How to make a revolute joint turn in one direction in Simscape?

Hello all,

I am building a bicycle on Simscape. The tyres are connected to axles which in turn are connected to the bicycle frame, and I used a revolute joint to connect the tyre and the axle (as shown in the picture below). I would like my tyres to only turn in one direction (clockwise) but it always turn in both senses when I simulate it. I saw another post saying that we needed to put a Rigid Transform before and after the Revolute joint to change its direction and I tried it but it didn't work (the only thing it did was make the tyre go clockwise or counter-clockwise at first depending on the rotation parameters of the joint, but the tyre still kept going both senses).

Could someone help me on this matter please? I am still new to Matlab and Simscape and I am learning new things and how to use it as I progress in my project, but that problem has kept me stuck for quite some time now.

Thank you very much.

Here is a screenshot of the model built with Simscape Multibody. I apologize for the the way it looks as it is quite messy but I hope you can still understand it.

Answer 1

[Yifeng Tang]

There may be better solutions, but just some ideas.

If you have access to Simscape Driveline, there is a one-way clutch (Unidirectional Clutch) that only transmit torque in one direction. When one end is fixed to the ground, the rotation become only clockwise, or counterclockwise, depending on how you set up the block. You can use this clutch block together with Multibody. Something like this:

Check out the documentation page of that Rotational-Multibody Interface block if you haven't used it before.

This does require you go between 1D Simscape world and 3D Multibody world.

Another possibility I can think of is to use the "mode" in the Revolute Joint. You can go between "normal" and "locked" mode based on an input. Enable that in the Revolute Joint dialog. Then you can set up algorithm externally to determine when to "lock" the joint, i.e. when it attempts to rotate backward (velocity close to zero, torque negative). You also need to set up logic for it to unlock (e.g. velocity zero, torque positive). And be careful with a small buffer region to ensure the trainsition is smooth.

Again, just some thoughts. I've tried neither myself.

[Henintsoa]

Hello, thank you very much for your answer. I tried your first recommendation and I got this error:

An error occurred while running the simulation and the simulation was terminated

Caused by:

• ['BicycleF4/Solver Configuration']: Transient initialization at time 4.42857148993639, solving for consistent states and modes, failed to converge.
• Nonlinear solver: Linear Algebra error. Failed to solve using iteration matrix.
• The model may not give enough information to make it possible to solve for values of some of its variables. Specific advice is given below. all components and nodal across variables involvedCannot solve for one or more variables: 'Rotational_Multibody_Interface1.t' (t) 'Rotational_Multibody_Interface1.torque' (Torque) 'Rotational_Multibody_Interface1_t0' (Rotational_Multibody_Interface1_t0) 'Unidirectional_Clutch.fundamental_clutch.clutch.t' (Torque) 'Unidirectional_Clutch.fundamental_clutch.torque' (Torque) Equations (including switched-linear equations) of one or more components may be dependent or inconsistent. This can cause problems in transient initialization. Here is the set of components involved: all components involved'BicycleF4/Rotational Multibody Interface1'Equation location is: 'C:\Program Files\MATLAB\R2023b\toolbox\physmod\simscape\library\m\+foundation\+mechanical\+interface\rotational.ssc' (line 42) 'BicycleF4/Unidirectional Clutch'Equation locations are: 'C:\Program Files\MATLAB\R2023b\toolbox\physmod\sdl\sdl\+sdl\+clutches\+fundamental_components\one_way_clutch.sscp' (no line number info) 'C:\Program Files\MATLAB\R2023b\toolbox\physmod\sdl\sdl\+sdl\+clutches\+fundamental_components\one_way_clutch.sscp' (no line number info)

As for your second recommendation, I am still figuring out how to do it. I also tried to put a constant torque at the port t of the revolute joint instead of linking it to the port t of the Rotational Multibody interface, and I didn't get any error but I am not sure it resolved the problem.

[Yifeng Tang]

Hi Henintsoa,

Are you able to share a model for diagnostics?

[Yifeng Tang]

Oh, I also realized that the "lock" mode for the joint is only available in R2024a. You may need to try it there. Otherwise it won't lock up to prevent back rotation.

[Henintsoa]

Hello again,

Here are the models. The first one is the one where I did what you said (at least I hope so) which is the one with that error, and the second one is where I added a constant at the port t. I apologize for the messiness of the model. Also, there is an electric part linked to the rear tyre. It is not supposed to go together with what I am trying to do here but for convenience purposes I decided to see how to make the wheel turn with the front wheel, and once I figure it out I will use it on the rear wheel without the electrical system that is there. Thank you very much.

[Henintsoa]

I have the student version of the R2023b, do I need to uninstall it and download the R2024?

[Yifeng Tang]

It's OK to have multiple versions installed on one computer. I have 8 :p

[Henintsoa]

Oh I see I see, okay, I will install it, thank you

[Yifeng Tang]

Hi Henintsoa,

Do you think the attached model does what you envision?

It's a wheel rotation on the ground and there is a force pushing on the shaft of the wheel so it may go forward or backward. I put a feedback force on the revolute joint such that when the rotation is positive, no torque is applied, but when the rotation is negative, a torque proportional to the angular velocity will be applied. So it's like a large one-way damper. The directionality is achieved using a simple lookup table and I include two tables to allow rotation in either positive or negative direction.

What you think?

[Henintsoa]

Hello,

Thank you so much for your answer and for taking the time to make this. Yes, this is what I was looking for, thank you so much. I'm sorry but could you please tell me how to make the wheel go faster? I increased and decreased the torque (the gain) but it goes very slowly no matter what.

[Yifeng Tang]

The gain is there to adjust the "strength" of the one-way brake as a large damping coefficient. It's not what makes the wheel move. The force (PS Step) applied on the rectangular joint (i.e. on that shaft) is what pushes things forward. Then the friction between the wheel and the ground make the wheel rotate. Change the final value of the PS step will make the wheel go faster.