Heat Treatment After Welding: A Complete PWHT Guide

The weld passed dye penetrant, the inspector signed the traveler, and the customer still called six weeks later: the vessel cracked right at the weld during hydrotest. Nothing was wrong with the weld metal itself — it was the heat around it. That gap, between a sound weld and a sound structure, is exactly what post-weld heat treatment closes.

Why a Weld Needs Heat Treatment After It Cools

The weld passed dye penetrant. The inspector signed the traveler. Six weeks later the customer called: the vessel cracked right at the weld during hydrotest.

Nothing was wrong with the weld metal itself. The bead was clean, penetration was sound. What failed was the heat around it, because welding is not merely joining: it is a metallurgical event that happens to include melting. That gap between a sound weld and a sound structure is what post-weld heat treatment (PWHT) closes.

The Thermal Cycle: Steel Heated, Melted, and Quenched by Its Own Heat

As the arc travels the joint, the base metal heats toward its melting point, the filler melts into it, and the weld metal freezes. Steel ahead of the arc expands; behind it, it contracts. In between, a narrow band of parent metal never melted but changed anyway.

That band is the heat-affected zone (HAZ): base metal altered by welding heat but not melted. On carbon steel it might run 5 to 30 mm wide depending on heat input, thickness, and travel speed, with a fine-grained zone next to the fusion line and coarser grain further out.

The lower critical temperature (Ac1) for carbon steel, where austenite begins to form on heating, sits around 1,330 to 1,370 °F (720 to 740 °C). A welding arc spikes local metal well above that, and HAZ cooling rates can run on the order of 100 to 1,000 °C per second, far faster than steel can reach equilibrium. The weld’s own quench locks in a non-equilibrium condition.

Untempered Martensite and the Hardness Problem

In grades that can harden, the trouble concentrates in the HAZ. Cooled fast enough from austenite, carbon steel forms untempered martensite, a supersaturated, body-centered tetragonal structure that is extremely hard and brittle. An as-welded HAZ in a hardenable grade can measure well above 300 HBW locally. No radiograph flags that, because it is a metallurgical condition in the parent metal itself.

There is a second, quieter problem in the same zone. Martensitic transformation causes local volume expansion, and in a rigid HAZ that expansion is accommodated by plastic deformation of nearby grains. Under enough restraint it generates localized micro-cracking, so fine and closed that dye penetrant may not detect them. The joint looks flawless and still carries a crack that grows under service loading.

Residual Stress: Contraction Into a Cage

Every weld has shrinkage. The molten pool solidifies and contracts while the surrounding cold metal holds it in place. The weld wants to shorten, and something is holding it.

In a lightly restrained joint the assembly pulls in slightly and stresses stay modest. In a highly restrained joint, a nozzle neck, a heavy-to-thin wall joint, a closure ring, or a fillet-welded attachment to a rigid structure, that restraint builds. The peak tensile residual stress in the weld and HAZ can approach the yield strength of the base metal, and it stays there permanently, stacked on whatever service loads arrive later.

The Four Problem Families PWHT Exists to Fix

Physically, PWHT addresses four separate families of damage, and a good cycle often addresses more than one at a time.

Problem family	What it is	What it leads to
Residual stress	Stresses locked in by welding shrinkage and restraint	Brittle fracture risk, reduced fatigue life, cracking at restraint points
Microstructure and hardness	Untempered martensite, coarse grains, HAZ micro-cracking	Low ductility and toughness, crack initiation sites
Hydrogen	Diffusible hydrogen from moisture, coatings, rust, or contaminated consumables and gas	Delayed hydrogen cracking, also called underbead cracking
Distortion	Weld shrinkage and angular distortion of the assembly	Out-of-tolerance dimensions, misaligned bores and flanges, extra stress raisers

Why “The Weld Is Sound” Is Not the Same as “The Assembly Is Fit”

Weld quality and structural fitness are different questions. Nondestructive examination answers only one: is there a discontinuity in the fused metal, and is it acceptable? It says nothing about hardened base metal, residual stress, or hydrogen still trapped in the HAZ.

That is why the HAZ, not the weld metal, is usually the weakest link. Weld metal is deliberately chosen to be tougher than the base metal it joins, with extra alloying and, in most classifications, a deliberate yield strength overmatch. The HAZ gets no such benefit: it is base metal accidentally heat-treated by a moving arc. A better bead cannot fix that. If the HAZ is hard and stressed, the crack simply initiates there, at or just outside the fusion line.

The Core Idea: A Controlled, Repeatable Thermal Cycle

PWHT is simpler than its reputation suggests. Heat the finished weldment slowly and uniformly to a specified temperature, hold it there long enough, and cool it under control. That single cycle lets the steel recover and recrystallize so residual stresses redistribute, gives diffusible hydrogen time to escape, and tempers untempered martensite back into something ductile.

The word that matters most is controlled. The heat must reach the entire heat-affected region, not just the surface near a burner, and it must be measured on the part, not on the furnace wall. Results are documented so a third party can verify them, which is why PWHT runs on written procedures and recorded charts rather than shop-floor judgment.

Set the expectation early: whether a weldment needs PWHT, and to what temperature, is an engineering decision, not a shop decision. The design authority, the code, the contract documents, and the welding procedure specification make that call. The shop’s job is to execute it faithfully and prove it. The air gas welding fundamentals and the arc itself are only the beginning. The rest is metallurgy, and metallurgy is where failures come from.

What PWHT Actually Does — and What It Cannot Fix

Post-weld heat treatment has a mysterious reputation: the weldment goes into a furnace, comes out “relaxed,” and the stress is supposedly gone. Pull the curtain back, because PWHT is not a cure-all. It does three specific metallurgical things to the metal around a weld, and most shops achieve all three with one temperature window, one soak time and a controlled cool. Knowing what it does and what it leaves untouched keeps expectations honest.

Mechanism 1: Stress Relief

What people call “stress relief” is three overlapping processes. At elevated temperature the metal recovers: the dislocation structure rearranges and part of the stored elastic energy is released. Higher in the range, recrystallization replaces strained, elongated grains with new, softer, equiaxed ones. Across the cycle the part expands and contracts more uniformly than during welding, softening the gradients that locked stress in.

The payoff is measurable. In high-restraint welded joints, a proper stress-relief cycle has been reported to cut peak tensile residual stress by roughly 30 to 80 percent, with the biggest gains at the weld toe where the peak lives. That is precisely where a service crack likes to begin.

Mechanism 2: Hydrogen Removal

The second job is bake-out. Diffusible hydrogen gets into a weld from many sources: moisture in the shielding gas, rust and mill scale on the joint, wet or contaminated consumables, paint, cutting oil, even moisture in the base metal. Held at soak temperature long enough, that hydrogen diffuses out of the heat-affected zone and escapes.

Timing is why bake-out is scheduled rather than reactive. Hydrogen-related cracking, the classic cold or delayed crack, does not show up while the torch is hot. It can surface days later, once hydrogen concentrates at stress concentrations and the material cools into a brittle condition. NACE documents how hydrogen cracking and weld metal hardness interact, so bake-out is a defined hold in the procedure, not a reaction to a visible crack.

Mechanism 3: Tempering the Hardenable HAZ

Hardenable grades, including quenched-and-tempered steels, martensitic stainless and the 9Cr-1Mo-V creep steels used in power plant, can self-quench in the HAZ. The result is untempered martensite: very hard, very brittle and crack-happy. PWHT tempers that martensite into a tougher condition and pulls peak hardness back toward the base metal.

In 9Cr-1Mo-V Grade 91 pipe, as-welded HAZ metal commonly measures 350 to 420 HV, while the limit under ASME pressure piping codes is typically 250 HB. The ASME weld hardness criteria are why PWHT is not optional on those grades. The same tempering logic applies to martensitic stainless, and consumable selection should be settled before the heat treatment plan is fixed. If you are choosing electrodes for alloy steel, that decision and the planned temper window need to agree.

Here is the surprise: a fabricator normally runs one temperature window to achieve all three effects. A carbon-steel stress-relief cycle in the neighborhood of 1,100 to 1,200 °F per AWS D1.1 is hot enough to bake out hydrogen and temper hardenable HAZ metal, and long enough to relieve stress. When a grade needs something different, the code or the material data sheet says so and the window moves.

What PWHT Cannot Fix

• Weld defects. Porosity, undercut, lack of fusion and cracks are mechanical problems. PWHT is a thermal cycle, not a repair method: it will not close a crack, and it can even redistribute stresses around a flaw. Fixing a defect means grinding, gouging or re-welding, and air arc welding and carbon arc gouging is the usual shop answer.
• Out-of-tolerance geometry. PWHT relieves stresses, which often means parts move. It does not pull a distorted weldment back to print. Corrections still have to be machined or ground after treatment, and the code may require re-verification of mechanical properties on reworked areas.
• Strength already lost above the tempering temperature. Heat a quenched-and-tempered or martensitic stainless part above its tempering temperature and hardness and strength drop permanently. No repeated cycle brings them back, which is why AWS treats stress relief of these grades as a specialist decision.
• Missing preheat. PWHT cannot replace proper preheat on martensitic grades. Preheat controls hydrogen and cracking during the weld itself; heat treatment only addresses what happened after the metal cooled.

One more honest caveat: PWHT is not free metallurgically. It marginally reduces yield strength in the treated material and can slightly soften weld metal deliberately over-matched to the base. EPRI’s technical basis for minimizing PWHT sets out the trade-offs behind reducing or eliminating the operation. Sometimes that slight softening is welcome; sometimes the design has already spent its strength margin.

Problem	PWHT help?
Residual stress around the weld	Yes, the main purpose
Diffusible hydrogen and delayed cracking	Yes, if baked out in time
Hard, untempered HAZ metal	Yes, it tempers the martensite
Porosity, undercut, existing cracks	No, and it cannot re-heal them
Distortion and strength lost above tempering temperature	No, the loss is permanent

Temperature, Soak Time and Heating Rates: The PWHT Recipe

Every heat treatment after welding comes down to four numbers and two conditions. The numbers are target temperature, soak time, heating rate and cooling rate. The conditions are furnace uniformity and how accurately you measure the part instead of the air around it.

Target temperature: the decision that drives everything else

For carbon steel, AWS D1.1 Table 5.2 puts stress relieving at 1,100 to 1,200 °F (about 593 to 649 °C). Below that band you get partial recovery and little stress reduction; above it you start eating into the material’s tempering margin.

The hard ceiling is the lower critical temperature, Ac1, roughly 1,330 °F for carbon steel. Above it you re-austenitize the steel and get a fresh, uncontrolled transformation in the heat-affected zone on the way down, the opposite of what you paid for.

Alloy steels push the window higher, and ASME B31.3 Table 331.1.1 shows it clearly:

• P-No. 1 (carbon steel): 1,100 to 1,200 °F (595 to 650 °C)
• P-No. 4: 1,200 to 1,300 °F
• P-No. 5A and 5B: 1,250 to 1,400 °F
• P-No. 15E (Grade 91): 1,300 to 1,425 °F with a 30-minute minimum soak

Thermocouple selection stops being trivial at those temperatures. Type K covers the common ranges, but anything approaching 1,400 °F wants Type N or a higher-rated sheathed element. Owner and plant standards tend to restate code requirements rather than invent new ones, as the Los Alamos National Laboratory general welding standard does when it folds D1.1 into its own post-weld heat treatment clauses.

Soak time: measured in inches of control thickness

Soak time is not per inch of nominal wall. It is calculated on control thickness, the thickest welded section measured at the weld. A 3/8-inch pipe with a 1-inch flange or socket weld at the end is a one-inch control thickness job, and the treatment time follows the one inch.

ASME B31.3 Table 331.1.1 sets the rule for P-No. 1 carbon steel at one hour per inch of control thickness, with a 15-minute minimum. Above 2 inches, the soak becomes 2 hours plus 15 minutes for each additional inch. The Grade 91 entry carries its own 30-minute floor.

Then the detail that causes more argument than any other number in the standard: the soak clock starts when the part reaches temperature, not when the furnace does. A 4-inch plate trails the furnace indication badly. Count from the furnace and the part gets short-changed by exactly the lag you ignored.

AWS D1.1 recognizes the full Table 5.2 range is not always practical, so Table 5.3D permits an alternate stress-relieving treatment at a lower temperature held longer. It is an engineered choice, not a shop shortcut, and it must appear in the drawing, the specification and the WPS before work starts.

Heating rate: the 400 °F per hour rule

AWS D1.1 sets two limits on the way up. Furnace air temperature at loading cannot exceed 600 °F. Above 600 °F, the heating rate is capped at 400 °F per hour divided by the maximum thickness in inches, and never more than 400 °F per hour regardless.

Do the arithmetic: a 1-inch part may climb at 400 °F per hour, a 2-inch part at 200 °F per hour, a 4-inch part at 100 °F per hour. Slow heating keeps the gradient between the hot weld zone and the cold surrounding metal small, which is the difference between a stress-relieved assembly and one that distorts or cracks while you relieve it. Published work maps how thermal parameters drive residual stress and distortion.

Cooling rate: the last free variable

AWS D1.1 permits air cooling from 600 °F downward once the soak is complete. Higher-strength and quenched-and-tempered grades differ: ASME requires controlled cooling for the P-3, P-4 and P-5 groups, because an uncontrolled quench through the transformation range can re-harden the very heat-affected zone you just paid to soften.

Furnace uniformity and measurement lag

These two conditions quietly sink treatments. Industry practice and most owner specifications call for a furnace uniform to about ±25 °F with no more than a 50 °F spread between any two points in the heated zone, verified by a current uniformity survey. A furnace that cannot hold that spread cannot deliver the code temperature.

Measurement lag is the other one. Furnace air tells you what the burner is doing; the part tells you what the steel is doing. Put thermocouples on the part itself, at or near the weld and heat-affected zone, never on the furnace wall, plus at least one independent furnace thermocouple for control. Without a part thermocouple you are measuring the oven, not the treatment.

The recipe in plain language

Parameter	AWS D1.1 carbon steel	ASME B31.3 P-No. 1	Shop reality
Soak time	Per spec; Table 5.3D alternate	1 hr per inch of control thickness, 15 min min; over 2 in, 2 hr + 15 min per inch	Control thickness is the thickest welded section, not nominal wall
Heating rate	Above 600 °F: 400 °F/hr ÷ max thickness in inches, cap 400 °F/hr	Controlled ramp per code	Slow ramps limit gradient, distortion and cracking
Cooling	Air cool from 600 °F downward	Controlled cooling for P-3, P-4, P-5	Stops re-hardening of the HAZ

Where charts are reviewed line by line and tolerance is measured in thousandths, the discipline tightens further, so the habits in our aerospace welding standards and careers guide are worth studying even if you never weld for an airframer.

Who Actually Requires Heat Treatment After Welding

Ask five fabricators “Does this job need PWHT?” and you get five answers: the code requires it, the drawing calls for it, the customer forbade it, the material will not tolerate it, or “we do what the inspector says.” The order of authority is a chain, not a matter of opinion.

The Hierarchy of Requirements

Strongest first:

1. Contract and procurement documents — purchase specifications, data sheets, owner specifications.
2. Design engineer of record — the licensed engineer who stamped the design and owns fitness for service.
3. Governing code — AWS D1.1, ASME VIII, B31.3.
4. Welding procedure specification — the WPS, which turns code and engineer instruction into an executable heat treatment cycle.
5. Welder or shop preference — which counts for nothing on its own.

Most disputes come from treating items 3 and 5 as equals. A code is a floor, a minimum set of rules a jurisdiction enforces; a contract can demand far more, and then the contract wins. The design authority can always exceed the code, because the code prevents known failure modes rather than certifying a structure for service.

Structural Steel: AWS D1.1

For most structural steel, AWS D1.1 does not mandate PWHT. Clause 5.8 makes it instruction-based: the engineer of record specifies the treatment in the contract documents, and the shop performs it under a qualified WPS. Tables 5.2 and 5.3 give the stress-relieving temperature ranges and the alternate lower-temperature, longer-time treatment.

The same clause warns that stress relief is not recommended for quenched-and-tempered grades such as ASTM A514 (T-1), A517, A709 Grade HPS 100W and A710. Heating those steels into a stress-relief window can undo the temper and leave the joint harder and more crack-prone.

Pressure Equipment: ASME Section VIII and the B31 Codes

ASME Section VIII Div 1 UG-29 covers PWHT for pressure vessels, and Div 2 carries its own alternative-rules requirements, including the Part 4 and Part 5 approaches many fabricators use for high-strength or low-temperature service.

ASME Section IX is the one people misunderstand most. It does not tell you whether to heat treat, it tells you how to qualify. PWHT is an essential variable in the WPS, so a qualification that did not cover the heat treatment range does not cover the job.

• ASME B31.1 — Power Piping, requiring thermal stress relief of carbon steel welds under defined conditions.
• ASME B31.3 — Process Piping, with holding temperatures in Table 331.1.1 and exemptions in Table 331.1.3 for certain thicknesses and weld types.
• ASME B31.4 — Pipeline Transportation Systems for Liquids, covering heavier, higher-pressure transmission service.
• ASME B31.8 — Pipeline Transportation Systems for Gas, calling for stress relief of welds in all carbon steel when nominal wall exceeds 1-1/4 inch, and for carbon steel above 0.32% carbon or a carbon equivalent (C + 1/4 Mn) above 0.65%.

That trigger is a chemistry-and-thickness test, not a service test. A thick, ordinary-carbon line pipe can be mandatory PWHT work because it crosses 1-1/4 inch nominal wall.

Tanks, Pipelines and Repair Work

API 650 governs welded storage tanks, setting heat treatment requirements for shell and nozzle welds based on material, thickness and service. API 1104 covers field welding of pipeline piping and requires stress relief of grade B carbon steel and higher in the specified thickness range. Pipeline work usually follows the operator’s own specification over the code, which often adds requirements.

For repairs and alterations, PCC-2 is what shops meet when a code does not cover the repair adequately, covering repair welding of ASME-listed equipment and pressure vessels and the associated heat treatment requirements. Internationally, ISO 15614 and ISO 15617 handle WPS qualification and production weld testing, so PWHT appears there as a defined qualification range, while European pressure equipment runs on EN 13480 for unfired vessels and EN 13445.

What Actually Triggers PWHT

Codes set minimums, but the reasons for PWHT are consistent across all of them:

• High restraint — a joint welded into rigid structure cannot contract freely, so residual stress stays high.
• Low-temperature service and brittle fracture risk — notch toughness matters more than strength.
• Sour service — NACE MR0175 / ISO 15156 hardness limits mean the HAZ must not be left hard; PWHT is a primary tool.
• Creep and fatigue service — sustained loads at temperature, and cyclic loading at a known stress concentration.
• Thick sections — more restraint, more hydrogen, deeper HAZ.
• Dissimilar joints — different materials need different temperature windows.

The rule is settled: the higher stress-relieving requirement governs. LANL’s welding standard GWS 1-08 states this explicitly for dissimilar joints.

How to Find Out What Your Job Requires

Before the steel is cut, read in this order: the purchase specification and any owner or data sheet addenda, the general notes on the drawing, the code the project is built to, and the WPS. If they disagree, ask in writing before you weld. At traveler review that answer costs an email; after the weldment is in the oven it costs a rejected load.

For anyone heating exotic assemblies: heat treatment windows differ by material, and a carbon steel stress-relief cycle can be ruinous to stainless, aluminium or nickel alloys. If the weldment mixes materials, the answer is not “the highest temperature wins” but a separate metallurgical decision, as the alloy steel welding electrode guide shows how wide those windows run.

Methods of Applying Heat After Welding

With fabrications differing in size, location, and material, there is no single way to apply heat after welding. The correct method is the one that can prove the weldment reached its required temperature and meets the governing code. Each option carries its own tradeoffs, so the choice must always favor compliance over convenience.

Full Furnace PWHT: The Reference Standard

For carbon and low-alloy steels, placing the entire assembly into a controlled electric or gas-fired furnace is considered the gold standard for post-weld heat treatment. With heat distributed to every surface at the same time, a furnace allows for tightly controlled heating and cooling rates with far fewer hot or cold spots. This level of uniformity is why ASME codes lean towards furnace cycles when they are practical.

Furnace PWHT also lends itself to the most defensible documentation. Uniformity surveys, calibrated thermocouples, and chart recorders can all be validated together to produce a straightforward time-temperature chart. The only real drawback is mobility. While smaller assemblies fit with ease, moving a multi-ton weldment to a furnace is often unrealistic for field repairs, which is what drives the need for approved alternatives.

Local or Spot PWHT

When a full furnace cycle isn’t feasible, local or spot PWHT can be used to heat only a calculated band of base metal around the weld. This is most often done with resistance coil heaters, induction coils, radiant heaters, or flexible ceramic and electric heating blankets. For many field jobs, fabricators source these consumables and instrumentation from suppliers such as Airgas Welding Supplies.

Local heating is well-suited for small repairs, nozzle attachments, or branch connections, but it creates a steep thermal gradient. That gradient increases the risk of distortion and means uniformity will never match a furnace. Most importantly, local PWHT relieves only what the heater can reach. For this reason, it must be approved by the design authority, use thermocouples on the weld and adjacent base metal, and produce a recorded chart to prove the soak band met the required temperature.

Self-Radiation and Flame Methods

In theory, the residual heat from welding (self-radiation) or limited fuel-gas flame heating could be used to apply a post-weld cycle. In practice, open-flame heating is nearly impossible to control across a surface, carries a high risk of overheating pockets of metal, and rarely produces reliable, repeatable chart evidence. For these reasons, flame heating is only ever considered under an extremely specialized, written, and approved procedure. For most pressure-retaining applications, it simply isn’t a viable option.

In-Process Tempering Plus a Post-Weld Temper Cycle

Not all tempering happens after the final pass. On select hardenable grades, a fabricator can control interpass temperatures for in-process tempering, then follow with a dedicated post-weld temper cycle. This is a metallurgical strategy, not a universal replacement for stress relief. Its success depends heavily on the welder’s control, uniformity can vary between passes, and validating the cycle with chart evidence requires full instrumentation under strict supervision. Because of this, it’s only used for very specific alloys and qualified procedures.

Solution Annealing: A Distinct Operation From Stress Relief

It is critical to never confuse stress relief with solution annealing. Austenitic stainless steels are normally solution annealed at roughly 1,020–1,100 °C (1,870–2,010 °F) followed by rapid cooling to restore corrosion resistance. Duplex stainless steels require their own precise solution annealing window to prevent the formation of brittle intermetallic phases. Both are almost always furnace-based with atmosphere control, have minimal tolerance for error, are ill-suited for unapproved local heating, and are designed to restore corrosion resistance, not solely to relieve residual stresses.

Comparing the Options at a Glance

Method	Temp. Control	Uniformity	Chart Production	Distortion Risk	Field Suitability	Relative Cost
Full Furnace PWHT	Excellent	Excellent (entire assembly)	Straightforward, highly defensible	Lowest when supported correctly	Poor (limited by furnace size)	Highest
Local/Spot PWHT (Blankets, Coils)	Good, with diligent TC placement	Moderate (band only)	Possible with strict monitoring	Moderate-High (thermal gradients)	Excellent, with engineer approval	Mid
Self-Radiation/Flame	Poor, very difficult to verify	Very poor	Often not feasible without a qualified procedure	High	Extremely limited, rarely approved	Low labor, high compliance risk
In-Process + Post-Weld Temper	Variable, procedure dependent	Variable between passes	Difficult, needs full instrumentation	Variable	Very limited, only for qualified hardenable grades	Mid-High (engineering heavy)
Solution Annealing (SS/Duplex)	Excellent, with tight setpoints	Excellent (furnace required)	Mandatory, often with cooling rate records	Low if fixtured correctly, but rapid cooling can induce distortion	Poor for field work. Material-specific, not a general stress relief	High

Ultimately, when the assembly allows, furnace PWHT is the most defensible option. For field repairs, an approved local treatment can be valid, but its thermal gradients demand extra care. Flame heating, unqualified self-radiation, or unqualified in-process cycles simply don’t offer the repeatability required for most pressure-retaining applications. No matter the route, the part must meet the specified temperature, not the heating source, with calibrated instruments, thermocouples on the weld and adjacent base metal, and a recorded time-temperature chart as proof of compliance.

Materials You Should Not Heat Treat Like Carbon Steel

Copying a carbon steel post-weld heat treatment (PWHT) cycle to other alloys is one of the quickest ways to ruin a weldment. While carbon steel has a well-established relief window, many other materials have much tighter tolerances. Going outside of them can lead to lost corrosion resistance, embrittlement, or a permanent drop in mechanical properties.

Austenitic Stainless Steel: The Sensitization Trap

For grades such as 304/304L and 316/316L, the greatest threat is sensitization. Exposure between 800–1,600 °F (425–870 °C) allows chromium carbides to precipitate along grain boundaries, most notably in the heat-affected zone. This window directly overlaps with the typical carbon steel stress relief range of 1,100–1,200 °F.

The proper fix is often a solution anneal at 1,900–2,000 °F (1,040–1,100 °C) followed by a rapid quench to restore corrosion resistance. In rare cases, a code-approved low-temperature stress relief below the sensitization range may be considered. However, as noted by AWS, time and temperature are both factors, so a long soak at an assumed “safe” temperature is not guaranteed to be safe.

Duplex Stainless: Protecting the Phase Balance

Duplex grades like 2205 and Super Duplex rely on a roughly 50/50 ferrite-austenite balance to achieve their strength and corrosion resistance. This balance is only held by a solution anneal around 1,870–2,010 °F (1,020–1,100 °C) with a fast quench.

A carbon steel style cycle at 1,100–1,200 °F can cause sigma phase to precipitate in the ferrite, leading to severe embrittlement. This risk is well documented in published research on duplex phase behavior. Because of this, requirements from standards such as NORSOK M-630 and applicable ASME code cases must be followed in place of a generic PWHT.

Martensitic Stainless: Temper, Don’t Stress Relief

Grades such as 410, 420, and 17-4 PH are designed to be heat treated, not stress relieved. For 410 and 420, a preheat followed by welding and tempering between roughly 1,000–1,200 °F is used to bring the hard HAZ back to a usable condition.

17-4 PH is even more particular. When in conditions like H900, H1025, or H1150, a double temper is typically required. The first handles untempered martensite in the HAZ, while the second restores the exact hardness and toughness called for. These are precision cycles, so the material supplier’s data sheet is mandatory, not optional. For those working in stricter environments, our aerospace welding overview expands on these requirements.

Quenched and Tempered Steels: Don’t Undo What You Bought

Some quenched and tempered steels are explicitly not candidates for stress relief. AWS D1.1 calls out grades such as ASTM A514 (T-1), A517, A709 Grade HPS 100W, and A710. A PWHT cycle will simply temper the already-hardened weld HAZ, undoing the strength and toughness that the buyer specified.

For other high-hardness alloys, such as 4340, the same principle applies. If stress relief is unavoidable, a full re-quench and re-temper is often required to regain properties. Proper consumable selection is critical for these grades, which is covered further in our guide to alloy steel welding electrodes.

Aluminium, Copper, and Age-Hardenable Nickel Alloys

Many non-ferrous and superalloys are far more sensitive to heat than carbon steel. Precipitation-hardened aluminium, especially 7000-series grades, can over-age if exposed to their aging range. For these, stress relief windows are incredibly narrow, roughly 300–450 °F (150–230 °C), and often, no stress relief is the correct choice. For process details, refer to our aluminum MIG welding guide.

Copper alloys, from brass and bronze to copper-nickel, may allow low and short stress relief cycles, but age-hardenable bronzes must avoid their precipitation temperatures entirely. The sharpest trap, however, is Alloy 718. The standard carbon steel PWHT window falls directly within its aging range. A routine cycle can unintentionally age the part while failing to produce its specified property set. For 718, the correct route is typically a solution anneal near 1,850–1,900 °F (1,010–1,040 °C) followed by its specified double aging cycle.

Material	Typical approach	Temperature guidance	Main risk
304/304L, 316/316L	Solution anneal + quench, or reviewed low-temp stress relief	Solution anneal 1,040–1,100 °C; relief only below ~425 °C	Sensitization, loss of corrosion resistance
2205, Super Duplex	Solution anneal + rapid quench	1,020–1,100 °C	Sigma phase, embrittlement
410, 420	Preheat, weld, temper	Temper roughly 1,000–1,200 °F	Cracking, hard untempered HAZ
17-4 PH	Solution + double temper to condition	Per H900/H1025/H1150 spec	Wrong condition, brittle HAZ
A514 (T-1), A517, A710, HPS 100W	Stress relief generally not recommended	Specialist review required	Unwanted tempering of the HAZ
Aluminium	Low-temperature stress relief, or none	~300–450 °F	Aging 7xxx grades
Alloy 718 and similar	Solution anneal + specified aging cycle	Solution ~1,010–1,040 °C	Uncontrolled aging in the PWHT window

Before sending any weldment to a furnace, identify every material within it, including dissimilar joints, filler metals, hardware, and weld wire. A single-recipe approach works for carbon steel, but for everything else, multi-material assemblies may require separate cycles, separate components, or a written heat treatment plan tailored to each alloy.

Preparing the Weldment: A Pre-PWHT Checklist

Post-weld heat treatment leaves little room for error. If a weldment goes into the furnace dirty, poorly supported, or with incorrect instrumentation, the thermal cycle won’t relieve stress as intended, and in some cases, it can create new problems. The following checklist is what experienced shops rely on to ensure the furnace can do its job properly.

Finish All Welds and Plan for Inspection

Every weld, repair, gouge, and temporary shop attachment must be complete before loading. A post-PWHT repair means grinding the weld out, rewelding, and running the entire heat treatment again, as a repair after PWHT will invalidate the original treatment. Temporary items, such as lifting lugs or boring bar mounts, should be removed or accounted for as required by procedure.

Inspection timing is just as critical. ASME codes and most industrial specifications require nondestructive examination, such as radiography, ultrasonic, magnetic particle, or liquid penetrant testing, to be performed after PWHT, not before. Many plant specifications explicitly prohibit pre-PWHT liquid penetrant, wet magnetic particle, or ultrasonic examination, as their residues can be masked by the thermal cycle or provide misleading results for the final, as-treated condition.

Clean the Weld, HAZ, and Surrounding Metal

A clean surface is required for even heat transfer and accurate temperature readings. Slag, spatter, grease, paint, water, rust, and mill scale should all be removed from the weld, the heat-affected zone, and a reasonable band of base metal. Slag can insulate the weld, causing steep thermal gradients, while oil, grease, or paint can outgas or ignite as temperatures rise. Rust and scale can flake off mid-cycle and fall onto the load below.

Pay extra attention to corrosion-resistant and non-ferrous materials. For example, when prepping aluminium, the surface oxide must be addressed alongside degreasing. For those looking to refine their process, our guide to aluminum cleaner for welding and the aluminum MIG welding guide cover practical prep habits that carry over to heat treatment.

Lastly, grind any sharp gouges, undercut, or notches smooth. These are stress concentrators, and PWHT is meant to relieve stress, not compensate for geometry defects. Where the code requires a minimum wall thickness at the weld, verify it before loading, as heat treatment will not restore lost material.

Fixturing, Support, and Expansion

At elevated temperatures, steel loses a significant amount of its strength. Long, unsupported vessels or beams can sag or twist under their own weight at soak temperatures. For this reason, the weldment must be blocked and supported to maintain its intended shape throughout the cycle.

However, restraint must be controlled. Fully welded-in, rigid fixtures can create enough thermal stress during ramp-up to crack the weld. The part must be allowed to expand. For carbon steel, thermal expansion is roughly 11 to 13 microinches per inch per degree Fahrenheit, meaning even a modest length will move several inches over the course of a cycle. Never tack a fixture to the weld itself, and always plan for expansion with your blocking layout.

Thermocouple Placement and Instrumentation

Furnace air temperature alone cannot prove that the metal has reached soak temperature. During ramp and soak, furnace and part temperatures can differ by tens of degrees, especially on heavy weldments. Part-contact thermocouples are required for proof of compliance.

• Placement: Attach at least one thermocouple directly to the metal, as close to the weld and HAZ as practical.
• Type: Use Type K for most carbon steel PWHT cycles between 595 to 760 °C. For higher alloy or hotter cycles, use a thermocouple rated for that service, such as Type N or Type S.
• Accuracy: Use correct extension wire with cold-junction compensation, and run a separate furnace thermocouple for control.
• Protection: Attach in a way that won’t create a stress raiser or damage a corrosion-critical surface.

AWS D1.1 and API 1104-style specifications require the time-temperature chart to record thermocouple locations, furnace ID, operator, procedure, and the full curve. A chart without these details is not a compliance chart.

Furnace Readiness

Before loading, verify the furnace is qualified for this exact load:

• Uniformity: Confirm a current furnace uniformity survey for the load size and configuration.
• Calibration: Check calibration dates for the chart recorder, controller, thermocouples, and all indicating instruments.
• Utilities and backup: Ensure fuel, power, ventilation, and any backup systems are available for the full cycle. A power interruption mid-soak is an out-of-cycle condition and is often cause for rejection.
• Safety: Verify fire protection, clear access, and review applicable safety requirements, such as those outlined by OSHA.
• Atmosphere control: For vessels that held flammable or toxic hydrocarbons, a documented purge, gas-free certificate, and gas testing are required before entering a heated furnace.
• Loading: Arrange the load for proper air circulation, with thermocouple leads routed through penetrations without pinching or dragging across hot surfaces.

With each of these steps complete, the preparation sheet becomes just as binding as the WPS. The furnace gets the credit for stress relief, but it’s the pre-PWHT preparation that determines whether that relief was ever achieved.

The Mistakes That Cost Fabricators the Most

Most costly PWHT failures don’t come from skipping the treatment. They come from small procedural errors that go unnoticed until an inspector asks for the charts. Here are the most common pitfalls, broken down as symptom, cause, and fix.

Recipe and Timing Errors

Symptom: The part was held for what looked like enough time, but test results show little to no stress relief.
Cause: The soak clock started when the furnace hit setpoint, not when the part reached temperature. For heavy sections, that can under-soak the weldment by an hour or more.
Fix: Use part thermocouples placed on or near the weld and HAZ. Soak time starts the moment the control thickness of the part reaches the minimum specified temperature, not when the furnace air does. Verify it against your chart every cycle.

Symptom: Hardness comes back high, or the material turns brittle.
Cause: The temperature window was wrong. Too low gives no measurable relief; too high causes over-tempering, grain coarsening, or sensitization in stainless steels.
Fix: Follow the temperatures dictated by your code and WPS. When a full cycle is impractical because of size or geometry, check for approved alternatives such as AWS D1.1’s alternate low-temperature stress relief (Table 5.3D), since these lower-temperature, longer-time cycles exist for exactly that scenario.

Heating and Measurement Errors

Symptom: Distortion, warpage, or in severe cases cracking during the heat-up phase.
Cause: Heating too fast creates steep thermal gradients between the hot exterior and cooler interior of the weldment.
Fix: Follow your procedure’s ramp rate to the letter. Most codes cap it to prevent thermal shock, giving the mass of the part time to expand uniformly.

Symptom: The chart looks perfect, but the inspector isn’t convinced the weld zone actually hit temperature.
Cause: Relying on furnace air temperature as proof, measuring at a single location, or using the wrong thermocouple type or a miswired cold junction.
Fix: Use multiple part thermocouples, with at least one directly on the weld or HAZ of the thickest section. Make sure they’re the correct type for your range, properly extended, and that your recorder has cold-junction compensation. Never substitute furnace TC readings for part readings.

Symptom: A thick transition joint passes temperature readings, but a thinner adjacent area doesn’t, or vice versa.
Cause: Confusing nominal wall thickness with control thickness, the thickest welded section of the weldment. Soak time must be calculated off that control thickness, not the average or thinnest member.
Fix: Identify your controlling section before you calculate the cycle, and place a thermocouple there. If thicknesses vary drastically, local considerations must be reviewed by the engineer.

Method and Material Errors

Symptom: Overheating or material degradation in a non-carbon steel assembly.
Cause: PWHT-ing a dissimilar assembly where one material needs a much higher temperature than the other can tolerate. This is especially dangerous for austenitic stainless steels, duplexes, or quenched and tempered alloys.
Fix: The material data sheet and governing document always win. Some materials cannot take a standard carbon steel PWHT cycle at all. If a dissimilar joint must be treated, an approved procedure with metallurgical review is mandatory.

Symptom: A field repair is attempted with nothing but a rosebud torch.
Cause: Using open flame without an approved procedure. Flame heating is nearly impossible to control, monitor uniformly, or document to code requirements.
Fix: Only use methods that can prove the part reached temperature with a recorded chart. If local PWHT is the only option, it must be approved by the engineer and performed with controlled heaters that meet the same verification requirements.

Cracks, Repairs, and Documentation

Symptom: Cracking appears or was found, and PWHT is being considered as a fix.
Cause: Treating a weldment that already has hydrogen cracking. Heat treatment cannot heal an existing crack. It relieves stress and diffuses hydrogen that is yet to form a crack.
Fix: Find and remove the cracks by gouging or mechanical removal, verify with NDT, repair with proper preheat and low-hydrogen practices, and only then consider PWHT if required.

Symptom: A repair is made after PWHT, but no re-treatment is performed.
Cause: Failing to re-heat-treat after a post-PWHT repair. Repair welds made after PWHT generally require the weldment to be re-heated and re-examined, and most codes require re-NDT after any repair.
Fix: If the repair is significant enough to reintroduce stress or hydrogen, re-treat the affected area, or the whole weldment if required, and re-examine it to the code’s satisfaction.

Symptom: An inspector rejects the job over paperwork, even though the part looks fine.
Cause: Using a furnace without a current uniformity survey and calibration, or missing required documentation.
Fix: Your time-temperature chart, thermocouple locations, furnace ID, procedure or WPS reference, operator, and calibration certificates matter just as much as hitting the temperature. If a PWHT temperature is exceeded beyond the allowable limit, the usual remedy is to re-heat to at least the previously achieved maximum and re-examine the welds by NDE, with the specifics set by the governing code.

PWHT is a precise operation. Cutting corners on any of these details doesn’t save time. It almost always costs more in rework, delays, and, most importantly, the integrity of the structure you’re putting into service.

Verification, Inspection and the Paper Trail

The furnace cycle is the easy part of post-weld heat treatment. What gets a job through inspection is the evidence. A PWHT performed perfectly but never recorded is, on paper, a PWHT that never happened. Two things make heat treatment defensible: measurements proving the part got hot enough and stayed long enough, and the records that prove it.

The Heat Treatment Chart

The most important document in the process is the time-temperature chart, a continuous strip or electronic trace showing the ramp up, soak plateau and controlled cool down. A credible chart lets someone outside the shop reconstruct the cycle. Look for:

• Thermocouple locations, marked on a sketch of the part, tying each reading to a physical spot.
• Furnace or oven identification, linking the chart to a machine with its own calibration and uniformity history.
• Operator name and date, plus the procedure, work instruction or WPS number the cycle was run to.
• Set point, actual soak temperature and actual soak duration, recorded separately. The soak clock starts when the part reaches temperature, not the furnace.
• Any deviation or excursion, such as an overshoot or power interruption, flagged in writing rather than quietly corrected.

Calibration, Uniformity and the People Involved

Charts are only credible if the instruments behind them are trustworthy. Expect calibration certificates for the chart recorder and every thermocouple, traceable to a national standard, plus a current furnace uniformity survey. Surveys repeat on a set interval and after any furnace repair, relocation or thermocouple change, because a hot spot or cold corner wrecks the chart’s meaning. Also expect proof the operator and the thermal procedure itself were qualified. Codes such as ASME’s pressure vessel and piping codes treat heat treatment as a controlled process with controlled records, not an informal shop favour.

Hardness Testing: The Fastest Proof the Cycle Worked

Hardness is the cheapest, fastest screen that a PWHT cycle did its job. Tempering and stress relief measurably reduce peak hardness in the heat-affected zone, so pre- and post-treatment readings at the same locations show whether the metal changed. Testing follows AWS and ASTM methods: ASTM E18 for Rockwell, ASTM E10 for Brinell, ASTM E384 for Vickers. The HAZ, not the weld metal, normally governs acceptance, since that is where untempered martensite forms.

For sour service, hardness limits come from NACE MR0175 / ISO 15156. As a rule of thumb, 22 HRC is the commonly cited maximum for carbon and low-alloy steels under many sour-service conditions, with relaxed limits for specific qualified materials. Test procedures, equipment and technician qualification should follow NACE and applicable ASME Section V provisions.

Impact Testing, NDT and Dimensional Checks

Where toughness is a stated acceptance criterion, Charpy V-notch testing is required after PWHT, on the weld and HAZ, at the specified test temperature per the code. Nondestructive examination normally follows heat treatment rather than preceding it: MT and PT for surface defects, UT for internal flaws, and RT where the code or owner requires it, with radiation safety controls under OSHA’s welding guidance and site rules. Two rules catch people out: an overheat above the specified limit requires re-examination of the affected area, typically radiographically; and a repair made after PWHT means the weldment is re heat-treated and re-examined, so the original chart no longer tells the whole story.

Surface preparation for MT and PT matters as much as the procedure; a poorly cleaned area reads as an indication nobody should trust. On aluminium, use a dedicated cleaner, not a carbon-steel product; prepping aluminium properly for welding and inspection is worth reading before the part enters the shop.

Dimensional verification is easy to forget because PWHT is treated as a stress-relief operation rather than a forming operation. It is not. Heating and controlled cooling of a restrained assembly moves it, so check straightness, flatness and critical dimensions against the drawing, and account for any correction work.

Record Retention and the Turnover Package

Everything above has to survive the project. ASME Section VIII Div 1 and Div 2 require quality control records, including heat treatment documentation, as part of the manufacturing record, and the same applies to piping turnover packages, vendor turnover files and owner data books. Retention periods follow the contract and governing code, not shop habit. Keep the chart, the survey, the calibration certificates and the hardness or impact results together, indexed to the weldment’s unique identifier, so the file can be pulled at turnover or audit years later.

When You Can Skip It, What to Do Instead, and the Bottom Line

Every fabricator eventually asks the same thing: do we actually have to heat treat after welding? Sometimes yes, sometimes no.

Code Exemptions: Permission, Not Obligation

Most piping codes, ASME included, carry exemption clauses that relieve certain joints from mandatory PWHT. ASME B31.3 exempts based on material grade, nominal wall thickness and service temperature: thin, lightly restrained carbon steel joints in benign service have historically performed acceptably without stress relief.

The TWI review of postweld heat treatment code exemptions questioned several of those blanket rules, showing that PWHT is a significant cost and that necessity depends on a cluster of factors: fracture toughness, service temperature, restraint, and hydrogen sensitivity. A joint exempt in benign conditions can fail in cold, highly restrained service. EPRI’s technical basis for minimizing postweld heat treatment reached a related conclusion: code temperature limits can be unreasonably low for some materials, so fabricators over-treat. EPRI recommended adjusted temperature ranges.

Alternatives That Sometimes Work

Something has to compensate for the residual stress PWHT would have relieved.

Mechanical and thermal stress improvement. Vibration stress relief is the best known: vibrating the weldment at a frequency and amplitude that produces small plastic strains in the weld metal redistributes peak residual stress. It needs no furnace and does not distort the part. The catch: it is hard to verify, only partly effective at peak stress locations, and does nothing for hydrogen or untempered martensite, so the code still governs acceptance. Thermal stress improvement is the same idea taken gently: heat lower and cool under controlled conditions to build a compressive surface layer.

Peening and in-process tempering. Shot or hammer peening at the weld toe and adjacent HAZ introduces compressive surface stress, which works well against fatigue cracking. It does nothing for hydrogen, complicates surface finish, and supplements rather than replaces. In-process tempering, where the electrode’s alloy content tempers the HAZ as it solidifies, is genuinely effective on some grades, especially the low-hydrogen, nickel-bearing consumables used for quenched-and-tempered and martensitic steels. It removes the need for a post-weld cycle.

Welding engineering that removes the need. This is where the real savings live. Modern low-hydrogen consumables cut diffusible hydrogen at the source. Controlled preheat and interpass temperature keeps the cooling rate down so hardenable HAZ metal does not form untempered martensite. Better fit-up cuts the metal melted and cooled. Symmetrical sequences and restraint management cut locked-in thermal stress, and thinner sections, where the code allows, remove most of the driving stress.

Our guide to alloy steel welding electrodes covers the low-hydrogen and nickel-bearing consumables that let an in-process temper do the work of a post-weld cycle.

Cost and Schedule: The Part Nobody Quotes Accurately

PWHT is one of the more expensive post-fabrication operations, and the bill is never just the furnace. It includes the cycle itself, handling and fixturing, thermocouple attachment, chart recording, calibration and uniformity survey, and returning the part to fit-up for closure welds. On a large vessel it can add days to the schedule, and those days are not contiguous because the part is handled twice. The decision belongs at design review, when wall thickness, material and welding sequence are still open; afterwards the need usually cannot be removed.

A Decision Framework for the Next Job

1. Does a controlling document require PWHT? Contract, purchase order, design specification, code edition and the engineer all outrank shop judgment.
2. If the code exempts the joint, does the service still argue for it? Cold service, brittle fracture, high restraint, sour service hardness limits, creep, fatigue, dissimilar joints and thick sections push back toward treatment.
3. Is the material sensitive to the treatment temperature? Stainless, duplex, martensitic, quenched-and-tempered, aluminium and nickel alloys can be damaged by the wrong cycle.
4. Can the need be engineered out? Low-hydrogen consumables, tighter preheat and interpass control, better fit-up, a symmetrical sequence and less restraint often remove it at no added cost.
5. What do the alternatives actually deliver? Stress improvement reduces residual stress but does nothing for hydrogen or untempered martensite.
6. What will it cost to prove it either way? A documented cycle with charts is defensible and auditable; an undocumented exemption is not, and missing charts trigger rejected work.

The Bottom Line

Heat treatment after welding is engineering, not a finishing touch: a controlled thermal cycle with a purpose, a temperature window, a documentation trail, and a code or an engineer behind it. It relieves residual stress, removes diffusible hydrogen, and tempers hardable HAZ metal. That can be the difference between a structure that reaches service and one that cracks on hydrotest or in the field.

Skipping it is sometimes right, and the codes and the literature give you real grounds. But it must be a documented decision by the design authority, supported by an engineering case, not a scheduling choice in the weld bay. The cheapest heat treatment is the one you designed out of the job; the most expensive is the one you skipped and had to redo.

Frequently Asked Questions

Is heat treatment after welding always required?

No. It depends on the governing code, the design engineer’s specification and the service conditions. ASME Section VIII, B31.3, B31.8, API 650 and similar codes set minimum requirements, and codes such as AWS D1.1 generally leave the decision to the engineer of record. Even where a code allows an exemption, the design authority can still require PWHT.

What temperature is used for heat treatment after welding?

For carbon steel, the common stress-relief window is 1,100–1,200 °F (595–650 °C), with ASME B31.3 and AWS D1.1 tables setting the exact range by material group. Alloy steels step up to 1,200–1,400 °F, and Grade 91 creep steels to 1,300–1,425 °F. Stainless steels and aluminium require entirely different cycles.

How long should the soak time be for post-weld heat treatment?

Most codes use one hour per inch of control thickness with a minimum of 15 minutes for the standard range, with heavier sections requiring additional time — for example, ASME B31.3 requires 2 hours plus 15 minutes per additional inch over 2 inches for P-No. 1 carbon steel. Soak time begins when the part reaches temperature, not when the furnace does.

Does stress relief heat treatment fix weld distortion?

Not the distortion that already exists. PWHT relieves the driving residual stresses so the part is less likely to move further in service, and it can slightly change dimensions, but a bent or out-of-tolerance weldment generally still needs mechanical straightening or replacement. The real value is preventing new distortion from high-restraint service loads.

Can you heat treat stainless steel after welding the same way as carbon steel?

No. Carbon-steel stress-relief temperatures fall in the carbide precipitation range for austenitic stainless and can destroy corrosion resistance; duplex stainless can embrittle in that range. Stainless typically needs solution annealing at roughly 1,020–1,100 °C with rapid cooling, or a metallurgically approved low-temperature stress relief, and the choice must be specified.

What is the difference between stress relief and a full post-weld heat treatment temper?

Stress relief is a subcritical cycle below Ac1 that relieves residual stress without changing the microstructure, while a temper or PWHT temper is specified to modify the HAZ microstructure — tempering untempered martensite, restoring toughness and meeting hardness limits. For hardenable and quenched-and-tempered steels, the temper cycle is what controls properties.

Can vibration stress relief replace a code-required PWHT?

Usually not without written approval. Mechanical stress improvement methods such as vibration stress relief are recognized in some specifications, but their effectiveness for critical service is debated, and codes generally require the design authority to accept them in writing before they are used in place of a specified furnace cycle.

What happens if a part goes above the specified PWHT temperature?

It depends on the material and the amount of overheat. Slight overtemp on carbon steel may be acceptable, but exceeding the tempering temperature can permanently soften quenched-and-tempered or martensitic material, and overheating austenitic stainless can sensitize it. Codes typically require the part to be re-heated to at least the previously achieved maximum temperature and the welds re-examined by NDE.

Should nondestructive testing be done before or after heat treatment?

After, in most cases. Many specifications explicitly prohibit liquid penetrant, wet magnetic particle or ultrasonic examination before PWHT, and codes generally require the welds to be re-examined after the cycle because properties and geometry have changed. Doing NDT first and never looking again after treatment is a common documentation and quality mistake.