CardiaNova · Chiang Mai University

Artificial Heart
Mechanical Reliability

A continuous-flow pump, modelled end to end. Orbit it, cut it open, then change any number and watch the whole analysis move.

Panitan Khwankaew (Ton) · design-exploration model
Revolutions since you opened this
0
at 3,000 rpm
Reliability at 5 years
—
—
drag to orbit · scroll to zoom

The analysis

Seven views over the same model. Every input is live — change one and every result on every tab recomputes.

Design-exploration model — engineering only. No hardware has been tested and no clinical claim is made. Inputs are handbook properties, physiological conditions, or engineering estimates; subsystem life values are placeholders, not measurements. Results rank design options and size a test programme — they are not reliability predictions and cannot support a regulatory or clinical statement. Governing equations are cited on the Methods & references tab, where every input is tagged with its provenance.

Architecture

Operating point

System reliability over time
Selected architecture Other architectures Target R(5 yr) = 90%

Inputs

Pump speed and service life are on the Overview tab.

Load cycles accumulated in servicelog scale

Impeller geometry

S-N curve and the operating pointTi-6Al-4V · Basquin
Allowable alternating stress Applied (Goodman-equivalent) Beyond test data

Resonance check

Contact bearing

Wear state at the selected time in service
Predicted life vs wear coefficientboth axes log
What actually moves the answerone-at-a-time
Range of predicted life when each input is swept over its plausible range, everything else held at its current value.

Hermetic enclosure

Time to critical moisture vs leak rateimplanted at 37 °C

The other kind of seal

The percutaneous driveline is not a mechanical sealing problem. The skin barrier is permanently broken, making it a chronic infection pathway — a biological failure no amount of mechanical sealing fixes. Switch to Architecture B on the Overview tab to see it become the dominant failure mode.

Evidence note. This study asserts the mechanism and models its consequence. It quotes no infection rate, and the life value used for the driveline is a placeholder like every other subsystem. Source a rate from the registry and review literature before stating one to a clinical audience — see reference tier VERIFY on the Methods tab.

Subsystem characteristic life

Every value here is a placeholder awaiting test data. Tune them and watch which one actually controls the outcome.

Reliability of the current configuration
Which subsystem fails firstbefore 5 years
Required improvement to reach the target

Test conditions

Demonstration

Units required vs test durationzero-failure demonstration
Combined acceleration Real time Your test duration
Does the plan survive acceleration-factor uncertainty?20,000 trials

Read this before presenting any of it clinically

This tool computes mechanical engineering quantities — stress, wear, hermeticity, and a reliability roll-up over assumed component-life distributions. It makes no clinical claim of any kind: not safety, not efficacy, not haemocompatibility, not any outcome or event rate in patients.

The governing equations below are classical and cited precisely. The subsystem life distributions are placeholders chosen to demonstrate the method, not measurements — they are the reason the reliability percentages must never be quoted as predictions for this or any device. Replace them with test data, then the numbers mean something.

Citations marked VERIFY are pointers to where evidence lives. No number in this tool was taken from them. Read the primary source and cite it directly before repeating any clinical statement to a clinical audience.

Governing equations, as implemented

Provenance of every input

Each parameter is tagged by where its value comes from. Anything tagged ESTIMATE is an engineering judgement that must be replaced with measured data before it supports a decision.

What this study does not cover

  • Haemocompatibility — haemolysis, platelet activation, von Willebrand factor degradation, pump thrombosis. Shear-field and blood-damage modelling is a separate discipline and is not attempted here.
  • Biocompatibility and materials leaching — the ISO 10993 programme.
  • Electrical safety and EMC — IEC 60601 series.
  • Anatomical fit, cannulation, surgical technique and anything downstream of them.
  • Software and control-system safety, including failure-mode behaviour of the levitation controller — modelled here only as an opaque life distribution.
  • Human factors, alarms, and the external system the patient actually lives with.

References

Physics ported 1:1 from the Python study (params.py · analysis.py · run_all.py), which remains the reference implementation. Series-system reliability is solved in closed form rather than by Monte Carlo, so these figures are exact for the model.

CardiaNova · Chiang Mai University · Revision B · August 2026
Design-exploration model. Not validated device evidence. Not a reliability claim.