HYDROGEN COMPRESSOR SEAL CASE STUDY UTILIZING HALO (NON-CONTACTING, COMPLIANT) INTER-STAGE, IMPELLER EYE, BUFFER AND FAIL- SAFE SEALS

HYDROGEN COMPRESSOR SEAL CASE STUDY UTILIZING HALO (NON-CONTACTING, COMPLIANT) INTER-STAGE, IMPELLER EYE, BUFFER AND FAIL- SAFE SEALS

Author Information John Justak, President/CEO of ATGI Over 28 years of high speed turbomachinery design experience (1M+rpm) Cryogenic and high temperature applications 7 patents granted related to seals & bearings Several more applied for

The Problem An end user of a hydrogen compressor in an oil refinery application has identified existing labyrinth seals as a source of compressor inefficiency and a potential cause of rotordynamic instability. The objective of this effort was to investigate the applicability of a compliant/non-contacting seal developed for aerospace applications in the 5-stage hydrogen centrifugal compressor. This effort identified the potential seal performance of the compliant seals in the hydrogen compressor application 4 primary locations (shaft seal, impeller eye seal, thrust balance seal and barrier/buffer gas seal).

Existing seal turbomachinery seal designs Labyrinth Sealsare the most commonly used form of turbomachinery gas path sealing. They are a fixed clearance and subject to contact and wear. Carbon ring sealscan provide reduced leakage however, they are prone to damage and require extensive close tolerancing and care during install which leads to increased cost. 4 Brush Seals- Are compliant seals that can provide reduced leakage over laby seals. However, they are prone to wear, flutter and are high cost.

Existing seals in case study H2 Photograph of existing hydrogen compressor labyrinth seals which utilized rotating teeth and a Fluorosint land. -The Labyrinth seals are installed with a 0.018 radial clearance. compressor

Labyrinth seal inherent issues Split case design requires tedious assembly procedure to install the rotor hardware and align with stator seals During compressor start-up, rotor-dynamic vibes facilitate large shaft excursions Large shaft excursions cause rotor/seal contact and premature wear causing larger than desired radial gaps Labyrinth seals are known for larger than desired cross coupled stiffness The cross coupled stiffness causes rotor vibes and premature wear/maintenance of the hardware

Desired Seal A seal that will allow ease of hardware build-up a seal that can function well with an offset rotor A seal that is not impacted by rotor excursions during compressor start-up A seal that eliminates or reduces cross coupled stiffness A seal that has a reduced running clearance

The Solution A non-contacting, compliant, dynamic seal. The seal is installed with a large radial clearance that reduces in clearance under low pressure (less than 5psid). The dynamic seal then self-adjusts to maintain a close clearance under all operating conditions, including dynamic rotor offsets. The seal is pressure balanced. In addition to low leakage the seal also provides rotor-dynamic damping.

Compliant, dynamic seal characteristics Installed with a clearance Reduce/eliminate crosscoupled stiffness from seals Temperature -252 C to 650 C+ Pressure Tested to 660psid Designed To 450 bar (6600psi) Speed +365 m/s (1200 fps) Operating clearance: 0.0005 to 0.030

Theory of Seal Operation (force balanced) Forces: 1. Spring force between cavity and carrier 2. Static pressure over seal shoe (Cavity) 3. Dynamic pressure between shoe and rotor Axial pressure drop results in high velocity fluid under shoe High velocity gas results in a pressure drop The pressure drop under the shoe creates a force balance toward the rotor As the seal/shoe surface approaches the rotor, the gas becomes choked Choked gas in the nozzle results in pressure rise/force balance away from the rotor Designed for fixed gap after minimal dp (~5 psi) is achieved across seal. Basic Design Inputs: -Leakage across mission profile -Relative motion between rotor and static structure -Local P,T conditions 10

H2 Compressor Interstage seal Front Plate 2 nd seal plate holder 2nd seal (3 x 0.01 ) Seal Seal running parameters Stage1 Stage2 Stage3 Stage4 Fluid gas Hydroge n mixture Hydroge n mixture Hydroge n mixture Hydroge n mixture Max Shaft rotation [RPM] 8650 8650 8650 8650 Inlet seal Pressure [psi] 130 141 153 165 Outlet Pressure (Reference) [psi] 119.6 130 141 153 Pressure ratio 1.0869 1.0846 1.0851 1.0784 Delta P [psi] 10.4 11 12 12 Fluid Temperature In [F] 115 129 143 157 Fluid Temperature Out [F] 100 115 129 143 Back plate - All 4 interstage seals are identical - Welded Assembly

Interstage Shoe Geometry 0.005 Split line cut 0.025 stops 0.02 stops 0.012 Cut Swirl brakes

Interstage Stress at Max Displacement (0.020 )

Results Inter-stage CFD (predicted and test) CFD prediction of Mass flow versus radial gap. Test results for dynamic seal installed with a 0.02 radial gap (flow v/s delta P Seal effective clearance v/s pressure drop

Impeller Seal conditions Seal running parameters Stage1 Stage2 Stage3 Stage4 Stage5 Fluid gas Hydrogen mixture Hydrogen mixture Hydrogen mixture Hydrogen mixture Hydrogen mixture Max Shaft rotation [RPM] 8650 8650 8650 8650 8650 Inlet seal Pressure [psi] 130 141 153 165 179 Outlet Pressure (Reference) [psi] 119.6 130 141 153 165 Pressure ratio 1.0869 1.0846 1.0851 1.0784 1.0848 Delta P [psi] 10.4 11 12 12 14 Fluid Temperature In [F] 115 129 143 157 171 Fluid Temperature Out [F] 100 115 129 143 157 Identical geometries Shaft Radial Tolerance [in] +/- 0.0005 Rotor Radial Vibration (in the center ) [in] +/- 0.003 Rotor Radial centrifugal growth (Max) + 0.0105 (Min) [in] + 0.0095 Max. Bearing Radial clearance [in] +/- 0.00275 Seal installed clearance [in] + 0.02

Impeller Stage 1 Geometry Front Plate Back Plate 2 nd seal plate holder 2 nd seal Anti rotation pin 2 nd seal Sea l Bolted assembly

2ndary seal split line geometry Impeller STAGE 1 Anti-rotation pins 2 nd seal (3 x 0.01 ) Flush line

Impeller Stage1 Shoe geometry 0.025 stops 0.02 radial stops (seal rest on back plate when shaft sits on seal) 0.02 stops 0.012 Cut Swirl brake

Shaft weight on seal analysis The compliant seal was designed to support the weight of the rotor during Compressor build-up 625 lb on longest tooth on seal Front plate Back plate

Results CFD Impeller Stage 1 3500 Impeller Seal - Gap vs Leakage 3000 Stage 1 -Gas Mixture- delta P=10.4 psi- Ref P=119.6 psi Mass flow [scfm] 2500 2000 1500 1000 Stage 1 -air- delta P=10.4 psi- Ref P=119.6 psi 500 0-3.12E-17 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02 Seal clearance [in]

Seal Installation

Compressor Start-up During Start-up the compliant seals are retracted (0.02 gap) Seals remain non-contacting As pressure is produced the seals close down to their running clearance Compressor end user stated this is the quietest machine at the refinery, there are no vibes

Compressor performance improvement Compressor has been running 24/7 since September 2012 Compliant seals installed with 0.02 radial gap are running at 0.006 radial gap. Hydrogen compressor efficiency was improved Compliant seals reduced compressor assembly time Compressor remains vibe free In addition to the operational seals a fail-safe seal was installed along with the low leakage (0.0015 radial gap) compliant buffer seal in the buffer seal cartridge. The f-s seal clamps down on the rotor if a large pressure drop occurs. This reduces or eliminates the escape of H2 gas.

Lessons Learned 4 inter-stage, 5 impeller eye, and buffer gas seals were installed in the hydrogen compressor. The impeller and inter-stage seals were manufactured for the split case design. The large installation clearance (0.02 radial gap) facilitated a smooth and less time consuming installation when compared to existing labyrinth seals. During compressor start up and run in, the compressor did not experience the typical vibrations on the way up to full speed. At full speed the compressor was more efficient than with the labyrinth seals. The refinery end user stated that he has never had a compressor run more quietly and vibration free. At this point in time, the compressor has been operating since September 2012, 24 hours a day, 7 days a week without incident and has maintained as built performance. This is the first installation of a split case non-contacting, dynamic, force balanced seal of this design.