The 'What' — What is a Thermowell?

A thermowell is a protective metal sleeve installed in a process pipe or vessel. It houses temperature sensors (such as RTDs or Thermocouples), isolating them from high fluid pressures, corrosive environments, and high-velocity flows.

By protecting the sensor inside its hollow core, the thermowell permits safe sensor removal for calibration or replacement without draining the line or shutting down the entire facility. However, because it extends directly into the path of the flowing fluid, it behaves as a cantilever beam, absorbing significant static and dynamic mechanical loads.

The 'Why' — Why Vortex Shedding Induces Vibration

As fluid flows past the cylindrical thermowell, it separates, creating alternating low-pressure swirls — a Kármán vortex street — detaching from opposite sides of the stem.

Each vortex detachment exerts an asymmetric dynamic force on the stem, alternately pushing it transversely (perpendicular to flow) at the wake frequency \(f_s = S_t \cdot U / d\) and in-line (parallel to flow) at \(2f_s\). The Strouhal number \(S_t \approx 0.22\) for Re > 1000.

The 'Which' — Which Failure Modes Occur? (Resonance Limits)

Every thermowell has a natural frequency \(f_n\) — the frequency at which it vibrates when disturbed. If the wake frequency \(f_s\) approaches \(f_n\), the thermowell enters resonance (lock-in), exponentially growing the vibration amplitude until fatigue failure.

ASME PTC 19.3 TW-2016 defines the frequency ratio \(r = f_s / f_n\) safety limits:

• Safe zone: \(r < 0.8\) — vortex frequency safely below natural frequency.
• Forbidden zone: \(0.8 \le r \le 1.25\) — highly amplified bending stress; FAIL.
• In-line limit: \(r < 0.4\) required for dense liquids to avoid in-line resonance at \(2f_s = f_n\).

The 'Where' — Where Do Thermowells Fail?

As a cantilever beam, the thermowell concentrates maximum bending stress at its root — the weld-neck-to-nozzle junction. Microscopic fatigue cracks initiate in the heat-affected zone (HAZ) of the root weld, propagating under cyclical loading until the stem fractures and enters the flow stream.

The bending moment formula: \(M = F_L \cdot L\), where bending stress at root \(\sigma = M \cdot c / I\). For a hollow cylinder, \(I = \pi(D_o^4 - D_i^4)/64\).

The 'How' — How to Resolve Sizing Failures

When \(r \ge 0.8\), the design fails the frequency check and must be corrected. Engineers apply one or more of these proven solutions:

1. Shorten insertion length \(L\): Since \(f_n \propto 1/L^2\), halving \(L\) quadruples \(f_n\). Most powerful remedy.
2. Increase stem diameters: A larger root diameter \(Q\) raises section modulus \(I\), stiffening the beam and raising \(f_n\).
3. Use stiffer material: Higher Young's Modulus \(E\) (e.g., Hastelloy at 205 GPa) raises \(f_n\) over lower \(E\) materials.
4. Angled / elbow mounting: 45° nozzle or elbow-facing installations reduce the effective normal velocity on the stem.
5. Helical strake fins: Three helical fins welded to the stem break vortex correlation and prevent resonant lock-in without changing \(f_n\).

Q1: What is the Strouhal number and why is it critical for thermowell design?

The Strouhal number \(S_t\) is a dimensionless parameter relating vortex shedding frequency to flow velocity and body diameter:

Rearranging: \(f_s = S_t \cdot U / d\). For \(\text{Re} > 1000\) (typical industrial flow), \(S_t \approx 0.22\). This means vortex shedding frequency is directly proportional to velocity and inversely proportional to the stem diameter — making the tip diameter \(B\) (the smallest dimension in the flow) the most critical dimension for vortex frequency calculation.

Q2: How is the natural frequency of a cantilevered thermowell calculated?

Using Euler-Bernoulli beam theory for a cantilever with a hollow circular cross-section:

Where: \(\lambda = 1.875\) (first mode cantilever shape factor), \(L\) = insertion length (m), \(E\) = Young's Modulus (Pa), \(I = \pi(D_o^4 - D_i^4)/64\) = second moment of area (m⁴), \(\rho_s\) = material density (kg/m³), \(A = \pi(D_o^2 - D_i^2)/4\) = cross-sectional area, and \(m_a\) = added (virtual) fluid mass per unit length.

Key insight: \(f_n \propto 1/L^2\) — doubling length reduces \(f_n\) by 4×.

Q3: What is the "added mass" correction and when does it matter?

When a thermowell vibrates in a fluid, it must also accelerate the surrounding fluid with it — this is the virtual (added) mass effect. It acts as an apparent increase in mass, reducing the natural frequency:

Where \(C_m \approx 1.0\) for a cylinder in unbounded flow. This correction is significant for dense liquids (e.g., water at 1000 kg/m³) but negligible for gases (e.g., steam at <10 kg/m³). Ignoring this correction for liquid service could overestimate \(f_n\) by 20–40%, leading to unsafe designs.

Q4: Why is the frequency limit set at r < 0.8 (not 1.0)?

The safety factor of 0.8 accounts for the phenomenon of "lock-in". When \(r\) approaches 0.8–1.2, vortex shedding does not simply continue at its natural Strouhal frequency — it synchronises (locks in) to the thermowell's structural vibration frequency. This dramatically amplifies the dynamic force and extends the resonance bandwidth far beyond a pure mathematical singularity.

Additionally, the safety margin accounts for:

• Uncertainties in fluid velocity (±15–20% in field conditions)
• Material property scatter (Young's modulus ±5%)
• Manufacturing tolerances in dimensional accuracy
• Temperature effects on \(E\) reducing \(f_n\) in service

Q5: How does operating temperature affect thermowell natural frequency?

Young's Modulus \(E\) decreases with rising temperature. The temperature correction applied in ASME PTC 19.3 TW-2016 is approximately linear:

For SS 316 at 400°C vs 25°C: \(E\) drops from 193 GPa to ~166 GPa — a 14% reduction. Since \(f_n \propto \sqrt{E}\), natural frequency drops by ~7.5%. This is why thermowell calculations must always use the operating temperature material properties, not room temperature values. At very high temperatures (>500°C), the material may creep, requiring the use of stress-rupture limits instead of yield strength.

Q6: What is the difference between static and dynamic stress checks?

Static (steady-state) stress is caused by the steady drag force of the fluid pushing the thermowell in the flow direction. The maximum bending stress at the root:

Dynamic (cyclic) stress is caused by the alternating lift force from vortex shedding. This cyclical stress, even at amplitudes below the yield strength, can cause fatigue failure over thousands or millions of cycles. ASME PTC 19.3 TW-2016 requires dynamic stress amplitude to remain below the material's endurance limit (typically ~0.4 × Ultimate Tensile Strength for steel).

Q7: When should a velocity collar (support collar) be used?

A velocity collar is a snug-fitting support ring located partway along the thermowell insertion length. It acts as an intermediate support point, effectively dividing the cantilever into two shorter spans. This changes the boundary condition from a simple cantilever to a propped cantilever, dramatically altering the mode shape:

Where \(L_{eff}\) is the effective (reduced) unsupported length. A collar at the mid-point can raise \(f_n\) by a factor of 4–8× over a simple cantilever. Collars are used when process velocity cannot be reduced and shortening the thermowell would result in the sensor tip falling out of the representative flow stream.

Q8: Why do stepped thermowells offer better performance than straight wells of the same insertion length?

A stepped thermowell has a thick, stiff upper shank (near the root) that transitions to a thinner lower shank (near the tip). This provides:

• Higher \(f_n\): The thick root provides high \(EI\), making the well stiff where bending moment is highest.
• Lower tip mass: The thin lower shank reduces the effective participating mass, further raising \(f_n\).
• Lower wake frequency: The smaller tip diameter \(d\) in the flow reduces \(f_s = S_t \cdot U / d\) — wait, a smaller tip actually increases \(f_s\). But since \(f_n\) rises faster than \(f_s\), the ratio \(r = f_s/f_n\) still improves.
• Better thermal response: The thinner tip reduces thermal mass and improves sensor response time \(\tau\).

Q9: Explain the hydrostatic (collapse pressure) check for thermowells.

A thermowell experiences full process fluid pressure on its outer surface and near-atmospheric pressure in the bore (where the sensor sits). This creates a net external pressure that can cause the thermowell to buckle or collapse like a tube under external pressure.

ASME PTC 19.3 TW-2016 checks that the process pressure \(p\) does not exceed the collapse rating:

Where \(t\) = wall thickness, \(\nu\) = Poisson's ratio (~0.3 for steel), and SF = safety factor (typically 4 for long-term service). For thin-walled thermowells in high-pressure service (>100 bar), this check can be the governing failure mode, overriding the frequency limit.

Q10: What are helical strakes and how do they suppress resonance?

Helical strakes are three continuous helical fins spiralling along the thermowell's outer surface at 120° pitch intervals. Borrowed from offshore riser engineering (where they suppress vortex-induced vibration on subsea pipelines), strakes work by disrupting vortex spanwise correlation:

• Smooth cylinders shed vortices coherently along their span — this creates a resonant lift force across the full length.
• Strakes cause the vortices to shed at different phases at different axial locations — the forces partially cancel, reducing net lift force by 70–90%.

Key design rule: strakes must span the full immersed length of the thermowell to be effective. Partial straking can sometimes worsen performance by creating a discontinuity that triggers coherent shedding over the unstraked portion. Strakes increase drag coefficient \(C_D\) by ~60–80% but eliminate the alternating lift force.