Model accuracy

VirtuCath™ uses two distinct model layers, each characterized separately:

1. The catheter calculator (the calculator): predicts each section's stiffnesses (EA, EI, GJ), their compliance counterparts (L/EA, L/GJ), and derived failure-mode KPIs (burst pressure, external crush pressure, tensile and torque failure loads) from the layer stack, materials and geometry. It keeps a layer-by-layer laminate structure, but its correction factors are learned from 2,228 solutions of a wire-level finite-element model that places every braid and coil wire on its real path through the polymer wall. It gives a result in milliseconds. No constant in either model was fitted to bench data.
2. The simulations built on top of it. Three simulations start from what the calculator resolves:
   • The steering simulator (the simulator) predicts steering behavior under actuation (pull-wire deflection, gravity loading, multi-body equilibrium) with VirtuCath's own quasi-static multi-body solver: a chain of rigid segments joined by bending, twist and axial springs set from the calculator's section stiffnesses, solved to a converged static equilibrium at every frame.
   • The Kink Radius simulation, a 2D finite-element model of one cross-section, finds the bend limits: kink, collapse, and permanent set (its own section below).
   • The Pushability simulation, a 2D finite-element push test of the shaft's free length, finds the buckling behavior (its own section below).

The layers feed each other: the simulations consume the section properties and layup the calculator resolves. Each model is still characterized independently, against benchmarks appropriate to its job. This page documents all of them.

Two words recur below, and they are not interchangeable. Verification asks whether the software solves its equations correctly; it is checked against established analytical and numerical references. Validation asks whether the predictions match physical reality; it is checked against bench measurement. VirtuCath publishes both kinds of evidence, and each claim on this page is labeled as one or the other.

How accurate is VirtuCath? · Watch on YouTube

Evidence at a glance

Prediction	Benchmark	Headline result	Full document
Section stiffness: bending (EI), torsion (GJ), axial (EA) (the calculator)	Published bench data for a 7 Fr braided shaft in three braid patterns (EI, GJ) and a third-party coil-under-braid shaft (EA) (validation); a sealed test of 400 designs across six construction families, scored once after the calculator was frozen (verification)	Bench: median error 6.9%, largest 17.7%, all 7 trusted comparisons inside the plan limits (15% median, 25% largest). New designs: median 7.9% (EI), 8.3% (GJ), 2.6% (EA)	Validation of the VirtuCath Stiffness Model, version 1.0: 📄 PDF
Kink and collapse bend radii (the Kink Radius simulation)	Exact beam theory, the exact thin-shell limit point, a refinement study on the bundled designs, and a collapse estimate from the wire-level stiffness model (verification)	Limit point within 0.014% of the exact thin-shell solution on thin walls; bending stiffness within 0.16% of beam theory; collapse onset converged to a median 0.022% at the app's mesh (sections with pull-wire lumens up to 10% tight, a deliberate mesh choice that keeps the run at a reasonable duration); inside the wire-level estimate on 6 of 7 catheters, at most 1.8% cautious on the other. Radii are screening values: rank designs in the app, confirm the finished design on the bench	Verification of the VirtuCath Kink Radius Simulation, VR-VC-004 version 1.3: 📄 PDF
Buckling under push (the Pushability simulation)	Euler's exact elastica for the four ideal tip restraints, two-step columns, a refinement study, and contact and friction checks (verification)	Push force after buckling within 0.13% of the exact elastica at the default mesh (error falls with the square of the element size); buckling knee within 1.4% of the Euler load at the default starting bow; all 2,682 solved steps converged and balanced	Verification of the VirtuCath Pushability Simulation, VR-VC-003 version 1.1: 📄 PDF
Deflected shape and wire tension under gravity and pull-wire actuation (the steering simulator)	Exact beam theory for a segment under its own weight (500 cases) and the exact constant-curvature arc under a pull wire; a refinement study; a published Cosserat-rod reference (Rao et al., 2021) on 949 one- and three-segment catheters bent up to 180° (verification)	Small deflection: median error 0.18%. Against the reference: median tip-angle error 0.13° (one segment) and 0.06° (three segments), median tension error 1.0% and 0.33%; no catheter off by more than 4.3° or 4.8%	Verification of the VirtuCath Steering Simulation, VR-VC-001 version 5.0: 📄 PDF

Every quantitative claim on this page traces to one of these documents. The sections below give the context each number needs: what was tested, where the model is strong, and where you should reach for bench testing instead.

Quick reference: when to trust the predictions

• ✅ Single-braid shafts and steerable shafts with pull-wire lumens (median error 5–10% on new designs)
• ✅ Coil-only sections in bending and axial stiffness (torsion is weaker: median 11%, one coil in three more than 15% off)
• ✅ Axial stiffness (EA) in every construction family (median under 7.5%)
• ✅ Comparing two similar designs: the ratio between them is predicted 3–4× more closely than either absolute value (see Comparing designs)
• ✅ Pull-wire-driven deflection within the published parameter space (small and large deflection verified, tip angle within a fraction of a degree of a high-fidelity reference; see Steering simulator accuracy)
• ✅ Push and buckling behavior under the ideal restraints (verified against Euler; see Buckling); absolute push forces inherit the stiffness accuracy above
• ⚠️ Bending stiffness of a coil under a braid (median error 18%): take more care
• ⚠️ Braids with three or more filaments per carrier, and micro-catheters below 3 Fr (1 mm OD): take more care
• ⚠️ Torsion compared across braid densities on steerable and two-overlapping-braid shafts: take more care
• ⚠️ Torsion of coil-only and triaxial sections (triaxial: 7 sealed designs, median 17%): take more care
• ⚠️ One design in four is more than 15% off in bending or torsion (90th percentile: 28% EI, 37% GJ). The calculator does not yet flag a design outside the range it was trained on; treat unusual constructions as the tail
• ⚠️ Bend limits (kink / collapse): the failure mechanism is the most stable output; treat radii as screening values and confirm the finished design by bench test (see Bend limits)
• ⚠️ Materials that differ from datasheet values: characterize and override the Materials Library entry

---

Catheter calculator accuracy (EA / EI / GJ)

Reference document: Validation of the VirtuCath Stiffness Model, version 1.0 (October 2026). 📄 Download (PDF)

Catheter design needs the stiffness of each shaft section before anyone builds a sample. A catheter wall is not a classical laminate: the wires cross, press on each other and can stick or slide where they cross, and the polymer is 100 to 10,000 times softer than the wire, so the way the wires engage sets much of the bending and torsion response. VirtuCath resolves that wire mechanics in a wire-level finite-element (FEM) model, with friction and stick-slip behavior at the crossings, and carries it into the fast calculator you use in the app. The calculator learned from 2,228 FEM solutions across six construction families: single braids, steerable shafts with pull-wire lumens, coils, coils under braids, two overlapping braids and triaxial braids. Both models use datasheet material properties, and no constant in either model was fitted to bench data.

These stiffnesses are also the main inputs to the steering, kink and pushability simulations, so their accuracy carries into every result built on them.

How the models were checked

The checks follow the ASME standards for verification and validation of computational models (V&V 10, 20 and 40). Each acceptance criterion was written before its check was run.

Question	How it was checked	Result
Does the code solve the equations correctly?	Exact solutions, scaling laws, symmetry and limit cases	All within tolerance
Is the numerical error small?	Grid convergence index on 36 designs	Median 0.1–2%
Does it agree with independent theory?	Six benchmarks against closed-form theory and independent models	0–5% (one bending-dominated case 10–15%)
Does it agree with measurements?	Seven trusted bench comparisons	Median 6.9% (calculator), all inside the plan limits
Does the calculator reproduce the FEM model?	1,433 designs it did not learn from, four kinds of split	Median 8.7% / 6.6% / 2.1% (EI / GJ / EA)
Does the frozen calculator hold up on a blind test?	Sealed test of 400 designs, scored once after the freeze	Median 7.9% / 8.3% / 2.6% (EI / GJ / EA)
Blind bench check	Shafts built and measured without access to the predictions	Future work

Bench validation

The trusted bench data are a published 7 Fr shaft built in three braid patterns (regular, half-load and diamond), with measured EI and GJ, and a third-party coil-under-braid shaft with measured EA. The limits were set in the validation plan before the final evaluation: a median error of 15% or less, and no error above 25%.

Shaft	Stiffness	Calculator error
Regular braid	EI	−3.1%
Half-load braid	EI	+4.0%
Diamond braid	EI	−6.9%
Regular braid	GJ	+9.1%
Half-load braid	GJ	+2.2%
Diamond braid	GJ	−17.7%
Coil under braid	EA	+7.4%
Median / largest		6.9% / 17.7% (limits 15% / 25%)

All seven comparisons are inside the limits, for the calculator and for the FEM model it learned from (median 9.0%, largest 20.7%). The largest error is the torsion of the diamond-pattern shaft, a sparse net whose torsion depends on the details of the wire joints. Shafts tested on an in-house three-point-bend rig agree in direction and are reported in the paper.

Accuracy on new designs

The sealed test is the blind check of the calculator on designs it never saw. The 400 sealed designs were kept apart from all training and model selection, the calculator was frozen before any sealed result was read, and the set was scored once with metrics fixed in advance. The calculator in the app is the frozen one.

Construction family	EI	GJ	EA
All usable designs (397 of 400)	7.9%	8.3%	2.6%
Steerable (pull-wire lumens)	5.2%	4.4%	1.2%
Single braid	9.8%	9.7%	2.1%
Coil only	5.1%	11.1%	7.0%
Coil under braid	18.4%	9.5%	6.9%
Two overlapping braids	7.0%	4.1%	4.6%
Triaxial braid (7 designs)	3.2%	16.6%	0.7%

Median error of the calculator against the FEM model. Across all designs, 74% were within 15% in bending, 73% in torsion and 91% in axial stiffness; the 90th-percentile errors are 28% (EI), 37% (GJ) and 13% (EA), so one design in four is more than 15% off in bending or torsion. Two overlapping braids rest on 10 sealed designs, triaxial braids on 7, and the physics engine itself failed to converge on a third of the triaxial designs it was given, so those two families carry the least evidence. Within each family, the calculator ranks designs in nearly the same order as the FEM model. Bending of a coil under a braid is the weakest point; treat those values with more caution.

For a design decision: on a four-way steerable catheter, a 10% error in EI changes the predicted tip angle by about 6% and the pull-wire force by 4–5%.

Comparing designs

Even when a design's absolute value is off, two similar designs carry much of the same error, so the difference between them is predicted much more closely. Twelve base designs were each varied one input at a time (braid density, wire width, jacket stiffness, wall thickness, coil pitch), and the ratio between each variant and its base was compared with the FEM model:

	EI	GJ	EA
Error of each design on its own (median)	9.7%	7.7%	2.4%
Error of the ratio between two similar designs (median)	2.4%	1.6%	0.8%

The calculator also moved in the right direction on every bending and axial change and on 160 of 169 torsion changes. This is what makes VirtuCath suited to A/B design comparisons (two braid densities, two jacket grades, two wall thicknesses) and to iterative design, where the question is which configuration is stiffer, and by how much? Use the calculator to screen and compare, then build and test the selected design. VirtuCath does not replace design verification tests.

Limits

• No blind bench validation yet. The trusted bench shafts helped choose between modeling options during development (no constant was fitted to them, but which terms the calculator keeps was decided with them in view). A blind bench campaign, with shafts built and measured without access to the predictions, is planned.
• Ideal construction. The models do not include voids, unwetted wires or process variation.
• Datasheet material values. Datasheet moduli are measured on standard specimens, dry, at room temperature. Extrusion, orientation and reflow change the polymer in a real shaft, and so do test conditions such as body temperature (37 °C) or a hydrated shaft (see Material properties matter).
• Scope. Linear-elastic section stiffness at the curvature and twist of a standard bench test. Less common constructions (triaxial braids, three or more filaments per carrier, micro-catheters below 3 Fr) have the least evidence.

Closed-coil pull-wire lumen liners

A pull-wire lumen can be lined with a closed metal coil instead of a polymer tube (Catheter Setup, Construction). The coil is modeled as a plain closed spring, not as a composite: Wahl bending stiffness for a fully closed single-filar helix, the torsion-spring rate in torsion, a solid metal annulus in axial compression (the turns already touch) and a helical spring in tension. At the lumen offset the coil is credited through the composite-beam neutral axis rather than a parallel-axis term about the centroid, because a metal coil whose compressive stiffness dwarfs the polymer wall pulls the neutral axis onto itself; the resulting credit is roughly the wall's own axial stiffness times the offset squared, always below a solid metal tube of the same wall. On the shipped default steerable a 0.05 mm 304V coil liner raises the proximal shaft's EI by about 1.7x and its compression stiffness by about 3x, while a coil in the flexible section stiffens it by about 1.9x. The tension side is published separately as axial_stiffness_EA_tension_N. The bending form is the free-helix value and ignores coil-to-coil bearing on the compressed side, so it is a lower bound. These are simulated values, not measurements.

Strength-limit KPIs in the Failure Mode panel

The remaining Failure Mode Analysis values (Tensile: Peak Force, Torque at Failure, Burst Pressure, and Crush Pressure, all marked estimated) are governed by material strength limits, wall geometry, and the wire contribution rather than the composite stiffness chain. They inherit the usual datasheet-strength uncertainty (~10–20%) plus the error of the underlying stiffness prediction.

Tensile: Peak Force and Torque at Failure estimate the load at which the first material in the section reaches its strength limit, rather than ultimate rupture. Polymers are taken as linear up to their tensile strength, so a stiffer polymer that yields first starts to deform permanently at a lower load. In tension a braid or coil wire strains as the wall around it does, axial stretch less the wall's narrowing. For tension in particular, real rupture is higher, since ductile wires keep carrying load past first yield. These are typically accurate within a factor of two and usually conservative. Tensile predictions can be optimistic for sections with thin liners, where stress concentrations cause real failure to occur sooner than an analytical model predicts.

Each of the four has a First Yield (estimated) row above it: where the first layer starts to take a permanent set, on the same model with each layer at its published yield. These are estimates from published material limits (each material's datasheet or literature yield strength, or nitinol's set onset), with no bench anchor and no standard behind them (no catheter standard defines a yield endpoint). They are shown only when every layer that could yield first has a published yield. A published yield strength is itself a low-end datasheet value measured on a standard specimen, so the same material caveats apply as for the failure values.

Burst Pressure is the pressure at which the wall ruptures, with the tip closed off as in a burst test. The wall is solved as a laminate free to lengthen, so a braid shares hoop load only as far as its wire angle allows, and beside a pull-wire lumen only the wall around the lumen carries load. Each layer carries load only up to its own limit: the reinforcement wires reach theirs early and stop carrying load, and the polymer takes the rest until it bursts. The reported value is the highest pressure the wall holds along the way, which for most reinforced walls is the polymer bursting. Against exact multilayer elasticity on 174 walls of the same materials, the median difference is 7 % (90th percentile 29 %). No bench burst measurement is available to compare with, so treat it as a comparative estimate.

Crush Pressure is the lower of two limits: the wall buckling inward as a ring, and the wall crushing through its strength. The ring is taken at its thinnest point, beside any pull-wire lumens, and its braid and coil bands carry the hoop stiffness the wire-level model gives them. Against the section crush simulation on the same wall, on walls that collapse before any layer creases, the ring limit has a median error of 7 % (4 % up to 10 atm, the vacuum range) and reads more than 30 % off on one wall in ten, mostly thick walls and walls with two braids. Most of that accuracy comes from the ring formula itself; the trained hoop stiffness of the bands changes what the simulation predicts (by a median 22 % on those walls) and has no bench crush data behind it yet. Walls with two overlapping braids have no trained hoop block and keep their bands as built. On a very soft wall, one squashed more than about 10 % around the hoop at the reported pressure, treat the value as optimistic. On a thick wall with pull-wire lumens (wall thickness above about half the radius), the ring limit can read up to about 3 times the section crush simulation; treat it as optimistic there too and confirm with the Kink Radius section model or a bench test.

Bend limits are not in this panel: they come from the Kink Radius simulation, covered next.

---

Bend limits: the Kink Radius simulation

The app's bend limits are simulated on the real cross-section, not estimated from a formula: the Kink Radius simulation bends each section's real cross-section step by step (the actual layup, the pull-wire lumens and their liners, including a closed-coil liner as a hoop-stiff metal ring, the bend direction) and reports the events it observes: the moment peak that leads to collapse, 50% lumen loss (the ISO 25539-2 / EN 13868 kink definition), and each layer's permanent set. Because the events are observed on your geometry rather than predicted by a formula, the simulation sees what no closed form can: which layer gives first, how the answer moves with bend direction, and what failing actually looks like for this design.

Reference document: Verification of the VirtuCath Kink Radius Simulation, VR-VC-004 version 1.3 (October 2026). 📄 Download (PDF)

The section mechanics are verified against exact solutions, following ASME V&V 10:

• Exact solutions. At small curvature the section's bending stiffness matches exact beam theory to within 0.16%. For a thin tube, the moment peak (the Brazier limit point) converges to the exact thin-shell solution, solved without Brazier's small-flattening assumption: within 0.014% in curvature and 0.012% in moment. Thicker walls depart from it smoothly, as theory that leaves out the wall thickness predicts.
• Numerical error. On the sections of the bundled designs, doubling the mesh changed the collapse onset by a median of 0.022%. Sections with pull-wire lumens are meshed more finely; there the collapse onset is at most 10% tight: it is where the liner beside a lumen first folds, and a finer mesh resolves that fold at a slightly gentler bend. That mesh is a deliberate choice that keeps the run at a reasonable duration.
• Wire-level model. Each braid and coil band's stiffness against flattening is set from VirtuCath's wire-level stiffness model, by a law derived from other designs (the law's form was chosen with the seven catheters below in view; no constant was fitted to them). On seven thin braided and coiled catheters not in that law's training data, the collapse onset falls inside the wire-level estimate for 6 and errs on the cautious side by 1.8% for the other.

The verification does not yet include a pre-registered comparison of predicted kink radii with physical catheters.

The failure mechanism is the output the tool leads with, because it is the most stable. Across a sweep of solver settings, how a section ends (collapse, lumen closure, self-intersection) agreed in 29 of 30 runs. That stability is numerical, not bench validation: it says the mechanism belongs to the design rather than to the solver settings. The tool reports it first and lists the radii beneath it as the evidence.

Radii are shown at the precision the physics supports. Displayed radii are rounded to the nearest millimeter (0.05 in): a bend radius is a size, and the display is honest about that. Where a section ends by snapping rather than by a smooth moment peak, the snap radius is reported as an endpoint (the section snaps near here) rather than dressed up as a converged number. Exports carry full precision.

Permanent set comes from published material limits. Each layer's permanent-set limit is the strain past which its material stays deformed (a lasting set): the published set onset where one exists (nitinol), a metal's 0.2 % yield strength over its modulus, or, for a polymer, its published yield strain (0.7 of it for glassy polymers), or yield strength over modulus for an offset yield and for PTFE-class liners. These are estimates from published material limits, not measured on catheter shafts. Elastomers (TPU, silicone, PVC, soft Pebax), for which no set strain is published, keep a stiffness-based estimate anchored on observed kink events on production catheter constructions and consistent with the structural strain limits the flexible-pipe industry codifies for the same polymer classes (API 17J). Either way permanent set is reported below the Kink headline rather than as it. The behavior this produces (reinforced constructions limited by strain, unreinforced tubes by collapse) matches the pattern across several hundred published minimum-bend-radius specifications mined from manufacturer catalogs: Freelin-Wade, Parker Parflex, NewAge Industries and Zeus for plastic tubing; Alfagomma/Kuriyama, Parker, Aflex and Festo for hydraulic hose, alongside the SAE 100R-series tables; and Performance Pipe and DuraLine field-bending guidance for PE pipe.

Bench validation of the simulated radii is an early, growing program. The bench-kink dataset today is a handful of production constructions: not yet enough to state accuracy statistics that would mean anything, and this page does not publish statistics it cannot stand behind. What the data so far supports is the trust ordering below, and it is why the tool's own guidance is to rank in the app and confirm on the bench.

The tool does not report what it cannot measure. A solid multi-lumen core has no open bore to ovalize into a limit point, so for cored sections the tool reports material strain and the geometric endpoints and states that collapse does not apply, rather than printing a number that is not a measurement. The same rule runs through the results: an event the ramp did not reach says not reached, and a quantity that was not computed is left blank rather than shown as zero.

How far to trust a given prediction

The simulation solves the cross-section mechanics (ovalization, collapse, lumen loss) from the actual layup, so the tiers below are not about the geometry: they are about the material strain limits the solved strains are judged against.

• Most reliable: layers whose limit is published for the material itself (metals, nitinol, ETFE); the Basis column on the Kink Radius tab reads published.
• Reasonable: polymer layers whose limit is derived from a published yield strength (Basis derived), and elastomer layers on the stiffness-based estimate (Basis estimated (stiffness law)). Treat the permanent-set radius as a good estimate rather than a number to design a margin against. The Kink headline (collapse and lumen loss) does not depend on this limit and is unaffected.
• Indicative: thick-walled sections (wall approaching a third of the section radius). Creasing there is governed by plasticity the elastic material model does not carry, and the tool flags it on the run. Rank designs here; confirm the winner on the bench.

Predictions assume a single monotonic bend at room temperature; cyclic kink limits run roughly 2× tighter. Where a specific assumption is strained on a specific section, the app says so in that value's tooltip rather than leaving you to work it out.

The intended workflow is the one that works everywhere in this tool: use the simulation to explore, rank and narrow the design space (it will tell you which construction is better, which bend direction is weaker, and why), then confirm the finished design's bend limit on the bench.

---

Buckling: the Pushability simulation

Reference document: Verification of the VirtuCath Pushability Simulation, VR-VC-003 version 1.1 (October 2026). 📄 Download (PDF)

The Pushability simulation pushes the free distal length of the shaft against a tip restraint and predicts the push force, the buckling knee, the buckled shape and any wall contact. The report checks that the solver solves its equations correctly, following ASME V&V 10:

• Exact solutions. For a uniform column under each of the four ideal tip restraints (free, pinned, guided, clamped), the push force after buckling matches Euler's exact elastica to within 0.13% at the default mesh of 50 elements, and the buckled shape to within 0.07%. Two-step columns match as closely.
• Buckling knee. At the default starting bow, the knee of the force–stroke curve is within 1.4% of the Euler load for every tip restraint.
• Numerical error. The force error falls with the square of the element size, and barely moves with the number of stroke steps or the stiffness of the tip contact.
• Contact and friction. A buckling shaft first touches a channel wall where the exact shape reaches the wall, and the tip friction stays at the Coulomb limit while the tip slides.

As everywhere in this tool, relative comparisons between designs and scenarios are the most reliable output; absolute push forces inherit the accuracy of the section stiffnesses they are computed from (see the bench validation above). The verification does not yet include a comparison of predicted push forces with physical catheters.

---

Steering simulator accuracy

Reference document: Verification of the VirtuCath Steering Simulation, VR-VC-001 version 5.0 (October 2026). 📄 Download (PDF)

The steering simulator powers the runtime physics on the Steering Simulation tab. It takes each section's stiffness and finds the catheter's static equilibrium as the pull wire is drawn back, giving the full deflected shape, the tip angle and the wire tension. Unlike the calculator, which predicts section properties, the simulator's job is to predict the full-length deflected shape given those properties as inputs. The report checks that it solves its equations correctly, following ASME V&V 10, in four steps:

• The code. The solver's analytic first and second derivatives of the catheter's energy agree with finite differences to better than one part in ten million, and automated tests cover release, gradual drive and refinement.
• Exact solutions. For 500 random segments bending under their own weight, the solver matches classical beam theory with a median error of 0.18% in tip deflection. For a segment bent by a pull wire, whose exact shape is a circular arc at any deflection, it departs from the exact tip angle by a median of 1.2%, slightly on the stiff side.
• Numerical error. Every tenth case was re-solved with two to four times as many bodies. The results changed by a median of 1% or less, and the error falls steadily as the chain is refined.
• A high-fidelity reference. The reference is the Cosserat-rod model of Rao et al. (2021), the most accurate model in that benchmark study, run at full resolution. Over 949 one- and three-segment catheters bent by up to 180°, the median tip-angle difference was 0.13° and 0.06°, and the median wire-tension difference 1.0% and 0.33%. No catheter differed by more than 4.3° in tip angle or 4.8% in tension; the largest differences are on the tightest bends.

What this means for design work

When the pull is the input, as in the app, expect the tip angle within a fraction of a degree of the exact mechanics of the model, and the wire tension within about 1%. The solver's small discretization error has the same sign in almost every catheter (it bends slightly less than the exact arc for a given tension), so it largely cancels when two similar designs are compared. These errors are small next to the uncertainty of the section stiffnesses the simulator is given, so the stiffness inputs, not the solver, set the accuracy of a steering prediction for a real catheter.

What the simulator verification covers, and what it doesn't

These studies verify the mathematical accuracy of the steering solver against exact solutions and a continuum-rod reference of the same idealized problem: one pull wire, planar bending, uniform segments, an inextensible backbone and wire, and no friction, contact or external load. The app's options for wire friction, several wires and the proximal fixture are covered by the code checks only. They do not yet include direct validation of the deflected shape against physical catheter prototypes; that is the next step. (The section stiffnesses the simulator consumes as inputs are separately validated against bench measurement; see Bench validation above.)

---

Material properties matter

Both the calculator and the simulator rely on the polymer and metal properties from the Materials Library matching your production materials. Library values are sourced from manufacturer datasheets and the published literature (hover a value in the Materials Library to see its published range), but real-world material behavior varies: lot-to-lot variation, processing history, moisture content, aging, and strain-rate sensitivity can shift modulus by 5–15% from datasheet values, especially for polymers.

If your production materials deviate from the library defaults, predicted stiffnesses (and therefore the simulated dynamic response that depends on them) will shift accordingly. Datasheet moduli are measured on standard test specimens, dry, at room temperature; the polymer in a finished shaft has been extruded, oriented and reflowed, and may be used at body temperature or hydrated. Because bending stiffness scales roughly in proportion to the constituent moduli, a difference in modulus carries almost directly into predicted EI. For tighter agreement with your build, characterize your specific materials under your use conditions and override the library entries with measured values. Material characterization is the single most effective lever you have to tighten agreement.

Related pages

• Kink Radius: the simulation behind the bend limits covered above.
• Pushability: the simulation behind the buckling results covered above.
• Catheter Analysis: where the predicted stiffnesses are shown.
• Simulation Readouts: runtime simulator metrics that depend on the calculator outputs covered above.
• Materials Library: the polymer and metal property defaults referenced above.