Meadville Blower Pulsation — Global Argument Map

Air Products / Vitro Glass, Meadville PA  ·  K111-1 & K111-2 GL720L-2  ·  Review of Wood Group Report US04418-01-1

High impact finding
Medium impact finding
Low / supporting finding
Rebuttal / new argument
New / refined — current session
Key / decisive argument

Overall Claim

The Wood Group report US04418-01-1 contains elements that need discussion and clarification to ensure the robustness of its key findings and recommendations to ensure that the proposed dampening arrangement addresses the root cause of the failures. The 1/4 wave resonator may not be working due to its placement. Measurement data should be reviewed.

High impact findings
HIGH · #1
Branch resonance — not main pipe pulsation
The ~360 Hz dominant peak in PdA1/PdA2 cannot measure true pipe pulsation — the same peak would appear at the blower discharge if it were real. PdA3 and PdB1 show it does not.
Why does the blower discharge sensor show a clean LPF-dominated spectrum if 360 Hz is the dominant source?
HIGH · #2
Coherence analysis not performed
The causal chain pulsation → vibration → noise was never tested using coherence analysis, despite 48 simultaneous channels being available. This is the standard tool to test causation.
Why was coherence analysis not performed with 48 simultaneous channels available?
HIGH · #3
Figure 4.1.11 — correlation ≠ causation
Three different quantities at different locations plotted on one chart cannot prove one causes another — all share the blower as a common source and will peak at the same frequencies regardless of causal relationship.
How do you distinguish causation from common-source correlation in this chart without coherence data?
HIGH · #4
Structure-borne vibration path ignored
Fluid-borne pulsation treatment may not fix structure-borne vibration. The two paths are typically independent. Silencing the pipe will not stop vibration travelling through the steelwork.
What analysis was done of the structure-borne vibration path independently of the fluid-borne path?
KEY FINDING
HIGH · #5
Resonator junction at pressure node — acoustically inert
The 1/4 wave resonator centre is 1577mm from the blower discharge port. λ/4 at 59 Hz ≈ 1507–1560mm across the PSA cycle. The junction sits essentially at the pressure node throughout — acoustic pressure there is 1–10% of maximum. Energy absorption ≈ 0–1% of correctly placed resonator.
The acoustic analysis justifying its placement needs to be reviewed.
HIGH · #6
Inlet condition not measured
Inlet pressure was not measured during the survey. Blower starvation cannot be ruled out as a primary or contributing cause. An unmeasured quantity cannot be excluded from the causal analysis.
How was inlet starvation ruled out without measuring inlet pressure during the survey?
HIGH · #7
Mechanical condition not assessed
Progressive failure (bolt loosening, weld cracking) indicates chronic stress, not transient overload. Sub-harmonic content may indicate rotor asymmetry. Neither was assessed as a primary mechanical cause.
What is the explanation for the sub-harmonic at 29.6 Hz and what does it indicate about rotor condition?
Medium impact findings
MED · #8
Two units — different root causes
Units A and B have worst vibration at different ends of the machine. Same root cause would produce similar machine-level vibration patterns. They must have different primary causes.
MED · #9
Sensor fell off the pipe
A sensor that detached during measurement cannot be trusted. Readings produced while bouncing around are contaminated. The report does not identify which readings are affected.
MED · #10
1293 psi reading is physically impossible
The system operates near atmospheric pressure. 1293 psi would have burst the pipe catastrophically. This corrupted reading invalidates the entire test row in which it appears.
MED · #11
Day 2 data — measurement not machine change
The noise microphone — immune to measurement problems — shows no change between Day 1 and Day 2. The apparent change in pulsation data is an instrumentation artefact, not a machine behaviour change.
MED · #12
Clean Day 1 spectrum cannot be ignored
PdB1 Day 1 shows a clean LPF-dominated spectrum at 4.29 psi — the machine's true acoustic output. Ignoring the cleanest reliable data because it does not fit the narrative is not acceptable engineering.
MED · #15
Any pulsation study must be fed the correct input data
Any pulsation analysis based on the readings from PdA1/PdA2 data must take account of the dead leg resoannces to ensure that the study conclusions are reliable.
Lower impact / supporting findings
LOW · #13 & #14
There are 3 freqeuncies 179 / 359 / 538 Hz which need further expalanation
These three frequencies may form an exact harmonic ratio which would normally require a single physical explanation, the report does not provide one. Flo-Dyne identified possible rotor cavity resonance, but this is a weak argument.
LOW · #16
Noise microphone is primary evidence
The noise mic is immune to dead-leg resonance, sensor detachment, and PSA cycle corruption. It is the most reliable source in the dataset and should be given greater weight in the report as it directly reflects what is happening in the pipe.
LOW · #17
Resonator and dampener may not be able to coexist sensibly
If the 1/4 wave resonator were working, the definition of the dampener nmay change. Using two devices to targeting the same problem implies neither is expected to work, which appears contradictory.
LOW · #18
A pulsation dampener may not fix structure-borne vibration
A gas pulsation dampener stops airborne transmission, not structure-borne. The structural path through the pipe walls and supports is unaffected by changes to gas-phase pulsation.
LOW · #19
Inlet pressure drop not checked against specification
Inlet pressure drop is expected for a Roots blower but must be within Aerzen design limits. Nobody measured it during the survey. It remains an unexcluded variable.
PdA1 & PdA2 measurement validity — detailed argument map

Sub-Claim: PdA1 & PdA2 do not represent main pipe pulsation

The reported 9–18 psi at 356 Hz may well be amplified by a two-stage acoustic resonance in the measurement installation. In which case true main pipe pulsation at 356 Hz is approximately 0.3–0.6 psi — too small to warrant any remediation.

Arguments against validity
AGAINST
Dead-leg geometry — fundamental mechanism
Dead-leg resonates at ~356 Hz. The branch acts as narrow bandpass filter. The peak seems to be rock-solid across all 12 test conditions regardless of temperature, load, or which blower is running. A machine-driven peak would shift — this does not.
NEW · KEY FINDING
Two-stage amplification — quantified
Sensor sits at closed end of ½" NB tapping off the main dead-leg branch:

P_sensor = P_main × 2 × Q_branch × 2 × Q_tap

Stage 1 coupling (main pipe → dead-leg): ×2
Q_branch practical: 5–30 (central estimate 12)
Stage 2 coupling (dead-leg → ½" tapping): ×2
Q_tap if ½" tapping at λ/4: additional 5–8×

Conservative total: ≥24×
Data-implied total: ~50×
Area ratio main pipe / ½" tapping: 2448:1

True main pipe pulsation at 356 Hz: 0.3–0.6 psi
NEW · REFINED
Two-rotor discharge geometry — frequency-selective modal excitation
The GL720L-2 has two rotors, upper and lower, each with two lobes — giving two discharge locations at approximately 180° separation around the casing circumference. The rotors are geared 90° apart, so their discharge events are anti-phase at the LPF and odd harmonics.

At LPF (59 Hz) and odd harmonics: two anti-phase sources at 180° separation → breathing mode (0,1) cancels, spinning mode (1,1) reinforced. Single sensor reading at these frequencies is circumferentially position-dependent and unreliable.

At even harmonics (118, 237, 356 Hz…): sources move toward in-phase → breathing mode reinforced, spinning mode suppressed. Circumferentially uniform — single sensor more representative.

356 Hz ≈ 6×LPF (even harmonic) → breathing mode preferentially excited by the blower at this frequency. Wood Group's symmetry argument has partial validity at 356 Hz specifically — but see rebuttal.
NEW · DECISIVE
Spinning mode defence falsified by Wood Group's own data
Even if the breathing mode is driven at 356 Hz (even harmonic), a spinning mode contribution would produce sidebands. More importantly: if any propagating transverse mode were the dominant mechanism at PdA1/PdA2, the mode pattern rotating in the main pipe would amplitude-modulate the signal at the dead-leg junction, producing sidebands at:

356 ± 14.8 Hz (shaft)  or  356 ± 59.3 Hz (LPF)

PdA1 and PdA2 show a single clean peak and do not appear to show sidebands. This rules out any propagating transverse mode as the dominant mechanism. The only mechanism consistent with a clean single peak at a fixed frequency is a local fixed-geometry resonance — the dead-leg branch.
AGAINST
Plane wave cut-on — frequency-dependent validity
356 Hz is above main pipe cut-on of 262–283 Hz. At odd harmonics (59, 178, 296 Hz) the spinning mode dominates → single sensor is particularly unreliable — reading depends on unknown circumferential position of sensor relative to mode orientation.

At even harmonics including 356 Hz, breathing mode is more symmetric → single sensor is less unreliable — but amplitude is only ~0.3–0.6 psi in the main pipe regardless of mode shape.

In either case the 9–18 psi reading cannot be attributed to main pipe pulsation at 356 Hz regardless of modal structure.
AGAINST
Amplitude evidence — independent confirmation
PdA2 reads 15–18 psi at 356 Hz. PdA3 reads ~0.3 psi at 356 Hz at blower discharge. Factor ~50× higher downstream than at source — no propagation mechanism explains this. Only resonant amplification is consistent. PdB1 reads 4.29 psi at 59 Hz — the blower's true dominant output.
AGAINST
Peak Hold averaging — systematic overstatement
PSA cycle temperature sweeps resonance frequency ~18 Hz during each cycle. Peak Hold Max captures maximum at each bin independently — artificially broadening and elevating a sweeping resonance. True cycle-averaged RMS level significantly lower than reported.
AGAINST
PdB1 & PdA3 contradict PdA1 & PdA2 completely
Both blower discharge sensors show LPF at 59 Hz dominant. Neither shows significant content at 356 Hz. The dead-leg sensors and discharge sensors tell diametrically opposite stories. The trustworthy sensors (discharge) are consistent; the dead-leg sensors are contaminated.
Arguments for validity
FOR
Real pulsation at 356 Hz does exist
Dead-leg resonance requires real main pipe excitation at 356 Hz. The branch confirms 356 Hz exists in the main pipe at approximately 0.3–0.6 psi. This is real, present, and machine-generated — just amplified ~50× by the installation.
FOR
API 618 accepts single sensor measurements
Standard industry practice does not require spatial averaging. Wood Group are following normal methodology for this class of measurement campaign.
FOR
Within branch the field IS plane wave
½" NB branch cut-on well above 1000 Hz. At 356 Hz the branch interior is plane wave. Sensor accurately measures branch tip pressure. Error is in interpretation, not instrument accuracy.
FOR
356 Hz is an even harmonic — breathing mode driven
With two rotors at 180° separation and anti-phase timing, even harmonics (including 6×LPF ≈ 356 Hz) produce in-phase discharge events → breathing mode (0,1) is preferentially excited → circumferentially uniform pressure → single sensor is more representative at this specific frequency. The symmetry argument has some technical validity at 356 Hz.
FOR
Consistency across 12 conditions
All 12 test conditions show same dominant frequency. Consistent result is not random measurement error — it demonstrates a systematic physical phenomenon.
Rebuttals to the FOR arguments
REBUTTAL
Conceding 356 Hz exists ≠ conceding 356 Hz dominates
The branch is a bandpass filter with gain ~50 at 356 Hz and gain ~0 at all other frequencies. It cannot tell you whether 356 Hz is more or less severe than the LPF at 59 Hz. The filter creates an apparent dominance that does not exist in the main pipe.
REBUTTAL
API 618 assumes main pipe wall sensor
No standard endorses using dead-leg branch sensors as proxies for main pipe pulsation. The standard practice argument does not apply to this installation geometry regardless of modal uniformity.
REBUTTAL
Breathing mode at 356 Hz — but amplitude is still only ~0.3–0.6 psi
Even if the assertion that that the breathing mode is driven at 356 Hz (even harmonic), making the single sensor more representative at that frequency — the true main pipe pulsation amplitude at 356 Hz is still only ~0.3–0.6 psi. A circumferentially uniform reading of 0.3–0.6 psi does not justify a dampener. The amplitude argument is decisive regardless of modal structure.
REBUTTAL
No sidebands — modal arguments are moot at PdA1/PdA2
Whether spinning or breathing modes are driven in the main pipe is irrelevant to PdA1/PdA2 — both sensors show a clean single peak with no sidebands, confirming the dominant mechanism is fixed-geometry dead-leg resonance, not any propagating transverse mode from the main pipe. The modal discussion applies only to direct main pipe wall sensors such as PdA3.
REBUTTAL
Consistency proves geometry — not machine
Fixed resonance = fixed frequency across all temperatures and loads. A machine-driven frequency would shift slightly with conditions. The absolute consistency of 356 Hz is the signature of a fixed-geometry resonator — the dead-leg — not of any modal excitation from the blower.
Quarter-wave resonator placement — Key finding
KEY FINDING
Resonator junction at 1577mm — pressure node
From drawing: nozzle 250mm + spool 254mm + pipe run 1073mm = 1577mm from blower discharge port.

λ/4 at 59 Hz across PSA cycle:
ΔT=15°C → λ/4 = 1483mm → resonator 94mm PAST node
ΔT=47°C → λ/4 = 1558mm → resonator 19mm past node
ΔT=64°C → λ/4 = 1597mm → resonator 20mm before node

Acoustic pressure at junction: 1–10% of maximum throughout entire PSA cycle.
Energy absorption: 0.01–1% of correctly placed resonator.
The acoustic analysis justifying placement at 1577mm from the discharge port needs review.
KEY FINDING
Resonator is acoustically inert — confirmed by data
Figure 4.1.5 doesn't show a clear insertion loss notch at 59 Hz — the expected signature of a working resonator is absent. The resonator body itself vibrates at over 100g — it is being shaken structurally by the pipe, not absorbing acoustic energy. The device is correctly tuned (stub ≈ λ/4) but correctly positioned at the node — it cannot function.
CORRECTIVE ACTION
Optimal position: within 300–400mm of blower flange
Pressure antinodes at: 0mm (blower), 3010mm, 6020mm…
Resonator should be relocated to ≤400mm from the blower discharge flange where |P| ≈ 95% of maximum.

This is a piping modification, not a new device purchase. PULS XM model will confirm the node placement and demonstrate the benefit of relocation.

Anyone with a PULS licence and can run this model independently.
Was an acoustic model used to select the 1577mm junction position?

Global Conclusion

The measurement report contains five high-impact main areas thast warrant further evaluation. The dominant 356 Hz peak in PdA1/PdA2 is a dead-leg branch resonance artefact amplified by approximately 50× — the true main pipe pulsation at that frequency may likely be only 0.3–0.6 psi. The quarter-wave resonator was correctly identified and correctly tuned to 59 Hz but may have been placed at a pressure node of the standing wave it was designed to suppress — rendering it acoustically inert throughout the PSA cycle. The proposed pulsation dampener and silencer targets a measurement artefact at 356 Hz rather than the true dominant pipe pulsation at 59 Hz. The coherence analysis that would have revealed the causal structure was not performed despite 48 simultaneous channels being available.

On the question of source symmetry: the two-rotor geometry of the GL720L-2 means 356 Hz (an even harmonic of LPF) is driven by an approximately in-phase pair of sources, preferentially exciting the circumferentially uniform breathing mode. The symmetry argument has some validity at this specific frequency — but appears to not be conclusive because (a) the true amplitude in the main pipe is only 0.3–0.6 psi regardless of modal uniformity, (b) the absence of sidebands in the PdA1/PdA2 spectra confirms that the dead-leg branch resonance — not any propagating transverse mode — is the dominant mechanism at those sensors, and (c) at the LPF and odd harmonics the two anti-phase sources drive the spinning mode, making single sensor measurements at those frequencies particularly unreliable.

The primary corrective action is to relocate the resonator junction to within 300–400mm of the blower discharge flange — a piping modification that addresses the root cause. This can be confirmed and demonstrated using PULS XM software.

Meadville Global Discussion Map  ·  Ref: Wood Group Report US04418-01-1 review  ·