One-Way and Two-Way Contact Definitions

This post will give you a brief summary of one-way and two-way contact definitions. The one-way/two-way refers to the treatment of slave nodes impacting the master segments.

One-Way Contacts
One-way was one of the first implementations in contact treatment and was done so such that it is computationally efficient. In the one-way contact treatment, the slave nodes are represented by discrete points with no physical connectivity between the nodes. The master is however defined is represented by segments (either 3 or 4-noded). In one-way, only the slave side is checked for penetration against the master segments. Since the master nodes do not undergo any checking, it may be possible that it could penetrate the segments or elements on the slave with no application of contact forces to prevent the penetration. Consequently, users are encouraged to define the “finer� mesh as the slave side and the coarser mesh as the master side to reduce the possibility of master node penetrating the slave side. Following figure shows the condition of contact involving a coarse-slave against a finer-master and a fine-slave against a coarse-master to illustrate the need for the slave side to be finer for better contact penetration detection in one-way contacts. Due to the nature of the one-way treatment, the resulting contact-impact is dependent on slave/master definition. Switching slave and the master sets, could result in different behavior depending on the geometry and the mesh sizes. Examples of one-way type contact definitions are all contacts with either NODES or ONE_WAY string present in the contact keyword. A complete list of one-way contacts are listed here.


Two- Way Contacts
Two-way contact treatment is identical to the one-way except that the penetration detection is symmetric. This requires that both slave and master definitions have connectivity information and can only be defined as either part or segments. In two-way contacts, first the slave nodes are checked for penetration against the master segments after which the nodes of the master segments are checked for possible penetration against the slave segments. This dual treatment eliminates the need for the user to carefully choose the slave and the master side since switching the slave/master will result in identical behavior. Two-way contact definitions are highly recommended unless the slave side is much finer that the master. All contact keywords with the string “SURFACE_TO_SURFACE� without “ONE_WAY� or having the string “TWO_WAY� are examples of two-way contact definitions. Below is list of two-way contact definitions available in LS-DYNA.


Tied Contacts
The terms one-way/two-ways also refers to all tied contacts and affects which nodes are selected for tying. In the case of tied contact definitions with the string “NODES�, only the slave nodes are processed to determine the closest master segments and if a master segment is found and meets certain criteria such as the projected distance is within certain “tolerance�, then the slave node is tied to the mater segment. In the case of two-way tied contacts, determined by the presence of the string “SURFACE_TO_SURFACE�, both the slave nodes and the master nodes are processed to determine “potential� segments to which the node can be tied. Like mentioned earlier, unless a slave side is much finer than the master, it is always recommended to use the two-way tied contact definitions to remove the dependence on the mesh disparity and also to ensure the two-surfaces (slave and master) are completely tied.



  • Sven says:

    When an _OFFSET option is included in a TIED contact, no nodes (slave nor master) are projected onto the segment it is tied to, and a virtual spring is set up from master to slave.

    With no _OFFSET option for a 1-way contact, the slave nodes are “projected” onto the master surface it is tied to.

    With no _OFFSET option for a 2-way contact however, how are nodes “projected” onto their tied surfaces?
    For example, can it be assumed that both slave and master node positions are likely to be adjusted by dyna? In this case, moving master nodes would change the position of the master segments, which would in turn affect the slave node projections. Does this result in the situation that the final configuration of the projected nodes depends on the order in which they are processed by dyna in the initialization phase?


  • Suri Bala says:

    Yes, both slave and master nodes are repositioned. In the first pass, only the slave side nodes are repositioned and in the second pass, only the master side nodes are repositioned. When using a two-way tied contact, either side cannot be used as SLAVE in any other TIED contact which would violate the constraints. However, they can be defined as master in a ONE-WAY definitions which then will only cause the slave side to be repositied which then would result in one unique reposition irrespective of the order of the contact defintions. Not sure if this addresses your question.

  • David says:

    Hi Suri,

    I think question of Sven is asking how to tie nodes from one surface to segment of another surface in 2-ways TIED contact if there’s no _OFFSET option defined. In such case, a projection of node to (closest) segment will take place.

    Let discuss two case scenarios that the order of projecting node on segment from both slave and master surface are interchanged.

    501 (500) 502
    master: X————–X
    slave: X————–X
    102 (100) 101

    X: Node
    —: Segment

    Case 1:
    If a slave node (say node 101) is first projected on a master segment (say segment 500), its position is thus changed and will in turn change the slave segment (say segment 100) position. (Let has slave node 101 connected to segment 100, see figure drawn above). On the other hand, a master node (say node 501) connected to master segment 500 has the slave segment 100 as its closest segment. So master node 500 is then projected on the slave segment 100’s new position.

    Case 2:
    In another scenario, we first project the master node 501 on slave segment 100, followed by projecting slave node 101 on master segment 500.

    The diff. order employed in afore-cases will definitely give the different projection configuration. So, how does LS-DYNA handle such kinda situation?


  • Suri Bala says:


    For a given slave and master definition, LS-DYNA first projects all the slave nodes to their respective closest master segment. It then does the same thing to the master nodes by finding the respective closest segments on the slave side. After this, each of the nodes from the slave and master segments will have a unique closest segment on which it will be constrained to remain for the duration of the simulation.

    Based on this, it is clear that the nodal coordinates will certainly be different when you switch the slave and the master if the offset distances are non-negligible. If its non-negligible, then it is best modeled using _OFFSET method which will yield identical nodal coordinates in both cases.

Leave a Reply to Suri Bala Cancel reply

Your email address will not be published. Required fields are marked *