No pages match that search.
t = 0 · design intent enters the solver
Oltrion turns simulation into a design engine. From nonlinear materials and turbulent flow to coupled multiphysics and physics-informed machine learning, we help engineering teams explore more of the design space — and commit with evidence.
02 · the nonlinear frontier
Plasticity, contact, turbulence, heat — and the way each feeds the others — are where designs are won or lost. Hand calculations stop at the edge of that frontier. A well-posed simulation goes past it, and shows why a part behaves the way it does, not only that it does.
Stress · von Misesillustrative field
Peaks at the root fillet and along the trailing edge.
Temperatureillustrative field
Hottest at the leading edge and the tip; cooling holes run along the span.
Velocity magnitudeillustrative field
In the flow around the section at 60% span: fastest over the suction side, stagnating at the leading edge.
03 · the connected physics stack
Choose a discipline: its field takes over the part, and the quantities it exchanges with the others draw between them.
Stress, strain, contact, fatigue and failure — linear and nonlinear, static and explicit dynamic.
Exchanges
Steady and transient flow, turbulence, aerodynamics and hydrodynamics.
Exchanges
Conduction, convection and radiation, and the thermal stress they cause.
Exchanges
Electromagnetic fields, eddy currents, Joule heating and plasma devices.
Exchanges
Noise, vibration and harshness — modal, harmonic, aero- and vibro-acoustic.
Exchanges
Sprays, cavitation, boiling, particulates and combustion.
Exchanges
Two-way coupling of flow and structure: pressure in, deformation back.
Exchanges
Physics-informed networks, surrogates and reduced-order models, trained on simulation and checked against it.
Exchanges
04 · mission selection
The equations are shared; the boundary conditions are not. Choose an industry and the part takes its load.
Aero + centrifugal load · illustrative
Impact load · illustrative
Thermal + pressure load · illustrative
Seismic lateral load · illustrative
Thermal cycling · illustrative
Centrifugal + gas bending load · illustrative
Hydrodynamic load + cavitation risk · illustrative
Physiological load · illustrative
Crash pulse · illustrative
05 · physics-aware AI
Thousands of degrees of freedom in one full-order field (synthetic here).
Proper orthogonal decomposition finds the few shapes that carry most of the energy.
Three coefficients rebuild the field in real time. Move them.
surrogate = x̄ + c₁·φ₁ + c₂·φ₂ + c₃·φ₃
ε: relative L2 error against its own full-order field.
Synthetic snapshots; POD and error computed in your browser. Illustrative.
AI does not replace validated simulation. It makes more of it affordable.
06 · fluid intelligence
Separation, recirculation and heat pick-up decide efficiency long before a prototype exists. This is a live incompressible solve in your browser — drag across it to stir the flow.
2D solve
07 · structural intelligence
Topology optimisation removes material that carries no load and keeps what does. The root and the aerofoil are fixed; only the platform may change. Strike the part to see how it rings.
Iteration
Mode shapes amplified for visibility.
08 · applied industrial innovation
Problems we write about in depth. Each opens the full method.
Reacting Flows & Combustion
In order to accurately model combustion, it is important to account for the interactions between turbulence, mixing, and chemical reactions.
Read the method: Detailed ChemistryAcoustics and Vibration
The noise generated by turbulent flows can come from various sources, including air conditioning systems, rotating fans, and other devices that create turbulence in the airflow.
Read the method: Acoustics in Automotive IndustryComposite Materials
Cohesive zone models (CZMs): CZMs are commonly used to model the initiation and propagation of cracks and delamination in composite materials.
Read the method: Composite modeling approach in numerical computationCrash Test Simulation
Frontal Impact: In this type of crash test, a car is propelled into a barrier at a specified speed to assess how well it can protect occupants in the event of a head-on collision.
Read the method: The main scenarios of crash testsEach piece shown in one physics — illustrative.
TensorLynx · OLTRION’s engineering platform
Fast predictions are easy to buy. Answers an engineering lead can sign are not. TensorLynx takes a design from parametric geometry to a result re-checked on the full model, in one loop, with every step on record.
01 Define
Refuses
A value that makes no sense. It is refused, with a clear reason.
02 Simulate
Refuses
A coupled set-up that would give a wrong answer. It is refused, with its correction.
03 Improve
Refuses
A gradient that has not passed its finite-difference check. It may not steer a new case.
04 Learn
Refuses
An answer the evidence does not support. The prediction is withheld and the solver decides.
05 Optimise
Refuses
A recommendation that ignores a constraint. Feasibility is designed in, not checked at the end.
06 Decide
Refuses
To choose for you. The trade-offs are visible, and the choice is recorded with its reasons.
07 Verify
Refuses
To call a design Verified because an optimiser liked it. Verified means re-run on the full model.
08 Record
Refuses
A claim that cannot be traced. Every run seals what produced it; the next cycle starts from the record.
09 · engagement model
Independent simulation and analysis of your problem: we set up, run, check and explain the model.
Solvers, automation and AI/ML tooling built into your engineering workflow.
Discuss a buildPractical FEA and CFD training for your engineers.
Plan trainingTools we work with
10 · converged
Tell us the geometry, the physics and the decision it has to support.
