Validation of the Pipe2008 Network Modeling Software

December 2008

There have been several improvements to the hydraulic modeling program from KYPIPE LLC (Formerly from University of Kentucky) in the past 4 years. The Pipe2008 program which was validated in this report is version 4.107f. Previously, a similar third-party validation was conducted with Pipe2000 version 2.101. The purpose of this report is to validate the Pipe2008 program by comparing the results from several different models generated by an already validated Pipe2000 program (predecessor of Pipe2008) with those generated by Pipe2008 on the same set of models. The report highlights the differences between the modeling approaches between the new Pipe2008 and the old Pipe2000 program and then proceeds with the validation exercise.

Major Differences:

  • The current version of the program allows for pump with no check valves while the previous version included a check valve on all pumps by default. Therefore, it is possible that there could be reverse flow though certain pumps if the check valve is NOT specified explicitly in the current version of the program. However, all models tested had flows in the positive direction (or zero flows where the pumps were closed) when analyzed with Pipe2008 and therefore no differences arising from this change were noted in the results.
  • The above change had resulted in convergence problems with certain in-house models tested by KYPIPE LLC using the previous convergence check algorithm. The current version of the program has a slightly modified convergence check algorithm and therefore there could be a slight variation in the final results when compared to the previous version. In addition, the number of trials required for a converged solution could be slightly higher with the current version because of the new convergence algorithm. However, the results with the current version are expected to be more accurate than the previous version.
  • The current version of the program allows for more than one pump (in parallel or series) at a location. Therefore, the pump data and results sections have additional columns of information in the current version.
  • The current version has an extra column in pipe results section. This column prints unit headloss values for each pipeline in the form of HL/1000 where HL is the frictional headloss (in feet or meter) in the pipeline excluding minor losses.
Minor Differences:
There have been several formatting changes to the report file generated by the new model.

  • Most of these are cosmetic in nature to improve the clarity of the report.
  • Certain changes were made to print more details (data and results) of the model.
Validation:
Third-party model data was provided (*.p2k files) from the previous validation of the Pipe2000 program along with the corresponding set of results files saved in Microsoft Word format for each of the models. The names of old p2k files were appended with “(Pipe2000)” so they may be distinguished easily with the Pipe2008 files. The old (Pipe2000) files were imported into the new Pipe2008 program, one at a time, and saved with “(Pipe2008)” appended to the file name. Hydraulic analysis was performed on the Pipe2008 files, and the report files generated were compared with the corresponding report files from the Pipe2000 program.
The comparisons were made by importing results files generated by Pipe2008 program into MS WORD and comparing them with the old results file (using “Compare and Merge Documents” feature under Tools Menu of MS Word). The merged files showing the differences between the two documents are the downloadable Results_compared.zip file. Though there appears be several differences between the two results from the same model, most of the differences are format-related or changes to headers etc. Numerical differences between the two results files are highlighted in yellow. As evident from these merged files, the numerical differences in pipe flowrates and node pressures between two sets of results are negligibly small (less than 0.01% on an average).
Conclusions
Based on rigorous comparison of results from several diverse network models on old (Pipe2000) and new (Pipe2008) network modeling software, it is evident that the differences in numerical results are negligibly small and are primarily due to the more stringent default accuracy and improved convergence checking algorithm used by the new program (Pipe2008). Therefore, it may be concluded that the Pipe2008 has no errors in the modeling approach or in the results reported when compared to the results from the already-validated Pipe2000 program. It is recommended that the Pipe2008 data files included on this page may now be used for validating future releases of the Pipe2000 program.