Emerson Exchange 365

• Not Answered

Use of .dat file to create a model in MPC function block

I'm trying to use a MPC (not MPCPro or MPCPlus, just the simple MPC function block), but I am running into a problem: the plant is reluctant to let me make a life-test with the MPC, so I decided to use a data file (.dat) to create the model: I followed the instructions to create and give format to the .dat file, but every time I'm trying to load the file I got an error message saying "This is not a valid DeltaV Model Predict data file." I honestly have no idea what could be wrong... Help pleas!!!

cmendozal

• 24 Apr 2017 1:45 PM
• Cancel

9 Replies

• Devin Marshman

  

  • 24 Apr 2017 2:17 PM

  Hi CMendozal. It's tough to say what's wrong with the .dat file without seeing it, but the load data feature does work if formatted correctly.
  
  What I can suggest is switching to a MPCPro block. With MPCPro in DeltaV PredictPro, you can create a blank model by right clicking "Models" and selecting "new". Make sure that you have set your time to steady state appropriately first. Then select options -> expert mode, and you can manually enter model parameters (gain, time constant, deadtime, second order time constant). I'm not aware of any advantage at all of using the vintage MPC block over the MPCPro, so you should be fine.

  • Cancel
• John Rezabek

  

  • 24 Apr 2017 2:50 PM

  Here's a suggestion . . .
  1.) Select your MPC block, right click and open Predict
  2.)Move the green bars to include some dynamics and cover a few time constants
  3.) Make sure you have selected "expert" mode under options
  4.) In the "File" menu, select "create model" (don't worry if you get an error message about checking the model)
  5.) Look in the \DvData\Predict\Data directory. Predict will have created a DAT file you can now use as a template for your data
  
  . . . the DAT file will have the time / date of the session when you created it.

  • Cancel
• Lou Heavner

  

  • 24 Apr 2017 3:19 PM

  In reply to Devin Marshman:

  I think MPCPro is generally the best solution. The supporting infrastructure is better. But not all problems are constraint optimization problems. If you are only trying to improve the control response for interacting loops or deadtime dominant loops, then MPC is just fine. There is really no advantage to MPC, except where you don't need the LP optimizer. Commissioning is easier without having to do the LP. But a small dimension problem that is possible in MPC (4 MV, 4 CV) is going to be relatively easy in MPCPro.

  • Cancel
• fairchdm

  

  • 24 Apr 2017 7:48 PM

  In reply to John Rezabek:

  I agree with John. I took 7201 Advanced Control Class last year and the documentation to import data was incorrect. I didn't look to see if Books Online were correct. The best way to make sure your format is correct is to do as John said. Save a "fake" *.dat file and import data into it.

  • Cancel
• cmendozal

  

  • 25 Apr 2017 1:03 PM

  In reply to Devin Marshman:

  Thank you very much for your suggestion, I'm not expecting a quote a MPCPro license.

  • Cancel
• cmendozal

  

  • 25 Apr 2017 1:04 PM

  In reply to John Rezabek:

  Thank you very much for your suggestion, I'll try it today!!

  • Cancel
• cmendozal

  

  • 25 Apr 2017 1:06 PM

  In reply to Lou Heavner:

  The process for which I need the MPC is a double plus a single effect evaporator, with significant deadtime, It has 1 controlled variable, 1 constraint, and 4 disturbances, so probably the MPC will be enough, anyways I'm expecting a quote for MPCPro. Thanks for your answer!

  • Cancel
• cmendozal

  

  • 25 Apr 2017 1:07 PM

  In reply to fairchdm:

  I'll try it today, thanks!

  • Cancel
• Lou Heavner

  

  • 25 Apr 2017 3:07 PM

  In reply to cmendozal:

  If you have only 1 MV, you don't need a license for MPC, MPCPro, or MPCPlus. If you do get a license for multiple MVs, they can be used on any of the 3 flavors. Evaporators are an excellent process for 1 MV controllers. Usually the feed is wild (DV) and the heat (steam) is the MV. You are probably controlling density or some product quality, but have worked on projects where the steam is wild (waste heat) and the feed is the MV.

  • Cancel