12 Replies Latest reply on Feb 23, 2016 2:53 AM by Torkil Bladt

    Wrong reactions in report

    Giulio Malventi

      Hello, when I generate a report from a nonlinear analysis the reactions do not match the loads. Anyone else have this problem? Could it be that the values are extracted from the wrong time step? Using SW 2013.

        • Re: Wrong reactions in report
          Jared Conway

          Be more specific. How are you measuring? How do you know it is wrong ? How have you checked the software?


          I would recommend a test case to confirm it is software and not expectations or setup.


          For example a block fixed on one end and a force on the other.

          • Re: Wrong reactions in report
            Amirtharaj Vk



            You may use time curve to define load with respect to time. You need to check peak reaction force during the time period where the load peak takes place. If you are using contact with friction, you have to account, friction forces also. If you are using connectors with pre-loads, you have to account those pre-loads also.



            V K Amirtharaj

            EGS Computers India Pvt Ltd



              • Re: Wrong reactions in report
                Giulio Malventi

                Hi, my load is constant. I am using nonlinear because I am doing a plastic analysis. I am using none of the others.

                  • Re: Wrong reactions in report
                    Jared Conway

                    so where are you at relative to your original problem?


                    note, just because you are using plastic does not mean you need to use nonlinear. if you are in the linear part of the curve and do not have large displacement, it is unlikely that nonlinear will provide you any improved accuracy.

                      • Re: Wrong reactions in report
                        Giulio Malventi

                        Jared, not sure what you are asking. I am looking to determine the the minimum breaking load of a steel structure, so I have large displacements.


                        Could you please go back to my question?

                          • Re: Wrong reactions in report
                            Jared Conway

                            With a nonlinear analysis, there will be no breaking. Simulation does not support fracture and failure. Nonlinear provides large displacement analysis and a nonlinear E based on your curve. If using plasticity, it also provides plastic deformation. I provide this information because while you may not have asked about it, it may not be worthwhile to go down the path that you are going down.


                            Regarding your original question, you still have not answered my question or provided an example.


                            Where are you reading it?


                            How are you reading it?


                            Screenshots and steps taken would be helpful.


                            Then upload an example with a known solution. This way we have something common to look at, can compare between versions and computers and also can help you confirm the problem is or isn't software related .

                              • Re: Wrong reactions in report
                                Giulio Malventi

                                Jared, thanks for your concern. I am following DNV-RP-C208 and intend to use strain criterion to determine failure.


                                As for the original question, here is a screenshot of the report I was talking of. As you can see, the applied force is 8500000 N and the reaction 85000. This happens only with nonlinear, static is as expected.Senzanome.jpg

                              • Re: Wrong reactions in report
                                Bill McEachern

                                A Large displacement analysis may not be required if the displacements are not large but I would use it if there is uncertainty with respect to the stiffness of the structure changing due to changes in geometry as the load is applied- this is not a certainty in this case depending on the ductility of the steel (assumed) in question. If it were me I would have nlgeom on and you also need a very good approximation of the true stress strain curve. A reasonable estimate would be when the average strain or stess at the section normal to the load direction is at the failure criterion. Assessing this requires the gemoetry to be split so the section can be defined and assessed.

                        • Re: Wrong reactions in report
                          Bill McEachern

                          If I were you I would plot the reaction forces as function of time and see what the behavior is over the load application cycle. It certainly looks like it is reporting a time step that is not t=1 (assuming that is the "time" that corresponds to full application o the load.

                          • Re: Wrong reactions in report
                            Torkil Bladt

                            I have experienced the same problem in SW 15 (and earlier versions)

                            I think that the report generator uses the reaction force from the first time step wich is 1% of full load.

                            I am applying a pressure  of 90.2kPa  to the internal of thin walled duct piece (shell model), calculated as fixed at one of the round flanges:



                            If I do a non linear analysis I get the following loads in the report:


                            But a linear analysis with the same load, gives:


                            The rectangular inlet has a dimension of 1150x450, so a hand calculation of the X force shows that the last result is correct.

                            The difference is a factor 100, corresponding with the first time step of 1% of full load.

                            The other reactions are not exactly a factor 100 wrong, I think that it is due to the large deformation, wich is different in the two cases.


                            I hope that this explains the error, but I do not know how to solve it.

                            Best regards

                            Torkil Bladt, M.Sc. Denmark