10 Replies Latest reply on May 9, 2015 10:13 AM by Manuel Perarnau

    Pump simulation

    Manuel Perarnau

      Hello Solidworks forum,

       

      I have used solidworks many times before to do sketches and blueprints, but this is the first time I have use it to simulate flows.

       

      Overview:

      I'm trying to evaluate the performance of a custom pump that manages a very dense and viscous fluid through a rather small piping system.

       

      System:

      Assuming a flow required and a small reservoir to hold the fluid, I calculated the head needed from the pump. The pump itself is basically a horizontally placed Archimedes screw inside a cylinder.

       

      Modeling:

      Once the screw was sketched, assuming a certain length and number of coils, the pump was assembled. According to the tutorials, a solid cylinder was created concentric to the screw to simulate the rotating region, but made invisible in the "Component control" option.

      Later, according to the calculations made earlier, I set an inlet and outlet pressure. The inlet pressure was the pressure produced by the column of fluid in the reservoir, while the outlet was set to be the pressure that the pump has to overcome.

      The system is such that the pump needs to create 73421 Pa of gauge pressure on the outlet, while having a tank on the inlet that translates to 1234 Pa of gauge pressure.

       

      Simulation:

      After several simulations using a mesh in the third setting, I got the following results using different boundary conditions:

       

      Inlet

      [Pa]

      Outlet

      [Pa]

      Fluid

      [m3/s]

      Type

      Pressure

      Type

      Pressure

      Direction

      Flow

      Static

      1234

      Static

      73421

      Backwards

      3,80E-04

      Environment

      101325

      Static

      73421

      Correct

      1,60E-04

      Environment

      101325

      Static

      174746

      Backwards

      3,86E-04

       

      It appears that  the pump isn't really doing anything (it rotates at 30RPM), because the flow just goes from high to low pressure, but I am still not convinced that the boundary conditions are set correctly. Given that the fluid is glycerin (the closest thing to the actual fluid I'm using) I assumed that the viscosity would help  "adhere" to the slow rotating impeller and carry it to the end of the pump where it would encounter a sudden rise in pressure.

       

      Are the conditions set properly or is the pump itself that isn't creating any pressure?Flow lines 1.jpgFlow lines 2.jpg

      Basic layout.jpg

      Thank you for your patience and help.

        • Re: Pump simulation
          Amit Katz

          1. Your fluid flow direction is going with the pressure gradient. I suggest trying to set both the inlet and outlet to be environmental pressure, that will allow you to isolate your pump and see exactly how much flow it pushes and what the pressure is at the pump outlet.

          2. You can define your own custom fluid using the engineering database, check the "Materials" tab.

          3. "I assumed that the viscosity would help  "adhere" to the slow rotating impeller and carry it to the end of the pump " - Flow Simulation does not simulate fluid-structure interactions. I don't remember the exact BL model but I'm pretty sure that no-slip condition is used, so the fluid properties only affect the fluid-fluid interactions.

           

          4. I am curious as to what kind of rotating region model you're using (global, averaged, sliding) and how your stators are defined. Could you go into a little more detail about that?

            • Re: Pump simulation
              Manuel Perarnau

              Hello,1.jpg2.jpg3.jpg

               

              1.- I see. I've repeated the experiments and the flow is around the value I needed in the first place, but without the outlet pressure boundary condition, how can I know that the pump can deliver that flow once plugged into the system?

               

              2.- Yes, I defined the liquid before I started the simulation.

               

              3.- In general settings/wall conditions there's an option in which you can fiddle with the roughness of the wall. I haven't touched that option, but could it be possible to set the roughness according to the material for each element like the rotor or inner cylinder?  If that's so, would the shear forces help the fluid move inside the pump?

               

              4.- Where can I find the rotating region model options? I just made a cylinder with the same diameter and length as the rotor. Maybe I'm missing something

               

              Thank you for your help

                • Re: Pump simulation
                  Christopher Schaefer

                  Manuel,

                  Your rotating region- the boundary of the screw up to the inside of the cylinder wall- both rotates and translates.  SWx Flow Simulation's rotating region is not intended for such geometries; not even the sliding mesh new to SOLIDWORKS 2015.

                   

                  But something else to consider is any possible caviational effects along the blade.  What viscosities are you working with?

                    • Re: Pump simulation
                      Amit Katz

                      I am confused as to why a screw type impeller would not be a good fit for the software. One of the case studies in the tutorials (the CPU cooler) includes a rotating region to simulate a twisted blade fan. This is a similar application.

                        • Re: Pump simulation
                          Christopher Schaefer

                          Amit,

                          The rotating region of Flow Simulation assumes a number of things. One is that flow moves along the axial direction of rotation.  This is why scroll caged fans and other impellers with a scroll house sometimes provided trouble with Flow Sim (not impossible but as you may have experienced, the fluid domain has to be very tightly gridded).  And this is addressed with the new sliding mesh feature of 2015.  Secondly, the fluid domain using the standard rotating region stays "in plane." The cells themselves aren't moving but the swirl is calculated as part of the cell's behavior at each iteration.


                          The motion of a screw pump is such that the boundary between the domains (solid vs fluid) moves or translated through the plane and so the fluid domain actually needs to update during the rotation.

                           

                           


                          Also, Flow Sim uses an averaging technique with their RRF approach, also nullifying the screw-type pump originally asked about.

                            • Re: Pump simulation
                              Manuel Perarnau

                              Hello Christopher,

                               

                              I'm fairly new to the flow simulations of Solidworks, but does this means that the program is not capable of managing axial pumps?

                               

                              On the other hand, given that it can simulate fans with their housing, would it be reasonable to put a set of fans on top of each other, each with its own rotating region,  with the blades offset to create a continuous impeller?

                               

                              If they are all rotating at the same speed the mass moved by one stage can be transported to the other  with little disturbance, or at least I hope.

                               

                              Thank you for all your help

                                • Re: Pump simulation
                                  Christopher Schaefer

                                  Manuel,

                                  Axial discharge is not the question so much as it is the nature of the rotating region.  And no, I do not believe you could "stage" a series of rotating regions as the boundary where each region is stacked would have to share fluid cells with a neighboring boundary.  This would produce numerical errors that would only propogate from region to region. Flow Simulation does support multiple RRFs but the intention was never to stack them in series.

                                  (that said, I have done this with a "pre-charger" but their impellers were phyiscaly separated by a flow separation device)

                                   

                                  If one knew the swirl well you could possible treat the impeller/screw as a static solid and describe the motion of the fluid through/across said solid.  But you're mesh refinement at the boundary layer of the impeller would have to be quite fine. And I would still be concerned with the effects of cavitation depending on velocity and viscosity.

                                   

                                  In your initial post you stated that you're trying to evaluate the performance of a pump.  Do you have physical evidence to correlate against the Flow Sim results?  How do you plan on validating the CFD model?  Are you simply looking for trends or are you anticipating that the analyses will arrive at actual values?

                                    • Re: Pump simulation
                                      Manuel Perarnau

                                      Hello, and sorry for the delay

                                       

                                      There is no physical evidence to compare the performance of the pump. The idea was to setup a system and use a computer simulation to design a pump showed earlier. I still don't know if I could work, but given the amount of problems that I'm having, I think I will switch to another type of pump like a positive displacement.

                                       

                                      Of course this type of pump, according to what I saw, can't be simulated using Solidworks flow simulation, so instead I'll use the 3D drawing to manage the packaging of the components, blueprints and structural analysis.

                                       

                                      Thank you all for your help!

                                • Re: Pump simulation
                                  Jared Conway

                                  the tech reference and swx KB do a good job of explaining what Chris has outlined.

                          • Re: Pump simulation
                            Jared Conway

                            I would recommend connecting with your reseller and discussing this with them and the flow developers. Like Christopher mentioned, I don't think that the rotating regions in flow will work for this type of geometry.