brtietz
commented
8 years ago The actual length is the actual distance between nodes. Rest length will change based on pretension.
We assume that the system is a linear spring with a cable attached to it (F = k * dx + bv), which is true for most of our robots. dx is the difference between rest length and actual length. The cable is assumed to be stiff in tension. Therefore, we assume constant, linear stiffness. See the attached figure for a visual example.
If you increase the rest length, RL is closer to AL, and force will decrease since dx is smaller - excluding any effects of other rods.
As far as benchmarking - the files in test_integration are what we use. The simplest one for this case is the TimestepIndependence test, which is based on src/dev/btietz/timestepTest In this case, you could turn off the motor and make sure gravity stretches the spring appropriately.
On Sun, Oct 4, 2015 at 11:10 PM, chiku00 notifications@github.com wrote:
I tried to figure out why there was a decrease in tension when I increased the rest length. I thought that was due to the fact that stiffness = Youngs_Modulas*Area/Rest_Length. Since rest length increased, stiffness decreased. So the force might have decreased for the same amount of extension. But it could also be that the amount of force being transmitted through the ties had changed due to the re-orientation of the ties and struts. Fair enough. But, then I noticed that the rest length initially wasn't the same as the one specified via the initial nodal positions. Node 3 = (0, 0,10); Node 4 = (-5,10,0); So, the initial rest length should be 15. But it's not. It's around 10.07. Why this discrepancy? Or is that not how the initial rest-length is decided? Does that mean 'AL' in the log file does not mean 'Actual length', i.e. 'Actual Nodal Distance'? Moreover, I had set the stiffness value at 100000 in 'actuatorConfig', but the initial stiffness is being shown as (after calculation, see file below) as 97485. Why is that? I'm doing all this to make sure that the software is giving me correct results. If you have some sort of benchmark that you used to validate your code, could you direct me to it?
— Reply to this email directly or view it on GitHub https://github.com/NASA-Tensegrity-Robotics-Toolkit/NTRTsim/issues/179.
chiku00
I tried to figure out why there was a decrease in tension when I increased the rest length. I thought that was due to the fact that stiffness = Youngs_Modulas*Area/Rest_Length. Since rest length increased, stiffness decreased. I tried to validate this by comparing the relative change in the stiffness to that of the relative change in the rest length (see the excel sheet below for the formulas used). They do not match. Then what other factors have contributed to the change of the stiffness of the ties? The stiffness is showing a -3% change while the corresponding change in rest length is 12%. Is the area of the string changing due to Poisson ratio (non-linear)? Youngs Modulas should remain constant for the ties (not that it is being declared or anything).
Also, I had set the stiffness value at 100000 in 'actuatorConfig', but the initial stiffness is being shown as (after calculation, see file below) as 97485. Why is that? I'm doing all this to make sure that the software is giving me correct results. If you have some sort of benchmark that you used to validate your code, could you direct me to it?
https://www.dropbox.com/s/lzdh2febwc4h7u3/Actuator%20A%201.xlsx?dl=0