From bottlenecks to flow in integrated production systems
Every production system has a constraint, and it moves over time. How do we fix the operation instead of endlessly firefighting bottlenecks?
GMs will tell you they are hitting their targets, and most will have KPIs to prove it. But total throughput still sits 5 to 15% below nameplate, quarter after quarter.
The performance gap often lives in the handoffs between assets.
The pattern we see
Local optimisation wins at the expense of system throughput.
“Our mines might hit 110% of a weekly target by clearing waste early and producing ore later but starving the crusher on some shifts.”- Mine GM
Lean principles designed for tight-tolerance industrial manufacturing get applied to high-variability operations and “eliminate waste” slowly becomes “eliminate the buffers that protect flow”. Stockpile strategies designed for an earlier era still govern operations running at much higher throughput. Buffers sit in the wrong places and constraint starvation becomes a way of life.
“We are great at protecting the plan shift by shift, but we have lost the ability to recover the weekly plan for the losses that regularly happen.” – Mine Planning GM
Quarterly forecasts assume smooth production. Daily reality deals with breakdowns, weather, and geology. Without the freedom to run hard when conditions are good, each disruption becomes a permanent loss.
What it costs
I have seen these patterns play out from inside three recent engagements. At one large iron ore operation, this pattern ran for a decade. Nameplate capacity was never reached. The system bled over $1 billion a year in revenue across 12 to 15 million tonnes of lost production. Three failure modes drove it.
- Stockpile strategy had not scaled with throughput growth, starving the constraint of feed.
- Low-cost ROM rehandle moves that would have enabled high-value shipments were avoided.
- Over-specified blend requirements locked out lower-grade ore bodies.
At an aluminium smelter, the constraint moved three times in twelve months. It moved from the rodding furnace to the tapping cycle and then to the main furnaces. The team followed the constraint each time it moved. Utilisation lifted from 80% to 95% in one year.
Industry research shows that 80% of major capital projects blow their budget by 40% and their schedule by 20%. On one offshore platform construction project we worked on, digital twin simulation testing during design slowed the early stages but contributed to an overall CAPEX cut of roughly 30%. The facility hit schedule and reached 10.5 million hours without a lost time injury.
Why “fix the bottleneck” is not enough
Fixing one bottleneck shifts the constraint somewhere else. In a tightly coupled system, the constraint migrates. If you only solve locally, you chase it from one handoff to the next and never reach nameplate.
The mindset required is to protect the plan and recover it by adapting as conditions change. Treat disruptions as part of the operating rhythm and build recovery into every shift. Align the team around system throughput, not individual targets. Size and place buffers to protect the constraint.
This is where Value Stream Mapping meets Theory of Constraints. Both have been around for decades. The difference is applying them end to end. Pit to port. Well to refinery. Smelter pot to cast house.
See the whole chain, then act on the constraint
Applied end to end, the value stream map makes the constraint visible. It shows where material waits, where handoffs break, and where lead time accumulates. It also reveals where Lean has been misapplied, buffers stripped out in the name of waste reduction that were protecting throughput.
This is where Theory of Constraints and Lean work together rather than against each other. Lean maps the flow and eliminates genuine waste. Theory of Constraints identifies the bottleneck and protects it with strategic buffers. In a high-variability operation, both disciplines are needed. The map finds the problem. The buffers keep the system moving.
Argon uses 4D+R
Diagnose. Define. Design. Deliver. Reset.
We start by diagnosing the real constraint in the field and where to focus, define the target condition, design the buffers and controls, deliver the change with operations, and reset when the constraint moves.
Each pass tightens the system and gives operations more control over what actually drives throughput. Bring data, analytical tools, and operational experience. Map the current state on the ground. Not from a conference room. Walk the flow, stand and observe, time it yourself. When the constraint shifts, and it will, reset and run the cycle again.
“A day in the field is worth five in the office.” At one smelter, walking and timing the process end to end with the new GM uncovered performance gaps that no data system had surfaced. – David Rotor
From static maps to digital twins
A value stream map captures a snapshot. A digital twin lets you act in real time. But digitise bad logic and you don’t solve the problem.
At one lead smelter, management invested in a digital twin that failed. Our team found over 2,000 sensors, temperature, noise, vibration, flow, that had never been properly calibrated since installation. Once replaced or recalibrated, recoveries lifted from the mid 80s to the low 90s where they continue to sit. The twin did not fail because the technology was wrong. It failed because the fundamentals underneath it were broken.
Start with the manual VSM. Lock in good logic before you digitise it. Walk the process and get operator insights. When the foundation is right, the digital twin gives you what a quarterly mapping workshop never can. Continuous visibility, automated recovery triggers, and predictive intervention instead of reactive problem solving.
Where to start
Three entry points, depending on what the operation needs.
If a single bottleneck is losing value, start with the constraint. If coordination across the system is the problem, start with flow and VSM. If you are building new, start with both the VSM and the digital twin. Each builds on the discipline before it. You cannot digitise what you have not mapped, and you cannot map what you have not diagnosed.
From constraint to control
Our method works because it follows the physics of the system, not the politics of the organisation.
We have run operations and worked in the field at some of the largest mining, metals, and energy assets in the world. We also know how to see what operators miss when they live inside one system every day.
If the issues in this article sound familiar. We can help. reach out today
Share:
David Rotor
Partner, Australia
