Here are some notes I prepared to help Strand7 beginners where I work. No responsibility for any errors. Apologies for the length.
- - - - - - Extract starts - - - - - -
Contact problems
Background
Strand7 can model problems where parts of the structure that were initially in contact move apart, or (alternatively) where parts of the structure that were initially apart come into non-penetrative contact. These situations are modelled using the point contact element. The two most commonly used types of point contact element are the normal contact element and the zero gap element, used in situations of initial contact and initial gap respectively.
With the zero gap element the entire length of the element represents the gap. The element will be inactive (ie generating no force) until the gap between its two end nodes is closed. When the gap is closed the element will generate an axial force whose magnitude will be such as to prevent penetration, and if friction is specified the element can also have a lateral force.
Normal contact elements
The normal contact element is capable of generating a compressive axial force within itself, but not a tension. It can also generate a lateral friction force when it is in compression. The modelling of friction is discussed separately in the final part of this section: all comments in the earlier parts assume that you have specified zero for the friction coefficients, except where it is specifically stated to the contrary.
The following points are relevant to the normal contact element.
[ul square]
[li]The elements should be directed perpendicular to the surface whose contact is being modelled.[/li]
[li]The element must have a non-zero initial stiffness specified, or it will be completely ignored.[/li]
[li]For static problems you should usually utilise the dynamic stiffness option, which removes the need for you to think much about what value to use for the initial stiffness. For dynamic problems the dynamic stiffness option does not apply (despite the apparent eponymity), so you will need to give consideration to your initial stiffness value.[/li
[li]You must use a nonlinear solver.[/li]
[li]It makes no difference whether or not you invoke the solver's nonlinear material option (unless some other aspect of your model requires the use of the option).[/li]
[li]If you invoke the solver's nonlinear geometry option, you need to be cognisant of the setting for update direction in the element definition.[/li]
[ol circle]
[li]By default, update direction is NOT operative. Under these circumstances the contacting surface is assumed to remain perpendicular to the element's undisplaced direction, and the component of the element's displaced length along its undisplaced direction is used to determine whether contact is being made. So far, so good. However whilst the element's internal forces are (correctly) calculated relative to its displaced location, they are transferred to the surrounding structure as if they had been calculated relative to its undisplaced location. This leads to the structure not being in overall moment equilibrium, because the displaced element will generally not be parallel to the undisplaced element. Note that this dis equilibrium is due to the axial force, and will be additional to any dis equilibrium caused by lateral friction forces (an effect that is described below).[/li]
[li]If you activate the update direction option, the contacting surface will at all times be perpendicular to the element's displaced direction, and the element's true displaced length is used to determine whether contact is being made. There will be no out of equilibrium effect. If you have not specified any friction, the behaviour of the element when it is active will be the same as that of a pinned link.[/li]
[li]Direction updating is ignored unless the solver is set for nonlinear geometry.[/li]
[li]The writer has found it extremely difficult to achieve solution convergence for problems involving the quadrella of geometric nonlinearity, contact elements, direction updating, and non-zero friction.[/li]
[/ol]
[li]You should always check your results in regions of potential contact by looking at displaced shapes in true scale.[/li]
[/ul]
Zero gap elements
These are very similar to normal contact elements, except that they do not become active until the compression deformation in the element equals or exceeds its original length. All of the comments on normal contact elements apply, with the exceptions / additions / amplifications noted below.
[ul square]
[li]Unless update direction is being used, the issue of whether the element is active is based on the component of the element's displaced length along its undisplaced direction.[/li]
[li]The writer has found it extremely difficult to achieve solution convergence for problems involving the trifecta of geometric nonlinearity, gap elements, and direction updating. This applies irrespective of the presence or absence of friction. He suspects that this difficulty is somehow related to the fact that when contact occurs the length of the element has the potential to be zero, under which circumstances the gap direction becomes undefined. However the convergence difficulties seem to arise even in structures where the gap never closes perfectly, in which case one would expect the element to remain inactive. More investigation is required here.[/li]
[/ul]
Modelling friction
In association with either of the above-mentioned point contact elements, you can specify friction. This is done by entering a pair of friction coefficients, one for each of the element's principal transverse directions. Points to bear in mind.
[ul square]
[li]Any resulting friction force will be acting in the plane perpendicular to the element's axial direction.[/li]
[li]The two principal directions are treated independently and are uncoupled: if the relative lateral movement of the element's end nodes includes components in both directions, then the total friction force generated may be greater than the force that would be generated were the movement along one principal direction only.[/li]
[li]If a friction coefficient is equal to or greater than 1, a condition of zero slip will be enforced. In other words, the friction coefficient will actually be assumed to be infinite.[/li]
[li]Any generated friction force will be transferred to the rest of the structure as a pair of equal and opposite lateral forces at the element's end nodes. Since these forces are not colinear, but are apart by a distance equal to the axial length of the element, they apply a net external couple to the structure. Hence the structure will not be in overall moment equilibrium.[/li]
[/ul]
Thanks Denial!
It sounds great and I'm going to try to follow your advice, but I still have a question about normal contact elements between two surfaces: since the two surfaces are supposed to be in a situation of initial contact, shall I interface them with zero-lenght normal contact elements (is that eventually possible?), or shall I 'force' a miniminum, though finite, distance between the two surfaces? Or what?
Hope I'm not asking a stupid question ... I'm a beginner.
Hoping to get your precious advice soon, thanks in advance.
Anyway I'll let you know about my steps forward (if there'll be any).
Bye. Luigis
You will need to have a small but finite gap between the contacting elements, across which you put the "gap elements". Without this Strand7 will merge the coincident nodes at every opportunity you give it, and will delete the zero-length element.
Ok, it's very clear now. I have already made a big step forward. Now I'll have to work on it and to cope with convergence difficulties. It'll take a little time I guess, but I won't give up.
Thank you.
