Learning Forum

[abtharpe42]

I'm trying to simulate a simple gas ejector, only several inches in size, with ideal gas air (constant properties) using geometry and conditions from a disseration I've found online. The author simplified his domain to 2D axisymmetric. My plan is to use his design as a starting point for an alternative design I'm working on that will require the domain be 3D, so I'm trying to replicate the author's results in 3D to act as a benchmark for the alternative design. From my own past experience, modeling a simple gas ejector has been confusingly difficult to converge in the density-based solver. Below is the geometry and settings of the simple replication case that I'm currently trying to run:

 

Mesh:

• Polyhedral
• Body Sizing is 0.02 inches
• Body of Influence is 0.005 inches
• Curvature Normal Angle is 10 degress
• Boundary Layer is Smooth Transition with 5 layers, Transition Ratio of 0.272, and a Growth Rate of 1.5
• Minimum Orthogonal Quality is 0.5
• Cell count is 360,771 cells

Model Settings: 

• Density-Based Solver
• K-Epsilon Realizable with Enhanced Wall treatment and Production Limiter
• Formulation is Implicit with a Courant Number of 5
• All Discretizations set to Second Order Upwind
• Gradient is Green-Gauss Node Based
• Flux Type is Roe-FDS
• Enabled High Speed Numerics, Warped-Face Gradient Correction, and Convergence Acceleration for Stretched Meshes (CASM is probably unnecessary)
• Under-Relaxtion Factors left at default values
• All Residuals set to 1e-6

Boundary Conditions:

• Operating Pressure is 0 psi
• Primary Inlet: Total Pressure is 164.7 psi, Supersonic/Initial Gauge Pressure is 164 psi, and Total Temperature is 300 K
• Secondary Inlet: Total Pressure is 59.3 psi, Supersonic/Initial Gauge Pressure is 59 psi, and Total Temperature is 300 K
• Outlet: Gauge Pressure (Static Pressure) is 59.7 psi, Backflow Total Temperature is 300 K

Residuals and Entrainment Ratio Plots:

Mach Number and Static Pressure Contours:

 

Vector Plot:

As shown in the Residuals and Entrainment Ratio plots, I had decently low convergence but not all the way to 1e-6. My Entrainment Ratio plot is oscillating, and you can even see in the Vector plot at the diffuser exit that I had slight backflow. These results were actually some of the better quality that I've obtained so far. I've also found that a Poly-Hexcore mesh for some reason performs slightly worse than a Polyhedral mesh of equal size. Any advice on what more I need incorporate or change to acheive complete convergence of a simple 3D ideal gas ejector? Information online of 3D ejector CFD is scarce.

Thanks in advance.

[Rob]

If it's only Mach 2 or so I'd use PBCS over the Density based solver, the latter is for really fast or very tightly coupled flow systems. 

You may also need more mesh in the core, but I can't tell from the images:  we have zoom for a reason! I'd also review how much of the upstream domain you really need: what do you gain from including so much of the high & low pressure region? 

As an aside, ejectors (also called jet pumps) are anything but easy to get working well. They are very simple, but that doesn't mean it's easy! There's decades of research into them as the various flow pressures & dimensions are fairly critical. 

Another aside, I always find decimal inches to be highly amusing. I'm English, and even we use metric, well, most of the time........ 

[abtharpe42]

Thanks for the reply! And yeah, when I made the forum post the images were way bigger, so I'll keep that in mind next time. The base of the geometry was given in the dissertation I was following. The only part I changed was that the primary and secondary inlets now enter a complete chamber and the fact that the domain is now 3D instead of 2D axisymmetric. As far as using the PBCS solver, I've tried that and it does work considerably better. But from everything I've read regarding gas ejector CFD the density-based solver is what everyone runs with, so that's why I was using it. As far as why the domain has to be 3D, I'm wanting to eventually include a third inlet to the starting chamber. I have a task involving the mixing of certain gas species using the ejector, and some of those species are to enter through a third inlet, which will have its own temperature and pressure conditions. When using the PBSC solver, what would you recommend regarding the settings for Discretizations schemes, URFs, and the Time Scale Factor for good convergence?