Oil-Free Bearings and Seals for Centrifugal Hydrogen ...

Tribology Online, 8, 1 (2013) 44-63. ISSN 1881-2198

DOI 10.2474/trol.8.44

Copyright © 2013 Japanese Society of Tribologists 44

Article

Oil-Free Bearings and Seals for Centrifugal Hydrogen Compressor

Hooshang Heshmat, Andrew Hunsberger, Zhaohui Ren, Said Jahanmir* and James F. Walton, II

Mohawk Innovative Technology, Inc. Albany, NY 12205, USA

*Corresponding author: [email protected]

( Manuscript received 30 April 2012; accepted 19 September 2012; published 31 January 2013 ) ( Presented at Symposium S8: Hydrogen Tribology for Future Energy in the International Tribology Conference Hiroshima 2011 )

Deployment of a safe, efficient hydrogen production and delivery infrastructure on a scale that can compete economically with current fuels is needed in order to realize the hydrogen economy. While hydrogen compression technology is crucial to pipeline delivery, positive displacement compressors are costly, have poor reliability and use oil, which contaminates the hydrogen. A completely oil-free, high-speed, efficient centrifugal compressor using 4th generation compliant foil bearings and seals has been designed for hydrogen pipeline delivery. Using 6-12 MW drives operating at speeds to 56,000 rpm, a modular, double entry compressor was configured to deliver 500,000 kg/day at pressures greater than 8 MPa. Each of the two or three multi-stage compressor frames operate above the bending critical speed of the rotating group since speeds are 5 to 7 times faster than conventional compressors. To assure a structurally and economically feasible design, the rotor of each compressor frame spins at the same speed with blade tip velocities near 600 m/s. An iterative aerodynamic/structural/rotordynamic design process was used, including both quasi-three dimensional inviscid internal flow and Computational Fluid Dynamic (CFD) analyses. The flow field was carefully analyzed for areas of excessive diffusion, sudden velocity gradients and flow separation. Excellent correlation between the preliminary design and CFD analysis was obtained. Structural and rotor-bearing system dynamic analyses were also completed to finalize the compressor system configuration. Finite element analysis of the compressor impeller was used to verify structural integrity and fatigue limits for selected materials. Rotor-bearing system analysis was used to define acceptable bearing locations and dynamic coefficients, system critical speeds and dynamic stability. Given the high speeds, supercritical operation, and required reliability, efficiency and freedom from contaminants, compliant foil gas bearings and seals were designed and evaluated. Since hydrogen will be used as the lubricant for the foil bearings, substantially lower power loss than oil lubricated bearings will be experienced and the auxiliary supply or scavenge system is eliminated. Keywords: bearings, foil bearings, seals, foil seals, hydrogen compressor, oil-free, bearing design

1. Introduction

In today’s low volume hydrogen economy, delivery infrastructure consists of high-pressure tube trailers and small pipeline infrastructure, extending several 100 miles from the point of production. The cost to deliver hydrogen by such means varies from $4 to $9 per gallon of gasoline equivalent (gge) [1]. To become a reality, safe and efficient hydrogen production and delivery infrastructure on a scale that can compete economically with current fuels is needed. The United States Department of Energy (DOE) has set a goal of $1/gge for the cost of hydrogen delivery by the year 2017 [2]. To achieve this goal, significant reduction in capital and operating costs of hydrogen delivery is needed.

The most cost effective method for large volume hydrogen distribution is with gaseous pipeline delivery, similar to today’s natural gas pipeline [2]. It is estimated that a single compressor station would require a capacity of up to 1,000,000 kg/day and discharge pressures greater than 8 MPa [2]. Today’s positive displacement (PD) hydrogen compression technology is very costly, and has poor reliability and durability, especially for components subjected to wear (e.g., valves, rider bands and piston rings). Due to the poor reliability of PD compressors, current hydrogen producers often install duplicate units in order to maintain on-line times of 98-99%. Such machine redundancy adds substantial capital costs. Therefore, reliable, efficient and low capital cost hydrogen compressor technology is needed. Given this need, centrifugal compressors, such as those used for


Japanese Society of Tribologists (http://www.tribology.jp/) Tribology Online, Vol. 8, No. 1 (2013) / 45

natural gas (NG) pipelines and other approaches, are being considered. However, present day NG compressors are limited in speed and would require many more stages in order to achieve the necessary discharge pressure with hydrogen. In order to achieve comparable system performance (i.e., pressure ratio, efficiency, flow, etc.) to an NG system, a hydrogen centrifugal compressor must operate much faster than anything presently available. In addition to their speed limitations, present day NG compressors are oil-lubricated, which would contaminate the hydrogen gas or require costly filtering. Finally, the potential for hydrogen embrittlement of materials used in present-day NG compressors makes these systems incompatible with hydrogen.

To overcome NG compressor limitations for hydrogen and meet the requirements for high operating speeds, contamination free process gas, hydrogen embrittlement of exposed componenets and cost constraints, new advancements in compression technology are needed. One approach, which addresses the above mentioned design constraints, is the use of compliant foil bearings. Foil bearings are capable of operating reliably at very high speeds, including above the bending critical speed of the rotating group, while using the hydrogen gas as its lubricant. Since foil bearings permit very high speed operation, the number and stages and overall diameter can be reduced, resulting in overall lower pumping station capital costs. Furthermore, by using hydrogen to lubricate the bearings, oil contamination is eliminated and maintenance actions reduced.

Even though foil bearings have been around since the early 1950s, as is evident from the work of Prof. Blok, et al. [3,4], the working mechanism was first described in the late 1940s based on observations made during the development of high-speed tape recording devices for computers [4]. These observations resulted in the tape-like or tension dominated gas foil bearings in which the hydrodynamic action between the moving tape and stationary head were simply reversed. The earliest leaf type foil bearing patent was issued to Silver et al. [5]. Initial developments with these early patented foil bearings were primarily applied to high performance air cycle machines for aircraft cabin cooling and pressurization. At about the same time as the initial patent was issued, a preliminary model for foil bearings was published by Eshel and Elrod [6]. The model used a simplified approach and assumed the bearing to be infinitely long with the governing fluid film and elasticity equations being uniquely solved. This early theoretical work was followed by the experimental work of Lichet [7]. It was the analytical and experimental work by the academicians during the 1960s and 1970s that paved the way for development of various, novel foil bearings. Inventions, such as bump type or bending dominated foil bearings by Cherubim [8] and his contemporaries, attest to that effect.

The energy crisis of the early 1970s’ was the

impetuous behind new and expanded interest in the advancements of foil bearing technology for high speed rotating machinery systems as well as a change in the locale for technology development. Government and industry began to lead the innovation and advancements in the development of foil bearings. The work of Walowit and his team, [9,10] laid a solid foundation for further theoretical investigation of the bending dominated foil bearings using nested and coupled governing equations of elasticity and hydrodynamics. Leaf type foil bearings were then investigated by two groups using two different numerical techniques, namely Finite Element Analysis (FEA) and Finite Difference Analysis (FDA). These newer analytical models began to more realistically represent foil bearings [11,12].

Over the years major advances have lead to improvements in load carrying capacity and damping characteristics of the foil bearings. While the early foil bearings with a uniform corrugated bump foil (1st generation) had limited load capacity, the newer foil bearings use a more intricate bump foil design with multiple slits in the foil and varied spacing of bumps across the foil (3rd generation), resulting in a much higher load capacity [13].

Integral to the continued development of foil bearings was the need to understand the surface morphology and tribological coatings associated with foil bearings. While efforts of the 1970’s did establish some unique tribo-material coating systems, the developed and used coatings were limited to lower temperatures or used carcinogenic materials that made them unusable for all practical purposes. However, substantial contributions to the field of tribology and foil bearings, including bearing surface treatments, came to fruition beginning in the early 1990s and have continued to the present. Development of extremely flexible coatings [14-16] has paved the way for advancements in foil bearings to the point that they can sustain the extreme environment of advanced gas turbine engines, namely, the foil bearing operating temperatures surpassing 870°C are possible.

The work of the lead author and his research team is also recognized for its simultaneous developments in the theoretical realm, experimental methods and results, and in the methods needed to successfully apply foil bearings and seals in advanced oil-free rotating machinery. These works have resulted in a better understanding of the performance, design and application methodology of compliant foil bearings/seals, as well as the development of better analytical tools to integrate advanced foil bearings/seals into high performance propulsion and power generation systems [17-27]. These advanced foil bearings with improved load and temperature performance having variable compliant structures in all three dimensions and time-dependant damping and stiffness are designated as 4th generation foil bearings. These advanced foil bearings have enabled tribological components to operate supersonically and have opened the window of opportunity for the development of most

Hooshang Heshmat, Andrew Hunsberger, Zhaohui Ren, Said Jahanmir and James F. Walton, II


efficient modern turbocompressor systems with smaller footprints.

To enhance the economic viability of hydrogen as an energy carrier and to substantially improve the efficiency of centrifugal compressors so that they become feasible for use in transportation and delivery of hydrogen, a close clearance, non-contacting, dynamic compliant foil seal [28-35] has been suitably sized for hydrogen and/or natural gas pipeline compressors. A gas foil seal has a compliant smooth top foil as the seal surface supported by corrugated strip foils. The corrugated strip foils act as compliant backing springs and the top foil, when loaded, deflects, conforming to the pressure profile. Conceptually, axial foil seals comprise foil thrust bearings with extremely small land over diameter ratios, and seal plates with edges extended to form skirt like secondary static seals sandwiched in-between seal cartridges. Radial foil seals are comprised of foil journal bearings and smooth top foils with flange like fingers extending radially outward.

An analytical model is used to design high-efficiency foil seals and bearings. This model is readily and efficiently modified to include various hydrodynamic compliant surfaces under hydrostatic loading (e.g., differential pressure across the seal) plus hydrodynamic pressure (due to runner velocity and eccentricities). A provision for the reaction of the top elastic and smooth foil to the applied pressure and the deformation of the underneath compliant elements is included in the analysis. By including smooth top foils and possibly surface coatings, we have a better understanding of fluid film thickness variations between the seal/bearing interfaces.A numerical technique has been developed to merge finite element analysis (FEA) with finite difference analysis (FDA) and proven to provide

successful results [36-40]. This technique combines governing equations of elasticity with hydrodynamics and solves multi-level, non-linear equations using the above mentioned numerical methods.

The computational technique was used for the design of foil bearings and foil seals using variable boundary conditions and large differential sealing pressures. Elastohydrodynamic pressure profiles were iteratively computed using a finite difference method and input to a finite element model of multilayer compliant foil seal. This technique provides an efficient and robust method to analyze and evaluate high-speed seal/bearing designs so that the best design can be selected and fabricated for testing. The performance of the designed foil bearings and seals was then evaluated under various testing conditions. Both static and dynamic testing were performed on the seals using air and helium at operating speeds up to 60 krpm. The results of these tests, in which various seal parameters were varied, indicated that very low values could be obtained for the flow factors used to characterize seal leakage [41]. It was then demonstrated that performance capabilities of the designed centrifugal hydrogen compressor met the needs for the pipeline delivery of pure hydrogen [42].

2. Experimental and Analytical Methods

The feasibility of a multi-stage centrifugal compressor for hydrogen pipeline delivery was investigated using an iterative design process, as shown in Fig. 1. Design targets were selected based on a review of DOE established targets and recent literature to identify hydrogen demands and distribution in the future hydrogen economy (See Table 1). The most critical design targets identified for a centrifugal compressor for this application are capacity, pressure ratio, contamination, leakage, foot print, and system cost. In order to design a centrifugal compressor to satisfy these design targets, the following methods were applied.

2.1. Preliminary Design A preliminary design study was performed to define

the overall system envelope. From there, more detailed design of rotordynamics, bearing and seal sizing, detailed

Design Targets

Final Design

Preliminary Design

AerodynamicsRotor/System Dynamics

Structural Analysis

Thermal Management

Fig. 1 Iterative design approach used in the development of a multi-stage hydrogen compressor

Table 1 Design Targets for hydrogen pipeline compressor [2]

Category Design Targets

Capacity (kg/day) 100,000 – 1,000,000

Inlet Pressure (MPa) 2.0 - 4.8

Discharge Pressure (MPa) 8.3

Contamination None

Leakage (%) <0.5

Footprint (m2) 28-33



aerodynamics and structural analyses were performed. A preliminary design was conducted with a commercially available compressor design software package (CompAero, RH Aungier). The program used non-dimensional analysis to make preliminary performance predictions based on a library of empirical data. This approach gives the designer a reasonable first approximation of impeller diameter, rotational speed, efficiency and stage pressure ratio. The results from the preliminary design provided an overall system layout, including approximate geometry, number of stages, rotor length and maximum tip velocities. These results provided input to the more detailed design analysis.

2.2. Aerodynamic Design The role of the aerodynamic design was to determine

if performance predictions made in the preliminary design were reasonable, to refine the single stage design and to investigate the performance of the multi-stage system. While the preliminary design provides approximate stage geometries, the details of the shroud, hub and blade profiles require significant refinement. Aerodynamic design was performed using mean-line analysis, quasi three-dimensional flow analysis and computational fluid dynamics (CFD). Mean-line analysis and quasi three-dimensional flow analysis were performed with CompAero software. The results were used to develop detailed geometry of the impeller, diffuser and return channel. Aerodynamic performance was evaluated for a range of inlet pressure and flow conditions to validate stable system performance. The results were used to update rotordynamic models, and structural analysis. The final design was also evaluated using CFD software (Ansys Inc., Canonsburg, PA) at a single design point. The results of the CFD analysis were used to validate the predictions of mean-line and quasi three-dimensional analyses.

2.3. Rotor/System Dynamics The rotor/system dynamic analysis included design

and evaluation of the rotor, bearings and seals of the compressor system. A series of rotordynamic studies was performed using finite element analysis (FEA) with DyRoBes software (Eigen Technology, Inc, Davidson, NC). Based on the rotor geometry, mass and operating speeds determined from aerodynamic analysis, key design parameters for the foil bearings were identified. Rotordynamic analysis was then performed to assess critical speeds, stability and response as a function of the bearing speed dependent stiffness and damping properties. Various materials were investigated to evaluate their influence on rotordynamic behavior. More detailed material selection was conducted as part of the stress and structural analysis.

2.4. Structural Analysis Structural analysis of the proposed compressor

system was conducted using linear elastic Finite Element Method (Nastran solver, MSC Corp., Santa Ana, CA) to

determine centrifugal stresses and dimensional growth. A review of several candidate materials was performed. Material selection is particularly critical in this application due to potential for hydrogen embrittlement. A thorough material review was conducted to select relevant materials for evaluation.

2.5. Foil Bearing/Seal Design The fundamental steps involved in the design of foil

bearings are outlined in a recent publication [43]. The flow diagram in Fig. 2 presents the program logic flow chart for the analysis of a compliant surface foil seal and/or bearing. The static force-displacement relationship plays an important role in analyzing compliant foil seals/bearings, as the compliance of the surface governs the operational characteristics. The solution of the governing hydrodynamic equations must deal with a compressible fluid that is coupled with a

Fig. 2 Calculation procedure for foil bearing/seal

analysis



compliant seal/bearing surface. A solution to the structural compliance is provided in two stages. In formulating the first level elasticity solution of the compliant element stiffness matrix, it is assumed that the smooth top foil follows the global deflection of the backing springs and will not follow the indentation between bumps of the corrugated backing springs. It is also assumed that deflection of the top foil and backing springs in responding to the hydrodynamic pressure is dependent on local effects only. The film thickness variation is due both to initial geometry (eccentricity and initial seal surface shape) and the deflection of the compliant surface under imposed hydrodynamic pressure. Two-dimensional compressible Reynolds equation is written in finite difference form with the dependent variables represented by a finite number of points located at intersections of a grid mesh. The flow is assumed to be compressible and isothermal. The gas is assumed to be ideal; and steady state flow conditions are assumed. The resultant form is linearized by the Newton-Raphson method, represented in matrix notation and solved by the column method to satisfy boundary conditions. The resultant pressure field is then taken to the next level via finite element elasticity analysis to include the effects of the top foil in the compliant foil seal/bearing analysis. In this higher level elasticity analysis, the mathematical model of the foil seal/bearing consists of a staggered, thin top foil supported via springs subjected to the pressure field. The combined top foil and backing spring deflection is computed for the given pressure, resulting in a deflection matrix and overall structural stiffness. The bearing stiffness coefficients are computed with the use of small perturbation on a primary steady-state journal rotating frequency and position. FEA and CFD computer codes are used as external codes to predict bearing overall deflection under hydrodynamic operation. The latter is primarily used to enhance the entrance fluid low and top smooth foil deflections.

Since the foil bearing system relies on the process gas for lubrication and internal cooling of seals, foil bearings and any other rotating component that are difficult to access from the outside of the system housing, thermal management must be utilized to prevent thermal runaway within the rotating components/stators. Secondary flow paths are designated to remove excess heat and also stabilize the temperatures. The bearings, seals and other rotating components, due to asymmetric loading, have tendency to heat up non-uniformly creating asymmetric temp distribution, which results in a non-uniform thermal expansion and thermal stress concentrations. Therefore, thorough thermal analysis needs to be conducted iteratively in conjunction with the other design steps to assure and sustain thermal stability of the systems, Fig. 1.

2.6. Foil Bearing and Seal Testing Following the design of the foil bearings and seals for

the hydrogen compressor, the required full-scale components were fabricated and tested at full speed.

Special bearing and seal tribometers were designed and assembled for these tests. The tribometers and testing protocols are described under the results section.

3. Results

A multi-stage centrifugal compressor was designed for a large-scale hydrogen pipeline using the iterative design approach, described in the previous section. Because the design target estimates in Table 1 vary significantly, an average capacity and inlet pressure were chosen for this investigation. The design targets chosen are 500,000 kg/day of hydrogen with an inlet pressure of 3.4 MPa and discharge of 8.3 MPa.

Given the required high flow rates, both single and double entry compressor systems were considered. While both design approaches could achieve the desired flow and pressure ratio, the single entry system would have higher thrust loads, which present a challenge for the thrust bearings. Consequently, the double entry design with inherently lower net thrust loads was considered for this study. Each frame consisted of three compression stages with six impellers (165 mm diameter) on a single rotor (Fig. 3). Inlet flow was divided in half and each compressor wheel designed to handle 250,000 kg/day. Two frames were required to achieve the target discharge pressure with an operating speed of 67 krpm (580 m/s). In order to achieve a high performance in this design approach, the margin of safety with respect to material stresses was estimated between 1-2 times the ultimate yield strength. This was, therefore, considered a high-performance but high-risk design.

An alternative design approach was also considered in which greater margin was provided with respect to discharge pressure, rotordynamics and material stresses. This high margin design operates at a speed of 56 krpm (490 m/s) and provides a greater safety margin with respect to rotational stress of the rotor but will require three compressor frames. In either design approach, the rotor operates above the first bending critical speed. In order to accommodate the large rotor excursions of super-critical rotor operating speeds, compliant foil bearings with high damping and load capacity are

Fig. 3 Three-quarter cut-away view of a single frame

of the hydrogen compressor concept



required. The following results describe the detailed analysis, which was performed in order to arrive at these design choices.

3.1. Preliminary Design The goal of the preliminary design was to determine a

reasonable estimation of single centrifugal stage geometry and performance. Results were obtained using the non-dimensional analysis techniques with the aid of the empirical data library in CompAero. An unshrouded or open impeller was assumed because of the very high operating speeds anticipated. Vaned diffusers were employed in the preliminary design. Based on the design points specified above, the influence of operating speed and impeller diameter on stage efficiency was determined and results are shown in Fig. 4a. The results show very little influence of operating speed for the range evaluated; however, a clear maximum efficiency is achieved with respect to impeller diameter. Peak efficiency occurs at an impeller diameter between 160-170 mm. The influence of operating speed and impeller diameter on stage pressure ratio was also determined (Fig. 4b). For a given operating speed, maximum pressure ratio was achieved at impeller diameters greater than 150 mm. Above these diameters, little additional pressure ratio was gained.

The results of the parametric study were used to determine the total number of stages required to achieve the design target of 8.3 MPa discharge pressure, assuming an inlet pressure of 3.4 MPa (Fig. 5). To account for pressure losses associated with piping, intercoolers and other unknowns in the final system, 5% percent pressure margin was added to the design discharge pressure. For impeller diameters between 160-170 mm, a total of 6-9 stages would be required depending on the selected operating speed and, hence, tip speed.

3.2. Aerodynamic Design Aerodynamic design was conducted to confirm the

performance predictions of the preliminary design and to refine the geometries of rotating components for detailed rotordynamic and stress analysis. Detailed blade design was conducted using a quasi-three dimensional inviscid internal flow analysis with CompAero design software. The quasi three-dimensional flow field analysis was a cost effective design tool to quickly iterate between aerodynamic performance and the detailed flow field behavior. The flow field was carefully analyzed for areas of excessive diffusion, sudden velocity gradients and flow separation. The design was optimized through minor changes in impeller blade geometry. Once a design was found to have satisfactory flow field characteristics, structural and rotordynamic analyses were performed. Four key design iterations were conducted between aerodynamics, structural and rotordynamics in order to arrive at the final compressor design.

A CFD analysis of a single compression stage was performed with the high margin design and was used to validate aerodynamic performance predictions made with mean-line analysis. The 3D solid model of the impeller design and the velocity streamline results from CFD analysis are shown in Fig. 6. The results of the CFD analysis showed overall acceptable flow field behavior with no significant turbulence or diffusion. The correlation of the mean-line analysis to CFD results was excellent, with 98.5% agreement with respect to stage

0.76

0.78

0.80

0.82

0.84

0.86

0.88

0 100 200 300

Adi

abat

ic E

ffic

ienc

y (%

)

Impeller Diameter (mm)

488 m/s

549 m/s

610 m/s

1.04

1.08

1.12

1.16

1.20

0 100 200 300

Stag

e Pr

essu

re R

atio


488 m/s

549 m/s

610 m/s

(a)

(b) Fig. 4 Preliminary design results for design point B

(500,000 kg/day; 4.9 MPa) for adiabatic efficiency as a function of impeller diameter in (a) and for pressure ratio as a function of impeller diameter in (b)

0

5

10

15

20

0 100 200 300

N0.

Sta

ges

Req

uire

d


488 m/s

549 m/s

610 m/s

Fig. 5 Total number of stages required to achieve the

design target of 8.3 MPa discharge pressure



pressure and efficiency. Following validation of the mean-line analysis

method, the multi-stage compressor performance was analyzed. The results of the individual stage discharge pressure and shaft power are shown for both the high margin design and the high performance design (Fig. 7). At the exit of frame 3 (i.e. stage 9), the high margin design discharge pressure was 9.9 MPa and required a shaft power of 11.4 MW. A 13% discharge pressure margin was achieved. For this analysis, intercoolers placed between frames were assumed to have a 70 Pa pressure drop. For the high performance design, a discharge pressure of 8.7 MPa was achieved and total shaft power requirement was 10.7 MW. For the high performance design, 5% pressure margin was achieved.

3.3. Rotor/System Dynamics Once the rotor system concept, including number and

sizing of impellers, was defined, a finite element model was developed for rotor/bearing system dynamic analysis. The estimated stiffness and damping properties of the designed foil bearings were used for rotordynamic analysis of the compressor rotor with DyRoBes FEA software (Eigen Technology, Inc). One foil journal bearing was located at each end of the shaft (Fig. 8a). The initial iteration from aerodynamic analysis suggested a shaft minimum diameter of 50 mm. The critical speed

analysis results indicated a first bending mode of 22 krpm and the second bending mode of 61 krpm. In this

(a)

(b) Fig. 6 3D impeller model in (a) and CFD analysis of

the flow within the impeller at the design point in (b)

0

4

8

12

16

20

0

1

2

3

4

5

6

7

8

9

10

Pow

er (M

egaW

att)

Dis

char

ge P

ress

ure

(MPa

)

Discharge Presure Goal

Discharge Presure

ShaftPower

0

4

8

12

16

20

0.00

1.00

2.00

3.00

4.00

5.00

6.00

7.00

8.00

9.00

10.00

Pow

er (M

egaW

att)

Dis

char

ge P

ress

ure

(MPa

)

Discharge Presure Goal

Discharge Presure

ShaftPower

(a)

(b) Fig. 7 Multi-stage compressor performance of the

high margin design operating at 490 m/s in (a) and of the high performance design operating at 580 m/s in (b)

(a)

(b) Fig. 8 Foil journal bearings (a) and foil thrust bearings

(b) selected for use in the hydrogen compressor



case, the high performance design would be required to operate above the second bend. In addition, the operating speed of the high margin design (56 krpm) would not provide sufficient margin from second bend.

In order to avoid operation near the second bend, the diameter of the shaft in between the compressor impeller was increased. The increase in shaft diameter resulted in an increase of the first and second mode. The increase in diameter provided a sufficient safety margin away from the second bending mode. Rotordynamic analysis was performed with three candidate materials that were identified during stress analysis. As seen in Table 2, the material type had little impact on the rotordynamic behavior.

Since operation above the first bending critical speed was expected, iterative design was used to achieve bearing dynamic coefficients that would insure stable rotor-bearing system operation passing through the first bend. Although the selected dual entry compressor design provides inherent balanced thrust loads, some transient thrust loads may be present during operation. To handle such transients, two foil thrust bearings were located at the center of the shaft, Fig. 8b.

3.4. Foil Bearing Design The range of analytical parameters and design

variables involved in the present method of solution is fairly wide. The overall program, therefore, involved a number of computer codes namely FEA and CFD and specialized structural elasticity analysis, which included non-linear solutions and effects of surface interactions with the interface layers, some of which had to be iterated in order to arrive at integrated results for the

bearing system [43]. Design of the journal bearings followed the combined

structural and hydrodynamic analysis method outlined in the Section 2.5. Note that foil journal bearings do not permit the magnitude of hydrodynamically generated pressure profile to dip below the boundary or bearing chamber pressures (i.e., the ambient pressure). Whenever the diverging portions of the film tend to produce subambient pressures in the fluid film, that is, on top of the foil, the prevailing ambient pressure underneath the foil will lift it up until the pressures on both sides of the foil are equalized. From continuity requirements, the film must fulfill both the zero pressure and zero pressure gradient boundary conditions. For a complete description of boundary conditions the readers are referred to Reference [43]. There are six geometric, structural and operational parameters relevant to a foil journal bearing. These are pad angular extent, structural compliancy length to diameter ratio, speed parameter, load angle, and number of pads. There are also the eccentricity ratio and the attitude angle, the latter tied to the load angle. In the analysis of foil journal bearing, the governing hydrodynamic equations dealing with compressible fluid is coupled with the structural resiliency of the bearing surfaces [43]. The solution to the structural resiliency is provided in two stages. The first level elasticity solutions deal with compliant elements (bump foil/backing spring). In this scheme, the bearing backing springs were modeled and a resiliency matrix was determined. The film thickness in line with the deflection of the bearing surface was substituted into the Reynolds equation and solved by finite difference method. The first level computation provided hydrodynamic pressures obtained

Table 2 Summary of predicted bending critical speeds for different materials

1st Bend Critical Speed (rpm) 2nd Bend Critical Speed (rpm) Aluminum Alloy 27,100 74,633 Titanium Alloy 26,092 72,044 High-Strength Steel 27,700 76,321

0.81

1.21.4

1.61.8

0

20

40

606000

7000

8000

9000

10000

Radius (inch)Angle (degree)

Stiff

ness

(psi

)

0.81

1.21.4

1.61.8

0

20

40

600

5

10

15

20

Radius (inch)Angle (degree)

Pres

sure

(psi

)

(a) (b) Fig. 9 Bump stiffness from finite difference program in (a) and pressure profile from hydrodynamic analysis in (b) (1

inch = 25.4 mm, 100 psi = 689 kPa)



from the solution of the modified Reynolds equation. The hydrodynamic pressures were then substituted into the second level elasticity equation to solve for the bearing surface deformation and film thickness using finite element method. Several iterations were followed to obtain convergence.

The compressor system was designed to operate at speeds above the first bending critical speed with rigid body frequencies. The dynamic and steady state bearing performance was analyzed at rotational speeds up to 60,000 rpm (+15% overspeed) at speed increments of 5,000 rpm. For the dynamic performance the computations considered the resonant frequencies for a given running speed.

In designing the thrust foil bearing, the combined structural and hydrodynamic analysis technique was applied to one pad of a candidate foil thrust bearing. Based on the initial geometry and boundary conditions, stiffness of the bump foils shown in Fig. 9 was calculated using a custom FD program. Applying a local pressure-deflection relationship, the pressure profile shown in Fig. 9b was predicted through hydrodynamic analysis. Linear elastic FEA was then conducted to determine the elastic deflection of the top and compliant bump foil elements. The smooth top foil pad was modeled as a 2D shell and the bump foils were modeled as grounded spring elements at nodes of the shell elements. The leading edge was fixed in all degrees of freedom, the pressure profile was applied at the nodes of the shell elements and the stiffness values accounting for the bump foils were converted to spring constants at the nodes of the shell elements. Figure 10 compares the un-deformed (white meshes) and elastically deformed (colored contours) smooth top foil pad, illustrating that the smooth top foil pad develops an elastic pocket, as expected. Using the pressure profile provided by hydrodynamic analysis and the deformation results computed by structural FEA, the stiffness values that include the effects of both the smooth top foil and bump foils were calculated. These new stiffness values were then used in the next iteration of FD hydrodynamic

analysis. This procedure was repeated until the pressure solution converged, which was typically three iterations or less.

The analytical technique was improved to address elasticity effects of combined smooth top foil and elastic foundation consisting of multilayer compliant stacks of bump foils and thin plates. Therefore, instead of simply modeling the structural layers underneath the smooth top foil as backing springs with initial compliance calculated from the closed-form elasticity solution, the stack was fully modeled with FEA. The pressure profile was still solved in the elastohydrodynamic (EH) analysis using the FD method and applied to the smooth top foil as the only input to the FE model.

Figure 11 shows the exploded view of a sample foil thrust bearing assembly. The FE model, shown in Fig. 12, includes the smooth top foil, top plate, bump, crown plate and spring. The adjacent layers are modeled as touching contact bodies that allow for relative motions with assigned friction coefficients, except at welding locations where they are fixed. The stack sits on a rigid body fixed in all degrees-of-freedom. In addition, the top plate and crown plate are tied together at the holes to simulate the pin connection.

The same initial pressure from EH analysis was applied as load to each of the smooth top foil pads in the FE model. The elastically deformed shapes of the smooth top foil and top plate under this pressure are shown in Fig. 12 c and d, where colors depict magnitude of the displacement. The stiffness matrix was then calculated from the pressure applied to the smooth top foil and its corresponding deflection. Using this stiffness, the new pressure profile was solved for Iteration 2 from the EH analysis. Applying the new pressure to the smooth top foil in the FE model, the smooth top foil deformation resulting from Iteration 2 was solved. The calculated stiffness using EH pressure and FE displacement from Iteration 2 are shown in Fig. 13a. The EH pressure for Iteration 3 using the stiffness resulting from Iteration 2 is shown in Fig. 13b.

Fig. 10 Un-deformed (white mesh) and deformed

(colored contours, maximum deflection of 52 µm) of smooth top foil

1. Smooth Top Foil & Stiffener

3. UpperBump

4. Lower Bump

5. Crown Stiffener

6. CrownPlate

7. Crown Spring& Block

2. Top Plate

Fig. 11 Exploded view of foil thrust bearing assembly



3.5. Foil Seal Design The iterative foil bearing design technique was also

used for the design of both face foil seal and radial foil seal. As shown in Fig. 14, the seal plate of the face seal replaced the top plate of the foil thrust bearing in the FE model. The result from Iteration 1 showed that the smooth top foil of the face seal deformed slightly less than the foil thrust bearing under the same initial pressure, indicating a slightly stiffer supporting stack. Therefore, without going through Iteration 2, the pressure for

Iteration 3 shown in Fig. 13b was directly applied to the smooth top foil. Figure 14c and d shows the corresponding deformed shapes of the smooth top foils and seal plate from Iteration 3.

The basic design of the radial foil seal is shown in Fig. 15. It consists of smooth top foils made of high strength and high stiffness material supported by a series of compliant/spring like corrugated bumps. The smooth top foil forms the sealing surface and is located closest to the shaft. The compliant foil seal can move with the shaft and still maintain an extremely small gap, thereby

0.5

1

1.5

2

00.5

11.50

2

4

x 104

x (inch)

y (inch)

Cal

cula

ted

Stiff

ness

from

Iter

atio

n N

o. 2

(psi

/in)

0.5

1

1.5

2

00.5

11.50

20

40

x (inch)

y (inch)Pres

sure

for I

tera

tion

No.

3 (p

si)

(a) (b) Fig. 13 Calculated stiffness using EH Pressure and FE displacement in from Iteration 2 in (a) and computed EH

pressure for Iteration 3 using stiffness from Iteration 2 in (b) (1 inch = 25.4 mm, 100 psi = 689 kPa)

(a) (b)

(c) (d) Fig. 12 FEA modeling of foil thrust bearing (a) top view, (b) bottom view, (c) elastically deformed shape of the

smooth top foil (maximum deflection of 126 µm) and (d) elastically deformed shape of the top plate (maximum deflection of 148 µm) from Iteration 1. Displacement magnitudes are shown in color



(a) (b)

(c) (d) Fig. 14 FEA modeling of face seal: (a) top view, (b) bottom view, (c) elastically deformed shape of smooth top foil

(maximum deflection of 56 µm) and (d) elastically deformed seal plate (maximum deflection of 59 µm) from Iteration 3. Displacement magnitudes are shown in color

Shaft

FlangeEnd

FreeEndFreeEnd

Seal

Primary Bumps

hFace Bumps

CR

Top Foils

Shaft

FlangeEnd

FreeEndFreeEndFreeEndFreeEnd

SealSeal

Primary Bumps

hhFace Bumps

CR

Top Foils

Fig. 15 Schematic diagram of the MiTi® radial foil

seal

(a)

(b) Fig. 16 FEA modeling of radial foil seal



making it possible to substantially reduce the leakage rates compared to state-of-the-art seal technology. Because the compliant foil seal is supported on compliant bump springs, the whole seal assembly can move with the shaft. In essence, the multi-layer smooth foils of the sealing surface act like a foil bearing where increasing hydrodynamic pressures are generated with increasing speed, such that the hydrodynamic stiffness

exceeds the bump spring stiffness. Besides accommodating shaft dynamic motions, the compliant foil seal design also permits shaft thermal and centrifugal growth since only one end of the smooth top foil is fixed to the static housing. By fixing only one end of the smooth top foils, system extensibility is possible to accommodate shaft growth with speed, something that is especially important for the high-speed hydrogen compressor design under consideration.

The computation technique was applied to model and analyze a simplified radial foil seal consisting of two overlapped smooth foils with flange fingers and two bump foils with a septum in between. As shown in Fig. 16, each layer was meshed with 2-D shell elements with appropriate thicknesses. The inner smooth foil was meshed in such a way that the pressure profile from FDA could be interpolated and applied accordingly.

The hydrodynamic pressure was calculated from elastohydrodynamic analysis using the FD method, resulting in an initial pressure profile, seen in Fig. 17a. Next, hydrostatic loading due to differential pressure across the seal was added to the calculated hydrodynamic pressure due to runner velocity and eccentricities. To provide better accuracy for the reaction of the top elastic and smooth foil, a finer mesh was used for the inner smooth foil than the finite difference grids. Therefore, the pressure was interpolated first and then applied to the inner smooth foil in the finite element model. The interpolated pressure taking both hydrostatic and hydrodynamic pressures into consideration is shown in Fig. 17b. The static structural analysis was conducted using implicit nonlinear solution. Figure 18 compares the un-deformed and elastically deformed inner smooth foil under the combined hydrostatic and hydrodynamic pressures. It is clear that the smooth top foil is fairly compliant under load due to the compliant elements underneath.

3.6. Seal Performance Evaluation Preliminary tests were first conducted to verify the

structural stiffness of the foil seal. These tests were performed with the seal located in place but without rotation of the shaft. Using the clearance values

00.2

0.40.6

0.8

0100

200300

4000

100

200

300

Length (inch)Angle (degree)

Pres

sure

(psi

)

(a)

(b) Fig. 17 Calculated EHD pressure profile in (a) and

interpolated hydrodynamic and hydrostatic pressure in (b) for a radial foil seal (1 inch = 25.4 mm, 100 psi = 689 kPa)

(a) (b) Fig. 18 Un-deformed (a) and elastically deformed (b) inner smooth top foil under combined hydrodynamic and hydrostatic

pressures (maximum defelction of 1.15 mm)



estimated from the load deflection curves of the seal assembly measured prior to testing as inputs to an in-house Compliant Foil Seal (MiTi-CFS) program, the mass flow leakage was predicted for both air and helium under varying differential pressures. The corresponding flow factor was plotted as a function of the differential pressure and compared with the experimental results. As shown in Fig. 19, the predicted flow factors match the measured values for both air and helium. Using the same clearance and stiffness values used for helium and replacing with the viscosity of hydrogen, the seal leakage was then predicted for hydrogen. Helium was used as a simulant gas for hydrogen due to its physical properties as compared with hydrogen.

Following static testing, a high-speed seal tribometer fully operating on foil bearings was designed and manufactured to evaluate seal performance under various conditions, Fig. 20. The custom high-speed motor with integrated turbine could drive the spindle to 60 krpm through a coupling. A double or back-to-back seal configuration with high pressure gas was adopted to balance the thrust load such that two separate seals could be tested simultaneously. The two seal assemblies and test journal are also shown in Fig. 20. In order to characterize the seal leakage as a function of operating speed and inlet pressure, the tribometer was fully instrumented to measure the inlet mass flow and pressure, seal foil temperatures and rotor orbits, among others.

The initial dynamic testing was carried out with four seal configurations in progressive settings in order to safely run to full operating speed and high inlet pressure. Figure 21a compares the flow factor of the four configurations as a function of speed at a fixed inlet pressure of 100 psi. The seal leakage reduced as speed increased due to centrifugal growth of the rotor and the developed hydrodynamic pressure. The leakage was reduced when face bumps were installed, while preloading the face bumps did not improve the seal performance. The most significant contribution to seal performance was from the radial clearance between the

seal surface and test journal; the smaller the clearance the less leakage. The configuration with loose radial clearance and loose face bump was selected for further tests.

To characterize seal performance at elevated temperatures, four heat torches were installed at the gas inlet. The gas passed through the heaters before entering the seal cavity. Figure 21b shows the flow factor as a function of inlet pressure at different rotational speeds ranging from 0 to 60 krpm at ambient temperature. The static test produced more leakage than higher speed dynamic tests and the leakage increased as inlet pressure increased. The heaters were then turned on and the temperature was set at 260°C (500°F). After the temperature reached steady-state, both static and dynamic performance testing were repeated. Figure 21c compares the seal performances at an operating speed of 60 krpm under ambient and 260°C temperatures and it clearly shows that the seal leakage was significantly reduced at elevated temperature. Seal performance with helium was characterized under both ambient and elevated temperatures. Figure 21d compares leakage of air and helium as a function of operating speed at an inlet pressure of 483 kPa (70 psi) when the temperature was set at 130°C (250°F). The flow factor with helium was smaller than that with air under the same conditions.

3.7. Foil Bearing Tests A series of tests was conducted with a high-speed foil

bearing tribometer under high side loads and with air and helium gas to demonstrate foil bearing liftoff and operating speed performance. The purpose of these tests was to assess the ability of the design tools to predict bearing performance for operation in hydrogen. The test setup is shown in Fig. 22.

For the start/stop testing, a static load was applied by a pneumatic loader through the isolation springs and a long flexible cable with a swivel joint and a short rod mounted to the loader bearing housing bottom dead center position. Since the bearing housing was allowed to rotate freely about the shaft spin axis, a strain gage load cell was mounted to the tribometer base and pressed

0

0.001

0.002

0.003

0.004

0.005

0.006

0.007

0.008

0.009

0 20 40 60 80 100 120Differential Pressure (psi)

Measured and Predicted Flow Factor for Air & Helium

Helium Measured (2-Foil Coated, T2R1)Air Measured (2-Foil Coated, T1R7)Helium Predicted (Load Def lection & MiTi-CFS)Air Predicted (Load Def lection & MiTi-CFS)

Flow

Fac

tor

Fig. 19 Measured and predicted flow factor in static

testing (i.e., without shaft rotation) with air and helium (100 psi = 689 kPa)

MiTi® High-Speed Motor with Integrated Turbine on Foil Bearings2.5” Seal

Assemblies & Detachable Test Journal

Coupling

Robust Spindle on Foil Bearings for Variable Size Seal Testing

Fig. 20 Fully assembled and instrumented seal

tribometer



into contact with the loader bearing housing rod to measure the start/stop torque. Loads of 35, 89, 133 and 178 N were applied to the loader foil bearing with the rotor at rest. The drive was then activated and torque measured to identify the speed for liftoff and touchdown. As seen in Fig. 23, the torque rises rapidly as the speed is increased during start-up due to initial contact and boundary lubrication. Once fluid film lubrication is

established, the torque is reduced to a small value. At the shut down, the torque increases again due to boundary lubrication and touch down. Using this approach and tracking rotor vibration frequency content, it was possible to estimate the liftoff/touchdown speeds as a function of load. These values for the test in Fig. 23 are 8,000 and 6,000 rpm, respectively. Testing was conducted in air and helium and predictions for bearing performance in hydrogen were made based on the test results (see Fig. 24). Thus, under maximum design, static loading of 133 N per foil, journal bearing rotor liftoff speed would be less than 6,000 rpm or only 10% of the maximum operating speed. Since operating speed is selected at 60,000 rpm, the low liftoff speed indicates that hydrodynamic film pressures would be fully developed and would provide a significant load carrying margin at the operating speed.

Further tests were conducted with helium and air at 260°C (500°F) at 30,000 rpm, Fig. 25. The bearing temperatures remained fairly constant under load. The difference between the bearing temperature, measured at inlet and outlet positions under the loaded region, was due to the heat generated from power loss as the air passed axially through the bearing. Bearing testing was then repeated with helium at ambient temperature (Fig.

Dynamic Testing of Four Seal Configurations at 100 psi

Speed (krpm)

0 10 20 30 40 50 60 70

Flow

fact

or

0.002

0.004

0.006

0.008

0.010

0.012

Loose radial clearance No face bump (T5R2)Loose radial clearance Loose face bump (T6R1)Loose radial clearance Loaded face bump (T7R1)Small radial clearance Loose face bump (T8R1)

Dynamic Testing at 60 krpm at Ambient and 500F Temperatures

Inlet Pressure (psi)

0 20 40 60 80 100 120

Flow

fact

or

0.002

0.004

0.006

0.008

0.010

0.012

Ambient (T9R2)500F (T10R2)

Static and Dynamic Testing at Ambient Temperature

Inlet Pressure (psi)

0 20 40 60 80 100 120

Flow

fact

or

0.002

0.004

0.006

0.008

0.010

0.012

Static (T9R2)20 krpm (T9R2)30 krpm (T9R2)40 krpm (T9R2)50 krpm (T9R2)60 krpm (T9R2)

Dynamic Tesing with Air and Helium at 70 psi under 250F Temperature

Speed (krpm)

0 10 20 30 40 50 60 70

Flow

Fac

tor

0.002

0.004

0.006

0.008

0.010

0.012

Air (T12R5)Helium (T16R2)

(a)

(b)

(c)

(d) Fig. 21 (a) Flow factor versus speed at 100 psi for four seal configurations, (b) Flow factor versus inlet pressure at

various speeds at ambient temperature, (c) Flow factor versus inlet pressure at 60 krpm under both ambient temperature and 500°F and (d) Flow factor versus speed with air and helium at 70 psi under elevated temperature of 250°F (100 psi = 689 kPa)

Fig. 22 High speed Tribometer for start/stop and side load testing of foil bearings



26). As with tests in air, only minor temperature rise was observed in the bearing indicating that a fully developed gas film was supporting the rotor.

Recognizing that the final application will be in hydrogen, bearing experimental data at different operating ambient pressures for air, helium and argon were obtained and used to assess the potential impact for hydrogen operation. Figure 27 represents the experimental data used to extrapolate and assess the expected performance of foil bearings in hydrogen at pressures up to 10 MPa (1500 psig). Using the above data, the projected foil bearing performance in hydrogen and air was predicted up to 300 bar pressure. As seen in Fig. 28, load capacity and power loss increase with increasing pressure, and minimum film thickness decreases following a maximum value at approximately 25 bar. Only minor changes in dynamic stiffness are observed.

3.8. Dynamic Testing In order to demonstrate the performance of compliant

foil bearings operating through and above the first bending critical speed, a compressor rotor simulator was designed and tested (Fig. 29). The simulator rotor was a solid three thrust disk shaft designed to exhibit similar rotordynamic behavior as the rotor group in the hydrogen compressor. Full-scale compliant foil bearings, similar

to those selected for use in the hydrogen compressor, were used in the simulator test rig. The tests successfully demonstrated stable and repeatable performance of the rotor above the first bending critical speed, as seen in Fig. 30, in which a composite of eight successive decelerations through the first bending critical speed (at 34,000 rpm) are shown. Testing was continued above the bending critical speed up to a maximum speed of 54,820 rpm (Fig. 30). These tests successfully demonstrated the damping capability of the foil bearings as the rotational speed was increased passing through the first bend.

3.9. Structural Analysis Rotor and impeller designs created through

aerodynamic analysis and modeled with rotordynamic software were verified for structural integrity using FEA. Structural analysis was focused on the rotor, impeller hub and impeller blades. Three materials were investigated in the structural design analysis: aluminum alloy; titanium alloy and high-strength steel. A summary of results for each material for the high margin design is given in Table 3. The results indicate significant stress safety margins for all three selected materials. Stress distribution for the high-strength steel impeller is shown in Fig. 31.

Lift Off/Shut Down 40 lbs Load

Time

03:47 03:48 03:490

5000

10000

15000

20000

25000

30000

0.4

0.5

0.6

0.7

0.8

0.9

1.0

1.1SpeedDate/Time vs Normalize TQ

6000 rpm8000 rpm

61186-F020

Fig. 23 Trace of rotor speed and normalized bearing

torque during rotor start up and shutdown

Speed

Bearing Temp(X=0)

BearingLoad

Bearing Temp(X=L)

20211-F002Time

14:20:00 14:25:00 14:30:000

5,000

10,000

15,000

20,000

25,000

30,000

35,000

0

20

40

60

80

100

120

420

440

460

480

500

520

540

560

Spee

d (R

PM) Load (lbs)

Temperature

( oF)

Fig. 25 Side load applied to foil bearing at a rotor speed of 30,000 rpm and air temperature of 260 oC (20 lb = 89 N)

Normal Load (lb)10 20 30 40

2000

4000

6000

8000

10000

12000

10

20

30

40

Liftoff AirTouchdown(via Torque)

Touchdown(via FFT)

Touchdown (He)

61186-F018

Predictionin H2

Fig. 24 Foil bearing liftoff and touchdown speeds as a

function of static load for air, Helium and Hydrogen (10 lb = 44.5 N)

Speed (RPM)0 5,000 10,000 15,000 20,000 25,000 30,000 35,000

0

20

40

60

80

100

120

140

160

75

80

85

90

95

100

105Load

Loader NDE

Loader DE

Helium Gas @ 0.50 lbs/min

61186-F021A

Temperature

( oF)

Load

(lbs

)

Fig. 26 Dynamic load test with helium gas at 3 gr/s

flow rate (40 lb = 178 N)



170%

100%

50%

0%0 0.5 1.0 1.5 2.0 2.5 3.0

BearingLxD= 1"x2"Korolon™ Coating

Ambient Pressure (atm)

Speed @ 18 Krpm, Temp 500 °C

Argon

Helium

Air

07-0012-HSA

Non

-dim

ensi

onal

Loa

d

100%

75%

50%

0%0 50 100 150 200 250

Bearing Load (N)

25% 2.5 atm

1.0 atm

1.5 atm 2.0 atm

Speed @ 18 Krpm, Temp 500 °C

07-0010-HSA

Non

-dim

ensi

onal

Tor

que

0 50 100 150 200 250Bearing Load (N)

2.5 atm

1.0 atm

1.5 atm2.0 atm

Speed @ 18 Krpm, Temp 500 °C100%

75%

50%

0%

25%

07-0011-HSA

Non

-dim

ensi

onal

Tor

que

Speed @ 18 Krpm, Temp 500 °C100%

75%

50%

0%

25%

0 50 100 150 200 250

Bearing Load (N)

2.0 atm

2.5 atm

1.5 atm

07-0013-HSA

Non-

dim

ensio

nal T

orqu

e

(a)

(b)

(c)

(d) Fig. 27 Experimental foil bearing load capacity as a function of ambient pressure in (a), experimental foil bearing

torque as a function of load and ambient pressure in Argon in (b), experimental foil beating torque as a function of ambient pressure in air in (c) and experimental foil bearing torque as a function of load and ambient pressure in Helium. The load and torque values are normalized (non-dimensional) with respect to a reference load.

0

100

200

300

400

500

600

700

0 50 100 150 200 250 300 350Atmospheric Pressure (atm)

Foil

Bea

ring

Load

Cap

acity

(psi

)

BearingLxD= 1.85"x2.5"Korolon™ CoatingN=50,000 rpm

Hydrogen

Air

07-0014-HSA

50

100

150

200

250

300

350


Min

Film

Thi

ckne

ss (µ

in)

175

177

179

181

183

185

187

189

191

193

195

Atti

tude

Ang

le (º

deg)


Hydrogen (µin)

Air (µin)Air (ºdeg)

Hydrogen (ºdeg)

07-0016-HSA

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1


Pow

er L

oss

(Hp/

10)

Hydrogen

Air


07-0015-HSA

Air KXXAir KYXAir KXYAir KYYH2 KXXH2 KYXH2 KXYH2 KYY

-2

-1

0

1

2

3

4

5

50 100 150 200 250 300 350

Atmospheric Pressure (atm)

Stiff

ness

(lb/

in)x

106

07-0017-HSA

(a)

(b)

(c)

(d) Fig. 28 Load capacity of foil bearing in air and hydrogen as a function of ambient pressure in (a), foil bearing minimum

film thickness and attitude angle as a function of ambient pressure for air and hydrogen in (b), theoretical foil bearing power loss in air and hydrogen as a function of ambient pressure in (c) and foil bearing dynamic stiffness coefficients as a function of ambient pressure in hydrogen and air in (d). In an axial cross-section of the rotor or bearing, X refers to the axial horizontal axis and Y refers to the vertical axis.



Fig. 29 Full scale simulator rotor for testing above the

first bending mode

(a)

(b) Fig. 31 Impeller Von Mises stress predictions; front

view in (a) and back view in (b), maximum stress of 758 MPa

010203040506070

20,000 25,000 30,000 35,000 40,000

Mag

nitu

de (µ

m)

Frequency (RPM)

Test 1Test 2Test 3Test 4Test 5Test 6Test 7Test 8

0

5

10

15

20

25

10,000 20,000 30,000 40,000 50,000 60,000

Mag

nitu

de (μ

m)

Frequency (RPM)

MaximumSpeed of 56,820 RPM

(a)

(b) Fig. 30 Rotor response during deceleration through

the bending critical speed for seven different tests in (a) and rotor response during high speed operation above the bending critical speed in (b)

Table 3 Stress analysis results for the high margin design (490 m/s) with three candidate materials

Alloy Type Aluminum Alloy Titanium Alloy High Strength Steel

Density (kg/m3) 2,768 4,484 7,750

Modulus of Elasticity (GPa) 73.1 110.0 213.7

Ultimate Strength (MPa) 579 1,172 1,999

Max von Mises Stress at 490 m/s (MPa) 272 441 758

Stress Safety Factor at 490 m/s 2.1 2.7 2.6

Fatigue Strength (MPa) 58 696 1,144

Max Blade Stress at 490 m/s (MPa) 182 266 507

Fatigue Safety Factor at 490 m/s 0.3 2.6 2.3



The results confirmed low stresses in the impeller blades. Peak stress in the impeller was on the backface of the hub; however, stress magnitudes were well below the yield strength of the material. While all three materials appeared to be acceptable with respect to the maximum stresses, the low fatigue limit of the aluminum alloy was of major concern. In fact, the blade stress is three times larger than the fatigue limit of this high strength aluminum alloy. However, both the steel and titanium alloy provide safe margins of 44% and 38% against low cycle fatigue in the blades.

Stress and structural analysis was also performed for the high performance design at tip velocities of 580 m/s. For all materials studied, the margin of safety with respect to the yield strength was reduced compared to the high margin design, as illustrated in Table 4. Thus, it appears that titanium and high strength steel are possible material candidates; however, high strength steel alloys are generally susceptible to hydrogen embrittlement. Therefore, high strength titanium alloys would be safer in the high pressure hydrogen gas. Nevertheless, it would be prudent to also consider coatings for further protection against hydrogen embrittlement.

3.10. Overall System Layout Once all components of the compressor system were

designed and evaluated, an overall system layout was conducted, including identification of drive systems, power transmission elements and intercoolers. The system package size was approximated and system cost estimates were calculated. The next step in this program consists of development of a single-stage compressor system based on the proposed design. The single stage compressor will be used to demonstrate the aerodynamic performance of the designed hydrogen compressor. Testing is currently in progress and the results will be published in due course.

4. Conclusions

The current philosophy in design of high speed rotating machines, such as centrifugal compressors, dictates the use of rolling element bearings with ceramic rolling elements, impregnated lubricious composite retainers, an oil-lubricated system, squeeze film dampers, externally cooled systems, with unmanageable thermal

conditions, Due to the zero contamination requirements and high rotational speeds, the current design approaches cannot meet the demands of advanced hydrogen compressors. In this paper we have shown that a new design approach based on foil bearings can be successfully applied to design a novel hydrogen compressor.

A completely oil-free, high-speed centrifugal compressor using 4th generation compliant foil bearings and seals has been designed for hydrogen pipeline delivery. An iterative aerodynamic/structural/rotordynamic design process was used, including both quasi three-dimensional inviscid internal flow and Computational Fluid Dynamic (CFD) analyses. The flow field was carefully analyzed for areas of excessive diffusion, sudden velocity gradients and flow separation. Excellent correlation between the preliminary design and CFD analyses was obtained.

Structural and rotor-bearing system dynamic analyses were also completed to finalize the compressor system configuration. Finite element analysis of the compressor impeller was used to verify structural integrity and fatigue limits for selected materials. Rotor-bearing system analysis was used to define acceptable bearing locations and dynamic coefficients, system critical speeds and dynamic stability. Given the high speeds, supercritical operation, and required reliability, efficiency and freedom from contaminants, compliant foil gas bearings were selected and designed. Since hydrogen will be used as the bearing lubricant for the foil bearings, substantially lower power loss than oil lubricated bearings will be experienced and the auxiliary supply or scavenge system will be eliminated.

Acknowledgments

The authors gratefully acknowledge the continued support of the US Department of Energy Fuel Cell Technologies Program, and particularly, program managers Dr. Monterey Gardiner, Mr. Paul Bakke, Dr. K. Scott Weill and Dr. Sara Dillich.

References

[1] Chemical and Market Reporter, February 24, 2003, 43.

Table 4 FEA results of the three candidate materials at various tip velocities

Tip Speed (m/s)

High-Margin High Performance Material

Max Von Mises Stress (MPa) Safety Margin Max Von Mises Stress (MPa) Safety Margin

Aluminum Alloy 272 2.1 385 1.5

Titanium Alloy 441 2.7 624 1.8

High-Strength Steel 758 2.6 1072 1.8



[2] Hydrogen, Fuel Cells & Infrastructure Technologies Program, United States Department of Energy, 2007, p.3.2-1.

[3] Blok, H. and van Rossum, J. J., “The Foil Bearing – A New Departure in Hydrodynamic Lubrication,” Lub. Engr., 9, 6, 1953, 316-320.

[4] Recorded Communication/Discussion on the Concept of Tension Dominated Foil Bearings with Professor Blok, IMECE, London, UK, 1995.

[5] Silver, A., et al., “Foil Bearings,” U.S. Patent Documents, No. 3,215,479, November, 1965.

[6] Eshel, A. and Elrod, H. G. Jr., “The Theory of the Infinetly Wide, Perfectly Flexibl, Self-Acting Foil Bearing,” J. Basic Engr., Trans. ASME, 87, 4, 1965, 831-836.

[7] Lichet, L., “An Experimental Study of Elasto-Hydrodynamic Lubrication of Foil Bearings,” J. Lub. Tech., Trans. ASME, 90, 1, 1968, 199-220.

[8] Cherubim, H., “Foil Bearings,” U.S. Parent Documents, No. 3, January, 1974, 366,427.

[9] Walowit, J. A., Murray, S. F., McCabe, J., Arwas, E. B. and Moyer, T., “Gas Lubricated Foil Bearing Technology Development for Propulsion and Power Systems,” WPAFB Technical Report AFAPL-TR-73-92, December, 1973.

[10] Walowit, J. A. and Anno, J. N., “Modern Developments in Lubrication Mechanics,” Applied Science Publishers, Ltd., 1975.

[11] Zorzi, E. S. and Koepsel, W. F., “Gas Lubricated Foil Bearing Development for Advanced Turbomachines,” AFAPL-TR-76-114, vols. I and II, November, 1978.

[12] Heshmat, C. A. and Heshmat H., “An Analysis of Gas Lubricated, Multi-Leaf Foil Journal Bearings with Backing Springs,” ASME Trans., J. Trib., 117, 3, 1995, 437-443.

[13] Anonymous, (http://www.nasa.gov/centers/glenn/about/fs14grc.html) accessed August 9, 2012.

[14] Heshmat, H., Jahanmir, S. and Walton II, J. F., “Coatings for High Temperature Foil Bearings,” ASME Paper GT2007-27975, Proceedings of the ASME Turbo Expo: Power for Land, Sea & Air, 5, May 2007, Montreal, Canada, 971-976.

[15] A low friction coating designed by MiTi - Korolon; http://www.korolon.com; http://www.miti.cc/korolon/index.html.

[16] Heshmat, H., “Tribology of Interface Layers,” CRC Press, New York, 2010.

[17] Heshmat, H., “Compliant Journal Bearing with Angular Stiffness Gradient,” U.S. Patent Documents, No. 4,262,975, April, 1981.

[18] Heshmat, H. and Chen, H. M., “Hybrid Foil-Magnetic Bearing”, Presented at Northern Europe Tribology Conference, NORDTRIB’96, Oslo, Bergen, Norway, Proceedings of the 7th

Nordic Symposium on Tribology, Volume II, June 16-19, 1996.

[19] Heshmat, C. A., Xu, D. S. and Heshmat, H., “Analysis of Gas Lubricated Foil Thrust Bearings using Coupled Finite Element and Finite Difference Methods,” Paper No. 99-Trib-34, STLE/ASME Tribology Conference, Orlando, FL, October 10-13, 1999, Journal of Tribology, 122, 1 Paper No. 99-Trib-34, 2000, 199-204.

[20] Heshmat, H., “Compliant Foil Seal,” U.S. Patent Documents, No. 6,505,837, 2003.

[21] Heshmat, H., Zhaohui R., Hunsberger, A. Z., Walton, J. F. II and Jahanmir, S., “The Emergence Of Compliant Foil Bearing And Seal Technologies In Support Of 21st Century Compressors And Turbine Engines,” to be presented at the ASME 2010 International Mechanical Engineering Congress & Exposition, Paper No. IMECE2010-40598, November 12-18, 2010, Vancouver, British Columbia, Canada.

[22] Walton II, J. F. and Heshmat, H., “Application of Foil Bearings to Turbomachinery Including Vertical Operation,” ASME Paper No. 99-GT-391, ASME Turbo, Land, Sea and Air, June 1999, Indianapolis, India.

[23] Heshmat, H., “The Integration of Structural and Fluid Film Dynamic Elements in Foil Bearings – Part I: Past Approaches to the Problem,” ASME Paper No. DETC99/VIB-8271, ASME Design Engineering Technical Conference, September 12-15, 1999, Las Vegas, Nevada.

[24] Heshmat, H., “Operation of Foil Bearings Beyond the Bending Critical Mode,” ASME/STLE Joint Tribology Conference, Orlando, FL, October 10-13, 1999, Journal of Tribology, 122, 1, 2000, 192-198.

[25] Heshmat, H., Walton II, J. F., Della Corte, C. and Valco, M., “Oil-Free Turbocharger Demonstration Paves Way to Gas Turbine Engine Applications,” ASME Paper 2000-GT-0620, ASME Turbo Expo, Land, Sea and Air, IGTI, Munich, Germany, May, 2000.

[26] Heshmat, H., Walton II, J. F., Tomaszewski, M., Cruzen, G. S. and Rabe, L., “A Turbojet Engine Demonstration with Air Foil Bearing,” UVS Tech 2003, Royal Military Academy, Brussels, Belgium, December 3-5, 2003.

[27] Tomaszewski, M., Walton II, J. F. and Heshmat, H., “On the Coupling of Foil Bearing Supported Rotors: Part 2 – Experiment,” ASME Paper GT2007-27825, Proceedings of the ASME Turbo Expo: Power for Land, Sea & Air, Volume 5, pp. 951-960, May 2007, Montreal, Canada.

[28] “Compliant Foil Seal.” U.S. Patent No. 6,505,837, Jan. 14, 2003.

[29] Salehi, M., Heshmat, H., Walton II, J. F. and Cruzen, S., “The Application of Foil Seals to a Gas Turbine Engine,” AIAA Paper 99-2821, 35th AIAA/ASME/SAE/ASEE Joint Propulsion



Conference and Exhibit, June 20-24, 1999, Los Angeles, CA.

[30] Salehi, M. and Heshmat, H., “On the Fluid Flow and Thermal Analysis of a Compliant Surface Foil Bearing and Seal,” Tribology Transactions, 43, 2, 2000, 318-324.

[31] Salehi, M. and Heshmat, H., “High Temperature Performance Evaluation of a Compliant Foil Seal,” AIAA 2000-3376, 36th AIAA/ASME/SAE/ASEE Joint Propulsion Conference and Exhibit, July 16-19, 2000, Huntsville, Alabama.

[32] Salehi, M. and Heshmat, H., “Performance of a Complaint Foil Seal in a Small Gas Turbine Engine Simulator Employing a Hybrid Foil/Ball Bearing Support System,” STLE Transactions, 44, 3, 2001, 458-464.

[33] Heshmat, H. and Chen, H. M., “Hybrid Foil-Magnetic Bearing,” Presented at the Northern Europe Tribology Conference, NORDTRIB’96, Oslo, Bergen, Norway, June 16-19, 1996, Proceedings of the 7th Nordic Symposium on Tribology, Volume II.

[34] Heshmat, C. A., Xu, D. S. and Heshmat, H., “Development of Advanced Thrust Foil Bearings,” Presented at the 54th STLE Annual Meeting, Las Vegas, NV, May 23-27, 1999. Submitted for publication in the Tribology Transactions.

[35] Heshmat, H. and Walton II, J. F., “Principles of Gas Turbine Bearing Lubrication and Design,” Chapter 7 in the “Handbook of Lubrication and Tribology”, Volume I, Application and Maintenance, Second Edition. Sponsored by the STLE and published in June 2006 by CRC Press, Taylor & Francis Group.

[36] Heshmat, H. and Chen, H. M., “Principles of Bearing Design,” Chapter 19 in the “Compressor Handbook”, published in January 2001 by McGraw-Hill.

[37] Heshmat, H. and Shapiro, W., “High-Temperature, Unbalanced, Dry Contact Face Seal. Interfacial

Phenomenon and Design Considerations,” Journal of STLE, 45, 4, April, 1989, 235-242.

[38] Salehi, M. and Heshmat, H., “Evaluation of Large Compliant Foil Seals under Engine Simulated Conditions”, AIAA Paper 3729, 38th AIAA/ASME/SAE/ASEE Joint Propulsion Conference, July 7-10, Indianapolis, India, 2002.

[39] Salehi, M. and Heshmat, H., “Analysis of a Compliant Gas Foil Seal with Turbulence Effects”, AIAA 2001-3482, 37th AIAA/ASME/SAE/ASEE Joint Propulsion Conference & Exhibit, July 18-11, Salt Lake City, 2001.

[40] Salehi, M., Heshmat, H., DellaCorte, C. and Valco, M., “High Temperature Operation of a Foil Seal in a Small Gas Turbine Simulator Employing a Hybrid Foil /Ball Bearing Support System”, Presented at 2000 International Mechanical Engineering Congress and Exposition (IMECE), November 5-10, 2000, Orlando, Florida.

[41] Heshmat, H., Ren, Z., Hunsberger, A. Z., Walton J. F. and Jahanmir,S., “The Emergence of Compliant Foil Bearing and Seal Technologies in Support of 21st Century Compressors and Turbine Engines,” IMECE2010-40598, ASME International Mechanical Engineering Congress Proceedings, Nov. 12-18 2010, Vancouver, BC.

[42] Heshmat, H., Hunsberger, A. Z., Ren, Z., Jahanmir, S. and Walton, J. F., “On the Design of a Multi-Megawatt Oil-Free Centrifugal Compressor for Hydrogen Gas Transportation and Delivery Operation beyond Supercritical Speeds,” IMECE2010-40575, ASME International Mechanical Engineering Congress Proceedings, Nov. 12-18 2010, Vancouver, BC.

[43] Heshmat, H., “Compliant Foil Bearing Technology: An Overview,” Bruce, R. H., editor, Handbook of Lubrication and Tribology, Vol. II, Second Edition, CRC press, New York, 2012, Chapter 47.