How to Use the Steel Phase & Transformation Explorer

Before you begin

Purpose and limits

The explorer is a teaching and preliminary engineering-screening environment. It is organized around the causal sequence:

Student and professional workspaces

Guided start and professional workflow

Choose the goal first

The Guided Start page separates recommended starting points for Students and Steel professionals. The selected workspace now stays active throughout the tool. It changes navigation, guidance, and information priority; it never changes calculations, inputs, warnings, or the modules available to you.

• Student goals begin with phase diagrams, thermal paths, TTT/CCT concepts, chemistry, connected heat treatment, and practice.
• Steel professional goals begin with heat-chemistry screening, transformation response, austenitize–quench–temper review, through-section hardenability, thermal-data analysis, and metallography.

Each goal card states which workspace it opens. Use Browse all guided workflows to reach every original practical question, including:

• What phases are present?
• What happens during heating or cooling?
• Will the centre of the section harden?
• Is the austenitizing cycle adequate?
• Which quench medium should be compared?
• What does the uploaded thermal record show?
• How should a thermocouple or pyrometer result be interpreted?
• What surface-concentration or transformation-response trend, tempering response, microstructure, or alloy-family behaviour should be expected?

Workspace mode

Professional decision workflow

The professional workspace adds a structured review layer around the existing tool. Move through the stages in order, revisit any stage as evidence changes, or open any of the 12 original tabs directly.

1. Define material/process: establish the chemistry, material family, section, thermal history, and other case inputs in the existing modules.
2. Select engineering question: state the screening decision so the workspace can prioritize the most relevant modules and checks.
3. Review applicability: inspect each versioned model record, output classification, validity state, evidence basis, approval state, source references, assumptions, uncertainty, and required verification before interpreting an estimate.
4. Compare scenarios: capture read-only Baseline and Alternative snapshots, plus an optional third case, for side-by-side review. Capturing or viewing a comparison does not apply old values or change the active scenario.
5. Inspect sensitivity/uncertainty: run a bounded low–baseline–high screen for carbon or property cooling rate against estimated hardness or martensite, then record known uncertainty sources. This one-factor view is non-mutating and is not a confidence interval.
6. Add plant evidence: record an evidence type, date, traceable reference, observation or result, and your assessment as Supports, Conflicts, Inconclusive, or Not assessed. The tool does not validate that classification.
7. Calibrate and monitor: optionally import a separate governed calibration package, fit an additive-bias candidate on development heats or batches, compare generic, calibrated, and actual values on untouched validation groups, and screen later monitoring residuals. A candidate never changes the generic model or approves itself.
8. Generate report: create a separate engineering screening report containing scope, applicability, comparison, sensitivity, evidence classification, calibration audit summary, cautions, model/version traceability, and the same governed values and warnings shown on screen. Download the compact governance JSON or a checksummed replay JSON when a machine-readable record is needed.

Model governance and output classes

Every result used by the professional workflow is linked to a model ID and version. Applicability is reported as In range, Near boundary, Out of domain, or Not evaluated. Out-of-domain values are withheld rather than carried forward as an unqualified engineering result.

Explanation detail

Interpretation basis

Use Equilibrium for Fe–Fe₃C phase fields and lever-rule fractions. Use Rapid cooling for generalized ferrite, pearlite, bainite, and martensite trends. The rapid-cooling selection does not convert the equilibrium diagram into a CCT diagram; it changes how the scenario is interpreted.

Critical-temperature coach

• Ae is the equilibrium critical temperature.
• Ac is the observed transformation temperature during heating.
• Ar is the observed transformation temperature during cooling.
• Ms and Mf refer to martensite start and finish.

The heating- and cooling-rate shifts in the coach are teaching approximations, not process setpoints.

Step 5 capability · Professional workspace

Plant calibration, group-separated validation, and drift monitoring

This optional workflow keeps the original generic property model intact and builds a separate local correction candidate around controlled plant or test data:

1. Prepare and govern the source: use one quantitative target, one declared measurement method, a controlled source reference, and stable record and heat/batch identifiers.
2. Import: load a version 1 calibration-package JSON or CSV. Invalid imports are rejected without replacing the prior valid session.
3. Calibrate: within each calibration heat or batch, average actual − generic residuals; then give each group equal weight when fitting the single additive correction. Validation and monitoring results do not influence it, and the generic model is not refitted or overwritten.
4. Validate: compare generic, calibrated, and actual values on validation heats or batches that never appeared in calibration. Review group-equally-weighted bias, MAE, RMSE, and R²; a better fit on development groups is not validation.
5. Monitor: aggregate corrected residuals at the later-in-time heat/batch level and screen their exponentially weighted moving average (EWMA). Treat any signal as a prompt for investigation.
6. Record and review: retain the package fingerprint and audit summary, define the candidate’s local scope, and obtain qualified review outside the browser workflow. This version does not authorize applying the correction to new inputs.

Separate calibration-package contract

A calibration package is not a saved Explorer scenario, thermal-process upload, evidence note, or specification dataset. JSON may place the following fields at the top level or under metadata; if a field appears in both places, the values must agree. CSV uses the same row fields plus consistent metadata columns or # key: value comment lines.

Contract 1.0 does not collect Nb or Ti. Zero is supplied only to the existing governance-envelope check; it is not an inferred plant value, predictor input, or fitted variable. Do not use this candidate where omitted microalloying or another unmodelled variable makes the generic property screen inapplicable.

Minimum data separation is enforced: at least five calibration rows across three groups and at least three validation rows across two different groups. Replicating rows within one group does not give that group more influence: fitting and reported performance first aggregate by group, then weight groups equally. These are software minimums, not evidence that the sample is representative or statistically adequate.

Supported first-release quantitative targets

Actual inputs outside these sanity ranges are rejected. Corrected values are deliberately left unclamped so an invalid correction stays visible for diagnosis. The physical-output gate covers only the imported calibration, holdout, and monitoring comparisons: if any of those corrected values is outside its target range, evidence comparison is withheld. A range-safe imported dataset can make the evidence audit current; it never permits applying the correction to a new chemistry or cooling-rate input. This version has no candidate input envelope or prediction-time output gate, so new-input application remains withheld.

Group separation and leakage

All specimens or repeated tests from one heat or batch must stay in a single role. A row-level random split can place related specimens on both sides, making validation look better because the model has already seen that group. The importer rejects this heat/batch leakage. Freeze the split, cleaning rules, scope, and model choices before viewing holdout results; if holdout results influence any of those choices, reclassify those groups as development data and obtain fresh validation groups. Fingerprints make revisions detectable but cannot enforce that discipline. The browser result is an internal, group-separated holdout only; independent validation requires a separate controlled external campaign or artifact, and neither result by itself proves that future production is covered.

SPEC_DATA is identity context only

The optional specContext accepts only grade identity and traceability fields: schema version, specification body and grade key, display name, edition, PSL/category where applicable, material form, and limited verification metadata. The calibration engine does not read specification chemistry, mechanical limits, Charpy/DWTT requirements, footnote rules, overlays, or acceptance logic as observations, targets, labels, or approval evidence.

Candidate and approval gate

A successful candidate analysis is labelled Calibration candidate with an Engineering screening classification. The base model and version remain frozen and traceable. At this stage it records Plant calibrated: false, new-input application: withheld, pending qualified review, and acceptance withheld. Neither good training metrics, improved internal holdout metrics, a specification identity, nor a user acknowledgement can promote it.

Step 7: independent validation and Plant calibrated promotion

The separate JSON validation package pins the candidate-model fingerprint, training fingerprint, scope fingerprint, target, and heat/batch grouping basis. Its protocol ID, revision, independent organization, independence statement, lock timestamp, and numeric acceptance limits must be recorded before every validation result. At least five completely new heats or batches are required; any overlap with calibration, internal holdout, or monitoring groups fails closed. Five groups is a software floor, not a claim of statistical sufficiency.

The candidate is not refitted. External bias, MAE, RMSE, and optional R² are calculated across equal-weighted external group means and compared with the predeclared limits. Every external row must also remain within the candidate’s calibration-data chemistry/cooling-rate envelope and declared thickness scope, and every un-clamped prediction must pass the physical-output gate. A failed pin, chronology check, independence check, envelope check, output gate, or acceptance criterion blocks approval.

After all validation checks pass, a reviewer may record their name, role, organization, controlled authority reference, approval time, scope-limited statement, and explicit authority attestation. The browser records that attestation but cannot verify identity, credentials, organizational authority, laboratory chain of custody, or source-system integrity. Only then is the frozen candidate labelled Plant calibrated for its recorded envelope. Prediction-time use is fail-closed outside that envelope or if the active candidate changes. Specification acceptance, product release, and qualified-procedure status remain withheld.

Raw-data privacy and report record

Imported rows live only in the calibration engine’s current page session. They are not written to professional browser storage, saved scenarios, share links, the HTML report, or governance JSON. Clearing the calibration session or reloading the page removes them. The compact audit may be stored locally and exported; it contains:

• package ID/revision/title, source reference, measurement method, import/analyse timestamps, and package/model fingerprints;
• the declared local site/line/process/product/thickness/test/cooling scope and its scope fingerprint;
• target and unit, frozen generic model ID/version, equal-group additive correction, fitted row/group counts, and role/group totals;
• grouping basis, disjoint-group result, later-than-reference check, full monitoring chronology status, and optional traceability-only specification identity;
• group-equally-weighted generic-versus-calibrated calibration and validation bias, MAE, RMSE, and R²;
• monitoring row/group counts, timestamp-ordering rule and chronology status, EWMA setting and time-varying limits, trigger count, governance state, reasons, and warnings.

A fingerprint detects a changed package or scope; it is neither encryption, anonymization, nor proof against tampering. Browser memory and local storage are not encrypted records-management systems. Apply your organization’s data classification, authorization, retention, and device rules before loading controlled information.

Interpret drift correctly

Monitoring uses the mean calibrated residual (actual − calibrated) for each later heat or batch. A group’s latest row timestamp defines its position. An individual group-mean residual is screened against ±3s; with λ = 0.2, the EWMA calculates:

Here, s is the sample standard deviation of the equally weighted calibration-group residual means. The time-varying limit is ±0.6s at the first point and approaches ±s. The result is stable, review trigger, or insufficient. Unverified chronology, no monitoring groups, zero/unavailable reference variation, variation indistinguishable from floating-point precision, or fewer than 20 calibration groups produces an insufficient result. When numeric variation cannot support limits, inferential residual and EWMA triggers are withheld. When group timestamps tie, order-dependent EWMA values, limits, and triggers are withheld while order-independent residual diagnostics remain visible. Fixed group-mean limits also require one constant replicate count across the calibration baseline and the same count in every monitoring group. Unequal calibration groups, unequal monitoring groups, or a baseline-to-monitoring count mismatch therefore also force an insufficient result in this version.

The fixed 3σ screen assumes a stable Phase-I residual process, comparable independent subgroups, and no material serial correlation. Skew, heavy tails, autocorrelation, measurement drift, or changing product mix can invalidate nominal false-alarm behavior; a qualified practitioner must select a different chart or limit design when those assumptions are not defensible.

Step 6 capability · Professional workspace

Revisioned reports, checksums, and deterministic replay

The Report stage can now preserve a governed screening result as a named revision and prove whether the same tool build reproduces it. This does not approve the engineering conclusion; it makes the calculation record traceable.

1. Complete the record: enter a conclusion, interpretation, next action, action owner, prepared-by identity, optional reviewer, and target date.
2. Name the revision: use a meaningful label such as “Baseline review” or “Cooling-rate alternative.”
3. Save or download: a local revision is stored unencrypted in this browser; Download replay JSON creates a portable package. At most five local revisions are retained.
4. Verify before replay: import accepts only the supported schema, matching tool version, matching model-registry version, and an unchanged package checksum.
5. Replay: the tool restores the original scenario and decision controls, reconstructs embedded thermal chart data when present, reruns the shared calculation engine, and compares the scenario fingerprint, governed values/statuses, summary, and chart data with the saved signature.

Step 8 capability · Complete dual-audience coverage

Full module expansion and role-specific tools

Step 8 connects every original Explorer module to a complete Student or Professional purpose without changing the shared calculation engine. Switching workspace, opening a guided path, building an activity, or opening a review pack changes navigation and guidance only. It does not change chemistry, process inputs, model results, calibration state, qualification state, saved/shared scenarios, or exports.

Expanded guided learning

The Student workspace now provides twelve six-stage learning paths. Every path follows Predict → Manipulate → Observe → Explain → Check → Apply and saves progress only in the current browser:

• Phase Diagram Foundations; Heating, Cooling & Transformation; and Chemistry & Heat Treatment;
• Through-section hardenability; Austenitization control; and Quenching and tempering trade-offs;
• Thermal data and measurement; Mechanical testing and metallography; and Surface treatment and diffusion;
• Solidification, rolling, and alloy families; Reference-diagram literacy; and Model calibration and validation.

The expanded lessons teach evidence discipline as well as metallurgy. A learner must distinguish equilibrium from kinetics, surface response from centre response, a sensor trace from a confirmed transformation, a schematic image from representative microscopy, and a model estimate from plant or specification acceptance.

Activity brief builder

The Student workspace includes an instructor-ready activity builder covering all twelve original modules. Select the module, a 20-, 45-, or 90-minute duration, and the expected evidence type. The generated brief contains a learning objective, all six learning stages, an application prompt, and a fixed safety/governance guardrail. Downloaded briefs are planning aids; they contain no answer key, student identity, score, plant data, model approval, or product-release authorization.

Professional review packs

The Professional workspace includes five question-led review packs: heat chemistry screening, austenitize–quench–temper review, thermal-record investigation, microstructure/property investigation, and plant-model qualification review. Each pack opens the relevant existing modules in sequence and provides a completion checklist. The downloaded JSON is an unapproved review checklist, not a governed report, replay record, electronic signature, calibration package, validation package, or specification decision.

Step 10 capability · Controlled specification traceability

Move from metallurgy screening to governed specification work

The Explorer can load a session-only SPEC_DATA version 1 JSON package and select a specification body, grade, and applicable product form. It validates the identity contract before exposing a selection. Imports are limited to 2 MiB and are never saved in the Explorer.

1. Explore: use the metallurgy modules to examine mechanisms, assumptions, and directional process–structure–property response.
2. Identify: load a controlled schema package and select the intended grade identity and product form.
3. Export identity context: download the traceability-only context for a controlled calibration package or engineering record.
4. Verify and decide elsewhere: use the Material Specification Lookup and Material Specification Compliance Checker with the governing licensed standard, actual test evidence, purchaser requirements, and responsible approval.

The exported identity contains only schema version, specification body, grade key and display name, edition, PSL/category, applicable form, and limited verification metadata. Chemistry and mechanical limits, Charpy/DWTT requirements, testing rules, overlays, footnotes, equivalence claims, and acceptance logic are deliberately excluded.

Core module

Equilibrium diagram

Plot and manage points

1. Select metric or imperial temperature display.
2. Enter carbon and temperature for P1.
3. Use + Add point to create P2, P3, and additional comparisons.
4. Select a row, marker, or result card to make that point active.
5. Click or drag on the diagram to move only the active point.

Display controls

Lever rule

Inside a two-phase field, the tool estimates equilibrium mass fractions from the tie-line endpoints. For ferrite plus austenite:

The schematic microstructure is educational. It does not reproduce actual grain morphology, inclusions, banding, or transformation kinetics.

Thermal-cycle builder

Heating and cooling paths

Select a preset—normalizing, annealing, spheroidizing, quench and temper, intercritical annealing, austempering, martempering—or build a custom cycle.

1. Enter carbon.
2. Add or remove cycle steps.
3. Set target temperature, duration, and step type.
4. Use Play, Pause, or the time slider.
5. Review boundary crossings and the instantaneous equilibrium-reference panel.

Transformation kinetics

TTT and CCT

Chemistry from the Chemistry module shifts the generalized curves and critical temperatures. Final-temperature martensite formation below M_s uses the Koistinen–Marburger relationship with α = 0.011 °C⁻¹, while untransformed material is shown as retained austenite. These are not certified, grade-specific diagrams.

Chemistry and causality

Chemistry, properties, and causal chain

Select an illustrative preset or enter verified heat chemistry. The module estimates CE IIW, Pcm, A_c1, A_c3, M_s, hardenability, weldability, and directional property trends.

The Guided Start causal dashboard updates the five-stage chain and warns when the active phase-diagram carbon differs from the chemistry-model carbon.

Release 2

Jominy and section-size hardenability

This module separates maximum hardness from depth of hardening.

1. Confirm chemistry.
2. Select round or plate geometry.
3. Enter diameter or thickness.
4. Select quench medium and agitation.
5. Enter prior-austenite ASTM grain number and target hardness.
6. Read the estimated Jominy curve and through-section result.

Outputs include equivalent Jominy distance, surface/quarter-depth/centre martensite, hardness, and effective target-hardness depth.

Release 2

Austenitization and grain-growth window

The module balances four competing needs: heat the core, form austenite, dissolve and homogenize carbides, and avoid excessive prior-austenite grain growth.

• Set temperature, hold time, section size, heating rate, and initial grain size.
• Select starting microstructure, carbide burden, and grain-boundary pinning.
• Review core heat-through time, effective soak, dissolution, homogenization, final ASTM grain number, and grain-growth risk.
• Use the heatmap to identify incomplete, balanced, and grain-growth-dominated regions.

Release 2

Quench-medium and cracking-risk comparison

Compare still air, forced air/gas, conventional oil, accelerated oil, polymer solution, water, brine, and a molten-salt-bath analogue under common assumptions.

• Set bath temperature, agitation, geometry, size, transfer delay, corner severity, and final temperature.
• Review surface, quarter-depth, and centre cooling curves.
• Compare 800→500 °C cooling rate, martensite, hardness, maximum thermal gradient, martensite mismatch, cracking risk, and distortion risk.
• Use Apply to hardenability to transfer the selected setup.

Release 3

Process-data analysis

Upload a thermal record

1. Open Process Data.
2. Drop or select a CSV, TSV, TXT, or semicolon-delimited file.
3. Map the time and temperature columns.
4. Select source time and temperature units.
5. Adjust smoothing and arrest sensitivity.

The tool calculates the smoothed temperature history and:

Read the results

• Crossing-based average cooling rates for 800–500 °C and 500–300 °C, calculated as ΔT/Δt.
• Candidate thermal or transformation arrests.
• Candidate proximity to the A₁ equilibrium reference and estimated M_s; cooling A_r temperatures are not predicted.
• Measured-versus-simulated cycle overlay on the physical elapsed-time basis.
• Root-mean-square error over overlapping timestamps without duration normalization.

Release 3

Temperature-measurement assistant

Thermocouple mode

Select thermocouple type, wire/sheath diameter, attachment method, and shielding or thermal mass. Review the uncalibrated relative-lag tendency and attachment-bias risk. The tool does not calculate response time in seconds; that requires sensor construction, junction and sheath details, surrounding medium and velocity, mounting, and calibration or step-test data.

Infrared-pyrometer mode

Enter indicated temperature, instrument emissivity, estimated actual emissivity, reflected temperature, target width, spot diameter, and surface condition. Review the radiation-balance correction, target-to-spot ratio, and background contamination risk.

Release 4

Surface treatment and diffusion

Select carburizing, carbonitriding, nitriding, ferritic nitrocarburizing, induction hardening, flame hardening, or laser hardening.

Diffusion-controlled treatments use an Arrhenius diffusivity and complementary-error-function profile:

For diffusion modes, review the solute-concentration profile, concentration-threshold depth, and diffusion-distance indicator; these are not a universal hardness conversion. For induction, flame, and laser modes, review only the unitless exposure and penetration-response indices. The tool does not calculate physical hardened depth without validated energy or frequency, coupling or absorptivity, travel speed, geometry, thermal properties, and quench-boundary inputs.

Release 4

Tempering simulator

Enter steel response family, as-quenched hardness, tempering temperature, time, number of cycles, retained austenite, and post-temper cooling mode.

Review tempered hardness, toughness recovery, residual-stress relief, retained-austenite reduction, secondary-hardening tendency, and embrittlement risk together. Do not assume hardness changes monotonically for every alloy family.

Release 4

Mechanical-property laboratory

Select microstructure, carbon, ASTM grain number, test temperature, notch severity, and residual stress. Switch among:

• Engineering stress–strain curve
• Charpy transition and DBTT curve
• Fatigue S–N curve

The module provides directional yield strength, tensile strength, elongation, hardness, Charpy energy, fatigue strength, and elastic modulus. It does not invent universal HBW or HRC conversions; use the applicable ASTM E140 or ISO 18265 table for the material and range.

Release 4

Virtual metallography laboratory

Generate schematic ferrite, ferrite–pearlite, bainite, martensite, spheroidite, and decarburization fields. Adjust grain size, magnification, etchant, banding, inclusion population, and section orientation.

Use the mystery-field exercise to practise phase and microconstituent identification. The displayed fields are original teaching schematics—not reference photomicrographs.

Release 4

Solidification and rolling inheritance

Use carbon, segregation tendency, thermal gradient, equiaxed nucleation, reduction ratio, soft reduction, cleanliness, and process-stage controls to animate:

Review segregation, porosity, closure, columnar-grain tendency, banding, and anisotropy indicators. Rolling can close voids and change morphology without eliminating inherited chemical segregation.

Release 4

Alloy-family modes

Release 5

Reference diagrams

This tab holds two separate reference charts. Choose between them with the selector at the top of the panel. They share a carbon axis but not a coordinate system, so points do not carry across when you switch.

Which chart is which

Plotting and moving points

1. Use + Add point to create P2, P3, and as many further points as you need. Existing points stay visible.
2. Select a point by clicking its P button in the list, or by clicking its marker.
3. Drag any marker directly, or click anywhere inside the plot to move the active point there.
4. Type exact values into the carbon and temperature fields for precise placement.
5. Remove deletes the active point. At least one point is always kept.
6. Reset examples restores the seeded points for the current chart.

Reading a region

Move the pointer over the diagram. The crosshair reports carbon content, temperature in both Celsius and Fahrenheit at once, and the interpreted region. The What does this region mean? panel updates as you move, covering what is present, how it forms, typical morphology, and the property tendencies that follow.

On the rapid map, the painted colour and reported product come from the same teaching classifier; the unsupported interval below 0.20 wt% C is marked unavailable rather than extrapolated. On the poster, the interaction layer reports the metastable Fe–Fe₃C interpretation even where the underlying artwork also displays a stable graphite field.

Experience levels

Critical lines and legend

Critical lines draws A₁, approximate A₃/A_cm, M_s, and approximate M_f on the rapid-quench map, and the eutectoid, eutectic, peritectic, liquidus and solidus on the poster. Overlay labels toggles the Explorer-added field and critical-line annotations; labels printed in the poster artwork remain visible. Selecting a legend entry loads that region into the explanation panel. Rapid-map regions are directly keyboard reachable; poster regions are available through the keyboard-accessible HTML legend buttons.

Units

The °C and °F buttons change the display only. Internal values stay in Celsius, so a point does not move when you switch units. The poster's own printed scales remain in their original dual-unit form, since the artwork is reproduced unaltered.

Zoom, pan and full screen

Use the zoom buttons, or hold Ctrl and scroll. While zoomed, pan with the diagram container’s scrollbars or touch scrolling. Dragging inside the plotted field moves the active point; it does not pan. Fit returns to the whole chart and Full screen gives the diagram the entire viewport.

PNG export

Export PNG writes the diagram as currently displayed, including plotted points, visible overlays, critical-line emphasis and labels. Raster elements are embedded at export time, so the file is complete and self-contained. Attribution and the copyright line are included in the exported image and must not be removed.

On a phone

Controls stack into a single column and touch targets are enlarged. The poster is dense by design; use zoom and pan to inspect it rather than expecting the whole sheet to be legible at once. Poster artwork is only fetched when you open that chart, so the rapid-quench map stays fast on a mobile connection.

Engineering limitations

• A₁ is a reference invariant and M_s is a carbon-only Andrews estimate. The A₃ and A_cm curves are calibrated to standard anchor points rather than presented as exact equations. Alloying additions that move these boundaries are not represented on this map.
• M_f is an illustrative fixed offset below M_s, not a quantitative finish prediction. The real transformation is asymptotic and retained austenite can persist below the line, increasingly so at higher carbon.
• The pearlite, bainite and austenitising divisions are educational approximations conveying field topology only. They carry no implied precision.
• In this simplified drawing, the lower-bainite field closes near 0.42 wt% C where its illustrative divider meets M_s. This is not a physical prohibition: bainite can form from remaining austenite during below-M_s holding, so use a grade-specific TTT diagram.
• Real boundaries depend on chemistry, grain size, section size, homogeneity and holding time. Validate against grade-specific data before any process decision.

Source and copyright

The iron-carbon chart is the Iron-Carbon/Cementite Phase Diagram published by Buehler, an ITW company. Copyright © 2006 ASM International. All rights reserved. It is reproduced here under permission held by UpSkill Sprint Consulting and is served from this site rather than a third-party server. The artwork is reproduced unaltered; a link to the official PDF is provided beneath the diagram. Attribution and the copyright line must be preserved in any export or reuse.

The rapid-quench map is an original teaching reconstruction. A₁ and the carbon-only Andrews M_s estimate use published reference values; A₃, A_cm, M_f, and the product divisions are anchor-calibrated or illustrative and are not traced grade-specific transformation boundaries.

Learning and records

Learn, save, share, and export

Learning mode

Follow four six-step paths, generate phase-field challenges, use hints, inspect boundary explanations, and practise microstructure identification. Model calibration and validation is a synthetic evidence-literacy lesson: it uses the existing chemistry/property screen to teach Generic → Calibrate → group-separated Validate → Monitor, but it never imports plant data, fits a real candidate, or changes Professional calibration state.

Save and share

• Save scenario stores settings in the current browser.
• Load saved restores the local scenario.
• Copy share link embeds supported configuration data in the URL.

Uploaded raw process data remains local to the current browser session and is not embedded in the share link.

Professional notes, plant evidence, comparison snapshots, and explicitly saved audit revisions are kept only in this browser and are not added to ordinary saved scenarios or share links. Comparison and sensitivity views do not modify the active scenario.

Calibration imports are even more restricted: raw rows stay only in page-session memory. The Professional record may keep a compact audit summary, but it does not persist or export those rows.

Export

• Point and phase-analysis CSV
• Equilibrium-diagram PNG
• Derived process-data CSV
• Downloadable process-analysis HTML report
• Separate professional engineering-screening report
• Compact governance JSON
• Checksummed deterministic-replay JSON
• Browser print/PDF output

The professional report documents a screening review and its evidence status; it does not certify a grade, validate a model, qualify a process, or approve material. All pre-existing exports remain available and keep their original purpose.

Practical applications

Recommended workflows

Accuracy controls

Engineering cautions

• Equilibrium versus kinetics: equilibrium phase fields do not predict transformation time or cooling-rate response.
• Phase versus microconstituent: ferrite and cementite are phases; pearlite and bainite are microconstituents.
• Model validity: empirical equations have chemistry and process ranges.
• Section effects: centre and surface experience different thermal histories.
• Measurement: sensor lag, attachment, emissivity, reflected radiation, and calibration can dominate the result.
• Morphology: identical phase fractions can produce different properties when grain size, shape, distribution, and banding differ.
• Alloy families: plain-carbon models are incomplete for stainless, tool steel, and cast iron.
• Calibration leakage: keep every result from one heat or batch in one role; reused groups do not provide valid internal holdout evidence, and independent validation requires a separate controlled external campaign or artifact.
• Compatible test basis: calibration rows must share a controlled target, unit, specimen/test method, sampling basis, and cooling-rate definition.
• Calibration versus compliance: grade identity is traceability context only; use the dedicated specification tools for requirements and acceptance checks.
• Drift: a residual or EWMA signal identifies a change worth investigating, not its cause or a product verdict.
• Safety: the tool does not provide safe operating instructions for hazardous heat-treatment, atmosphere, quench, flame, induction, laser, or molten-salt processes.

Common issues

Troubleshooting

Before relying on an output

Final checklist

• Confirm chemistry and material family.
• Confirm units and sensor calibration.
• Separate equilibrium phases from kinetic products.
• Document prior microstructure and grain condition.
• Record section size, geometry, and sampling location.
• Document heating, holding, transfer, cooling, and tempering history.
• Check model and equation applicability.
• Preserve raw process data and analysis settings.
• Keep calibration and validation heat/batch groups disjoint.
• Confirm a compatible measurement and cooling-rate basis.
• Treat calibration as local to its documented scope.
• Investigate drift signals before assigning cause.
• Use the specification lookup/checker for standards work.
• Keep calibration-candidate approval and acceptance withheld until qualified review.
• Verify predictions using appropriate testing.
• Include uncertainty and cautions in reports.
• Use approved safety procedures.
• Obtain technical approval where required.