Root cause failure analyses of sub synchroneous vibration in the centrifugal compressor. pdf

RCFA

1. Root Cause Investigation of Sub-Synchronous
Vibration in a Multi-stage Centrifugal Compressor
Jun Inai, Project & Development Engineer
Kawasaki Heavy Industries, Japan

2. Outline
•Introduction
•Findings
•Root Cause Analysis
•Modification
•Validation of modification
•Site confirmation test
•Conclusion
•Lesson learnt

3. Introduction
 During the commercial operation of an eight-stage back to back
GT driven centrifugal compressor locates at the end-user’s off-shore
platform, high level shaft vibration alarm under specific operating
conditions was reported from end user.
 According to the site evaluation test with dynamic measurements,
sub-synchronous vibration (SSV) was observed under higher load
conditions of high pressure compressor for every operating speed.
 This case study features the root cause analysis of SSV problem
using large scale unsteady CFD and the final result of site
confirmation test after improvement.

4. Findings (1)
Site Evaluation Test
GAS
COOLER KNOCK OUT
DRUM
Casing drain
GAS
COOLER
PT
TT
FT
TT
PT
DPT
KNOCK OUT
DRUM
TT
PT
FT
TT
PT
DPT
LP COMPRESSOR(LPC) HP COMPRESSOR(HPC)
XE
XE
ANTI SURGE VALVE ANTI SURGE VALVE
 Site evaluation test was conducted to understand the circumstances.
 Dynamic measurements of rotor vibration and discharge pressure at
HPC casing drain and downstream piping were implemented.
GT
DPT XE : Dynamic measurement points
Pressure transducer
Vibration transducer

5. Findings (2)
Operable Range and Shaft Vibration 1.0
1.2
1.4
1.6
1.8
2.0
2.2
2.4
2.6
2.8
3.0
3.2
3.4
3.6
3.8
4.0
0 1000 2000 3000 4000 5000 6000 7000 8000
Suction Volume Flow - m3/hr(corr)
Pressure Ratio
Pressure ratio

SSV not present
105% speed
100% speed
80% speed
90% speed
Suction volume flow
97.9% speed
Shaft vibration
alarm detected !!
SSV onset line
 Operable range was restricted by high level shaft vibration.
 SSV onset points correspond to increase of shaft vibration.
HPC operating map

6. Findings (3)
Sub Synchronous Vibration
 SSV around 20~30 Hz were dominantly present. They were
approximately 1/6~1/7 times the machine rotational speed.
 Same frequency of discharge pressure fluctuation were also
detected at casing drain and down stream piping.
 Is this a typical vaneless diffuser rotating stall ?
At first we suspected it as the most possible root cause.
80% speed 90% speed 100% speed
Observed radial shaft vibration (Y-NDE)
16μm
1N
=146Hz
28Hz
8μm
18Hz
20Hz 1N
=130Hz
1N
=164Hz

7. Root Cause Analysis (1)
Rotating stall check at vaneless diffuser inlet
Rotating stall at diffuser inlet is checked at the design phase
based on Senoo criteria. And it was re-confirmed that sufficient acceptable
margin were secured. Flow angle at diffuser inlet at design condition
0.0
2.0
4.0
6.0
8.0
10.0
12.0
14.0
16.0
18.0
20.0
22.0
24.0
26.0
28.0
30.0
32.0
34.0
36.0
0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
b2/r2
Flow angle at diffuser inlet (deg)
Flow angle of each stage of HELANG
Critical flow angle, rotating stall to occur
Stable
Unstable
HPC
Senoo criteria
Flow angle at diffuser inlet [deg]

LPC
 It indicates the root cause is not a rotating stall at diffuser inlet.
Impeller exit blade height / Impeller diameter b2/r2

8. Root Cause Analysis (2)
CFD Analysis of 8
th
stage
IGV
Impeller
(φd=0.0118)
Parallel wall
vaneless diffuser
Diffuser outlet
with spacer vane
Eye seal
Center seal
Discharge Volute
Shunt holes
Deflector
•Number of vane
- Impeller = 22 w/t splitter
- IGV = 16
- Spacer vane = 8
•Number of shunt holes = 3
•Rotating speed =9514 rpm
•Calculation time = 1day/rotation
Birdview from upstream
from 7R
to 4R
Center seal leakage
Rabbet fit
Large scale unsteady CFD analysis was carried out for the 8
th
stage.

9. Root Cause Analysis (4)
Static pressure fluctuation at spacer vane inlet Hz
mT
f
Hz
T
f
T
T
m
r 23
10145.441
11
23
10145.44
11
11.1
10145.44
10248.10
75
360360
3-
3-0
3-
-3














2 rev.
000000ddeegg
006600ddeegg
118800ddeegg
336600ddeegg
112200ddeegg
224400ddeegg
330000ddeegg
ΔT=10.248ms ΔT=10.248ms
ΔΔθθ==7755°°
1N
ΔT=10.248ms
T=44.145ms
TIME
LOCATION Hz
mT
f
Hz
T
f
T
T
m
r 20
103065.68
11
159
106.3065
11
88.7
103065.6
10248.10
75
360360
3-
3-0
3-
-3













 Hz
mT
f
Hz
T
f
T
T
m
r 21
10766.153
11
63
10766.51
11
31.3
10766.15
10248.10
75
360360
3-
3-0
3-
-3














Time & space distribution of static pressure at spacer vane inlet
Station VD5
000deg
090deg 270deg
180deg
Rotation
Circumferential position

20Hz 21Hz 23Hz
Close frequency as observed at site test could be simulated.

10. Root Cause Analysis (5)
Static pressure fluctuation at other stationary region
Static pressure time & space distribution & FFT spectra
Strong pressure fluctuation at spacer vane inlet affects to the other
stationary region.

11. N=17.5
TH
REVOLUTION N=18.0
TH
REVOLUTION N=18.5
TH
REVOLUTION










N=19.0
TH
REVOLUTION N=19.5
TH
REVOLUTION N=20.0
TH
REVOLUTION


Root Cause Analysis (6)
Unsteady pressure distribution across the stationary region
Static pressure distribution
Periodic pressure fluctuation was shown across the whole stationary
region.

12. Root Cause Analysis
Summary
A large scale unsteady CFD analysis achieved to simulate the
sub-synchronous phenomena as close frequency as measured
SSV at site test and indicates strong flow fluctuation due to
large flow separation at the diffuser outlet with spacer vane at
the final stage.
It was considered that the root cause is complete stall induced
from diffuser outlet due to excess flow passage expansion
between diffuser outlet and discharge volute at the final stage.
Therefore, configuration of ‘diffuser outlet with spacer vane’
shall be re-designed.

13. Modification
Improved diaphragm of 8
th
stage
Cross-sectional configuration of diffuser outlet was changed from expanded
shape to parallel wall shape
Spacer vane shape was also changed from cusped to elliptical blunt.
Spacer vane
Pre-modified Modified

14. Validation of modification (1)
Static pressure fluctuation at other stationary region
Static pressure time & space distribution & FFT spectra
Confirmed no presence of noticeable time & space distribution of static pressure
at spacer vane inlet and other stationary region.

15. N=17.5
TH
REVOLUTION N=18.0
TH
REVOLUTION N=18.5
TH
REVOLUTION










N=19.0
TH
REVOLUTION N=19.5
TH
REVOLUTION N=20.0
TH
REVOLUTION


Validation of modification (2)
Flow stability across the stationary region
Static pressure distribution
Pressure fluctuation at the stationary region completely disappear.

16. Validation of modification (3)
Rotor excitation force
Pre-modified Modified
View from upstream View from upstream
Rotor excitation force at the stage 8
th
occurs in the direction of
discharge nozzle.
Excitation force time averaged/dynamic have both decreased in
association with modification.
Dynamic Dynamic

17. Site confirmation test (1)
Operable range & Shaft vibration
Surge line
Confirmed wide operable range is secured as estimated
Overall vibration is less than 25μm for whole operable range
Pressure ratio

105% speed
100% speed
80% speed
90% speed
Suction volume flow
97.9% speed
93.4% speed
Former SSV onset line
Surge control line

18. Site performance test (2)
SSV presence
Pre-modified Modified
16μm
1N
=164Hz
28Hz
20μm
SSV  Negligible small (<1μm)
Confirmed no dominant SSV presence for all operable range.
5μm
1N
=160Hz

19. Conclusion
With regard to the natural gas export compressor on the off-shore
platform which was restricted its operable range due to SSV as 1/7
times the machine rotational speed, a large scale unsteady CFD
analysis was carried out in order to investigate the root cause.
The CFD analysis achieved to simulate those sub-synchronous
phenomena. And it was found that the root cause was a typical
stall at diffuser outlet due to excess flow passage expansion
between diffuser outlet and discharge volute at the final stage.
Modified stationary flow passage was designed and validated its
effectiveness by CFD analysis in the same manner as root cause
analysis.
Modified diaphragm was already installed to the site machine.
The followings were confirmed through the site evaluation test.
* No presence of dominant SSV for whole operable range
* Operable range is secured as estimated

20. Lesson & Learnt
Even the stalls in such a stationary flow passage region apart
from the rotor can be the excitation force of shaft vibration
especially under high pressure condition.
Sufficient consideration and care with a broad view shall be
taken during the engineering phase.