Improving accuracy for simple shapes

[tegnamo]

I am trying to obtain accurate drag force results for some simple shapes being analyized in Floworks. I am currently using a sphere 1.5m in diameter, as well as a cylinder 1.5m in diameter (they are being analyzed separately). The cylinder is experiencing flow normal to its axis.

While I can get what appears to be relatively close results (same order of magnitude, but still off by 70-100N for the cylinder) to that predicted by regular hand calculations, I am wondering about a few of the values and options.

I've got most of the conditions figured out, using air flow in the x-direction of 27m/s, and external flow being selected. Some of the options I can't quite grasp, however, include the turbulence parameters, intensity, and length. Is there some method for choosing the proper values? I really don't know how to predict what the intensity should be, nor the length.

Additionally, I am slightly confused about what surface goals I should specify when trying to obtain the total drag of the object in question. There is the x-component of force, along with the x-component of normal force, and so on. I am looking for the same value which is yielded by your basic drag force equation for these aforementioned shapes. For instance, a cylinder is Fd=Cd*.5*u^2*p*d, and a sphere is similarily simple in execution. Perhaps I need several of those surface goals and I would then sum them together?

Lastly, I wonder about the result resolution, and minimum gap size settings. It seems that the gap size should be something like 0.1m, given the 1.5m diameter of my object, but I really don't know. I suspect that the result resolution should be relatively high. I had good results with 6 and above, though CPU time increases significantly.

If someone has advice on the above or on how to, in general, make your generic CFD airflow calculation more accurate, I'd appreciate the input.

Thanks!

 

[tegnamo]

I had forgotten to add one more thing to my huge list of concerns.

Is Floworks opposed to analyzing flow over two solids running in series? I have two cylinders, and I expect the drag in the trailing cylinder to be less, since it is drafting the first, but the surface goals state that the x-component of drag is about 100N higher on the leading cylinder. I attached a picture with the flow trajectory visible, to give you an idea.

Thanks again!

 

[tegnamo]

Argh...I meant that it is 100N higher on the trailing cylinder.