This tutorial for a fluid mechanics engineering office builds a traceable calculation file, from the archived case to the report.

Start with tutorial no. 1

A fluid mechanics engineering office does not deliver numbers: it delivers defensible numbers. What sets the two apart lies in what accompanies the result — the assumptions, the verification, and the possibility for a third party to redo the calculation.

This last tutorial in the series computes nothing new. It shows how to turn the calculations from the previous nineteen into a file.

• Prerequisites: one or more computed cases;
• Duration: the time to gather the pieces of already-calculated cases, to be done once to establish a reusable template;
• What you get out of it: a file that withstands scrutiny, and a case you can reopen in two years.

• Tutorial outline:

1. Archiving the case, not just the result
2. Documenting mesh convergence
3. Stating the assumptions and limitations
4. Producing the report
5. Documenting the machine
6. Establishing a reusable template

1: Archiving the case, not just the result

A case folder is self-contained: it holds the geometry, the mesh, the conditions, the settings and the fields. The File > Save CFD case as… menu puts it wherever you want, and Open a CFD case reopens it later with automatic detection of the case type and the phase status.

Heliciel also has a compact CFD project format, which packages the case and its metadata into a single file: this is the format used to send to a client or to file in an archive.

Two habits to take:

• Tick "Save logs" in the Parameters tab before meshing. The detailed outputs of each tool are deleted by the end-of-cycle cleanup after a successful run: if you want them, you have to request them beforehand and copy them elsewhere;
• Keep the link to the BEM project. The File > Open Heliciel project the current case originates from entry reopens the project the case came from. File this project alongside it: without it, the computed geometry can no longer be reconstructed.

2: Documenting mesh convergence

This is the most important piece of the file, and the most often missing.

A result obtained on a single mesh is not verified: nothing says it would not change at the next level. The demonstration comes down to a table: the same quantity, computed at three increasing preset levels, with the corresponding cell count.

This last deviation is the only uncertainty you can honestly state. It is better to write "thrust is established at such a preset level, and varies by so much between the last two levels" than an invented error bar.

Three captures of the results panel at mesh levels 1 (15,245 points), 3 (45,759) and 5 (85,085): thrust goes from 917 to 882 then 819 N, torque from 247 to 207 then 193 N·m. It is not stabilised at this stage: the value cannot be quoted yet, which is precisely what this check reveals

3: Stating the assumptions and limitations

An honest file states what the calculation does not say. The list is short, well known, and the same from one file to the next:

• Incompressible flow: valid, as a general rule, up to a Mach number of about 0.3, i.e. roughly 100 m/s in air. For a blade, it is the relative velocity at the tip that must be looked at, not the advance speed;
• Modelled turbulence, not resolved: no detail of the structures, and no prediction of laminar-turbulent transition. At low Reynolds number, drag is pessimistic;
• Qualitative results beyond stall: past the stall angle of the profile, the steady-state calculation indicates a trend, not a value;
• Numerical diffusion: first-order upwind schemes smooth out gradients and tend to overestimate drag. Specify the scheme used;
• In liquid: no free surface, uniform reference pressure, cavitation assessed but not simulated.

If you have carried out the approach in tutorial no. 18, add your reference deviation: this is the element that gives a file the most weight, because it shows that the chain has been qualified on a known case.

4: Producing the report

Two test bench tools feed the document directly:

• the toolbar's Capture button records the current view: compose the scene — pressure, cut planes, iso-surfaces, orientation — then capture;
• the File > Create PDF report entry assembles a document from the case.

One rule of composition: every image carries its scale and its conditions. A pressure map without its bounds, without the preset used and without the convergence status is an illustration, not a result. Two maps to be compared must share the same colour scale, otherwise the comparison is about the palettes.

Also specify, for each image, whether it is a Phase 1 steady state or a Phase 2 instant. In Phase 2, the quantities quoted are averages over the last revolution: say so.

First page of a CFD report (aircraft propeller, Phase 1 MRF): case header, results table and 3D view with streamlines — the conditions that make these figures defensible must be attached. This page contains no section cut (image to be cropped: it shows a development path)

5: Documenting the machine

Calculation time is a piece of data in the file, and it only makes sense together with the machine that produced it. The Hardware diagnostic button in the Guide tab inventories the configuration: this is the source to cite.

Also note the number of processes used, set in the Process frame. A useful reminder: the number of physical cores, not the number of logical cores.

This information serves two purposes: allowing a third party to estimate the cost of reproducing the run, and allowing you to size future campaigns. The propeller preset catalogue moreover grades its relative cost, from ×1 for the fastest level to ×117 for the reference level: this is a planning basis directly usable in a quote.

6: Establishing a reusable template

Do this work once, and keep its structure as your file template:

1. Purpose and scope: the question asked, in one sentence;
2. Geometry: origin of the model, attached Heliciel project;
3. Conditions: fluid, speed, rotation speed, immersion depth where applicable;
4. Mesh: preset, cell count, quality indicators;
5. Solver: preset, phases run, criterion and convergence status;
6. Mesh convergence: the three-level table;
7. Results: values, with the breakdown by boundary and by type of stress;
8. Comparison: comparison against the BEM, and against an external reference if one exists;
9. Assumptions and limitations: the list stated above;
10. Appendices: archived case, logs kept, hardware configuration.

This structure is the same whatever the machine studied. It is what makes a calculation reviewable, contestable and reproducible — and that is what marks out an engineering office's file.

End of the series. The twenty tutorials in this collection cover the complete chain, from the first simulation to the deliverable file. They can be read in order to get up to speed, or separately depending on the question of the day.

The CFD bench overview: the 3D view of the propeller on the left, the tabs and results panel on the right (too small to read the tab names)

The collection of twenty CFD tutorials

« Previous: From the hull to the propeller

1. First CFD propeller simulation
2. Reading the OpenFOAM case folder
3. Choosing the mesh preset
4. The MRF zone and Phase 1
5. Phase 2 with sliding mesh
6. Comparing BEM and CFD
7. Reading the convergence of a run
8. Turbulence models
9. Boundary layer and y+
10. Reading a pressure map
11. Where the forces come from
12. Checking an airfoil polar
13. Marine propeller and cavitation
14. Wind turbine and tidal turbine
15. Fan in a closed duct
16. Aircraft propeller, cruise and static thrust
17. The optimisation loop
18. Calibrating on a reference
19. From the hull to the propeller
20. The deliverable calculation file (you are here)

This series accompanies the Heliciel design tutorials, which cover the BEM side: blade design, rotation speed selection, performance curves. CFD comes after them, to verify and to see.