Expert Tribology Consultancy
Loading...
Loading...
The diagnostic tools on this site are implementations of published methods, and an implementation can be corrected. When a correction changes the result a tool gives for the same input, the tool says so with a version. This page lists the current version of each method implementation and when it last changed.
A new version does not recalculate anything you saved earlier. If a saved result matters, re-run it.
Versioning starts on 30-08-2026. A result produced before that date carries no version at all, and there is no way to tell from the result itself which implementation produced it — treat such a result as unknown rather than as version 1.0.
Last changed: Borderline flag beside `confidence` (#944). Every triangle `result` block on /api/tools/duval-triangle now carries a `borderline` object: `state` (stays, stays-partial, borderline or not-assessed), `zonesReached`, `gasesHeld`, `convention` and, when no check was made, `notAssessedReason`. It says whether the figure’s own gases, varied by the analytical uncertainty IEC 60599:2022 §6.2 (printed p. 17) states for DGA values, with S per gas from the service-test column of IEC 60567:2023 Table 5 (printed p. 49), give any result other than the reported zone: another zone, or no zone at all (`UNKNOWN`, for example Triangle 2 near its gas floor). Gases below 5 × S are not varied, and the check is then a lower bound. The construction is TriboTech’s reading of the clause, and it is the same check the triangle tool page shows. The `registry` block does not carry it. `confidence`, in the result block and in the `registry` block, is deprecated: it is a software heuristic, not a probability, and no standard defines it. It stays in the response, and its values are unchanged. Triangle 2 no longer emits the warning "The point lies close to a zone boundary; measurement uncertainty may move it into the adjacent zone." It fired when the point lay nearer to a zone edge than the distance the confidence heuristic uses, a distance no source states; `borderline` takes its place. A result saved under 1.30 differs for the same input only in the new `borderline` object and, on Triangle 2, in that warning. No zone, polygon, vertex, boundary, zone code, severity or recommendation changed, and the SVG figures are unchanged.
Last changed: Borderline flag beside `confidence` (#944). Every pentagon `result` block on /api/tools/duval-pentagon (Pentagons 1 and 2, their improved forms, the Unified Pentagon and Pentagons 3, 4 and 4b) and the `result` block of the classic /api/tools/duval-pentagon-2 route now carry a `borderline` object: `state` (stays, stays-partial, borderline or not-assessed), `zonesReached`, `gasesHeld`, `convention` and, when no check was made, `notAssessedReason`. It says whether the five gases, varied by the analytical uncertainty IEC 60599:2022 §6.2 (printed p. 17) states for DGA values, with S per gas from the service-test column of IEC 60567:2023 Table 5 (printed p. 49), give any result other than the reported zone: another zone, or no zone at all (`UNKNOWN`). Gases below 5 × S are not varied, and the check is then a lower bound. The construction is TriboTech’s reading of the clause, and it is the same check the pentagon tool pages show. The `registry` block does not carry it. `confidence`, in the result block and in the `registry` block, is deprecated: it is a software heuristic, not a probability, and no standard defines it. It stays in the response, and its values are unchanged. The pentagons no longer emit the warning "Result is near region boundary, consider additional testing". It fired whenever the point lay within a fixed distance of any edge of its zone, the outer rim included, and no source states that distance or that advice; `borderline` takes its place. A result saved under 1.18 differs for the same input only in the new `borderline` object and that warning. No zone, polygon, vertex, boundary, region id, severity or recommendation changed, and the SVG figures are unchanged.
Last changed: Borderline flag beside `confidence` (#944). The /api/tools/iec-60599 `result` block now carries a `borderline` object: `state` (stays, stays-partial, borderline or not-assessed), `zonesReached`, `gasesHeld`, `convention` and, when no check was made, `notAssessedReason`. It says whether the five gases, varied by the analytical uncertainty IEC 60599:2022 §6.2 (printed p. 17) states for DGA values and notes carries over to the gas ratios, with S per gas from the service-test column of IEC 60567:2023 Table 5 (printed p. 49), give any result other than the reported Table 1 case: another case, ratios outside every Table 1 case (`DT`), or a set for which R3 cannot be calculated (`NS`). Gases below 5 × S are not varied, and the check is then a lower bound. When the reported `faultCode` is DT or NS there is no Table 1 case to check, and the state is not-assessed. The construction is TriboTech’s reading of the clause, and it is the same check the IEC 60599 tool page shows. `confidence`, the string tier HIGH, MEDIUM, LOW or INDETERMINATE, is deprecated: it is a software heuristic, not a probability, and IEC 60599 does not define it. It stays in the response, and its values are unchanged. A result saved under 1.12 differs for the same input only in the new `borderline` object. No Table 1 case, ratio limit, ratio, fault code or warning changed.
Last changed: Change to a new IEEE guide (#786): the IEEE path now scores mineral oil, natural ester and silicone against IEEE C57.166-2026, which supersedes IEEE C57.106, C57.111 and C57.147, so the same input can give a different verdict. In service in the main tank, all three liquids take Condition I/II/III (§8.2); a Condition III that rests only on rows §8.2 does not name is labelled as TriboTech’s reading. Natural ester in service gains dissipation-factor, interfacial-tension, acidity and colour limits. Silicone needs the rated voltage, and its breakdown voltage counts only as ASTM D1816; a D877 result is no longer scored. The mineral oxidation-inhibitor limit is scored only for oil declared Type II. The mineral tap-changer table splits into vacuum and non-vacuum columns, with a 230 kV class. IEEE C57.106-2015 stays selectable for mineral oil, marked superseded.
Last changed: Wording only: the methodology block names its two data files without a directory path, and the isotherm cross-check cites Oommen 1983 (EIC) Table II and Eq. (2) without a file path or catalogue number. No value, band, category or estimate changed — a result saved before this date can differ only in those three strings.
Last changed: Citation wording only (#804). Duval & Paulhiac 2026 is now cited at the printed pages of its published issue, IEEE TDEI vol. 33 no. 4, pp. 3053–3060: Table V at p. 3056 and § III.D at p. 3055. The two § III.D sentences quoted in the method-validity notices follow the published wording, which abbreviates "concentration" as "CON". No value, limit, gate, zone or verdict changed. A result saved before this date can differ only in those citations and quotes.
Last changed: Wording only: when a sample does not report both O2 and N2, the delta and rate disclosures now say that the O2/N2 ≤ 0.2 column carries the higher allowances and so gives the fewer findings, that using it is TriboTech's choice, and that IEEE C57.104-2019 prints a default for a missing ratio only in §6.1.1 Step 3 (printed p. 35), the > 0.2 section. They no longer call that column the stricter one. The column used did not change, and no rate, cell, band, gate or verdict changed — a result saved before this date can differ only in the wording of those disclosures.
Last changed: Wording only: the legal wording note now states that the action prose of §6.1.2.2 and §6.1.2.3 is not shipped on any surface, whether or not the terms of use have been accepted. No status, band, threshold, gate or verdict changed — a result saved before this date can differ only in that note.
Last changed: New method family (#467). A single mineral-oil sample read against the CO/CO2 paper-involvement criteria of IEC 60599:2022 §5.5 (printed p. 15), IEEE C57.104-2019 Annex D.8 (printed p. 69) and CIGRE TB 771 (2019) App. E.2 (printed p. 58), shown side by side with no verdict: each row states its own bounds, whether the sample meets each of them, and the source’s own conclusion sentence. The three disagree on the ratio bound (10 vs 20), on the CO2 bound (10 000 vs 15 000 cumulative), on how many conditions there are, and on what a ratio below 3 means, and the disagreement is the output. CO2/CO is withheld when CO is zero or not reported, IEEE D.8 c) can never read as met without a furan value, TB 771’s cumulative CO2 is never formed from a spot value without a declared degassing history, and any fluid other than mineral oil is a named gap. Nothing existed before this version.
Last changed: Page anchors only (#734). The IEC 60599:2022 Table A.10 NOTE 1, NOTE 2 and A.5.3 anchors are re-derived against the published edition (printed p. 32 → 31). Provenance only — no screen, bound, verdict or ratio moved.
Last changed: Sequence change (#925): Triangles 3 and 3b are no longer applied to a sample declared as rapeseed ester, because no rapeseed-ester boundaries are published for them (Duval 2008, Table 3; CIGRE TB 443, Table 4; CIGRE TB 771, App. H.8). The row is now shown as not applied, with that reason. Version 1.1 showed a zone computed on the synthetic-ester (MIDEL) boundaries instead. No other row and no agreement statement changed.
Last changed: Disclosure wording only: the scope and gap disclosures are rewritten as plain statements, and some now cite the correct clause. No zone, band, threshold, gate or verdict changed — a result saved before this date can differ only in the wording of its disclosures.
Last changed: Reported zone on the Triangle 1 boundary lines (#961). On Triangle 1, in its standard and OLTC modes, the envelope's reported zone comes from the triangle calculator, which moved to duval-triangle@1.30. For a point exactly on a boundary line of IEEE C57.104-2019, Table 6 (6.2.3, printed p. 42) whose percentages came out a floating-point rounding step below the line, `reportedZone` now gives the zone of the Table 6 row that holds at it, not the zone on the lower side. For example, 68.6 ppm CH4, 0.7 ppm C2H4 and 0.7 ppm C2H2, which is exactly 98 % CH4, 1 % C2H4 and 1 % C2H2, now reports PD; it reported T1. A result saved under 1.0 can differ for the same input only for such points. The perturbation convention, the perturbed points and the hull are unchanged, and no pentagon figure changed.
Last changed: First version (#466): the key gas of the principal fault (Unified Pentagon) and of the secondary faults (Triangles 1b and 4) is read against the fault-specific L1/L2 limits of Duval & Paulhiac 2026, Table X, p. 3058. Applied to mineral oil only (TriboTech's reading of the paper's scope); concentration only; no rate-of-change grade.
Last changed: Constituent change (#944): the families this route composes moved to duval-triangle@1.31, duval-pentagon@1.19 and iec-60599@1.13. `primary.result`, every `secondary[].result`, every `pentagons[].result` and `iec60599.result` now carry a `borderline` object, the same object the single-tool route returns for the same gases with the fluid the workflow’s lineup uses and otherwise that route’s defaults (`state`, `zonesReached`, `gasesHeld`, `convention` and, when no check was made, `notAssessedReason`). It says whether that figure’s own gases, varied by the analytical uncertainty IEC 60599:2022 §6.2 (printed p. 17) states for DGA values, with S per gas from the service-test column of IEC 60567:2023 Table 5 (printed p. 49), give any result other than the reported one: another zone or Table 1 case, no zone (`UNKNOWN`) or, on IEC 60599, `DT` or `NS`. The `registry` blocks do not carry it. `borderline` is not `gating.nearBoundary`: that block keeps its own convention and citation, reads Triangle 1 only, and appears only when a reported D+T receives Triangles 4/5 because its uncertainty box reaches a zone that licenses them. A primary block can therefore be borderline while `gating.nearBoundary` is null. `confidence` in every block is deprecated, and its values are unchanged. The pentagon blocks no longer carry the warning "Result is near region boundary, consider additional testing". The retired Triangle 2 warning cannot appear here, because this route does not run Triangle 2. Composition, tool selection, every gate rule, `gating`, `recommendedTools` and `faultLimits` are unchanged, and no zone in any block moved. A result saved under 1.26 differs for the same input only in the new `borderline` objects and that warning.