The Hardness Trap — What API 5CT Hardness Limits Do and Do Not Certify

API Specification 5CT, 11th Edition sets a maximum hardness for exactly four grades: L80 at 23.0 HRC / 241 HBW, C90 and T95 at 25.4 HRC / 255 HBW, and C110 at 29.0 HRC / 279 HBW (Table C.5). Every one of those is a maximum mean, not a maximum reading — footnote c to Table C.5 says so — and each carries a separate single-indentation reject threshold under 6.7.1: 24.0 HRC for L80, 27.0 HRC for C90 and T95, 31.0 HRC for C110. The hardness trap is not a gap between two numbers. It is the habit of reading an API 5CT acceptance criterion as though it were a statement about whether the pipe suits your well.

API 5CT says plainly that it is not. The NOTE to 6.14.1 reads: "The SSC test is for quality control purposes only and does not qualify the material for any specific sour service application; it is the product user's responsibility to ensure that the product is suitable for the intended application." Clause 6.14.1 then directs the purchaser to NACE MR0175 / ISO 15156-1 and ISO 15156-2 for guidance on using C90, T95 and C110. Two documents, two questions. This page covers what API 5CT specifies and verifies; the environmental limits — H₂S partial pressure, pH, temperature, chlorides — are set in ISO 15156-2 and in your project specification, and we do not reproduce them here.

ZC Steel Pipe supplies L80, C90, T95 and C110 casing with hardness means reported against the serial numbers that 5.4.2 makes mandatory for C90, T95, C110 and Q125, and with the SSC test method, specimen count and applied stress stated on the certificate rather than implied. If you are still deciding which grade the well needs, start with the OCTG sour service grade selection guide.

The Two Standards and Their Purposes

API 5CT defines manufacturing requirements for casing and tubing — chemistry, mechanical properties, dimensional tolerances, testing and marking. Its hardness limits exist so that a grade's heat treatment can be verified as consistent and within band. A mill that produces T95 within the Table C.5 ceiling, without breaching the single-indentation threshold or the variation allowance, has manufactured the pipe correctly per API. That is the whole scope of what those numbers certify.

NACE MR0175 / ISO 15156 answers a different question: is this material acceptable in a specific H₂S-containing environment? That answer depends on the environment as much as the metal — H₂S partial pressure, in-situ pH, temperature, chloride content, and the presence of elemental sulphur. ISO 15156-2 covers carbon and low-alloy steels and ISO 15156-3 covers corrosion-resistant alloys. Neither is reproduced in API 5CT, and API 5CT does not attempt to summarise them.

The trap closes when a purchase order writes "API 5CT compliant, NACE MR0175 compliant" and treats the two phrases as one requirement. The first is checkable at the mill against numbered clauses. The second, without a stated environment and a stated limit, is checkable against nothing.

API 5CT Hardness Limits, Grade by Grade

Everything in the table below comes from API Specification 5CT, 11th Edition — Table C.5 for the ceilings and 6.7.1 for the single-indentation thresholds and retest bands. No figure from any other document appears in it.

Three readings of this table matter more than the numbers themselves. First, the ceiling column is a mean — a joint can be accepted with individual indentations above it, provided the average holds, and can be rejected on a single indentation without any mean being calculated. Second, L80 differs from the C90/T95/C110 family not only in ceiling but in how much of the order is measured at all: two tests per lot against one test on every length. Third, the four grades in the list are the only ones API 5CT limits — P110 has no hardness ceiling anywhere in the standard, which is a statement about what the mill must verify, not a verdict on any well.

For the complete grade ladder with API 5CT hardness limits, see the API 5CT specification tables →

To work through grade options against your well conditions, use the Pipe Grade Selector → and the sour service selector →

L80 — Two Tests per Lot, and What That Means

L80 Type 1 is the most widely ordered of the hardness-limited grades — see the full L80 grades specification for the complete property set, and the N80 vs L80 comparison for why N80 is not an equivalent. Its Table C.5 ceiling is 23.0 HRC / 241 HBW as a maximum mean, with 6.7.1 a) 1) rejecting the length or piece if any single indentation exceeds 24.0 HRC. Unlike C90, T95 and C110, L80 has no retest band between the two — 6.7.1 a) 1) gives an acceptance and a reject with nothing in between, and 9.6.13 governs L80 retests separately.

The frequency is the part most buyers never look at. Clause 9.6.3 states that on L80 pipe, coupling stock, coupling blanks, coupling material and accessory material, "hardness testing shall be carried out at the same frequency as tensile testing" — a lot basis. Table C.35 sets that at 2 tests per lot, through-wall in one quadrant at the body tensile test location, with a lot running up to 200 lengths at Label 1 of 4½ and below and up to 100 lengths above 4½. On a 100-joint order of 9⅝″ casing, two joints are measured and 98 are not.

Named failure mode — the unmeasured 98. This is a data failure before it is ever a metallurgical one. When a certificate shows "API 5CT compliant" against an L80 lot, the hardness statement behind it may rest on two through-wall tests representing up to a hundred lengths. Nothing is wrong with that — it is exactly what 9.6.3 and Table C.35 require — but it does not support a per-joint claim, and no amount of document review after delivery will turn it into one. The diagnostic is simple: count the hardness rows on the certificate and compare them to the joint count. If there are two rows and a hundred joints, you have lot data. The fix belongs at the order stage, under 9.6.1, which states that additional hardness testing on the outside surface and through-wall hardness testing of pipe and upsets "may be carried out as agreed upon between the purchaser and the manufacturer", with the test procedures likewise agreed. Agreement item, before manufacture — not a request afterwards.

L80-13Cr, 9Cr and 3Cr deserve a separate note, because they share L80 Type 1's 23.0 HRC / 241 HBW row in Table C.5 and are routinely mistaken for sour grades on the strength of it. They are not. They are CO₂ corrosion grades, distinguished by chromium — 12.0–14.0 % for 13Cr, 8.00–10.0 % for 9Cr, 2.50–3.90 % for 3Cr under Table C.4 — and they are also chromium-bearing CRAs, which puts their H₂S service limits under ISO 15156-3 rather than -2. API 5CT specifies no SSC test for any L80 variant; 6.14 and 9.10 are titled "Sulfide Stress Cracking Test—Grades C90, T95, and C110" and L80 is not in scope. Sharing a hardness row with a carbon-steel grade is not shared service applicability. See the sour service grade selection guide for where each sits.

T95 and C90 — Every Length Tested, and What the Number Means

T95 has no types. API 5CT 11th Edition clause 1.3 assigns types to L80 and N80 only, the Type column against T95 is blank in Table C.4 and Table C.5, and "T95 Type 2" does not occur anywhere in the standard. If that phrase is on your purchase order, correct it before placing — the T95 specification page and the L80 vs T95 selection guide cover the grade and when it is the right step up.

On testing frequency, T95 and C90 are covered better than L80, not worse — the reverse of what is usually assumed. 9.6.4 a) requires, for non-upset C90 and T95 pipe, "one through-wall hardness test in one quadrant shall be made from one end of each length," with approximately 50 % of the test rings cut from front ends and approximately 50 % from back ends. Every length is measured. 9.6.5 adds that upset T95 pipe is through-wall tested in all four quadrants on the body of each length tensile-tested, with the upset itself tested one in every 20 lengths within each lot. And A.19 (SR 47), which the purchaser must specify, requires a test ring from both ends of each pipe rather than one — which is the real supplementary lever for hardness on this grade, adding the second end rather than adding the first.

What the resulting number means is the part that goes wrong. Three distinct criteria sit behind a single printed figure:

That last row is the one nobody asks about. Table C.5 column 10 sets the allowable hardness variation for C90, T95, C110 and Q125 by wall thickness: 3.0 HRC at wall up to 12.70 mm (0.500 in.), 4.0 HRC from 12.71 to 19.04 mm, 5.0 HRC from 19.05 to 25.39 mm, and 6.0 HRC at 25.40 mm and above. Clause 6.8 defines the variation as the difference between any two mean hardness numbers within one quadrant, and states it does not apply between specimens. A joint sitting well inside the ceiling on every reading can still be non-conforming if the spread inside one quadrant is too wide — which is a heat treatment uniformity signal, and arguably a more useful one than the ceiling.

So when a certificate prints one hardness figure per length with no label, ask two questions before reading anything into it: is this a mean, and how many indentations does it represent? Without those, you cannot tell which of the four rows above was applied.

Couplings are a separate product with a separate sampling rule, and buyers frequently specify one and forget the other. API 5CT does cover them — 9.6.6 sets the frequency for C90, T95 and C110 coupling blanks, coupling stock, coupling material, pup joints and accessory material: for thick-wall tube used to make more than one coupling blank, through-wall hardness tests are made on two test rings, one from each end; for pieces heat-treated in individual lengths, the piece with the highest surface hardness in the lot is selected for through-wall testing; and those through-wall tests are made in four quadrants, not one. Clause 6.9 requires every individually heat-treated coupling blank, pup joint or accessory to be surface hardness tested for process control, and states that those results need not be provided by the manufacturer unless specified in the purchase agreement.

Named failure mode — the coupling data gap. The consequence is documentary and entirely avoidable. Coupling hardness is measured under a different clause, at a different frequency, in four quadrants rather than one, and a whole tier of it — the 6.9 process-control surface hardness used to select which piece gets the through-wall test — is withheld by default unless the purchase agreement calls for it. The diagnostic is the certificate: if there is no coupling section separate from the body section, coupling hardness was not reported, whether or not it was measured. The fix is one line invoking 6.9 reporting and requiring the 9.6.6 four-quadrant through-wall results to appear as their own block on the certificate.

HAZ Hardness — The Hidden Trap

The HAZ hardness trap operates outside the API 5CT system entirely — it applies at installation, not at manufacture, and requires action at the project engineering level rather than the pipe procurement level.

When a girth weld is made during installation, the heat input rapidly heats and then cools the base metal adjacent to the weld. In the zone immediately beside the fusion line — the HAZ — the parent metal is taken above transformation temperature and then cooled at a rate set by the surrounding heat sink. Depending on alloy content and heat input, that can produce a martensite-rich microstructure considerably harder than the pipe body it came from. The HAZ is not the weld metal; it is unmelted parent material that has been thermally transformed.

API 5CT sets hardness requirements for product the pipe manufacturer makes: pipe body, coupling stock, coupling blanks, pup joints and accessory material. It does not address a girth weld made on site months later, and it does not pretend to. Note also that API 5CT does address the one weld it owns — for electric-welded quenched and tempered pipe, 9.15.10 requires weld seam inspection after final heat treatment and rotary straightening, with full volumetric inspection over a 3 mm (⅛ in.) wide zone centred on the fusion line. But that is the mill's longitudinal seam, not your installation girth weld.

So a casing string with complete per-length hardness documentation can still have uncontrolled HAZ hardness at every field weld, and nothing on the pipe MTC speaks to it. Any HAZ hardness limit comes from the project welding specification and from ISO 15156-2, and it is verified on welding procedure qualification coupons — typically by a Vickers hardness traverse across the weld cross-section — before production welding starts. The levers are heat input control, run sequence, preheat and interpass temperature, and post-weld heat treatment where the procedure calls for it.

Keep the two documents apart on the page. The pipe manufacturer cannot prevent HAZ hardness failures on your installation welds, and a pipe MTC is not evidence about them. That responsibility sits with the installation contractor and the project welding engineer, under a qualification record of its own.

C110 — the Grade API 5CT Warns About in Its Own Text

C110 carries the highest hardness ceiling of the four limited grades: 29.0 HRC / 279 HBW maximum mean under Table C.5, with a single-indentation reject above 31.0 HRC and a 29.0–31.0 HRC retest band under 6.7.1 a) 3) and 9.6.14. It also carries the tightest sulphur limit in the whole standard at 0.005 % maximum under Table C.4 — half of T95's 0.010 % and one sixth of P110's 0.030 % — alongside Mo at 0.25–1.00 %, Cr at 0.40–1.50 % and Ni at 0.99 % maximum. The full C110 casing specification covers where the grade sits in the ladder.

What makes C110 worth a separate section is that API 5CT itself flags it. Clause 6.14.1 states: "The purchaser should refer to NACE MR0175/ISO 15156-1 and ISO 15156-2 for guidance on the usage of Grades C90, T95, and C110. Particular attention should be given to the application of Grade C110 in NACE MR0175/ISO 15156-2 SSC Regions 2 or 3, as this material is not suitable for all sour (hydrogen sulfide-containing) service applications." That is the specification warning the purchaser about a grade the specification itself contains — and pointing at a different document for the answer.

API 5CT backs that with heavier testing rather than a different hardness philosophy. C110 gets three SSC specimens per lot under 9.10.2 a) 2) for Method A, taken from the ends of three different products drawn from the front, middle and back thirds of the lot, against one specimen for C90 and T95. Its Method A applied stress is 85 % of minimum yield — 644 MPa (93,500 psi) — per Table 12, against 80 % for C90 and T95. And 9.6.4 b) requires a through-wall hardness test on both ends of every C110 length by default, reducible to the C90/T95 single-end frequency only where the manufacturer demonstrates a process control plan to the purchaser's satisfaction.

Whether C110 may be used in your well is not a question API 5CT answers, and its own 6.14.1 says so. That determination belongs to ISO 15156-2 and to the project specification, against the well's H₂S partial pressure, pH, temperature and chloride content. What we can put on the certificate is what API 5CT requires and what your order added on top: the SSC test method, specimen count and applied stress; the hardness means with their indentation counts at the stated frequency; and the serial number 5.4.2 requires on every C110 length.

Worked Example — Applying API 5CT's Own Hardness Criteria

A T95 certificate for 7 in × 26 lb/ft casing reports through-wall hardness under 9.6.4 a), one test ring per length, four indentations per test, in a single quadrant. Wall thickness is 9.19 mm (0.362 in.), which puts the joint in the Table C.5 band at or below 12.70 mm — allowable hardness variation 3.0 HRC. Applying 6.7.1 a) 2), 9.6.14 and 6.8 in order:

Four points come out of this that a flat "25.4 HRC max" column would have hidden. Joint 1 is accepted with every single indentation above 24 HRC, because the criterion is the mean. Joint 2 is not rejected — it goes to retest, and if the new mean from three fresh indentations lands at or below 25.4 HRC it is accepted. Joint 3 has the same mean as Joint 2 and is rejected outright on a single 27.4 HRC reading; the mean is never the deciding factor once 27.0 is exceeded. Joint 5, at 25.3 HRC, is accepted and is a tighter, better-controlled joint than Joint 4 on variation even though it sits 4 HRC higher.

Joint 4 is the interesting one. Its mean of 21.0 HRC is comfortably inside the ceiling, but the spread within the quadrant is 3.3 HRC against a 3.0 HRC allowance at this wall thickness. Under 6.8 — "hardness variation is defined as the difference between any two mean hardness numbers within one quadrant" — a full evaluation requires a second mean within the quadrant to compare against, which is precisely why the number of indentations and their positions matter and a single printed figure does not carry enough information. Ask for the raw indentation data, not the summary.

None of the above says anything about whether any of these joints suits a particular well. It is API 5CT acceptance, applied correctly. The environmental question is decided against ISO 15156-2 and the project specification, and requires the well conditions — not just a hardness column.

Correct Purchase Order Language

Two phrases do most of the damage. "NACE MR0175/ISO 15156 compliant" names a document without naming a requirement or an environment, so the mill builds to API 5CT and certifies to API 5CT — correctly. And "PSL-2" does not exist in API 5CT at all; Product Specification Levels are an API 5L construct, the phrase does not occur anywhere in the 11th Edition, and every "PSL" string in the document is an end-finish code inside the dimension tables. A line reading "API 5CT T95 Type 2, PSL-2, NACE compliant" contains three terms the standard cannot act on, and the mill's compliant response is plain T95 at default test frequencies.

Write clauses and numbers instead.

L80 PO requirements:

• Standard: API Specification 5CT, 11th Edition, Grade L80 Type 1 (state the type — L80 is one of the two grades that has them)
• Acceptance criterion: API 5CT 6.7.1 a) 1) — maximum mean 23.0 HRC / 241 HBW, single indentation above 24.0 HRC rejects
• Hardness testing frequency: additional through-wall testing beyond the 2 tests per lot of 9.6.3 and Table C.35, agreed under 9.6.1 with test procedures likewise agreed. Settle this before the order is placed; it cannot be created afterwards
• Traceability: 5.4.1 traceability to heat and lot with all applicable chemical and mechanical test results is purchaser-specified for L80 — say so explicitly
• NDE: Table C.37 puts L80 at acceptance level L4 under Table C.38. Specify A.3 (SR 2) if you want L80 inspected to the P110 regime at level L2
• Any project hardness limit differing from API 5CT: state the number and cite the project specification document it comes from
• MTC: EN 10204 3.2, with hardness records tabulated against the identity the mill is maintaining under 5.4.1

T95 PO requirements:

• Standard: API Specification 5CT, 11th Edition, Grade T95 — no type designation applies
• Acceptance criterion: API 5CT 6.7.1 a) 2) — maximum mean 25.4 HRC / 255 HBW, retest band 25.4–27.0 HRC per 9.6.14, single indentation above 27.0 HRC rejects; hardness variation per 6.8 and Table C.5 column 10
• Hardness testing: per-length body testing is already required by 9.6.4 a). Specify A.19 (SR 47) for a test ring from both ends of each pipe, and require the individual means with their indentation counts reported against the 5.4.2 serial numbers
• Couplings: require the 9.6.6 four-quadrant through-wall results as a separate certificate block, and invoke 6.9 so that process-control surface hardness results are provided rather than withheld
• SSC testing: state the NACE TM0177-2016 method. Specify A.18 (SR 46) to move from the 9.10.2 a) 1) default of one Method A specimen per lot at 80 % of minimum yield — 524 MPa (76,000 psi) for T95 — to three specimens from the front, middle and back thirds at 90 %, which is 590 MPa (85,500 psi)
• Toughness: specify A.9 (SR 16) if Charpy testing is to be mandatory rather than qualifiable by documented procedure under 6.5.5
• Any project hardness limit differing from API 5CT: state the number, cite the project specification, and place it under 6.7.1 c) with the SSC testing that clause references
• MTC: EN 10204 3.2

C110 PO requirements:

• Standard: API Specification 5CT, 11th Edition, Grade C110
• Acceptance criterion: API 5CT 6.7.1 a) 3) — maximum mean 29.0 HRC / 279 HBW, retest band 29.0–31.0 HRC per 9.6.14, single indentation above 31.0 HRC rejects
• Hardness testing: both ends of each length under 9.6.4 b). If the mill proposes the reduced single-end frequency, require the process control plan demonstration that 9.6.4 b) makes a condition of it
• SSC testing: three Method A specimens per lot from the front, middle and back thirds under 9.10.2 a) 2), at 85 % of minimum yield — 644 MPa (93,500 psi) per Table 12. State whether any reduction to one per lot under 9.10.2 a) 3) ii) is acceptable, and on what process control evidence
• Environmental suitability: this is an ISO 15156-2 and project specification determination, not an API 5CT one — 6.14.1 says so. State the well's H₂S partial pressure, temperature, pH and chloride content in the enquiry so the project's own limits can be applied to the grade
• MTC: EN 10204 3.2, with serial numbers per 5.4.2

Write these before manufacturing begins. Retesting after delivery, where it is possible at all, adds weeks and still leaves you with data the standard never required anyone to collect.

MTC Review Checklist for Sour Service Casing

Reviewing an MTC against both API 5CT and NACE MR0175 requires checking fields that a standard API acceptance review does not typically touch.

Before accepting any sour service casing consignment, verify on the MTC:

Grade and standard: Confirm the grade is explicitly stated and that the type designation is used only where the standard has one. Under clause 1.3, "L80" used alone covers L80 Type 1, 3Cr, 9Cr and 13Cr, so a certificate reading plain "L80" has not told you which of the four you received — and 3Cr, 9Cr and 13Cr are CO₂ grades, not carbon-steel sour grades. T95, C90 and C110 have no types at all; a certificate that prints one is describing something the standard does not contain. Check that the API 5CT edition cited matches your PO.

Heat treatment record: Table C.3 requires quench and temper for L80, C90, T95, C110 and Q125, with a minimum tempering temperature against each — 566 °C (1050 °F) for L80 Type 1 and 3Cr, 621 °C (1150 °F) for C90, and 649 °C (1200 °F) for T95 and C110. The record should appear as its own field showing the actual tempering temperature, not a tick against "Q+T". If it is missing, the mill may have met the requirement but has not demonstrated it.

Hardness — pipe body: Establish what the hardness column actually represents before reading anything into it. Is the figure a mean or a single indentation, and how many indentations sit behind it? For L80 the default under 9.6.3 and Table C.35 is two tests per lot, not a value per length, so a per-length column exists only if it was agreed under 9.6.1 at order stage. For C90, T95 and C110 it should be one per length under 9.6.4, or both ends where 9.6.4 b) or A.19 (SR 47) applies. Check the figure against the Table C.5 mean ceiling and the 6.7.1 single-indentation threshold separately, and check the spread against the Table C.5 column 10 variation allowance for the wall thickness. Confirm the scale matches your PO — Table C.5 footnote a makes laboratory Rockwell C the referee method in case of dispute.

Hardness — coupling: This belongs in its own section, because it is sampled under a different clause. For C90, T95 and C110, 9.6.6 requires through-wall tests in four quadrants, on two test rings one from each end of thick-wall tube used for multiple blanks, or on the piece with the highest surface hardness in the lot where blanks were heat-treated individually. The 6.9 process-control surface hardness results that drive that selection are withheld unless the purchase agreement calls for them. If there is no coupling block on the certificate, coupling hardness was not reported.

SSC test record: For C90, T95 and C110 this is mandatory under 6.14 and 9.10 and should state the NACE TM0177-2016 method, the number of specimens per lot, where in the lot they were drawn from, and the applied stress as a percentage of minimum yield with the MPa and psi figure behind it. One Method A specimen at 80 % is the C90/T95 default; three at 90 % means A.18 (SR 46) was invoked; three at 85 % is the C110 default. If the certificate says only "SSC tested", it has not told you which of those you bought. Note that L80 has no SSC test in API 5CT at all — 6.14 and 9.10 are both headed "Sulfide Stress Cracking Test—Grades C90, T95, and C110".

Serial numbers: For C90, T95, C110 and Q125, 5.4.2 requires each length to be uniquely numbered so that test data can be related to individual lengths, and the same for coupling stock, coupling material, coupling blanks, pup joints and accessory material. This is mandatory, not purchaser-specified — a certificate for those grades without serial numbers tied to the test data is a genuine non-conformance, and it is the cheapest thing on this list to check.

What the certificate does not certify: An API 5CT certificate is evidence of API 5CT conformance and nothing else. Environmental suitability is determined against ISO 15156-2 and your project specification, using the well's H₂S partial pressure, pH, temperature and chloride content — none of which appear on a mill certificate. If your project requires a documented material selection for the environment, that is a separate document with a separate author, and it is not something a mill can issue.

The review takes fifteen minutes per certificate if the data is there. If the data is missing, it is better to discover that at the mill than at the wellsite.