Mr. Joseph Jhun 0 Report post Posted April 16, 2019 Hi Team, I'm quite surprised that the reverse normals on a trimesh with V4.4 doesn't seem to work. To confirm it, I've tried with V4.3 by following exactly the same procedure that I've done with V4.4 with no problem, i.e., I was able to switch the surface normal back and forth by simply applying the reverse normals on a trimesh with V4.3. Has anyone ever noticed this? Thanks in advance,Joseph Share this post Link to post Share on other sites
Mr. Carsten Fuetterer 9 Report post Posted April 17, 2019 Hi Joseph, I guess you are using the Converge Export. This one now automatically changes the normals so that they always point inwards, which is needed by Converge. So the post processing functionality might not work for this case. Why do you need this? Is the normal direction wrong? best regards Carsten Share this post Link to post Share on other sites
Mr. Joseph Jhun 0 Report post Posted April 17, 2019 Hi Carsten, As you know already, we've been running blood pump simulations containing a rotating body (i.e., impeller), and the vectors on the rotor surface should point outward, thus, the computational domain is made only between the pump casing/housing and the rotor surface. It makes me worried if you made the Converge Export's vector normal being always inwards by default and there's no option to control it. Thank you,Joseph Share this post Link to post Share on other sites
Mr. Carsten Fuetterer 9 Report post Posted April 17, 2019 Hi Joseph, I guess we did not thought of this. Do you export impeller and flow domain extra and merge it in Converge? best regardsCarsten Share this post Link to post Share on other sites
Mr. Joseph Jhun 0 Report post Posted April 17, 2019 Hi Carsten, Yes, that's what I do or at least that is what I prefer because of the controlability of the vector normals on both (i.e., rotor and pump casing) parts independently.In fact, I guess we could simply use STL export option to get around this circumstance (with either colour or multibody), but it still seems weird that Converge export works only that way... Thank you,Joseph Share this post Link to post Share on other sites
Mr. Joseph Jhun 0 Report post Posted April 17, 2019 Hi Carsten, I've also found a problem with trimesh being exported as STL for Converge. I've used STL Multi Body to be used for Converge, and there were so many "Intersctions" and "Open Edges" from it. Yet, one that's through Converge export got no issues like that at all. If I reiterate the issues briefly: Trimesh exported with Converge export: It creates well-connected triangles, i.e., no issues with intersections, open edges, etc. However, it doesn't have controllability on vector normals. Trimesh exported with STL Multi Body: It has controllability on vector normals, yet, it creates number of issues with triangles. Please let me know what you think. Thank you,Joseph Share this post Link to post Share on other sites
Mr. Carsten Fuetterer 9 Report post Posted April 18, 2019 Hi Joseph, why don't you combine them into one flow domain in CAESES? I will ask the developers how we can handle this. cheersCarsten Share this post Link to post Share on other sites
Mr. Joseph Jhun 0 Report post Posted April 18, 2019 Hi Carsten, In fact, that was the method that I've tried the very first time...The reason I began to separate them was to test the controllability of vector normals independently...I've sent you the project file via your direct email, and please let me know if you need any further information. Thank you so much for your support again!Joseph Share this post Link to post Share on other sites
Mr. Carsten Fuetterer 9 Report post Posted May 3, 2019 As a general remark, since this issue was solved via mail: Our export still automatically points the surface normals inwards and therefor ignores the post processing operation "reverse normals". This does also work, if you substract parts from each other. cheersCarsten Share this post Link to post Share on other sites