Knee Joint Reaction Analysis problem with force.sto ouput file from Static Optimization

[Lorenzo Capetta]

Hello everyone,
I'm encountering an issue with the Joint Reaction Analysis tool applied to the knee joint. If I'm not wrong, we expect the moment around the flexion axis to be near 0, but this isn't happening.

I conducted an analysis using the Joint Reaction example from the Examples and Tutorials section of simtk.org.

After some research, both on the internet and here on the forum, I figured out that the Analyze tool is pretty sensible to its inputs; more in particular, the last columns of the the force.sto file (given in output by the Static Optimization tool), which are the columns containing the external forces, should not be taken into consideration by OpenSim while executing this analysis.

However, removing these columns leads to results that are not only different, but also closer to the expected results.
Through the forum I also found the hypothesis that the Analyze tool may consider the external forces two times (since it requires them as input, but they are also contained in the force.sto file). Looking at this idea, I tried the four possible scenarios:
1. Analyze with both external_forces input and external forces in the force.sto file (complete pipeline.png).
2. Analyze with external_forces input, but external forces removed from the force.sto file (cleaned force_file.png).
3. Analyze with no external_forces input, but external forces in the force.sto file (analyze with no external forces.png).
4. Analyze with no external_forces input, and external forces removed from the force.sto file (cleaned force_file + analyze with no external forces.png).

I will copy in the attachments the plots of these four combinations. As you can see, each one is different from the others.
I also noticed how this happens only for the knee joint. The hip, for example, as expected, gives moments euqal to 0 in all the four possible cases; also for the ankle the results are the ones expected (moments near to 0 around the flexion axis).


Am I focusing on the wrong file? Am I misunderstanding of the Analyze tool, or could this be an issue with OpenSim itself?