A shell-and-tube heat exchanger can have several thousand tube ends. Each one forms the boundary between shell-side and tube-side fluid, and tube-to-tubesheet joints are a recurring location for leakage and mechanical problems in service.
The choice is usually framed as expanded versus welded. That framing is incomplete: a third construction — expanded and welded — is common, and "welded" covers two things that carry different loads.
IN BRIEF
An expanded joint is mechanical. The tube is deformed plastically against the tubesheet hole, and the residual contact pressure there provides both retention and sealing.
A welded joint is metallurgical, and ASME Section VIII, Division 1 separates these welds by the design load each carries: full-strength, partial-strength, and seal. "Seal weld" is not a synonym for the other two.
Expanded-only joints are not defined by a universal pressure limit. Suitability depends on exchanger design, materials, service conditions and the applicable specifications.
The two get combined for several reasons — leak tightness, load sharing, closing a crevice, an owner specification — and the fabrication sequence follows whichever element carries the load.
Joint construction sets the repair path. Pulling an expanded tube is tooling work; pulling a strength-welded tube means machining the weld off the tubesheet face first.
On this page: what the joint does · the constructions compared · expansion and grooves · seal vs. strength weld · why both · service conditions · materials and clad tubesheets · over-expansion · inspection · retubing · selection · FAQs
What a tube-to-tubesheet joint has to do
A tube-to-tubesheet joint is the connection between a heat exchanger tube end and the tubesheet hole it sits in. The tubesheet separates the two fluid circuits, so the joint is the only barrier between tube-side and shell-side fluid at that location.
It has four jobs, and one construction does not automatically satisfy all four:
Retain the tube against pressure, differential expansion and bundle handling.
Separate the fluids. Tolerable leak rates vary by orders of magnitude between services.
Survive the local environment, at the transition between two chemistries and often in the tube's highest-turbulence region.
Stay serviceable. Someone will eventually plug a tube or retube the bundle.
A joint that retains the tube adequately can still leak. A joint that seals at hydrotest can loosen under thermal cycling. Treating retention and tightness as one requirement is the most common way this decision goes wrong.
Expanded, welded, and expanded-and-welded: the short answer
A mechanically expanded joint is made by deforming the tube outward into the tubesheet hole until the tube yields, while the surrounding tubesheet stays elastic enough to spring back against it. The residual interface pressure between tube OD and hole wall holds the tube and seals it. No filler metal, no fusion.
A welded joint fuses the tube to the tubesheet, normally at the tube-side face. The weld can be sized to carry the full axial strength of the tube, a defined lower load, or none of it.
An expanded-and-welded joint uses both, and they are not redundant: which one is credited with the design load changes the weld size, the expansion specification and the fabrication sequence.
| Construction | How it is made | Carries the load | Where it fits | What to watch | Retube |
|---|---|---|---|---|---|
| Mechanically expanded | Roller or hydraulic expansion into the hole, usually into machined grooves | Residual interface pressure | Where achievable retention and tightness suit the duty | Relaxation under cycling; sensitive to hole finish and expansion control; a crevice remains beyond the expanded length | Generally more straightforward |
| Seal welded (expanded first) | Expansion makes the joint; a weld adds tightness | The expansion | Expansion retains adequately, but the leak requirement is tighter | The expansion still has to be right; the weld does not substitute for it | Weld must be removed |
| Strength welded | Weld sized to a design load per the Code rules | The weld | High differential pressure, severe cycling, costly cross-contamination | Weld quality is the joint: fit-up, heat input, fusion | More involved; weld removal required |
| Strength welded + light expansion | Weld carries load; light expansion closes the tube-to-hole gap | The weld | Where the crevice behind the weld, or tube movement under vibration, is a concern | Sequence; clearance between expanded zone and weld | More involved; weld removal required |
How mechanical expansion works
Expansion is controlled plastic deformation. The tube is pushed past yield so it takes the shape of the hole; the tubesheet around it loads elastically and springs back when the tool releases. What remains is the joint — which is why the tubesheet has to be thick enough to store that springback, and why expansion is specified per material combination. Peer-reviewed testing across expanded, welded and welded-expanded joints shows why expansion requirements cannot be generalized: achievable joint strength depends on tubesheet geometry, material combination, expansion method and degree of expansion.
Roller expansion drives tapered rolls around the tube ID, working material outward progressively. It suits field work and retubing, and imposes axial elongation as the rolls advance. Hydraulic expansion pressurizes a sealed length of tube ID so the tube conforms to the hole without rolling contact — more uniform, less elongation, and more sensitive to hole surface condition.
What tube hole grooves do
A groove is a shallow circumferential recess machined in the tubesheet hole. Expanded tube material flows into it, interlocking the tube against pull-out and lengthening the path an escaping fluid must travel.
TEMA addresses grooving for expanded joints, keyed to tubesheet thickness — thicker tubesheets call for more than one groove — with separate treatment for hydraulic and explosive expansion. Take actual widths, depths and thresholds from the governing edition rather than a rule of thumb. TEMA also limits how far expansion may extend, so the expanded portion does not reach the shell-side face and remove the unexpanded collar supporting the tube where it enters the shell.
Grooves are not universally required: experimental work on hydraulic expansion has shown ungrooved tubesheets meeting pull-out requirements at sufficient pressure.
Seal weld vs. strength weld
These are not two names for the same detail. They describe which element carries the design load.
A strength weld is sized to carry a defined axial load. ASME Boiler and Pressure Vessel Code Section VIII, Division 1 addresses tube-to-tubesheet welds in UW-20, which separates them on exactly that basis. A full-strength weld is sized to carry at least the full axial strength of the tube. A partial-strength weld is sized to a lower design load established from actual design conditions, permitting a smaller weld where the service does not demand full tube strength.
A seal weld is not credited with carrying the joint's design load. Its function is leak tightness. In a seal-welded joint the mechanical joint — normally the expansion — retains the tube and is what the design calculation relies on.
So the terms are not interchangeable on a drawing. Specifying a seal weld while relying on the weld to carry load leaves the joint without a qualified load path. Specifying a full-strength weld where a partial-strength weld is adequate buys weld size, heat input and distortion the service never asked for.
Weld geometry follows the required load: fillet, groove, or a combination at the tube-side face. The minimum leak path — the shortest distance from the weld root to a free surface — is a governed dimension, and specifications such as API 660 set their own floor for it.
Why a joint would be both expanded and welded
Combining them is not automatically better. It answers a specific requirement:
The weld carries the load; a light expansion closes the gap. Without expansion an annular gap remains behind the weld — a crevice, and a place the tube can move under vibration.
The expansion carries the load; the weld provides tightness. Where expansion retains adequately but the tolerable leak rate is lower than expansion alone reliably delivers.
Thermal cycling, or the specification. Expanded joints can relax as interface pressure is worked by repeated heating and cooling; welds do not. Owner and EPC standards also mandate both outright for some service categories.
What the combination does not do is rescue a poorly executed element: a seal weld over a loose expansion has an unqualified load path.
Expand then weld, or weld then expand?
There is no universally correct sequence. Expanding first lets the expansion be verified before it is committed, and closes the gap at the weld root, which can influence fit-up and weld quality; the welding heat then works against the expansion it was placed over. Welding first protects the weld from being disturbed by expansion work, but gives up the ability to confirm expansion quality beforehand.
The resolution follows which element carries the load: protect expansion quality when the expansion is the mechanical joint, protect weld quality when the weld is the strength member. Either way, a clearance between the end of the expanded length and the weld is normally specified, so expansion-induced stress does not land in the heat-affected zone. Expansion carried up to the weld has been identified as a contributing factor in documented tube-end fatigue failures.
Service conditions decide it
Pressure and temperature alone cannot answer this. What moves the decision:
Differential pressure across the joint, not the higher of the two design pressures
Thermal and pressure cycling — startup frequency, load following, batch operation
Consequence and tolerable rate of cross-contamination, often controlling, and unrelated to mechanics
Differential expansion, far larger in a fixed-tubesheet unit than in a floating-head or U-tube
Flow-induced vibration at the tube end, and the owner's maintenance philosophy, including whether a retube is expected
Pressure by itself is a weak discriminator. The Heat Exchange Institute's guidance for closed feedwater heaters reports expanded-only joints performing reliably at 300–1,200 psig tube side. It suggests supporting the joint with a weld at around 2,000 psig and above, or where daily load transients, frequent cycling or tube-end erosion are expected. The transients are doing as much work there as the pressure.
Leakage and cross-contamination
A failed joint leaks from the higher-pressure side to the lower-pressure side. Depending on the exchanger, that may send tube-side fluid into the shell side or shell-side fluid into the tubes — showing up as a level rise, a composition shift, or nothing detectable for a long time. Tube-to-tubesheet leakage is frequently entirely internal, producing no external leak, because both sides of the joint sit inside the pressure envelope. Where cross-contamination is expensive, tolerable leak rate rather than mechanical load drives the choice to a welded construction.
Materials, clad tubesheets, and the crevice
For expansion, what matters is the relative yield and work-hardening behavior of tube and tubesheet. Expansion targets vary by tube material because yield behavior, ductility and work hardening differ significantly among alloys — the target belongs to the material combination, not to the expander. For welding, what matters is weldability of the pair and the qualified procedure — nominally similar alloys can still need a specific filler or buffer layer. The alloy selection logic belongs upstream of this decision, not inside it.
A clad or weld-overlaid tubesheet carries pressure in the backing material and corrosion resistance in a layer on the process face, and the joint is made into that layer. That introduces joint-design considerations a solid tubesheet does not have: the thickness of the corrosion-resistant layer, the joint configurations it allows, and the welding requirements that follow. TEMA and API 660 both address minimum overlay thickness at the tube-side face. In many clad or weld-overlaid configurations the tube-to-tubesheet weld is made at the corrosion-resistant layer, though the exact detail depends on the specified construction. A leak reaching the backing material behind the cladding can attack it from inside long before anything shows externally — clad construction logic, at a barrier the joint penetrates thousands of times.
Unless the tube is expanded through the full tubesheet thickness, an annulus remains between tube OD and hole wall over the unexpanded length. That annulus is a crevice: stagnant fluid, restricted oxygen transport, and aggressive species — chlorides in particular — concentrated well above the bulk stream. A tight joint can still have one, and crevice attack appears repeatedly in tube-end failure investigations. Full-depth expansion is one response, geometries that move the weld away from the crevice another. Both are design decisions, not adjustments available later.
Over-expansion and under-expansion
Expansion is specified as a target tube wall reduction. Both directions off target cause problems.
Under-expansion leaves insufficient residual interface pressure. The joint may pass a hydrostatic test and still loosen in service, and the shortfall is not visible from the tube-side face.
Over-expansion is less intuitive. Past a point, more expansion stops helping: experimental work found that increasing expansion percentage without limit does not continue to increase pull-out strength. Beyond that it thins the tube wall, work-hardens the material, can crack work-hardening alloys, and can distort the ligaments between adjacent holes.
The correct target depends on tube material, wall thickness, hole tolerances, tubesheet thickness and expansion method; TEMA and API 660 both give material-specific values. There is no single correct percentage, which is why expansion is measured and documented rather than judged by feel — and why hole diameter, finish and freedom from scratches matter as much as the expander. Harris Thermal performs tubesheet drilling and machining in-house, where hole quality is established or lost.
Inspection and testing
Examination scope belongs to the joint type, and Code-required examination should be kept separate from purchaser-specified.
Visual examination of every joint, before and after welding where applicable
Liquid penetrant examination, where specified, applied at intermediate and final weld stages to identify surface-breaking discontinuities while they are still repairable
Macro examination of qualification mock-ups under ASME Section IX QW-193, where sectioned and etched joints confirm fusion and minimum leak path
Shell-side pressure testing with the tube side open, a practical shop check of tube-to-tubesheet tightness, because leakage can be observed at the tubesheet face
Hydrostatic testing, which proves the pressure boundary but is not a sensitive small-leak test, and helium leak testing where the tolerable rate is below what hydrostatic or pneumatic testing resolves
Volumetric examination of these welds is limited by geometry, which is the practical argument for the mock-up: what cannot be examined in production must be proven in qualification. Harris Thermal's quality program covers liquid penetrant examination, positive material identification, helium leak testing and hydrostatic testing, with third-party inspection by an authorized agency.
Repair, plugging, and retubing
Joint construction sets the repair path — where a design decision gets paid for fifteen years later.
Plugging takes a failed tube out of service without removing it. Tapered and mechanical plugs are addressed in ASME PCC-2, and the applicable method depends on service conditions and on whether the joint is expanded or welded. Enough plugged tubes and the exchanger no longer makes its duty; Harris Thermal's guidance on repair versus replacement covers those thresholds.
Retubing is where the constructions diverge most. Removing an expanded tube is a tooling operation. Removing a strength-welded tube means machining the weld off the tubesheet face first, one tube at a time, before anything can be pulled. Repeated weld removal and machining can progressively reduce the available tubesheet face material, and each pass puts the holes at risk of damage.
None of that makes welded joints wrong. It makes the expected number of retubes a legitimate input to the original decision, alongside pressure and chemistry. Retubing and exchanger repair begins with identifying the existing construction, since repairs and replacement units both start from what is already in the tubesheet.
A selection sequence
Define both fluids, including contaminants, trace species and solids.
Establish design and operating pressure and temperature both sides, and the differential across the joint.
Determine the consequence of cross-contamination, and state the tolerable leak rate as a criterion rather than as the words "leak tight."
Fix tube and tubesheet materials, and whether the tubesheet is solid, clad or overlaid.
Evaluate thermal and pressure cycling, transients, and flow-induced vibration at the tube end.
Establish the axial joint load from the tubesheet design, including differential expansion on fixed-tubesheet units.
Select the construction and, for a welded joint, the weld category — full-strength, partial-strength or seal.
Define expansion method, length, grooves and wall reduction target; then examination and testing, separating Code-required from purchaser-specified.
Confirm against TEMA class, the Code of construction and the owner or EPC specification.
Review maintainability and expected retube history before the construction is fixed.
Steps 3 and 8 are easy to under-specify, yet both can materially change the joint selection. For replacement equipment, add the existing drawings and the tube-joint failure history — often the most useful document in the package.
Harris Thermal fabricates custom shell-and-tube heat exchangers and ASME Section VIII, Division 1 pressure equipment, with in-house mechanical and thermal design, tubesheet machining and drilling, orbital and manual tube welding equipment, and retubing tooling. Documented experience covers carbon and low-alloy steels, stainless and duplex stainless, 6-moly grades, nickel alloys, titanium, zirconium and clad materials. Where a joint specification is still open, or a unit keeps leaking at the tubesheet, the conditions and failure history are worth reviewing before the construction is fixed.
Tube-to-tubesheet joint construction is a small detail on an exchanger drawing. It can influence leakage risk, fabrication, inspection and repairability across the equipment's whole service life.
FAQs about tube-to-tubesheet joints
What is a tube-to-tubesheet joint?
It is the connection between a heat exchanger tube end and the tubesheet hole it sits in. Because the tubesheet separates the tube-side and shell-side circuits, the joint is the only barrier between the two fluids there. It must retain the tube, seal against leakage, survive the local chemistry, and stay serviceable for retubing.
Is it better to expand or weld heat exchanger tubes?
Neither is better in general. Expansion produces a mechanical joint whose strength comes from residual contact pressure; welding produces a metallurgical joint sized to a defined load. Welding may be selected where the tolerable leak rate is very low, cross-contamination is costly, differential pressure or cycling is severe, or an owner specification requires it.
What is the difference between a seal weld and a strength weld?
A strength weld is sized to carry a defined axial load — the full axial strength of the tube, or a lower design load from actual design conditions. A seal weld is not credited with carrying the joint's design load; its function is leak tightness, and the expansion retains the tube. ASME Section VIII, Division 1 treats these as distinct categories.
Why are some heat exchanger tubes both expanded and welded?
Because the two do different things. Where the weld carries the load, a light expansion closes the annular gap behind it, removing a corrosion crevice and preventing tube movement under vibration. Where the expansion carries the load, a seal weld supplies leak tightness expansion alone cannot guarantee. Thermal cycling and owner specifications also drive the combination.
Can a heat exchanger tube be over-expanded?
Yes. Past the target wall reduction, additional expansion stops adding strength: experimental work on hydraulically expanded joints found pull-out strength failing to increase with further expansion. It then starts causing harm — wall thinning, work hardening, cracking in work-hardening alloys, and distortion of the ligaments between holes.
Can welded tubes be replaced during a retube?
Yes, but the weld has to come off first. Each weld is machined or ground from the tubesheet face before the tube can be pulled, which adds time and consumes face material. Expanded joints can be more straightforward to retube, because there is no tube-to-tubesheet weld to remove — a reason to treat expected retube frequency as a design input.