Chilled Water Pipe Welding: Codes, Procedures, and the Field Guide

There is a specific sound a chilled water system makes when a weld lets go — a soft, hissing tick behind the drywall that only the building engineer can hear at 2 a.m. A single 4-inch butt weld in a hospital chiller room had passed the hydrostatic test, passed visual inspection, and still opened a pinhole in the carrier pipe. The repair cost three times the original weld, and the downtime cost more than that. Chilled water pipe welding is one of those trades where “good enough” quietly becomes “very expensive.”

The 2 A.M. Ceiling Leak: Why One Chilled Water Weld Costs More Than the Weld

Most incident reports start after the water hits the floor. By the time the building engineer can track down the source, it’s usually above a ceiling with no easy access. More often than not, the culprit is a pinhole at the root of a butt joint. From the outside, the bead looks like a textbook weld. From the inside of the bore, it’s porous, crusted with mill scale and oxides that were welded in because nobody purged the pipe before tacking. The outside sold it. The inside failed it. As one twenty-year journeyman put it, the weld didn’t fail, it was never complete.

What Chilled Water Pipe Welding Actually Is

Chilled water pipe welding is joining the steel piping that moves chilled or condenser water between chillers, cooling towers, pumps, air handlers, and heat exchangers. In most central plants, there are two loops at play. The chilled water loop is a closed, pressurized circuit moving cold water to air handlers and back to the chiller. The condenser water loop moves water between the chiller’s condenser and the cooling tower, which may be open to its basin or a closed loop depending on the design.

For commercial work, that pipe is most often welded Schedule 40 carbon steel with an abundance of fittings, flanges, and branch connections. Much of it will be hidden behind insulation, so the welder’s job isn’t just to make a joint that looks good on the floor, it’s one that has to stay sound for years with zero visibility.

Where the Piping Lives

The location of these welds is half of the expense. A joint in a central plant is a completely different job than one buried above an occupied ceiling.

Location	What Makes It Tricky
Chiller galleries and central plant rooms	Tight access around heavy equipment with shutdowns that require facilities approval.
Ceiling plenums and above-ceiling spaces	Concealed once insulated, with little room to swing an electrode holder or maneuver equipment.
Mechanical rooms and penthouse risers	Vertical runs, branch take-offs, and added stress from heavy hangers.
Industrial and process areas	Harsh water chemistry, higher design temperatures, and often an extra layer of plant-specific inspections.

Why These Welds Are Different

Three factors set chilled water welds apart from a typical pipe job.

First, constant operation. Many chilled water loops run year round. There is no dry season to give a small flaw a break. Every joint is thermally cycled and flooded for its entire service life, so imperfections don’t have time to go dormant.

Second, total concealment. Once insulation is on, a failure becomes demolition. To reach a single joint above a ceiling, crews have to remove insulation, access panels, and sometimes framing. That access is almost always more costly than the weld itself.

Third, an oxygen-controlled environment. Closed chilled water loops are treated with corrosion inhibitors and oxygen scavengers to prevent pitting. Because of that, the inside of the pipe is chemically protected. Introducing trapped mill scale, moisture, or oxides from a non-purged weld creates a corrosion site in the one place the treatment can’t reach.

These joints must also meet pressure demands. Typical closed chilled water loops run at roughly 30 to 80 psig of static fill plus pump head, with 125 psig commonly used as the design pressure. Per ASME B31.9, hydrostatic testing is required at 1.5 times the system design pressure, meaning a 125 psig system is tested to about 188 psig.

The True Cost of a Failure

The weld is cheap. Everything to fix it isn’t.

• Demolition and reinstatement: Insulation, ceiling tiles, access panels, drywall, and potentially framing all need to be removed and replaced to reach a concealed joint.
• Access and labor premiums: Repairs at 2 or 3 a.m. in an occupied building often require lifts, temporary lighting, and shift premiums on top of a qualified welder’s rate.
• Cooling loss: Taking a chiller offline can strip cooling capacity from a hospital, data center, or office tower, turning a leak into a facility-wide issue.
• Water damage: Even a small pinhole can release a substantial amount of water before isolation, damaging finishes and equipment below.
• Liability and reporting: For mission-critical spaces, downtime isn’t a comfort complaint, it’s a reportable event with serious consequences.

Time and time again, the repair itself ends up costing multiple times the original weld, with downtime far exceeding both. Nobody budgets for a purge. Everybody budgets for a Tuesday night in an occupied building. That’s exactly why every step that follows in this article exists, from qualification to inspection, to prevent that one pinhole from ever happening.

How a Chilled Water System Works — and Why That Changes How You Weld It

A chiller is a heat pump built around a metal wall, usually a tube bundle. Refrigerant boils at very low temperature on one side, water on the other. Heat crosses the wall into the water, which is pumped away, cools, and returns. The rest is pipe, pumps, valves, and air handlers carrying that water to where the cooling is needed.

Two water circuits surround that machine, and telling them apart explains most of what matters about the welds.

Loop	Carries	Typical conditions	Implication for the welder
Chilled water (closed)	Evaporator to air handlers and back	45°F supply, 55°F return	Cold, treated, pressurized, hidden
Condenser water (open)	Condenser to cooling tower and back	Ambient, aerated, often 90°F plus	Corrosive, dirty, exposed to weather
Condenser water (closed)	Same duty, sealed loop, separate cooling tower	Warm, isolated from atmosphere	Treated like chilled water, but hot

Open versus closed is a different corrosion universe. Open condenser water is sprayed into warm, humid air at the cooling tower, picking up oxygen, airborne solids, biological growth, and whatever the tower fill media leaches. The warm water also speeds up chemical reaction. A closed loop never touches the atmosphere, so it stays clean and controlled.

Delta T is the whole point of the system

Closed chilled water runs 45°F supply and 55°F return, a delta T of roughly 10°F. That difference is the reason the system exists: more delta T moves more heat and costs less pumping energy. When an engineer chases capacity, they are chasing delta T, which means chasing flow velocity, fouled strainers, or a chiller loading with scale.

Why a leak on chilled water is a building problem, not a plumbing problem

That 45°F supply keeps the piping surface below the dew point indoors, so a chilled water main in a plant room feels cold and damp with moisture condensing on it. That changes what a small leak looks like. A weeping joint does not spray. It sweats, wets the insulation, and the insulation holds the water until it saturates and drips into the ceiling below. By the time a stain shows on a tile, the leak has run for weeks, the pipe looks wet and rusty around the bead, and nobody worries until the drywall comes down. A weld that passes a hydro test and then weeps is a serious finding, not cosmetic.

Velocity: the number that ages the system

Flow velocity matters because water is not gentle. Design velocity for chilled water mains falls in the 3 to 7 fps range, and it is not unusual to see hydronic piping design guidance treat that band as the practical working envelope. Above it you accelerate erosion corrosion on pipe walls, erode tube surfaces inside the evaporator, and wear the weld bead. The bead is the roughest surface in the run, so velocity attack shows up first as a groove in the reinforcement, then a thinning of the section.

Flushing is a separate event with its own requirement. Under UFGS 23 64 26, flushing runs at a minimum velocity of 3 fps until strainer baskets retain no particles larger than 1/32 inch. Here is the trap: construction debris, weld slag, and mill scale released during flushing will foul the chiller tube sheet within the first operating season if temporary startup strainers are omitted.

What the water does to the metal

Closed loops are treated with corrosion inhibitors and oxygen scavengers, which is why closed chilled water can stay in service for decades with minimal wall loss. Open condenser water gets a heavier treatment program because it is aerated and warm. In either case, initial corrosion inhibitor must be dosed per the water treatment program before chillers are placed in service.

Now the connection to welding. A system aerated during installation, or a joint purged badly with oxygen trapped in the bore, attacks the carbon steel at the root of the weld. That is where pitting corrosion starts, because that surface has no mill scale, no rolling marks, and a heat-affected zone with different chemistry than the parent metal. Pitting at the root of a weld is a leak waiting for a reason. Where stainless or aluminum is used, the metallurgy changes again, making surface preparation and cleaning discipline far less forgiving, as detailed in this guide to prepping aluminum for welding.

Movement, anchors, and the load on the joint

A chilled water run between an anchor and a chiller is never static. From cold start to operating temperature, and summer to winter, the pipe grows and shrinks. Build a loop between a fixed anchor and a chiller header and the pipe between them must take that movement, which arrives at the joints as bending and axial load.

Branch connections see the worst of it. A tee or a reinforcing saddle is where thermal movement, internal pressure, and vibration land on the same piece of metal, so a brittle or poorly shaped weld there is the first thing to crack. The joint needs strength and ductility to take that load, and a rod chosen for all-position work matters where the joint sits overhead or in a tight corner.

Bring it back to the weld

Everything above converges on one uncomfortable fact. Pipe, fittings, hangers, and valves are manufactured products with known properties. The weld is the only part of the system with a metallurgical discontinuity in it: a column of grains that did not exist before the arc struck, a heat-affected zone with altered properties, and a bead surface with different roughness than the surrounding pipe.

That is why the system fails at the weld first when the procedure is wrong. A weld that was not purged can pit. A weld that was not qualified can crack under movement. A weld with excess reinforcement and undercut is a stress riser waiting for the anchor to push. A weld inspected from the outside only may have a root defect you never saw. Get it right and the weld is the most durable part of the run. Get it wrong and it becomes the weak link in a system that has to run quietly, continuously, for years.

The Rulebook: ASME B31.9, Section IX, and the Codes That Govern the Work

Chilled water pipe welding sits inside a stack of codes, standards, and contract documents, and most field problems trace back to somebody picking up the wrong one, or to nobody checking the adopted edition at all. The good news: the governing framework is smaller and more logical than it looks. There is one primary design and installation code, one qualification code, one contract specification that ties them together, and a handful of safety standards. Once you know which document answers which question, the paperwork stops being mysterious and starts being a checklist.

ASME B31.9: the code written for building services piping

ASME B31.9, Building Services Piping, was first published in 1982 specifically because building services piping did not fit neatly into the power or process codes that already existed. It covers the design, materials, fabrication, installation, inspection, examination, and testing of piping that serves buildings. B31.9 covers water and antifreeze solutions for heating and cooling, condensing water, steam and condensate, nontoxic nonflammable liquids and gases, vacuum service, and compressed air. That is a direct description of a chilled water system: the supply and return mains, the condenser water loop, and the auxiliaries that go with them.

The code sets hard limits on where it applies, and these limits decide which code a given system falls under:

• Liquid service up to 350 psig
• Steam and condensate service up to 150 psig
• Carbon steel piping up to NPS 48 (DN 1200) with 0.50 in (12.7 mm) wall
• Stainless steel piping up to NPS 24 with 0.50 in wall
• Aluminum up to NPS 12
• Brass and copper up to NPS 12

Now the part that catches people out. B31.9 explicitly does not cover refrigeration piping, potable water piping, sanitary and storm drainage systems, or fire protection piping. In a big building you can be welding chilled water, domestic water, and a sprinkler main in the same mechanical room, each under a different code. If a chilled water system falls outside the B31.9 envelope (high pressure, high temperature, unusual fluid, or a size beyond the limits above), it moves up to a higher-tier code such as ASME B31.1 for power boiler piping or ASME B31.3 for process piping.

ASME BPVC Section IX: the qualification engine

Every code in the ASME family that requires welded joints points at the same engine: ASME BPVC Section IX, Qualification of Welders, Welding Operators, and Welding Procedures. B31.9 requires that procedures and welders be qualified in accordance with Section IX. Section IX decides whether your Welding Procedure Specification, on the QW-482 form, is acceptable, and whether your welder holds a valid performance qualification. It is also where the PQR, the Procedure Qualification Record on form QW-250, is recorded, which is the evidence that the procedure actually works.

Section IX is deliberately narrow. It does not tell you what pipe to use, what pressure to test to, or what to inspect. It answers one question: is this weld made by a qualified procedure, by a qualified welder, to a qualified standard?

AWS alternatives: D10.9 and B2.1/B2.1M

American Welding Society standards do the same job for different industries. AWS D10.9 covers welding of ferrous and nonferrous metals used in building construction piping and is referenced by several building codes for hydronic and general service piping. AWS B2.1/B2.1M, Specification for Welding Procedure Specifications and Welding Performance Requirements, is the general shop and field standard across SMAW, GMAW, FCAW, GTAW, and brazing, and it is the standard most likely to show up where the owner has added supplemental welding requirements.

Either one works, but only when the referencing document actually calls for it. B31.9 does not automatically hand the job to AWS. And if a spec points to B2.1/B2.1M, check the edition: AWS B2.1/B2.1M:2026 was approved as an American National Standard on December 15, 2025, superseding the 2021 edition. Qualifications written to the old edition are not automatically valid under the new one, and a crew can be unqualified on a Monday morning because of a paperwork detail nobody noticed.

For context on how large the qualified-procedure ecosystem has become, the NCPWB has developed more than 100 ASME Section IX compliant welding procedures covering SMAW, GTAW, GMAW, FCAW, and torch brazing on mild steel through titanium. Those are prequalified recipes a contractor can adopt instead of testing from scratch, and they are the reason a mechanical contractor can mobilize on a Schedule 40 header without spending three weeks in a test lab.

UFGS 23 64 26: the contract document that ties it together

Codes tell you what is technically required. The contract document tells you what you actually have to deliver. For building piping, that is the Uniform Facility Guide Specification 23 64 26, Piping Systems, the source most federal, institutional, and many commercial specifications borrow from. It sits on top of B31.9 and Section IX and adds the project-specific requirements: material submittals, shop drawings, hydrostatic test pressures and hold times, NDE scope, and workmanship rules.

One line in that specification causes more cracked elbows than anything else on this list. UFGS 23 64 26 states that field-fabricated direction changes and branch connections formed by mitering or notching pipe are not permitted. No field mitered elbows, no saddle-notched branch connections, no “we’ll fit it and grind it in” on a running system. Every direction change and branch is a manufactured fitting. If the drawing shows something that requires a field miter, the answer is a change order and a shop-fabricated spool, not a torch and a grinder in a ceiling plenum.

Safety and consumable handling

Two more standards belong on the list. AWS Z49.1, Safety in Welding, Cutting, and Brazing, is the reference for fume control, ventilation, fire watch, hot work permits, and confined space awareness, and it lines up with NIOSH and OSHA guidance on welding fume. On a chilled water job in an occupied building, ventilation and fume control matter as much as the weld, because the welder is frequently kneeling in a plenum with a return air duct nearby.

Electrode storage and drying also gets written into most specs and shop procedures, referencing AWS D1.1/D1.1M practice for handling, storage, and reconditioning of welding consumables. Electrodes are not warehouse inventory. Once a container is opened, the electrodes get a holding oven, a controlled issue rate, and a documented reconditioning routine. A wet or stale electrode is the single most common cause of porosity in carbon steel weld metal, and porosity on a chilled water joint is a future leak.

For a shop that runs aluminum work on the side, our guide to aluminum flux core welding covers the process differences worth understanding, though keep in mind that B31.9 caps aluminum piping at NPS 12, so most aluminum chilled water work falls under project-specific specifications rather than standard B31.9 requirements.

Choosing the Joint and the Material: Carbon Steel, Stainless, and Copper

Walk into a chiller plant and nearly everything looks identical: black carbon steel pipe, Schedule 40, matching wrought fittings, bolted flanges. Chilled water is not an aggressive fluid, and the industry settled on carbon steel because it is cheap, stocked in every size, easy to weld, and easy to insulate. The real decisions happen at the seams, where a note on a drawing must become a joint a welder can make at 2 a.m. in a crowded plant room.

Carbon steel is the default, and for good reason

Most chilled water distribution is ASTM A53/A53M Grade B black pipe, welded or seamless. Where the same system also carries hot water or a higher design temperature, ASTM A106/A106M Grade B or C is the usual substitution. Fittings have to match the joint method you intend to use:

• ASME B16.9 for butt-welded wrought fittings
• ASME B16.11 for socket-weld fittings, generally NPS 2 and smaller
• ASME B16.5 for flanges, NPS 1/2 through 24
• ASME B16.47 for large diameter flanges beyond the B16.5 range
• ASME B16.34 for valves

One detail gets lost on drawings: fitting material and schedule must line up with the pipe, and the joint type has to be the one your welding procedure actually qualifies. A B16.9 fitting dropped into a run assembled with B16.11 socket welds creates a qualification problem nobody finds until the inspector is standing in the gallery.

The size split that decides your joining method

Many building specifications divide the system at NPS 2-1/2 for a straightforward reason. Below that size, pressure thrust and hanging loads are small enough that a soldered or threaded joint holds, and the speed of a brazed connection is worth more than structural strength. Above it, those forces make welded and flanged joints the standard.

Pipe size	Typical material	Typical joint
NPS 1/2 to 2	Type L drawn copper	Soldered (sweat)
NPS 1/2 to 2	Schedule 40 steel	Threaded
NPS 2-1/2 and larger	Schedule 40 steel	Welded and flanged

Comparing the joining methods

Each method trades one advantage for another.

Method	Strengths	Limitations
Butt weld	Full structural strength, matches pipe, clean bore, easy to inspect and radiograph	Needs beveling, purging, and a qualified procedure; slowest to make
Socket weld	Fast, no beveling, good for small sizes and equipment hookups	Lower strength than butt weld, rougher bore, small diameters only
Grooved coupling	No hot work, quick install, reusable, good in occupied spaces	Verify groove dimensions and key engagement before torque; not ideal where weld inspection is mandated
Brazed or soldered copper	Fast, permanently leak-tight, compact	Surface cleanliness critical, no flux allowed, purge needed for silver brazing
Threaded	Cheap, no heat, simple to correct	Limited size range, corrosion-prone threads, poor for concealed service

Grooved couplings deserve their own caution, because they are only as good as the groove. Follow the coupling manufacturer’s published procedures: verify groove dimensions against the maker’s chart and confirm key engagement before any fastener is torqued. Skip the groove check and the joint either leaks at the gasket or quietly splits at the housing.

Stainless steel when the water is the problem

Condenser water in coastal plants, cooling towers with poor bleed-off control, or systems with high chloride levels sometimes push engineers toward stainless. Inside ASME B31.9 the stainless envelope is narrow: NPS 24 (DN 600) maximum size and 0.50 in (12.7 mm) maximum wall thickness.

Stainless brings its own welding problems. It is normally joined with GTAW or SMAW, purge requirements are tighter than for carbon steel, and interpass temperature must be held down to avoid sensitization, where the grain boundaries of the weld area lose corrosion resistance. Filler metal matters as much: matching or over-alloyed filler preserves the passive layer. Read up on TIG technique before you set up the purge.

There is a respiratory concern carbon steel does not have. Stainless welding fume contains hexavalent chromium, a recognized carcinogen, which makes local exhaust ventilation a condition of the job rather than a courtesy. The OSHA chemical data entry for hexavalent chromium and NIOSH’s welding fume resources lay out the hazard and the controls. SMAW stainless in a closed plant room is a ventilation problem before it is a welding problem.

Copper, and where it stops

Copper earns its place on small-bore branches, risers, and equipment connections where space is tight and a torch is faster than a welder. But it has hard limits inside B31.9: NPS 12 and 12.125 in outside diameter maximum for copper tube. Past that size the code envelope no longer reaches, and the spec must reference something more stringent.

In practice, field copper is rarely welded. It is brazed or soldered, and the preparation rules are unforgiving: UFGS requires joint surfaces be cleaned with a wire fitting brush until bright and clean before brazing, and states that brazing flux must not be used. The filler metal is self-fluxing and the torch work follows standard practice, as covered in this guide to air acetylene welding and brazing.

For stick welding stainless or carbon steel in tight spaces, this piece on why SMAW still wins in the field is a useful counterweight to the argument for switching machines.

Match the joint to the load, the size, the environment, and the inspection plan the owner will demand. Get that right on paper, and the weld becomes routine.

The Weld Itself: A Step-by-Step Field Walkthrough

Most chilled water failures are not exotic metallurgical disasters. They are ordinary ones: a pipe never cleaned before assembly, a root gap set by eyeball, a purge never established, an electrode left in a damp cooler for a week. Here is the sequence a good crew runs.

Step 1: Confirm What You Are Welding

Read the markings before a fitting goes on. Carbon steel pipe should carry a mill mark showing ASTM A53/A53M Grade B or ASTM A106/A106M, plus size, schedule, and heat number. Wrought fittings carry the manufacturer mark and the ASME B16.9 or B16.11 designation. Check both against the limits and schedules in ASME B31.9 and the qualified WPS.

Step 2: Clean the Pipe and the Bore

Scale, rust, mill slag, oil, pipe dope, and insulation adhesive must come off the pipe, the fitting, and the inside of the bore. Mill scale is the dangerous one: it sits inside where nobody looks and flicks straight into the weld pool. A file, a power brush, and degreaser with lint-free rags is the realistic minimum.

Step 3: Square It, Bevel It, Set the Gap

Cut the pipe square, then form a single V-groove. On Schedule 40 carbon steel, an included angle of 37.5 to 45 degrees is typical, with a root face and root gap near 1/16 in. (1.6 mm) each. Check gap and alignment with a root gap gauge and straightedge before tacking. Allowable misalignment scales with pipe diameter, so read the B31.9 figure rather than trusting your eye.

Step 4: Purge the Joint Properly

This step gets skipped, and it decides whether your weld survives. Purging fills the pipe and fitting with an inert gas so the root pass is made in an atmosphere that cannot oxidize. Nitrogen is the common choice. Put a flowmeter on the purge line so flow can be set and repeated on every joint, and vent the far end so gas actually travels through. Operators monitor oxygen at the vent and hold it in the low hundreds of ppm.

Step 5: Pick the Process for the Access You Have

Process	Filler	Where It Wins
SMAW	E7018	Cramped plant rooms, dirty or rusted steel
GMAW	ER70S-6	Production speed and a clean root pass indoors
FCAW	E71T-1	Field-friendly deposition, tolerant of mill scale
GTAW	ER308L / ER316L	Stainless, thin wall, and root-critical work

SMAW is the fallback when space is cramped or the stock is ugly, which is why stick welding stays common in chiller plants. GMAW is faster but needs shielding gas and clean air, while FCAW tolerates dirty field conditions. GTAW gives the cleanest root, at the cost of slower travel and a stricter purge. All are covered for procedure and performance qualification in AWS B2.1/B2.1M, whose current 8th edition spans electrogas, electron beam, electroslag, flux cored, GMAW, GTAW, laser beam, oxyfuel, plasma arc, SMAW, stud arc, and submerged arc welding. For gases and consumables, this welding gas and supply guide covers the cylinder and shielding basics.

Step 6: Preheat, Interpass, and Heat Control

For B31.9 carbon steel in normal chilled water service, preheat is rarely required. It becomes necessary with thicker wall, higher carbon content, or cold weather that pulls pipe temperature down faster than heat can spread. The interpass limit applies on every joint and is stated in the WPS. Too hot gives a coarse, soft root and a cap that undercuts; too cold gives a hard, crack-sensitive heat-affected zone. Use stringer beads, keep travel speed steady, and let the joint cool between passes.

Step 7: Weld Out, Pass by Pass

Root, fill, cap. Between passes, chip off slag and clean the bead, including inside the bore. A flake of slag left in the pipe eventually sits on a chiller tube sheet. Watch root penetration closely: back gouge and grind out slag, porosity, or a cold root before filling over it. The cap pass should be smooth and only slightly convex, with reinforcement inside the limits of your WPS and no undercut. Undercut looks cosmetic but behaves like a machined crack starter. Use carbon arc gouging to remove a rejected root rather than welding over it.

Step 8: Handle Electrodes Like They Are Perishable

Electrodes come out of the oven and go back into it. UFGS 23 64 26 requires electrodes to be stored and dried per AWS D1.1/D1.1M or the manufacturer’s recommendation, and prohibits electrodes that have been wetted or have lost their coating. A damp E7018 gives you porosity, and porosity in a chilled water root is a leak waiting for a season to mature. Keep a rod oven or storage heater on site and discard anything dropped in a puddle or left open overnight. OSHA’s welding safety guidance covers the handling and fume side of this.

Step 9: Look at the Bead Before You Call It Done

Walk the finished weld and look for cracks, incomplete fusion, slag, porosity, overlap, excessive reinforcement, and undercut. A joint that fails visual gets gouged out, the area re-examined, re-welded, and re-examined by the same method. Doing it with the scaffold still up is cheap; doing it after insulation goes on is the 2 a.m. ceiling leak.

Qualification: Writing the WPS, Qualifying the PQR, and Certifying the Welder

Qualification is where chilled water pipe jobs collect their most expensive surprise, and it is almost never the welding itself. A crew can lay a beautiful 8-inch butt joint all afternoon and still get the job rejected because nobody can produce a current WPS, a matching PQR, and a performance qualification for the welder holding the stinger. The fix is not more paperwork for its own sake. It is three documents that line up.

Three Documents, One Chain of Logic

Think of qualification as a chain with three links. The Welding Procedure Specification (WPS) is the recipe: base metals, welding process, filler metal, groove design, preheat, interpass temperature, and post-weld heat treatment. ASME’s QW-482 form is the standard format for writing one. The Performance Qualification Record (PQR) is the test record proving the recipe works. The welder performance qualification proves a particular individual can make sound welds following that WPS. Break any link and the chain no longer supports the weld you are about to make.

Document	The question it answers	Who owns it
WPS (QW-482)	How will this joint actually be welded?	Contractor
PQR (QW-250)	Does this procedure produce conforming welds?	Contractor
Welder performance qualification	Can this welder make conforming welds to that WPS?	Employer

Under ASME B31.9, each employer is responsible for all welding performed by its employees, and the employer must supply its welders with a WPS describing base metals, process, filler metal, groove design, preheat, and PWHT. AWS B2.1/B2.1M:2026, now the 8th edition, says the same thing in plainer language: employers are responsible for the welding performed by their organization, including use of qualified welding procedures, qualified welders, and qualified welding operators.

Which Code Governs the Qualification

B31.9 does not invent its own qualification system. It requires qualification in accordance with ASME BPVC Section IX, and that is where the welder qualification standard and forms come from. When a referencing document calls for it, AWS D10.9 or AWS B2.1/B2.1M are acceptable alternatives. B2.1 also permits welding procedure and performance qualifications that meet the requirements of other codes and specifications, provided they also meet B2.1 itself. That last clause is why so many chilled water contractors run a Section IX qualification and still satisfy a specification that names AWS. Always confirm which edition the project adopted, including for B31.9 itself.

Qualifying the Procedure

Procedure qualification starts with test coupons cut from production pipe or plate. You weld them using the same process, filler metal, and parameters the WPS specifies, then examine what came out. A macroetch cross-section shows weld profile, root penetration, and fusion in a single look. Guided bend tests take strips cut from the coupon and bend them around a former, and the face bend and root bend must show no cracks or incomplete fusion beyond the allowable amount. Tensile testing pulls the weld apart to measure strength. The payoff for all of that is range: a procedure qualified on thicker material and a larger diameter generally covers a band of thinner and smaller production welds, so one qualified WPS can cover a large share of the system instead of one joint at a time.

Qualifying the Welder

The welder test is simpler to describe and easier to fail. The welder makes test welds in the positions the real work will demand, and those welds are examined by guided bend or macroetch. Position limits catch people out. A 6G qualification covers a range of pipe positions and diameters; a 2G or 1G qualification does not cover overhead work. If your crew has to reach up into a ceiling plenum to land a header, a flat-and-horizontal-only ticket will not cover it. Base metal and thickness ranges work the same way, with qualifying on carbon steel covering the carbon steel group and testing on thicker material covering thinner material. Consumable selection matters too, which is why shops keep a solid all-position welding rod guide on hand before qualification day.

Records, Continuing Qualification, and the Real Cost

Two administrative realities keep this honest. Records must be current and on file before production welding begins, not filed afterward once the inspector asks. And the money is real: the National Certification Board for Welding Examiners notes that in some areas of the United States it costs about $2,000 to qualify a welder. That cost is exactly why its welder interchange provision matters, letting qualified welders move between member contractors without repeating the qualification tests. Continuing qualification provisions exist for the same reason: a qualification is a snapshot of what a welder demonstrated on a given date, so welders who have been away from a process may need to requalify before they weld production work.

Pre-Production Checklist

Before a single production weld is lit, a supervisor should be able to produce:

1. A current WPS covering the joint, process, filler metal, and thickness range you will weld.
2. A matching PQR with macroetch, bend, and tensile results attached.
3. Welder performance qualifications valid for process, base metal, thickness, and position.
4. Records physically on site, not back in a trailer folder or an email.
5. Electrode and consumable storage and drying practices matching the WPS.
6. The purge procedure written down, with gas on hand and a flowmeter ready.
7. A plan for keeping qualification current as the crew rotates and grows.

None of this guarantees a weld that will survive a decade in a closed loop. What it does is remove the embarrassing possibility that the work was beautiful and undocumented, which is the version of failure that turns into a dispute instead of a repair.

Proving the Weld: Visual Inspection, NDT, and the Hydrostatic Test

For chilled water pipe welding, the last pass of a butt weld is not the finish line. With hundreds of feet of piping often hidden behind walls and above ceilings, verification is mandatory. To prove a leak free pressure boundary, three layers of examination must be satisfied: visual inspection, nondestructive examination (NDT), and the hydrostatic test.

Layer One: Visual Examination

Before any gauges or specialized equipment are brought out, a qualified inspector will conduct a visual examination. For the majority of building service piping, this is required under ASME B31.9 and UFGS 23 64 26. Attention is focused primarily to the root, toe, and cap of each joint for any surface breaking discontinuities.

The most frequent causes for rejection are cracks, incomplete fusion, slag inclusions, excessive reinforcement, undercut, and overlap. Though small at the surface, any of these flaws can become a starting point for fatigue or corrosion over the life of a chilled water loop.

Layer Two: Nondestructive Examination (NDT)

While visual inspection covers the surface, many of the most detrimental flaws lie beneath it. For this reason, NDT associated with weld examination must be conducted prior to pressure testing by an AWS Certified Weld Inspector (CWI/SCWI) or by NDT personnel qualified for the test method per SNT-TC-1A.

Not every joint will require NDT, however, owners will often mandate it for critical pressure boundaries, hot taps, or high-consequence systems such as hospitals and data centers. For chilled water scope, the three methods most commonly used are:

• Radiographic Testing (RT): Utilizing x-ray or gamma radiation, RT produces a permanent image of the internal weld profile. Inspectors look for linear indications such as lack of fusion and cracks, as well as rounded indications such as porosity and slag.
• Ultrasonic Testing (UT): High frequency sound waves are used to locate volumetric and planar flaws. UT is especially effective for thicker carbon steel butt welds and is well suited for long, straight runs.
• Magnetic Particle Testing (MT): Limited to ferromagnetic materials, MT is best used to examine weld toes, attachments, and surface connected flaws that may have been missed during visual inspection.

Much like the strict acceptance criteria required for precision work within aerospace welding, NDT results must be evaluated against the applicable code before moving to pressure testing. A passing NDT report is clearance to pressurize, not a formality.

Layer Three: The Hydrostatic Test

Once all required visual and NDT examinations have been accepted, the entire isolated segment is proven leak tight through hydrostatic testing. Water is the preferred test medium due to its incompressibility, allowing for a much safer and more controllable test than compressed gas.

Per ASME B31.9, every point in the system is required to be hydrostatically tested to 1.5 times the design pressure or greater for a minimum of 10 minutes. For institutional work, these minimums are frequently increased. NIH design requirements call for a 4-hour minimum hydrostatic test on all hydronic systems, and prefer ASME B40.100 Grade 1A gauges with at least a 3-inch face and error not exceeding 0.75 percent of test pressure.

Pressure alone isn’t the only limit. Test stress at the bottom of vertical runs must not exceed 90 percent of the specified minimum yield strength or 1.7 times the SE value in Appendix A of ASME B31.9. This accounts for the hydrostatic head of the water column to prevent overstressing components at the lowest elevations.

For a safe and controlled test, the following must be observed:

• Gauge Selection: Calibrated gauges must conform to ASME B40.100 Grade 1A or better, with at least a 3-inch dial, a range of at least 1.5 times the test pressure, and combined calibration and reading error not exceeding 0.75 percent.
• Relief Protection: A safety valve set no more than one-third above the test pressure is required to prevent accidental over pressurization.
• Air Removal: All air must be vented at high points prior to raising pressure. Trapped air is compressible, can mask leaks, and creates a serious hazard in the event of a rupture.
• Exposure: All joints must remain uninsulated and exposed for the duration of the test and inspection. Concealing a joint prior to a successful hold is one of the quickest ways to turn a pinhole into a major repair.
• Component Limits: Test pressure must not exceed the maximum pressure rating of any component within the isolated segment, including valves and fittings.

During the hold, the inspector will methodically walk every exposed weld looking for weeping, sweating, or pressure loss. After a successful hold, review of all test documentation, and clearance from the inspector, the system may be drained, flushed, and prepared for insulation. Visual inspection weeds out the obvious, NDT uncovers the hidden, and the hydrostatic test validates the entire pressure boundary as a whole. Together, these three steps are what keep a chilled water system watertight for decades to come.

The Mistakes That Cause Failures — and the Fixes That Prevent Them

Most chilled water leaks are not caused by a careless welder. They are caused by a good weld made in the wrong conditions, with nothing downstream capable of catching it. These failure modes keep repeating in plant rooms nationwide. Each has a cause you control, a consequence you can predict, and a way to find it before it finds you.

The Twelve That Show Up Most

1. Skipping or botching the purge. Without inert gas sweeping the bore, the root pass oxidizes and internal scale forms; the bead looks perfect from the outside and pits internally within a season. Nickel and stainless filler metals oxidize at the root faster than carbon steel fillers, so stainless lines need faster, tighter purging. Fix: fill the pipe and fittings with inert gas during welding, as UFGS 23 64 26 requires, and track oxygen on the purge flowmeter. Detect: borescope the bore after blow-down and look for scale.
2. Using nitrogen as shielding gas. Fix: reserve nitrogen for purging only and shield with argon or CO2, as explained in this guide to welding gases. Detect: a radiograph showing dense root porosity on a GTAW or GMAW joint points to nitrogen contamination.
3. Incomplete root penetration and lack of fusion. A wrong root gap, misaligned joint, or runaway heat input on thin wall leaves no sound root. Fix: set the root gap with a gauge, tack to correct alignment, and control heat input to the qualified WPS. Detect: radiography or ultrasonic testing shows incomplete penetration where a visual exam only shows a good-looking bead.
4. Porosity from wet electrodes, dirty pipe, or lost shielding. Wind strips the gas cover and the arc atmosphere contaminates the puddle. Fix: dry and store electrodes per AWS D1.1/D1.1M, clean the pipe of scale and grease before assembly, and screen the joint from wind; stick welding still wins in tight plant rooms. Detect: round indications on a radiograph, or visible oxidation and soot on the bead surface.
5. Undercut and excessive reinforcement. They sound cosmetic, but they are crack initiation sites and make every NDE interpretation harder. Fix: keep travel speed and work angle steady, then remove defects with carbon arc gouging and re-weld. Detect: visual examination, plus ultrasonic indications near the surface where reinforcement piles up.
6. Field mitering and notching. UFGS 23 64 26 prohibits field-fabricated direction changes and branch connections formed by mitering or notching pipe, and mitered elbows crack in service. Fix: use manufactured wrought fittings or shop-fabricated bends. Detect: visual check for mismatched wall thickness and an oval bore, confirmed by radiography.
7. Skipping NDE because the hydrostatic test passed. A hydro test proves leak-tightness at test pressure for a few hours, not long-term integrity. Fix: treat nondestructive examination as a separate requirement on critical pressure boundaries and hold welds for B31.9 acceptance criteria. Detect: RT or UT finds the subsurface defect the leak test never saw.
8. Insulating and concealing before testing. A weeping joint behind insulation and drywall becomes a ceiling disaster. Fix: keep every joint uninsulated and exposed for the duration of testing and inspection, and insulate only after accepted records. Detect: the extended hydrostatic hold itself, run before concealment rather than after.
9. Slag and mill scale left in the bore. Debris travels downstream and fouls the chiller tube sheet. Fix: clean each pass, including the inside of the bore, and remove all scale and foreign matter before assembly per the specification. Detect: strainer inspection after flushing and a plate strainer full of rust and weld debris.
10. Dissimilar metals and galvanic attack. Where copper, stainless, and carbon steel share a loop, the less noble metal corrodes first. Fix: install dielectric unions at transitions and keep water treatment running; brazing copper properly matters too. Detect: telltale pitting at the joint and ultrasonic wall-thickness readings trending thin.
11. Expansion stress cracks at branches and anchors. The branch connection sees the highest stress in a moving run, and over-restrained headers split at the toe. Fix: design loops and offsets before the welder shows up, and do not weld a branch onto a fully restrained header. Detect: hairline cracking near the anchor on visual exam, confirmed by radiography.
12. Water left standing in the bore during weld-out. Trapped liquid flashes to steam, blows out the root, and rusts the interior behind the bead. Fix: drain and blow out the line, and vent before the arc comes on. Detect: porosity and rust blowout visible on borescope inspection of the completed joint.

None of these twelve failures is exotic. All are settled with prep, discipline, and a hold point in the right place. Handle them on the front end and the 2 a.m. leak never has a chance.

Field Wisdom: Productivity, Cost Control, and the Pre-Startup Checklist

Ask a chiller-plant welder what separates a smooth week from a rough one and you will rarely hear about the arc. You will hear about access, light, fit-up, and the paperwork that must close before anyone can put insulation over a joint. That is where the money is saved or lost.

Working condition	Process	Reason it wins
Tight plant room, awkward height, limited electrode swing	SMAW with E7018	No wire feeder, no spatter cleanup, tolerant of dirty steel and awkward positions
Open floor or shop spool with room for a gun lead	GMAW with ER70S-6	High deposition, clean root, minimal slag and cleanup time
Cold, windy, dirty field conditions	FCAW with E71T-1	Works in open air where wind shielding is hard to hold, tolerant of surface condition
Stainless or root-critical joint	GTAW	Control over root pass and heat input, which stainless sensitization demands

That is why the case for stick welding still holds up in real plant rooms, and why consumable selection decides arc time as much as amperage does. A shop with a good all-position rod lineup and spare machine leads beats one crew with a single perfect setup and nothing spare.

Shop Spools, Strong Root Passes, and Controlled Travel

Whenever the geometry allows, fabricate in the shop: better light, real ventilation, a comfortable welding position, and the chance to fit up and tack a spool square on a table instead of hanging it off a scaffold. Chilled water mains are mostly straight runs and elbows, which makes them ideal spool work.

In the field, three habits carry the productivity. Spend time on fit-up: a clamped joint with a consistent root gap lets a welder lay a root pass in one clean movement instead of chasing a wandering puddle. Invest in the root pass: a strong, fully penetrated root speeds every pass above it because nobody stops to back-gouge. Control travel speed deliberately. Too fast gives a wide, cold, undersized bead; too slow a fat, wavy reinforcement that costs grinding time later.

The Cost Conversation: Verification Now Versus Rework Later

Here is the arithmetic experienced contractors run in their heads. Qualifying one welder costs roughly $2,000 in some regions, and a procedure qualification plus its paperwork trail costs a comparable amount. A single rework cycle on a failed joint, meaning removing insulation, cutting out the weld, re-fabricating, re-testing, and re-insulating, routinely costs several times that. Testing early is cheap insurance. Testing late is a construction event.

That is also why the difference between a 10-minute code minimum and a widely adopted 4-hour institutional hold is not a technical debate. UFGS 23 64 26 and owners’ specifications set a 4-hour hold because of historical leakage behavior a 10-minute test does not expose. Both can be code compliant. One is a risk decision.

Supports, Anchors, and Why Branch Connections Crack First

A weld can be textbook perfect and still fail, because the loads around it were never resolved. Support, hanger, guide, anchor, and seismic-bracing layouts are specified using MSS SP-58 type designations and spacing, and expansion compensation, whether loops, offsets, or expansion joints, arrives as an engineer-of-record submittal with calculated anchor and guide forces. None of that is field improvisation, and bending pipe to make it fit undoes the design.

Branch connections see the highest stress in the run because the main line is restrained on both sides while the branch tries to move with the loop, the offset, and the equipment it serves. Good fit-up, sound consumables, and a continuous bead matter more there than anywhere else.

The Pre-Startup Checklist

Pressure testing, flushing, and chiller-connection flow verification should be witnessed by the Engineer of Record or the commissioning authority before anything is insulated or concealed. That is what keeps a small weep from becoming a ceiling disaster.

Item	What good looks like
Documentation	Current WPS, PQR, and welder performance qualifications on file before production welding starts
Condition of the joint	No scale, grease, or mill scale in the bore; correct root gap and alignment; consumables dried and stored per AWS D1.1/D1.1M
Purge records	Purge documented, flowmeter readings captured, vent open at the far end for each joint
Visual and NDE results	Visual report signed; radiograph or UT results closed and accepted
Hydrostatic test records	Test pressure, hold duration, gauge calibration, and no visible leakage recorded
Flushing and strainers	Temporary fine-mesh startup strainers at each chiller evaporator and condenser connection during flushing, then cleaned and swapped for the permanent screen before startup
Water treatment	Corrosion inhibitor and oxygen scavenger dosed, and the chemical charge confirmed before filling for service
Final decision to insulate	All inspection, testing, and witnessing complete and signed off before insulation or concealment

Conclusion

Every item on that list is cheap when done early and expensive when done late, and each traces back to one habit good crews practice without being asked: leave the joint better documented and better protected than you found it. The hospital chiller room in the opening story did not fail because of bad steel or a lazy welder, but because a small amount of water found a path nobody checked for.

Frequently Asked Questions

What code governs chilled water pipe welding?

ASME B31.9, Building Services Piping, is the primary code for building chilled water, condenser water, and heating water systems, and it requires welding procedures and welders to be qualified in accordance with ASME BPVC Section IX. If the system falls outside the B31.9 pressure, temperature, or material limits, it moves to a higher-tier code such as ASME B31.1. Note that B31.9 does not cover refrigeration piping, potable water, sanitary and storm drainage, or fire protection.

Do you need to purge with nitrogen when welding chilled water pipe?

Yes — most specifications, including UFGS 23 64 26, require the pipe and fittings to be filled with an inert gas such as nitrogen during welding to prevent internal scale formation, which otherwise becomes a corrosion initiation site inside the bore. Critically, that same specification notes that nitrogen must not be used as a shielding or purge gas if it is directly exposed to the arc, because it dissociates to reactive atomic nitrogen and can increase porosity and embrittlement. Use nitrogen for purging only, and use argon or CO2 for shielding.

What hydrostatic test pressure and duration does ASME B31.9 require?

B31.9 requires hydrostatic testing at 1.5 times the system design pressure for a minimum of 10 minutes, with no leakage and no unacceptable pressure drop. Many institutional and government specifications raise the hold to 4 hours because of historical leakage problems. The test pressure must never exceed the maximum pressure rating of any component in the segment, and the stress at the bottom of vertical runs must stay under 90 percent of specified minimum yield strength or 1.7 times the SE value in Appendix A.

Is NDT required on chilled water welds?

Under ASME B31.9 and UFGS 23 64 26, visual examination is the default and nondestructive examination is performed only where the engineer of record specifies it. Owners typically add radiography or ultrasonic testing on critical equipment connections, plant-room mains, or joints near hot taps. When NDT is required, it must be performed before pressure testing, by an AWS Certified Weld Inspector or by personnel qualified to the test method per ASNT SNT-TC-1A.

Can I field miter chilled water pipe instead of using a fitting?

No — UFGS 23 64 26 explicitly states that field-fabricated direction changes and branch connections formed by mitering or notching pipe are not permitted. Branch connections must use manufactured pipe fittings that match the piping material and schedule, with butt-welded fittings to ASME B16.9. Field miters are a recurring source of cracks and long-term leaks because they put a notch in the pipe wall right where stress concentrates.

What electrode and process should I use for Schedule 40 carbon steel chilled water pipe?

SMAW with an E7018 electrode is the dependable general-purpose choice for butt welds in carbon steel chilled water piping, because it handles field conditions, has forgiving slag characteristics, and produces a ductile bead. GMAW with ER70S-6 wire is faster and cleaner where there is room for the gun and the machine, and FCAW with E71T-1 is a good field option. Whatever you choose, the process, filler metal, position, and thickness range must appear in a qualified WPS with a supporting PQR, and the welder must be qualified to that procedure.

Does the wall thickness of chilled water pipe change the qualification requirements?

It can, and it often does. AWS B2.1/B2.1M and ASME Section IX both allow one qualified groove weld procedure to cover a range of production thicknesses, and one qualified welder test to cover a range of diameters and positions — for example, a 6G performance test on a larger pipe can cover smaller pipe and a broad range of positions. However, thickness and position limits must be checked against the specific qualification tables before you commit a welder to a job, and fillet weld size limits apply to branch connections.

Why do chilled water welds fail years after they pass the hydrostatic test?

The usual culprit is internal scale or oxide left on the root pass because purging was skipped or poorly monitored. The joint passes a hydrostatic test because it is leak-tight at test pressure, but the inside surface is already compromised, and once the closed loop picks up oxygen and the water treatment program is imperfect, pitting corrosion starts at that defect and eats through the wall. The second common cause is mechanical: inadequate expansion compensation or over-constrained anchors put cyclic bending stress on a joint that was never designed for it.

What are the welding fume hazards for chilled water welders?

Welding carbon steel produces fume containing iron oxides and manganese, and the OSHA permissible exposure limit for welding fume is 5 mg/m³ as total particulate for an 8-hour time-weighted average. Manganese has a much stricter ACGIH threshold of 0.02 mg/m³ respirable, and excessive exposure is associated with Parkinson-like neurological effects. Welding stainless or chrome-plated components adds hexavalent chromium, a confirmed carcinogen, with an OSHA PEL of 5 µg/m³ and a NIOSH REL of 0.2 µg/m³ — so source control, local exhaust, and correct process selection matter more than the respirator alone.

When should chilled water piping be insulated, and does it matter for the welder?

Insulation goes on only after the hydrostatic test, the flush, and the witness inspection are complete. Hydrostatic testing must be completed before insulation is applied and before any portion of the piping is concealed, and all joints are to be left uninsulated and exposed for the duration of testing and inspection. For the welder, this means planning the sequence so that a joint that might need repair is still accessible, and it means checking for a sweating or weeping joint behind the insulation before the concealment is closed up.