March 5, 2021 at 1:13 am #3677
Regarding to Floor Load features has been implemented, I have some curiosity. These are especially in two ways load transfer and distributed trough perimeter beams with non equal length ratio. NextFEM transfer as triangular with improper peak values rather than trapezoidal according to tributary area methods. This condition could be erroneous and unsafe if used as design input or investigated decisions.
Below an example calculation I mentions, also an input files attachment. Based on these quick review, it can be shown how triangular loads by NextFEM need to be reversed and reduced for such cases.
' Uniform Loads W = 10.0 kN/m^2; ' Length La = 4.0 m; Lb = 6.0 m; ' Shortest Length L = min (La, Lb) = 4.0 m ' Peak Loads q = (1/2) * W * L = 20.0 kN/m ' Check for Total Loads : Tw = W * La * Lb = 240.0 kN ' Floor Loads by NextFEM qa = 30.0 kN/m; qb = 20.0 kN/m; Tw1 = qa*La + qb*Lb = 240.0 kN ' Floor Loads by Tributary Area Methods Lx1 = La/2 = 2.0 m Tw2a = q*Lx1*2*2 = 160.0 kN Lx2 = Lb - 2*Lx1 = 2.0 m Tw2b = q*Lx2*2 = 80.0 kN Tw2 = Tw2a + Tw2b = 240.0 kN ' Floor Loads by NextFEM (if reversed only) qa = 20.0 kN/m; qb = 30.0 kN/m; Tw3 = qa*La + qb*Lb = 260.0 kN ' Percent Differences in Total Loads pdiff1 = (1-(Tw/Tw3))*100 = 7.7% ' Percent Differences in Shear Forces pdiff2 = (1-(40.0/45.0))*100 = 11.1% ' Percent Differences in Bending Moment pdiff3 = (1-(76.67/90.0))*100 = 14.8% ' Floor Loads by NextFEM (if reversed and reduced) qa = 20.0 kN/m; qb = 26.66 kN/m; Tw4 = qa*La + qb*Lb = 240.0 kN ' Percent Differences in Total Loads pdiff4 = (1-(Tw/Tw4))*100 = -0.0% ' Percent Differences in Shear Forces pdiff5 = (1-(40.0/39.99))*100 = -0.0% ' Percent Differences in Bending Moment pdiff6 = (1-(76.67/79.98))*100 = 4.1%March 5, 2021 at 3:26 pm #3685
thanks for the hand-calculation comparison. You applied a “quadPlane” load, that distributes the forces “by centroid”, hence it will not generate trapezoidal loads. That’s why you see these peak loads, that are correct in a “centroid”-like distribution (in your hand calculation, Lx1=0).
If by “reversed” you mean that you have to press “Reverse” once the loading plane has been drawn, this happen because you have “Normal to the plane” option active: this means that, for vertical loading, you have to draw the plane in counterclockwise way in order to have it directed on -Z.
If this is not the case, please clarify what you mean for “reversed”.March 5, 2021 at 7:20 pm #3686
Apologize if I’m using unusual word ‘reversed’ and make misinterpreted. In load case name ‘FLR’ as shown in pictures of floor load distribution by NextFEM, shortest beam produce large peak values (30kN/m) and longest beam has low (20kN/m). This is far away from to tributary areas methods which I shown in load case ‘NU’, longest beam should be trapezoidal shapes with peak values equal to ‘q’ (20kN/m) and shortest beam should be triangular shape with the same peak values.
It may be okay if NextFEM using triangular shape for both longest and shortest beams, with some exception in properly distribution of peak values. I’m using ‘reversed’ word to notions these condition, a peak load 20kN/m and 30kN/m for longest and shortest beam need to be change by switching each other. Shortest beam should have a peak values 20kN/m and longest beam are 30kN/m.
I applied these values in load case ‘NUrev’, please refers to input files I previously attached. This switching values method between shortest and longest beam not actually best approach and correct, total reaction or loads (260kN) is higher than the panel load itself (240kN).
Based on these above conditions so a reduction is required, since load in shortest beam (20kN/m) was correct according to tributary loads then reduction is applied to longest beam (30kN/m) and founded a proper value is ~26.66kN/m. I checked again by create load case ‘NUredrev’ and after running it shown reliable results. Only bending moment has small differences (~4%), total loads and shear force are identical compared to results by tributary area load methods ‘NU’.March 6, 2021 at 11:16 am #3693
Thanks for clarifications, this will be taken into consideration and let you know.March 7, 2021 at 6:33 pm #3702
We implemented a new method to distribute floor load to beams based on quadrilateral 2-way distribution (classic tributary area method). This will produce your NU distribution.
It will be published in the next patch.March 7, 2021 at 8:31 pm #3703
Great news, many thanks for appreciation to consider these features.March 10, 2021 at 3:58 pm #3704
I have been doing minor updates for these features. However, it seems these new twoWayQuadPlane does not working properly on spitted members. please refers to attachment file inputs. This problems does not happen when using old quadPlane methods.
Splitting a beam members is required in using OpenSees as as solver to get spans member outputs, or it need to be captures a second order effect.
Regards,March 10, 2021 at 7:27 pm #3706
Thanks for the tests, we extended the support in the patch released some minutes ago.
For usage with OpenSees, remember that equivalent point loads works only with Euler-Bernulli beams (deselected the proper option prior to export) and for fiber elements. Point load inside beams are not supported in OS by elastic Timoshenko beams.March 11, 2021 at 10:09 am #3707
Now, it’s working properly for split members. Also, thanks for the hint in compatibility problems of Timoshenko beam element using OpenSees solver. Answering my previous question in strange graphical output cause of large discrepancy between these solver, default’s shear deformation activated makes OpenSees result shown smaller values.
- You must be logged in to reply to this topic.