Issues with 3-2-1 Constraint Method

[Jieve]

Hello,

I’m conducting a linear static analysis on a ribbed plate made of 6082-T66 Aluminum and am having issues with 3-2-1 constraint method. I have applied balanced forces to the plate based on my FBD, all are pressure loads, and used the 3-2-1 method of constraint. While the sum of the reaction forces is small, something like 4N, I’m getting high localized forces at the reactions, up to almost 300N. There may be some small round-off due to slight discrepancies between FBD force locations and the centroid location of the pressure forces, but these should be negligible. Any idea why this might be happening? I’m not using large displacement. The software is Solidworks Simulation 2011.

Also, despite all of the posts I’ve read from this forum and the stuff from the Roshaz website, I still cannot convince myself of the validity of the 3-2-1 method for displacement analysis… or at least that one can choose any 3 points in a plane. If I am essentially constraining 3 points to lie in the same plane, without knowing for sure that these points should remain plane after deformation, how can I know that the deformation is correct? If I choose 3 points on a plane at the center of the plate, the edges of the plate can bow upward. However, restraining 3 points toward the outside of the plate forces those to remain in the same plane, giving a different result. Any have a good explanation?

Thanks for the input.
Jieve

 

[franc11HD]

Jieve,

AS far as the issue with 3-2-1 constraints giving you localized reactions, have you checked the reacting moments? It is usually the moments that creates those localized reactions. Maybe there is one bending moment that is not constraint correctly.

As far as your concern about displacements, I came to the same conclusion as you. It is good for stress and load distribution but as far as displacements are concerned, it is not appropriate. I convinced my self with a simple cantilever beam. I did FBD then changed the constraint along the length of the beam and it gave me different displacements, but the stresses and load distribution were the same. I thinkg it is because the local model is taken out of context. for displacement analysis, you would need to put it in an assembly or give the right boundary conditions in order to represent how the model behaves in terms of displacements.

 

[rb1957]

doesn't 3-2-1 refer to translational constraints ?

if you're applying balancing loads then there Should be zero reactions at the constraints, no local stresses, no load, no nothing (other than what would be happening in the structure due to the loading).

i'd review your loads/reactions. maybe run a test of your model with a single point force (and its reaction, sharing a common line of action).

 

[Jieve]

Thanks everyone for the input.

Yeah, 3-2-1 refers to translational constraints. Since it is a solid model, the nodes were fixed translationally so there are/can be no reaction moments at the constrained nodes. I plugged the numbers back into my FBD and they come out ok. I find it weird that the sum of the reaction forces in the vertical direction are almost equal to zero. Isn't that the equivalent of saying that the loads ARE balanced? I'm thinking that the problem is that the force locations on my FBD are slightly different than on the model due to the pressure loads (I assumed them to be at the centroids of the areas but there were some areas with small dowel holes that I ignored), and there must be some large moment arm or something causing the large forces. I'll have to re-check this.

From a load standpoint it makes sense that having no reactions would minimize the effects of the reactions on the stress distribution of the model. But wouldn't confining the displacement of the three points to a plane do the same? The stress is ultimately calculated from the deflections, it would seem to make sense that if the deflections are confined to a specific shape, then so is the stress contour. I'm still not finding this entirely clear.

Jieve

 

[rb1957]

maybe they're balanced in that there's no Z component ... i suspect you don't apply a Z- load ?

since you understand where you're applying load and where you're reacting it ... you could start by replacing your reaction forces with x- y- constraints.

else, look at the X- and Y- constraint forces ... they'll show you how much you need to add to your reaction forces.

they points you constrain (with your 3-2-1 constraints) only define the orientation of the body in space, they don't react any load. changing them only rotates the body. it'll only affect the problem if you want the deflection of the body at a node, with respect to the outside world; then you'll need to be sure that you're constraining the part in a real world sense.