When calibrating large models, it takes lot of time to run. How can performance be improved?
One problem can arise when using a very large amount of field data. The number of field data snapshots used in Darwin Calibrator will affect the run time and speed of the model, also the steady state of model i.e. number of trials required to solve the model.
Here are some things to be checked when you have a large model with a large amount of field data:
Ensure that the model is well balanced (solves as fast as possible) even before running Calibrator. See: Troubleshooting Network Unbalanced or Cannot solve network hydraulic equations
Test which engine mode (based on 2.00.10 or 2.00.12) gives the fastest solve times. See: Engine Compatibility Mode and related Calculation options
Use as fast of a computer as possible (CPU clock speed, hard drive, memory. Note: multi-threading is current not used during calibration)
Be very prudent with the application of field data sets.
Include only good field data (cases where the system is being stressed / high head losses are occurring).
Aggregate field data where possible. For example rather than have one field data set at 12:02 AM and one at 12:03 AM, put field data together in one field data set).
Eliminate any field data sets that are insensitive to adjustment parameters.
Consider running an iterative optimization process. The iterative optimization process might be to optimize a small subset of field data, then use the generated result as a starting point to optimize another subset and so on.
Using Darwin Calibrator
Water Model Calibration Tips