Hydrogen and ammonia
Pipelines, storage, crackers and electrolysers
Surface Physics Analysis
How will this joint, coating or barrier behave under its duty cycle?
Give Surphysis the layer stack, the duty cycle and your own characterisation files. It returns the fracture margin and the breakthrough time, each traced to the source it came from, with the modelled probability of meeting the criterion under those inputs. Where the evidence will not carry a verdict it says so, and names the evidence that would settle it.
Free for academic labs. No AI at runtime: every number comes from a numbered physics rule you can audit.
Inconel 718 built on sapphire, graded interlayer
R051 200 K excursion, 5 laser pulses
- Inconel 718laser built
- Cr barrier150 nm
- Graded interlayer5 steps, 3.2 µm
- MoO2 anchorcritical interface50 nm
- Sapphirec-plane substrate
Cr barrier on X65 pipeline steel
R137 MPa H2, 50 °C, 20 a
- Cr barriercritical interface150 nm
- X65 pipeline steelbulk
7YSZ thermal barrier on CMSX-4
mock1100 °C, 1 500 cycles
- 7YSZ top coat180 µm, columnar
- Thermally grown oxidecritical interface4.2 µm
- NiCoCrAlY bond coat120 µm
- CMSX-4 substratesingle crystal
Die attach on a copper substrate
mock-40 to 150 °C, 1 000 cycles
- Silicon die0.3 mm
- Attach layercritical interface40 µm
- Copper substratebulk
Industries
Six sectors. One question about the interface.
A case begins from the environment the part actually sees, not a blank form. You set the duty cycle from your own service record on the first screen, and the engine works to it.
Oil and gas
Sour service, downhole tools, LNG cryogenic cycling
Power
Gas turbine hot sections, nuclear instrumentation
Aerospace
Engine coatings, sensor windows, brazed assemblies
Additive manufacturing
Multi-material builds, metal on ceramic, repair
Electronics
Die attach, packaging, thin-film sensors
Three service doors and an evidence record
Design, analyse and decide with the evidence in view
8 scoped workflows are available. Availability does not establish physical validation. Other entries identify demonstrations or pending capability.
Design
Explore materials and build your design.
Compare material systems and develop layer stacks for your operating conditions. Examine thermal mismatch, interface compatibility and processing constraints.
- Material PassportView material properties, their sources, applicable conditions and evidence gaps.live
- Thermal Mismatch CheckCalculate thermal stress in two bonded materials during heating or cooling, with bending neglected, prevented or free to occur.live
- Stack DesignerDefine three or more bonded layers, or two with a blended layer, and calculate their thermal stress and, where supported, curvature.live
- Graded Joint DesignerDesign a graded material transition between a selected ceramic and metal under stated constraints.limited / pending
- Material Pair ScreeningScreen ceramic–metal pairs and candidate anchor metals against defined design criteria.limited / pending
- Process WindowAssess candidate processing conditions against defined thermal and material limits.limited / pending
Analyse
Understand how your design behaves.
Connect characterisation to calculation using XRD, EBSD, SEM and other measurement data. Investigate stress, microstructure, phase composition and diffusion, with assumptions and limitations visible.
- Measurement readerRaw file preview and observed XRD peak fitting available; calibrated property analysis pendinglimited / pending
- Barrier lifeConditional fixed-temperature diffusion timescale from your inputslive
- Fracture verdictDelamination and channelling margin by mode mixitypending review
- Chemistry checkDoes the oxide or barrier form and stay stablepending review
- Failure investigationRanked hypotheses and the discriminating measurementpending review
- Repair-cycle effectHow each repair moves the margin and the breakthrough timepending review
Decide
Evidence to support your decision.
Turn results into evidence for your next engineering decision. Compare alternatives against requirements, identify critical data gaps and produce traceable technical reports.
- Datasheet comparatorManual property comparisons with units, limits and per-property applicabilitylive
- Uncertainty and measure-nextFixed-example sensitivity and measurement-priority heuristic; customer workflow pendinglimited / pending
- Verdict reportFixed example report; manual scenario exports available in the calculation pageslimited / pending
- Specification sheetDraft layer schedule, user acceptance requirements and traceable inputs for technical reviewlive
- Qualification reportDraft evidence dossier linking requirements, source references, gaps and proposed testslive
- Standards taggingAPI 571, ISO 15156, ASME IX referencespending review
- FMEA exportPlanned failure-mode worksheet; scoring method and customer workflow not implementedpending review
Evidence
Inspect the recorded evidence separately from design decisions.
Public scorecard: Ledger status and worked examples; no accuracy established
The engine
Physics you can audit. No AI at runtime.
Twenty-five numbered rules cover thermo-elastic mismatch, minimax graded interlayers, grain-boundary diffusion, interface fracture mechanics and global sensitivity. Each rule cites its equation and its sources; each run prints the numbers it used.
- Continuum mechanics and fractureBiaxial mismatch stress, minimax step rule, Hutchinson and Suo energy release rates, mode-mixity toughness
- Microstructure and kineticsScherrer and Williamson-Hall grain size, Hart effective diffusivity, Harrison regimes, barrier breakthrough
- Uncertainty that ranks the next measurementMonte Carlo over stated error bars, Saltelli Sobol indices, measure-next by variance reduction per cost
R00 · preferred value selection
Stages left to right; every line is a dependency declared in the rule registry.
426property rowseach with its source, method, form and temperature window, from 32 references a reader can follow. Across every table: 71 A, 376 B, 64 C.
29rulesR00 to R24, every one unit-tested against a worked case
7duty cyclesreal service conditions, so a case starts from the environment the part sees
How it works
Four steps, one audit trail
What a verdict reads like
One page. The weakest interface, its breakthrough time, and what would move the answer.
No score out of ten. A named interface, a named mechanism, a number with its band, and the assumption that most controls it. Written for the engineer who has to sign.
Verdict · case C-0412engine 0.1 · mock data
Inconel 718 built on sapphire, graded interlayer, LPBF
Indeterminate, case being rebuilt
| Weakest interface withheld, case being rebuilt | withheld, case being rebuilt pulses |
| Probability of meeting the barrier criterion over 5 pulses Modelled probability, Monte Carlo over the input distributions, 4 000 draws. Conditional on the model and on those distributions, and not calibrated against outcomes. | P = withheld, case being rebuilt |
| Interlayer step stress Minimax grading, 5 steps | withheld, case being rebuilt · indeterminate |
| Fracture margin Edge delamination, Al2O3 to step 1 | G / Gc withheld, case being rebuilt · indeterminate |
What would settle itNot a measurement. This case is indeterminate because the case has an input chosen to make its output match a target, which cancels whatever it was tuned against and which no test can find, and no measurement reaches that. It is resolved by rebuilding the case on inputs that each trace to a published page a reader can obtain.
Why Surphysis
The tools you own stop at properties or at stress. The verdict lives in between.
Surphysis runs on top of the simulation and database tools you already have. It replaces the six-week consultancy report, not your stack.
| Capability | FE packages | CALPHAD tools | Materials databases | Consultancy report | Surphysis |
|---|---|---|---|---|---|
| The interface is the unit of analysisIts own passport, toughness data and fracture margin | |||||
| Your measurement files move the answerXRD, EBSD, SEM, permeation tighten the priors | |||||
| Barrier life from grain-boundary kineticsBreakthrough time, not thickness alone | |||||
| Ranked measure-next listWhich instrument, which sample, how much it tightens the verdict | |||||
| Public scorecard of blind predictionsHit and miss record against published failures | |||||
| Answer in minutes, in the browserData stays on your machine until you export |
built in possible with expert work not offered
Public evidence record
The scorecard separates logged records, numerical checks and outcome-known examples. An empty ledger establishes no prediction accuracy.
0 ledger entries
No physical validation claimed.
Your data stays yours
Uploads are parsed in your browser. Nothing leaves your machine until you export a report. No model is trained on your files, because there is no model.
Runs in the browser. No account needed to try a case.
Built from the literature
Every property carries its source, method, form and temperature window. Disputed values stay disputed and widen the band rather than pick a side.
Every value in a verdict traces to a citation you can open.
Engine, liveMeasure, model, decide. Every step keeps its number.
Request a demo session
Bring one interface you worry about. We run it live.
Forty minutes on a call. You describe the stack and the duty cycle, we run the case in front of you and hand you the verdict report afterwards. Academic labs get an account on the same call.
- BeforeSend the layer stack and the duty cycle, or bring them to the call. Characterisation files are welcome and not required.
- On the callWe build the case in front of you and run it. You see every rule fire, with the numbers it used and the assumptions it made.
- AfterwardsThe verdict report, the assumptions table and the ranked measure-next list, by email the same day. Yours to circulate.