Stabilizing Contact
Abaqus Standard Contact Stabilization
Why Difficult Contact Problems Can and Should Be Solved in Standard
Companion Reader
Joseph P. McFadden Sr.
The Holistic Analyst | Combating Engineering Mind Blindness
McFaddenCAE.com | McFadden@snet.net
April 2026 | Version 20.3
Developed in collaboration with Claude (Anthropic)
This companion reader accompanies the audiobook of the same title. The words are the same — no break tags, no pauses, just the conversation on the page. Sit down with it. Read it at your own pace.
Part One — The Problem With Taking the Easy Way Out
I want to start with a situation I have seen play out dozens of times in simulation groups across the industry.
An analyst has a model with complex contact. Multiple parts. Dissimilar materials. Some initial gaps. Maybe a gasket or a foam interface. They run it in Abaqus Standard. It doesn't converge. They cut the increment size. Still doesn't converge. They add *CONTACT CONTROLS, STABILIZE. Still struggling. Someone says: just run it in Explicit. Problem solved.
Except the problem is not solved. The problem has been hidden.
Explicit analysis with mass scaling can absolutely produce a result. The solver will not stop. It will march forward through every increment and deliver stress contours and force-displacement plots that look exactly like engineering output. And if the added mass fraction is too high, or if the loading rate is too fast, or if the analysis time is orders of magnitude shorter than the physical event, that output can be spectacularly wrong — and it will give no indication that anything is amiss. No convergence warning. No energy balance failure flag. Just confidently incorrect numbers.
A poor simulation is worse than no simulation at all. An unconverged implicit solution at least tells you something is wrong. An explicit solution with excessive mass scaling tells you nothing — it simply lies convincingly.
This guide — and the audiobook it accompanies — is about the alternative. Understanding why Standard contact problems fail to converge, and then working through a structured ladder of approaches that address the root cause rather than circumventing it. Every technique has a physical basis. None of them are magic parameters to scatter through your INP file hoping something sticks. Used correctly, they allow you to solve contact problems in Standard that most analysts assume require Explicit — and the solutions you get are physically meaningful, computationally efficient, and defensible.
The Abaqus INP Comprehensive Analyzer has specific capabilities that support this workflow, and I note them throughout.
Part Two — Why Standard Struggles With Contact
The Newton-Raphson Method and Contact Discontinuities
Abaqus Standard solves nonlinear problems using the Newton-Raphson iterative method. The concept is straightforward: starting from a known equilibrium state, apply a small load increment, then iterate to find the displacements that satisfy both the constitutive law and global equilibrium. Each iteration requires the solution of a linearised system of equations built around the current tangent stiffness matrix.
The method works extraordinarily well for smooth nonlinearities — material plasticity, geometric nonlinearity, large deformations. These produce stiffness matrices that vary smoothly with displacement, and Newton-Raphson converges quadratically in the vicinity of the solution. Each iteration typically cuts the residual error by orders of magnitude. A well-formulated plasticity problem converges in three to five iterations per increment with almost mechanical reliability.
Contact is a different animal entirely. When a node transitions from open to closed — from no contact to contact established — the stiffness contribution at that node changes instantaneously from zero to a finite value. This is not a smooth nonlinearity. It is a discontinuity. The tangent stiffness at iteration k has no information about what the stiffness will be at iteration k+1 if a new set of nodes comes into contact. The Newton-Raphson method can predict a correction that overshoots the contact surface, which changes the contact state for the next iteration, which changes the stiffness again, which produces another overshoot. The solver cycles.
Technical note: Abaqus handles contact discontinuities through separate Severe Discontinuity Iterations (SDIs). Before standard convergence checks are applied, Abaqus iterates to stabilise the contact state. If contact status keeps changing — nodes opening and closing with each iteration — SDIs never resolve, and the increment is abandoned and cut back. Repeated cutbacks with persistent SDI activity are the characteristic signature of contact chattering.
Initial Rigid Body Motion
Before contact is established, there is nothing preventing rigid body motion of an approaching component in the direction of contact. The stiffness matrix is singular in that direction. Newton-Raphson cannot solve a singular system. It will either abort immediately with a zero-pivot warning, or apply a displacement correction of essentially infinite magnitude before being reined in by the line search algorithm.
This is not a failure of Standard. It is a mathematical statement of physical reality: a body with no constraints and no contact has infinite compliance in the unconstrained direction. The solver is correct. The model description is incomplete for the period before contact is established.
The Temptation to Switch to Explicit
Explicit analysis avoids all of this by construction. It uses an explicit time integration scheme that does not require a system solve at each step. Contact is enforced incrementally through a penalty method that adds a force proportional to penetration. There are no iterations, no convergence checks, no Newton-Raphson cycles. The solver always advances.
This is genuinely the right choice for a specific class of problems: true high-speed dynamic events where the inertia of the system is physically important. Drop tests. Impact. Crash. High-rate forming. It is not the right choice for quasi-static loading events where the analyst is simply avoiding a difficult convergence problem.
And the tell — the sign that someone has made this mistake — is mass scaling. When you see *VARIABLE MASS SCALING with a DMASS energy fraction above five percent, you are looking at a static problem dressed in explicit clothing. The Recommendations tab in the INP Analyzer flags this and reminds you to monitor the DMASS fraction. Two output variables belong in every Explicit quasi-static validation: DMASS, which measures the added mass fraction directly, and ALLMW, which measures the artificial kinetic energy introduced by the mass scaling work. Both should be requested as history output and both should be negligible. DMASS > 5% of total model mass is the standard rejection threshold. A ALLMW that is not essentially zero means the artificially added mass is actively doing work on the model — the dynamics are corrupted.
Part Three — Before You Reach for a Parameter, Step Back
Here is something I want to say before we go through any of the technical stabilization techniques.
If you are having serious trouble stabilizing a contact analysis in Standard — if you are cutting increments, adding damping, trying different enforcement methods, and still fighting the solver — that is a signal. Take it seriously.
It may be telling you that something is wrong with the physics. Or that what you are asking the software to do does not actually make physical sense. Or that the design itself is so geometrically or mechanically marginal that the simulation is correctly reflecting a real instability in the system.
Before you add another parameter, stop. Step back. Visualize the situation. Ask yourself: what is actually happening at the contact interfaces in this model? Is there a part that is completely free — no constraints, no contact yet — and being loaded by a force with nowhere to react it? Is there a compliant component being pushed against a rigid surface before any lateral constraint exists? Is the contact sequence physically realistic, or have I set up a mathematical problem with no physical analogue?
Treat convergence difficulty as a prediction error. The solver may not be failing — it may be correctly reflecting that the configuration you have described has no stable equilibrium.
Sometimes the answer is obvious once you look at it. A component is free to translate until it contacts another surface. Before that contact, the stiffness in that direction is zero. The solver is correct to fail. The fix is not stabilization — the fix is to give that component a physical reason not to move freely before contact is established.
And this is where engineering creativity matters more than keyword knowledge. You do not always need a physical constraint. Sometimes you can model a very soft pad pushing the free component toward its contact surface. A foam pad, a rubber interface, an elastomeric gasket — something that is physically present in the assembly but perhaps simplified out of the model. Model it with extremely low stiffness — low enough that it adds essentially nothing to the model's overall stiffness, but high enough to prevent the rigid body mode. This is not artificial damping. It is engineering judgment about what is physically present that you chose to simplify away.
You can do the same thing with soft springs. A grounded SPRING2 element with a stiffness one percent of the expected contact stiffness. It constrains the degree of freedom that is singular before contact. Once contact is established and friction takes over, the spring force becomes negligible. Remove it in subsequent steps if needed, or leave it if its contribution to the result is below measurement noise.
The deeper question is worth asking out loud. If the simulation struggles to find equilibrium for this contact configuration, is the physical system actually stable in this configuration? Contact problems that are mathematically ill-conditioned are often physically ill-conditioned too. A thin-wall component being compressed between two surfaces with no lateral support is genuinely sensitive to small perturbations in the real world. The solver is not wrong. The design may be marginal.
One specific failure mode worth recognising: thin shells and thin-wall structures that buckle or snap through under contact loading. If the convergence difficulty involves negative eigenvalue warnings and repeated cutbacks as the shell approaches a surface, this is a structural instability problem — not a contact problem. The correct solution is not contact stabilization; it is an arc-length method (the Riks procedure in Abaqus, *STATIC, RIKS) which can trace the post-buckling equilibrium path through the unstable regime, or an implicit dynamic step which uses inertia to traverse the snap. Applying contact stabilization to a genuinely buckling structure will produce a result, but it will be the wrong result — you will have stabilized past the instability rather than through it.
Treat convergence difficulty as a prediction error and treat it honestly. Do not paper over it with stabilization until you have looked at the physics and confirmed that the analysis you are trying to run makes physical sense.
With that said — when the physics is sound, the setup is correct, and the difficulty is genuinely numerical, here is the structured approach.
Part Four — The Stabilization Ladder
Work through these steps in order. Stop at the step that solves the problem. Each is progressively more interventional, and each carries progressively more responsibility to validate that the results are physically representative.
Step 1 — Use General Contact and Surface-to-Surface Discretization
Node-to-surface contact is the legacy approach. It has three well-known failure modes: contact snagging at element edges and corners, contact chattering from node oscillation near the surface, and inconsistent contact pressure for non-matching meshes. Surface-to-surface contact discretization enforces constraints based on surface integrals rather than individual node projections. It is significantly more robust for curved surfaces, non-matching meshes, and contact across material interfaces. SIMULIA now recommends general contact for all Standard analyses.
There is an important practical benefit that is easy to overlook: general contact in Abaqus Standard defaults to penalty constraint enforcement rather than the Lagrange multiplier method used by contact pairs. Switching to general contact therefore often brings improved convergence for free — no additional keywords required — because penalty enforcement smooths the stiffness discontinuity at contact onset. If you are using legacy node-to-surface contact pair formulations and experiencing chattering, switching to general contact alone may be sufficient to resolve the problem.
CONTACT, OP=NEW and CONTACT INCLUSIONS, ALL EXTERIOR activate general contact with automatic surface detection, surface-to-surface discretization, and penalty enforcement by default. No manual surface pair enumeration required.
Step 2 — Switch to Penalty or Augmented Lagrange Enforcement
The default constraint enforcement method for hard contact in Standard is the direct method using Lagrange multipliers. This enforces exactly zero penetration. The price is an asymmetric contribution to the stiffness matrix whenever contact status changes, which degrades Newton-Raphson convergence. For models with many nodes cycling in and out of contact, Lagrange multiplier enforcement amplifies the instability.
Penalty enforcement replaces the hard constraint with a contact stiffness. A small, controlled amount of penetration is permitted, which dramatically smooths the stiffness transition at contact onset. The penalty stiffness defaults to a value that keeps penetration below one percent of a characteristic contact face dimension — physically negligible for most engineering purposes.
The augmented Lagrangian method is a hybrid: it uses penalty enforcement within each iteration but augments the contact pressure if penetration exceeds the tolerance, then iterates again. It achieves near-zero penetration with better convergence than direct enforcement. For most difficult contact problems, augmented Lagrangian is the recommended first upgrade from the default.
Caution: Reducing penalty stiffness below 0.1× the default allows physically significant penetration. Always check CPRESS and COPEN output at converged contact nodes.
Step 3 — Use the Friction Onset Control and Slip Tolerance
If friction is defined and the model converges without friction but fails with it, friction is driving the instability. The stick-slip boundary: nodes near the transition between sticking and sliding produce a strongly asymmetric Jacobian. The slip tolerance parameter regularises this transition, replacing the mathematically sharp threshold with a smooth zone of finite width. A value of 0.005 — half a percent of element characteristic length — is a reasonable starting point. The FRICTION ONSET=DELAYED option delays friction to the increment after contact establishes, preventing the combined discontinuity of simultaneous contact establishment and friction activation.
One additional tool specifically for friction-driven convergence problems: the unsymmetric equation solver. Friction inherently produces an asymmetric Jacobian — the force at a slave node depends on the motion of the master surface in a non-symmetric way. Abaqus Standard automatically uses an unsymmetric solver when it detects this condition, but you can force it explicitly. This is particularly useful when friction is combined with large-deformation contact or when the default symmetric solver is struggling with residuals.
*STEP, UNSYMM=YES forces the unsymmetric equation solver for the step. It increases memory and CPU cost but can significantly improve convergence in friction-dominated contact problems. Use it when friction is confirmed as the driver and other friction controls have not resolved the issue.
Step 4 — Use Softened Contact
Hard contact — zero penetration, infinite stiffness — is mathematically clean but numerically demanding. When two surfaces establish contact with even a small initial overlap, or with mismatched mesh densities, the instantaneous enforcement of zero penetration creates a very large contact force in the first iteration.
Softened contact replaces this step function with a smooth pressure-overclosure relationship. In the linear softened model, contact pressure rises linearly with penetration distance. In the exponential model, pressure approaches a target value asymptotically as the clearance approaches zero. Neither produces physically significant penetration in a converged solution. Calibrate the parameters so that the contact pressure at zero clearance is an order of magnitude above the expected contact pressure. Softened contact is particularly effective for initial tight clearances, non-matching meshes, and multi-body assemblies where the load path passes through many sequential contacts.
Step 5 — Apply Contact Controls Stabilization (Targeted)
Contact controls stabilization is the targeted application of viscous damping to a specific contact pair. It is distinct from static stabilization at the step level. The damping is active only when surfaces are within a characteristic distance of each other, it applies only in the contact normal direction by default, and it ramps down to zero over the step so that the converged solution at the end has no artificial damping acting on it.
The correct use case is narrow: a situation in which contact will clearly be established at some point in the step, but rigid body motion of the approaching body before contact makes the initial increments numerically singular. The stabilization provides just enough resistance to suppress the rigid body mode until contact is established.
Critical validation: Compare ALLSD (stabilization energy) to ALLIE (internal energy). ALLSD must stay below 5% of ALLIE throughout the step. If it does not, the damping is too high and is corrupting the results. Never accept results where ALLSD > 5% ALLIE.
Step 6 — Split Into a Contact-Establishment Step
One of the most effective and least used techniques is to split the load application into two sequential steps. The first step applies one to ten percent of the total load with stabilization active. Its only purpose is to get the surfaces into contact. The second step applies the remaining load with stabilization turned off.
This is physically motivated. Contact establishment is typically the most numerically difficult phase. Once contact is firmly established across the interface, the subsequent loading is often much more tractable because the stiffness matrix no longer has singularities. Validate energy separately in each step. In step one, some ALLSD contribution is expected. In step two, ALLSD should be negligible.
Step 7 — Use an Implicit Dynamic Quasi-Static Step
When structural instability — not just contact — is causing convergence failure, the implicit dynamic procedure with a quasi-static application is the most physically honest stabilization approach available in Standard. Instead of artificial viscous damping, it uses the actual mass distribution of the model to provide inertial resistance to sudden configuration changes.
The inertia term in the dynamic equation provides natural resistance to rapid accelerations — which is exactly what prevents convergence in a snapping or chattering contact problem. The time scale must be chosen so that inertia forces are small relative to internal forces — typically by applying the load slowly enough that kinetic energy stays below one percent of internal energy. This is why it is called quasi-static: dynamic in formulation, quasi-static in character.
The validation check: ALLKE / ALLIE < 0.01. If kinetic energy is significant, the loading rate is too fast. The implicit dynamic procedure uses backward Euler integration, which is unconditionally stable — no mass scaling is needed.
Step 8 — Global Static Stabilization (Last Resort)
*STATIC, STABILIZE adds artificial viscous damping to every node in the entire model. It is the nuclear option. The model is effectively immersed in a viscous fluid for the duration of the step. This is appropriate for a very narrow set of problems: global structural instability such as buckling, snap-through, or collapse.
It is not appropriate for contact convergence problems unless you have exhausted every other approach and confirmed that the instability is global in nature. When used for contact problems, it corrupts the deformation of parts that should be moving freely. The energy validation is the same: ALLSD / ALLIE < 5%.
The strongly preferred form is adaptive stabilization, where Abaqus adjusts the damping factor in real time to keep the ALLSD/ALLIE ratio below a user-specified tolerance. This provides automatic protection against over-damping — the solver adds only as much artificial energy as it needs, then backs off. The default dissipated energy fraction of 2 × 10⁻⁴ is conservative and will often need to be increased slightly; start there and monitor the actual ratio. A fixed damping factor specified directly is much harder to calibrate and offers no automatic protection.
*STATIC, STABILIZE=2e-4, ALLSDTOL=0.05 — adaptive form: Abaqus adjusts damping in real time to keep ALLSD/ALLIE below 5%. Preferred over fixed-factor stabilization.
Part Five — How to Know If It Worked
A converged solution with stabilization is not the same as a correct solution. Convergence means the solver found a configuration that satisfies the equations of the stabilized model. Whether that configuration represents the physical behaviour depends on whether the artificial energy is small enough to be considered negligible.
Three output variables are mandatory for any analysis that uses stabilization. Request them in your output definitions. Do not omit them.
ALLSD — Stabilization Dissipation Energy
This is the total energy dissipated by the artificial damping. The acceptance criterion is ALLSD < 5% of ALLIE at all times during the step. In practice, ALLSD is typically highest early in the step when contact is being established and drops as equilibrium is reached. A plot of the ratio over step time should show it starting elevated and dropping toward zero. If ALLSD is persistently elevated above five percent, the stabilization is doing too much work. Either reduce the damping factor, narrow the application region, or accept that the model has a more fundamental problem that stabilization is masking.
ALLIE — Total Internal Energy
ALLIE is the total internal energy of the model — the sum of recoverable elastic strain energy, plastic dissipation, creep dissipation, and all other energy stored or dissipated in the material constitutive response. It is not the same as the elastic strain energy alone (ALLSE). For a well-behaved model ALLIE should rise monotonically with applied load. A drop in ALLIE — or oscillation — is a signal that the solution is not tracking the physical load path correctly, regardless of convergence. Always use ALLIE (not ALLSE) as the denominator in the ALLSD/ALLIE ratio check, because ALLIE correctly captures energy dissipated through plasticity and other inelastic mechanisms that would not appear in ALLSE.
CDPRESS and CDSHEAR — Contact Damping Stresses
When contact controls stabilization is used, the contact damping stresses CDPRESS (normal) and CDSHEAR (tangential) should be compared to the actual contact stresses CPRESS and CSHEAR at the end of the step, after contact is firmly established. If CDPRESS is more than ten percent of CPRESS, the contact result is being driven by the damping rather than by the physics. Reduce the damping factor.
INP Analyzer check: The Recommendations tab flags missing energy output variables. For any model using CONTACT CONTROLS, STABILIZE or STATIC, STABILIZE, the output requests must include ALLSD and ALLIE. If these are absent, there is no way to validate the stabilized result. The tool flags this as a finding requiring resolution before submission.
Part Six — Fixing the Model Before Reaching for Stabilization
Every technique in the stabilization ladder is a way to help the solver handle a difficult contact problem. But many contact convergence failures are caused by modeling errors that would be resolved far more cleanly by correcting the model.
Initial Gap or Overlap
If contact surfaces are defined with a small initial gap, the approaching body is unconstrained until the gap closes. If they have an initial overlap, a large contact force is generated in the first increment that may be many times larger than the applied load. Both conditions drive instability. Use *CONTACT PAIR, ADJUST=value to close initial gaps up to the specified tolerance before the step begins. The BC and Load Viewer in the INP Analyzer shows the assembly position and will reveal obvious overlaps.
Master/Slave Assignment
Using the stiffer material as the slave and the softer material as the master produces asymmetric penalty stiffness distributions that increase the number of SDI iterations required to stabilise contact. The Contact Analysis tool in the INP Analyzer checks and flags master/slave assignment using elastic modulus, material class inference for materials without *ELASTIC, and rigid body detection. Getting the assignment right before running is worth more than any amount of stabilization tuning.
Slave Mesh Coarseness
A slave mesh that is two or more times coarser than the master allows master nodes to penetrate slave elements to a depth of one element before the constraint detects the violation. This requires multiple SDI iterations to resolve. Refining the slave mesh in the contact zone — not necessarily globally — is often the single most effective improvement available. The Contact Analysis tool reports the slave-to-master mesh size ratio and flags it CRITICAL above 2:1 and WARNING above 1.5:1.
Unconstrained Rigid Body Motion Away From Contact
Parts that have no stiffness before contact are common in assembly models: a fastener approaching a hole, a clip snapping onto a rail, a seal about to seat. A SPRING2 element with a stiffness one percent of the expected contact stiffness is often enough to suppress the singular mode without significantly affecting the post-contact result. Or a soft pad representing a component that is physically present in the real assembly but simplified out of the model. Remove the spring or pad in subsequent steps once contact is established, or verify that its contribution is below the accuracy threshold you care about.
Wrong Constraint Enforcement for the Mesh Type
Second-order tetrahedral elements (C3D10M) and surface-to-surface contact with Lagrange multipliers can produce oscillating contact pressure distributions when mesh densities are non-matching. The augmented Lagrangian method with a relaxed penetration tolerance — 0.5% instead of the default 0.1% — resolves these oscillations cleanly without affecting the integral contact force accuracy.
Part Seven — The Decision Framework and Closing Thoughts
The following sequence is the practical framework. Engineering judgment is required at every step, but it represents the order in which interventions should be attempted — from least invasive to most interventional.
1. Verify the model is physically correct before any stabilization. Check master/slave assignment, mesh density ratio, initial conditions, unit system. Use the INP Analyzer Contact Analysis tool.
2. Visualize the problem. Ask whether the contact sequence makes physical sense. Look for free components with no lateral constraint. Consider whether a soft pad or grounded spring can provide physical support that exists in the real assembly but is absent from the model.
3. Switch to general contact with surface-to-surface discretization if using legacy contact pair formulations.
4. Change constraint enforcement from Direct (Lagrange multiplier) to Penalty or Augmented Lagrange.
5. If friction is present, add *FRICTION, SLIP TOLERANCE=0.005. Test without friction to confirm friction is the driver.
6. Replace hard contact with softened (exponential or linear) pressure-overclosure. Calibrate so that contact pressure at zero clearance is 10× the expected contact pressure.
7. Apply *CONTACT CONTROLS, STABILIZE to the specific problematic pairs only. Validate ALLSD / ALLIE < 5%.
8. Split into a contact-establishment step (1–10% of load, stabilization on) followed by the main load step (stabilization off). Validate energy in both steps.
9. Replace STATIC with DYNAMIC, IMPLICIT, APPLICATION=QUASI-STATIC. Validate ALLKE / ALLIE < 1%.
10. Apply global *STATIC, STABILIZE only if all previous steps are exhausted and the instability is confirmed to be global. Use adaptive stabilization.
11. Consider Explicit only if the problem is genuinely dynamic — the loading has inertial significance at the physical time scale. If you reach for Explicit to avoid convergence difficulty on a quasi-static problem, you have not solved the problem.
The pattern I described at the start of this reader — jump to Explicit, crank up mass scaling, deliver results — is genuinely common. It is not a failure of the analyst's intelligence. It is a consequence of deadline pressure and the seductive ease of a solver that always converges.
But the path of least resistance is not the path of least error. An explicit analysis with mass scaling inflated to thirty percent of the model mass is not a simulation. It is an animated guess. The contours will look smooth. The forces will have plausible magnitudes. The report will have a cover page. And the product decision made on the back of that analysis may be wrong in a way that will only become visible when the physical product is tested — or worse, when it is in the field.
The techniques in this guide are not difficult. They require understanding, not heroics. Surface-to-surface contact. Penalty enforcement. Slip tolerance. Contact controls stabilization with energy validation. A quasi-static implicit dynamic step. These are the tools that Dassault Systèmes built for exactly these problems. They work. They produce physically defensible results. And they keep the analysis in Standard — where the physics is static and the results mean what they appear to mean.
The INP Analyzer will not tell you which stabilization technique to use. That is engineering judgment, and it belongs with the analyst. But it will tell you whether your contact setup has the basic modeling errors that make convergence impossible regardless of technique. It will tell you whether your energy output is present so you can validate the stabilization. And it will tell you whether the analysis you are about to run has the setup quality to be worth running at all.
When Explicit Quasi-Static Is the Right Choice
The guidance in this reader is not never use Explicit for quasi-static problems. It is: do not use Explicit to escape a Standard convergence problem that proper modeling would resolve. There is an important and professionally significant distinction between those two situations.
Explicit quasi-static with verified mass scaling is the established industry approach for a specific category of manufacturing simulation problems. What these problems share: deformation is genuinely large, contact geometry changes continuously over large distances during the event, and Standard cannot efficiently trace the load path even with optimal incrementation.
Metal forming and stamping. Deep drawing, stamping, roll forming, progressive die operations, and hydroforming involve 50 to 80 percent nominal strain in key regions and continuously evolving die-sheet contact. Standard cannot handle these. Explicit quasi-static is the industry standard. Examples: automotive door panels, beverage can forming, roll-formed sections, hydroformed tube components.
Rubber and foam large compression. When hyperelastic or foam materials compress beyond 30 to 40 percent nominal strain, Standard often loses equilibrium. Examples: seating foam crush, rubber bump stops at end of travel, extreme gasket compression.
Clinching, riveting, and crimping. Self-piercing rivets, clinched joints, and terminal crimping involve simultaneous large deformation, material fracture, and multiple sequential contact operations. Standard requires prohibitively fine increments for these processes.
The springback workflow. The accepted automotive practice is to run the forming step in Explicit — large deformation, die contact — then import the deformed shape and residual stresses into Standard for elastic springback relaxation. Neither solver does both well. The two-step workflow exploits the strength of each.
The distinguishing criteria: deformation genuinely large (over 20 percent nominal strain); contact geometry changes continuously; Standard requires prohibitive increments even with correct setup; ALLKE/ALLIE below 5 percent throughout (below 1 percent preferred); and DMASS below 5 percent of total model mass. If the last two criteria are violated, you do not have a quasi-static analysis. You have an artificially accelerated dynamic simulation — and the results are unreliable regardless of whether the solver completed.
Validation for Explicit quasi-static: ALLKE/ALLIE < 5% (< 1% preferred) • DMASS < 5% model mass • ALLMW negligible • Force-displacement smooth, no oscillations • Velocity ramped from zero, not stepped • Run at half speed: results must change < 5%
A complete guide to running proper Explicit quasi-static analyses — velocity ramping, mass scaling validation, energy balance, smooth tool motion definition, and output strategy — is in preparation as a separate essay and audiobook in this series.
The Analyst Who Struggles Learns More
There is a reason this guide asks you to stay in Standard when you reasonably can, and it goes beyond the technical argument. It is cognitive.
Neuroscience research on learning — Robert Bjork's work on desirable difficulties in particular — has established clearly that struggle produces more durable and more transferable knowledge than ease. When you have to work hard to construct an answer, the mental model you build is deeper and more connected than one received without effort. The knowledge sticks. It generalises. It becomes part of how you think, not just something you did once.
When a Standard contact analysis fails to converge, you are forced to stop and reason. What is physically happening at these interfaces? Why is the stiffness matrix singular in this direction? What does the model actually look like before contact is established? You visualize. You ask whether the physics makes sense. You build a mental model of the event. That model lives in you — not in the solver, not in the input file. You carry it into every subsequent model for the rest of your career.
When you switch to Explicit and it converges, none of that happens. The solver found a path through the increments. It did not require you to understand the rigid body mode or the chattering mechanism or the friction Jacobian. It just ran. And the question worth asking honestly is this: who solved the problem? You did not. The computer did. And the understanding it used to get through the increments stayed with the computer.
"The easy path through the solver is the hard path through your career. The analyst who has fought Standard contact and won understands something about the physics that no amount of Explicit results can teach."
Every time you step back and ask — is there a soft pad missing? Is this a genuine physical instability? What is the rigid body mode telling me about the design? — you are practising the diagnostic skill that separates engineers who understand their simulations from engineers who run them. One produces defensible engineering. The other produces reports.
That is the deeper case for staying in Standard. Not just that the results are more trustworthy. Not just that the energy balance is cleaner. But that the work of getting Standard to converge makes you a better analyst. The struggle is not the obstacle. The struggle is the education.
Simulation done correctly — Standard, explicit, or otherwise — is how you find failures on a computer screen instead of a test floor. Simulation done incorrectly, however convincingly it converges, is how you find them in the field.
Joseph P. McFadden Sr.
The Holistic Analyst | 45 Years in CAE/FEA Engineering
McFaddenCAE.com | McFadden@snet.net
Abaqus INP Comprehensive Analyzer V20.3 — Free at McFaddenCAE.com
Developed in collaboration with Claude (Anthropic)
