Questions about WaterCAD calibration of a large distribution system

Hello James,

 

You can import the SCADA data into the Darwin Calibrator directly, as shown in the below screenshot. You need to define the SCADA connection in the model first.

 

Please note that, in order to calibrate a model, Darwin Calibrator makes adjustments to the pipe roughness, demand, and/or element status, provided you have entered the demand and roughness data in the model first.

Please see the below technotes on Darwin Calibrator and SCADA Connect, which you may find useful. 

http://communities.bentley.com/products/hydraulics___hydrology/w/hydraulics_and_hydrology__wiki/5910.using-darwin-calibrator 

http://communities.bentley.com/products/hydraulics___hydrology/w/hydraulics_and_hydrology__wiki/24272.scadaconnect-simulator-for-watergems-v8i-selectseries-6 

James,

You don't want to have one pipe per roughness group. You want to group pipes with similar properties (e.g. age, material). The way that genetic algorithms search the solution space depends on the size of the solution space. If you have say 5 roughness values at 10 groups, the solution space is 5^10 = about 10 million possible solutions which Darwin can handle. When you have 5^1000 the solution space becomes larger than the number of atoms in the universe and the chance of hitting a good solution in our lifetime becomes tiny. You want a small number of groups.

More importantly, are you sure that the only source of error in you model is pipe roughness? Making adjustments to a parameter with Darwin is easy. Determining which parameters is needing adjustment and whether certain data points are valid is much harder and requires a lot of simulation runs before you think about optimization to get a feel for the sensitivity of the model to each parameter. Think of all the sources of error in a model: demands, demand patterns, pump curves, roughness, elevations, closed valves, connectivity, tank levels, inaccurate SCADA data, ... Roughness only becomes important when you are dealing with high velocity in the pipes. For most pipes, the velocity is so low that model results are insensitive to roughness.

If you send me your email (tom.walski@bentley.com), I can send you some references.

Tom

Sushma,

That is what I thought I should do but I wasn't sure. It looks like all I would need to do is convert my SCADA data from Excel sheets into an acceptable format to import into WaterCAD.

Thank you!

James M.

Tom,

Thank you for the help... Your words about the roughness groups helps me out a lot.

As far as the sources of error goes, let me give you a little more information. We are calibrating the model to try to determine the cause of some problems the client is having (water is bypassing a certain area.) The client said they checked for closed valves and they don't have any leaks or anything that would be caught be adding demand groups. Our plan was to only adjust the roughness values and look for any areas that have a high roughness value, then investigate further from there.

As far as other sources of error (outside of any adjustment groups) I can't say whether or not there is error in the model for demands, pumps, closed valves, etc... The water distribution system is poorly documented (about 80 years old) and there is no documentation from the group that originally created the WaterCAD model.

I will email you shortly for those references.

Thank you,

James M.