Challenge with Simulating Loader Mechanism with Closed Kinematic Loop

I'm currently working on simulating a loader mechanism that includes a closed kinematic loop in its arm, actuated by two jacks. From my research, I understand that conducting a full multibody simulation in systems with a closed kinematic loop while all joints are actuated is impossible. Therefore, one of the actuators must be converted to an unactuated joint (Here is the Link for reference).

To address this, I left the right jack of the loader unactuated in my simulation and forced the left side to stay in a specific position, effectively conducting a static analysis (inverse dynamic) and recording the calculated forces on other joints, such as those between the chassis and the arm. Something like pictue below:

Here is my challenge:

1. When simulating the arm with only one active jack, the forces generated on the unactuated side of the loader are not accurate because the unactuated jack does not apply any force to the mechanism, which it does in reality. Am I correct in this understanding?
2. If so, how do others typically address this issue in their simulations?

Any insights or suggestions on how to properly simulate this mechanism would be greatly appreciated!

[Yifeng Tang]

Did you ever TRY to actuate both joints? If you provide forces as inputs, or identical positions, I suspect it may just work OK. I've definitely seen models that do that. If you use Translational Multibody Interface (available since R2021a) to actuate the joint from 1D Simscape hydraulic/isothermal liquid network, you should be able to do that with both joints as well.

[Medalan]

Thank you @Yifeng Tang for your answer, and I apologize for the delayed response.

Actuating both joints does indeed work in general, but unfortunately, it does not meet my specific requirements.

Here's what I am trying to achieve:

My goal is to determine how much load the bucket can carry at specific positions before the hydraulic jacks reach their maximum force, as well as to calculate the forces generated in the joints of the mechanism. This requires me to fix the position of the mechanism, apply an external force to it, and then calculate the forces on the jacks. In my model, the jacks are represented by prismatic joints.

If I set both joints to be actuated, I would need to determine the appropriate forces to achieve the desired positions. I’ve tried resolving this issue using a PID controller to set the jack lengths to the target positions. However, this approach feels inefficient and impractical, as it requires constant PID tuning whenever the external load changes.

Therefore, I am seeking a solution based solely on the inverse dynamics method to calculate the joint forces accurately without relying on actuating both jacks. Unfortunately, I am working with version R2020b, which limits some of the more recent functionality you mentioned.

[Yifeng Tang]

I wonder how you "... calculate the forces on the jacks" right now. When you provide a motion (position input), you can turn on the sensing port for the actuator force in the prismatic joint and find out how big the force is at any given moment. If you keep the motion constant, you'll get the static force.

If all the joints are "planar", meaning prismatic, revolute, etc., it's indeed not necessary to actuate both prismatic joint. One is sufficient to support the whole mechanism. That also means the force needed to actuate one should be exactly twice of what it takes to actuate both. You can try prescribe position to one or both joints and see what values you get from the joint force sensing port.

[Medalan]

Hi again, @Yifeng Tang!

Thank you for your response. I apologize if my previous explanations were unclear. To clarify, I am indeed following the approach you described. For a better understanding, let me provide an example based on a skid loader model:

As you mentioned, for both the arm and bucket jacks, I have set one of the prismatic joints to be actuated and left the other unactuated.

Your logic regarding force calculation is correct, and I am using this strategy in my project. However, I am encountering a challenge related to the interaction forces in the revolute joints (as shown in the picture below). I need these forces as input for FEM analysis to ensure that the designed part is strong enough.

This strategy works well when the external force is applied to the plane of symmetry of the model. However, when the external force is applied to the corner of the bucket, I believe the logic doesn't hold. While the model runs without issues, the calculated forces in the joints do not seem realistic, which makes them unreliable for FEM analysis. This is the challenge I'm facing.

Answer 1

[Yifeng Tang]

Hi @Medalan

Thanks for the clarfication on your question. Let's see if this idea helps.

I created two simple models to demonstrate the idea. Both models are pushing a block upwards using two prismatic joint:

In "example_actuate_one.slx", one of the joints is actuated with a motion signal. The other one kept at zero force and automatic motion. One of the sensed actuator force is zero and the other at whatever force needed to move the masses.

Then in "example_actuate_both.slx", one joint is still actuated with a motion, and the other one is actuated with a force. The force is taken from the sensed actuator force from the first joint with a PS transfer function (delay). The resulting forces are very close to each other, except for the initial transient, and the values are about half of the original one when actuating just one joint.

Maybe a similar setup can help you apply the right forces and find out the constraint forces at other joints?

[Medalan]

I checked your model and believe it's the solution! I tested it in my model, and the results appear logical and sensible.

Thank you very much @Yifeng Tang.

[Yifeng Tang]

Good to know. Happy modeling.