Bug in v3.5
Several bugs have been found in the Monte Carlo feature in version 3.5. Improvements are expected in future versions.
Monte Carlo analysis is a feature in ResSim that enables you to evaluate the impacts on simulation results due to the uncertainty associated with certain input information in your reservoir model, as well as the subsequent uncertainty in those results. This feature uses random sampling of user-selected input variables within specified probability distributions.
Monte Carlo analysis is performed using a ResSim Alternative type called Monte Carlo. When you compute a Monte Carlo alternative, ResSim orchestrates iterative simulations of the alternative based on random sampling of one or more user-selected input variables. The following six types of variables can be selected for random sampling:
- Time Series Multipliers,
- Input Time Series,
- Reservoir Rule Parameters,
- Lookback Values,
- Rating Curves, and
- Scripts.
For each input variable, you can select from seven probability distribution types for defining the statistical distribution of each random input variable including: Normal, Log-Normal (natural logarithm and base 10), Gamma, Empirical, Triangular, Uniform, and Discrete distributions. Based on your input parameters for a selected distribution type, HEC-ResSim's Monte Carlo analysis computes the Probability Density Function (PDF) and Cumulative Distribution Function (CDF) for the random variables, and then randomly samples values from the CDF. Output variables and convergence criteria are specified for the Monte Carlo analysis and may use time-series summary values (Maximum, Minimum, Mean, or Volume). You can also specify options for defining the minimum and maximum number of simulation iterations, for continuing the analysis when additional iterations are desired to improve convergence, for restarting the analysis when previous simulation results are desired to be cleared, and for which iterations should save their full standard output.
