The cost of FEA or CFD analysis is driven mainly by engineering time, and engineering time is set by six things: how clearly the questions are defined, how much the geometry needs preparing, which physics are involved and how non-linear they are, how many load cases and design variants are compared, whether the behaviour is steady or changes over time, and how much reporting and design iteration is expected. A linear stress study of one clean bracket under one load sits at one end of the range. A transient conjugate heat transfer study of a complete enclosure, across several duty cycles and design options, sits at the other.
That is why this guide gives no figures. A number without a scope means very little, and two studies described in the same words can differ several times over in effort. What you can control is the clarity of the brief, the state of the CAD and the number of options you need compared, and those three decide most of what a quote contains.
The questions the study must answer
Scope is the largest single driver, and it starts with the question. “Is this strong enough?” leaves open which loads, which failure modes, which parts and what counts as strong enough. “Does the bracket stay below yield with a factor of 1.5 at 3 g vertical, and survive 10⁶ cycles at ±1 g?” defines the load cases, the criteria and the output. The second question is easier to price, quicker to answer and less likely to grow once work has started.
CFD works the same way. “How does the air flow through this unit?” is open-ended. “What is the pressure drop at 400 m³/h, and does each of the six outlets receive within 10 % of the mean flow?” describes a study with a defined end.
Geometry complexity and CAD clean-up
Design CAD is built for manufacture. It carries threads, logos, tiny fillets, small gaps and overlaps between parts, and sometimes hundreds of fasteners in an assembly. Each has to be simplified, repaired or deliberately represented before a usable mesh can be built. For CFD, the fluid region usually has to be extracted from the solid model as a closed volume, which exposes every gap and leak path in the assembly. On a large or untidy assembly, preparing the geometry can take longer than solving it. Our guide to preparing CAD for simulation explains what helps.
Physics and non-linearity
Each increase in physical complexity adds set-up effort, solution time and interpretation, in FEA and CFD alike:
| Driver | Lower effort | Higher effort |
|---|---|---|
| Structural behaviour | Linear static, parts bonded together | Contact that opens and slides, plasticity, large deformation, buckling |
| Dynamics | Natural frequencies and mode shapes | Harmonic response, transient dynamics, drop and impact |
| Fatigue | Constant-amplitude loading at one feature | Variable load histories across many locations |
| Flow | Steady, single fluid, simple inlets and outlets | Separation, rotating parts, multiphase flow, transient flow |
| Thermal | Conduction with assumed surface coefficients | Conjugate heat transfer with radiation |
| Coupling | A single physics | Thermal-structural, fluid-thermal or electromagnetic-thermal |
| Materials | Linear elastic, with published data | Hyperelastic, rate-dependent or anisotropic, needing test data |
Non-linearity matters because the load has to be applied in increments with the stiffness updated as it goes, and because the modelling choices, such as contact definitions and material models, call for more care and more judgement. A sensible scope often starts linear: a linear study can show whether a non-linear one is needed at all.
Load cases and design variants
Once a linear model is built, extra load cases add relatively little, because the same model is solved again with different loads, and linear results can be scaled and combined. In a non-linear study each load case is a separate solution, and the order in which loads are applied can change the answer, so each one costs more.
Design variants cost more than load cases, because each new geometry has to be prepared and meshed again. Comparing three heat sink designs is closer to three set-ups than to one. Where variants differ only in a few dimensions, building the model parametrically from the start reduces the cost of each extra option. Deciding at the outset how many options you want compared is one of the easiest ways to control cost.
Steady state or transient
A steady-state study finds the condition a system settles into: the temperatures at full continuous power, the flow at a fixed fan speed, the stress under a constant load. A transient study follows the system through time, with hundreds or thousands of time steps, larger result sets and more to interpret. Some questions need it: a motor warming up over a duty cycle, a pressure surge when a valve closes, a product hitting the floor. Many do not. If the worst case is a steady condition, a steady study answers the question for less.
Reporting and design changes
The deliverable affects the cost as much as the analysis does. A short summary with key results and a recommendation is a different piece of work from a formal report written for a customer or a technical file, with every assumption, load case and result set out. Agree which you need at the start.
Design changes are the other variable. Some studies end with a recommendation; others continue through two or three rounds of redesign until the part meets its targets. Iteration adds cost, but it is often where most of the value lies, particularly when the team that ran the analysis can make the changes in CAD.
Where computing time fits
Solver time is the part of a study people often expect to dominate, and it rarely does. Large transient CFD and explicit impact studies do use substantial computing resource, but for most studies the larger share of the effort goes into defining the problem, preparing the geometry, setting up loads and materials, interpreting the results and writing them up. Where studies run on cloud hardware, a large model, or several variants run in parallel, no longer depends on the size of anyone’s workstation.
Other things that move a quote
- Material data. Non-linear, fatigue, high strain rate or elastomer data may not exist for your grade, and sourcing or estimating it adds work.
- Codes and standards. Working to the code or standard you specify can prescribe load combinations, assessment methods and report content.
- Uncertain inputs. When loads or boundary conditions are not known, a study may need several cases to bracket them instead of one.
- Changing scope. New load cases, revised geometry or extra questions after work has started are a common reason for a study to grow.
Ways to reduce cost without losing the answer
- Use symmetry. Where geometry, supports and loads are all symmetric, a half or quarter model can answer a static question for less. Modal and buckling studies are the exception, because their modes need not be symmetric.
- Model only what matters. A bracket, its bolts and a representation of the stiffness of the frame it attaches to can answer a question about the bracket without modelling the whole machine.
- Start with the governing case. If one load case or operating point clearly dominates, study it first; the others may not be needed.
- Freeze the geometry. Agreeing one CAD revision for the study avoids rework when the design moves on mid-study.
- Match the deliverable to the decision. An internal design choice may need a summary and a recommendation rather than a formal report.
How to get a tighter quote
The more of the following you can send with an enquiry, the closer a quote can be to the work actually needed:
- The decision the study supports, and the specific questions it must answer.
- CAD in STEP or native format, or images and overall dimensions if files are not ready.
- Loads and operating conditions: forces, pressures, accelerations, temperatures, flow rates, duty cycles and how often each occurs.
- Materials and grades for each significant part.
- Acceptance criteria: allowable stress or deflection, required factor of safety, temperature limits, target pressure drop, and the code or standard you specify.
- The number of design options you want compared.
- The reporting you need, and who will read it.
- Any history: failures in service, problems found in testing, or earlier analysis of the same part.
Not all of this needs to be ready for a first conversation. Often the most useful starting point is a short description of the problem, which can then be turned into a defined scope.
When to bring in a specialist
If you are planning a study and want a realistic picture of what it involves, tell us about the project. SixGate™ shapes each study around the decision you need to make and returns results with the design changes they point to. Simulation can be bought on its own or as part of the detail design and simulation stage, in which case the design changes go straight into the detailed model, the engineering pack and production with no hand-over between firms.