김경종
230 lines
20 KiB
Markdown
230 lines
20 KiB
Markdown
<!-- source-page: 841 -->
|
||
|
||
|
||
|
||
<details>
|
||
<summary>text_image</summary>
|
||
|
||
Overhanging nodes
|
||
Slave surface
|
||
Master surface
|
||
Problem node
|
||
</details>
|
||
|
||
Figure 39.1.1–7 Two surfaces in a region of nonconverging force equations.
|
||
|
||
<!-- source-page: 842 -->
|
||
|
||
<!-- source-page: 843 -->
|
||
|
||
# 39.1.2 COMMON DIFFICULTIES ASSOCIATED WITH CONTACT MODELING IN Abaqus/Standard
|
||
|
||
Products: Abaqus/Standard Abaqus/CAE
|
||
|
||
# References
|
||
|
||
• “Defining general contact interactions in Abaqus/Standard,” Section 36.2.1
|
||
• “Defining contact pairs in Abaqus/Standard,” Section 36.3.1
|
||
• \*CONTACT
|
||
• \*CONTACT PAIR
|
||
• \*CONTACT INITIALIZATION DATA
|
||
• “Defining general contact,” Section 15.13.1 of the Abaqus/CAE User’s Guide, in the HTML version of this guide
|
||
• “Defining surface-to-surface contact,” Section 15.13.7 of the Abaqus/CAE User’s Guide, in the HTML version of this guide
|
||
• “Using contact and constraint detection,” Section 15.16 of the Abaqus/CAE User’s Guide, in the HTML version of this guide
|
||
|
||
# Overview
|
||
|
||
This section highlights the difficulties that are most commonly encountered when modeling contact interactions with Abaqus/Standard. Recommendations on how to circumvent these problems are presented.
|
||
|
||
# Difficulties resolving initial contact conditions
|
||
|
||
It is important to understand how Abaqus/Standard interprets and resolves contact conditions at the start of a step or analysis. If necessary, you can check initial contact conditions in the message file (see “The Abaqus/Standard message file” in “Output,” Section 4.1.1). Unintentional contact openings or overclosures can lead to poor interpretations of surface geometry, unintentional motion in a model, and failure of an analysis to converge.
|
||
|
||
# Removing initial contact openings and overclosures
|
||
|
||
When modeling the contact between two faceted surfaces, it is often possible for small gaps or penetrations to occur at individual nodes. This problem is particularly common when the two surfaces have dissimilar meshes. Abaqus/Standard uses two default methods for dealing with initial penetrations:
|
||
|
||
• In general contact small initial overclosures are automatically adjusted to remove the penetrations.
|
||
• In contact pairs initial overclosures are interpreted as interference fits and resolved accordingly (see “Resolving large interference fits” below).
|
||
|
||
<!-- source-page: 844 -->
|
||
|
||
You can improve the accuracy of a contact simulation by having Abaqus/Standard adjust the position of the slave surface to ensure that all slave nodes that should initially be in contact with the master surface start out in contact without any penetration (see “Controlling initial contact status in Abaqus/Standard,” Section 36.2.4, and “Adjusting initial surface positions and specifying initial clearances in Abaqus/Standard contact pairs,” Section 36.3.5). When an intended initial clearance or overclosure is small compared to typical dimensions of the bodies in contact and a small-sliding contact pair is used, you can specify the clearance or overclosure precisely (see “Defining a precise initial clearance or overclosure for small-sliding contact” in “Adjusting initial surface positions and specifying initial clearances in Abaqus/Standard contact pairs,” Section 36.3.5).
|
||
|
||
The small-sliding contact tracking approach is more sensitive than the finite-sliding tracking approach to initial local gaps at the contact interface. In small-sliding contact each slave node interacts with a contact plane defined from the finite element approximation of the master surface, as discussed in “Contact formulations in Abaqus/Standard,” Section 38.1.1. Abaqus/Standard can define these planes only when each slave node can be projected onto the master surface. Having these slave nodes start the simulation contacting the master surface allows Abaqus/Standard to form the most accurate contact planes for the slave nodes.
|
||
|
||
# Large unintended initial overclosures
|
||
|
||
The contact initialization algorithm may occasionally infer large initial overclosures where you do not intend initial overclosures to exist. For example, specifying incorrect surface normals can cause the contact initialization algorithm to interpret a physical gap as a penetration, as discussed in “Orientation considerations for shell-like surfaces” in “Defining contact pairs in Abaqus/Standard,” Section 36.3.1. Minor changes to the surface or contact definition will typically avoid undesired overclosures, but these situations typically call for some diagnosis to determine how to avoid the problem.
|
||
|
||
# Identifying the location of unintended overclosures
|
||
|
||
The first step in resolving a large initial overclosure is to identify the location of the problem:
|
||
|
||
• If initial overclosures are treated as interference fits to be resolved in the first increment (which is the default behavior for contact pairs; see “Modeling contact interference fits in Abaqus/Standard,” Section 36.3.4), a contour plot of the contact opening distance output variable (COPEN) for the initial output frame will show which regions have initial overclosures (penetrations correspond to negative values of COPEN).
|
||
• If initial overclosures are resolved with strain-free adjustments, a contour plot of the output variable STRAINFREE for the initial output frame will show where adjustments occurred (see “Contact diagnostics in an Abaqus/Standard analysis,” Section 39.1.1, for further discussion of this output variable). However, large strain-free adjustments may cause the mesh to become highly distorted, making it difficult to fully diagnose the problem; in such cases, perform a datacheck analysis (see “Abaqus/Standard, Abaqus/Explicit, and Abaqus/CFD execution,” Section 3.2.2) with initial overclosures instead treated as interference fits to be resolved in the first increment to facilitate diagnosis (as discussed above).
|
||
|
||
<!-- source-page: 845 -->
|
||
|
||
Once you identify the location of an unintended initial overclosure, limiting the display in the Visualization module of Abaqus/CAE to the master and slave surfaces of the interaction involved in the initial overclosure is helpful for identifying the cause of an unintended initial overclosure (see “Managing display groups,” Section 78.2 of the Abaqus/CAE User’s Guide, for a discussion of the display group options). Viewing the surface normals (see “Displaying element and surface normals,” Section 55.7 of the Abaqus/CAE User’s Guide) may help determine whether unintended overclosures are due to incorrect surface normals.
|
||
|
||
# Overclosures on discontinuous surfaces
|
||
|
||
Figure 39.1.2–1 shows an example with a large, unintended initial overclosure. In this case a single contact pair with discontinuous surfaces is meant to enforce contact in two distinct regions (Table 36.3.1–1 “Orientation considerations for shell-like surfaces” in “Defining contact pairs in Abaqus/Standard,” Section 36.3.1, shows which contact formulations allow discontinuous surfaces). The arrows in Figure 39.1.2–1 show the positive normal direction for each surface region. The surface-to-surface contact formulation searches along the slave-surface normal direction (in the positive and negative directions) for potential interaction points on the master surface. The search emanating from point A identifies point B as the only potential interaction point for point A in this example. The contact pair interprets this as a valid penetration because no better candidate interaction location is found and surface normals are opposed at points A and B. Methods to avoid this unintended overclosure include:
|
||
|
||
• defining separate contact pairs with continuous surfaces for each of the two distinct contact regions; and
|
||
• specifying general contact, which filters out nearly all unintended initial overclosures.
|
||
|
||
|
||
|
||
<details>
|
||
<summary>text_image</summary>
|
||
|
||
Interpreted as a penetration for a single
|
||
contact pair with discontinuous surfaces
|
||
Slave
|
||
Master
|
||
B
|
||
A
|
||
Slave
|
||
Master
|
||
</details>
|
||
|
||
Figure 39.1.2–1 Example of an unintended initial overclosure due to a modeling error involving discontinuous surfaces.
|
||
|
||
<!-- source-page: 846 -->
|
||
|
||
# Overclosures on three-dimensional surfaces
|
||
|
||
The cause of unintended initial overclosures may be less obvious for three-dimensional models with complex surfaces. The most important step in overcoming this problem is identifying which regions of respective surfaces are involved in an unintended initial overclosure. For a surface-to-surface contact pair without strain-free adjustments, a portion of the master surface should be apparent behind the slave surface (opposite the slave surface normal direction) at a distance consistent with the reported (negative) COPEN value. For a node-to-surface contact pair, the direction to the interaction point on the master surface typically corresponds to a local minimum distance between the slave and master surfaces.
|
||
|
||
# Resolving large interference fits
|
||
|
||
As previously discussed, Abaqus/Standard optionally interprets initial overclosures as interference fits. You should use one of the methods discussed above to remove any initial overclosures that are an unintended result of mesh discretization or errors in defining contact surfaces. In some cases the interference fit may be intended but may be too large to be resolved robustly with the method that is used by default for contact pairs in Abaqus/Standard (which is to resolve overclosures in a single increment). In this situation you should modify the contact model to allow resolution of overclosures over multiple increments (see “Modeling contact interference fits in Abaqus/Standard,” Section 36.3.4, for more information). If you choose to have initial overclosures treated as interference fits for general contact, they are automatically resolved over multiple increments (see “Controlling initial contact status in Abaqus/Standard,” Section 36.2.4).
|
||
|
||
# Preventing rigid body motion in contact simulations
|
||
|
||
Rigid body motion is generally not a problem in dynamic analysis. In static problems rigid body motion occurs when a body is not sufficiently restrained. “Numerical singularity” warning messages and very large displacements indicate unconstrained motion in a static analysis. Therefore, if contact is used to constrain rigid body motion in static problems, ensure that the appropriate surface pairs are initially in contact (see “Controlling initial contact status in Abaqus/Standard,” Section 36.2.4, and “Adjusting initial surface positions and specifying initial clearances in Abaqus/Standard contact pairs,” Section 36.3.5). If necessary, define the model geometry to give a small initial overclosure to the contact pair, or use boundary conditions to move the structures into contact in the first step. The boundary conditions, which are unnecessary in subsequent steps, can be removed after the body is adequately constrained through contact with other components. Similarly, if a rigid body is meant to translate only, constrain its rotational degrees of freedom.
|
||
|
||
Frictional sticking can constrain rigid body motion. However, contact pressure must develop before friction can be generated. Therefore, friction is not effective in constraining rigid body motion when surfaces first come into contact. You must temporarily eliminate rigid body motion by defining a boundary condition or by grounding the body with soft springs or dashpots.
|
||
|
||
If you are unable to prevent rigid body motion through modeling techniques, Abaqus/Standard offers some tools to automatically stabilize rigid bodies in contact simulations. These tools are discussed in
|
||
|
||
<!-- source-page: 847 -->
|
||
|
||
“Automatic stabilization of rigid body motions in contact problems” in “Adjusting contact controls in Abaqus/Standard,” Section 36.3.6.
|
||
|
||
# Poorly defined surfaces
|
||
|
||
Over the course of an analysis, you may notice undesirable behavior between contact surfaces (excessive penetration, unexpected openings, inaccurate application of forces, etc.). This behavior often results in nonconvergence and termination of an analysis. These problems can arise from a number of causes related to mesh, element selection, and surface geometry.
|
||
|
||
# Defining duplicate nodes on the master surface
|
||
|
||
When defining three-dimensional surfaces for use in finite-sliding applications, avoid defining two surface nodes with the same coordinates. Such a definition can give rise to a seam, or crack, in the surface as shown in Figure 39.1.2–2.
|
||
|
||
|
||
|
||
<details>
|
||
<summary>natural_image</summary>
|
||
|
||
Abstract geometric pattern with interconnected nodes and curved edges, no text or symbols present
|
||
</details>
|
||
|
||
Both vertices have the same coordinates.
|
||
They are separated to show the crack in the surface.
|
||
|
||
Figure 39.1.2–2 Example of doubly defined surface node.
|
||
|
||
If viewed with the default plotting options in Abaqus/CAE, this surface will appear to be a valid, continuous surface; however, if this surface is used as the master surface for finite-sliding, node-to-surface contact, a slave node sliding along the surface may fall through this crack and get “stuck” behind the master surface. Similar problems can occur for finite-sliding, surface-to-surface contact. Typically, convergence problems will result that may cause Abaqus/Standard to terminate the analysis.
|
||
|
||
Use the edge display options in the Visualization module of Abaqus/CAE to identify any unwanted cracks in the surfaces used in the model. The cracks will appear as extra perimeter lines in the interior of the surface. Duplicate nodes can be avoided easily by equivalencing nodes when creating the model in a preprocessor.
|
||
|
||
<!-- source-page: 848 -->
|
||
|
||
# Avoiding problems with contact along the perimeters of surfaces
|
||
|
||
When modeling finite-sliding contact, ensure that the master surface definition extends far enough to account for all expected motions of the contacting parts. Contact along the perimeter of master surfaces should be avoided with the node-to-surface contact formulation.. Abaqus/Standard assumes that the mating slave surface nodes can fall off the free edge of the master surface, which can cause problems if a slave node wraps around and approaches its mating master surface from behind. Figure 39.1.2–3 illustrates appropriate and inappropriate master surface definitions.
|
||
|
||
|
||
|
||
<details>
|
||
<summary>text_image</summary>
|
||
|
||
trimmed master surface
|
||
slave surface
|
||
Inappropriate master surface definition
|
||
untrimmed master surface
|
||
Appropriate master surface definition
|
||
</details>
|
||
|
||
Figure 39.1.2–3 Example of master surface extension.
|
||
|
||
A slave node that falls off a master surface in one iteration may find itself contacting the surface in the very next iteration; this phenomenon is known as chattering. If chattering continues, Abaqus/Standard may not be able to find a solution. This problem is less likely with the surface-to-surface formulation approach, because each contact constraint is based on a region of the slave surface rather than individual slave nodes. Request detailed contact printout to the message (.msg) file to monitor the history of a slave node that might slide off the master surface (see “The Abaqus/Standard message file” in “Output,” Section 4.1.1). The message file output will show the cyclic opening and closing of contact at a slave node, which will indicate where the master surface needs to be modified.
|
||
|
||
For node-to-surface contact you can extend the master surface beyond the perimeter of the physical body that it approximates to avoid chattering problems. Chattering can also occur with some contact elements, such as slide line and rigid surface contact elements. Slide line contact elements can also be extended. See “Extending master surfaces and slide lines,” Section 36.3.8, for details.
|
||
|
||
# Falling off small-sliding master surfaces
|
||
|
||
Falling off the edge of a master surface in small-sliding contact problems is not an issue since slave nodes do not slide on the actual surface of the model. Instead, each slave node interacts with a flat, infinite contact plane. This plane is associated with the set of master surface nodes that are closest to
|
||
|
||
<!-- source-page: 849 -->
|
||
|
||
the slave node in the undeformed configuration. For details about small-sliding contact, see “Contact formulations in Abaqus/Standard,” Section 38.1.1.
|
||
|
||
Falling off surfaces modeled with interface elements
|
||
|
||
Falling off the edge of a surface modeled with interface elements is not an issue since the slave nodes slide on a flat, infinite contact plane.
|
||
|
||
# Using poorly meshed surfaces
|
||
|
||
Several problems are caused by surfaces created on very coarse meshes. Some of these problems depend on your choice of contact discretization, as discussed later in “Discrepancies between contact formulations.”
|
||
|
||
Penetrations with coarsely meshed slave surfaces
|
||
|
||
When a coarsely meshed surface is used as a slave surface for node-to-surface contact, the master surface nodes can grossly penetrate the slave surface without resistance (see Figure 39.1.2–4). This situation is common when nonmatching meshes come into contact. Refining the slave surface tends to alleviate this problem.
|
||
|
||
|
||
|
||
<details>
|
||
<summary>flowchart</summary>
|
||
|
||
```mermaid
|
||
graph TD
|
||
A["master surface (segments)"] --> B["slave nodes cannot penetrate master segments"]
|
||
B --> C["penetration"]
|
||
C --> D["slave surface (nodes)"]
|
||
D --> E["master node can penetrate slave segment"]
|
||
E --> F["gap"]
|
||
F --> A
|
||
```
|
||
</details>
|
||
|
||
Figure 39.1.2–4 Master surface penetrations into the slave surface due to a coarse mesh of the slave surface for node-to-surface contact.
|
||
|
||
Surface-to-surface contact will generally resist penetrations of master nodes into a coarse slave surface; however, this formulation can add significant computational expense if the slave mesh is significantly coarser than the master mesh (see “Contact formulations in Abaqus/Standard,” Section 38.1.1, for further discussion).
|
||
|
||
<!-- source-page: 850 -->
|
||
|
||
# Contact occurring at a single element
|
||
|
||
If the mesh on a surface is too coarse, it is possible for a contact interaction to occur entirely within the bounds of a single element. This typically happens when the two contacting surfaces have dissimilar curvature, as depicted in Figure 39.1.2–5.
|
||
|
||
|
||
|
||
<details>
|
||
<summary>text_image</summary>
|
||
|
||
Master surface
|
||
Slave surface
|
||
</details>
|
||
|
||
Figure 39.1.2–5 The master surface contacts the slave surface at a single element face.
|
||
|
||
The results from such an interaction are unreliable and generally unrealistic. If the model in Figure 39.1.2–5 uses node-to-surface contact, the master surface penetrates the slave surface without resistance until it encounters a slave node, as discussed above. If the master and slave designations are reversed, the contact constraint is applied at a single slave node; this concentration creates inaccurately high calculations of the contact pressure. If the model uses surface-to-surface contact, excessive penetration is not likely to occur. However, with only a small number of constraint points involved in the interaction, the averaging algorithm used to enforce surface-to-surface contact performs poorly. Inaccurate contact stress and pressure calculations result.
|
||
|
||
If contact is occurring at a single element, refine the mesh to spread the interaction across multiple element faces.
|
||
|
||
# Coarsely meshed master surfaces and small-sliding contact
|
||
|
||
Coarsely meshed, curved master surfaces in small-sliding simulations can lead to unacceptable solution accuracy due to the approximate nature of the “master planes.” Using a more refined mesh to define the master surface will improve the overall accuracy of the solution in small-sliding problems. However, unless perfectly matching meshes are used, local oscillations in the contact stress may still be observed, even in refined models.
|