FOR COOs, VPs AND GENERAL MANAGERS
Lean, OEE and the tonnes between departments
Most improvement programmes in mining lift each department's utilisation and expect the mine to follow. In our experience the tonnes are lost between departments, when one stage runs short and the next has nothing in hand. The stock and spare capacity that prevent that loss are often the first things an efficiency programme removes.
SIGNS TO LOOK FOR
Signs the protection between departments has gone
From head office most of these look like good cost control. At site they show up as lost shifts and catch-up work.
-
Every department meets its utilisation target, and the mine still misses the month.
-
Inventory is reported as one figure for the whole operation, not as days of work in front of the bottleneck.
-
A wet month is followed by months of catch-up stripping or development.
-
Fleet or crews were parked in the last cost review because they looked underused.
-
Each department is benchmarked against the best result from another site or shaft.
-
The bottleneck moves from week to week between the fleet, the plant and development.
If three or more sound familiar, your mine is probably paying for its variability in lost time, because the stock and spare capacity that would absorb it have been cut.
CURRENT PRACTICE AND THE ALTERNATIVE
Two ways to improve a mine
Both approaches want more tonnes at a lower unit cost. They give different instructions on the calls a site makes every day. We think most programmes in mining follow the first, whatever they are called.
CURRENT PRACTICE
Raise every department
-
Benchmark each department and machine against the best site
-
Set utilisation or OEE targets near 90 per cent for all of them
-
Treat stock between stages as waste
-
Park fleet that looks underused
-
Watch one inventory figure for the whole operation
-
Spread capital across the areas that ask for it
THE BOTTLENECK-LED WAY
Protect the stage that sets the pace
-
Choose one bottleneck and hold it still
-
Size the buffers in front of it to the variability the site actually sees
-
Give the stages that feed it spare capacity to refill those buffers
-
Judge other departments by whether the bottleneck kept working
-
Watch the days of work in front of the bottleneck, every day
-
Spend capital last, and at the bottleneck
The first set of habits suits a factory, where the operator controls most of the variability. On a mine much of it comes from the orebody and the weather.
On a mine, the idle time around the bottleneck is what refills the buffers after a bad shift.
THE PATTERN
The buffer-cut cycle
Each step looks reasonable on its own. Trimming stock and fleet protects the period's cost, and the shortfall that follows is put down to the shock alone. In our experience the catch-up costs more than the buffer would have.
Public reports from Queensland coal producers describe rebuilding depleted in-pit inventory after wet seasons, in one case over several years. Underground, the same cycle shows up as development falling behind until the production crews run out of ready stopes.
Australian mining productivity, 2024–25
About 37% below its 2000–01 peak
Hours worked in mining, 2024–25
+5.6%
The ABS links the rise to restoring operational capacity after weather-related disruptions to coal, iron ore and gas.
Multifactor productivity derived by Stratflow from ABS growth rates. Source: ABS, Estimates of Industry Multifactor Productivity, 2024–25. The figures cannot show the cause; they do match the cycle.
WHERE THE PROTECTION SITS
A buffer only protects what is downstream of it
In this illustrative strip mine the dragline is the bottleneck. The buffers that protect output are the blasted and pre-stripped ground in front of it, kept full by drill and pre-strip fleets with capacity to spare. Uncovered coal and the ROM stockpile protect the coal fleet and the plant. They cannot keep a dragline working when drilling falls behind.
THE DAILY QUESTION
How many days of work does the bottleneck have in hand, and how fast can the stages behind it refill that after rain?
A single inventory figure for the whole operation can look healthy while the ground in front of the bottleneck is thin.
Illustrative. The dashed rope means drilling is released only as fast as the dragline can use the ground.
Underground, the same logic
The buffers are ready headings and broken stock in front of the production crews or the hoist, and the spare capacity sits in development and haulage. Choosing the bottleneck is a deliberate call. In a small, tightly confined orebody it is often the available headings. In a large one it is usually fixed infrastructure that cannot be expanded, such as the hoist or the mill.
A common version of the cut: a mine runs four jumbos at about 60 per cent utilisation. A review finds three could do the work at 80 per cent, so the fourth is parked to save cost. Then capital development falls behind, or backfill shuts a working front, and the three remaining jumbos have no room to catch up. A stoping front falls behind for good.
THE ARITHMETIC
Why efficiency targets remove the protection
1 Variability is always paid for
Operations science has a law for this. Hopp and Spearman's variability buffering law says variability is paid for with some mix of stock, spare capacity and time. Take out the stock and the spare capacity, and the mine pays in time: queues and lost shifts.
2 Queues grow fast near full utilisation
For a single machine, work waits about four times as long at 80 per cent utilisation as at 50 per cent, and nine times as long at 90 per cent. On a mine that queue looks like trucks waiting at a shovel, or headings waiting for the bolter.
3 Benchmarks ask for more than any site can hold
A case we use in our training: a gold producer measured five similar shafts against one another and made the best result in each department the standard for all five. Even if the shafts were truly alike, the best of five monthly results would sit well above their common average by chance alone. The standard asks every shaft for more than it can sustain, in every department at once.
4 Run it and see
In a simple five-stage teaching model, every stage averages 100 tonnes a shift. Each loses a whole shift one time in ten, and on working shifts its output varies between about 70 and 150 tonnes.
Illustrative teaching model. Buffers start half full and each figure is an average over at least 10,000 simulated shifts. The model shows the mechanism; it is not a forecast for any mine.
The 93-tonne design carries 20 per cent more capacity in four stages, and that spare capacity does the work. An efficiency review would still rate it worse. Its four other departments run at about 78 per cent utilisation, against 85 per cent in the balanced chain.
The design that produces the most looks the least efficient on a utilisation report.
See it for yourself in One mine, two scoreboards, a five-minute simulation.
WHAT THE NEW WAY DELIVERS
What changes when the bottleneck is protected
CASE: A NSW LONGWALL
The stockpile became the buffer
As longwall capacity grew, the bottleneck at this mine had moved to the coal handling and preparation plant. Pushing coal from the face into the plant produced stop-start flow.
From August 2023, alongside the start-up of a new longwall, the mine made the plant its managed bottleneck. It paced the face to what the plant could process, used the ROM stockpile as the buffer with clear triggers for when the longwall should run, and coordinated departments in a daily Flow Room looking 12 to 48 hours ahead.
CASE: AN AUSTRALIAN UNDERGROUND MINE
Cost cuts first, then one drill rig
The mine had produced at much the same rate for years. When commodity prices fell it ran a conventional programme: redundancies, a pay cut, lower supplier prices and truck utilisation benchmarked against peers. Production did not move. The trucks were never the problem; there were no more broken tonnes to haul.
A review of the life-of-mine schedule then showed that production drilling was holding the mine back. The mine hired one more production rig with operators, for day shifts only. Ore to surface rose about 20 per cent and held, quarter after quarter. Nobody was asked to work harder or faster, so the gain came almost at once.
Annual production, NSW longwall coal mine
+33%
From a 2.4 Mt average over 2018 to 2023 to 3.2 Mt in 2024, with no additional people and only minor capital spend.
More ore to surface, Australian underground mine
About 20%
From one extra production drill rig on day shifts, at about 1 per cent of operating cost.
Our historical range
10 to 40%
Often within 12 weeks. Context from past work, never a forecast for your mine.
FROM THE METHOD'S LINEAGE
Two buffers: one defended, one cut
These accounts come from talks by the late Arrie van Niekerk about work his teams did in southern African mines.
OPEN-CUT COAL
The loader that ran at 30 per cent
The team built a stockpile in front of the crusher and bought a large front-end loader to feed from it. The loader never reached 30 per cent utilisation, so the team made sure head office understood exactly why it was there. When heavy summer rain in 2008 stopped many coal mines in the region, this one kept loading and took every train it could. After that, Arrie said, head office stopped tampering with the arrangement.
UNDERGROUND GOLD
The shaft that rose by two-thirds, then fell back
A shaft facing closure built buffers of locos, hoppers, spares and maintenance skills, alongside changes to tramming and to how supervisors worked. Output rose from 9,500 to 16,000 square metres a month and held for two years. A new manager then closed the stores and cut crews back to the standard templates. The shaft fell to 7,000 square metres a month.
Protective capacity has to be written down and agreed with finance, or the next cost review will remove it.
WHERE LEAN, SIX SIGMA AND OEE FIT
The methods work in sequence
The rule of thumb is short. Where a site controls most of its variability, start with Lean. Where it cannot, start with the bottleneck, then use Lean and Six Sigma in the stable areas that creates.
Lean and TPS. Remove waste and standardise work where conditions are stable. On a mine, aim it first at the bottleneck: its cycle, changeovers and maintenance.
Theory of Constraints. Sets the structure: the bottleneck, its buffers and the measures that serve it. It decides where the mine absorbs the variability it cannot remove.
Six Sigma and OEE. Six Sigma suits recovery and reagent control in the plant. OEE is worth chasing hard at the bottleneck. Set for every machine, it rewards running whether or not the chain needs the output.
In practice that gives a sequence.
1 Set the structure first
Choose the bottleneck, size its buffers and agree the measures that serve it. Lean work can start alongside.
2 Aim Lean at the bottleneck
Standard work for its cycle, faster changeovers and maintenance planned around it raise output directly.
3 Use Lean and Six Sigma in the plant
Fixed equipment and steady feed suit them well.
4 Shrink buffers last
Only after variability has fallen, and in measured steps. In our experience, cutting a buffer first brings the pressure back to the crews within weeks.
BEFORE YOU SIGN ANY PROGRAMME
Questions to put on the table
These work whichever method a provider sells. A good answer to each is short and specific to your site.
-
Which stage sets this mine's output, and how will the programme protect it?
-
How many days of work does that stage have in hand today, and who watches that number daily?
-
What utilisation will each other department be asked to run at, and why?
-
Which stocks and which spare capacity will be cut, and what variability will they then stop absorbing?
-
How quickly could the mine refill its buffers after a wet month, and what would that cost?
-
How soon should we see a change at the bottleneck, and what is the stop rule if we do not?
-
Who in finance will understand why the protective capacity is there?
WORKING WITH STRATFLOW
A contained first look at one operating asset
The daily Flow Room makes this routine. Department heads meet for about 30 minutes each morning to look at the bottleneck and its buffers, and settle cross-department calls in the room. It runs beside your existing management operating system.
Through many interventions we have found that in the absence of a Flow Room, the gains fade. Our part is deliberately small. The site team does the work and keeps the result.
1
Check the gap
Run the Best-Day Diagnostic on six months of daily production. The calculation stays in your browser.
2
Talk it through
A 30-minute confidential call about the gap, the bottleneck and its buffers.
3
Five days on site
A discovery with your site team to find the bottleneck and see where its protection has gone.
4
Decide
A go or no-go decision before any wider commitment. If it is a go, the Flow Room starts beside your existing routines.
EVIDENCE AND LIMITS
Held to what it shows
The queueing and buffering results on this page are established operations science. The teaching model shows the mechanism with its assumptions stated; it is not a forecast for any mine. The site results are case evidence, not controlled trials, and our historical range is context, never a forecast. Production results are not safety results, and we report safety measures separately.
The method was developed by the late Arrie van Niekerk over more than 20 years and more than 90 mining interventions. Hendrik Lourens, FAusIMM, COHSProf and TOCICO certified, trained under him and has applied it in eight interventions over 12 years.
RELATED READING
On this site
Green departments, red mine
Why a sound management operating system still stalls, and the daily routine that sits beside it.
More capital, or a lower target?
How to point existing capital at the bottleneck, with a payback calculator.
Full Potential: when every cost centre improves and the mine loses
Why benchmark-driven savings can remove the capacity a mine needs.
One mine, two scoreboards
A five-minute simulation of a mine where every department reports green.
Australian longwall coal mine
Annual production up from a 2.4 Mt average to 3.2 Mt.
Figures marked illustrative are examples only, not forecasts for any site. Sources: Australian Bureau of Statistics, Estimates of Industry Multifactor Productivity 2024–25; Hopp and Spearman, Factory Physics; Kingman (1961) on queueing; Harter (1961) on order statistics.