Why Maximising Every Machine Can Reduce Factory Output

Keeping every machine busy sounds like great manufacturing management, it sounds like the best way to run, the most sensible.

If a machine, a team, is available for 40 hours a week, surely we should try to use as many of those 40 hours as possible.

If it can produce 1,000 parts a day, producing 1,000 must be better than producing 800, right?

And if every department does the same, the factory should become more productive? 

Except it doesn't necessarily work like that.

A factory is not a collection of independent machines. You have a manufacturing system.

Incoming raw materials feed your stock, which in turn feeds the lines, the teams, the machines. Machines and teams at the front of the process, feed those further on, until you reach the end process of pack and dispatch.   

But if every process produces whenever it has capacity, the time, regardless of what the next process step needs, you create queues, Work in Progress and even longer lead times.

That creates an important distinction:

Maximising the utilisation of every individual resource is not the same as optimising the performance of the factory.

In fact, trying to keep every machine busy can make orders take longer.

The aim is not to maximise every machine.

The aim is to maximise the flow of finished, saleable product through the factory.

Capacity in manufacturing is the maximum volume/speed of products a machine, team can produce in a set time using its current resources. We might talk about parts per day or bottles per hour for a single team or machine.

Utilisation measures how well we are using a resource (a machine, a team). 

We get to the end of the month and find we haven't shipped any more orders or certainly no where near as many as we thought. 

Capacity = what the machine, team could produce under defined conditions.

Utilisation = how much of an individual resource's available time you're using.

These first two are often measured on single machines, single teams.

Throughput = the rate at which the factory produces saleable output. Crucially, this is measuring what you can actually sell.  This is what most manufacturers want to increase.

Imagine a simple three-stage manufacturing process.

Machine A can produce 100 units an hour.

Machine B can process 60 units an hour.

Machine C can process 80 units an hour.

Machine Utilisation vs Factory Flow

If we measure Machine A on utilisation, the obvious thing to do is keep it running at 100 units an hour.

But the factory cannot turn those 100 units into 100 finished products every hour.

Machine B can only process 60.

So the additional 40 units don't increase factory output.

They become work in progress. They sit there waiting for machine B to come free.  Machine C could do 80/hr and yet only gets 60/hr. It takes a hit on it's numbers.

After one hour there are 40 waiting.

After five hours there could be 200. Now you've got to find somewhere to store the parts.

After a full shift there may be hundreds of parts sitting between the two processes.

Machine A looks highly productive.

The factory isn't. Walk your factory you'll see WIP it's normally an indicator of one machine finishing work faster than next one in the process can cope.  

It's a great indicator of where the bottleneck is.

And yes this can be done for teams and factories where parts are custom made, bespoke, made to order. 

This is one of the traps created by measuring manufacturing performance locally.

The first machine reports:

• high utilisation;
• good efficiency;
• lots of units produced.

But downstream we may now see:

• queues of work;
• increasing WIP;
• more material handling;
• longer manufacturing lead times;
• greater difficulty identifying priorities;
• space consumed by stock;
• more opportunity for damage and quality problems;
• people expediting supposedly urgent jobs through the queue.

Nothing about producing those additional parts, on A, has increased the rate at which the factory can produce finished goods.

It has simply produced work earlier than the rest of the system can use it.

This is why factory performance comes from flow, not busyness.

In the simple example above, Machine B is the constraint.

Ignoring downtime, product mix and other real-world complications, the system cannot sustainably produce more than Machine B can process.

Making Machine A (Or C)  faster does not change that.

This matters when businesses make improvement decisions.

If Machine B remains the constraint, making Machine A faster will not increase factory throughput.

You may spend money increasing local capacity while simply creating WIP faster.

Before investing, ask what actually constrains the whole system, ask:

What prevents the whole manufacturing system from producing more? Where is the bottleneck?

That question is usually much more valuable than:

How Do We Make All Machines faster? which is normally where improvement training starts.

Many factories have inherited measures that encourage local maximisation.

Production departments may be measured on:

• machine utilisation;
• labour utilisation;
• units per hour;
• efficiency against standard hours;
• departmental output.

None of those measures is inherently wrong.

The problem comes when achieving the measure becomes more important than the overall result.

A production manager may therefore continue making parts because stopping a machine appears inefficient.

But if those parts are not required yet, cannot move to the next operation or are being produced ahead of a more important order, the factory may simply be converting available capacity into inventory.

The resource is busy.

None of these measures is inherently wrong.

The problem starts when achieving the local measure becomes more important than improving the factory result.

Work in progress is frequently treated as something that simply exists in manufacturing.

It doesn't.

At least some WIP is the result of decisions about when work is released and how much is produced.

If one process consistently makes work faster than the next process can consume it, a queue forms.

That queue then increases the time between starting an order and finishing it.

This is one reason a factory can appear extremely busy while customers are still waiting for their orders.

There may be plenty of activity.

There may even be plenty of capacity in individual departments.

But work is spending too much time waiting rather than flowing.

Think about this simple idea

If your factory finishes 10 jobs a day and has 20 jobs in progress, there are roughly two days of work in the system.

If you speed up a single machine and WIP rises to 30 jobs but crucially throughput stays at 10 a day, you now have roughly three days of work in the system.

More WIP without more throughput means longer lead time.

Local Optimisation in Manufacturing

Sometimes, yes.

And this is one of the hardest ideas for some manufacturing businesses to accept.

Idle capacity is not automatically waste.

If a non-constrained resource has already produced everything the system currently requires from it, producing more simply to keep it busy may create unnecessary inventory.

The question should not be:

How do we keep this machine busy?

It should be:

What does the factory need from this machine?

There are, of course, good uses for spare capacity.

It might be used for:

• planned maintenance;
• training;
• improvement activity;
• producing genuinely required future demand where that is appropriate;
• trials and development;
• cleaning and workplace organisation.

But producing unnecessary work purely to generate a high utilisation percentage is not automatically productive.

We tend to measure labour and machines by activity. But the business earns its money from finished, saleable output.

If we make parts the factory does not currently need, we have converted cash, labour and material into inventory and may not recover that money for some time.

Optimising the factory means understanding how its resources work together.

That includes questions such as:

• What is actual customer demand?
• What production rate does that demand require?
• Where is the true production constraint?
• Where does work wait?
• In what sequence should work be produced?
• When should work be released?
• Which resources require spare capacity to protect flow?
• Which processes genuinely need improvement?

Tools such as TAKT time can help establish the production rate required by demand.

Push and pull production systems can influence when work should enter the system.

But neither should be applied blindly.

The correct approach depends on the type of manufacturing system being operated.

A high-volume production line, a machine shop producing batches and a manufacturer assembling complex bespoke equipment do not necessarily require the same controls.

Before selecting the solution, understand the factory you're actually running.

There is another problem with focusing heavily on local utilisation.

It can distract management from the resource that actually controls output.

If ten departments are all being pushed to maximise utilisation, management attention gets spread across all ten, even if only one actually controls factory output.

If one stage determines the flow of the factory, improving or protecting that stage may be worth considerably more than improving the other nine.

This is why bottleneck analysis should come before many conventional improvement projects.

If the constraint can produce another five units every day and those units become saleable finished goods, the improvement affects factory throughput.

If a non-constraint produces another five units which simply enter a queue, it doesn't.

The value of an improvement depends on what it changes in the whole system.

This seems contradictory until you separate process performance from system performance.

Suppose a changeover improvement allows a machine to run much larger campaigns.

Locally, that may look excellent.

The machine runs for longer between changeovers.

Utilisation rises.

Cost per unit may fall.

But larger batches may also:

• increase WIP;
• make other products wait longer;
• increase finished stock;
• reduce responsiveness to changing demand;
• lengthen overall lead time.

The process has become more efficient.

The factory may have become less responsive.

That doesn't mean the improvement is wrong.

It means its effect has to be judged against the whole manufacturing system, not only the process where the improvement took place.

Improvement should improve the result, not just the process.

There is one resource where high productive utilisation can be extremely important:

the constraint.

If a process genuinely limits the rate at which the factory can turn demand into finished output, lost productive time at that resource can mean lost factory output.

That makes questions such as these particularly important:

• Is the constraint waiting for material?
• Is it waiting for information?
• Is it processing the correct priority?
• Is it losing time to avoidable changeovers?
• Is quality failure consuming its capacity?
• Is work reaching it in the correct sequence?
• Is it being used for work that another resource could perform?

The objective is not low utilisation.

Nor is it maximum utilisation everywhere.

The objective is to use capacity where it creates the greatest value for the whole manufacturing system.

What metrics should I track to measure factory capacity improvement?

If you suspect your factory is busy but isn't producing what you expect, start with the system rather than the individual machine.

If you want to know whether the factory is actually improving, look beyond individual machine utilisation.

Useful measures include:

1. Finished-goods throughput – how much saleable output is the factory producing?
2. Constraint output – is the resource controlling the system producing more?
3. Manufacturing lead time – how long does work take from release to completion?
4. WIP – how much work is currently inside the system?
5. Queue or waiting time – where is work spending time waiting?
6. On-time delivery – is improved flow translating into better customer performance?

Local utilisation can still be useful. But it should not be allowed to override these whole-system measures.

The question is not simply “How busy were we?”

It is “Did more finished product flow through the factory?”

This gives you a much better insight for deciding what to improve.

The next time someone says:

"We need to get the utilisation of that machine up."

Ask another question first:

"If we produce more there, what happens next?"

Keeping every machine busy can feel like good manufacturing management.

But the factory is not paid for activity. It is paid for producing and delivering what customers need.

That means understanding the whole system, protecting the constraint, controlling when work is released and allowing spare capacity where it helps flow.

A resource can be highly utilised while the factory performs badly.

And a resource can occasionally be idle while the factory performs exactly as it should.

The goal is not maximum activity. You've been trying it for years.

The goal is to get the right work to flow through the factory layout at the rate the business needs.

A factory can contain dozens of apparent problems while only a small number are materially limiting output, lead time or delivery performance.

The Factory Performance Diagnostic Review looks across the complete manufacturing system to identify why performance is being lost and what should be fixed first.

Mark Greenhouse is a UK manufacturing consultant with more than 30 years' experience across manufacturing and operations, working with businesses to improve capacity, flow, delivery and manufacturing performance.

A fabrication business was keeping a laser cutting machine busy, the team were constantly asking for the latest orders to process and pushing to buy raw materials in. They were cutting metal and putting the parts onto pallets for orders that weren't due out for two months.

The teams in the fabrication and welding had enough work to process. The teams prepping the structures for painting and assembly were constantly under pressure to keep up. The assembly teams had orders they could start but not finish, they didn't have all the parts to finish.

By slowing the cutting team down, we reduced the amount of metal being bought and re-directed the team into helping with metal prep before the paint process. We also asked the fabrication and weld teams to prepare some parts themselves to increase the amount of work ready for paint.

Finally we made sure the sequence of work at the prep team matched the assembly team. 

Within weeks the output had risen by over 55% and orders started going out on time, instead of 4 weeks late.