USE OF SIX SIGMA TECHNIQUES IN GAS
SEAL PANEL TROUBLESHOOTING AND SEAL
FAILURE ANALYSIS
BY
ROBERT A. HUFFMAN
MANAGER, CENTRIFUGAL COMPRESSOR ENGINEERING
GE Energy - Conmec
AND
MATTHEW J. DIGIACOMO
CENTRIFUGAL COMPRESSOR DESIGNER
GE Energy Conmec
2004 Turbomachinery Symposium
Houston, TX
Six Sigma Overview
Six Sigma is a process for applying critical thinking to; solve
problems, improve processes, and improve products.
Six Sigma uses a collection of tools that have been used by other
Quality Initiatives, but focuses on a process driven implementation
of those tools.
DMAIC is the Key Acronym
> Define Clearly Define the Problem and Expectations
> Measure Gather Data to enable Conclusions / Decisions to be
made
> Analyze Analyze the Date to make Decisions
> Improve Implement a method to Improve the Situation
> Control Establish Controls to keep from repeating the past
Two Case Studies
First Case Study is Trouble Shooting a DGS Panel that Experience
Operational Irregularities
> Buffer Gas Supply Pressure Regulator was being driven to full
open due to low Delta P between buffer gas supply and buffer
chamber pressures.
> Customer afraid had lost the process labyrinth seal in the
compressor and facing an outage to correct.
Second Case Study is Trouble Shooting Multiple Dry Gas Seal
Cartridge Failures
> Customer experienced three outages at very short intervals to
replace DGS
> Failures did not appear consistent in symptoms of the failure
beyond high primary vent leakage
> After third failure, Conmec becomes involved to trouble shoot
Case 1 - GAS SEAL PANEL IRREGULARITIES
Step 1- Define the
Problem
FOLLOWING
INSTALLATION OF
NEW GAS SEAL
ASSEMBLIES,
UPON START-UP,
ALARM SIGNALS
LOW BUFFER GAS
SUPPLY PRESSURE
DELTA
STEP 2, GATHER DATA - Measure:
PRIMARY VENT
FLOWRATES
APPEAR NORMAL
PDT 1114 & PDI 1115
SHOW < 1 PSID OR
REQUIREMENT OF
>27 PSID.
PDI 1113 TRACKS
SIGNIFICANTLY
WITH CHANGING
BUFFER GAS
FLOWRATE
FILTER DP TRACKS
WITH CHANGING
BUFFER GAS
FLOWRATE, 1-5
PSID.
PDI 1113 INDICATES
50-180 PSID OVER
INLET PRESSURE
VERSUS 1-2 PSID
NORMAL
PERFORMANCE,
(P1, P2, T1, T2,
FLOWRATE)
APPEAR NORMAL
THRUST LOAD AND
POSITION ARE
NORMAL AND
STABLE
VIBRATION LEVELS ARE
NORMAL AND STABLE
SHOULD WE SHUTDOWN??
LOOK AT THE KNOWN FACTS
VIBRATION LEVELS APPEAR NORMAL AND STEADY
PERFORMANCE APPEARS NORMAL
GAS SEAL LEAKAGE APPEARS NORMAL
BEARING TEMPERATURES ARE NORMAL AND STEADY.
BUFFER GAS FLOW APPEARS NORMAL THROUGH THE FILTERS
IF A SEAL HAS NO BUFFER PROLONGED EXPOSURE TO PROCESS GAS WOULD
ALLOW DEPOSITS IN THE SEAL FACES TO BUILD UP LEADING TO SEAL FAILURE.
WITH THE CONTROL VALVE FULL OPEN, THE GAS SEAL SUPPLY, (COMPRESSOR
DISCHARGE) COULD NOT GENERATE ENOUGH PRESSURE DELTA IN THE SUPPLY TO
CAUSE SEAL DAMAGE.
ANSWER.UNLESS A SIGNIFICANT INCREASE IN PRIMARY
SEAL LEAKAGE RATE IS SEEN, NO!!*
*SHORTLY AFTER THIS DECISION, AN INCIDENT WITH THE DRIVER FORCED A SHUTDOWN.
STEP 2 Cont, BRAINSTORM:
DAMAGED PRIMARY SEAL
BOTH ENDS.
MISSING INBOARD BUFFER
CAVITY LABYRINTH SEAL
DAMAGED PRIMARY SEAL
DISCHARGE END ONLY
BLOCKED ORAFICE PLATES
FO1127 & FO1126 (ONE)
DAMAGED PRIMARY SEAL
INTAKE END ONLY.
BLOCKED ORAFICE PLATES
FO1127 & FO1126 (BOTH)
FAULTY BALANCE CAVITY
PRESSURE INDICATOR PI 1113
BLOCKED BALANCE LINE
FAULTY BUFFER SUPPLY
DELTA INDICATOR PDI 1115
FAULTY BUFFER SUPPLY
DELTA TRANSMITTER PDT 1115
BALANCE PISTON SEAL
DAMAGE
HIGH PRESSURE SIDE DELTA
TRANSMITTER VALVE BLOCKED
BLOCKED CONTROL VALVE
PDV 1114
FLOW BYPASSING PRESSURE
DELTA INDICATOR PDI 1115 & PDT
1114
MISSING INBOARD BUFFER
CAVITY LABYRINTH SEAL
LOW PRESSURE SIDE DELTA
TANSMITTER VALVE BLOCKED
3 = SOME INFLUENCE
9 = LARGE INFLUENCE
IN THE NEXT STEP WE
WILL CONCENTRATE
ON THE LARGE
INFLUENCES.
PI 1113 FAULTY
PDI 1115 FAULTY
PDT 1114 FAULTY
PDT 1114 & PDI 1115 FAULTY
PDCV 1114 BLOCKED
FO 1126 BLOCKED
FO 1127 BLOCKED
FO 1127 & F0 1126 BLOCKED
BALANCE LINE PLUGGED
BALANCE PISTON SEAL DAMAGE
DAMAGED SEALS, BOTH END (EITHER
ELASTOMER OR FACE DAMAGE, SINGLE
OR DUAL FACE FAILURE)
MISSING BUFFER LABYRINTH SEAL
EITHER OR BOTH ENDS
GLOBE VALVE BETWEEN PDCV 1114 &
PDI 1115 BLOCKED
LEAKAGE BYPASSING PDT 1114 & PDI
1115
NEEDLE VALVE BETWEEN PI1113 &
BALANCE CAVITY BLOCKED
COMPRESSOR
DISCHARGE
PRESSURE
PDT 1114, & PDI 1115
READING (BUFFER
SUPPLY DELTA)
PRIMARY VENT
FLOWRATE
DAMAGED SEAL DISCHARGE END
(EITHER ELASTOMER OR FACE
DAMAGE, SINGLE OR DUAL FACE
FAILURE)
DAMAGED SEAL INTAKE END (EITHER
ELASTOMER OR FACE DAMAGE, SINGLE
OR DUAL FACE FAILURE)
THRUST BEARING
DIRECTION AND
BEARING TEMP.
CAUSES
PI 1113 READING
(BALANCE CAVITY
PRESSURE REF)
COMPRESSOR
EFFICIENCY
1 = NO INFLUENCE
PDT 1108, PDI 1107
READING (FILTER DP)
IN THIS CASE WE WANT
TO RANK THE EFFECTS
BY THEIR LEVEL OF
INFLUENCE ON THE
OBSERVATIONS.
INFLUENCE
EFFECT:
9 = LARGE
3 = SOME
1 = NO INFLUNENCE
STEP 3 Analyze,
CAUSE AND EFFECT:
CAUSES
9
1
1
9
3
1
1
9
1
1
9
1
1
1
COMPRESSOR
DISCHARGE
PRESSURE
PDT 1114, & PDI 1115
READING (BUFFER
SUPPLY DELTA)
PRIMARY VENT
FLOWRATE
THRUST BEARING
DIRECTION AND
BEARING TEMP.
COMPRESSOR
EFFICIENCY
NEEDLE VALVE BETWEEN PI1113 &
BALANCE CAVITY BLOCKED
PDT 1108, PDI 1107
READING (FILTER DP)
FLOW BYPASSING PDT 1114 & PDI 1115
PI 1113 READING
(BALANCE CAVITY
PRESSURE REF)
INFLUENCE
EFFECT:
9 = LARGE
3 = SOME
1 = NO INFLUNENCE
STEP 4, ANALYZE
CONCENTRATING ON THE HIGH EFFECTS WE ENDED UP WITH ONLY 2
LIKELY CAUSES FOR OUR OBSERVATIONS.
Case 1 - SUMMARY, PANEL IRREGULARITIES
INSPECTION OF THE
COMPONENTS SHOWED
DEBRIS BUILDUP IN THE
NEEDLE VALVE ASSEMBLY.
FLOW BYPASSING
GAUGES CONTRIBUTED
TO THE IREGULARITIES.
THE NEEDLE VALVE WAS CLEANED AND THE BYPASS VALVE PRESSURE
TESTED. UPON RESART, SYSTEM RESPONSE WAS NORMAL
Case 2 Multiple Dry Gas Seal Failures
Background
> Natural Gas Pipeline Compressor
> New Bundle installed with new DGS
Compressor ran ~ 1500 hours
Other Issues caused train to come down Driver Auxiliary
System Failure. No work done to Compressor.
On Start-up, High Primary Vent Leakage at Shutdown Level
on both ends of compressor
Replace both DGS
> Run ~ 450 and again experience High Primary Vent
Leakage Shutdown
Both DGS replaced
> Run ~ 450 hours and experience third High Primary Vent
Leakage Shutdown
Both DGS replace, High Leakage alarm on opposite end of
Compressor from last failure.
Case 2 - Define
Three failure events that occur at 1500, 450 and 450
approximate hours intervals
First Failure
> Inspection of DGS shows possible explosive
decompression of primary O-ring in DGS on both ends
> Barrier Seal shows damage to carbon rings and face
springs
> Oil contamination from Driver Auxiliary failure
Aux oil pump continued to run while compressor
shut down over night, no barrier seal nitrogen
buffer present.
Case 2 Define Cont
Second Failure
> Discharge End DGS O-ring
shows signs of heavy
nibbling and a white
substance in the compressor
ports, some small amount of
residual oil in ports found
> Barrier seal shows signs of
damage to carbon ring and
face springs
Third Failure
> Suction End DGS O-ring
shows signs of heavy
nibbling, a white substance
and small amount of oil in the
compressor ports is found.
> Barrier seal shows signs of
damage to carbon ring and
face springs.
Primary Seal O-Ring and Tungsten Face As
removed from Compressor
Secondary Seal As removed from Compressor
Case 2- Define Cont
Use of Time Line assists in
sorting out the facts.
Are the failures same root
cause or are multiple root
causes occurring?
Use of Fishbone Diagram
help Identify possible root
causes Input from all Stake
Holders
Case 2 Measure & Analyze
Case 2 - Analyze
Case 2 Field Inspections
The Timeline, Fishbone Diagram and FMEA Tools helped to identify
several possible root causes
Planning of Shutdown to Inspect the machine for these root causes could
then be accomplished Down time, equipment and personnel
Do not stop at first smoking gun this method prepared all involved to
avoid jumping to a conclusion at the first source of problem found keep
digging to rule out possible root causes or identify multiple root causes.
Case 3 - Summary
Multiple Root Causes were Found
> First DGS failure (1500 hours)
Slight nibbling and heavy cracking
of the primary O-Ring
DGS Shim Ring run out axial
oscillation of seal
Explosive Decompression blow
down rate several times faster
then recommended
Oil Flooding of Seals caused
initial shut down fault of GT
auxiliaries symptom
> Second DGS failure (~450 hours)
Heavy nibbling of DGS primary ORing
Sever misalignment of
compressor to GT
Root cause of misalignment
not discovered
Check of alignment records
and piping strain made
Case 2 Summary Cont.
> Third DGS failure (~450 run hours)
Heavy nibbling of DGS primary O-Ring
Sever misalignment of compressor to GT
Root cause of misalignment not
discovered
Check of alignment records and
piping strain made
> Other Factors found and addressed
Barrier Seal Failures buffer gas supply
regulation
Oil getting to secondary vent port but not
found at primary seal face
Not a root cause but would lead to
failure
Detergent cleaning of DGS
White residue found in DGS port cavities
Testing determined that detergent used
to wash the seals was reacting with
Methanol Alcohol that was also used to
wash the seals
Not found on seal faces, could lead
to a failure practice stopped
Case 2 Summary (Improve)
Seals have now logged ~2500 run hours with out
signs of problem per End User
> Changed Shim Rings
> Changed blow down orifice size
> Aligned machinery monitored alignment
Changes through first 500 hours of operation
> Changed buffer supply regulation system
> Stopped use of Detergent cleaning of DGS
