Simulation of Centrifugal Compressor in Ansys Workbench

Here I am going to explain the method to simulate the centrifugal compressor, the long awaited blog. Main attraction of this blog is to demonstrate the technique to handle the CAD model with blades and hub only imported as IGES/Step or parasolid file. Usually people convert the IGES or parasolid file in turbogrid (*.tur) format and get high quality hexa mesh in Turbogrid. The main problem with this approach:

1. Consistency problem. Different people may get different blade shape, casing and hub profile.

2. No easy / straight forward approach. I have been using Gambit to get coordinates. And some also use Gridgen or other softwares long with excel to do this work. While working in Gambit there are number of times when files get corrupted and some times Gambit crashes and wasting efforts of days (sometimes even weeks). There is also problem of persistence e.g. If some one want to change the number of profiles along span from two to five or ten. In Gambit or Gridgen it means starting from scratch each time.

3. Gambit and gridgen do not support extensive geometric operations while designmodeler provides CAD like environment.

Brief steps are :

1. Open workbench and drag the geometry icon into main window and open the designmodeler.

2. Apply operations in blade modeler mode of DM like creating flow passage and export points.

3. Open Turbogrid and use ATM optimized approach for best quality mesh.

4. Process mesh in CFX-Pre like setting up solver parameters, boundary conditions and turbulence model.

5. Start solver and solve problem

6. Post process results.

https://i0.wp.com/s30.postimg.org/gg3dhe8w1/image.png

https://i1.wp.com/s27.postimg.org/3pz6xkk8z/image.png

https://i1.wp.com/s11.postimg.org/7k22vi0tv/image.png

https://i1.wp.com/s7.postimg.org/gcdp15arv/image.png

https://i0.wp.com/s24.postimg.org/43oxzw6px/image.png

https://i1.wp.com/s23.postimg.org/ey36a5ht7/image.png

https://i0.wp.com/s7.postimg.org/pj9rtnk5n/image.png

https://i0.wp.com/s25.postimg.org/9ahl2jkwf/image.png

https://i2.wp.com/s25.postimg.org/bt3a386mn/image.png

https://i0.wp.com/s25.postimg.org/9psuvk6tr/image.png

https://i1.wp.com/s25.postimg.org/z9v51zs7j/image.png

https://i0.wp.com/s25.postimg.org/a544oksr3/image.png Continue reading

Advertisements

Solver settings for Conjugate Heat Transfer

Conjugate heat transfer (CHT) can be defined as the simulation of solid zone and fluid zone simultaneously . In solid zone we will only solve conduction equation (simplified form of energy equation) and in fluid zone continuity, momentum and energy equation. We can activate body forces due to gravity and buoyancy forces by defining operating density in operating pressure panel and thereby activating natural convection mode. During initial phase of solution due to different behaviour of solid and fluid zones, solution diverges immediately when solved in steady state mode due to buildup of initial transients (they are not real and will have no effect on final solution). Therefore in this blog we have discussed two approaches i.e. transient solution with only one iteration per time step and pseudo transient solver. It is FAQ that what solver settings I should use for any CFD problem including conjugate heat transfer modeling. Even advanced users get confused, let alone the “novice”. This question is even more demanding with inclusion of advanced models such as heat transfer, multiphase flow modeling etc. For current problem Fluent 15.0 is used with two different settings discussed later. In this blog I have tried to clarify different flow schemes for the conjugate heat transfer problem. In this simulation heat transfer from computer chip (producing 75 W) is to be dissipated into the still air via natural convection heat transfer mode. Fins are attached to CPU chip to enhance the heat transfer rate. External air modelled with zero velocity using pressure inlet (zero gauge pressure) and pressure outlet (zero gauge pressure). For the fins, chip and board solid zones are defined. Inter fluid-solid or solid-solid coupling is done via coupled boundary conditions. Chip is modelled as source term to energy equation using volumetric heat generation. First we have tried steady state solver but in vain. After that we have used two settings with similar results with different no of iterations to converge solution. These cases are: 1. PISO coupling scheme with transient solver: Time step was chosen as 10 s arbitrarily and one iteration per time step. It is found lower time steps induces unsteadiness and due to that solution does not converge. It is recommended to initially use 1000s as time step to make the solution stabilize. 2. Coupled pressure based solver with pseudo transient solver: Time step method is user defined with 10s time step for fluid and 1000s for solid. Conclusions: 1. Solution converges faster (10x sometimes) with Pseudo transient. 2. Overall results are same.

Recommendation: For steady state coupled problems always use pseudo transient Solver with coupled pressure based solver 

heatsink Setup Output1 mesh a001 CAD model Solver settings for transient + Piso b photo hosting a Solver settings for transient solver Solver settings for Coupled pressure based + pseudo transient c d Results for Pseudo transient and coupled pressure solver contour1 contour2 contour3 contour4 Results for transient solver with PISO scheme. trasient 1 jpg Kostenlos Bilder hochladen trasient 2jpg trasient 3jpg trasient 34pg

Turbomachinery modeling in Ansys workbench: Axial Turbine or Compressor

Steps : 1. Open workbench and create cell for each program 2

2. Import geometry in designmodeler and create flow path. Export coordinates on blade in turbogrid format 

3

3.  Create mesh in Turbogrid   

4

4. Setting boundary conditions and problem setup in CFX-Pre 6   

5

5. Getting solution in CFX-Solver   

8

6. Post processing in CFD-Post 

1

7

Low Pressure Turbine T106A Transition Modelling

  1. We have used Fluent 14.5
  2. Solver is coupled pressure based
  3. Flow scheme is 2nd order accurate
  4. Time scheme is also 2nd order accurate
  5. Meshing was done in Gambit. Mesh is hybrid. Mesh in boundary layer is map and in the rest of domain is triangular
  6. Two meshes are used: Baseline and fine
  7. Time step is 0.001, 0.0001 and 0.00001
  8. Total simulation time is 0.1 second and data is averaged over all time steps.

ImageImage

Image

Image

Image

Image

Image