A quite rare Innocentive Challenge popped-up lately in front of me. The “Seeker” is looking for a solution to decrease their tremendous hydraulic valve prototype tests. This is a very typical “reduce trial-and-error testing by simulations” scenario.
Excerpt from the description:
The Seeker designs and produces hydraulic valves for use in its products.
The design process is currently trial-and-error in that for each new valve design a prototype must be produced in order to test the pressure and flow characteristics of the design. If the prototype does not have the desirable flow and pressure characteristics it must be re-designed and re-tested until its characteristics are within the required range.
The cost and time involved in designing a new valve could be significantly shortened if the Seeker had a way to accurately model and predict the valve characteristics before producing a prototype and thus is seeking software to accomplish this task.
The Solution should not require complicated and expensive commercial software packages that would be overkill for this specific application.
But this is exactly what opensource CFD is for! Simulating valves is pretty straightforward with Computational Fluid Dynamics, especially if OpenFOAM is used as a tool. The great flexibility it provides helps to automate the simulations through scripts: this is what the “Seeker” of this challenge is looking for!
My main concern is still that even somebody gives them a working framework (simulation case setup, guidelines for meshing the hydraulic valve etc.), they will still need somebody, also known as a CFD Engineer to do the following:
- CAD geometry preparation for simulation (cleanup, healing of the hydraulic valve etc.)
- Meshing. While this can be automated, in special cases (thin gaps for instance) it can not.
- Simulation setup. Yes, a template can be used, but what if there are special needs (moving parts, etc.)? What about convergence, residuals?
- Postprocessing. This can be scripted/automated, too, but are they sure they are extracting the right data from the results?
If they don’t want invest in a CFD Engineer, I would rather outsource the work to a CFD company, that is capable to do the above steps. This way the Seeker can still benefit from the reduced cost and time of prototyping, and enjoy the biggest advantage of on-demand outsourcing: it would be unnecessary to hire a CFD engineer.
Just my fifty cents.