SECTOR / BM · BIOMEDICAL DEVICES
Medical device CFD, built to survive review.
Flow, thermal and structural analysis for medical devices, run so that the credibility of the model is established for its context of use before anyone asks. We fix the acceptance thresholds first, close the analysis against them, and hand you a check matrix your reviewer can audit line by line.
The reference study
The FDA benchmark nozzle, in both directions, at five Reynolds numbers.
The CDRH benchmark nozzle compared against the inter-laboratory PIV measurement, framed as an ASME V&V 40 credibility assessment. This is the study we run to show a device buyer exactly what our evidence package looks like, on the one geometry their reviewer already knows.
five laboratory datasets per Reynolds number, both flow directions, published by the FDA
THE FULL STUDY →Who this is for
A CM&S engineer, a Regulatory Affairs lead preparing a submission that contains a simulation section, or a consultancy principal who has been asked whether the model will survive review.
What do I hand the reviewer when they ask how the credibility of this model was established?
What we take on
- Hemodynamics: wall shear stress, oscillatory shear index, residence time and low-shear area fraction
- Thrombogenicity and stasis screening on blood-contacting geometry
- Cannula, catheter and connector flow, pressure drop and shear exposure
- Oxygenator, dialyser and blood pump flow paths
- Inhaled drug delivery: nebuliser and dry-powder particle deposition
- Sterilisation and thermal dose modelling
- Structural and fatigue analysis on implants and delivery systems
- Verification and validation evidence packages written to ASME V&V 40 form
What you hold at the end
- Context of use and question of interest in ASME V&V 40 form, with the model-risk placement and its reasoning shown
- A three-band check matrix: verification, requirements, validation, every row carrying its threshold and the published basis that threshold came from
- A comparator dossier naming the measured data, its publisher, licence position and stated experimental uncertainty
- A case archive that re-runs from a clean checkout, with a pinned container digest and a SHA-256 manifest
- The rows that failed, published at the same weight as the rows that passed
Run to
Naming the procedure is not the same as claiming approval under it. We run the work to these procedures and cite them on every threshold; certification and approval are the authority's to give, not ours to claim.
How to engage
Four rungs, stepping by depth of evidence. Most first engagements start at S0, which is priced so a manager can commit without procurement, and the whole fee credits into whatever comes next.
| Rung | Best for | Fee | Duration | |
|---|---|---|---|---|
| S0 | Threshold Study | A first, small, low-risk piece of work with a supplier you have not used | $3,800 - $6,500 | 2 weeks |
| S1 | Verification Study | Is our numerical answer sound on its own terms? | $16,000 - $28,000 | 3 to 5 weeks |
| S2 | Validation Study | Does it agree with a measurement we did not produce? | $38,000 - $68,000 | 6 to 10 weeks |
| S3 | Design Closure | Now use it to pick the design, and show what lost | $75,000 - $145,000 | 10 to 18 weeks |
Threshold Study
We write the check matrix for your case before anyone runs it. You get every threshold with the procedure number or published statistic it was set from, a named comparator with its provenance and a written answer on whether it is actually obtainable, the tooling and versions we would pin, and the wall clock and cost of the study that would close it. If no comparator exists for your context of use, we say so in writing and the study stops there. That answer is worth the fee on its own.
- Three-band check matrix, every row carrying quantity, threshold and published basis
- Comparator dossier: source, publisher, licence position, stated experimental uncertainty, and a go or no-go on obtainability
- Tooling and version pin list, hardware spec, and machine-hour estimate from a measured timing rather than a guess
- The checks we expect to be hardest, named in advance, with the mechanism
- A fixed-fee SOW for the rung above, valid 60 days
DONE MEANS: YOU HOLD A CHECK MATRIX YOUR REVIEWER CAN RED-LINE BEFORE A SINGLE CORE-HOUR IS SPENT. THE FULL FEE CREDITS INTO ANY RUNG ABOVE IT IF COMMISSIONED WITHIN 60 DAYS.
Verification Study
Your geometry, your conditions, closed against the verification and requirements bands. Mesh quality audited and published, near-wall resolution reported as an area fraction rather than a maximum, iterative convergence demonstrated by solution stationarity rather than a residual floor, a three-grid convergence study with the observed order of accuracy and the discretisation uncertainty stated as a number, and domain and boundary-condition sensitivity quantified.
- Case archive that re-runs from a clean checkout with one command
- Container definition with a published image digest
- Verification report with the full convergence tables, histories and mesh quality
- Verification and requirements bands filled with measured values against declared thresholds
- Per-run wall clock and core hours, and a SHA-256 manifest
DONE MEANS: EVERY VERIFICATION AND REQUIREMENTS ROW CARRIES A MEASURED VALUE AGAINST ITS DECLARED THRESHOLD, NO ROW REMOVED AND NO THRESHOLD EDITED AFTER THE FIRST RESULT WAS READ.
Validation Study
Everything in the verification study, plus the comparison against measured data with the delta stated as a number against the band it was judged on. Where the comparator carries its own experimental uncertainty, that uncertainty sets the band rather than a tolerance we picked.
- All verification deliverables
- The validation band filled, every row with its delta against the declared threshold
- Comparator provenance, retrieval date and licence position
- The rows that failed, published at the same weight as the rows that passed
- A report written in the form your reviewer or surveyor expects
DONE MEANS: AGREEMENT, OR DISAGREEMENT, IS STATED AS A NUMBER AGAINST A THRESHOLD THAT WAS FIXED FIRST, AND THE ARCHIVE REPRODUCES IT.
Design Closure
The model becomes a design tool. Candidates screened against the declared objective, the winner closed, the rejected candidates published with the number that disqualified each one, and release geometry generated from the closed solution.
- All validation deliverables
- The design space explored, with the screening criteria declared first
- Rejected candidates published with their disqualifying numbers
- Release geometry generated from the closed solution
- A design report that traces every figure to an artefact
DONE MEANS: YOU HOLD A CHOSEN DESIGN, THE EVIDENCE IT WAS CHOSEN ON, AND THE RECORD OF WHAT IT BEAT.