Factory Physics – Why modern production continues to stall (and what science teaches us)

1. Everyone is doing Lean. So why are problems still occurring?

Walk into any factory and you will see:

• 5S posters and marked floors everywhere

• value stream maps on the wall

• daily kick-off meetings around whiteboards

• teams that have been ‘working with lean for 10 years’.

And yet you hear the same thing almost everywhere:

• ‘Our lead time is a lottery. Sometimes 5 days, sometimes 5 weeks.’

• ‘We structurally cannot exceed 90–95% delivery reliability.’

• ‘We are drowning in WIP, but every time we reduce WIP, everything comes to a standstill.’

• ‘Planners are constantly working overtime, and yet we are still lagging behind.’

If lean is so widespread, but these patterns persist, then the logical question is: what is missing in the foundation on which lean operates?

It is certainly not because lean is ‘bad’. Lean is and remains powerful.

Lean works excellent as long as the system operates within physically feasible limits. It is essential for discipline, stability and standardisation and provides a language and culture of continuous improvement that is indispensable. But lean is an improvement philosophy, while Factory Physics is the design framework that determines where and within what limits those improvements have an effect. Together they form a powerful combination: Lean + Factory Physics.

What is often missing is the physical framework: a quantitative description of how a factory behaves under variability, WIP and high occupancy. Lean is providing the implementation and culture while Factory Physics provides the preconditions and the system ceiling. Lean tells you what usually works, not why it works – and certainly not when it no longer works.

It's like learning to drive without knowing what braking distance is: you often get away with common sense... until it's wet, it gets dark or someone suddenly crosses the road.

Aha moment: Most factories manage symptoms (late deliveries, too much WIP) instead of causes (variability, utilisation rate, buffer location). Lean + Factory Physics together give you both the tools to tackle symptoms and the insight into why they arise — and within what limits you can solve them structurally.

2. The real enemy: variability as a dominant force

Variability is mentioned in almost every improvement process:

• malfunctioning

• fluctuating demand

• quality issues

• changeover times that take ‘about that long’

• people who don't work at exactly the same pace every day.

But usually it remains anecdotal: ‘sometimes it takes longer, sometimes it doesn't’.

In Factory Physics, variability is quantified using the variation coefficient (cv):

cv = standard deviation / average

For example: an average process time of 10min with a standard deviation of 10min gives cv = 1. That is highly variable, but typical in many production environments.

As soon as cv approaches 1, variability determines:

• how much WIP you need at a minimum

• how much extra capacity (reserve) you need

• how short and stable your lead time can be.

And then comes the inexorable buffering law: in a system with variability, you must buffer with one or more of three things:

• WIP (inventory)

• Extra capacity (people, machines, overtime)

• Lead time (waiting)

If you demand low inventories, high utilisation and short lead times at the same time, you are asking for something that is impossible in practice, unless you have no variation in your process.

In practice, there are often too many or too large buffers in the system, built up by historical variations and symptom control. On the other hand, this offers opportunities: by reducing variability, buffers can be reduced without compromising performance.

Micro-case: A machine manufacturer in Eastern Flanders was instructed by management to ‘halve WIP, shorten lead times and not hire any additional staff.’ The planning team worked on a new schedule for three months. The result: WIP fell by 40%, but lead times increased by 25% and delivery reliability crashed to 68%. The buffering law had been violated – and physics had been fighting back.

3. The S-curve of utilisation: why 90% ‘efficiency’ is strangling your factory

One of the biggest misconceptions in manufacturing is that the higher the utilisation, the better.

The mathematics of queues shows the opposite.

The extra waiting time caused by variability increases by approximately:

u / (1 − u) (u = utilisation)

Let's do the maths:

• u = 0.70 → u/(1−u) = 0.7 / 0.3 ≈ 2.3

• u = 0.80 → 0.8 / 0.2 = 4

• u = 0.90 → 0.9 / 0.1 = 9

• u = 0.95 → 0.95 / 0.05 = 19

Imagine a bottleneck with an average process time of 1 hour and cv ≈ 1:

• at 70% utilisation: waiting time ≈ 2–3h → total throughput time ≈ 3–4h

• at 80%: waiting time ≈ 4h → total lead time ≈ 5h

• at 90%: waiting time ≈ 9h → total lead time ≈ 10h

• at 95%: waiting time ≈ 19h → total lead time ≈ 20h

So you only add 10–25 percentage points to occupancy, but your lead time becomes two to five times longer.

Meanwhile, management sees a machine that is temporarily idle and concludes, “This is wasteful!” or sees a bottleneck at 82% and thinks, “We are leaving 18% of capacity unused”. So, some extra work is pushed through and every micro-disruption suddenly becomes a structural problem.

Aha moment: At 95% utilisation, your bottleneck is no longer a production machine – it is a traffic jam that sometimes also produces something. Waiting time dominates; process time becomes secondary. And the more complex your product mix and routings, the more variability you generate – and the faster that high utilisation becomes a problem.

4. Lean, Six Sigma, TOC... and what's still missing

Lean, Six Sigma, TPM and TOC all provide valuable pieces of the puzzle:

• Lean → waste reduction, standardisation, flow principles

• Six Sigma → process stability, quality improvement

• TPM → more reliable machines

• TOC → focus on bottlenecks and global optimisation

But none of these frameworks provides the complete mathematical picture of how WIP, lead time and throughput are related, how variability spreads, how buffer capacity is best positioned and what utilisation is ‘healthy’ for a bottleneck.

Factory Physics actually does.

It is not just another method, but the underlying theory that all these methods use.

5. Why you should care

Without factory physics:

• you continue to rely blindly on average utilisation and OEE

• you undertake lean projects whose impact you cannot predict

• you invest in machines for which you have no system business case, even though there are cheaper alternatives

• and planning remains a mixture of experience, Excel and stress.

With factory physics:

• you can estimate in advance what reducing WIP really does to lead time

• you know why 82% utilisation sometimes works better than 93%

• you can rationalise buffer stock and capacity

• you work with lean within physically realistic limits

• and you can pinpoint the real root cause of substandard performance.

The key message: you cannot manage a factory without understanding the ‘physics’ of your factory.

Lean is the philosophy of improvement, providing a few basic concepts and tools. Factory Physics, however, is the physics book that allows you to truly understand your process. Factory Physics is the science of lean.

In the sequel to this article, we explain that physics in a structured way, without turning it into an academic thesis.