Look at the numbers first, read second. Everything here is a stored engine answer on a fictional foundry's stock, not figures written into the page.
1. Aim at the window, not the limit
A heat that lands right on the upper carbon limit is scrap at the first independent check. So the charge aims at the central part of each element's field.
Element
Grade field
Expected
Status
C
0.36–0.44
0.38690.3869
in field
Si
0.17–0.37
0.19540.1954
in field
Mn
0.5–0.8
0.5260.526
in field
Cr
0.8–1.1
0.84910.8491
in field
Ni
≤ 0.3
0.030.03
in field
Cu
≤ 0.3
0.06050.0605
in field
P
≤ 0.035
0.01950.0195
in field
S
≤ 0.035
0.01570.0157
in field
The charge aims at the working window — the central 70% of each element's tolerance field, leaving room for analysis scatter.
2. Mass: from charge to good casting
Metal yield and melting loss sit between what was charged and what was tapped; gates and risers sit between the tap and good casting. Cost is per kilogram of good casting.
824.5 kgCharged
→
799.8 kgTapped
→
519.9 kgGood casting
→
UAH 41.6Per kg of good casting
Melting loss in this heat: 1.84 kg; it depends on how the furnace recovers each element.
3. The stock falls short — we say what is missing
The same stock without ferrochrome. Linear programming does not answer “no solution” here: limits are broken at a penalty and the answer names the element and the shortfall.
Engine examplealloygen.charge_calc/1Data under review
Stock without FeCr
Input: same stock, no ferrochrome
The stock cannot make 40X: Cr is short by 0.8062%, expected 0.0388% against a minimum of 0.8%. The engine says what is missing instead of “no solution”.
Status
stock falls short
Cr: short by
0.8062 %
Cr: expected
0.0388 %
Cr: grade field
0.8–1.1 %
ImportantCr: short by 0.8062% — the stock cannot cover it.
This is not a charge for the furnace: limits are broken on purpose to show the shortfall. The fix is a carrier of the element in stock — ferrochrome or pure chromium.
Assumptions, warnings and sources
Warnings
ImportantCr: short by 0.8062% — the stock cannot cover it.
Mediumgrade 40X limits are checked against secondary sources, not the text of GOST 4543-2016 / DSTU 7806:2015 (table 1 as quoted elsewhere; the standard itself was not read). Not enough for a certificate.
Mediumbrittle temperature range LIT–ZDT 45 °C (at 1 °C/s); sulphur and phosphorus account for 28 °C (63%) of it: they segregate into the liquid between dendrites (at fs = 0.9: S 0.127%, P 0.115%). With S 0.0078% and P 0.00965% the range narrows to 31 °C. In a furnace without refining this is done only through the charge: scrap and ferroalloys lower in sulphur and phosphorus.
Mediumno product requirements given: there is nothing to select a grade by, so the calculation runs only for the grade named explicitly.
Mediumprices of 3 charge items have no price list or date (GRAPH, FEMN, FESI): the cost estimate is a demo — add your dated price list.
Mediumelement recovery in furnace IF_1T is a reference value, not derived from heat history. The expected analysis will differ from the ladle sample more, the more silicon, manganese, aluminium and titanium the charge has.
Mediumthe tariff of 8.5 UAH/kWh is not backed by a bill or contract — set your own tariff.
NoteCE(IIW) = 0.489 > 0.45: welding will need preheating.
Assumptions
The charge aims at the working window — the central 70% of each element's tolerance field, leaving room for analysis scatter.
Grade 40X limits come from the alloygen grade book via secondary sources (GOST 4543-2016, table 1); not checked against the official text — pilot, under review.
Charge compositions, prices and stock are the foundry's data; the service computes with them and does not verify them (warnings say where an item has no document).
Melting loss uses element recovery for furnace IF_1T (induction): reference values, not calibrated on this furnace's heat history.
Cost: charge at stock prices plus energy, lining and conversion from furnace data; lines and totals are in the response.
The problem was solved elastically: composition limits were broken at a penalty to show what is missing and by how much — this is not a charge for the furnace.
Sources
GOST 4543-2016 / DSTU 7806:2015 — structural alloy steel products, grades (via secondary sources)as of Oct 5, 2026Data under review
SCRAP_CLEAN price: foundry invoice No. 118as of Sep 20, 2026
4. A number without an interval is a lie
Critical points and hardness are empirical models with intervals. Below: the order asks for 240–280 HB, and each route's prediction interval is on the same scale.
737 °CAc1 · 717–757 · Andrews 1965
772 °CAc3 · 747–797 · Andrews 1965
335 °CMs · 310–360 · Andrews 1965
quench 850 °C, oil + temper 550 °C, 2.0 h your route
borderline
quench 810 °C, oil + temper 690 °C, 2.0 h
borderline
quench 810 °C, oil + temper 700 °C, 2.0 h
borderline
ordered prediction interval
5. Refusal is a result too
Without a sample analysis no passport is issued: the engine gives a stamped draft and a list of what is missing. A forecast never goes into a passport.
Engine examplealloygen.passport/1Data under review
Heat passport
Input: no sample analysis
Without a sample analysis it is only a draft passport for 40X: the calculated composition is in the field for 8 of 8 specified elements.
Document
draft
Certificate
no
Elements in field
8 / 8
Form fields
8
DRAFT · CALCULATED DOCUMENT · NOT A CERTIFICATE
What a passport needs
the ladle or product sample analysis of this heat: heat number, date, sample type, method, laboratory, instrument, manufacturer, a signatory independent of production, and all specified elements (C, Si, Mn, P, S, Cr, Ni, Cu)
the text of EN 10204 — to check the document form
Assumptions, warnings and sources
Warnings
Mediumgrade 40X limits are checked against secondary sources, not the text of GOST 4543-2016 / DSTU 7806:2015 (table 1 as quoted elsewhere; the standard itself was not read). Not enough for a certificate.
Mediumbrittle temperature range LIT–ZDT 45 °C (at 1 °C/s); sulphur and phosphorus account for 28 °C (63%) of it: they segregate into the liquid between dendrites (at fs = 0.9: S 0.128%, P 0.116%). With S 0.00785% and P 0.00975% the range narrows to 31 °C. In a furnace without refining this is done only through the charge: scrap and ferroalloys lower in sulphur and phosphorus.
Mediumthe “heat treatment” cost line 3797.87 (3173.97…4866.32): heat-treatment furnace efficiency, holding losses, furnace-hour rate and oil use and price are demo values. Replace them with your own furnace data.
Mediumno product requirements given: there is nothing to select a grade by, so the calculation runs only for the grade named explicitly.
Mediumonly a draft was issued: there is no actual ladle or product sample analysis — the composition in the document is calculated.
Mediumthe document form is not checked against the text of EN 10204: the standard was not bought, fields come from open descriptions.
Mediumprices of 4 charge items have no price list or date (FECR, GRAPH, FEMN, FESI): the cost estimate is a demo — add your dated price list.
Mediumelement recovery in furnace IF_1T is a reference value, not derived from heat history. The expected analysis will differ from the ladle sample more, the more silicon, manganese, aluminium and titanium the charge has.
Mediumthe tariff of 8.5 UAH/kWh is not backed by a bill or contract — set your own tariff.
Mediumtempering at 600 °C is in the reversible temper-embrittlement range with molybdenum below 0.20%: after tempering cool in water or oil, not with the furnace.
Notean induction furnace does not lower hydrogen: [H] is set by charge moisture and atmosphere and is not predicted; the safe level for large forgings is 2 ppm, reached at equilibrium only at pH2 ≤ 7.24 mbar (5.72…9.84) at 1599 °C — that is vacuum. For a large forging: vacuum degassing or anti-flake treatment; no section threshold was found in open sources — the foundry sets it.
NoteCE(IIW) = 0.65 > 0.45: welding will need preheating.
Assumptions
A passport is issued only from the actual sample analysis of this heat within the tolerance field; without a sample it is a draft stamped “not a certificate”.
Property forecasts never go into the passport: with no test report the line reads “not tested”.
The document form is not checked against EN 10204: the standard was not bought, fields come from open descriptions.
Grade 40X limits come from the alloygen grade book via secondary sources (GOST 4543-2016, table 1); not checked against the official text — pilot, under review.
Charge compositions, prices and stock are the foundry's data; the service computes with them and does not verify them (warnings say where an item has no document).
Sources
GOST 4543-2016 / DSTU 7806:2015 — structural alloy steel products, grades (via secondary sources)as of Oct 5, 2026Data under review
Sources with dates
Sources of the examples on this site. Grade limits are checked against secondary sources and flagged “data under review”.
The engine adds these warnings to every answer until the cause is fixed. A calculation on reference coefficients looks exactly like one on calibrated ones, so we do not hide them.
GRADE_SECONDARY_SOURCE
Grade limits
Checked against secondary sources, not a purchased copy of the GOST/DSTU text. Not enough for a certificate — the pilot runs as “a calculation is a forecast”.
RECOVERY_UNCALIBRATED
Your furnace's recovery
A reference value for now. Once your furnace has a history of expected vs actual heats, the engine calibrates it itself.
PRICE_UNSOURCED
Charge prices and tariff
Without a price list or invoice the cost is a demo. Add a dated price and the warning goes away.
PASSPORT_FORM_UNVERIFIED
Passport form
Fields come from open descriptions of EN 10204; the standard has not been bought yet. Without a sample analysis a passport is always a draft.