The Mantua Group

Simple Black and White Asset Management, Reliability Expertise, and Maintenance Execution Perfection.

Monte Carlo Simulation: Run Your Plant a Thousand Times

You cannot calculate the availability of a real plant with a formula. Real plants have redundancy, buffers, shared maintenance crews, spare parts with lead times, opportunistic maintenance, and failure distributions that interact in ways closed-form mathematics cannot follow.

Monte Carlo simulation is beautifully simple in concept: if you cannot solve the system analytically, sample it. Build the plant in software, its logic, failure models, maintenance tasks, resources and spares, then run it, drawing random failure times from the distributions. Then run it again with different random draws. And again, for hundreds or thousands of simulated lifetimes. Each run is one possible future for your plant. The distribution across all runs tells you what to expect and, just as importantly, how much variation to expect.

What the Simulation Actually Tracks

  • System and component availability, including which items dominate the downtime and how outage frequency distributes across the fleet.
  • Production losses, valued through the consequence structures described in our article on risk matrix modelling, including partial throughput states and buffer recovery.
  • Resources and spares: crew callouts, labour hours, spares draws, stockouts and the delay costs they cause, feeding the spares optimisation question directly.
  • Lifecycle cost, accumulated event by event across the mission, which becomes the boardroom currency covered in our lifecycle cost article.

The Model as Mirror

Here is the discipline simulation teaches. When the output produces a number that looks wrong, say an extruder down 6.8% of the time across 247 outages in 10,000 hours, that is not a software problem. That is the model holding up a mirror. Either your input assumptions are off, in which case you have found an error before it misled a decision, or your plant genuinely has that problem coming, in which case you have months of warning instead of a commissioning surprise. Both discoveries are valuable, and both are far cheaper to make in simulation than in operation.

Two practical disciplines follow. First, run enough simulations: single runs are anecdotes, and stable statistics need adequate run counts, which the software makes routine. Second, aim the model at a decision. Our white paper Backward Blueprinting argues that every analysis should be designed backward from the decision it must support, and simulation studies are the clearest case: the model scope, run length and outputs all follow from the question being asked.

How the Tools Help You Discover the Benefit

The AvSim availability simulation module of Availability Workbench runs Monte Carlo over RBDs and fault trees on a shared framework, with standard reports for predictions, component availability and production loss analysis that answer most questions out of the box, plus a SQL reporting layer for the rest. For process plants, the Process Reliability module adds production-focused analysis, and AWB Enterprise supports multi-analyst deployments with shared libraries. Our availability simulation services and training courses cover model construction, validation and the interpretive judgment that separates a forecast from a printout.

Go Deeper

How can we help? Talk to us at contact@mantua.group.