We have a detailed system to track when results vary between different version of Source. The details below will help you work out why the results have changed between the last production release 3.7 and this one (4.1).

The main areas with results changing include calibration, constituents, ordering, ownership, Resource Assessment (Annual accounting and Continuous accounting) and wetlands.

• Calibration - we have improved the default parameters for SCE, so they are now calculated based on the number of parameters being optimised.
• Constituents - main changes in constituents is improving how very low flows are handled close to and including 0 . We also improved some of the models such as the decay model. Outputs have also been improved to be more consistent across both constituent routing methods (lumped and marker tracking). There have been changes to the mixing of lumped constituents in links.
• Ordering - A number of improvements to ordering have been made, including ordering priorities, to allow prioritisation of supply in some areas of the system.
• Ownership - We now allow ownership in Geographic models as well as Schematic models.
• Resource Assessment (Annual accounting and Continuous accounting) - a number of recorders have changed names for consistency.  Various changes have been made to Resource Assessment to reflect the different practices of the different jurisdictions.
• Storages - we have increased the validation on storages to prevent user errors in setting up dimensions and outlets.  
• Wetlands - we have changed the solver used to improve performance of the wetlands module.

Models that depend on thresholds to make decisions can end up with large differences. Generally when systems such as Continuous accounting are used, when there is a small change, threshold triggers can cause a very small change in results to snowball into large differences for some models. 

To test your project:

1. Create a regression test in release 3.7 (Analysis tools).
2. Run it in version 4.1 (/wiki/spaces/SD41/pages/25822617) to test for any changes. Note some recorder names may have changed, in which case you need to manually update the recorder names in the regression test csv files. Open the model, run normally and save the results to a .res.csv to see the new recorder names.
3. If the results have change dramatically and would like to narrow down what caused the changes, you can run your project in each beta or you can do a binary search if you wish to know the exact reasons why your results have changed.  The release notes for each Beta release identify what caused changes to results. If you would like help with your projects or would like to have your projects included for testing and you have Support & Maintenance, please contact support: support.ewater.org.au 

4.1 Regression Test Changes since 3.8.22beta