Thus any one knew a commercial CFD program that can calculate spatial pressure distribution from a given velocity field ? The received velocity field can be regarded as laminar and incompressible, and the velocity data is structured in a 3-D Cartesian coordinate system.

Hi Svante, Please define the problem in detail. How is this velocity field got in the first place?

Thanks,

Thomas

Hi Thomas and thank you for responding to my question! I'm doing the final exam for the Master of Engineering degree at Linköping university hospital were my task is to find a way to calculate the time depended pressure distribution of a human heart cycle, or in some other place of the human cardiovascular system. The time I'll shall spend on this is 6 month ! The idea to calculate pressure have emerge from the possibility to obtain velocity data from NMR (Nuclear Magnetic resonance) cameras. That is, when placing a patient in to the camera for about 30 minutes, one resaves a velocity data file that e.g., describe the time dependent blood flow throw the heart of the patient. This data has already been utilised to visualise instationary pathline of the blood flow throw the heart. The received velocity data is organised in a 256X256X32 (u(x,y,z,t),v(x,y,z,t),w(x,y,z,t)) matrix system. This means, by solving the pressure gradients from NS-equation, one can use the PPE (Pressure Possion Equation) to solve the pressure distribution with in a constant. This sounds nice in theory but when looking for a practical way to solve this problem in 6 month it gets on to my nerves. Up to this day I have successfully calculate the pressure distribution for a stationary rotating cup by utilise Matlab. But this is just a 2-D stationary case with constant boundary condition. The two big problem I have now is to implement right boundary condition to a geometry that is moor irregular that one can imagine for a CFD problem, and the second problem is to get right Neuman condition for the problem to converge, that is if I shall use Neumann in stead of the Dirichlet condition? There have been several attempt to solve this problem and to my knowledge it has only been done for 2-D cases yet. The way that they have solve the problem in a 2-D case is to calculate PPE over the hole 256X256 region, even though the fluid region is a complementation off the hole region, and then state a Neumann condition on the edges of the region? In this way one is not forced to keep track on the fluid boundary region witch is a tremendous save on the implementation programming. In a similarly way I've solved the rotating cup but instead of using Neumann condition I've use zero condition on the edges and this seems to work well. But all explained above doesn't feel so good so that is way a ask if there is a commercial CFD program that can bee utilised to calculate the pressure distribution when one is using a known velocity field. My spelling and grammar is perhaps not so good but anyway I hope you understand what I have written. Regards Svante!

Hello Svante:

You might want to contact the people at CFD Research Corporation. CFD-ACE+ should be able to solve this problem pretty easily.

People you might want to contact there are: 1)Vinod Makhijani

e-mail address: vbm@cfdrc.com Vinod is incharge of Bio-Medical Group at CFDRC

2)Raza Mirza

e-mail address: rmm@cfdrc.com Raza should be able to guide you to solve this problem. He would know the current capabilities of CFD-ACE+ in detail.

Good Luck,

Thomas

May I ask why you want the pressure? That is, is the missing constant going to be a problem? If so you may have to get another piece of information from somewhere.

If your velocity data is coincident, then you not only have a problem with an unknown constant but also with controlling the shortest wavelength mode. Consider a constant velocity field (want a constant pressure field). If one uses a simple central difference to evaluate the pressure gradients then an arbitrary constant can be added to the solution without the differencing noticing but also an arbitrary highest order mode (plus/minus some value in a checkerboard pattern) can be added as well.

You are quite right to identify the pressure boundary conditions as a concern since simple conditions such as extrapolation, zero normal gradient and a fixed value are approximations and introduce error. Your Poisson equation applies at the boundary and so you would be better off working with that.

Does your data locally satisfy continuity? If not, you may need to "adust" it so that is does. Does you data globally satisfy continuity? If not, your Poisson equation will have no solution with Neumann conditions.

It looks like a nice project and performing it in Matlab seems a sound approach to me since you can control and sort out the issues indicated above (and all the other ones!). Not sure a commercial black box is necessarily a wise step since you may lose a true understanding of the errors present and hence the quality of the data.

yeah as thomas said any pressure based solver like fluent, fidap etc should be able to handle this. it would do you good to contact people at these companies to consult with