I have read thru the June 1999 paper describing the hinge and ball joints modelling and understand the concept and issues and have previously done a design based on this.

I am working on a new project and downloaded the referenced Barco.zip file to review the methodology.

In particular the Barco2 model is unclear to me and I would like if David D. could take a look at the current .zip file and explain in greater detail each of the nodes and in particular the restraint nodes and CNodes used.

Thank you

David D, Why is there a Y support at nodes 100 and 130?

I'm looking at BARCO2.
The ball joints, themselves, are modeled using restraints (using Node/CNode pairs) rather than pipe elements. Six restraints are required to completely define these connections - 5 rigids and one bi-linear stiffness (RZ2). The Y restraints at 100 & 130 (CNoded to 110 & 120) are required to hold the system together in the Y direction (just like the others in X, Z, RX & RY directions).

Thank you for your time.

Maybe I am not grasping the sequence of the nodes in the model

Here is what I see in the model
Element 80-90 is from the horizontal pipe to the elbow.
- Dx = 25 feet
- restraint on this element has node 90 with no support type indicated
- restraint on this element has node 100 with Cnode=110 as a type Rx and Ry

Element 90 - 100 is from the elbow downward to first ball joint
- Dy = -1 foot 11.25 inches
- restraint on this element has node 100 with Cnode=110 as a type Rz2, X, Y and Z

Element 110 - 115 is from the upper ball joint to the expansion joint
- Dy = -2 feet
- restraint on this element has node 110 with Cnode=120 as a type -YRod

Element 115 - 116 is the zero length expansion joint

Element 116 - 120 is from the expansion joint to the lower ball joint
- Dy = -2 feet
- restraint on this element has node 130 with Cnode=120 as a type Rx, Ry, Z

Element 130 - 140 is from the lower ball joint to the elbow
- Dy = 1 foot 11.25 inches
- restraint on this element has node 130 with Cnode=120 as a type Rz2, X and Y

Element 140 - 150 is from the elbow to the anchor
- Dx = 1 foot 3 inches
- restraint on this element has node 150 as a type Anchor


Please clarify
1. Element 80 - 90, please explain the restraint with node number 90 and no restraint type?
2. Element 116 - 120. Why is the restraint as node 130 with CNode=120 rather than as Node 120 with CNode = 130? Does this make a difference?
3. Please explain further about the y restraint holding the system together. I would model this without the Y type restraints at nodes 100 and 130. I thought the CNodes would "hold" the system together?

Thank you

1. That model was built before we had 6 restraint entries per spreadsheet. We had only 4, so 4 of the 6 restraints at 100 were entered on the element 90-100 and the other two were defined on the previous element 80-90. The restraint checkbox on 80-90 opens the restraint window but you can define any restraint in that window. There is no restraint at 90.
2. If the restraint is linear, the Node/CNode pair can be entered in either manner. If the restraint is nonlinear, this relationship is important - a +Y restraint at Node 10 with a CNode of 20 is the same as a -Y restraint at Node 20 with a CNode of 10. Reversing the Node/CNode pair will not affect this model.
3. Yes, these restraints hold the system together - but only in the direction defined. A single Anchor between 100 & 130 would work alone but we are manipulating individual directions so all 6 must be defined.

Thank you Dave,
1. So because 90 is entered in the restraint window and there is no restraint type indicated for node 90, does this mean that the program does not put any restraint at 90?
2. The Node/CNode explanation with the example (node 10 & 20) helps, however I will have to think more about this to fully grasp the reversal from +Y to -Y depending on the sequence of the Node/CNode pair. Perhaps you can describe this in more detail to help me better understand.
3. I now get the need for the Y restraint at nodes 100 & 130. Thank you for that.

Thank you

1. The check in the Restraints checkbox on element 80-90 indicates that restraints are entered but it does not imply that the restraint must be located at 90. Restraints have their node location as part of the restraint input.
2. A restraint without a CNode is what I would call an absolute restraint, the CNode is Earth and I cannot move Earth. We are working with a restraint definition between two nodes - this is what I would call a relative restraint. If I have a rigid +Y restraint at node 100 with a CNode defined as 101, this indicates that 100 is free to move in +Y away from 101. (I realize that this sounds confusing but it works.) I could also look at it the other way - 101 is free to move in the -Y from 100. So:
a +Y restraint @ Node 100 with a CNode 101 is identical to:
a -Y restraint @ Node 101 with a CNode 100

Dave,
2. Thank you for this explanation. That description helps me visualize the relationship.

Dave,
It seems I still do not understand fully.

I am trying to model an A-frame assembly / 3 swivel layout consisting of a RotaBall joint (axial rotation only), then a Flex Ball (full ball joint) and then another Rota Ball (axial rotation only).

I am trying to piece together the model using the information above, however after several failed attempts at modelling I need assistance. I know the model failed by viewing the animation and the +yrod and expansion joint on 1 half of the assembly are coming apart.

Attached is a drawing of a 3 swivel assembly.

CAESAR II is a stiffness program - you assemble a stiffness matrix to represent the system and then load that stiffness matrix.
Your three swivel (sub)system cannot be modeled using this sort of stiffness approach. As your system moves, the elements "re-orient" and the original global stiffness matrix would have to be reset for the new orientation. Your swivel subsystem is more of a linkage, I am sure there are approaches to address this sort of analysis but CAESAR II is not the best choice.
I call you subsystem geometry constrained. As long as the subsystem can achieve the imposed position of the piping, it works. The system would be safe/reliable if the loads associated with the original and final position are acceptable for both your subsystem and for the piping. (This is not unlike flexible hose.)
The ball joint example in the newsletter article works because there are only two nodes that are linked by geometry rather than stiffness. The geometry (or, the linkage) is maintained by that +YROD restraint, not elements in the stiffness matrix. The article points out the need to add a "slip joint" along the pipe span between the ball joints so that the axial stiffness of the pipe does not conflict with the rotation of that rod model. It works in relating two points in space but I don't expect that to function properly with three points (and two links).
I would approach this like a flex hose - run the system without the subsystem to find out where the pipe wants to go. Then make sure the linkage can accept that displacement and (from the manufacturer) get an idea of what load is created by the subsystem based on that calculated position. Apply these loads to your CAESAR II model of the piping to evaluate pipe stress and also use them to evaluate your equipment load.

Very good.
Thank you David for this insight