This article is part of the Ventilation Automation Pyramid series. Read the full overview here.
Level 4 changed what the ventilation system knows. The mine now tracks where every tagged machine is, and airflow responds to actual equipment presence rather than assumptions. Level 5 changes what the system does with that knowledge: instead of responding zone by zone to where equipment is, it optimizes airflow across the full network and anticipates demand before it arrives.
From Zone-Level Response to Network-Level Optimization
The defining shift at Level 5 is that the ventilation system stops managing zones and starts managing the network.
Underground ventilation is not a collection of independent circuits. Every fan speed change, every damper position, every regulator adjustment sends pressure ripples through the entire airway network. A decision made on 650 Level affects what is available on 700 Level. At Levels 1 through 4, those interactions are either ignored or handled conservatively, which usually means running more air than necessary to ensure every zone stays above its minimum regardless of what is happening elsewhere.
A Level 5 system has a calibrated model of the full ventilation network: every drift, raise, crosscut, and vent raise represented as a branch with a resistance value; every fan as a pressure source; every regulator as a variable resistance. With that model and a real-time picture of where all the load is, it can solve for the fan operating points and damper positions that satisfy every minimum airflow requirement across the whole mine simultaneously, at the lowest total power draw. Not zone by zone, but as a single optimization problem.
The energy potential is real. Ventilation accounts for anywhere from 30% to 50% of total electrical consumption at a typical underground operation. The gains at Level 5 do not come from one dramatic intervention. They come from continuous marginal improvements: running a main fan at 70% speed instead of 100% for the hours when a production level is at partial occupancy. The relationship between fan speed and power is cubic: cut the speed by 30% and power drops by roughly 66%. Compounded across a year at a large operation, those marginal corrections add up to material reductions in OPEX.
The Production Plan as a Primary Input
Location data from the RTLS drives Level 4. At Level 5, a new input comes in above it: the production plan.
Underground mines run on predictable cycles. Drill-and-blast sequences follow the same pattern shift after shift. Mucking follows blasting. Ore hauling follows mucking. The timing varies, but the structure is stable enough that a system trained on historical operational data can forecast with reasonable confidence which zones will be occupied, by which equipment categories, thirty to sixty minutes from now.
A Level 5 system ingests that forecast, runs it through the network model, and pre-positions airflow before the load arrives. The fan ramps up before the LHD enters the stope. The regulator opens before the haul truck reaches the ramp intersection. When the shift changes and crews reposition to a new development heading, the system has already begun redirecting capacity.
The important distinction from Level 2 scheduling: this is not a fixed timetable. A pure schedule does not adapt when the blast is delayed two hours or when a machine breaks down and pulls out of rotation. A Level 5 system continuously reconciles the forecast against the RTLS ground truth. If the crews are not moving where the model predicted, the discrepancy is detected and the pre-positioning adjusts. The production plan is the primary input. Real-time location data remains the correction signal.
The Network Model Problem
You cannot run predictive network optimization without a calibrated ventilation model, and maintaining that model is one of the practical barriers that separates Level 5 from everything below it.
Underground mine networks change constantly. Development advances. Raises connect. Temporary stoppings get erected and removed. A model calibrated six months ago may no longer reflect the actual network, which means its predictions will be wrong. And wrong predictions driving automated actuation are worse than no predictions at all, because the system will be confidently moving air in the wrong direction while operators assume it is working correctly.
Keeping the model current requires sustained discipline: updating it when the network changes, and periodically running physical measurements to validate that the model still matches reality. Mines that have built this into their ventilation management program as a standard operating practice are the ones positioned to run Level 5. For operations where the model lives in a rarely-opened spreadsheet or primarily in the ventilation engineer’s head, the jump to Level 5 is a data quality and process problem before it is a technology problem.
What You Can Do With a Live Model
Once you have a calibrated network model running alongside real-time operational data, you can do more than optimize fan speeds. You can run simulations.
The most common use case we saw on projects at this maturity level was fire scenario modelling. A fire underground is the worst thing that can happen in a mine. The ventilation system’s response to it is not obvious: depending on where the fire is, the instinct to flood the area with fresh air can actually spread smoke and combustion products further into the active workings, putting crews in danger rather than protecting them. The correct response depends entirely on the location of the fire relative to the airway network, where personnel are, and which fans and regulators can be used to create a smoke-free evacuation corridor.
With a live digital twin of the ventilation network, mine engineers can simulate exactly that. They pick a fire location, run the scenario through the model, and see how the airflow responds under different fan and damper configurations before anyone is underground. The emergency response procedure for a fire on 700 Level is not the same as for a fire on 550 Level, and a well-configured Level 5 system makes it possible to work that out in advance rather than improvise under pressure. Some operations go further and pre-program the specific ventilation responses for their highest-risk scenarios directly into the control system, so that the response to a confirmed fire at a given location can be activated as a single action rather than a sequence of manual interventions.
What Level 5 Actually Requires Organizationally
Three things have to be true at once for Level 5 to work.
The production schedule has to flow into the ventilation management system in a structured, timely way. At most operations, the mine plan lives in a planning tool managed by the mining department. The ventilation system has historically been managed by a separate team with a separate toolset. Getting those two systems to exchange data reliably, shift after shift, requires integration work and ongoing discipline that crosses departmental boundaries. The technical connection is often straightforward. The organizational alignment is harder.
Someone has to own the network model. That responsibility needs to live somewhere in the org chart, with the time, access, and tools to keep the model current as the mine develops. Without that, the model drifts and the optimization starts producing results that do not match reality.
And the system has to earn trust from the people underground. Ventilation is safety-critical. Miners and shift supervisors have a legitimate expectation that the air will be there when they need it. When a fan is running at a fraction of its capacity or a damper has closed unexpectedly, someone underground will notice and will want an explanation. Clear HMI displays, audit trails for automated setpoint changes, and acknowledgment workflows for unusual system states are not optional features. They are what makes the system acceptable to the people working beneath it.
Why Almost Nobody Is Actually Running Level 5
Fully realized Level 5 deployments are still uncommon. The components exist, and the platforms that support predictive optimization and digital twin simulation are available from several vendors. But the gap between a system that has the capability and a mine that is running it reliably day-to-day, with integrated production data, a maintained network model, and the organizational habits to match, is significant.
What is more common in practice is a hybrid: strong Level 4 RTLS-based control, a ventilation engineer who periodically runs optimization scenarios manually in a modelling tool, and some degree of schedule-based pre-positioning layered on top. That is not Level 5 in the full sense, but it borrows the core ideas and captures a meaningful share of the value.
The path to genuine Level 5 typically runs through a few years of Level 4 maturity first. Operations that have used RTLS data long enough to understand their occupancy patterns, that have a calibrated model they actually maintain, and that have built the cross-departmental habits around production data integration are the ones that can make the jump without the implementation falling apart six months in.
Closing the Pyramid
This article completes the series. The progression across all five levels follows a consistent logic: each level adds either the actuation capability or the new input that the levels below it did not have.
Level 1 is the actuation layer: the infrastructure that makes any of the rest possible: fans that can be speed-controlled, dampers that can be positioned remotely, a system that can receive a command and act on it. Everything above Level 1 is a question of what drives those commands. Level 2 answers that with time: the schedule. Level 3 answers it with environmental signals: sensors that trigger responses when thresholds are crossed. Level 4 answers it with location: the mine knows where the diesel load is and responds to actual equipment presence rather than assumptions. Level 5 answers it with the production plan and a calibrated network model, making it possible to anticipate demand, optimize across the full mine, and run emergency scenario simulations that turn a good response plan into a tested one.
Each level solves a real problem, and not every operation needs to reach Level 5. But understanding where a system sits in this progression, and what the next level actually requires, is what makes it possible to plan a meaningful path forward rather than chase technology because a vendor made it sound straightforward.
If you are working through a VoD implementation or mapping your current system against this framework, get in touch. No commitment and no sales pitch.
Previous in the series: Level 4 – Location-Driven Demand Control.
- The full series:
- Overview: The Five Levels of Ventilation Automation in Mining
- Level 1: Remote Control
- Level 2: Scheduled and Time-Based Automation
- Level 3: Event-Triggered Ventilation Automation
- Level 4: Location-Driven Demand Control
- Level 5: Predictive Control and Network-Wide Optimization
Leave a Reply