14 Replies Latest reply on Jun 9, 2014 2:46 PM by Jared Conway

    Centrifugal Pump simulation getting terminated

    Eljas Linna



      My company will start integrating SolidWorks Flow Simulation to the product development process and I will be the main user. We are simulation the flow channel and the impeller of a centrifugal pump to find shapes which cause inefficiency. I have so far been unable to get proper results in a single simulation and I suspect the problem lies in my setup and the chosen simulation settings. The calculation begins nicely and ticks for several hours, until it suddenly gets terminated for an unknown reason, or gives results that don't make any sense (see picture below). Memory and hard drive are not a problem, since I'm still testing with the lowest resolution which uses only 30% of my RAM and there's hundreds of Gigabytes of space in the hard drive.


      I have attached a set of files I am experimenting with.


      Here's how I'm doing the simulation:

      1. Assemble all parts and add lids to close the volume

      2. Use the Wizard tool to create new simulation project (internal, local rotating region, water, lowest resolution)

      3. Select the rotating region, a cylinder body which covers the whole impeller

      4. Select Fluid Subdomain, it goes through the model nicely

      5. Select boundary conditions: 1. Inlet volume flow (0,01m^3/s) 2. Outlet environment pressure 3. Select ideal wall for all parts except for the impeller

      6. Run and pray


      I am not quite sure if my rotating region is of proper shape and size. I did read through several tutorials and according to them this one should be correct. I am also not sure about the wall boundary condition. I read on another forum that all parts that are not rotating should be turned into walls, except for the lids. Is this correct?


      Here's a picture of the solver during the calculation. It seems as if there was no movement at all. The scale of the pressure and velocity cut plots keeps changing higher and lower, but the actual value seems to stay the same and is uniform throughout the model. There are some very tiny spots near the impeller where velocity is higher than 0m/s.



      What am I doing wrong? Any help is much appreciated.

        • Re: Centrifugal Pump simulation getting terminated
          Eljas Linna

          An update:

          Just like previously, the solver just got abnormally terminated near the end of the process, at around 500 iterations. After loading the incomplete results, they look like this:


          Some obvious problems I notice:

          -The rotating region goes through the pump wall and affects the suction side. I guess I should trim it to the outer shape of the impeller after all?

          -The velocity couldn't possibly be 10000 m/s when the rotation speed is 100 radian/s.

          -The flow trajectories get packed to the bottom as soon as they enter the pump inlet.

            • Re: Centrifugal Pump simulation getting terminated
              Piotr Regula

              The reason the calculation finishes abruptly is probably because you didn't specify any finishing conditions and Flow is using its default ones. To change them go to Flow Simulation -> Calculation Control Options.


              You should specify some goals in your analysis, you will be able to use them as finishing conditions and monitor them during the analysis. That will give you an idea of whether your analysis has converged or not.


              You might also consider simplifying the model, at least for the time while you are getting familiar with Flow Simulation. For example eliminate the fluid space on top of the impeller and the gap between the impeller and the intake.


              Definitely fix the rotating region.



              Good luck

                • Re: Centrifugal Pump simulation getting terminated
                  Eljas Linna

                  Ah indeed, setting up the finishing conditions and goals might do the trick. Thank you for the tip.


                  I did some testing with a simple model and managed to go thorugh with the calculation. The model was made in 5 minutes and is not based on anything real, so I can't say whether the calculation was successful or not yet. I will run an overnight calculation on a real model to see if it works out.


                  I noticed a problem while playing with the simple model though. In reality the rotating impeller should cause a vortex in the inlet channel, but in the simulation water goes straight until it hits the rotating volume and suddenly accelerates dramatically. The inlet vortex has a huge impact on pump's efficiency. Is there a way to simulate a vortex caused by rotation in FRONT of the fluid?

                    • Re: Centrifugal Pump simulation getting terminated
                      Jared Conway

                      If you are getting solver terminated abnormally with rotation it means that you have bad geometry or that you have and rotating region setup. There are lots of good posts about what should be done on the forum and also good documentation in the solving eng problems doc and the solidworks kb.


                      As for the vortex, it will be generated if the problem is setup properly. Another example of setup problem is that your outlet needs to be moved away because of the error you are seeing in the solver window.


                      You may want to look at getting some more direct training on rotating regions. We have one that is online here http://www.hawkridgesys.com/training/courses/solidworks-flow-simulation-rotating-machinery-online-course/


                      We also have mentoring which is much more extensive and specific to your application.


                      I also wanted to mention that your first project should be on something with a know answer.

                        • Re: Centrifugal Pump simulation getting terminated
                          Eljas Linna

                          Hi Piotr and Jared,


                          It took me a while to try the simulation again due to other unexpected matters. Thank you for the tips. The problem which caused the termination was indeed thte lack of goals and bad rotation geometry, so the original problem has been solved. The lack of vortex in the inlet channel still remains, however. Could the reason be in the rotating geometry or in the calculation settings? Here's a picture of the situation and the shape of the rotating region. As you can see the water in the inlet channel below the impeller does not have any signs of vortex. The rotating region follows almost exactly the outer shape of the impeller.



                          Since we are a rather small company we would like to use any other available resources before taking extensive training courses due to their high costs.

                            • Re: Centrifugal Pump simulation getting terminated
                              Jared Conway

                              are you using pressure pressure or flow inlet and pressure outlet?

                              if you're using flow inlet, if your inlet isn't a long ways away from the rotation, it will be very difficult to get a vortex to generate. remember you flow rate is basically a flat velocity profile so it needs time to develop into a vortex.

                              the problem is either that or you expect a vortex but it doesn't actually exist. are you working on a known problem or a new design?

                                • Re: Centrifugal Pump simulation getting terminated
                                  Eljas Linna

                                  Thank you again for the reply. I am using flow / pressure. I tried it with extended inlet and outlet with triple the length, but it still didn't seem to have any effect. Changing only the inlet length somehow brought the termination problem back. The water seems to get stuck and its velocity increases until the solver gets terminated. I suppose the the rotation geometry is still a problem. I have no idea how the inlet length could cause this, before it ran just nicely.



                                  We have ordered one flow simulation analysis in the past on a very similar pump from a CFD company in an emergency situation. The flow turned out to have great tangential speed in the inlet channel which caused the impeller to fill up unevenly. In my simulation the tangential speed is practically zero in the same location.


                                  At this time I'm working on already existing pumps with known measured results.


                                  I wasn't expecting flow simulation to be this complicated even in the basic level

                                    • Re: Centrifugal Pump simulation getting terminated
                                      Jared Conway

                                      Rotation is a complicated analysis. That is why we created a special class specifically on rotating regions




                                      Based on the best practices we teach in it and our mentoring, we rarely run into problems like you are. I suspect the issue is just exposure to the software and not following the recommendations provided by the developer.


                                      If you need quick turnaround and application specific training, don't hesitate to contact me. Or have you considered speaking with your var for some help? If you have a support contract, they can at least point you in the right direction or confirm it isn't a software bug or limitation.


                                      My guess on your problem is geometry. Also 3 lengths is a good first start, but it might take even more to soften the inlet flow condition.


                                      The other thing I would suggest is doing your first simulation on the one with known results. It is reasonably possible that you aren't seeing the flow because it doesn't exist in this design. In our consulting we run into a lot of people that think their design should do one thing or another and then test only to find out the design was the issue, not the software.


                                      So depending on the severity of this project, is recommend getting some more experience or testing a known problem if you have time, otherwise get some direct support on this.

                                        • Re: Centrifugal Pump simulation getting terminated
                                          Eljas Linna

                                          Thanks for the reply. We will be considering taking a course in simulation if nothinig else seems to help. The time of your class is very inconvenient though, it will be in the middle of the night where I'm located.


                                          Do you release any offline teaching material?


                                          The pump which was ran by CFD company is practically the same as this one, only scaled around 20% smaller (both the impeller and flow channel). Very minor differences in the outlet channel. We also know the head, efficiency and other results from the pump I am currently trying to run.

                                            • Re: Centrifugal Pump simulation getting terminated
                                              Jared Conway

                                              No doubt it isn't for everyone. Feel free to contact our training team to see if we can make accommodations. If you have a couple of people, they can probably make it convenient for you or come up withr an offline option.


                                              I'd you have those parameters for the current pump, why run cfd?


                                              Why not run the previous one?


                                              Would you agree that the performance of the new pump is unknown and that it may not be the software or the way you are setting it up but rather it is how it performs?

                                                • Re: Centrifugal Pump simulation getting terminated
                                                  Eljas Linna

                                                  I have now ran the simulation with the exact same model as used in the previous CFD case. I ran it with different boundary condition combinations. When given Inlet Volume flow and Outlet pressure the problem remains and no vortetx appears. Given inlet pressure and outlet pressure a nice vortex appears, but none of the fluid seems to be able to escape the rotating region. The same happens when given inlet pressure and outlet volume flow, the fluid keeps swirling around the area where it should shoot out from the impeller. EDIT: Actually, it does get out of the impeller, but it can't exit from the outlet it seems.


                                                  Could it be the rotating region after all? I have tried using several shapes; the same as the impeller outer shape, offset it halfway between the chamber and impeller, and having it offset well inside the chamber wall and making those surfaces stators. None have proven to be the all-solving answer though.


                                                  The inlet length shouldn't be the problem at least since it is now signifigantly longer than in the previous CFD case.




                                                  We want to run CFD on already existing and physically tested pumps to develop them further and to have a reference when learning the software, as you have suggested. We will not jump into new pump types before we can properly operate flow simulation.

                                                    • Re: Centrifugal Pump simulation getting terminated
                                                      Jared Conway

                                                      highly would recommend having someone review your rotating region setup. your var should be able to help you with this or you could post here or i've provided my contact jared@hawkridgesys.com.


                                                      there are some other things to consider:

                                                      1. what software was used previously?

                                                      2. do you know exactly how they set it up?

                                                      3. is it possible they used a different turbulence model?

                                                      4. their results were comparable to the physical world or was it only simulated?


                                                      i'd guess that since you're getting closer it is a setup issue rather than something limited in the code. however you haven't provided a lot of detail. when you say you set the pressure at inlet and outlet, what did you specify? what do you mean it doesn't escape the rotating region?


                                                      i do want to mention the path you're taking is a good one.

                                                        • Re: Centrifugal Pump simulation getting terminated
                                                          Eljas Linna

                                                          There will be a guy from SolidWorks coming to our office tomorrow to help us figure out what to do.


                                                          Previously the simulation was outsourced to a company specializing in CFD. I can't remember which software was used but it was not Solid Works, I also don't know how it was set up. The simulation did match the results of physical test runs quite well, they were within 5% difference from the physical results.


                                                          No doubt it is my faulty setup which is causing the problems.

                                                          I set the inlet condition to 3 Bar total pressure and outlet 10 Bar total pressure according to the real measured values. Also tried other values and changing to static pressure. This is what I mean by the fluid not getting out of the rotation;


                                                          When setting inlet volume flow and outlet environmental pressure the fluid goes nicely through the outlet.