As ventilation engineers, how many times have we asked ourselves how reliable is our model? Especially on a new project where we have no way to compare it to measurements! Or for example, why does our model sometimes seem not to be able to replicate the conditions in some parts of the mine?
Since version 5.1, VentSim has a sensitivity analysis tool. In this short article I share a strategy that will help us answer these and other questions related to the reliability of our ventilation model.
What is sensitivity analysis?
As the name implies, sensitivity analysis allows us to evaluate how a model responds when there are changes in the model’s parameters. There are several fairly technical definitions, however in this article I will limit myself to explaining this concept with a couple of relatively simple examples taken from the article that Craig (the creator of VentSim) presented at the mine ventilation symposium in Canada last year.
Sensitivity to resistance
Remembering our ventilation concepts:
P = R*Q^2
P = Pressure (Pa)
R = Resistance (N*s^2/m^8)
Q = Flow rate (m^3/s)
In this case, the pressure P depends linearly on R and quadratically on Q. In a certain way, the pressure is more sensitive to changes in Q than in P. Seen in another way, if the resistance of one of the tunnels varies by 10% while keeping the pressure fixed (on account of a fan for example) the flow in that tunnel would vary by about 3% (Anyone care to do the calculation?). For a single tunnel this variation may not seem too much but if this same variation is applied to multiple tunnels sharing the same pressure source, the error in the flow will accumulate and may in some cases change direction!
To estimate the effect of these variations, VentSim’s sensitivity module automatically runs hundreds of simulations “disturbing” the resistance of the mine tunnels at random during each of these simulations to subsequently calculate the confidence margin of the model. In summary, how sensitive is my model to variations in tunnel resistance?
The concept of sensitivity applies to any model. In the case of ventilation, two of the most common variables are: resistance and temperature.
Thermal sensitivity analysis can be a little more demanding and complicated. Those of us who have performed thermodynamic simulations in VentSim have experienced first-hand some convergence problems during these simulations especially in recirculation areas or areas with poor ventilation. In addition to this, thermal simulations can hardly be considered stationary and for this reason many more parameters need to be considered during the simulation.
As with resistance sensitivity, VentSim runs dozens of thermal simulations to determine the model’s confidence range for variations in flow and mine heat. In the end, thanks to the sensitivity analysis, we can see which areas of the mine have the greatest variation and therefore require our attention.
“Listen and you will forget, Observe and you will remember, Do and you will learn” Chinese Proverb
Until you see, you do not believe!
How to execute the sensitivity analysis? The following video shows us how:
Once the simulation is finished, we can see the results of the reliability of our model. In this case, any tunnel where reliability is less than 100% deserves our attention but obviously we will focus our efforts on the lower reliability tunnels.
Resistance Sensitivity VentSim – Isaeng
In this case, the red tunnels present the lowest reliability or the highest sensitivity (values of 98%), an intelligent decision would be to focus initially on increasing the reliability in those tunnels through different strategies such as increasing the frequency of the gauging, improving the measurements of the areas, calculating the resistances in more detail, among others.
As I mentioned before, it is also possible to perform the thermal sensitivity analysis of our model, but how do the results look?
Geen alternatieve tekst opgegeven voor deze afbeelding
In this case, and coincidentally, the highest thermal sensitivity occurs in the blue tunnels that show a rather low reliability index. What would be the strategy to improve reliability? I leave it to you to share it in the comments or write me directly to firstname.lastname@example.org!
As Craig mentions in his article, no matter how accurately or how well we have built our model, there will always be variations in some regions of the model. These variations can be
Poorly ventilated areas.
Areas far from the main ventilation circuits.
Variations in the physical or geometric parameters of the system
In summary, thanks to VentSim’s sensitivity analysis tool, we can take proactive actions to prevent problems with ventilation and to improve the quality of our model, in short, to stop crossing our fingers!
Remember that the sensitivity analysis is available in the latest version. You don’t have your VentSim up to date? What are you waiting for to update it and take full advantage of this and other improvements? Do you want to know more about the latest improvements or receive a maintenance quote? Write to us at email@example.com.
About the author: Nestor and his company are distributors of VentSim, PumpSim and Terramin. He also offers his training and consulting services in mine ventilation and pumping together with leading companies in Europe and South America.