How to choose optical simulation software?

Before you choose: what simulation actually delivers

Optical simulation is a numerical reproduction of light propagation in a system — ray paths or the full electromagnetic field — before the first physical element exists. Before comparing specific tools, it helps to know where the return on this investment comes from, because that determines the features you will need:

• Eliminating prototype iterations — machining a prototype aspheric lens or an injection mould can involve significant cost and lead time; in a digital model you run many geometry iterations in a comparable time.
• Shorter time to market — design errors surface at the model stage, not on the production floor.
• Scenarios hard to reproduce in the lab — extreme angles of incidence, sunlight for different latitudes, thermal conditions.
• Optimising manufacturing tolerances — statistical analysis shows which assembly tolerances can be loosened without losing the required performance, which lowers serial-production cost.
• Catching parasitic phenomena — stray light, internal reflections and ghosting, unwanted optical paths and scattering — before they reach the finished product.

Criterion 1: system scale

The most important question is: how large are the significant structures relative to the wavelength, and do wave phenomena — diffraction and interference — matter in the problem at hand? That indicates whether light can be treated as rays or whether a wave method is needed.

• Structures much larger than the wavelength, where diffraction and interference need not be modelled explicitly → ray tracing: TracePro or OSLO.
• Waveguides and photonic structures whose propagation is essentially directional → the Beam Propagation Method (BPM), e.g. OptiBPM — depending on the nature of the problem.
• Structures with dimensions comparable to the wavelength, and problems that require full-wave modelling of the electromagnetic field → FDTD, e.g. OptiFDTD or OptiOmega.
• Transmission systems → system-level simulation in OptiSystem, where transmission and signal-quality parameters (e.g. BER, Q-factor) are analysed at the level of the whole system.

Criterion 2: imaging vs. illumination

If your system falls within ray-tracing scale, the second question is: are you forming an image, or distributing light? This determines the choice between two ray-tracing modes.

• Imaging systems (lenses, microscopes, spectrometers) → sequential ray tracing in OSLO. Rays pass through surfaces in a fixed order; what matters is image quality, aberrations, MTF, and diffraction analysis.
• Illumination and non-imaging systems (reflectors, LED luminaires, stray light) → non-sequential ray tracing in TracePro. Rays hit any surface in any order, reflecting and scattering multiple times; what matters is illuminance, flux, and optical efficiency.

In practice TracePro is particularly strong in analysing illumination, stray light, scattering, and multiple ray interactions with 3D geometry, whereas OSLO is primarily a tool for designing and analysing optical systems, especially imaging ones. The functional ranges do overlap in part, though — TracePro now also has a Sequence Editor for sequential analyses, and OSLO supports selected non-sequential ray-tracing scenarios.

Criterion 3: design stage

In integrated photonics (PIC), the tool also depends on the modelling level you are working at — from a single component to the whole system:

• Component (waveguide, coupler, modulator) → OptiFDTD or OptiBPM; waveguide mode analysis — OptiMode; Bragg gratings and FBGs — OptiGrating.
• Circuit level → OptiSPICE, where you combine optical and electronic component models.
• System level → OptiSystem, i.e. evaluating the whole transmission link.

Good practice is to move from component, through circuit, to system; each stage checks the design from a different perspective.

Criterion 4: CAD integration and automation

A tool that does not plug into your data flow creates a hidden cost: manual geometry export and conversion errors. When comparing solutions, check:

• CAD integration — RayViz lets you define optical properties and run ray tracing directly in the SolidWorks model; the full analysis is then done in TracePro. Ray collisions with the housing are visible already at the model stage.
• Material and source libraries — ready-made characteristics of LEDs from leading manufacturers, laser parameters, sunlight models. The absence of such a library means hours of entering data by hand.
• Automation and scripting — macros (the Scheme language in TracePro) let you automate repetitive analyses: parameter sweeps, reports, optimisation loops. Key when you calculate the same type of system many times.
• Optimisers — automatic search for system parameters against defined criteria; in OSLO typically for imaging systems, in TracePro for illumination and non-imaging systems.

Criterion 5: tolerance analysis and serial production

A mathematically correct design can fail at assembly if it does not account for process reality. If you are planning serial production, the tool must allow:

• sensitivity analysis — which parameters (lens decentration, radius-of-curvature error, air-gap thickness) most strongly degrade system performance;
• statistical simulation (Monte Carlo) of production spread and determination of acceptable tolerances;
• an assessment of whether the system can be produced without individually tuning each unit.

OSLO and TracePro run this analysis in their respective domains — imaging and illumination.

Numerical methods in brief

• Ray tracing — geometric ray tracing; fast, for systems much larger than the wavelength. Geometric ray tracing alone does not directly model the wave nature of light; some programs, however, offer additional models or analyses that account for diffraction effects.
• BPM (Beam Propagation Method) — efficient for long waveguide structures whose propagation is essentially directional.
• FDTD — full-wave modelling based directly on Maxwell's equations; appropriate for structures on the order of the wavelength (nanophotonics, metasurfaces), but computationally expensive.
• Mode solver (e.g. FEM or the finite-difference method) — determines the modes and their parameters in a waveguide or fiber cross-section; the basis for bend-loss and inter-mode coupling analysis.

Quick cheat sheet

Licensing model and total cost of deployment

The licence price is not the only cost component. When planning the budget, account for:

• Licensing model — a time-limited/subscription licence gives flexibility for shorter projects or a changing team composition, while a perpetual licence can be more economical for multi-year use. The terms depend on the vendor and the product.
• Academic and educational licences — the terms depend on the vendor, the product, and the use case. Lambda Research runs a university programme for teaching use (including TracePro, RayViz, OSLO; time-limited licences) and separate terms for academic research. Feature availability and licence duration have to be checked for the specific program.
• Training — topic-focused workshops in Polish shorten the learning curve; omitting this item from the budget is the most common reason a tool's potential goes unused.
• Technical support — direct contact with an engineer who knows the tool shortens downtime with modelling and licence-configuration problems.
• Updates — keeping the current version through a local partner ensures a stable environment during long projects.

iSymulacje — a brand of SPECTROPOL, the Polish distributor of Lambda Research Corporation and Optiwave — helps match modules to real needs, so you do not pay for features your projects will not use.

The most common selection mistakes

• Ray tracing for wave phenomena — calculating diffraction or interference with geometric tracing alone; such problems need a method that accounts for the wave nature of light — e.g. FDTD, and where appropriate BPM, RCWA, or physical-optics methods.
• FDTD for a whole macroscopic system — a full-wave simulation of a centimetre-scale object can be computationally impractical because of the enormous number of grid cells and the cost of the computation; ray tracing is the right choice there.
• Confusing sequential with non-sequential mode — stray light analysis in sequential mode usually misses exactly the chance reflections and scattering you are trying to find.
• Skipping the system stage — a correct component does not guarantee a working link; the BER and Q-factor budget is decided at the system level.
• Buying an oversized package without a needs analysis — the features you pay for go unused.
• Skipping CAD integration — manual geometry export lengthens every cycle and introduces errors.
• No local support or training budget — without appropriate training, some of the advanced features may go unused.