Digital twins and Theory of Constraints (TOC) in Mining
Read time: ~5min
TOC is a philosophy that provides methodologies and ways of thinking that help miners to focus on production constraints, and to exploit those constraints to maximise throughput.
With a digital twin that utilises an always-on constraints model, users can identify and measure all of the constraints appearing and moving across value chains, and managers can prioritise changes to address those constraints, while continuously measuring and improving those changes.
What is TOC and how can it help us?
The Theory of Constraints (TOC) is a management philosophy that emphasises the identification and management of constraints that hinder a system's ability to achieve higher throughput. Throughput is the rate at which the system produces revenue, minus the variable costs of production. Throughput is not a replacement for other accounting measures, it is a focusing mechanism for ensuring that production processes and efficiency are maximised.
In mining, we use All In Sustaining Cost (AISC) as a primary measure of the cost structure and hence sustainability of an operation, particularly across fluctuating commodity prices and grades. AISC is comprehensive and critical to understanding the potential profitability of an operation.
TOC emphasizes the importance of understanding and leveraging the system's constraints to improve profitability. It's less about reducing costs and more about optimising the flow through the system to maximise throughput, and to iteratively elevate major constraints to lift overall system throughput. This makes TOC a very useful tool for production, engineering and improvement teams.
TOC is built on 3 fundamental principles that are very applicable to mining:
Every system has local constraints that reduce global performance;
Many constraints are created by local optima rules (maximising what’s best for one part of the system over the system as a whole, often inadvertently); and
Local optima are usually creating by limiting beliefs or assumptions that may not be true, or may be changeable, to improve global performance.
The essence of TOC is identifying the most significant constraints that limit system performance and then deciding on how to best manage or eliminate these constraints. TOC embodies three core questions: What to change? What to change it to? How to cause the change? These questions are instrumental in diagnosing and rectifying system bottlenecks.
Born and proven widely in manufacturing and projects, TOC has had only limited penetration into the mining industry. Unlike manufacturing, where processes are more controlled and predictable, mining takes highly uncertain inputs and still needs to produce multiple products to specification. The unpredictability of ore grades at the unit level, the impact of weather and geological issues on mobile plant, and the challenges of operating and maintaining complex systems of systems in remote areas, all add layers of complexity not typically encountered in manufacturing setups. Mining also has large buffers of raw or semi-processed material such as stockpiles, designed to separate functional areas and reduce the impact of unplanned downtime. The unintended consequence of these many buffers is to incentivise siloed decision making, and to spawn numerous local value chain constraints across any one site. The presence of so many constraints, and the limiting belief that “we can’t change the system”, often leads miners to dismiss TOC. Compounding this, the lack of a critical mass of managers and engineers versed in TOC techniques ensures that most TOC implementations are consultant driven, performed far in arrears and often at too high a level, and do not last long beyond the consultants tenure. As a senior business improvement manager said to me, “a fair summary of our TOC experience would be that process can be very powerful, but it’s not sticky”.
How does TOC really work?
Let's delve into the core of TOC - the Five Focusing Steps.
Identify the Constraint:
In a mining scenario, the constraint could be anything from a bottleneck in the ore extraction process to limited transportation capacity for moving mined material. There can be many local constraints, but only 1 global constraint, and constraints appear and move across any given day. Simultaneously, multiple local constraints can compound in seemingly unpredictable ways. Like so many things, the level and time periods that we measure impact our view of what constraints are visible and therefore impactful.
Example: The secondary crushing system might be the bottleneck when processing softer ore, resulting in lower than expected unit capacity and starving the fine ore feed bin, resulting in intermittent periods of no feed to the process plant and/or necessitating holding up dumping in the ROM bin. In harder rock, the bottleneck may be in flotation, and in transition ores, it could move in unexpected ways.
Example: Unplanned downtime at an excavator holds up loading, and it takes 15min to diagnose if loading can continue, so trucks go into holding patterns, the ROM bin is drawn empty, and the fine ore bin level begins to fall rapidly.
Exploit the Constraint:
Before investing in costly upgrades, it's prudent to ensure that the identified constraint is utilised to its fullest potential and that the issue isn’t up or downstream process or capacity.
Example: Adjusting the production schedule to account for lower unit capacity on certain days, so that the secondary crushing is always at maximum capacity, and measuring what the “true” capacity is across different ore types.
Subordinate Everything Else to the Constraint:
All other processes should be adjusted to support the operation of the constraint.
Example: Ensuring that mining operations are synchronised with planned downtime of the secondary crusher when campaigning soft ore, to maximise availability of secondary crushing.
Elevate the Constraint:
If after exploiting the constraint, it continues to intermittently cause production losses, then it's time to consider investments to elevate its performance, usually by considering the entire value chain, not just the item of equipment or measured process.
Example: Experiment with different primary crushing settings or changes to upstream screening or scalping, and modifications or replacement of the secondary crusher.
Prevent Inertia:
After overcoming one constraint, it’s essential to avoid complacency and be on the lookout for the next constraint. The elevation of a consistent constraint will always create a new constraint elsewhere; every system has bottlenecks, and they move due to changes in the environment, both natural and artificial.
Example: Repeating steps 1-4 and be on the lookout for recurring or cascading constraints.
Other key aspects of TOC that are instrumental for miners include:
Throughput Accounting: This aspect emphasises measuring the rate at which the system generates revenue. In mining, unlike many other industries, this is the norm and we are seeking to maximise throughput or to maintain throughput when reducing costs. Where throughput accounting becomes interesting is when we start to compare actual to nameplate and actual to planned throughput. Some sites can normalise lower planned outputs based on historical performance, to maximise the likelihood of achieving targets, rather than maximising throughput.
Buffer Management: Given the unpredictability in mining operations, having buffers to accommodate for delays or unforeseen issues is crucial. The management of these buffers, and the analysis of what creates so called “buffer penetrations”, provides key insights into upstream and downstream value chains.
Continuous Improvement: The ethos of never-ending improvement is central to TOC and mining, urging managers to continually seek ways to better operational processes, with a focus on value chains (i.e. systems) not just individual items of equipment, and an emphasis on lifting the throughput, not just the capacity or availability, of the entire system.
Beyond these, there are other large and important TOC paradigms such as Critical Chain Project Management (CCPM), which is an entire discipline in its own right with tools and supporting processes. CCPM is also used in some mining organisations and on large capital projects and programs.
In summary, the application of TOC to mining operations and maintenance must consider:
Multiple constraints, at multiple levels, that move and can disappear and reappear at seemingly random intervals;
Multiple unit capacities, to account for multiple ore types;
The simultaneous and interdependent processes of production/faults and maintenance/shuts;
Constraints create the need for change, and changes (inc improvements) move constraints; and
Every system has constraints, and will always have constraints, and elevating one constraint creates (moves) the constraint to elsewhere.
The list is hardly exhaustive, and yet appears exhausting even to read. It’s no wonder we measure everything at the unit/equipment and area levels – the site/global is too abstracted, and the value chain is too complex and fast moving.
How does Geminum use TOC?
Like many complex problems, the answer is: it depends. Site, geology, complexity, commodity price, system configuration – are all factors that must be considered, and all of them are dynamic. However, like all complex problems, the solution needs to be simple enough for busy people to use every day, and it needs to be robust and configurable enough to allow for the uniqueness that every site has.
A focus on value chains and buffers shows the impact or lack of impact of local constraints on global performance.
The Geminum solution to this challenge is the ImproveTwin:
Capture, fuse and compare data from the many systems we use to manage our assets and work; and
Analyse this data in near real time and provide decision makers with the context and advice to make high quality decisions.
Capture & Compare
The first step is plumbing. Bringing data together is almost a mantra today. Rather than fusing “all the data”, we start with the data that is required to enable the constraints data model and make decisions, and work backwards to connect the data. We also often have discussions with customers who haven’t yet digitised all of their data; this may not be the impediment you think it is. We only need the data required to make better decisions, and we only need it at the frequency and fidelity that those decisions require. In other words, if a daily shift analysis is done weekly, we don’t need to ingest work order schedule adherence every minute. In the game of data fusion, the weakest data link is the constraint on the use case. Conversely, when higher fidelity data at higher frequency is available, improvements in business process may be unlocked e.g. automated mid-shift analysis to enhance SIC.
Analyse & Focus
The second step is where we create change. Aligning to management operating systems (MOS), the fused data is transformed by a constraints model to measure every loss of production in each local value chain (e.g. each drill-blast-load-haul-dump production chain), and to show which local value chain constraint is contributing to global production loss (site production losses). The model is structured to accommodate at least 5 levels, but can be extended to allow for more complex operations:
Group (Each site)
Global (Site)
Area (e.g. Mine/Pit, MHS, Process)
Local Value Chain (e.g. drill-blast-load-haul-dump)
Unit (e.g. drill rig, loader, haul truck, stockpile, crusher, conveyor, bin)
At every local value chain, we measure the production loss vs plan, and at the unit level the production vs capacity. Local value chain production losses that don’t directly impact global production show users where they can improve efficiency, or capacity, or coordination. Local value chain losses that directly impact production are global losses; every tonne lost is a tonne of production lost forever. A cumulative view of losses at every level, whether they be 10t locally every hour, or 1000t globally for a single unplanned downtime, provides support to extrapolate what the likely annual impact of these losses will be. Users often find that the small, repetitive losses are creating more impact than the highly visible faults, and that the focus on equipment faults can miss the recurring but more subtle nature of value chain losses (e.g. shift change overs going long).
The final benefit of an always-on constraints model is that users can measure the impact of deliberate change on throughput. Improvement projects that target throughput can be prioritised for “real” (i.e. global constraints) returns, and the impact of the improvement is continuously measured. Elimination of the constraint should create higher throughput, and “successful” improvements that do not deliver throughput increases can be revisited. Similarly, where a defined annual constraint is identified (i.e. 25k tonnes / annum), a budget and schedule can be traded off vs the likely production increase, to either maximise ROI of the improvement, or to maximise production benefits.
To create the best possible MOS context, we create role based dashboard tabs to give area and functional supervisors, superintendents and managers the critical information they need at the level and area most useful to them. All users can search and view any level of detail, but pre-configuring data to make it simple to make decisions is the best way to help frontline leaders to adopt and drive value. When all functional teams have the same view of the real situation, prioritisation and coordination is enabled.
Summary
In conclusion, TOC as a philosophy provides us with tools and ways of thinking that help to focus on production constraints, and to exploit those constraints to maximise throughput. With a digital twin that utilises an always-on constraints model, users can identify and measure all of the constraints appearing and moving across every value chain, and managers can prioritise changes to address those constraints, while continuously measuring the impact of those changes.