AirShaper vs SimScale vs Ansys vs OpenFOAM

Four approaches to CFD software compared on accuracy, consistency, speed, cost and the expertise they need

The short answer

There is no single best CFD software: the four options below make different trade-offs between control, cost and the expertise they need. Ansys and similar packages such as Simcenter STAR-CCM+ and PowerFLOW give a specialist full control over every setting, at a high cost in licenses, hardware and expertise. SimScale moves that expert workflow to the cloud, which removes the investment in software and hardware but still leaves the setup to the user. OpenFOAM is free, open source and the most flexible of all, but needs the most expertise; AI assistants make it easier to get a case running, not to know whether the answer is right or consistent. AirShaper delivers a simulation result, not just an interface: a fully automated workflow built on OpenFOAM and validated against wind tunnel and other tests, which sets up every simulation the same way, with aerodynamicists on hand who know how to design cars, drones and other products.

AirShaper publishes this comparison, so its own product is described in the most detail. Information on the other tools comes from their public websites and documentation, as of November 2026.

At a glance

Ansys, STAR-CCM+, PowerFLOWSimScaleOpenFOAMAirShaper
What you getSimulation software you set up and runA cloud platform for setting up and running simulationsAn open-source CFD toolboxA simulation result from an automated workflow, not just an interface
Who runs itCFD specialistsEngineers with simulation experienceCFD specialists at ease with the command lineEngineers and designers, without a CFD specialist
InterfaceDesktop applicationWeb browserText files and command line; ParaView for resultsWeb browser, or an AI assistant through MCP
Model preparationCAD usually cleaned up or wrapped before meshingUpload CAD; geometry and mesh set up by the userClean surface geometry, meshed with tools such as snappyHexMeshNon-watertight models accepted; meshing automatic, with adaptive refinement
SettingsFull controlFull control within the platformFull control, nothing presetSet by automated algorithms; you simply set the orientation, speed, flow rates, rpm and so on
ConsistencyDepends on the user and in-house proceduresDepends on the user's setupDepends on the user's setupSame workflow for every run, including convergence detection and averaging
ValidationWidely validated solvers; each setup still needs correlatingValidated solvers; the setup is the user's responsibilityLarge body of published work; each case setup needs validatingThe workflow itself, in nine published studies
ResultsPost-processing tools included, set up by the userPost-processing in the browser, set up by the userParaView or your own scriptsPerformance figures and interactive 3D visuals, the same for every run; 3D comparison of designs; mesh and OpenFOAM data to download
PhysicsBroadest: CFD plus structural, crash, electromagnetics, electronics cooling and moreCFD, structural, thermal and low-frequency electromagneticsMany flow solvers, including multiphase, heat transfer and combustionAerodynamics, hydrodynamics and internal flow, including multiphase (free water surface) and thermal; transient in beta
HardwareYour own workstations or cluster, or cloud optionsCloudYour own, or cloud capacity you manageCloud
PricingQuoted on request; commonly tens of thousands of euros per seatFree Community plan with limits; paid plans quoted on request, usage counted in core hoursFree software; the cost is time, expertise and hardwareFree account to try it; annual plans an order of magnitude more affordable than a CFD license to get started
Design improvementAdjoint and optimisation tools, set up by the userAI add-ons on paid plans, such as Physics AIAdjoint solver in the openfoam.com line; optimisation loops scripted yourselfDesign Advice (adjoint) and Morphing, automated and stable in the browser
Aero and hydro application expertiseSoftware support; engineering services offered separatelyPlatform supportCommunity forums; support and training from third partiesAerodynamicists who know how to design cars, drones and more, and helped create world-record designs; consultancy paid with plan credits

What makes CFD useful in design work

CFD simulations are very useful in engineering, but notoriously complex. The quality of the result depends on many choices along the way:

  • the quality of the 3D model;
  • the mesh, which represents the model in the simulation;
  • the boundary conditions, which describe what drives the flow;
  • the equations that describe the physics of the flow, including the turbulence model;
  • the solver that finds an approximate solution to those equations, how many iterations it runs, and how it decides that the solution has converged.

Each step brings modelling, discretisation or iteration errors of its own (see Computational Fluid Dynamics Explained), and they can be small or large.

For CFD to be useful in industrial design, results have to be:

  • Accurate. With a wide error band, small changes in geometry or speed cannot be compared.
  • Consistent and repeatable. When results vary over time (because different methods are applied), between users (each with their own preferences) or between tools, it is hard to see the bigger picture.
  • Fast. Delivered within a time frame that fits the design process. That includes throughput: if preparing a model takes hours or days, or post-processing is done by hand, the whole workflow slows down.
  • Affordable. Costs under control for software, maintenance and hardware.
  • Accessible. If simulations need an expensive expert, a wait for the aero department, or a tool only a few people can operate, CFD becomes less useful.
  • Easy to share. Results that everyone involved can open and discuss.
  • Clear and comparable. Raw data has to be translated into the performance figures that matter and shown in clear, consistent visuals, so that non-experts can understand the results and different designs can be compared properly.

The four approaches below score differently on each of these.

Ansys, STAR-CCM+ and PowerFLOW: the conventional approach

Ansys, part of Synopsys since July 2025, has made engineering simulation software for decades and represents the conventional approach: buy the software, buy the hardware to run it on, and hire an expert to operate it. Siemens' Simcenter STAR-CCM+ and Dassault Systèmes' PowerFLOW, a lattice-Boltzmann code widely used for vehicle aerodynamics and aeroacoustics, follow the same model.

This has been the default for years, and it is still the right choice in many cases. If you run simulations day in, day out, and want precise control over the case setup, it makes sense. In automotive or aviation, setups are very similar from one run to the next, so it pays to fine-tune them to the maximum and develop your own post-processing workflow. If you want full control over every setting and have the expertise to take responsibility for the output, these packages give you that.

Ansys also covers far more than CFD. Next to its CFD solvers Fluent and CFX, the portfolio includes Ansys Mechanical for structural analysis (FEA), LS-DYNA for crash and other explicit dynamics, HFSS and Maxwell for electromagnetics, Icepak for electronics cooling, and AI tools such as SimAI. Siemens' Simcenter and Dassault's SIMULIA ranges are similarly broad.

The challenges:

  • Model preparation. CAD usually has to be cleaned up into a watertight model before it can be meshed, or wrapped with the package's own tools, which takes skill and time.
  • Investment. Licenses are quoted on request. A commercial CFD seat commonly costs tens of thousands of euros, plus yearly maintenance or lease fees, and the hardware to run it on.
  • Expertise. These packages are not easy to work with, and the largest cost is often hiring and paying an expert who can operate them.
  • Consistency. Different users produce different results, depending on their preferences, experience and the hardware available.

The investment pays off when CFD is a full-time activity.

SimScale: the expert workflow in the cloud

Simplified a great deal, SimScale is an online version of the conventional approach: a browser interface in which users with simulation expertise set up and run simulations, with a choice of solvers. Its CFD runs mainly on OpenFOAM, covering incompressible and compressible flow, conjugate heat transfer and multiphase flow, with a lattice-Boltzmann solver for incompressible flow and pedestrian wind comfort. It also offers structural analysis with Code_Aster and Marc, thermal analysis and low-frequency electromagnetics. Some analysis types need a higher plan.

Instead of buying a license, you buy an annual plan and run simulations on SimScale's cloud. That removes a big barrier, the investment in software and hardware. Running many cases in parallel speeds up the workflow, and storing projects online makes collaboration across teams easier. A free Community plan offers selected analysis types with a limited number of core hours and simulations. The paid plans (Mechanical, Professional and Enterprise) are priced on request, with usage counted in core hours, one core hour being one computing core for one hour. Optional AI features include a copilot (Virtual Engineer), Physics AI and, on the Enterprise plan, agentic Engineering AI.

It is still an engineering-oriented approach: the user sets up the case and is responsible for the quality of the output, so results depend on the quality of each user's input, which varies.

AirShaper: a simulation result, not just an interface

AirShaper takes a fundamentally different approach: what it offers is a simulation result, not just an interface to simulation software.

A fully automated workflow, validated against wind tunnel tests, accepts non-watertight 3D models, sets up the boundary conditions, runs the simulation in the cloud and presents the results in 3D in the browser. Settings specific to the simulation, such as meshing and solving, are not left to the user but set by algorithms: you simply set the operating conditions, such as orientation, speed, flow rates and rpm. Those algorithms, including adaptive mesh refinement, work the same way every time. They also detect when a simulation has converged and then average the results, in the same mathematically consistent way for every run, so results are highly consistent.

That consistency makes results comparable and precise, speeds up the workflow and opens CFD to a much wider group of people than specialists. It also avoids endless iterations between design and aero teams, as designers can check a change themselves. Results are translated into the figures that matter, such as drag, lift and downforce, and shown in interactive 3D visuals that look the same for every run, so non-experts can read them and designs can be compared side by side in 3D. The sample simulations show what a result looks like. Comparison and data downloads depend on the plan.

The workflow has been tested in nine published validation studies, most of them by engineers outside AirShaper. In DARcorporation's Cessna 210 study, AirShaper's curves were on average the closest to NASA's wind tunnel data of the three codes compared, one of them STAR-CCM+, though STAR-CCM+ came closer on drag at low angles of attack. AVL Racing compared AirShaper with its conventional CFD workflow on a touring car and found a 5.2% difference in drag and 3.6% in downforce.

AirShaper also goes beyond the analysis itself:

  • Design Advice runs an adjoint simulation that shows where reshaping the model would have the most effect (see the adjoint method video).
  • Morphing reshapes the model automatically within limits you set, as Aptera did for its solar electric vehicle.
  • AI assistants can connect to AirShaper through its MCP server, which gives their workflows a solid, repeatable and wind tunnel validated result to work with. Creating and running simulations from an assistant needs an annual plan.

Adjoint solvers and shape optimisation exist in conventional packages too, but they are hard to get running stably and usually need a specialist to set them up. AirShaper runs both automatically, in the browser, as part of the same workflow.

Aerodynamic and hydrodynamic application expertise

Another key difference is expertise in the application itself. Running a simulation is one thing; knowing how to design a faster race car, a more efficient drone or a lower-drag hull is another. AirShaper's aerodynamicists bring that experience to customer projects, and on Professional and Enterprise plans the same credits pay for both simulations and consultancy.

Their work includes electric vehicles (the TWIKE), Pikes Peak-winning race cars (Robin Shute's four wins), hypercars (the 7X Design Rayo) and the Peregreen V4, which set the Guinness World Record for the fastest battery-powered RC aircraft.

Pricing and limits

AirShaper works with annual plans that are an order of magnitude more affordable than a commercial CFD license to get started, and a free account lets you try it first. See pricing for the plans.

The trade-off is control. AirShaper sets the mesh and solver settings itself and does not let users change them, but it is transparent about the outcome: the mesh and the OpenFOAM data, such as velocity, pressure and turbulence fields, can be downloaded afterwards. It also focuses on flow: aerodynamics, hydrodynamics and internal flow, including multiphase and thermal simulations, with transient simulations in beta. For structural analysis, electromagnetics or physics outside that range, such as combustion, a general-purpose package is the better tool.

OpenFOAM: free, powerful and up to you

OpenFOAM is the most widely used open-source environment for CFD. It is free under the GNU General Public License and comes in two main lines, from the OpenFOAM Foundation (openfoam.org) and from OpenCFD (openfoam.com). Both SimScale and AirShaper use it at their core.

It is extremely powerful, but also complex to use. It has no graphical interface of its own: a case is a set of text files, run from the command line, and results are usually viewed in ParaView (see our ParaView tutorial). More than with any other package, the quality of the solution depends on the quality of the input and the setup.

It takes months to fully understand OpenFOAM, but if you have the time, skills and hardware, it is a very solid solution. Going through the learning curve yourself also gives you a deep understanding of the mathematics behind it, which is a solid base for interpreting results. AirShaper's course material covers CFD and OpenFOAM.

The software itself costs nothing. The real costs are longer lead times, the need for an expert, and hardware.

OpenFOAM with an AI assistant

AI models are trained on content from the internet. Forums such as CFD Online, code repositories and papers are a rich source of CFD knowledge, and it feeds into assistants such as ChatGPT and Claude.

That has made it a lot easier to set up your own OpenFOAM workflow with an assistant's help. Getting a working simulation still takes time: cleaning up the model, setting up the case and running it on your own or cloud hardware.

Even as AI models keep getting better, problems remain:

  • Consistency. AI keeps evolving, so the workflow it creates today is likely to differ from yesterday's or tomorrow's.
  • A working result is not a good result. You still need expertise to know which result is the right one.
  • Settings are a blend. The assistant chooses settings that are often a blend of what it found online, and even the smallest change in mesh or solver settings can shift results a lot, without you knowing.
  • Convergence. A simulation has to run until the solution has converged, and the results then need averaging. AI-generated workflows don't always take this into account, and when they do, not always in a mathematically consistent way.
  • No validation. These workflows are not validated against physical tests. AI can tune a setup automatically to match benchmark cases, which is useful, but that often leaves a setup biased towards a limited number of test cases.
  • Presenting results. Raw data still has to be turned into clear performance figures and consistent visuals, such as interactive 3D comparisons between designs. Building that, dashboards included, takes time with an AI-generated workflow too.
  • Missing industry knowledge. AI can suggest how to improve a design, but knowledge specific to an industry, such as race cars, drones, vehicles or propellers, is hard to find and often kept private, so it does not reach AI models. AI can perhaps get you from zero to 50%; the step from 50% to 100% is where the real competitive advantage lies.

Which one fits

  • Ansys, STAR-CCM+ or PowerFLOW if CFD is a full-time job for a dedicated team, you need full control over every setting or physics beyond flow, and the budget covers licenses, hardware and specialists.
  • SimScale if you have simulation expertise and want that control without buying hardware, or need structural and thermal analysis on the same cloud platform.
  • OpenFOAM if you have the time, skills and hardware, want full control without license costs, or want to understand CFD in depth.
  • AirShaper if you want reliable, comparable aerodynamic or hydrodynamic results quickly and consistently across your team, without a CFD specialist, and aerodynamicists who know your application when you need advice.

They are not mutually exclusive. A team with its own CFD department can use AirShaper to let designers test concepts early, leaving the specialists the cases that need full control. And as with any CFD, a physical test confirms the final design; see CFD vs wind tunnel testing.

Unlock High-Performance Designs with AirShaper