivery dates. We plan all machines at once and meet your delivery dates at the lowest cost.

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

In short: A production scheduling optimisation decides which order runs on which machine, and when. It plans all machines together, so you meet delivery dates with as little overtime, changeover time and half-finished goods as possible.

The problem

On Monday at 9 am, the press on production line two stops. You open the spreadsheet and start moving jobs. A rush order arrived on Friday, and two machines are now waiting for setups. A setup means changing a machine over for the next part. By noon there’s a new sequence, built by hand. What it costs in changeovers, or which delivery date it just put at risk, nobody knows.

A production schedule is a chain of decisions, each one limiting the next. Get the chain wrong, and the cost turns up everywhere. Machines wait for setups, rush orders jump the queue and you pay overtime to catch up.

Signs your production schedule needs a model

You rebuild the sequence in Excel whenever a machine stops or a rush order lands. Machines wait for setups while orders queue elsewhere. You promise delivery dates before you check capacity. Half-finished orders pile up in front of the same machine. Each sign costs you overtime, capacity or a customer’s trust.

Why your ERP can’t sequence the shop floor

Your ERP holds orders, due dates, routings and bills of material. Routings list the steps each product takes through your machines, and bills of material list its parts. Your ERP records what was ordered, produced and booked. It doesn’t work out which order should run on which machine, and when. That sequence is hard to find: every order you move forward changes the setup time of the next one. That in turn decides whether later orders still meet their due dates.

Tools built for that job are called advanced planning and scheduling (APS). They plan with the materials, staff and machine time you actually have. Generic manufacturing scheduling software, APS included, is built for the average shop. It sequences orders by fixed rules, such as earliest due date first. It doesn't know what only your shop has: which changeovers take long, which tool exists only once, which machine can run only certain parts. So you rework its proposal by hand, and the capacity lost along the way stays invisible.

We build the model for your shop. Each of your rules becomes a constraint. The model computes the sequence that meets every constraint at the lowest cost in setup time and lateness. A fixed rule tries one sequence; the model compares millions. Before you decide, we show you on your own data how much setup time and lateness that saves in your shop.

What changes

Today, a machine stops and you spend the afternoon rebuilding the sequence by hand. Our model repairs the plan around the breakdown. It keeps the jobs already running and re-plans everything still ahead. You check the new sequence and release it, and the delivery dates you promised stay realistic.

Production scheduling optimisation: what you get

You check every plan against the numbers you already watch: on-time delivery, changeover hours, planning time and work in progress, the orders started but not yet finished. Another one is overall equipment effectiveness (OEE). It says for how much of the planned time a machine makes good parts at full speed. Before you promise an order, you can see whether it fits. When a machine fails or a rush order lands, you still work from a valid plan.

To a model, the special cases a spreadsheet can’t handle are just more constraints. The model reads your data from your ERP, computes the schedule and writes the plan back. Your ERP stays as it is. We test the model with your orders and in your worst weeks before anything goes live.

  • A model written around your own production scheduling 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

Every shift, job and resource goes into one model, and your rules go in as hard limits. You get the best schedule that keeps all of them. What “best” means is your call: lowest cost, the fairest split or the fewest late orders.

Capacity checks at order intake

The chain starts when orders arrive. The first question is whether you can make an order at all, with the machines, tooling and labour you have. A rough-cut check compares new orders with the total capacity of your key resources. You see which orders can’t fit before you promise the customer a date.

A finite-capacity check then builds the detailed plan against the hours each machine is available. Finite capacity means two things. The plan uses only the hours each machine is available. And no two jobs run on the same machine at the same time. Skip these checks, and confident promises turn into missed delivery dates.

Job sequencing and setup times

Then comes sequencing, the core of the problem: the order of jobs on each machine. It sets your makespan, the time until the last job is finished. It sets your tardiness, how far orders run past their due dates. And it decides how much of the day goes to setups instead of production.

What matters most here are sequence-dependent setup times (SDST): how long a changeover takes depends on what ran before. A switch between similar parts is often quick, and one between very different parts takes long. Sometimes the direction matters too: from A to B is quick, from B back to A takes long. So the sequence itself decides how much capacity you throw away.

Worked example: three parts, one press

Take one press and three orders for parts A, B and C. From A to B and from B to C, a short adjustment of 20 minutes is enough. Any step back needs a full tool change, say 90 minutes. Run A, B, C, and you lose 40 minutes. Run C, B, A, and you lose 180.

Now say the order for part C is due first. C, A, B costs 110 minutes of setup. Putting C last saves 70 minutes but makes that order late. Which plan is better depends on what an hour of press time is worth against a late delivery. A model prices both and picks, on every machine at once. You do the same sum in your head, for one machine, when there’s time.

You can batch similar jobs into a campaign: you run every order of one product family back to back, with hardly any changeover in between. That saves setup time. But an order from another family waits until the campaign ends and may miss its due date. The model sets how long each campaign runs, so you change over as little as possible and still meet your due dates.

Job shop, flow shop or batch

The layout of your shop floor decides the method. In a job shop, every order takes its own route through the machines. That’s the hardest case and the one with the most to gain, covered on our page on job shop scheduling. In a flow shop, all jobs pass through the same machines in the same order. Because that order is fixed, faster, specialised methods can plan it.

Where changeovers dominate, batch scheduling groups jobs that can run together or share a setup. Most factories blend all three, so the model follows your layout. Shop floor scheduling optimisation only wins back unused capacity when it matches how your shop floor is laid out.

Work in progress and the bottleneck

Sequencing also decides your work in progress (WIP): jobs started but not yet finished. That’s capital tied up on the floor. Parts queue in front of the bottleneck, the slowest step in production.

Some schedules release new work only as fast as the bottleneck can process it. Then few half-finished orders pile up. One that maximises each machine’s utilisation stacks half-finished orders in front of the bottleneck. The model weighs throughput, the finished quantity per day, against WIP. You decide how much capital may be tied up in half-finished orders.

Real-time rescheduling

No schedule stays valid for long. The schedule that was optimal at 6 am is wrong by 9 am. Real-time rescheduling re-solves over a rolling horizon, a planning window that moves forward with the clock. It locks in decisions already in motion and re-optimises everything still ahead. When a machine goes down or a rush order lands, the model re-sequences the work, so you don't spend the morning shuffling jobs in a spreadsheet. You see right away what the new sequence costs in changeover time and which due dates it puts at risk. You notice the difference most on days with one disruption after another. After each one you have a valid schedule again, and you decide from numbers, not gut feel.

Exact methods and metaheuristics

Where exact methods fit the problem size, we use MILP, short for mixed-integer linear programming. It writes the schedule as yes-or-no decisions, such as: does order A run before order B on this machine? Your rules link these decisions. You get a plan, plus proof of how close it comes to the best possible one. Where the problem is too large, we use metaheuristics: search methods that find very good plans fast, without that proof. Either way, you get a sequence that keeps every rule and that your shop floor can run.

Common questions.

Something else? Ask us directly.

You need what an advanced planning and scheduling (APS) system does: plan against the machines, materials and people you really have. Your ERP records orders and due dates, but it doesn’t search for the best sequence. Whether you buy a packaged APS or build a model depends on how unusual your shop floor is. If you override the tool every day, it doesn’t fit.

You need a setup matrix: how long it takes to switch from each product family to each other one. You don’t need it per article on day one. Most shops start with families and rough minutes from the people who do the changeovers. Then they compare plan and reality for a few weeks and sharpen the numbers where the two differ most.

We recommend re-optimising the schedule in two cases. First, on a regular rhythm, for example at the start of every shift. Second, right away when a bigger disruption hits, such as a machine failure or a rush order. Small deviations like a five-minute delay don’t trigger a new plan. The model doesn’t touch jobs already running, and usually the next few hours stay as they are. That way the people at the machines don’t get a new plan every hour. You decide what counts as a big disruption.

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