* update: cross-section / segment model #5
Conversation
* update model
| @@ -1,5 +1,6 @@ | |||
| # Tutorial example simulating a 3D mass-spring system with a nonlinear spring (no spring forces | |||
| # for l < l_0), n tether segments and reel-in and reel-out. | |||
| using PyPlot | |||
| @@ -56,7 +57,7 @@ function model(se) | |||
| @variables spring_vel(t)[1:se.segments] = zeros(se.segments) | |||
| @variables c_spr(t)[1:se.segments] = c_spring0 * ones(se.segments) | |||
| @variables spring_force(t)[1:3, 1:se.segments] = zeros(3, se.segments) | |||
| @variables total_force(t)[1:3, 1:se.segments] = zeros(3, se.segments) | |||
| @variables total_force(t)[1:3, 1:se.segments+1] = zeros(3, se.segments+1) | |||
There was a problem hiding this comment.
I tried to understand the original code and found that this line is wrong:
eqs2 = vcat(eqs2, acc[:, i+1] .~ se.g_earth + total_force[:, i] / 0.5*(m_tether_particle)) <== brackets?
In my approach I assumed, that the mass of a segment is concentrated at the end.
As already mentioned:
Some details: there are N segments with (N+1) nodes with masses. Therefore (N+1) forces are needed to calculate the acceleration. For the first and last node the formulas are different. For all other, segment-forces from spring/damper are added.
There was a problem hiding this comment.
Well, the mass of a segment is NOT concentrated at the end... But if you think my approach is wrong, can you create a but report?
There was a problem hiding this comment.
Alternatively it would also be useful to have test cases...
There was a problem hiding this comment.
I tried to understand the original code and found that this line is wrong:
eqs2 = vcat(eqs2, acc[:, i+1] .~ se.g_earth + total_force[:, i] / 0.5*(m_tether_particle))<== brackets? In my approach I assumed, that the mass of a segment is concentrated at the end. As already mentioned:Some details: there are N segments with (N+1) nodes with masses. Therefore (N+1) forces are needed to calculate the acceleration. For the first and last node the formulas are different. For all other, segment-forces from spring/damper are added.
Can you perhaps just add your version as a new script, Tether_07b.jl ? Then it is easier to compare both versions.
There was a problem hiding this comment.
I'm working on a MTK/Modia3D comparison. For a special case I was running in a stability problem. See: ModiaSim/Modia3D.jl#136
There was a problem hiding this comment.
I prepared a summary of simulations which can be found on: https://github.com/johhell/TestTether.jl
- original MTK implementation
- modified
Tether_07a.jl(acc. PR) Modia3D- for comparison
There was a problem hiding this comment.
Thanks a lot, that is great! But I have to hand in a paper until 25th of January, not sure if I find the time to look it it earlier.
There was a problem hiding this comment.
After reviewing the MTK documentation, I now have a segment-based (not monolithic) model available that provides the same results but offers a better understanding.
The current models do not provide any damping. Rotational damping at each node seems to be very complicated to implement (refer to Modia3D).
To get a more realistic result, I added a velocity^2 dependent damping force for each mass segment.
This approach can also be used to study the effects of wind on the tether.
See here a first result of 15 segments model with friction-based damping
After some refactoring I will proved the model for discussion within the next few days.
|
updated version of the segmented based tether model new features are:
|
|
I added your commit as Tether_08.jl and fixed issues with the latest version of MTK and use ControlPlots instead of PyPlot. I also fixed most issues of Tether_07.jl. If you have the time, please check if the new version of Tether_07.jl matches your results. |
with ref to #4