FOR SENIOR LEADERS IN MANUFACTURING, DISTRIBUTION AND SERVICE OPERATIONS
Flow in job shops, plants and supply chains
The method Stratflow uses in mining began in factories. This page shows it where it started, in two short games you can run in a browser and in the roof-sheeting plant where Hendrik Lourens first put it to work.
SIMPLER WORK, STEADIER WORK
When everyone is busy and the work is still late
When work passes through a series of steps, each step depends on the one before it. Push every step to stay busy and the queues grow, priorities change by the hour, and supervisors spend the day chasing the jobs that are late.
Goldratt argued that the ultimate constraint in most organisations is management attention. A plant run this way spends that attention on chasing, expediting and re-planning. The implementations on this page are, at heart, about giving that attention back.
We hold, and Hendrik’s peer-reviewed article on the mining fatality plateau argues, that freed attention is tied to the behaviour of high reliability organisations. Weick and Sutcliffe describe five habits of those organisations: preoccupation with failure, reluctance to simplify, sensitivity to operations, commitment to resilience and deference to expertise. Each one needs leaders with time to notice and respond, and steadier flow gives them that time. We see it as a necessary step towards high reliability, though not a sufficient one.
The games below show the production side of the mechanism in settings any leader will recognise. Questions worth bringing to them:
-
What really limits what we deliver each week?
-
Which of our measures reward keeping people and machines busy over getting work finished?
-
What would we stop doing if we managed the one thing that sets the pace?
TWO GAMES
Same orders, different rules
Both games run two copies of the same operation side by side. They get the same orders and the same luck. Only the rules they run by are different.
FOR ANYONE WHO RUNS A JOB SHOP, A WORKSHOP OR A MAINTENANCE TEAM
Two Job Shops, One Order Book
Two identical shops start out keeping everyone busy. Every order goes to the floor the day it is booked, operators pick their own next job, and late jobs get chased. On day 21 Shop 2 changes its rules. It releases work at the pace of its bottleneck, works one priority list, and sends free hands to help at the bottleneck. Watch its floor thin out, then compare the two delivery charts.
Further rounds take the same idea into the situations managers ask about: a bottleneck that moves, machine trouble, a rush of orders, batching, starting only with a full kit, and work in the office.
Seven rounds and a debrief, more than 20 minutes. Round 1 is open to anyone; the later rounds need a session code from Hendrik.
FOR MANUFACTURERS AND DISTRIBUTORS WHO MAKE BOTH TO STOCK AND TO ORDER
Long Runs or Full Shelves?
This game is built on a real plant: Modek’s roof-sheeting operation in South Africa, which Hendrik ran. One extruder with no spare hours made nine profiles. Finance wanted long runs to hold down scrap and changeovers. Customers wanted the sheet on the shelf when they came for it. Both instructions made sense, and the plant could not follow both.
The game plays a year of orders through the plant as it was found and as it was changed, side by side. It then shows two scoreboards for the same year: the efficiency report and the bank statement.
Five rounds and a debrief, about 30 to 40 minutes. Round 1 is open to anyone; the later rounds need a session code from Hendrik. The mix of profiles, colours and regions comes from Modek’s own sales. The year on screen is simulated.
In the Modek game the efficiency report counts the changed plant’s standing hours and extra changeovers against it, and its cost report comes out about twice as high.
The bank statement shows the same change selling more and making about twice the net profit.
WHAT THE PLANT RECORDED
Modek, 2008 to 2012
The games are models. These two cases are what the plant recorded, from the implementation notes and the business’s own results.
CASE: POLYCARBONATE ROOF SHEETING, 2009
A competitor that delivered in two days
In early 2009 a competitor entered the market with sheet cut and delivered within two days of order. Modek first lifted the extruder from 200 to 300 kg an hour and halved the time for a profile change. It then spent the new capacity on a short, fixed production cycle with hours deliberately left free, instead of filling the warehouse.
-
Late orders: 10 to 15 at a time before, none in November 2009
-
Lead time on the slower-moving profiles: 8 to 12 weeks before, 4 weeks at most after
-
Stock thought necessary for good service: 225 t before, a target of 136 t after
-
Market share: 61 per cent before, about 67.5 per cent estimated after, with the competitor in the market
-
Yield: 92 per cent before, 99 per cent after
Modek kept no safety data on this change, so this case speaks to stability of supply only.
CASE: POLYESTER ROOF SHEETING
The unit a year from closure
Sales had fallen for five years in a row. Lead times ran six to ten weeks against customers who would wait less than two. Finance pressed for longer runs to cut scrap, sales pressed for shorter runs to cut lead times, and the planner was caught between them. Head office had concluded that the staff were the problem.
With two days a month on site, the work went into the measures and into a few leverage points chosen with the people on the floor. Supervisors, for example, were measured on keeping accurate set-up charts, and scrap fell once run-to-run consistency followed.
-
Lead times: six to ten weeks before, two weeks after, with no stockouts and almost no late deliveries
-
Margins: up 20 per cent in the first year and 66 per cent after three years, about 19 per cent a year compound
-
Sales volume: up 69 per cent over the three years
Within two years the staff had the highest morale in the company. The people were the same. The measures had changed.

Polyester (GRP) roof-sheeting unit at Modek: 12-month moving average of margin (left axis) and kilograms sold (right axis), 2007 to 2013. The red line marks the start of the intervention in early 2010.
The quickest lever at Modek was what people were measured on, with the same people doing the work.
THE METHOD
Four rules that change how work flows
Operations science has a law for this: variability is always paid for in stock, capacity or time. The four rules below choose where to pay it.
The measures follow the rules. Judge the operation on what it ships and when, and judge products on what they earn per hour of the bottleneck.
Machine utilisation and cost per unit at every step stop being the scorecard. That is usually the hardest part of the change, and the part that frees the most management time.
In practice the rules look like this.
1 Find what sets the pace
In most plants one machine, one team or one person limits what the whole operation can ship. Every other step has some spare capacity.
2 Release work at that pace
Work released before the bottleneck can take it only lengthens the queues. Hold it at the door and the floor clears.
3 Protect the bottleneck with a buffer
A small, deliberate amount of time or stock in front of the bottleneck absorbs the ordinary upsets. Green, amber and red tell everyone where help is needed today.
4 Replenish what sold
Where customers buy from the shelf, keep the main stock central, refill what was sold rather than what was forecast, and let branches pull what they need.
FOR LEADERSHIP NETWORKS AND MANAGEMENT TEAMS
A shared starting point for leaders from different industries
Each game takes 20 minutes or more and needs no data or preparation. Played by a mixed group, each leader tends to recognise their own operation in it: a workshop, a ward, a warehouse or a project office.
A facilitated session of about 45 minutes uses one game and finishes with questions like these. Hendrik can run it remotely.
-
What limits our output?
-
Which of our measures hide it?
-
What would we test first?
For the thinking behind both games, the three bottles video on our home page shows why a chain of busy steps delivers less than its parts suggest. It was made for mining, and the physics is the same in a factory.
EVIDENCE AND LIMITS
Simplified plants, real management questions
The games are teaching models with their assumptions disclosed on screen. They show the mechanism. They are not a forecast for any real operation.
The Modek figures are the plant’s own records from 2008 to 2012. They describe one business and are not a forecast for another. Modek kept no safety data on these changes, so we make no safety claim for them.
Hendrik Lourens, MBA, MSc, FAusIMM, COHSProf, ran the Modek turnaround before applying the same method in mining.
RELATED READING
On this site
Manufacturing differently (PDF)
Hendrik’s article in AMT magazine, February/March 2020, on management innovation and the roof-sheeting turnaround.
One mine, two scoreboards
The same lesson in a mine, with games for executives and site teams.
Capacity to be safe
Why spare capacity matters for safe work.
Breaching the mining fatalities plateau
Hendrik’s peer-reviewed article on management attention, stability and safety.
Critical chain and adaptive project management
The same rules applied to projects and shutdowns.
Modek figures are from the plant’s implementation notes and results, 2008 to 2012. Game results come from teaching models and are not forecasts for any operation.