A shell-and-tube exchanger can pass hydrotest, meet its duty and run for months before the first tube leaks. When the pulled bundle shows bright wear bands at the baffle holes, flow-induced vibration is one of the first suspects.
Heat exchanger tube vibration is not simply a high-velocity problem. Whether shell-side flow forces become damaging depends on how fluid properties, velocity, tube geometry, support spacing and damping interact. That makes vibration a design problem as much as an operating one.
IN BRIEF
Flow-induced vibration is an interaction, not a velocity limit. Whether flow forces cause damage depends on tube natural frequency, support, damping and fluid properties.
Four mechanisms act in shell-side crossflow: vortex shedding, turbulent buffeting, fluidelastic instability and acoustic resonance. Research identifies fluidelastic instability as the most important for tube bundles in crossflow.
Unsupported span is a primary structural lever. For an idealized tube span, natural frequency falls roughly with the square of its length.
Damage ranges from slow fretting at baffles to rapid impact and fatigue, and can end in a through-wall leak. Plugging the failed tube does not, by itself, address the cause.
Every mitigation moves another variable, so vibration is best addressed while thermal and mechanical design can still be iterated together.
On this page: causes · mechanisms · response · baffles · problem areas · damage · evaluation · codes · design changes · operating changes · repeated failures · FAQs
What causes heat exchanger tubes to vibrate?
Heat exchanger tubes vibrate because shell-side fluid flowing across them produces fluctuating forces, and each tube span is an elastic beam with its own natural frequencies. This is flow-induced vibration. Some motion is normal; the question is whether it stays harmless over the exchanger's life.
Three things are worth keeping separate:
Excitation mechanisms are how flow produces force: vortex shedding, turbulent buffeting, acoustic resonance.
Structural response is how the tube moves, including fluidelastic instability, a coupled fluid-structure instability.
Damage mechanisms are what the motion does: fretting, fatigue and tube-to-tube impact, which can end in leakage.
The main concern is shell-side crossflow. Tube-side flow is axial, and turbulence is generally its principal excitation, though pulsating or two-phase tube-side flow may warrant separate review. Tube-side fluid also adds mass, lowering natural frequency.
What are the four flow-induced vibration mechanisms?
Ziada's review classifies the mechanisms acting on tube bundles in crossflow as vortex-shedding resonance, acoustic resonance, turbulent buffeting and fluidelastic instability.
| Mechanism | What drives it | Where it matters | Typical concern |
|---|---|---|---|
| Vortex shedding | Periodic wake shedding; frequency set by velocity, diameter and array geometry | Liquid crossflow; in gas, as a trigger for acoustic resonance | Resonance if shedding locks onto a tube natural frequency |
| Turbulent buffeting | Random, broadband pressure fluctuations | Crossflow service generally | Generally small-amplitude motion; long-term fretting at supports |
| Fluidelastic instability | Coupled fluid-structure interaction; unstable above a critical velocity | Crossflow; more likely at low natural frequency and low damping | Steep amplitude growth: tube-to-tube impact, severe support wear, fatigue |
| Acoustic resonance | Standing sound waves across the shell, excited by flow periodicity such as shedding | Gas and vapor shell-side service | Intense tonal noise; vibration and fatigue of shell components |
Vortex shedding
Vortices detach alternately from each side of a tube's wake and produce a fluctuating lift force. The shedding frequency follows f = St × V / d, where V is a reference flow velocity, such as the gap velocity between tubes, and d is tube diameter. In a bundle, the Strouhal number St depends on layout, pitch ratio and Reynolds number.
When shedding frequency approaches a tube natural frequency, the tube can lock in and resonate, and rapid damage is possible, especially in liquids. In gas crossflow, AECL researchers found wake-shedding resonance of the tubes seldom a problem; there it matters mainly as a driver of acoustic resonance. Avoiding a frequency match does not address the other three mechanisms.
Turbulent buffeting
Turbulence produces random pressure fluctuations across a broad frequency band, so no precise frequency match is needed: the tube responds mainly near its own natural frequencies.
Buffeting amplitudes are typically small compared with fluidelastic instability, and design guidance treats buffeting mainly as a long-term fretting concern rather than a short-term failure risk. Evaluation therefore estimates the response and resulting wear rather than trying to avoid a single frequency.
Fluidelastic instability
Fluidelastic instability is a coupled fluid-structure instability. Tube motion changes the flow around the tube and its neighbors, and the resulting motion-dependent fluid forces can feed energy back into the tube. Below a critical velocity, damping dissipates that energy. Above it, the flow supplies more energy per cycle than damping removes, and amplitude can rise steeply with velocity.
A bundle can therefore run quietly at one flow rate and suffer tube-to-tube impact at a modestly higher one. AECL researchers identified it as the most important vibration excitation mechanism for tube bundles in crossflow, and design practice aims to keep every span below the threshold with margin.
Critical velocity is estimated with empirical correlations descended from H.J. Connors' 1970 work. The threshold rises with natural frequency and with tube mass and damping relative to shell-side fluid density. Its constants depend on layout and pitch, and test data scatter widely, so design values are conservative. There is no universal critical velocity.
Acoustic resonance
Acoustic resonance is a resonance of the gas in the shell. In gas or vapor crossflow, standing sound waves can form across the shell, normal to the tube axes and the flow. When vortex shedding or another flow periodicity coincides with one of those modes, the result can be an intense tone that Ziada notes may be loud enough to cause structural failure.
Because tube natural frequency does not control it, a bundle that passes its mechanical vibration checks can still resonate acoustically. Detuning baffles installed parallel to the flow and tube axes are an established remedy, positioned to change the frequency or shape of the specific acoustic mode.
What sets a tube bundle's vibration response?
Four things set the response: natural frequency, support conditions, damping and tube layout.
Natural frequency and span. A tube span behaves like a beam: frequency rises with stiffness and falls with effective mass and, most strongly, with span. For an idealized span it scales with 1/L², so doubling the span cuts frequency to roughly a quarter. Changing tube diameter or wall thickness generally moves it less.
Effective mass includes the tube metal, tube-side fluid and an added mass of shell-side fluid. In fixed-tubesheet units, differential expansion can put tubes in axial compression, which lowers natural frequency, so axial tube load belongs in the vibration evaluation.
Supports and damping. A tube in a drilled baffle hole is supported only while it contacts the hole, and in liquid service research attributes much of a bundle's damping to that tube-support contact. If clearance is large, or has worn larger, the support can become effectively inactive and the vibrating span lengthens.
Layout and pitch. Layout angle and pitch ratio change gap velocity, added mass and the correlation constants, so no layout is best for vibration in general.
How baffles influence tube vibration
Segmental baffles direct flow across the tubes, set much of the shell-side pressure drop, support the tubes and fix the unsupported span. Changing them for any one of those reasons changes the other three.
Spacing works on both sides of the problem. Closer spacing shortens spans and raises natural frequency, but it also reduces crossflow area and raises crossflow velocity. In a simplified screening view of tubes supported at every baffle, the frequency gain can outweigh the velocity increase and improve fluidelastic margin. That outcome is not guaranteed: window tubes, end spans, U-bends, shedding frequencies, buffeting response and pressure drop all change too, so each spacing change has to be rechecked.
Window tubes carry the longest spans. With single-segmental baffles, tubes in the window pass through only every other baffle, so their span is twice the baffle spacing. Argonne laboratory tests on baffled bundles found the tubes most strongly subjected to fluidelastic instability in the window region.
No-tubes-in-window (NTIW) designs remove tubes from the windows so every tube is supported at every baffle, and they allow intermediate support plates between baffles. The trade-off is a larger shell for the same tube count. Double-segmental baffles can reduce crossflow velocity to roughly half that of single-segmental baffles at comparable spacing, depending on geometry, with a lower heat-transfer coefficient.
More baffles are not automatically the fix. They raise velocity and pressure drop, and a design already at its allowable pressure drop may then need a larger shell. The Heat Exchange Institute calls baffle design and spacing the single most effective way to reduce vibration risk, and specifying tighter tube holes than HEI or TEMA the least effective method.
Meeting a maximum unsupported-span limit is not the same as completing a flow-induced vibration evaluation, which also depends on shell-side velocity, fluid properties and damping.
Where vibration problems tend to develop
Vibration problems tend to develop locally, where velocity is highest or support is weakest. AECL researchers observed that they often occur in inlet regions and around sealing strips. Locations to watch:
Inlet and outlet zones with high local velocity
Window tubes, supported at every other baffle
U-bends, which can hold a U-tube bundle's longest unsupported length
Bypass lanes, sealing strips and obstructions such as impingement plates and tie rods
At the inlet, end baffle spaces can be longer than the central spacing to clear nozzles, placing long spans in the highest-velocity zone. An impingement plate protects the first tube rows against erosion; it does not by itself address vibration in the inlet zone.
Why two-phase service is harder to predict
Boiling, condensing and flashing change density, velocity, void fraction and flow regime along the flow path, and with them excitation and damping. Behavior in intermittent flow differs markedly from bubbly or annular flow, and single-phase correlations carry more uncertainty.
Nuclear steam generators differ from industrial exchangers in scale, support design and regulation, but they show how local two-phase conditions can drive instability. At San Onofre in 2012, an NRC inspection team recorded that the licensee traced tube-to-tube wear in replacement U-tube steam generators to fluidelastic instability. The identified causes were localized high steam-water velocity, high void fraction and insufficient anti-vibration bar contact force. Industrial reboilers, vaporizers and two-phase condensers call for the same attention to local conditions.
What happens when tube vibration becomes damaging?
Damaging vibration does not have to be violent. Low-amplitude motion can wear a tube through over years, while instability can cause impact and fatigue damage much faster.
Tube-to-baffle fretting and wear
Fretting is wear from repeated small-amplitude sliding and impact between the tube and the baffle hole. Wear depends on contact force and sliding distance, and turbulent buffeting alone can drive it. The result can be a localized band of wall loss roughly as wide as the baffle is thick, sometimes deeper on one side. Holes in thin or softer baffles can wear too, loosening support and letting damage accelerate.
Fatigue cracking
Larger amplitudes from lock-in or fluidelastic instability can produce cyclic bending stress high enough for high-cycle fatigue. Bending stress tends to be highest where the tube is effectively clamped: at the tubesheet face, or at a support where tight fit or corrosion products lock it.
Vibration in the first span also loads the tube-to-tubesheet joint, and leaking joints have been reported among vibration problems, though corrosion, cycling and fabrication quality cause them too.
Tube-to-tube collision
Tubes strike each other when mid-span amplitude closes the gap between neighbors, which points toward instability or lock-in rather than buffeting. The evidence is mid-span wear flats on adjacent tubes, away from supports. Argonne observed tube-to-tube impact in its laboratory tests, and tube-to-tube wear led to the San Onofre steam generator leak.
Through-wall leakage
Once the wall is breached, fluid generally flows toward the lower-pressure side, which may be tube side to shell side or the reverse, depending on the exchanger and operating mode. Depending on service, the result may be process or cooling-water contamination, product loss, reduced performance, a safety or environmental concern, or a shutdown. A leak may show only as a composition, level or pressure change.
How engineers evaluate vibration risk during design
Evaluation compares, span by span, what the flow can do to a tube with what the tube can tolerate:
Define every operating case, including turndown, upset, future rates and two-phase conditions.
Find local velocities at the inlet and outlet, in windows, near U-bends and in bypass lanes, not just bundle averages.
Calculate natural frequencies for critical spans, with effective mass, realistic supports and axial stress.
Check each mechanism, including the fluidelastic velocity ratio and, in gas service, acoustic frequencies. Revise geometry where a check fails.
Thermal design programs include vibration screening: HTRI's Xist screens for acoustic and mechanical tube vibration, and Aspen Exchanger Design & Rating includes vibration analysis. Screening flags locations; engineering judgment still decides whether the inputs reflect real operation and whether the correlations fit the geometry. Because the correlations rest on widely scattered test data, no method guarantees a design free of vibration damage; that is why margins are applied.
What ASME, TEMA and API 660 do and do not require
Codes, standards and project specifications play different roles:
ASME Section VIII, Division 1 sets requirements for the design, fabrication, inspection, testing and certification of the pressure vessel, and its design loadings (UG-22) include cyclic and dynamic reactions. Code compliance and a tube-bundle flow-induced vibration evaluation are separate deliverables, so the purchase specification should call for both where both are needed. ASME Section III includes nonmandatory flow-induced vibration guidance for nuclear components.
The TEMA Standards have included a recommended flow-induced vibration procedure since the 1988 edition. TEMA issued its 2026 Edition on August 1, 2026; confirm current provisions against the edition the purchase specification names.
API Standard 660 (Ninth Edition, 2015, with Addendum 1, 2020) does not require a flow-induced vibration analysis by default. Clause 6.1.2 requires the vendor to submit one if the purchaser specifies it, with informative guidance in Annex A. Some owner specifications that supplement API 660, such as IOGP S-614, require the analysis whenever the vendor performs or check-rates the thermal design.
The project specification decides whether an analysis is performed, by which method, for which operating cases, and who owns the thermal design.
Specifying TEMA construction does not, on its own, tell the fabricator whether a vibration evaluation is required or which operating cases it must cover. The purchase specification should state both.
How can tube vibration risk be reduced in design?
Fixes generally shorten spans or lower local velocity. Each has a price and must be rechecked against every mechanism:
Closer baffle spacing: shorter spans, but higher crossflow velocity and pressure drop
NTIW baffles with intermediate supports: shorter spans for every tube; larger shell
Double-segmental baffles: lower crossflow velocity; lower heat-transfer coefficient
Divided-flow (TEMA J) or larger shell: lower velocity; more nozzles, diameter or cost
Inlet changes (larger nozzles, distributor belts, impingement rods, added support): better distribution; more shell hardware or fewer tubes
Tube diameter, wall or pitch: changed stiffness, mass and gap velocity, plus surface area and cost
U-bend supports: shorter U-bend spans; effectiveness depends on maintaining contact with the tubes
Because every option moves pressure drop, heat transfer or size, vibration is cheapest to resolve while thermal and mechanical design can still be iterated together.
Can operating or equipment changes create a vibration problem?
Yes. A vibration assessment holds for the conditions and geometry it analyzed. Not every change creates a problem, but these deserve a check:
Higher throughput from debottlenecking or capacity creep. ExxonMobil researchers describe an industrial project where analysis found a 9% flow increase made serious tube vibration highly probable.
A different fluid, changing velocity, added mass, damping and acoustic frequencies
New vapor in a unit evaluated as single-phase
Fouling or blockage that diverts flow and raises velocity elsewhere
Worn baffle holes that lengthen effective spans
A retube in a different material or wall thickness, which changes natural frequency. HEI lists possible tube vibration and tube staking among design considerations for a condenser retube.
The useful first step is to find the original vibration basis and check the new conditions against it.
What should be investigated after repeated tube failures?
Plugging or replacing a failed tube isolates or restores that tube, but on its own it does not change the conditions that damaged it. Repeated or patterned failures deserve an evaluation of the mechanism before the repair is simply repeated.
Where a region of the bundle is vulnerable, the next tubes can fail quickly. From industrial experience, ExxonMobil researchers reported plug-and-run repairs followed by further failures within weeks or even days. In nuclear steam generator service, the NRC described a plugged tube that severed from vibration fatigue and affected adjacent tubes. A plugged tube stays in the bundle and may need evaluation or stabilization.
A retube does not automatically need a redesign, but where vibration caused the failures, a retube in kind restores the geometry that let the tubes vibrate.
Reading the failure evidence
Wear at a baffle does not prove vibration. Crevice corrosion and erosion-corrosion can attack tubes at the same places, corrosion products can lock tubes at supports, and corrosion can accelerate fretting.
The pattern helps separate these mechanisms. Wear bands at supports suggest fretting, mid-span flats suggest tube contact, and clustering at the inlet, windows or U-bends suggests flow-driven damage. Timing matters too: a change in rate, fluid or tube material shortly before the failures is a strong clue.
Eddy current testing for nonferromagnetic tubes, remote field testing for ferromagnetic tubes and ultrasonic IRIS can map the damage, though support-plate signals complicate interpretation. Options then range from tube stakes or stabilizers to operating limits, a bundle with revised supports, or a replacement exchanger designed for actual service.
Designing thermal performance and mechanical reliability together
Heat exchanger tube vibration is a system-level design problem. Flow creates excitation, the bundle responds dynamically, and support geometry and operating conditions decide whether that response stays acceptable.
Harris Thermal designs and fabricates custom shell-and-tube heat exchangers to ASME Section VIII, Division 1, with in-house thermal and mechanical design, and repairs and retubes existing units. For new or replacement equipment, the most useful early inputs are every operating case, including future rates and two-phase conditions, and the failure history of any unit being replaced. Vibration can then be addressed while the geometry can still change.
FAQs about heat exchanger tube vibration
What causes heat exchanger tubes to vibrate?
Shell-side flow produces fluctuating forces through vortex shedding, turbulent buffeting, fluidelastic instability and, in gas service, acoustic resonance. Whether they cause damage depends on natural frequency, span, supports, damping and fluid properties, not velocity alone.
What is fluidelastic instability in a heat exchanger?
A coupled fluid-structure instability: tube motion changes the flow forces on the tube and its neighbors, and those forces can feed energy back into the motion. Above a critical shell-side velocity, amplitude can rise steeply, causing tube impact, support wear and fatigue. The threshold depends on natural frequency, mass, damping, fluid density and array geometry.
How does baffle spacing affect tube vibration?
Spacing sets the unsupported span, and natural frequency falls roughly with the square of span. Closer spacing raises frequency but also crossflow velocity and pressure drop, so the net effect must be checked. With single-segmental baffles, window tubes are supported only at every other baffle, doubling their span.
Why do heat exchanger tubes wear at baffles?
Tubes pass through baffle holes with clearance, so vibration makes them slide and impact against the hole, thinning the wall in a band. Wear at a baffle does not prove vibration, though; crevice corrosion and erosion can attack the same locations.
Can tube vibration cause a heat exchanger leak?
Yes. Fretting, fatigue cracking or tube-to-tube impact can breach the tube wall, and first-span vibration can load the tube-to-tubesheet joint. Fluid then generally leaks toward the lower-pressure side, possibly contaminating process fluid or cooling water.
How can heat exchanger tube vibration be reduced?
By shortening spans or lowering local velocity: closer baffle spacing, intermediate or U-bend supports, NTIW or double-segmental baffles, a larger or divided-flow shell, or better inlet distribution. Each option changes pressure drop, heat transfer or cost and must be rechecked, so it belongs in the design stage.