11 Replies Latest reply on Nov 25, 2014 11:29 AM by Jared Conway

    Relief Valve on a PD pump

    Mike Ramsey

      I'm working on designing up a new poppet style pressure relief valve for a PD (gear) pump. We have never attemped to use CFD to design one of these valves before so I'm trying to figure out the process for doing so. I'm starting out trying to get the CFD analysis to match up with an existing valve that I have performance curves for. I know from testing how much that poppet moves from it's cracking point to the full bypass point and the pressure it starts to move and the pressure were I see full bypass. I've figured out using a parametric study how far the valve needs to open to pass all the flow through it but the poppet moves a lot more than what I've seen from my physical test data so the flow is a combination of the pump clearances and the valve opening. What I'm thinking I need to do is to calculate the cross sectional area of the clearances in the pump and make an oriface with the same cross section and create a manifold for the CFD. I attach the valve to the manifold along with the oriface. Once I get that done how would I go about finding the point where the valve begins to open and at what point does the flow out of the manifold reach 0?

        • Re: Relief Valve on a PD pump
          Amit Katz

          From what you describe it sounds like you're trying to do a combined physics and CFD analysis (using pressure of the fluid to move the plunger inside your valve.) Is this correct? I'm just trying to understand what you're trying to accomplish.

            • Re: Relief Valve on a PD pump
              Mike Ramsey

              Amit,

               

              That's probably a good way to look at it. The valve stays closed until a certain pressure is reached. Once the system hits that pressure I want to know how far that valve moves and what the system pressure is when I get zero flow coming in and out of the system.

                • Re: Relief Valve on a PD pump
                  Jared Conway

                  In flow, there is no fluid structure interaction. So you can't model it popping up. The parametric study is a good step, what you may need to do is look at forces on the valve and evaluate at what pressure the forces balance...etc. It will nit be an exact science.

                    • Re: Relief Valve on a PD pump
                      Mike Ramsey

                      So could I handle it like this;

                       

                      Set up a parametric study adjusting the position of the poppet with a pressure differential across what I'm going ot call a manifold. Set a goal plot on the outlet sides trying to get the flow out down to 0?

                       

                      Just for a point of reference here is a picture of what the set-up looks like now. Some how I need to make the simulation think there is a flow going through the oriface I have in the manifold. My goal is to get the fluid circulating just internal to the system.


                      RV-CFD.jpg

                        • Re: Relief Valve on a PD pump
                          Amit Katz

                          What Flow will help you with is determining the flow rate and the pressure loss for a given valve gap. You will need to use this analysis in order to choose an appropriate stiffness for your valve.

                          • Re: Relief Valve on a PD pump
                            Jared Conway

                            a few indices on the graphic would help me understand what you mean by "Get the fluid to circulate internally" and to evaluate if your approach will work

                              • Re: Relief Valve on a PD pump
                                Mike Ramsey

                                I've given up on getting the analysis to work as shown above. I went back to just the valve and trying to figure out how much the poppet needs to move in order to get the correct flow. In order to check my method I dug out some old test data where we measured the flow through the valve and had a rod attached to the poppet so we could measure the movment of the poppet at a given flow rate and pressure. What I've found is that if I move the poppet to the position indicated by the test data I should be getting about 10X less flow through the valve that what the analysis is showing. For example I move the valve open by .045" my test data gives me a flow rate of 28.8 gpm with 130 psi differential; the software on the other hand gives me 290 gpm going through the valve at the same position and pressure. I flip that around and I set a target flow rate of 28.8 gpm (converted to a mass flow rate) and I show the valve opening by about .0039". My data only shows a couple GPM at that position.

                                  • Re: Relief Valve on a PD pump
                                    Mike Ramsey

                                    Just a ammendment to my previous statement. I found an issue with the fluid densisty I had entered in. That helped in reguards to the mass flow rate being way off but I'm still about 3X more flow than I should be.

                                      • Re: Relief Valve on a PD pump
                                        Jared Conway

                                        probably not an equivalent test, would need to know a lot more about the test to compare it to the simulation setup

                                         

                                        flow simulation has several validation examples if you are concerned about accuracy of the software

                                          • Re: Relief Valve on a PD pump
                                            Mike Ramsey

                                            Ok so the original test consisted of a pump with the valve connected to the pressure side. The inlet was unrestricted as best as possible. There were flow meters inline with the outlet on the valve and the outlet of the pump. A gate valve was used to restrict the flow between the pump and the flow meter (poppet valve was unrestricted). Line pressure was measured on the pump inlet and on the pump outlet. The valve had a rod attached to the poppet coming out so movement could be measured with a dial indicator. Using this set-up we were able to get flow through the valve when it started opening and how far the poppet traveled when all the flow was directed through the valve. Using this data I set up the parametic study as a what if study. For my boundry condtions I used the pressure on the inlet and outlet side of the pump taken from my data. I then set up a surface goal on the inlet side of the valve measuring the mass flow rate. So for the input variable I edited the distance mate such that it was .045" more than if the valve is closed off, then I set a second design point moving the valve to .050". My output parameter was the surface goal. So I ran the study and compared the results. My test data shows I should be getting the following:

                                            Valve open .045" - 28.6 GPM

                                            Valve open .050" - 42.6 GPM

                                             

                                            Comparing that to the Flow simulation results

                                            Valve open .045" - 82.479 GPM

                                            Valve open .050" - 88.359 GPM

                                             

                                            The fluid is No. 2 fuel oil.

                                              • Re: Relief Valve on a PD pump
                                                Jared Conway

                                                is problem sensitive to mesh? ie, do you have a converged result?

                                                is the problem sensitive to the fluid? ie are your fluid properties exactly the same?

                                                is your problem sensitive to testing conditions? ie are all your testing conditions exactly the same?

                                                "line pressure", is that total or static pressure, is your system sensitive to which you choose?

                                                is your system sensitive to the flow from the components that attach to it? ie do you need to add more components to ensure that your area of interest isn't influenced by the boundary conditions?

                                                what happens when you reverse the problem, flow in and pressure out, do you get the right pressure drop?

                                                 

                                                my suggestion to you would be to get confident in the software by solving a problem with a known solution with hand calcs and independent testing and well defined conditions and assumptions. once you're good there, you can move onto your project. again i highly doubt a software issue but rather more likely an issue with the setup.