Airfoil Generating Negative Torsional Stiffness Values

  1. Anthony Ferman

    I am attempting to use VABS to analyze several airfoils of the same shape and size, but with different twist angles.  When I analyze this airfoil with zero twist, the Torsional Stiffness generated by VABS is positive as it should be, however whenever I apply the appropriate kappa1 values for twist, VABS generates negative Torsional Stiffness values which seems to indicate that instead of having torsional stiffness, the opposite is true and when any torsion is applied, the airfoil will not resist the torsion.  This seems counterintuitive to me.  Please take a look at my VABS input file and see if you can help me determine what may be causing the issue?  I have used strictly aluminum properties in this case to attempt to troubleshoot my difficulty.  The input file was generated from PreVABS, but then I altered line 3 (for initial twist) and added line 4 (Kappa1 value).

  2. Su Tian

    Hello Anthony,

    I am working on your case.

    Could you provide a smaller input file (< 10 MB if possible) producing the same issue? It would be easier to locate the problem.

  3. Anthony Ferman

    This file is smaller (I decreased the mesh size) but still generates a negative torsional stiffness value after processing with VABS.

    Thank you for your help in troubleshooting this issue.

    It is important to note that most of the stiffness values between these two files are close to each other, however the torsional stiffness values (although still negative) vary significantly from -9.7463526026E+08  to -4.8605839785E+08

    Initial file Timoshenko Stiffness Values (.005 Mesh):

    6.9906398784E+07    1.3233191702E+07   -1.4668557013E+06   -6.7059355339E+07    3.9479879759E+07   -1.3056662383E+09
         1.3233191702E+07    2.0629916108E+07   -2.0052454121E+04   -3.2159301495E+07    7.0974914709E+06   -2.4834862125E+08
        -1.4668557013E+06   -2.0052454121E+04    8.8240794942E+04    8.4627500206E+05   -8.3424181703E+05    3.6867272614E+07
        -6.7059355339E+07   -3.2159301495E+07    8.4627500206E+05   -4.8605839785E+08   -3.4756970201E+07    1.7273406007E+09
         3.9479879759E+07    7.0974914709E+06   -8.3424181703E+05   -3.4756970201E+07    8.8873992417E+07   -7.3173002233E+08
        -1.3056662383E+09   -2.4834862125E+08    3.6867272614E+07    1.7273406007E+09   -7.3173002233E+08    3.2142500783E+10

    Smaller file Timoshenko Stiffness Values (.050 Mesh)

    7.0569204238E+07    1.3398894844E+07   -1.4652899631E+06   -6.6325376937E+07    3.9860231261E+07   -1.3188084722E+09
         1.3398894844E+07    2.0282604677E+07   -1.7863384209E+04   -2.9236322395E+07    7.1867210256E+06   -2.5041333069E+08
        -1.4652899631E+06   -1.7863384209E+04    8.4538957472E+04    7.3634005789E+05   -8.2962653009E+05    3.6838740464E+07
        -6.6325376937E+07   -2.9236322395E+07    7.3634005789E+05   -9.7463526026E+08   -3.4608144105E+07    1.7155637268E+09
         3.9860231261E+07    7.1867210256E+06   -8.2962653009E+05   -3.4608144105E+07    8.9086002041E+07   -7.3937043602E+08
        -1.3188084722E+09   -2.5041333069E+08    3.6838740464E+07    1.7155637268E+09   -7.3937043602E+08    3.2407716300E+10


  4. Su Tian

    Hello Anthony,

    This issue may be caused by the trailing edge. The upper and lower surfaces meet at a single point instead of a line. This may cause problems of the assembled stiffness matrix in FEM. The weird trailing edge is created by PreVABS, and I will fix it later. A possible workaround is to create more base points near the trailing edge. Since only straight lines are used in PreVABS, you can add more points through linear interpolation. The goal is to create a line connection, instead of a point connection, at the trailing edge.

    Please let me know if this does not work.

  5. Anthony Ferman

    I've updated the input file to preVABS with additional trailing edge points to eliminate the point connection at the trailing edge, but VABS still generates a negative torsional stiffness for the cross section.  Any other ideas as to what may be happening here?

    Thank you.

  6. Su Tian

    Hello Anthony,

    After some discussions with Prof. Yu, the issue of negative stiffness with initial twist was observed years ago. The cause is very complicated and still obscure to us. Solving this issue requires more studies and changes in the theory and takes time. A workaround for now could be using multiple cross sections along the beam, and rotating each of them by an angle calculated from the initial twist.

  7. Anthony Ferman

    By this to do you mean input the angle into PreVABS and generate the mesh for each airfoil using their twist angle as opposed to adding the initial twist after the fact in the VABS input file?

    Thank you again for your help.