Significant reactions at constraints in spite of model in equilibrium

[matsS5MN7]

I have a model that is loaded with two opposite forces equal in magnitude, acting at a common point but at different surfaces. Both forces are modeled as "remote forces", acting at a common point in "mid air" but that have opposite direction and are acting on different surfaces. In reality there are other components carrying these loads, but for the context here, I don't need to model those. The forces are modeled relative to a local coordinate system, with components Fx=1e6, and Fx=-1e6. 

So the net load on the component is zero - the forces cancel out. In spite of this, I get high reactions at the constraints, and high stresses near the constraints. 

This is a modeling approach that I have used in the past, both with Ansys and also with Solidworks simulation, and there it works just fine. 

What's wrong here? 

[John_Holtz]

Hi @matsS5MN7 

Lots of questions. You should attach the model if you can.

I'm not sure what the circled areas are in the contour plot, but I get a sense that the model is twisting. In other words, the loads may be equal and opposite, but they create a large moment. That moment is resisted by the constraints which causes the large reactions and stress.  Is that true?

John

[matsS5MN7]

That is correct, there should be zero net moment and net force acting on the model, so there should be no bending of the horizontal, flat section where the constraints are located. I can't share the model for IP-reasons, but I will create a small sample model, to illustrate the problem. 

[matsS5MN7]

Here is a sample model that deforms the way I want it to, but I had to resort to using "inertia relief" option... Shouldn't  this structure deform correclty also without "inertia relief" and regular constraints at the holes where the constraints are located? (Now deactivated in "Subcase 1")

I'm thinking with small displacements /linear theory and loads cancelling out (equal magnitude, opposite direction, same point of action) , should give zero reactions at constraints? 

(Version 2026)

[AstroJohnPE · Accepted solution]

Thanks for the model.

You asked "Shouldn't this structure deform correctly also without "inertia relief" and regular constraints at the holes where the constraints are located? (Now deactivated in "Subcase 1")".

It depends on how you are constraining the holes. (Since you have a rigid connector in each hole, perhaps you are constraining the center of the connector, maybe pinned or not.) The general answer is no: the deformed shape with constraints is going to be different because it takes force to prevent the holes from bending. The inertial relief is a good test and provides the answer to your question. This is the displacement with no constraints:

If you now imagine forcing the holes to have 0 displacement, it is going to create stress to bend the shape back. 

Or another way to think of it is as follows: you can constrain at most 3 nodes and not change the strain. (Three points define a plane. The displacement values will be different because you are "rotating the deformed shape" to agree with the chosen points.) But even with 3 point support, you have to be careful about which directions you constrain in each of the 3 nodes. Once you constrain more than 3 nodes, you have to warp the above shape to conform to all the points.

You asked "loads cancelling out (equal magnitude, opposite direction, same point of action) , should give zero reactions at constraints?"

The answer is the sum of the reaction forces is zero. (23 N total out of 2,000,000 N is numerically zero.) I deleted the rigid connectors and fully fixed the inside face of the hole.

This does not mean that the force at individual nodes is zero. Each hole has a moment in order to keep the flange fixed, and therefore the reaction force most likely has a positive and negative force in each hole. The reaction at any one hole is not zero.

John

[matsS5MN7]

Another strategy I just tested, and that works, is to support the model with relatively flexible springs. Drawback: if the springs are made too flexible, it becomes difficult to visualize the deformed state, since the structures' "rigid body motion" becomes much greater than its elastic deformation.  

[John_Holtz · Accepted solution]

Hi @matsS5MN7 

I just noticed another issue in your analysis. Although the sum of the reactions forces is zero (20 some Newtons), the reaction moment is quite large. This could be due to an issue with the remote force load. See Results of an analysis are unexpected when using a remote force in Inventor Nastran. 

I changed the setup from using a local coordinate system with the remote load defined in the local X direction, to using a global coordinate system with the remote load in X, Y, Z components to duplicate the local coordinate system. The moment reaction is small as expected. (The largest von Mises stress did not change much.)

Another option I tried in a test model was to apply the load to an interpolation "rigid connector". (I separated the two center nodes of the connector by a small distance along the line of action). This test model also gave a small moment reaction as expected.

I suggest you use either of the two alternate loadings in your real model to confirm the moment reactions are close to zero and the stresses do not change. (I did not check the other components such as principal stress.)

John

[matsS5MN7]

Thanks @John_Holtz and @AstroJohnPE , this was helpful.