Agent-Based vs Discrete Event Modelling

Every underground schedule looks achievable on paper. The tonnes add up, the metres are there, and the fleet is sized to the task. Then the week runs. A truck holds at an intersection waiting for right of way. A loader sits idle because the bogging front is not ready. A jumbo finishes early and the next face is unavailable. None of these events are in the schedule, yet together they are the difference between making tonnes and having to explain why you did not.

Simulation can be used by mine planning teams to test whether a schedule will hold before the week starts. Two approaches are commonly used to simulate mining operations: discrete-event modelling and agent-based modelling. Both can represent queues, and both can use probability distributions to represent variability. The difference is where each approach models the interactions between equipment, and that determines what the model can see in a confined network with limited passing room and shared resources.

Discrete-event modelling is a mature technique, widely used for structured, sequential processes such as crushing and conveying circuits. Agent-based modelling takes a different starting point. It has been used in defence, logistics, epidemiology, and traffic engineering for decades to understand systems where individual behaviour drives collective outcomes. idoba.sim brings agent-based modelling to underground hard rock access mining through advanced underground mining simulation software, purpose-built for operational teams who plan, assess, and validate mine schedules.

What is Agent-based Modelling.

Agent-based modelling simulates the behaviour of individual agents, such as trucks, loaders, jumbos, and shotcrete sprayers, within a defined environment. Each agent follows its own rules for how it moves, waits, loads, and gives way. The agents interact with each other and with the road network, and the overall behaviour of the operation emerges from those interactions rather than being assumed upfront.

That is the important distinction. In a discrete-event approach, you define the sequence of steps and the resources that equipment uses, and congestion is modelled at the resources you specify. In an agent-based approach, you define the equipment and its rules, and throughput and congestion are outcomes of the interactions within the system.

When simulating an underground mine with agent-based modelling, you define the network, the fleet, the priorities, and the rules, then observe the emergent behaviours. When a truck slows due to congestion on a decline, that delay ripples through the cycle times of every vehicle behind it, as it would underground.

Why This Matters for Underground Mining.

Underground access mining is one of the most interaction-heavy environments in the industry. Single-lane declines. Intersection priority. Shared loading points. A development front and a production front competing for the same equipment. Passing bays that either exist where you need them or do not. These are not edge cases. They are the daily realities that determine whether a schedule is achievable.

This is where the choice of simulation approach matters. A discrete-event model captures the queues that form where equipment arrives at the resources the modeller has defined, such as a loader, a stockpile, or a single-lane segment. An agent-based model also captures what happens in transit, as machines meet, give way, and hold at intersections, which surfaces where bottlenecks or congestion may emerge across the network.

How Each Approach Models Congestion.

Both modelling techniques can represent congestion. The useful question is where each one models it, because that determines what the model can and cannot see.

How Discrete-event Modelling Handles Congestion.

Discrete-event modelling moves a system forward using timers. Take a truck that leaves the dump and heads for a stockpile to load. The model calculates how long the trip will take and starts a timer. When the timer finishes, the truck arrives. That arrival is the discrete event. The travel in between is not modelled in any other way; it is simply the time between two events.

Congestion is captured at the destination. If several timers finish at around the same time, several trucks arrive at the stockpile together, and the model can show them queueing for the loader.

Well-built discrete-event models can go further than this. Travel times can be calculated from Rimpull curves, grade, and speed limits, with probability distributions to represent variability. The road network can be divided into segments that act as shared resources, such as a single-lane section that only one truck can occupy at a time. Calculated positions can then be used to estimate where vehicles meet and to build a heatmap of where queues are likely to form. The more points that are defined on the network, the more of the picture the model can see.

What Agent-based Modelling Adds.

Agent-based modelling includes everything above and adds the interactions that happen during travel. Each machine is an agent with its own rules. A truck that comes up behind a slow-moving jumbo on a single-lane decline slows to match it. An empty truck holds at an intersection until priority allows it through. A vehicle backs into a passing bay to let another by. Each of these responses changes when that machine reaches its next point and so changes what every other machine does.

This is why the distinction holds up even when a discrete-event model is finely segmented. Three things separate the approaches in practice.

Defined in Advance or Identified in the Simulation.
In a discrete-event model, the modeller decides where interactions can occur by placing resources on the network. A conflict at a location that was not defined as a resource is not seen. In an agent-based model, an interaction occurs wherever agents meet.
Potential Meetings or Resolved Meetings.
Positions calculated from travel curves describe where each vehicle would be if it ran unimpeded. Overlapping positions flag a potential interaction, but on their own they do not resolve what happens next: who gives way, who waits, and how far the delay travels back through the fleet. Capturing that requires rules that make vehicles respond to each other, which is the behaviour agent-based models are built around.
How Much Detail is Needed.
A discrete-event model can approach agent-like behaviour by adding more resources and rules, but every added point is another assumption to build, check, and maintain, and it erodes the speed that makes discrete-event modelling attractive. An agent-based model starts from the equipment and the network, and the interactions arise from the rules.

Where Discrete-event Modelling Fits.

Discrete-event modelling is a mature and effective technique, and in the right setting it is the better choice. Because it advances by timers rather than simulating every metre of travel, it is computationally very light and can cover long time periods quickly. Where mid-travel interactions are rare, such as two-way roads, rail, or a fixed material-handling chain from crusher to conveyor, that speed comes at little cost to accuracy.

Underground access mining is a different environment. Single-lane roads, slow-moving machines, intersection priority, and limited passing room mean mid-travel interactions are often the main constraint on the schedule rather than an exception. A model that treats travel as a timer is missing a large part of that picture, which is why idoba.sim uses agent-based modelling for underground mining simulation.

How Does idoba.sim Use Agent-based Modelling?

idoba.sim uses agent-based modelling to replicate how equipment and activities interact underground before a schedule is committed. It covers development advance, ground support activities such as bolting, meshing, and shotcrete, as well as truck loading and haulage. It runs across a one to fourteen shift horizon, helping planning teams identify where execution risk exists within their weekly schedules.

For each scenario, idoba.sim runs 25 statistically varied simulations and produces a distribution of outcomes expressed as P10, P50, and P90. The P10 represents a credible downside outcome, the P50 represents the median outcome, and the P90 represents a more optimistic outcome

Instead of relying on a single number that assumes certainty, teams receive a realistic range of possible outcomes. This helps identify how much confidence a plan has before it becomes difficult to achieve. The median scenario is available in detail for review, including 3D playback of equipment movement through the network, so mine planners can see where conflicts are likely to occur and how they develop.

Teams can compare what-if scenarios, test fleet sizing, and equipment allocation decisions, and understand how changes to activities or sequencing may influence outcomes before committing equipment, crews, or production plans underground.

idoba.sim runs in the browser, is cloud-based, requires no installation, and is designed for planning teams rather than simulation specialists.

Why This Means for Technical Services.

For a Technical Services Manager, the value is a schedule you can stand behind. When presenting a plan to leadership, you are not relying on a single optimistic number. You are presenting a probable range with visibility into where risk sits within the schedule.

When sizing a fleet or validating a development timeline, idoba.sim can be used to test the design against equipment interactions that occur in transit, as well as at loading points. In one case study, an underground hard rock operation used idoba.sim to model a dedicated stockpile before committing to it. In a single-shift comparison across 25 statistically varied runs per scenario, the simulation showed a 12 percent increase in ore to surface and a 39 percent increase in loader utilisation.

The simulation examined total material moved, ore tonnes delivered to surface, truck productivity, and shift transition performance across the P10, P50, and P90 range. The planning team had simulated data to support the decision before a single tonne of ore was moved.

What It Means for Mine Planning Engineers.

For a short-term planner, the value is fewer mid-week surprises. You can use idoba.sim to identify congestion and interaction conflicts that form in transit, not only those at a loading point or other defined resource, before they impact production.

You can test whether re-sequencing changes the outcome or simply moves the constraint elsewhere. When the schedule is finalised and crews are committed, you can do so with a clearer understanding of the range of outcomes it is likely to deliver, rather than relying only on the number that looked achievable at the start of the week.

The Short Version.

Discrete-event modelling captures the queues that form when equipment arrives at shared resources at the same time, and does so efficiently over long time periods. Agent-based modelling also models how equipment interacts on the way, which is where much underground congestion begins.

idoba.sim brings agent-based modelling to underground hard rock mining through advanced simulation technology, providing mining teams with a quantified basis for confident planning decisions.

Next Steps.

The weekly schedule that looks achievable on paper still needs to hold up underground. Book a discovery call and our team will talk through how your underground operation runs and show you how idoba.sim provides a realistic range of outcomes before you commit equipment, crews, or production plans.

Frequently Asked Questions

Agent-based modelling in underground mining simulates individual pieces of equipment, such as trucks, loaders, jumbos, and shotcrete sprayers, as independent agents that follow their own rules and interact with each other and the underground environment. System-level outcomes, such as cycle times, equipment interactions, and production outcomes, emerge from those behaviours rather than being assumed. This makes agent-based modelling well suited to complex underground mining environments where congestion, shared networks, and equipment interactions influence the schedule.
Discrete-event modelling advances a system using timers. When a truck leaves one point for another, the model calculates the travel time and the truck arrives when the timer ends, so congestion is modelled where equipment arrives at shared resources such as a loader or stockpile. Agent-based modelling also simulates the interactions that happen during travel, such as machines giving way, holding at intersections, or following a slower vehicle. Discrete-event modelling is computationally light and well suited to processes with few mid-travel interactions. Agent-based modelling is better suited to underground access mining, where single-lane roads and slow-moving machines make those interactions a major influence on the schedule.
Yes, at the points the modeller defines. Discrete-event simulation can model queues at shared resources such as loaders, stockpiles, and road segments, and detailed models can use travel curves and probability distributions to estimate where queues are likely to form. The limitation is that interactions are only captured at defined points, and travel between them is treated as a calculated duration. Agent-based modelling simulates the interactions themselves, wherever machines meet, which matters on single-lane underground networks.
idoba.sim is an underground mining simulation software platform that applies agent-based modelling to underground hard rock access mining. It replicates how equipment and activities interact underground across a one to fourteen shift horizon, runs 25 statistically varied simulations per scenario, and reports outcomes using P10, P50, and P90 ranges. This helps mining teams understand the realistic spread of possible outcomes before committing equipment, crews, or production plans. idoba.sim is cloud-based and designed for operational teams, planners, and technical services professionals rather than simulation specialists.
For a given scenario, idoba.sim runs 25 statistically varied simulations and reports the distribution of outcomes. P10 represents a credible downside outcome, P50 represents the median outcome, and P90 represents a more optimistic outcome. This range provides mining teams with a clearer understanding of potential outcomes and the confidence level of a plan, rather than relying on a single assumed result.
No. idoba.sim is purpose-built for underground hard rock access mining, where equipment interactions, shared networks, and operational constraints have a significant influence on achievable schedules.
No. idoba.sim is built for operational mining teams, planners, and technical services professionals, not simulation specialists. Users can model scenarios, review outcomes, and understand operational constraints without requiring specialist simulation knowledge.
idoba.sim supports underground mine scheduling by enabling engineers to model equipment interactions, test scenarios, and understand realistic ranges of outcomes before committing a schedule. Through agent-based simulation, teams can evaluate factors such as fleet sizing, equipment allocation, sequencing, and operational constraints within the mining environment.
Underground mining simulation software uses digital models to replicate mining operations, equipment behaviour, and operational constraints. idoba.sim uses agent-based modelling to simulate how individual pieces of equipment interact underground, helping mining teams evaluate scenarios, understand potential outcomes, and make more informed decisions.

Agent-based vs Discrete-event Modelling: Understanding Mining Simulation Approaches.