Member

Hi,

I am trying to simulate a single person walking along a floor which spans 9 m. I am modelling the human footstep force in the time domain. I have created nodes for each footstep based on the pace length of 0.76 m starting at 0.32 m up until 8.64 m. I am assuming the walker steps off the floor after the last footstep and also would like to observe the response of the floor after this point. The path is in the centreline of the floor:

To do this, I have made a transient analysis (MSUP method) with 14 load steps (1 load step before the footsteps start, 12 load steps for each footstep and the last load step to look at the free vibration response) – see below. Walking pace is 2 Hz so the length of each time step is 0.5 secs. I have output the acceleration response at certain node locations which were my accelerometer locations in my lab tests.

What I expect the response to be is that as the walker walks across the floor, the highest response would be at mid-span (as the first mode is the first bending mode). However, as shown below, when the walker is at mid-span is actually the lowest. Have I missed something here in the analysis settings that is causing this issue?

Any help or guidance would be much appreciated!

Thank you.

• Member
edited December 2018

When the load is applied at the quarter point of the beam, the center gets a larger displacement than when the same load is at the center due to the leverage of the load from the simply supported end.

• Member
edited December 2018

Is it possibly a result of my application of the walking force as a 'nodal force' rather than a 'force' on the surface?

• Member
edited December 2018

In the physical experiment, the footfalls also come with some mass, and the mass at the center has a larger effect on the displacement than in the simulation, where there is no added mass with the footfalls, just the force.

To make the simulation closer to the physical experiment, have a point mass move along with the footfalls. Not sure how to do that in a continuous transient simulation, hopefully you can think of a way or some other member might suggest how to do this.