Re: [femm] The contour creation for force calculation

[David Meeker <dcm3c@xxxxxxxxxxxxx>]

matthewc_98 wrote:

> Hi All,
>
> I use FEMM to simulate the solenoid and so far, it works pretty well.
> But there is a question I cannot understand and hope you can help me
> to explain it. The question is as following:
>
> Currently, when I would like to calculate the pull force of the
> solenoid, I always creat the contour path as close as possible to the
> surface of core steel bar (the contour is away from the steel bar
> which means the path is within air). But while I read the User's
> Manual, I found that Dave suggested user to creat the contour at
> least two elements from any interfaces and boundaries.
> Could you tell me that what the definition of the element (distance ?
> inch ???)?

If you push the toolbar button with the yellow triangles in the 
postprocessor, the elements will appear on the screen as yellow 
triangles. The elements don't have a fixed size--you can make them as 
small or you want, depending on the mesh size that you define for each 
block label. Anyhow, you don't want to draw your contour through 
triangles that touch the boundaries. In theory, the contour that you 
pick doesn't really matter, as long as it passes through air and 
encircles the object that you are interested in--there's no reason to 
try to draw your integration path close to the object. However, if you 
draw close to the surface of objects, you tend to pollute the force 
results with relatively inaccurate field results that can occur around 
sharp corners. Also, the part of H that is tangential to an interface 
tends to be less accurate than a bit farther out, owing to some aspects 
of the smoothing scheme that femm uses.

> So far, I used the force data of FEMM to campare force data that from
> our force tester in the lab, and found that the error is within 15%.
> (Assume the force data of force tester is ideally correct).
> Is there any way to narrow the error of force simulation of FEMM?

A good way to go is to try the simulation with progressively finer 
meshes, increasing the mesh density until the result from the force 
integral converges to whatever tolerance you want. 

In general, mileage may vary in comparing to experimental tests--even if 
the program is giving exactly the correct answer to the numerical 
problem that it is solving, there can always be significant differences 
between the experimental setup and the as-modeled problem, due to 
mechanical tolerances, differences in material properties, measurement 
errors, and so on.

> Anyway, I really appreciate and satisfy FEMM.
> Thank you so much, Dave.
>
> Best Regards,
>
> Matthew
I hope that the program proves to be useful to you.

Dave.