Flow-Induced Vibration Screening

Use this skill for public, educational flow-induced vibration (FIV) screening. It computes a fluid kinetic-energy index rho v^2 and compares it to a configurable kinetic-energy threshold so an agent can flag piping that may need a detailed Energy Institute style FIV likelihood-of-failure assessment before vibration design.

When to Use

• When a user asks whether a line could be prone to flow-induced vibration.
• When an agent needs a quick kinetic-energy triage to scope a piping vibration study.
• When examples must run without confidential piping classes, project line lists, or company piping specs.

Inputs

• fluid_velocity: actual flowing velocity in the line in m/s.
• mixture_density: flowing mixture density in kg/m3.
• kinetic_energy_threshold: screening kinetic-energy threshold in Pa, default 10000.0.
• small_bore_present: optional flag that a small-bore connection or thermowell is present, default False.

Outputs

• kinetic_energy_pa: fluid kinetic-energy index rho v^2 in Pa.
• threshold_ratio: ratio of the kinetic energy to the screening threshold.
• likelihood_of_failure_band: qualitative low, medium, or high band.
• fiv_warning: ok, watch, or high.
• small_bore_flag: True when a small-bore connection raises the screening sensitivity.
• assumptions: public assumptions used by the placeholder model.

Engineering Method

The Python class FlowInducedVibrationModel uses an open, published screening concept only:

• the fluid kinetic energy uses the widely published index FKE = rho v^2, the same quantity used as the primary driver in public Energy Institute style FIV likelihood-of-failure screening.
• the threshold ratio compares the kinetic energy to a configurable screening threshold.
• a small-bore connection flag lowers the effective warning thresholds because small-bore and thermowell connections are a common FIV failure location.
• the likelihood-of-failure band is a simple rule-based label derived from the threshold ratio.

This is educational and screening-only logic. It does not reproduce the proprietary Energy Institute Guidelines, scoring tables, or correction factors. It is not a vibration standard, a fatigue method, a modal analysis, or a replacement for a qualified piping vibration assessment.

Python Usage Pattern

from flow_induced_vibration_screening import FlowInducedVibrationModel

model = FlowInducedVibrationModel()
result = model.evaluate(
    fluid_velocity=20.0,
    mixture_density=60.0,
    kinetic_energy_threshold=10000.0,
    small_bore_present=False,
)

print(result.fiv_warning)
print(result.kinetic_energy_pa)
print(result.likelihood_of_failure_band)

If the optional neqsim Python package is available, the result records that fact so an agent can recommend moving to validated NeqSim property models for mixture density and velocity, followed by a detailed FIV assessment. If it is not installed, the example still runs with public placeholder logic.

Related NeqSim Functionality

NeqSim already implements a validated Energy Institute style FIV likelihood-of-failure model. Redirect real assessments to:

• neqsim.process.measurementdevice.FlowInducedVibrationAnalyser — likelihood-of-failure analyser attached to a pipe segment.
• neqsim.process.mechanicaldesign.manifold.ManifoldMechanicalDesignCalculator — acoustic-induced vibration (AIV) likelihood-of-failure for manifold piping.

This skill is a public rho v^2 triage layer that decides when to invoke FlowInducedVibrationAnalyser for a full assessment.

Calibrated LOF ratios when the line size is unknown

A very common real situation is that the design LOF is known but the line list is not: a project states "max rate X was set at LOF ≈ 1", yet the diameter, wall thickness and support-arrangement category cannot be retrieved. The assessment is still fully defensible, because the validated correlation is

LOF = rho_mix * v_mix^2 * FVF / F_v      with   F_v = alpha * (D/t)^beta

For two operating points on the same line, F_v is identical and the flow area A also cancels (since rho v^2 = mdot^2 / (rho A^2)), so

LOF_2 / LOF_1 = (rho_mix v_mix^2 FVF)_2 / (rho_mix v_mix^2 FVF)_1

is exactly independent of D, t and the support category. The recommended pattern is:

1. Reproduce the stated design point with an assumed geometry and record LOF_raw_anchor.
2. Report every other case as LOF = LOF_raw_case / LOF_raw_anchor * LOF_design.
3. Verify the cancellation numerically by re-running one case with a different setSupportArrangement(...) — the calibrated LOF must not move.
4. Use the assumed geometry only to report absolute velocities, and flag it as an assumption.

The same identity gives the operating envelope directly, since LOF ~ Q^2 at fixed pressure:

Q_allow(P) = Q_design * sqrt( LOF_design / LOF(Q_design, P) )

Sweeping P turns a single design rate into an allowable-rate-versus-pressure curve, which is usually what an operator actually needs.

Wet gas versus dry gas: which way does the driver move?

A recurring and consequential mistake is to assume that drying a wet-gas line makes flow-induced vibration worse. It makes it better. At the same standard gas rate and pressure:

• the mixture density falls a lot (entrained liquid is what makes the mixture heavy), while the velocity rises only a few percent, so rho v^2 falls slightly; and
• decisively, FVF drops from the two-phase branch (~0.3-0.4 at GVF 0.97-0.99) to sqrt(mu_gas [cP]) ~ 0.11 for a single-phase gas.

The net wet-over-dry driver ratio for a typical rich gas at 40-50 bara is about 3 to 4.

Sanity rule. If a calculation reports that removing liquid from a wet-gas line raises the LOF, the calculation is wrong. Check FVF first: it must fall, not rise, as GVF goes to 1. At GVF = 0.99 the two-phase branch gives FVF = 0.268, so a single-phase gas must come out below that. (A NeqSim defect that returned FVF ~ 0.61 for dry gas - an extra square root plus a Pa*s/cP unit mismatch - was found exactly this way and is fixed; the branch is now FVF = sqrt(mu_cP / REFERENCE_VISCOSITY_CP).)

The corollary that matters operationally

Because main-line FIV relaxes when a line goes dry, a wet-gas rate limit derived from FIV is conservative for dry service, and main-line FIV usually stops being the binding mechanism. Do not carry a wet-gas FIV rate derating into dry-gas operation. The dry-gas concern is a different mechanism - flow-induced pulsation of dead legs - see the hand-off below.

Hand-off: dry-gas service means screening dead legs, not re-deriving a rate limit

When a line is converted from wet gas to dry gas, screen closed side branches with neqsim.process.safety.vibration.FlowInducedPulsationScreening (quarter-wave branch modes, Strouhal lock-in band 0.2-0.6, mode-weighted severity). Three protections that wet gas provided all disappear at once:

1. Acoustic damping collapses - liquid films and droplets are strong absorbers, so a dry branch is a high-Q resonator.
2. The branch-mouth shear layer becomes coherent - wet/slugging flow continually disrupts it, which is what prevents sustained lock-in.
3. Liquid-filled legs empty - a condensate-filled drain resonates near 850-1000 m/s; once it drains to gas the sound speed falls to ~375 m/s and its modes drop by a factor ~2.3.

Run-pipe accelerometers are largely blind to branch pulsation, so a clean main-line vibration record does not clear this mechanism. Pair the two screenings whenever a dry-gas transition, an increased-velocity case, or a "we measured the main line and it was fine" argument appears.

Gotchas with the validated analyser

Field validation pattern

When permanent vibration probes exist, validate the driver before using LOF as a control parameter: correlate the measured velocity (mm/s rms) against a driver proxy built from historian tags, q^2 / P (proportional to rho v^2). A strong Pearson correlation confirms the response is flow-kinetic-energy driven rather than machinery or acoustic in origin, which is what justifies expressing the operating limit in rate and pressure.

Validation Checklist

• Inputs are positive and densities and velocities are in SI units.
• Example inputs are public and synthetic.
• Tests cover low, warning, high, small-bore, and invalid-input cases.
• Results are described as educational screening indicators.
• Real FIV assessment is redirected to validated methods, Energy Institute guidelines, and qualified review.

Common Mistakes

Limitations

• No proprietary Energy Institute scoring tables, correction factors, or fatigue calculations are included.
• No mechanical, modal, acoustic, or support-stiffness analysis is performed.
• No transient, slug, or two-phase intermittency excitation is modelled.

References

• NeqSim repository: https://github.com/equinor/neqsim
• NeqSim Skills Guide: https://github.com/equinor/neqsim/blob/master/docs/integration/skills_guide.md