From Numbers to Meaning
What Every Engineer Should Know About FEA
A Guide for Engineers and Designers Who Commission Simulation Work
PART 3
Drop Simulation
Reading the Results
Joseph P. McFadden, Sr.
The Holistic Analyst
2026
From Numbers to Meaning
The simulation has run. The event has played out in virtual time. The analyst holds a complete record of what the model says happened — stress, displacement, and acceleration at every element, at every moment of the impact. That record is now in front of you.
Many engineers feel least confident at exactly this moment. Drop simulation results arrive as colorful contour plots, multi-curve graphs, and animation sequences. Without a frame of reference, the output can feel like a foreign language. It is not. Every result type has a physical meaning that connects directly to your experience with real products. Some of the concepts you encountered in school — the ones that felt abstract at the time — are about to come alive in a way that makes them genuinely useful.
This part builds those connections, and introduces a framework of four questions that will change how you engage with stress results from this point forward.
But first — a distinction that shapes everything in this part.
There is a difference between a consumer of simulation results and a partner in the work. A consumer receives the report, looks at the color plots, and waits to be told whether the design passed or failed. A partner looks at the same plots and asks questions: What if we moved that rib? How about that corner — is that stress driven by tension or compression? Why is the concentration there and not somewhere else? What would change if we thickened that wall by half a millimeter?
Those questions cannot come from someone who has never been given the tools to ask them. And they cannot be answered by the analyst alone — because the analyst does not know what you know about the product. The analyst knows the method. You know the system. The partnership between those two bodies of knowledge is where the real engineering work happens. This part gives you the tools to show up for your side of it.
Stress: The Most Common Result and Its Hidden Nuance
Stress is the result most commonly shown in a drop simulation review. You already have physical intuition for it. When a plastic ruler whitens near a tight bend, that whitening signals stress at the material's elastic limit. When a housing cracks at a corner after a drop, the crack initiated where stress exceeded what the material could carry.
But before examining how stress appears in simulation results, it is worth describing what stress actually is — because this description makes everything that follows easier to understand. Stress is the internal redistribution of forces. When a load is applied to a structure — whether from outside the system or from within it, such as the inertial force of a heavy internal component trying to continue moving during impact — the structure responds by redistributing those forces internally, carrying them from atom to atom through the body of the material.
Atoms are simple. They know two things: being pulled apart in tension, and being pushed together in compression. What varies from location to location inside the structure is the intensity of that pulling or pushing — the internal force per unit area at any given point. That intensity is what we call stress. When you look at a red region on a stress contour plot, you are seeing a location where the internal redistribution of forces has concentrated and where the atoms are working hardest. Whether that intensity exceeds what the material can carry is the engineering question.
In the simulation, stress is calculated at every element throughout the event and displayed as a color contour map. Red is high; blue or green is low. The first question when looking at any stress plot: where is the red, and does it make physical sense for the loading condition applied? High stress near a corner-first impact point, at the edge of a cutout in a thin wall, or at the attachment points of a dense internal component — these are predictable locations. The stress plot should tell a story your engineering judgment can follow.
What Von Mises Actually Is — and What It Is Not
When the analyst shows you a stress contour labeled Von Mises — and it very often will be — it is important to understand precisely what that quantity represents. Von Mises is not a stress in the straightforward sense. It is an equivalent stress: a single number derived from a formula that combines all components of the three-dimensional stress state into one value for practical comparison against yield strength.
At any point inside a structure, the material is simultaneously experiencing stress in multiple directions — along three axes, and potentially in shear as well. The full stress state is a three-dimensional quantity. Von Mises distills that complexity into one number, which can be plotted on a color map and compared against the yield strength from a standard tension test. For ductile materials — metals like steel and aluminum that deform significantly before fracturing — this approach works very well. It is the right tool for that material class, and it is why the simulation world defaults to it so heavily.
But Von Mises has a critical limitation that shapes how every result using it should be read: it has no sign. Von Mises is always positive, regardless of whether the material at a given location is in tension or compression. The directional information has been absorbed into the formula and lost.
Cracks initiate and propagate under tension, not compression. Compression may smush a surface or cause buckling. A fracture requires tensile stress to open and drive it. Von Mises cannot tell you which regime you are in.
A note on the word smush. It is not in any mechanics textbook. It is a technical Joe term — and it sounds exactly as undignified as intended. But it is accurate. Smush describes a localized compressive failure that flattens or crushes a surface without fracturing it: a boss that has been permanently pressed in, a rib that has been flattened, a surface that has collapsed under direct bearing load. It looks different from a crack, fails through a different mechanism, and requires a different design response. Silly word. Specific meaning. Worth keeping.
This matters fundamentally for understanding failure mode. A Von Mises value driven primarily by compressive loading suggests a different failure mechanism — and a different design response — than the same value driven by tension. The Von Mises plot alone cannot make that distinction.
Principal Stresses: Where the Sign Lives
The directional information that Von Mises discards is recovered through the principal stresses. At any point in the structure, a mathematical rotation can be found along which shear stresses vanish. The stresses acting along those principal directions carry signs: positive for tension, negative for compression. The maximum principal stress — the largest algebraic value — is the quantity most directly connected to whether the material will crack in tension.
When reviewing Von Mises results, ask to see the maximum principal stress at the critical locations. Ask whether the high Von Mises region is driven by tension or compression. That single question reframes the conversation from a magnitude discussion to a mechanism discussion — and mechanism is what drives the appropriate design response.
In Abacus and some other simulation tools, the analyst can display a signed Von Mises stress: the Von Mises magnitude assigned the sign of the dominant principal stress at each location. This provides the practical efficiency of a single-number plot with enough directional information to identify the loading regime. If the analysis is running in Abacus, requesting signed Von Mises is a small step that meaningfully enriches the interpretation.
Failure Criteria: The Physics Behind the Numbers
Failure criteria are the formal answer to a deceptively simple question: given the full three-dimensional stress state inside a material, under what condition does it fail? Most engineers have encountered these criteria in coursework. Fewer have connected them to the physical mechanisms they represent — and that connection is what makes them useful in a results review rather than just names on a list.
Von Mises: Distortion Energy and Shear-Driven Yielding
The Von Mises criterion — also called the distortion energy criterion — is grounded in a physical picture that is worth understanding correctly, because it is one of the most important ideas in structural mechanics.
When a ductile metal is loaded, the internal stress state has two distinct components. The hydrostatic component is a uniform pressure acting equally in all directions — like being squeezed by the ocean at depth. The deviatoric component is the part that causes the material to distort and change shape, driven by shear. These two components exist simultaneously at every point in the loaded structure.
The critical insight is this: ductile metals do not yield under hydrostatic stress. A piece of steel subjected to enormous uniform pressure — squeezed equally from all sides — will compress slightly but will not yield. It will not flow. Yielding in ductile metals is a shear-driven event. It is the deviatoric component — the part of the stress state involving shearing and shape change — that causes the atoms in the crystal lattice to slip past one another and produce permanent deformation.
Von Mises is a measure of that deviatoric, shear-driven component. It deliberately excludes the hydrostatic part because the hydrostatic part does not cause yielding. When the Von Mises stress reaches the yield strength measured in a standard tension test, the shear-driven distortion has reached the threshold at which the crystal structure begins to slip. The criterion works for ductile metals because the physics behind the formula matches the actual mechanism of their failure.
Tresca: Maximum Shear Stress
The Tresca criterion predicts yielding when the maximum shear stress in the material reaches a critical value. It is slightly more conservative than Von Mises — predicting yielding at a somewhat lower applied load — and for most engineering applications involving ductile metals under general loading conditions, the two criteria give similar results. The difference is most significant in specific stress states such as pure shear or equibiaxial tension. Both Von Mises and Tresca are built for ductile materials and are not appropriate for brittle failure modes.
A Rule Worth Stating Explicitly
Do not use Von Mises stress to assess failure in brittle materials. Not for glass. Not for ceramics. Not for any material whose elongation at break — whose strain at fracture — falls below approximately five percent. That is the practical engineering threshold between ductile and brittle behavior for the purposes of failure criterion selection.
A more conservative threshold of around three percent is appropriate when working with polymers or materials whose failure behavior is less well characterized. The reasoning is direct: below those strain levels, the material lacks the capacity for the shear-driven plastic flow that Von Mises is built to detect. The failure mechanism is different. Applying a ductile failure criterion to a brittle material does not produce a conservative answer — it produces an incorrect one. Precise in its calculation, wrong in its physics.
When a glass component, a ceramic part, or a filled polymer with low elongation at break is being assessed using Von Mises stress in a simulation review, that is the moment to stop and ask. Ask the analyst what the elongation at break is for that material. Ask whether Von Mises is the appropriate criterion. Ask whether the maximum principal stress or strain has been examined. That single question may be the most valuable engineering contribution made in the entire review.
Maximum Principal Stress: Brittle Failure
For brittle materials — glass, ceramics, many polymers below their glass transition — the relevant criterion is the maximum principal stress: the largest tensile component at a given location, compared directly to the tensile strength of the material. There is no equivalent stress formulation and no energy argument. The material fractures when the most severe tensile component exceeds what it can carry. The simplicity of this criterion reflects the mechanism: brittle fracture propagates from a pre-existing flaw when the tensile stress at the flaw tip reaches a critical intensity.
Glass and Brittle Materials: Strain, Probability, and the Limits of a Pass/Fail Answer
When a drop simulation involves a glass component — a display cover, a lens, a glass panel — the failure assessment is fundamentally different from the ductile metal case. Glass does not yield. It carries the load elastically until the moment of fracture, then fails suddenly and completely with no warning and no post-yield reserve.
Why Strain Rather Than Stress
In the simulation, what the model actually computes first is strain — a direct measure of how much the material has deformed at each location. Stress is derived from strain using the constitutive equations: the mathematical relationship between deformation and the resulting internal forces. Strain is the fundamental quantity the simulation measures. Stress is what the simulation calculates from it, using the material model.
For glass and other brittle materials, examining the maximum principal strain directly is often more informative than relying on the derived stress. The strain is closer to the raw physical state of the material. The stress carries any uncertainty in the material model along with it, adding one layer of derivation between the result and the physics. When reviewing glass component results, ask to see the maximum principal strain. Ask where the largest tensile strains are occurring, and whether those locations correspond to where the glass is physically most vulnerable in the assembled product.
The Probabilistic Nature of Glass Fracture
Glass does not have a single deterministic failure stress. It has a distribution of failure stresses — and this fundamental characteristic changes the entire nature of the simulation assessment for glass components.
The strength of any piece of glass depends on the microscopic flaws present in that specific piece: surface cracks, scratches, and inclusions introduced during manufacturing and handling. Applied tensile stress opens and propagates those flaws, and larger flaws initiate fracture at lower stress levels. Because the flaw population varies from piece to piece, strength varies from piece to piece. A given piece of glass under a given load has a probability of surviving — not a deterministic pass or fail.
This probabilistic behavior is characterized by Weibull statistics. The Weibull modulus describes the tightness of the strength distribution: a high modulus indicates consistent glass whose pieces fail within a narrow range; a low modulus indicates wide scatter with high variability in failure stress. When Weibull parameters for the specific glass product are available, the simulation can estimate the probability of fracture under the predicted loading rather than rendering a binary verdict.
Questions to ask for any glass component in a simulation review:
What is the maximum principal strain at the critical location, and does it correspond to a region of physical vulnerability in the assembly? Has the glass supplier provided Weibull characterization data for this specific product? What failure probability does the simulation result correspond to under those parameters? Does the surface condition modeled — polished, coated, as-cut — match the actual production surface, and how does surface finish affect the flaw population and failure probability?
Yielded Elements and Plastic Strain: A Map of Permanent Damage
When a ductile material is stressed beyond its yield point, the elastic relationship between stress and strain breaks down and the material deforms permanently. In the simulation, elements that have yielded carry stress beyond the yield threshold, and the accumulated deformation is no longer recoverable. This permanent component of deformation is plastic strain.
Why Plastic Strain Tells a Richer Story
Stress results near the yield point have a ceiling effect. Once yielding begins, the stress in affected elements plateaus and redistributes to neighboring material rather than continuing to rise proportionally. Examining peak stress alone can therefore understate how severely a region has been overloaded — the stress plot does not reveal how far past yield the material went, or how large a volume of material crossed that threshold.
Plastic strain tells that story directly. A small, confined patch at a geometric stress concentration may represent cosmetically undesirable but structurally benign local damage. A large connected region spreading across a load-bearing section of the housing wall signals that a substantial volume of material has been permanently altered — and that the structural integrity of that region deserves careful attention. Very high plastic strain values in a ductile metal may indicate proximity to the material's fracture strain even if the simulation has not explicitly predicted a crack.
Ask to see the plastic strain distribution alongside the stress results. Ask how large the yielded region is and whether it is confined or spreading. Ask what the peak plastic strain value is relative to the material's fracture strain. The plastic strain map is one of the most honest indicators the simulation can provide of what actually happened to your design.
The Four-Question Framework for Every Stress Review
With all of the above in hand, the stress results section of any drop simulation review can now be approached with a consistent four-question framework.
First: what failure criterion is being used, and is it appropriate for the material at the critical location? Von Mises for ductile metals and tough plastics; maximum principal stress or strain for brittle materials and glass. The right criterion for the right material is the foundation.
Second: for Von Mises results, ask to see the maximum principal stress. Determine whether the high Von Mises region is driven by tension or compression. Cracks require tension. These are different failure mechanisms with different design implications.
Third: for any region that has yielded, ask to see the plastic strain distribution. Understand not just that the yield threshold was crossed, but how far beyond it the material went and how much structural volume has been permanently changed.
Fourth: for glass and brittle components, ask about failure probability. What do the Weibull parameters for this specific glass product say about the likelihood of fracture under the predicted loading, and what uncertainties affect that estimate?
You do not need expertise in continuum mechanics to ask these four questions. You need to understand enough of the physical story behind each result type to know which question belongs in the conversation. That is what genuine partnership looks like.
What Partnership Sounds Like in the Room
The four questions above are the framework. Here is what using them actually sounds like in a results review — and why that conversation produces better outcomes than a passive review ever can.
A partner looks at a Von Mises plot showing high stress at a corner and asks: is that tension or compression? If it is tension, we may have a cracking risk. If it is compression, the failure mode is different and so is the fix. Let us see the principal stresses before we decide what to do about it.
A partner looks at a region of plastic strain spreading across a wall section and asks: what if we added a rib along that span? Would it keep the wall from flexing enough to yield? Or is the root cause somewhere else — is that wall deflecting because the support at its boundary is too compliant?
A partner looks at a glass result and asks: how about that edge — is that where we expect the highest flaw density from the cutting process? Because if the Weibull modulus is low for this glass, the scatter in failure stress is wide, and a result that looks marginal on paper may represent a meaningful fraction of production parts at risk.
A partner looks at an acceleration result on the circuit board and asks: do we have qualification data for these components at that G level? And if we do not — what if we changed the mounting compliance to spread the pulse over a longer duration? Would that reduce the peak acceleration enough to matter?
None of those questions require a degree in finite element analysis. They require an understanding of the physical story behind the result — enough to be curious in the right direction. The analyst has built the model and run the analysis. What they need from you is the product knowledge and design context that only you can bring. The question you ask in that room is your contribution to the answer.
Displacement, Acceleration, and the Time Dimension
Displacement: Two Physically Distinct Forms
Displacement tells you how far any given point moved from its original position. Elastic displacement is temporary — the housing flexes and recovers. The key question is whether temporary deformation during the event causes interference with internal components or exceeds the available travel of any mechanism. Permanent displacement is the deformation that remains after the event, indicating the material was loaded past yield. Whether it matters depends entirely on the tolerances and functional requirements at the affected location — information only the engineer who knows the product can provide.
Acceleration: The Internal Load Story
When the housing decelerates at impact, internal components try to continue moving. Their attachments absorb the resulting forces, expressed in the simulation as multiples of gravitational acceleration. A housing that shows no visible post-drop damage can simultaneously have subjected its circuit board to acceleration levels that damaged solder joints or dislodged connectors. The simulation provides those internal loads at every location, in all three directions, throughout the entire event — detail that physical testing alone cannot economically replicate.
Time History: The Full Biography of the Event
All results exist as continuous functions of time. The timing of the peak, the shape of the loading curve, and the presence of structural oscillation after initial contact all carry physical information the peak contour plot alone does not convey. Ask to see time histories at the critical locations. Watch the animation — the physical plausibility of the structure's motion is a valid engineering observation regardless of background in finite element methods.
The Results Do Not Make Decisions — You Do
The simulation shows you where stress is high, which failure criterion applies, where material has yielded, where glass is at probabilistic risk, where internal components have been subjected to high acceleration. It does not determine whether any of that is acceptable for your application, your customer's expectations, your use environment, or your organization's risk tolerance.
The analyst tells you what the model predicts. You determine what those predictions mean for your design. The analyst's technical depth without your contextual knowledge produces an accurate answer to an uncertain question. Your contextual knowledge without the analyst's technical depth produces an informed question without a rigorous answer. The partnership — both contributions present, the four-question framework guiding the conversation — produces what neither can reach alone.
Not a consumer waiting for the verdict. A partner who understands the evidence and asks the right next question. The analyst has the method. You have the system. Together, you have the answer.
This Is a Journey — And Your Brain Has Been on It Your Whole Life
Before closing this part, something worth considering — because it reframes this entire series, and this entire discipline, in a way that may surprise you.
Think about what your brain does every moment of every day. It is not receiving information passively. It is predicting. Before your eyes land on an object, your brain has already made a prediction about what it will be. Before your foot touches the ground, your brain has already predicted how the ground will feel. Before you reach for a glass, your brain has already modeled the weight, the texture, the resistance. Your predictions come before your actions. Your actions are the test of those predictions.
Your brain is the ultimate prediction machine. And the predictions it makes are not random — they are generated by internal models built from a lifetime of experience. Every time a prediction is confirmed, the model is reinforced. Every time reality does not match what was expected — every time the world surprises you — the model is updated. The brain revises, learns, and builds a more accurate representation of the world it operates in. This is not a metaphor. This is what the neuroscience tells us the brain is actually doing, continuously, at every level of perception and action.
CAE Follows the Same Path
Now notice something. That is exactly what simulation-based engineering does.
The finite element model is a formalized representation of a physical system, built from the best available knowledge, used to generate predictions about how that system will behave. Those predictions are tested against reality — in physical testing, in field data, in observation. Where predictions match, confidence in the model grows. Where they do not match, the model is revised. Updated. Made more accurate. The gap between the model and reality is not a failure. It is the signal that drives learning.
This is why the work is best understood as a journey rather than a destination. Your understanding of your system deepens with every simulation, every test, every revision, every conversation between analyst and engineer. The model you carry in your mind of how your product behaves — that model started somewhere, and it is being updated right now, by every part of this series.
The more deeply you engage with simulation results — asking the what ifs, questioning the assumptions, bringing your product knowledge into the conversation — the more your internal model grows. And the better your internal model, the better the questions you ask. The better the questions, the better the simulation. The better the simulation, the better the design. The brain updating its models through experience. The simulation updating its models through testing. The engineer growing through engagement. It is the same process, and it never really ends.
We are not observers of the world. We are modelers of it. Constantly building, constantly testing, constantly revising. CAE follows exactly the same path.
Before closing this part, one idea deserves to be stated plainly — because it runs beneath everything in this series and is the most honest thing that can be said about simulation work.
These are models. Every one of them. No matter how carefully constructed, how experienced the analyst, or how well the material properties are characterized — the simulation is a mathematical representation of physical reality. And reality, in its full complexity, can never be completely captured in a model. There will always be a gap. There will always be something the model did not know, could not include, or could not represent exactly. That is not a weakness of the method. It is the nature of modeling itself.
The engineer who understands this is more valuable than one who does not — because they know what questions to ask, and they know when the model's answer needs to be tested against the physical world before a design is committed.
What Simulation Actually Gives You
What simulation provides is something genuinely extraordinary: the ability to virtually prototype. To build the design, load it, examine it, change it, and rebuild it — before a single physical part is manufactured. To explore ten design alternatives in the time it would take to fabricate one prototype. To see inside the structure at every location and every moment of the event, with a spatial and temporal completeness that no instrument array on a physical test article can fully replicate.
That capability compresses development time, focuses physical testing on the questions that matter most, and surfaces vulnerabilities early — when they are inexpensive to address rather than late when they are not. It is a powerful tool with a specific and irreplaceable role in the development process.
What Simulation Does Not Replace
But it does not replace physical testing. It never replaces physical testing. The physical test is where the model meets reality — where the gap between them is revealed, measured, and understood. The best use of simulation is not to avoid testing. It is to arrive at testing better prepared: knowing where to look, knowing which failure mode the model predicted, knowing whether the physical evidence agrees and what to learn when it does not.
Simulation and testing are partners. Each brings what the other cannot provide. The product that emerges from that full partnership is better than any single contributor could produce alone.
The glass discussion earlier in this part is the clearest illustration of that principle. The simulation provides a probabilistic prediction. The physical test tells you where in that distribution the real parts actually sit. Neither answer is complete without the other. And neither conversation — between model and test, between analyst and engineer — produces its full value without both sides showing up prepared.
Coming Up in Part 4
Part Four addresses the gap between what the simulation predicts and what the physical product actually does — where it is largest, where it is smallest, and what an informed partner does with that knowledge to calibrate confidence in the results.
— End of Part 3 —
© 2026 Joseph P. McFadden, Sr. | The Holistic Analyst | McFaddenCAE.com
Freely shared for the engineering community. Not for resale.
