Evaluating Impact Analysis Models
Abaqus INP Comprehensive Analyzer V24.2
Evaluating Impact Analysis Models
Audiobook Companion Reader
A Step-by-Step Walkthrough for Explicit Dynamic Drop Simulations
Joseph P. McFadden Sr. — The Holistic Analyst — McFaddenCAE.com
May 2026 — Developed in collaboration with Claude (Anthropic)
How to Use This Document
This document is the companion reader for the Abaqus INP Analyzer Impact Analysis Walkthrough audiobook. Use it in whichever way is most useful to you. Follow along chapter by chapter while listening to the audio. Read it independently as a standalone reference. Or print the checklist and common mistakes sections and keep them at your workstation.
The physics and engineering principles in this walkthrough apply to explicit dynamic simulation regardless of which solver you use. LS-DYNA, Radioss, Ansys Autodyn, or any other explicit dynamics code — the Courant condition is the same, hourglass modes exist in every code that uses reduced-integration elements, and polymer rate dependency is real in every tool. The Abaqus INP Analyzer gives the walkthrough its structure, but the engineering knowledge stands independently.
Introduction
This walkthrough guides you through the complete evaluation of an impact simulation model. It is written for analysts in consumer electronics, medical devices, automotive, and any field where products must survive being dropped. It is equally useful as a standalone educational resource for anyone learning explicit dynamic simulation.
Explicit dynamic simulations differ from every other analysis type in one critical way: the solver advances unconditionally through time with no built-in physical correctness checks. It will complete a run with inverted velocity, missing contacts, incorrect material data, or excessive mass scaling and produce results that look exactly like valid output. The only defense is a disciplined pre-submission checklist that builds a chain of verified evidence before the solver runs.
Two technical concepts receive extended treatment because they are both important and frequently misunderstood: the Cowper-Symonds rate-dependent plasticity model, and hourglass control in reduced-integration elements. Both are covered with enough depth that you leave this walkthrough understanding not just what the check is, but why it matters.
Chapter 1 — The Physics of a Drop Simulation
Impact Velocity
A product dropped from one meter arrives at the surface at approximately 4,429 mm/s (4.43 m/s) in mm-tonne-s units. At the instant of contact, a compressive stress wave originates at the impact interface and propagates at the material sound speed — approximately 5,000,000 mm/s in steel and 2,000,000 mm/s in a polymer housing. The wave reflects at free surfaces, generating tensile stresses that can cause internal fracture. Contact forces redistribute load through the assembly. Plastic deformation absorbs energy. The product decelerates, rebounds, and comes to rest.
The Explicit Solver and the Courant Condition
The explicit solver advances through time at increments small enough that the stress wave cannot travel across more than one element per step. This is the Courant stability condition. For a typical one-millimetre polymer element, the stable time increment is approximately 50 nanoseconds. A ten-millisecond drop event requires approximately 200,000 time increments. There is no equilibrium iteration and no convergence check. The solver simply advances.
This is why setup errors are invisible. The solver has no mechanism to tell you the model is wrong. Plausible-looking results are not evidence of a correct model.
The Four Error Categories
The first is incorrect initial conditions. Wrong velocity direction, magnitude, or application point. An inverted velocity sends the product away from the surface. Zero contact force throughout. The solver completes normally.
The second is contact definition gaps. Parts pass through each other without generating force. The load path is broken entirely.
The third is mass scaling misuse. Artificial mass changes the inertial response of the model. The details of this are covered in Step 10.
The fourth is missing energy output. Without the energy history variables, the energy balance cannot be verified after the run and the results are unverifiable.
Chapter 2 — Steps 1 Through 3
Step 1 — Load the INP File
Launch the Analyzer. Click Browse and navigate to the impact model INP file. All files referenced by Include statements must be in the same folder. Click Process and wait for the progress dialog. Missing material files break density definitions. Missing density means the center-of-mass calculation is wrong and the energy balance check is invalid before the first solver increment.
Step 2 — Confirm the Model Type Banner
The banner should read Impact, or Mixed with an explicit dynamic step present. In V24.2, open the Recommendations tab and click Simulation Intent Analysis. For an impact model you should see Drop Test or Free-Fall Impact as the primary result at HIGH or MODERATE confidence.
Step 3 — Review the Summary Tab
Confirm part count, element types (predominantly C3D8R or C3D4 for explicit analysis), step count, detected unit system, and contact count. A zero contact count in a multi-part drop assembly is a definitive error.
Chapter 3 — Step 4, Island Parts
Step 4 — Inspect Island Parts
In the Parts tab, check the Islands button. A disconnected part has no mechanical connection to the rest of the assembly. When the assembly decelerates from 4.43 m/s to zero, the island part continues at 4.43 m/s indefinitely. It passes through housing walls and the impact surface. Its kinetic energy is never transferred to structural work or dissipated through contact. The energy balance cannot close.
If you find island parts, visualize each one and determine whether it should be connected via a contact pair or TIE constraint, whether it is intentionally isolated, or whether it is a leftover from a previous model revision. In practice, the third case is the most common.
Chapter 4 — Step 5, The Drop Simulation Analysis Tool
Step 5 — Run the Drop Simulation Analysis
Open Tools, then Impact/Drop Analysis. The dialog reads the velocity keyword block to extract the velocity vector and identifies the gravity direction from the body load definition. The hit surface detection algorithm scores every rigid-element part by how well its outward surface normal opposes the velocity vector. The rigid flat plate should score highest.
An inverted velocity is the single most common setup error in drop models. If a product housing face scores highest instead of the rigid table, the velocity direction is almost certainly inverted. The drop visualization in Step 6 catches this immediately and visually.
Chapter 5 — Step 6, The Drop Visualization
Step 6 — Review the Drop Visualization
Click Launch Visualization. A green arrow shows the initial velocity vector. A yellow arrow shows the gravity vector. The hit surface is highlighted. An orange sphere marks the center of mass of the product assembly.
Confirm that the velocity arrow points from the product toward the hit surface. Confirm that gravity acts in the correct direction. Confirm that the center of mass is physically plausible given your knowledge of the product.
A center of mass outside the product geometry, or dramatically offset to one end, almost always indicates a density unit error. One part with density in kg/m³ in a mm-tonne-s model is off by a factor of ten to the sixth. The Material Consistency Review catches this.
Chapter 6 — Steps 7 and 8, Contact and Bonded Interfaces
Step 7 — Verify Contact Definitions
Open Tools, Contact Analysis. Verify master-slave convention — the stiffer surface should be the master. Verify mesh size ratios and friction coefficients. Open the Interaction Viewer for a complete list of all TIE constraints, contact pairs, general contact, couplings, and multi-point constraints in the model. For assemblies with many interfaces, this is the only practical way to confirm every expected connection is present.
Step 8 — Coincident Surface Detection for Bonded Interfaces
For multi-part assemblies with physically bonded interfaces, run Find Coincident Surfaces from the Parts tab. Focus on the Mean Gap between matched face centroids, which should be below 0.05 model units for a correctly modelled interface. A POOR coverage grade usually reflects mesh density differences between adjacent parts, not a modelling error.
A missing TIE constraint on a bonded interface means the two parts are mechanically decoupled during impact. The part connected to the impact surface decelerates with the assembly. The unconstrained part continues at initial velocity. Its kinetic energy is never transferred to structural work and disappears from the energy balance.
Chapter 7 — Steps 9 Through 11, Boundary Conditions and Materials
Step 9 — Verify Boundary Conditions
Open Tools, BC and Load Viewer. Confirm that fixed boundary conditions are applied to the rigid table or floor surface. If the table is not constrained, the impact force accelerates it away from the product. The product decelerates more slowly than in reality, the forces are lower than they should be, and the simulation produces an unconservative result.
Step 10 — Check Mass Scaling
Open the Recommendations tab. Conservative mass scaling — adding less than one to five percent of the model's total physical mass to the smallest elements — is an accepted technique. The effect on forces, wave propagation, and energy distribution is small enough to be tolerable for most engineering purposes.
Excessive mass scaling is a different matter. The artificial mass increases the total inertial mass of the model. For the same initial velocity, more kinetic energy must be dissipated during the impact event, which means larger inertial forces are required during deceleration. At the same time, the speed of sound is proportional to the square root of the elastic modulus divided by the density. Adding artificial mass increases the effective density of the scaled elements, which slows their local wave propagation speed. The contact force history changes. The deformation sequence changes. The model no longer represents the physical product in a predictable way. If mass scaling exceeds five percent of total physical mass, fix the mesh instead.
Step 11 — Verify Materials for Impact
Rate-Dependent Plasticity for Polymers
Standard polymer tensile tests are conducted at quasi-static strain rates, typically around 0.001 per second. That is the standard ISO cross-head speed. The stress-strain curve from that test is what most supplier datasheets report, and it is the data most analysts enter into the *PLASTIC keyword.
In a one-metre drop test, the strain rates in the critical sections of the housing reach 10 to 1,000 per second. At those rates, a polymer like PC/ABS behaves very differently from its quasi-static behaviour. The material's molecular chains cannot uncoil and reptate on the timescale of the loading. The material becomes stiffer. The yield stress increases significantly. For a typical PC/ABS grade, the yield stress at 1,000 per second can be 60 to 85 percent higher than the quasi-static value.
When you use only the quasi-static *PLASTIC data, the material model yields at a stress level that the real material, at those rates, would still be carrying elastically. Elements enter plastic deformation too early. The hardening curve is too soft for high-rate conditions. The simulation over-predicts plastic deformation and energy absorption. The housing appears more ductile and tougher than it physically is at drop-test rates. Parts that would fracture in the physical test appear to pass in the simulation. This is non-conservative for failure prediction.
The Cowper-Symonds Model
The Cowper-Symonds model is a power law that scales the quasi-static yield stress upward as a function of the current strain rate. The formula is:
σ_dynamic = σ_static × [1 + (ε̇ / C)^(1/P)]
Sigma dynamic is the yield stress at the current strain rate. Sigma static is the quasi-static yield stress from the first data point in your *PLASTIC table. Epsilon-dot is the current equivalent plastic strain rate in the element, in per second units. C is the strain rate coefficient, also in per second units. Think of C as the strain rate at which the material starts to show meaningful rate strengthening. A high C means the material needs very high rates before the multiplier becomes significant. P is a dimensionless exponent controlling the shape of the curve. Lower P means steeper rate sensitivity.
For PC/ABS, typical values are C around 8,750 per second and P around 2.7. At 100 per second, the rate multiplier is approximately 1.40, meaning the yield stress is 40 percent higher than quasi-static. At 1,000 per second, the multiplier is approximately 1.70, meaning 70 percent higher. These are not small corrections. They determine whether a housing cracks or deforms in the simulation.
In Abaqus, the Cowper-Symonds model is defined under the RATE DEPENDENT keyword with TYPE=POWER LAW, as a sub-option beneath PLASTIC. The Analyzer's Cowper-Symonds Estimator in the Materials tab generates preliminary C and P values for common polymer families when high-rate test data is not available. Always validate with a single-element simulation before using estimated values in production. And never define both a RATE DEPENDENT Power Law block and tabular multi-rate PLASTIC blocks on the same material. Abaqus applies both simultaneously, scaling yield stress twice. The Analyzer's Check 12 flags this.
Hourglass Control in Reduced-Integration Elements
Reduced-integration elements — C3D8R for solids, S4R for shells — use a single integration point at the geometric centre of the element. This makes them fast, but it creates a vulnerability. There are deformation modes — twelve for a C3D8R element — where the element can change shape without any strain appearing at that single central integration point. The stress is zero. The stiffness matrix contributes nothing to resisting the deformation. The element distorts freely, storing no strain energy and resisting no force. These are called hourglass modes.
Abaqus suppresses hourglass modes by adding artificial stiffness or damping to each reduced-integration element through hourglass control, specified in the *SECTION CONTROLS keyword. The energy this mechanism consumes is reported as ALLAE — the artificial strain energy variable.
ALLAE should stay below approximately five percent of ALLIE, the real internal strain energy, for the simulation results to be credible. Ideally it should be below two to three percent. If ALLAE climbs to fifteen or twenty percent of ALLIE, the hourglass control is working heavily to hold the mesh together and the stress results in those regions cannot be trusted.
The ENHANCED hourglass control method, specified in *SECTION CONTROLS, is more accurate than the default RELAX STIFFNESS method for large-deformation impact problems with significant plasticity. If ALLAE is excessive, the correct response is to refine the mesh and improve element aspect ratios — not to tune the hourglass control parameters to suppress the symptom.
Chapter 8 — Steps 12 Through 14
Step 12 — Confirm Energy Output Requests
The *OUTPUT block for the explicit dynamic step must include ALLKE for kinetic energy, ALLIE for internal strain energy, ALLAE for artificial strain energy, and ETOTAL for total energy. Without these, the energy balance cannot be verified after the run.
After the simulation runs: the initial kinetic energy should equal the sum of the final internal strain energy, the remaining kinetic energy, friction dissipation at contact surfaces, and the artificial energy, within a few percent. Then check ALLAE against ALLIE specifically. Above five percent indicates a hourglass problem. The Output Request Builder in the Analyzer generates the correct block and stages it in the Edits tab.
Step 13 — Review All Recommendations
Work through every item in the Recommendations tab. Every Error must be resolved before submission. Every Warning should be reviewed and either corrected or documented with a clear engineering rationale.
Step 14 — Use the Learning Center if Needed
V24.2 adds five new Learning Center topics directly relevant to impact simulation. Strain Rate Dependency covers the complete physics of rate effects including the WLF coupling between rate and temperature, which explains why the minus 40 degrees Celsius drop test is so much more aggressive for polymer housings than room temperature. The Holistic Materials Approach covers injection-moulded weld lines (which can reduce local strength by 30 to 70 percent), fibre orientation anisotropy in glass-filled grades, and moisture effects in polyamide materials.
Chapter 9 — Pre-Submission Checklist
Complete this checklist before submitting any impact model. Every item corresponds to a check available in the Analyzer.
Model Setup
• ☐ Model type banner shows Impact or Mixed with explicit dynamic step present
• ☐ Simulation Intent Analysis returns Drop Test or Free-Fall Impact at HIGH or MODERATE confidence
• ☐ Completeness score reviewed; missing items addressed
• ☐ Islands count is zero, or each island is intentional and documented
Initial Conditions and Geometry
• ☐ Initial Conditions Velocity keyword present with correct direction and magnitude
• ☐ Gravity definition is correct
• ☐ Drop visualization confirms velocity arrow points toward the hit surface
• ☐ Hit surface is rigid and correctly positioned
• ☐ Center of mass location is physically plausible
Contact and Constraints
• ☐ At least one contact definition covers the product-to-surface interface
• ☐ Contact Analysis confirms correct master-slave convention, mesh ratios, and friction
• ☐ Interaction Viewer shows all expected constraints and contacts
• ☐ Coincident Surface Detection confirms TIE constraints cover all bonded interfaces
• ☐ No bonded interface pairs are missing TIE constraints
• ☐ BC and Load Viewer confirms fixed boundary conditions on the rigid table
Materials
• ☐ Every structural part has ELASTIC and DENSITY
• ☐ Material Consistency Review run; Checks 1 through 16 reviewed
• ☐ Rate-dependent plasticity defined for all polymer parts in Explicit step (Check 9)
• ☐ No Cowper-Symonds Power Law and tabular multi-rate *PLASTIC on same material (Check 12)
• ☐ Density values in correct range for the detected unit system
Solver Settings and Output
• ☐ Mass scaling adds less than 5 percent of total physical model mass
• ☐ Hourglass control specified for all reduced-integration element sets
• ☐ Energy output variables ALLKE, ALLIE, ALLAE, and ETOTAL are in the output requests
• ☐ All Recommendations errors resolved or documented with engineering rationale
Chapter 10 — Common Mistakes
Every entry below is a real debugging session that could have been avoided. For each: what went wrong, the consequence, and where the Analyzer catches it.
1. Velocity Direction Inverted
The product moves away from the table throughout the entire simulation. Zero contact force. The results look exactly like a valid drop simulation. Caught by the drop visualization: the green velocity arrow points away from the hit surface.
2. Velocity Applied to the Table Instead of the Product
The product is stationary. The table accelerates into it. The force-time history is completely wrong. Also caught by the drop visualization.
3. Missing Contact Pair Between Product and Table
The product passes straight through the impact surface with zero contact force. Caught by the Contact Analysis tool and the Recommendations contact check.
4. Missing TIE Constraint at a Bonded Interface
Bonded parts decouple during impact. The unconstrained part continues at initial velocity while the rest of the assembly decelerates. Its kinetic energy disappears from the energy balance. Caught by Coincident Surface Detection.
5. No Rate Dependency on Polymer Materials
The quasi-static yield stress under-predicts the actual material strength at drop-test strain rates. Elements yield too early, follow a hardening curve that is too soft, and the simulation over-predicts plastic deformation and energy absorption. The housing appears more ductile and tougher than it physically is. Parts that would fracture in the physical test appear to pass in the simulation. Non-conservative for failure prediction. Check 9 flags this. The Cowper-Symonds Estimator provides starting values.
6. Double Rate Dependency
Both a Cowper-Symonds Power Law block and tabular multi-rate *PLASTIC blocks on the same material. Abaqus applies both rate scaling mechanisms simultaneously. The yield stress is scaled twice. The material appears dramatically stronger than any physical data would support. Check 12 catches this.
7. Density in the Wrong Unit System
Mass is off by orders of magnitude. The center of mass is wrong. The impact forces required to decelerate the assembly are wrong. The energy balance is wrong. Caught by the Analyzer's unit detection and material plausibility ranges.
8. Excessive Mass Scaling
The artificial mass increases the total inertial mass of the model. For the same initial velocity, more inertial force is required during deceleration. The added mass also slows wave propagation speed in the scaled elements, distorting the contact force history and deformation sequence in ways that cannot be predicted or corrected. Keep total added mass below five percent of physical model mass and fix the mesh instead.
9. Poor Hourglass Control
ALLAE exceeds five to ten percent of ALLIE after the run. Reduced-integration elements are deforming in non-physical hourglass modes. Stress and deformation results in those regions cannot be trusted. Refine the mesh, check element aspect ratios, and consider ENHANCED hourglass control for large-deformation impact regions.
10. Missing Energy Output Requests
The energy balance cannot be checked. ALLAE cannot be compared to ALLIE. Results are unverifiable. Caught by the output request check in the Recommendations tab.
11. Reversed Master-Slave Contact Convention
The stiffer surface is assigned as the slave. Contact interpolation is less accurate, potentially producing excessive element penetration or contact force oscillation during the event. Caught by the Contact Analysis tool.
Closing Note
Two concepts from this walkthrough — Cowper-Symonds and hourglass control — represent something larger than two solver keywords.
An analyst who uses only quasi-static plasticity data in a drop test model may simply not know that the polymer they are simulating has a yield stress 60 to 70 percent higher at drop-test strain rates than the quasi-static test shows. Once they know this, they never go back to a quasi-static-only material card for a drop test again.
An analyst who ignores ALLAE after an explicit run may simply have never been taught what that variable measures or why the five percent guideline exists. Once they understand it, checking ALLAE against ALLIE becomes automatic.
Knowledge closes these gaps. Not just knowledge of how to operate the software. Knowledge of the physics the software is trying to represent. Knowledge of what the numbers mean. Knowledge of when the simulation is faithfully capturing reality and when it has drifted away from it in ways the solver will never announce.
That is what Combating Engineering Mind Blindness means. Use this walkthrough. Use the Analyzer. Keep asking why, until the answer makes physical sense.
McFaddenCAE.com · McFadden@snet.net · Combating Engineering Mind Blindness
Developed in collaboration with Claude (Anthropic)
