DoaWise

2026-08-01 · EN

Post-Weld Heat Treatment (PWHT)

Post-weld heat treatment (PWHT) is one of the most misunderstood operations in the field. Most crews picture it as "put the weld in a furnace and heat it for a few hours." In code language, however, PWHT is an umbrella term. Its most common form is a stress-relief anneal, and in carbon steels the primary function really is stress relief. In martensitic chromium steels such as P91 and P92, the primary function is tempering, and stress relief comes along as a by-product.

Austenitic stainless steels are a different case entirely: for that family PWHT is, as a rule, neither required nor desirable. Where it has to be applied, three separate options are on the table — low-temperature stress relief with limited effect (~400–450 °C), stress relief in the 850–900 °C band, which carries sensitization and sigma-phase risk, or full solution annealing (~1040–1120 °C plus rapid cooling), which is a separate manufacturing operation rather than PWHT. In every case, holding in the 425–815 °C band risks chromium carbide precipitation — that is, sensitization. So "PWHT usually turns into a solution anneal" is a misleading shortcut.

Welding is a violent local process. Heat input melts and resolidifies the joint, transforms the surrounding parent metal through the heat-affected zone (HAZ), and leaves behind high residual stress and hard microstructures. PWHT anneals that region at a controlled temperature and puts the component safely into service. But the wrong temperature, the wrong time, or an uncontrolled heating rate turns a beneficial operation into a harmful one — especially in modern chromium-molybdenum steels such as P91.

The code references in this article follow the ASME BPVC 2023 Edition, the edition available to us when the article was prepared. The current edition in force is 2025 (published 1 July 2025, mandatory from 1 January 2026). In practice the edition in your hands always governs, so verify every number below against it.

1. What PWHT actually solves

  • Residual stress reduction: The weld bead shrinks as it cools, the surrounding mass restrains it, and tensile stresses close to the yield strength lock into the weld zone. At PWHT temperature the yield strength of the material drops, so those stresses relax through local plastic flow and creep relaxation. PWHT does not zero out residual stress; it typically brings it down to a small fraction of the yield strength — on the order of a quarter of the original level in most carbon steels. The exact ratio depends strongly on material, temperature, time and geometry: relaxation scales with the Hollomon–Jaffe / Larson–Miller parameter (LMP), which means temperature and time are not fully interchangeable and temperature is far more effective than time. For that reason, single-number claims such as "PWHT reduces stress to X%" are no substitute for a specification. Even this partial reduction clearly lowers the risk of stress corrosion cracking (SCC) and brittle fracture.
  • Hardness reduction (tempering): Martensite formed by rapid cooling in the HAZ is hard and brittle. It tempers at soak temperature. For sour (H₂S) service, the acceptance criterion for carbon and low-alloy steels under ISO 15156-2 is 250 HV10 in the weld and HAZ and 22 HRC in the parent metal. The two are not the same criterion, and an HRC measurement is not metallurgically valid in a HAZ that narrow. For refinery wet H₂S service, the governing document is NACE MR0103 / ISO 17945.
  • Hydrogen removal: Diffusible hydrogen dissolved during welding diffuses out of the structure at elevated temperature, which reduces the risk of delayed cold cracking. For hydrogen removal alone, a "bake-out" (DHT) is typically applied in the 200–350 °C band; the time depends on thickness and diffusion distance. Watch out for a common reference mix-up: Methods A and B in EN 1011-2 Annex C are preheat temperature calculation methods — they are not a DHT/bake-out provision, and AWS D1.1 contains no general "hydrogen removal heat treatment" clause either. Take DHT time and temperature values from the project specification, from API 582, and from diffusion-based sources such as ISO/TR 17844; use EN 1011-2 Method A/B for preheat and preheat maintenance calculations. There is no single universal number.
  • Toughness and dimensional stability: A tempered structure gives higher notch toughness and improves dimensional stability during subsequent machining and in service.

2. Soak temperature and time: thickness decides

PWHT is defined by two basic parameters: soak temperature and soak time. The most frequent mistake here is treating the common field practice band as if it were the code minimum. The two must be stated separately.

ASME VIII Div. 1 minimum soak temperatures: P-No. 1 → 593 °C (1100 °F), P-No. 3 → 593 °C, P-No. 4 (1Cr-½Mo, 1¼Cr-½Mo) → 650 °C (1200 °F), P-No. 5A (2¼Cr-1Mo) → 677 °C (1250 °F), P-No. 15E (Grade 91) → 732 °C (1350 °F). The 593 °C minimum applies only to P-No. 1 and P-No. 3; running P-No. 4 at 593 °C is a direct nonconformity.

Common field practice bands: 595–620 °C for carbon-manganese steel; 650–720 °C for 1¼Cr-½Mo; roughly 705–715 °C for 2¼Cr-1Mo. European practice may show a 550–620 °C band for carbon steel, but on ASME work you cannot go below 593 °C for P-No. 1 and P-No. 3. If a lower temperature is unavoidable, use the alternative lower-temperature / longer-time rows of Table UCS-56-1 (for example 2 h/in. at 566 °C (1050 °F), 4 h/in. at 538 °C (1000 °F), 10 h/in. at 510 °C (950 °F)). Those rows are not unconditional. They are recognized essentially for P-No. 1 (Gr. 1/2/3); no such option exists for groups such as P-No. 4, 5 and 15E; they are restricted where notch impact (toughness) requirements apply; and they cannot be used for PWHT mandated by service, such as sour service or lethal service.

A separate rule, violated far too often: on quenched and tempered (Q&T) materials the PWHT temperature shall stay at least 30 °C (50 °F) below the tempering temperature of the parent metal — and in every case below the tempering temperature stated on the material certificate. A loose margin such as "20 °C below" is misleading. Exceed it and the strength of the parent metal drops permanently, invalidating the properties the material certificate rests on. This rule can conflict with the code minimum: for a Q&T steel tempered at 600–620 °C, the "stay at least 30 °C below" rule points to 570–590 °C while the ASME P-No. 1/3 minimum is 593 °C. In that situation the tempering temperature on the material certificate governs. The answer is not to pick one arbitrarily, but to evaluate the code's alternative lower-temperature / longer-time rows (where permitted) or the no-PWHT exemption route, and to document the decision in the WPS/PQR.

Soak time is tied not to some arbitrary thickness of the component but to the code-defined governing nominal thickness of the weld; that definition changes where different thicknesses meet and at corner and set-in joints. The general rule is 1 hour per 25 mm. ASME VIII Div. 1 UCS-56 and ASME B31.3 Table 331.1.1 set an absolute minimum soak time of 15 minutes — "1 hour minimum" is field habit, not code text. For heavy sections the steps differ between codes and must not be confused: ASME VIII Div. 1 Table UCS-56-1 gives 1 h/in. between 50 and 125 mm (2–5 in.), and above 125 mm (5 in.) 5 hours plus 15 minutes for each additional 25 mm over 5 in.; ASME B31.3, by contrast, applies 2 hours plus 15 minutes for each additional 25 mm above 50 mm.

For P-No. 15E (Grade 91), code tables typically show a 30-minute minimum — verify this minimum against the edition in your hands (the UCS-56-11 row) — but 30 minutes is not enough to complete tempering in a creep-strength-enhanced ferritic steel. EPRI and common project specifications call for a minimum soak of about 2 hours regardless of thickness on Grade 91; as thickness increases the code table already imposes longer times (for example B31.1 durations rise toward 5 hours for sections over 50 mm (2 in.)), and in all cases choose the time so that the target hardness band is achieved. Viewed through Larson–Miller, a PWHT built on the 30-minute code floor typically leaves hardness above the limit.

What matters is that the entire cross section of the component reaches the target temperature and completes the specified time there — not the surface temperature, but temperature equalized through the section. Keeping the time short means inadequate stress relief. Extending it excessively means overtempering, loss of strength and creep life, and carbide coarsening (M23C6) — grain growth is not the right fear here, because PWHT is a sub-Ac1 anneal and austenite grains do not grow. (Temper embrittlement, by contrast, comes from the cooling regime rather than the soak time; see Section 3.) Also, the cumulative total PWHT time, including repairs and reprocessing, shall not exceed the total time qualified in the PQR. Fix the parameters in the WPS/PQR and record them on the field temperature-time chart.

3. Heating and cooling rate limits

In PWHT the rate at which you reach and leave temperature matters as much as the temperature itself. A fast, irregular cycle creates new thermal gradients — that is, new residual stresses — in the component, and can cause distortion and cracking.

In ASME VIII Div. 1 UCS-56 the threshold is single and clear: 425 °C (800 °F). Below that temperature the rate is free; above it the formula applies, and heating and cooling are tied to different numbers. Set the formula up without mixing units — dividing a metric rate by a thickness in inches is the sneakiest calculation error in the field:

  • Heating: above 425 °C, no more than 400 °F/h ÷ thickness (in.); the metric equivalent is 5,600 °C·mm/h ÷ thickness (mm). The rate shall in no case exceed 222 °C/h (400 °F/h). If the calculation falls below 56 °C/h (100 °F/h), the rate need not be reduced below that value.
  • Cooling: in a closed furnace down to 425 °C, no more than 500 °F/h ÷ thickness (in.), metric equivalent 7,000 °C·mm/h ÷ thickness (mm); the ceiling is 278 °C/h (500 °F/h), and likewise there is no obligation to go below 56 °C/h. Below 425 °C, cooling in still air is permitted.

The thicker the part, the slower the permitted rate. The furnace temperature at the moment the vessel is charged shall also not exceed 425 °C; that condition is often skipped in the field. In addition, during the heating period the difference between the hottest and coldest point within any 4.6 m (15 ft) interval of the heated zone shall not exceed 140 °C (250 °F).

Cooling regime and temper embrittlement. In Cr-Mo steels, temper embrittlement does not arise from a long soak. It arises from segregation of P, Sn, Sb and As to grain boundaries during slow cooling through, or holding within, the 350–575 °C band. The result is an upward shift of the transition temperature — in other words, the brittle fracture margin evaporates in cold start-up and shutdown scenarios. That is why heavy-wall Cr-Mo equipment (the API 934 context) calls for cooling quickly enough through the 600–300 °C range, with the rate balanced against the gradient rules above. The indices that predict susceptibility from chemistry are composition indicators, not time indicators, and the two are not the same thing: the J-factor for parent metal, X-bar (X̄) for weld metal.

4. PWHT's own risk: reheat cracking

PWHT does not only solve problems; it can also generate its own damage mechanism. In the 2¼Cr-1Mo-V and Cr-Mo-V family, and in steels stabilized or microalloyed with V, Nb, Ti or B, fine carbides precipitate within the grains during heating in roughly the 500–700 °C band. As the grain interiors harden, deformation concentrates at the grain boundaries and intergranular reheat cracking develops. Standard 2¼Cr-1Mo (P22) is relatively resistant compared with the vanadium-bearing derivatives of this family; even so, it can show susceptibility in certain heats and heavy sections — so the right approach is to look at its chemistry and section thickness, not to treat P22 as "exempt." The crack almost always starts in the coarse-grained HAZ and at stress-concentrating geometries: a sharp weld toe, an unfused root, abrupt transitions, unground notches. Control routes: specify a susceptible material only knowingly, grind the weld toe and soften the geometry, limit heat input and interpass temperature, pass through the sensitive 500–700 °C band as quickly as possible without violating the section gradient limits (slowing down in this band is counterproductive, because it extends the period in which carbide precipitation and stress relaxation coincide), and check these areas with surface NDT (MT/PT) after PWHT.

5. Temperature control and thermocouples

PWHT is not blind heating; it is a documented control operation. The evidence is the temperature-time chart from a calibrated recorder, and without that chart the operation has not been performed. One reference mix-up needs correcting up front: the scope of EN ISO 13916 covers measurement of preheat temperature, interpass temperature and preheat maintenance temperature; measurement of PWHT temperature is outside the scope of that standard. For PWHT, the documents to reach for are: EN ISO 17663 for quality requirements for heat treatment related to welding, AWS D10.10 for local heating arrangements, AMS 2750 or the project specification for the furnace and thermocouple calibration system, and the construction code itself for recording provisions (ASME VIII Div. 1 UCS-56, ASME B31.3 para. 331).

  • Attach the thermocouples (usually Type K) directly to the component, most often by capacitor discharge welding. The measurement must represent the workpiece itself, not the heated surface.
  • Placement is critical: monitor the thickest section, the thinnest section, different points along the weld, and any region where a gradient is expected. In local PWHT, track the edges of the heater band (the control zone and the gradient zone) with separate thermocouples.
  • Keep the control thermocouple and the recording (monitor) thermocouple separate; the controller must see both the coldest and the hottest point. That way no region stays below or climbs above the target.
  • After the thermocouple attachments are removed, grind the contact points and — especially on P91 and in sour service — check them with surface NDT; capacitor discharge can leave a local hard spot.
  • Attach the instrument calibration certificates to the report; the inspector asks for these first.

In local PWHT, take the insulation width, the heated band and the gradient control band dimensions from the code and the applicable practice documents. In ASME B31.3, the local heat treatment provision is para. 331.1.6, and it defines the soak band as "the weld width plus 1×t or 25 mm on each side, whichever is less." AWS D10.10 recommends a different minimum (t or 50 mm on each side of the weld, whichever is less); using the two documents as if they gave the same number leaves the heated band too narrow. Clarify which document governs from the ITP and the project specification. Heat flows away from the joint faster than crews expect, so a narrow band and inadequate insulation are the most common root cause of the section center and the inside surface (ID/root) never reaching target temperature.

6. ASME VIII PWHT requirements

In ASME BPVC Section VIII Division 1, PWHT is governed mainly by UCS-56; current editions use not a single table but Tables UCS-56-1 through UCS-56-11, split by P-number, plus supporting paragraphs.

  • The minimum soak temperature and the minimum time per unit thickness are tabulated for each P-number and group.
  • Material group and weld thickness determine whether PWHT is mandatory — but thickness thresholds are not the only criterion: UW-2 imposes service-driven requirements, and these three cases must be kept apart. These requirements apply to carbon and low-alloy steels; they do not apply to austenitic stainless vessels.
    • Lethal service — UW-2(a) (and UCS-68(c)): in carbon and low-alloy steel, PWHT is mandatory for all welds regardless of thickness.
    • Unfired steam boilers — UW-2(c): the requirement applies to boilers with a design pressure above 50 psi (≈345 kPa / 3.4 bar); this threshold is frequently overlooked.
    • Direct-fired vessels — UW-2(d): for P-No. 1 carbon steel, PWHT is mandatory only when the weld thickness exceeds 16 mm (5/8 in.); for low-alloy steels other than P-No. 1 it is mandatory at all thicknesses. Treating direct firing as "thickness-independent" is a common but incorrect generalization.
  • The procedure (WPS) carries PWHT condition as an essential variable: procedures with and without PWHT are qualified separately (Section IX, QW-407; QW-406 for preheat). Applying or skipping PWHT without WPS approval invalidates the procedure.

Div. 2 and the B31 series for piping carry their own PWHT rules; clarify from the outset which code applies to a given component. The acceptance decision always belongs to the construction/reference code.

7. PWHT exemptions: when it is not required

Codes do not force every weld into PWHT; exemptions are recognized under specific conditions. An exemption is not "skipping it arbitrarily":

  • For P-No. 1 Gr. 1/2/3 carbon steels, the typical exemption threshold is a weld thickness ≤ 38 mm (1½ in.), conditional on at least 95 °C (200 °F) preheat above 32 mm (1¼ in.). These numbers can change from edition to edition; verify them against the table in force.
  • Some exemptions are tied to a preheat condition: instead of PWHT, cold cracking risk is managed with preheat immediately before and during welding.
  • Carbon equivalent (CE), the service environment (for example wet H₂S) and the need for notch toughness change the exemption decision. A service code can override an exemption recognized by the construction code — always apply the most restrictive requirement.

An exemption must come from the construction code and be documented in the WPS/PQR; verbal decisions such as "it's thin enough, we don't need it" will not survive an audit.

8. P91 and the narrow temperature window: the Ac1 risk

P91/T91 (9Cr-1Mo-V, Grade 91, P-No. 15E) and its 9–12Cr martensitic relatives such as P92 and E911 turn PWHT into a precision operation. This family is a separate group from bainitic Cr-Mo steels such as 2¼Cr-1Mo (P22, P-No. 5A); managing both with the same recipe is a mistake. The target microstructure is tempered martensite.

Cooling before PWHT — the rule most often understood backwards. Good practice on P91 is to cool the weld in a controlled manner from preheat/interpass temperature (200–300 °C) below Mf (~100 °C; in practice aim for ≤ 90 °C) and hold it there before going up to PWHT; the purpose is to complete the austenite → martensite transformation (on P91, Ms ≈ 400 °C and Mf ≈ 100 °C, both varying with grain size and chemistry). If you go straight from interpass temperature to PWHT, untransformed retained austenite is never tempered and converts to fresh, untempered martensite as the part cools after PWHT — exactly the damage you were trying to avoid. This rule is conditional, not absolute: classical practice and the first edition of the EPRI guideline require cooling below Mf, while the second edition (2015) of the EPRI good practice guideline recognizes an exception — if the filler metal has Mn+Ni ≤ 1.2% and the preheat temperature is below 204 °C (400 °F), cooling to the preheat temperature can be accepted as sufficient. The requirement to go below Mf is binding where the chemistry is high and retained austenite is a real risk. The project specification and the WPS decide which applies. If hydrogen cracking is a concern during this hold, the answer is not to park the part at an intermediate temperature but to apply DHT (hydrogen bake-out) after cooling below Mf; on P91 this band is chosen in the upper half of the general 200–350 °C range, at 250–350 °C, because the goal is not only hydrogen diffusing out but also completing that diffusion in a heavy section within a reasonable time. Do not leave the part waiting without PWHT for a long period either.

The temperature window. The code band for ASME P-No. 15E is 732–774 °C (1350–1425 °F); ASME B31.1 gives the same 1350–1425 °F band for Grade 91, so there is no real band difference between the two codes. (That is why rounded values such as "730–775 °C" should not be written — rounding up a ceiling that sits next to Ac1 legitimizes going outside the code.) In practice, procedures narrow this and target 750–765 °C. Every P91 temperature quoted in this article follows that same reference band.

  • Lower limit: if the temperature is low, the martensite is not tempered enough; hardness stays high and toughness stays low.
  • Upper limit — the Ac1 hazard: if the temperature approaches or exceeds the lower critical transformation temperature Ac1, the material partially re-austenitizes and fresh martensite forms on cooling. Ac1 is typically ~820–830 °C in the parent metal; the real risk is in the weld metal: every 1% of Ni+Mn lowers Ac1 by roughly 20–40 °C, and in weld metal Ac1 can drop to ≈785 °C (about 792 °C has been measured in weld metal with 1.2% Ni+Mn at slow field heating rates).
  • Get the chemistry ↔ ceiling relationship the right way round. A common error is to treat low Mn+Ni as a restriction; in fact low Mn+Ni raises the ceiling. The steps in the B31.1 note for P-No. 15E are: base ceiling 774 °C (1425 °F); where the filler metal has Mn+Ni ≥ 1.0% and < 1.5%, the ceiling may be raised to 788 °C (1450 °F), and where Mn+Ni < 1.0%, to 799 °C (1470 °F). This permission depends on the provision in the applicable code and edition and is not automatic. Do not mix up where the numbers come from either: 1.5% is the Mn+Ni chemistry ceiling for SFA-5.5 B9 filler metal; 1.2% is the EPRI threshold for the pre-PWHT cooling exception (not a temperature-ceiling threshold); 1.0% is EPRI's established target. Values above 1.5% also pull Mf down and create a retained austenite risk.
  • Which ceiling is binding? On work under ASME VIII / B31.1, the ceiling of the soak band is 774 °C, and where the code note grants no explicit chemistry-based step, no application exceeds that ceiling. The 788 °C and 799 °C values are increases the code note recognizes as a function of chemistry; they are not a blanket permission. With no chemistry record on file, the governing ceiling is 774 °C.
  • What happens if Ac1 is exceeded: this damage is not repaired by another PWHT. A partially re-austenitized region can only be corrected by full renormalizing (~1040–1080 °C) plus tempering, or by cutting out and replacing the weld. (The case in the field note below is different: there the problem was low temperature and the wrong measurement point, not an Ac1 excursion.)

Hardness is controlled in both directions. The commonly accepted band on P91 is roughly 190–250 HV10, and it is limited at both ends: the upper limit catches inadequate tempering, the lower limit catches overtempering or the wrong material; an excessively soft weld loses creep strength. Some specification families write ≤ 265 HB instead. The two are not alternatives and do not give the same acceptance level — 265 HB corresponds to roughly 280 HV10, clearly looser than the top of the HV10 band. A scale conversion is not a substitute for acceptance criteria: measure on whatever scale the specification names (HV10 if HV10 is required), and a result above 250 HV10 cannot be accepted on the grounds that it "passes 265 HB." Check which scale is made binding in the project specification, and do not confuse these limits with the NACE sour service criteria of 250 HV10 / 22 HRC.

Type IV creep cracking. Explaining P91 only through hardness and Ac1 leaves the picture incomplete: most service damage appears as Type IV creep cracking in the fine-grained / intercritical region of the HAZ. This narrow strip has lower creep strength than both the parent metal and the weld metal; high heat input widens the region, and inadequate or borderline PWHT weakens the microstructure further. Heat input control, interpass temperature discipline and correct PWHT parameters directly determine Type IV life.

9. Common field mistakes

  • Placing the thermocouple near the heater band instead of on the workpiece, and missing the real metal temperature.
  • Keeping the time short relative to thickness, so the center of the section never reaches target temperature.
  • Using some arbitrary rate threshold instead of 425 °C, assuming the heating and cooling formulas are the same, or dividing a metric rate by a thickness in inches.
  • Running P-No. 4 at the P-No. 1/3 minimum (593 °C); the code minimum is 650 °C.
  • Confusing the heavy-section time steps of ASME VIII and B31.3.
  • Leaving the insulation and heated band width inadequate, so the inside surface never heats up in a local PWHT; assuming the B31.3 and AWS D10.10 band dimensions are the same.
  • Taking P91 straight to soak temperature without cooling below Mf beforehand where the specification requires it.
  • Working with a single "approximate" temperature on P91, entering the Ac1 range and forming fresh martensite; going above the 774 °C ceiling without a chemistry record.
  • Settling for the 30-minute code minimum on Grade 91 and missing the hardness band.
  • Exceeding the parent metal tempering temperature on Q&T material.
  • Using the HV10 and HB hardness limits interchangeably.
  • Calling the operation "complete" without a calibration certificate and a temperature-time chart.

Related standards

  • ASME BPVC Section VIII, Div. 1 — UCS-56 and Tables UCS-56-1 through UCS-56-11; UW-2 service-driven requirements (lethal service, unfired steam boiler, direct firing). This article is based on the 2023 Edition; the edition in force is 2025.
  • ASME BPVC Section IX — procedure qualification; QW-406 (preheat), QW-407 (PWHT essential variable). Verify the subparagraph number of the special provisions for P-No. 15E against the edition in your hands.
  • ASME B31.1 / B31.3 — PWHT bands and time tables for power and process piping (B31.3 Table 331.1.1), local heat treatment and soak band definition (B31.3 para. 331.1.6); in B31.1, the 1350–1425 °F band for Grade 91 and the Mn+Ni-dependent ceiling steps.
  • EN 13445-4 — PWHT requirements for EU pressure vessel manufacture.
  • EN ISO 15609-1 — documenting heat treatment parameters in the WPS.
  • EN ISO 15614-1 — welding procedure qualification testing (WPQR).
  • EN ISO 17663 — quality requirements for heat treatment related to welding; this is the measurement and recording reference for PWHT.
  • EN ISO 13916 — measurement of preheat temperature, interpass temperature and preheat maintenance temperature only; PWHT temperature measurement is outside its scope.
  • AMS 2750 — furnace temperature uniformity and thermocouple/recording system calibration (where the project specification invokes it).
  • EN 1011-2 (Annex C, Methods A/B) — preheat temperature calculation; not a DHT/bake-out provision.
  • AWS D10.10 — guide for local heating heat treatment (band widths, insulation); band dimensions may differ from B31.3.
  • API 582 / API 934 — process industry welding guidelines (including DHT) and heavy-wall Cr-Mo equipment requirements (temper embrittlement, cooling regime).
  • AWS D1.1 — stress relief provisions for steel structures; D1.1 does not make PWHT generally mandatory and contains no general hydrogen removal heat treatment clause; the clause number and time table vary by edition, so verify against the edition in your hands.
  • NACE MR0175 / ISO 15156-2 — sour service (production) hardness limits. NACE MR0103 / ISO 17945 — refinery wet H₂S service.

From the field

On a thermal power plant project, the site crew ran PWHT on P91 main steam line welds as "around 740 degrees, a few hours," with thermocouples attached only to the top surface of the weld. Through-section hardness testing came back at 270+ HV10 in the HAZ, against a project specification limit of 250 HV10. The investigation pointed to two root causes: the heated band width and the insulation were not enough to bring the inside surface of a heavy section up to target temperature, and the measurement was being taken only from the hottest surface. The surface was reading correctly while the center of the section was not tempering. The decision: rework the band width and insulation to the code tables, repeat the local PWHT with three thermocouples (surface, mid-wall, root side), and work within the procedure's narrowed 750–765 °C band for the verified minimum 2-hour soak the specification called for. The rework was accepted. Here the problem was low temperature; had Ac1 been exceeded, another PWHT would have fixed nothing and the joint would have needed renormalizing plus tempering, or replacement. The lesson is plain: in PWHT, if the point you measure is wrong, the metallurgy is wrong no matter how good the chart looks. Read the temperature at the coldest point of the workpiece, and on P91 never trust a single point.

Disclaimer

This article is for educational purposes and does not replace the official standard, the code, or the written procedure. Always work from the current edition of the applicable code or standard, and from the project specification — the final decision rests with the code in force and the responsible inspector.

Take it to the field: To keep the welding standards, procedure logic and field control steps in this article in your pocket — completely offline and free — take a look at the Doawise Welding Eng Guide app.

At DoaWise we carry out welding engineering and inspection services to international standards, documenting every step — from preheat to PWHT parameters, from thermocouple placement to hardness verification — with recordable, auditable results. The right temperature, the right time and the right measurement point are the three conditions that make PWHT worthwhile, and we secure all three in the field.