The corrosion tool that thinks like a metallurgist
and speaks in probabilities.

Microstructure-aware, probabilistic prediction of pitting, chloride-SCC and sour cracking — plus cost-aware CRA selection. Runs entirely in your browser.

composition → PRENσ-phase couplingCPT + credible interval Cl-SCC · sour SSCcost-aware selectionCCUS / H₂
PitCast · pitcast.austenite.org · screening estimate — not a substitute for qualified engineering

Material

Service conditions

Advanced — thermal / weld history (σ-phase)

e.g. 850 °C for 4 h precipitates σ-phase and sharply lowers CPT.

Enter conditions →

Service conditions

Recommends the cheapest grade whose overall risk clears the threshold — avoiding over-specification.

Set a service →

All real measured records from the cited open database (Nyby et al., Scientific Data 2021, CC BY 4.0). Search by alloy or UNS, filter by metric, and click Assess → to load any record's composition into the tool.

Probabilistic Material Selection Diagram — sweeps the validated engine across a service window and overlays the safe operating limit on the ISO 15156-3 boundary. The physics-vs-code diff (right) is the part no off-the-shelf calculator gives you.

Alloy

Diagram

Operating point

Pick an alloy and a map →
Sweet CO₂ corrosion of carbon steel — five canonical models on one operating point · de Waard 1975/1995 · NORSOK M-506:2017 · NESC · Multicorp · cited

Environment preset

Operating point

Corrosion allowance

Set an operating point →
Cathodic-protection criteria (AMPP SP0169) + AC-corrosion screening (ISO 18086) for buried carbon-steel pipe · holiday spread-resistance model · cited

AC interference (at a coating holiday)

CP criteria (mV vs Cu/CuSO₄)

Instant-off (IR-free) is the value the −850 mV criterion is judged on; the ON potential includes IR drop and can falsely "pass".

Sacrificial anode sizing (DNV-RP-B401 / NACE SP0387)

Galvanic couple (ASTM G82 / MIL-STD-889)

Groundbed (Dwight / Sunde — NACE SP0169 / IEEE 81)

Enter AC + CP data →
Pipeline integrity — ASME B31G / Modified B31G corroded-pipe fitness-for-service + general remaining-life. PE-grade screening for axial external metal-loss defects (ILI / inspection patches).

Pipe geometry

External metal-loss defect

Remaining-life (uniform CR)

CUI risk (API 583 / NACE SP0198)

MIC risk (NACE TM0194 / TM0212 / SP0775)

HIC / SOHIC risk (NACE MR0103 / TM0284 / ISO 15156-2)

HIC needs free water + pH₂S > 0.34 kPa (ISO 15156). SOHIC adds tensile-stress amplification — dominant above ~250 HV.

RBI screening (API RP 581 — 5×5 matrix)

Reuses pipe geometry (tnom, tmin) from above. Screening only — full API 581 uses detailed damage-factor tables.

Set pipe geometry →

How it works

1 · Composition → microstructure

PREN from real composition; ferrite balance (WRC-1992); σ-phase and sensitization from thermal/weld history.

2 · Microstructure → resistance

An effective-local-PREN coupling degrades the critical pitting temperature (CPT) — the link no standard tool carries.

3 · Probability, not a point

Every output is a distribution with a credible interval; P(pit) = P(CPT < service temperature).

4 · Decision

Chloride-SCC and sour SSC (ISO 15156), then a cost-aware recommendation of the cheapest grade that's safe enough.

About & honest limits

PitCast is a faithful in-browser port of a validated Python research package (124 tests; calibrated against cited literature — e.g. it reproduces an independent 2507 σ-phase CPT loss of 64 °C from a single 2205 anchor). The CPT correlation is leverage-aware and leave-one-out calibrated.

Validation & accuracy

90%
LOO interval coverage46 / 51
6.6 °C
LOO mean abs. errorRMSE 8.5 °C
51
cited G48 CPT recordsASTM G48, 6% FeCl₃
500 / 500
sanity / invariant testsall passing

The CPT model is a single-feature linear fit on PRENN30 (PREN is the exact definitional formula Cr + 3.3·Mo + 16·N). That form was chosen by leave-one-out cross-validation over 7 candidate models — it was the most accurate; richer multi-feature variants overfit. The 90% prediction interval covers ~90% of held-out records, so it is statistically well-calibrated. The 500 invariant tests check probability bounds, temperature & chloride monotonicity, that σ-phase ageing lowers CPT, and the ISO 15156 sour gate at pH₂S < 0.3 kPa.

A ~6.6 °C error sits close to the intrinsic lab-to-lab scatter of CPT measurement itself — this is honest calibration evidence for a screening tool, not a guarantee. Calibration set: n = 51 records from Nyby et al., Scientific Data 8, 58 (2021), CC BY 4.0.

Cited measured anchors

Independent literature data points the models are checked against — each row links to its primary source. CPT anchors span low-PREN 304 to 6Mo super-austenitic; CO₂ anchors span field flowlines, top-of-line condensation and HP/HT wells.

⚠ Screening / research-grade. Constants are cited or clearly flagged; some envelopes (welds, SCC, CCUS/H₂) are rule-of-thumb. Not a substitute for qualified materials engineering, standards compliance, or testing. Do not use unsupervised for design.

© austenite.org · PitCast research preview