The comments about the cost of running multiple trials, and the lack of time that is available to do them is exactly the difference that I see in using computer models versus real-world models of experimentation. Another valuable difference is the ability to control the controlled variables to a greater degree. Also, there would a greater opportunity to change the independent variables to see how multiple variables alter the outcomes. However, with real lab experiments, you are better able to evaluate if there were factors that have not been considered when analyzing outcomes. In other words, with differing outcomes for each trial, you often need to go back and see if there were other environmental factors associated with those outcomes. This allows you to re-evaluate initial hypotheses and/or establish controls on those factors as well and obtain more accurate data. With simulations, you may end up unaware of all of the factors involved in an event as not every situation involves a closed system.