Britain’s grid now runs on a lot of weather. When the wind drops, a gas-fired power station has to make up the difference in minutes, so how quickly it can start decides how often it gets called on, and how much gas it burns getting there.
At one UK gas-fired power station, starting a unit took a little over 90 minutes. Nobody thought that was unreasonable; it was simply how long the plant took. Mike Douglas, a Technical Training Specialist at the station, chose it as his Black Belt project. His question was whether that number was a real physical limit or just an accepted one.
It was the second. Here is what he achieved, and the route he took to get there.
The Black Belt project results: 20 minutes off every start
The goal was to save 15 minutes on every start. Measured on real starts through 2026, the plant is now saving around 20 minutes on two of its three most common start types, and almost 9 minutes on the third.
- £171,000 a year in gas savings, calculated from the gas saved per start applied to the number of starts the station made in 2025. That is an annual run rate rather than money already banked, and it counts gas only.
- The gains are holding. All three improvements are statistically significant across 30 starts to date, not a one-off good result.
- No capital expenditure. The change was made in the plant’s control settings, not its hardware.
- No trade-offs. Emissions per start remain within their normal range, and the change was made without touching any safety limit.
- A commercial upside on top, deliberately left out of the numbers: a station that starts faster is more attractive to National Grid, which should mean more running hours. The team judged it too difficult to quantify reliably, so it forms no part of the £171,000.
The problem: why a power station start-up took over 90 minutes
A start-up is a controlled warm-up. Several tonnes of metal have to reach temperature before a unit can safely take load. Heat it too fast and you shorten the life of expensive components; too slowly and you burn gas going nowhere.
Control software manages that trade-off automatically, holding the warm-up back to keep the strain on the metal inside safe limits. What the project found was that the software was leaving far more margin beneath those limits than the equipment needed. The constraint was not the metal. It was a cautious setting that had never been re-examined.
That is a problem shape you will recognise whatever industry you work in.
How the project ran: the five DMAIC phases, step by step
Mike used DMAIC, the standard Lean Six Sigma improvement cycle. In outline:
Define: agree what you are actually measuring.
“Start-up time” means nothing until everyone agrees where the clock starts and stops, so the team fixed it precisely: from the moment gas is first burnt, to the moment the unit is running steadily at its normal minimum output. They narrowed the scope to the station’s three most common start types, sorted by how long the unit had been switched off, and set a target of 15 minutes.
Measure: check the data before you trust it.
The plant’s control system timestamps every event to the second, so the numbers were sound. Handwritten operator logs were less consistent, with different people recording different times for the same event, so those were taken out of the equation. Only then did the team establish a baseline for each start type.
Analyse: find what is really driving the delay.
The team sat down together and came up with three possible explanations. They could take a safety step out of the sequence, improve a piece of hardware, or change the control software. The data pointed clearly at the third. Mike followed the trail from steam temperature, to the strain that temperature put on the turbine, to the software limit that was holding the warm-up back. What it showed was that the limit was setting the start-up time, not the equipment.
Improve: pick the option deliberately, then test it.
The three options were scored against criteria the team had agreed in advance: impact, cost, complexity, risk and long-term benefit. Changing the software won, mostly because the other two needed real money for less return. Before anyone touched anything, the team worked through what could go wrong and how each risk would be prevented. Then they updated the written procedure and trained the operators. Only after all of that did they run the pilot. It cut more than 20 minutes off two of the three start types and close to 8 minutes off the third. The team went back over those results against the risk assessment before agreeing, unanimously, to make the change permanent.
Control: make it stick.
The team now monitors the two factors that actually drive start-up time, not just the time itself. The new procedure was trained out so the faster start is the documented normal way of working. And the software change went through formal change management, so no operator can alter it or get it wrong.
Months later the gains were still there. The figures in the table below are what the plant is achieving now, not what it managed once under trial conditions. That is the part that turns a good pilot into a real improvement, and it is why the sponsor was able to sign the project off and hand it back to the operations team.
| Start type | Before | Now | Saved |
| Start type A | 84.3 mins | 64.2 mins | 20.1 mins |
| Start type B | 91.8 mins | 83.1 mins | 8.7 mins |
| Start type C | 108.5 mins | 89.1 mins | 19.4 mins |
Three lessons from this project you can apply in any industry
- Define the measure before you argue about the number.
Agreeing exactly where the clock started and stopped ended a debate the team had been having for years. - Check your measurement system before your baseline.
One of the two data sources here was not reliable enough to use. Had nobody checked, the improvement would have been lost in the noise. - Separate a safety limit from the margin around it.
The limit was never touched. What changed was how much unnecessary room was being left beneath it, and that margin turned out to be worth 20 minutes a start.
Mike’s coach on why this Black Belt project stood out
Dean Arnold, the Master Black Belt who coached Mike through the programme:
“From a coaching perspective, what impressed me most about this project was the way Mike and the team combined rigorous data analysis with deep operational knowledge. Rather than simply accepting long start-up times as an inherent constraint of the plant, they used the DMAIC methodology to identify what was really driving performance and then challenged it in a controlled and risk-based way. The result is a sustainable improvement that reduces start-up times, saves approximately £170,000 a year in gas costs and makes the station more competitive in an increasingly flexible energy market. It is an excellent example of Lean Six Sigma delivering measurable operational and financial value without compromising safety or asset integrity.”
Congratulations Mike
We would like to congratulate Mike Douglas on successfully completing his Black Belt certification with The Lean Six Sigma Company and delivering an outstanding improvement project for the business.
His project shows what becomes possible when rigorous data analysis, deep operational knowledge and careful risk management come together: a faster, more efficient start-up, delivered with no capital investment and without compromising safety or asset integrity.
We look forward to following Mike’s continued Lean Six Sigma journey and seeing the impact he creates in the years ahead.
Start your own Black Belt project
Every operation has a number nobody has questioned in years. Our Black Belt programme is built around finding yours and fixing it, on a live project in your own organisation with a Master Black Belt coaching you through it.
Frequently asked questions
1. Does Lean Six Sigma work in the energy sector?
Yes, and power generation is unusually well suited to it. Plants are instrumented to the second, so the Measure phase can draw on real process history instead of a manual study, and the operating decisions that matter are already governed by written procedures and control settings, which are things you can change and reverse. The obstacle in energy is rarely the data. It is the assumption that plant behaviour is fixed, which is precisely what the method is built to test. The same holds across utilities, oil and gas, water, and any asset-heavy, safety-critical operation.
2. What does a Black Belt project involve?
A live improvement project in your own organisation, run alongside your job over several months, with a Master Black Belt coaching you through each phase. It is the practical half of the qualification. You certify by delivering a real, measured result, not by passing a written exercise.
3. Do you need a Green Belt before taking a Black Belt?
No. Our courses are modular and have no fixed prerequisites, so Black Belt can be taken as a starting point. The distinction is scope rather than sequence: a Green Belt typically improves a process within their own department, while a Black Belt leads larger, cross-functional change. This project is a good example, spanning operations, engineering and management.
4. What kind of results can a Black Belt project deliver?
Around £100,000 or more in benefit is typical, though it depends entirely on what you point it at. High-volume or high-cost processes tend to return most, because a small per-cycle improvement multiplies. This plant starts up well over a hundred times a year, which is why 20 minutes turned into six figures.
