Constraining a body with spring elements

[bnret]

Hello everyone,
currently, I am having trouble sufficiently constraining a model of mine that consists of a large number of bodies. My thought was to use spring elements to slowly transition from a fully fixed support for each body to the bodies just being held by their interactions. Unfortunately, I seem to have misunderstood this concept completely, as I am failing to conduct such an analysis for single tetrahedral element. As far as I understand it, if each node of the element is connected to two grounded nodes with spring elements (and those nodes are not coincidental), rigid body motion should be prevented. This is not the result of my algorithm, as the static analysis fails with a pivot error and the modal analysis shows 0 as the lowest two modal frequencies, thus pointing at rigid body motion. I am very confused by this and I do not see at which point I misunderstood the concept of those spring elements. I am glad for any help/support provided here!
This would be a minimal (not) working example of my problem:

from ansys.mapdl.core import launch_mapdl

mapdl = launch_mapdl(print_com=True)
mapdl.clear()
mapdl.prep7()

mapdl.et(1, 187)  # solid
mapdl.mp("EX", 1, 10)
mapdl.mp("NUXY", 1, 0)
mapdl.mp("DENS", 1, 1000)

pos = [[-0.118498779833,0.0674275234342,0.0222713090479],
[-0.120090339784,0.0670430315701,0.0226928617557],
[-0.118498779833,0.0674275234342,0.0237713092938],
[-0.118498779833,0.0684275254607,0.0232713092119],
[-0.119294559809,0.0672352775021,0.0224820854018],
[-0.119294559809,0.0672352775021,0.0232320855248],
[-0.118498779833,0.0674275234342,0.0230213091709],
[-0.118498779833,0.0679275244474,0.0227713091299],
[-0.119294559809,0.0677352785154,0.0229820854838],
[-0.118498779833,0.0679275244474,0.0235213092528]]

for i,p in enumerate(pos):
    mapdl.n(i+1,p[0],p[1],p[2])

mapdl.e(1,2,3,4,5,6,7,8)
mapdl.emore(9,10)

mapdl.et(2, "COMBIN14", kop2=0, kop3=0)
mapdl.r(2, 5)
mapdl.type(2)
mapdl.real(2)

for i,p in enumerate(pos):
    mapdl.n(11+i,p[0],p[1],p[2]+0.1)
    mapdl.d(11+i,"all",0)
    mapdl.e(i+1,i+11)

    mapdl.n(22+i,p[0],p[1],p[2]-0.1)
    mapdl.d(22+i,"all",0)
    mapdl.e(i+1,i+22)

mapdl.allsel()
mapdl.finish()
mapdl.slashsolu()
mapdl.antype("MODAL")
mapdl.modopt("LANB", 6)
#mapdl.mxpand(6, "", "Yes")
mapdl.solve()
mapdl.finish()

mapdl.post1()
output = mapdl.set("LIST")
print(output)
mapdl.exit()

[mikerife]

Hi @bnret
I think you might have in mind that the spring elements constrain in all 3 directions; they do not as they spring force is only alongs its length. You would need 3 springs at each node, one each for each x, y, z direction.

Mike

[bnret]

Hi @mikerife
aah that makes sense, thank you! I was under the impression that when the node is connected with two springs, there is no way it could move without elongating one of the springs. But I guess that only applies to the solution if the solver is solving with respect to the current configuration, doesn't it? Therefore, turning on large deflections might have helped.
Or is the spring's force only applied in the direction of its original orientation, no matter what?
Anyways, I am glad to have found a way to make it work and very thankful for your help!

[mikerife]

Hi @bnret modal is always linear so the spring force is always in the direction of the spring as modeled. If the spring is along say global Z, and one end moves in gobal X and/or Y, the spring force is still in Z. Need a spring in X and Y too to really constrain in all translations...and not have them be colinear!