Mr. Juryk Henrichs__0001 0 Report post Posted September 30, 2015 Hallo, starting to use the inscribecircles() function, I am wondering, what determines how close the curve ends are approached? I remember you did intentionally not go to the limits of the curves, but is it possible to control the offset? ThanksJuryk Share this post Link to post Share on other sites
Mr. Claus Abt 14 Report post Posted October 1, 2015 Hi Juryk, You could simply create an imagecurve of the sides and restrict their parameter domain,e.g. [0.05,0.95] and then run the analysis. Another option would be to create a feature that analyses the curvature of the camberline and cuts it off where curvature increases again towards the ends.Claus Share this post Link to post Share on other sites
Mr. Juryk Henrichs__0001 0 Report post Posted October 4, 2015 Hi Claus, I actually was looking at the other direction, making the circles go to the very end of the domain. I succeeded bringing the circles much closer to the ends by using imagecurves with a domain of [0.0,0.1]. I can now see why the function is cut off, but I still would appreciate more control over the function as in: used domain of the base curves and number of circles or spacing What exactly is happening towards the leading edge of a profile? I had hoped you could use the function also to determine the leading edge radius of a closed profile and I would like to have the full thickness and camber curves as they could be used for easy variation of thickness and camber along the span of a wing etc. I also actually wonder, how section analysis is done in xfoil? Is this along similar lines? CheersJuryk Share this post Link to post Share on other sites
Mr. Claus Abt 14 Report post Posted October 5, 2015 HI Juryk, It depends a lot on the quality of the profile curves if you get all the way to the ends. If you specialize the functionality for analyzing profiles there might be quite some improvements possible. But there are many issues about the inscribed circles already in theory, and even more in practice. If profiles are created from thickness distribution and offset normally from camberline it becomes obvious when you approach the leading edge radius. The radius does not represent the normal offset.InscribedCircles.fdb Maybe you can supply a feature to the community that helps getting closer. CheersClaus Share this post Link to post Share on other sites
Mr. Juryk Henrichs__0001 0 Report post Posted October 5, 2015 Hi Claus, I was looking only briefly into creating a circle from two tangents and did not really find out how to do this. So, i was happy Mattia mentioned that function. Would it be a problem to extend the functionality and add domain bounds and number of circles or spacing as parameters? Attached is a feature for section analysis in which I use the curve parameter to determine the normals. It is assuming that once you found the leading edge, you can move along equal fractions of the two halves to determine thickness and camber. I used the inscribed circles on fractions of the suction and pressure side to determine the leading edge (as the furthest point form the resulting camberline). The result compared to the inscribed circles looks reasonable well, but it is tested with one profile only so far. It works with a uniform parametrization of the curve only (obviously - so the parametrization as input should be removed). section_analysis_example.fdb I'd be happy to receive some feedback. Share this post Link to post Share on other sites
Jörg 29 Report post Posted October 6, 2015 Hi Juryk, Basically, it is no problem to have additional parameters and domain bounds - but I am not sure when we can finally provide this functionality. I hope you can work with the current status for the next weeks. CheersJörg Share this post Link to post Share on other sites
Mr. Juryk Henrichs__0001 0 Report post Posted October 8, 2015 Hi Jörg, yes, sure. This was meant as a feature request only. It already improved what I had so far and I will rebuild the feature to use the circles as far as possible as it should be more accurate. Anyway, in the above example there seems to be a problem with the inscribed circles if the foil is rotated by 90 deg. Documentation says the curves should be in the XY-Plane. Is there some requirement with respect to orientation too? CheersJuryk Share this post Link to post Share on other sites
Jörg 29 Report post Posted October 9, 2015 When changing the transformation angle to 90 degrees, I cannot see any issues. Anyway, it is possible that this routine works properly only with foils that are located in the XY-plane (this is a must), aligned with the x-axis. At least, this was the assumption at that time when it was implemented. Share this post Link to post Share on other sites
Mr. Juryk Henrichs__0001 0 Report post Posted October 9, 2015 Hi Jörg, funny, but you're right. At 90deg it works well now for some reason. I had to search a little bit but here is an example where the problem occurs again. I put the angle to 268 deg and no circles are inscribed. If the angle is set to eg. 275 deg, everything works again as expected.section_analysis_example_268deg.fdb CheersJuryk Share this post Link to post Share on other sites