DESIGN STUDY OF A LOW-COST LOX TURBOPUMP

Dr. David Japikse

Dr. Nicholas Baines
Michael J. Platt
Concepts ETI, Inc.
217 Billings Farm Road
White River Junction, VT 05001-9486

ABSTRACT

A preliminary design study, focusing on potential component selections and Design for Manufacturing and
Assembly (DFMA1) analysis, is presented in this study. The investigation was focused on a nominal cost liquid
oxygen turbopump suitable for a private launch class vehicle. Utilizing a “turbocharger-like” design philosophy,
preliminary feasibility studies of the basic pump design class, the rotordynamic design class, and the turbine design
class were conducted with associated DFMA evaluations. Reasonable cost levels and sensible levels of product
assurance have been established.

NOMENCLATURE
β1bt Inlet blade angle at the shroud p00 Inlet stagnation pressure
β2b Impeller exit blade angle pexit Exit static pressure
ηcalc Calculated stage efficiency prtt Total-to-total pressure ratio
ηgoal Desired stage efficiency prts Total-to-static pressure ratio
b2 Impeller exit width r1t Inducer eye radius
b5 Diffuser exit width r2 Impeller exit radius
m Flow rate r5 Diffuser exit radius
N Rotational speed T00 Inlet stagnation temperature
NPSH net positive suction head

PROJECT OBJECTIVES

The fundamental objective of this preliminary scoping project was to establish a reasonable estimate of the most
likely cost of producing, in series production, both a LOX and a LH2 rocket turbopump for a 40K upper-stage
engine. The LH2 study will be presented elsewhere (see Japikse, et al., 2000[1]). The operating parameters for this
stage were established by an external study as displayed in Table 1. It is understood that these numbers will change
as further iterations on the design cycle are conducted. However, the range of conditions envisioned do not present
concerns; the feasibility questions will not be materially affected by such considerations.

1
DFMA is a registered trademark of Boothroyd Dewhurst, Inc.
 Table 1. Oxidizer Turbopump Turbomachinery Variables

Inlet Flow Rate lb/sec 78.16

Inlet Pressure psia 50
Inlet Temperature R 183
Inlet Vapor Pressure psia 42
Inlet NPSH ft 17.1
Discharge Flow Rate lb/sec 78.16
Discharge Total Pressure psia 585
Pump Head Rise ft-lb/lb 1133
Pump Head Coefficient 0.5
Pump State Ns ≈1500
Pump Torque ft-lb 93.5
Pump Horsepower Hp 212
Desired Pump Isentropic Efficiency 75%
Turbine Inlet Flow Rate lb/s 0.914
Turbine inlet flange tot temperature R .1132
Turbine Inlet flange tot pressure psia 75
Turbine disch flow rate lb/s 0.914
Turbine disch static pressure psia 30
Isentropic velocity ratio 0.131
Turbine pressure ratio (T-S) 2.5
Turbine torque ft-lb 93.5
Turbine horsepower hp 212
Desired turbine isentropic efficiency (T-S) 42%
Turbine speed (estimated) rpm 11905

In order to reach the appropriate determination of probable production costs, CETI has conducted a preliminary,
but comprehensive design of a LOX turbopump. The actual numbers should not be construed as suitable for any
construction at the present time; they represent a pump that would work, with nominal life, but definitely do not
represent a fully iterated design of final quality. Due to the existence of the Agile Engineering Design System2, it
was possible to look at all elements in a quick, initial design in order to establish sufficient details for costing
purposes. These details are reasonably assured of meeting a ten-minute life with one restart capability. While
further work must be conducted in order to effect the final design, it may be safely taken that the preliminary design
configuration gives fair confidence for establishing reasonable component costs.

FEASIBILITY APPROACH

The objectives presented above should lead to the eventual design, prototyping, and production of a liquid
oxygen turbopump. Early cycle studies suggested that a single-stage pump might be feasible and that a two-stage
turbine may be required. In order to conduct the necessary feasibility studies, a preliminary (but intentionally
incomplete) design of the full liquid oxygen pump was prepared. This is shown in Figure 1. This design focused
only on the broad aspects of the principal features of the design. A preliminary pump design, a preliminary turbine
design, a preliminary rotordynamic system including bearings and seals, and preliminary structural and cost
assessments were conducted. In order to perform these evaluations, the highly flexible, concurrent, Agile
Engineering Design System of Concepts ETI, Inc. (CETI) was employed. This Agile Engineering Design System is
illustrated in Figure 2.

2
Agile Engineering Design System is a registered trademark or Concepts ETI, Inc.

2
 Figure 1. Conceptual LOX pump layout (subject to detail revisions).

The Agile Engineering Design System includes all engineering tools necessary for the complete design and
subsequent development of advanced rocket turbopumps. A copy of this system is in use at NASA MSFC (Marshall
Space Flight Center), which has aided in its development through various SBIR projects. Section 1 refers to
meanline analyses, which are conducted after the cycle calculations (upper left-hand corner, “A”) are conducted.
The cycle calculations were conducted externally and iterated with CETI engineers. Initial one-dimensional
calculations, as indicated in Section 1, were conducted for all components of the LOX pump and then preliminary
blading designs were conducted, as illustrated in Section 2. CFD calculations (Sections 3-5) were not conducted or
pursued for this evaluation. Structural calculations of pump impellers and turbine rotors were conducted on a
scoping basis. This covers Sections 6-7 of Figure 2. Subsequently, rotordynamic calculations, as shown in Section
8 were conducted and appropriate bearings and seals were selected for this preliminary evaluation. In addition, the
design for manufacturing (DFM) and design for assembly (DFA) evaluations of Block B, upper right-hand corner of
Figure 2, were of great value.

CYCLE CODES DFM, DFA, DFE

STEAM PLANTS REFRIGERATION Design for Manufacturing
GAS TURBINES GEAR COMPRESSORS Design for Assembly
Design for the Environment
TURBOCHARGERS
3242R6
ROCKET TURBOPUMPS CETI

Compressor, Pump, and Turbine

Meanline (Quasi-2D)

SE Design Optimization:
BA COMPAL, PUMPAL, RITAL,
A

AXIAL 3D Geometry Generation

AT

and Throughflow Analysis

D

Design
Analysis - Full CCAD
L

A Performance Curves AXCAD

B
NA

"Synthesis":
Data Reduction
Semi-Automated Comparison and Comparison 2D and 3D Description
O

of Design Intent to Tested Design of Rotors and Stators

ATI

Rapid Loading Calculations

1
Laboratory Quasi 3D Flow Calculations
R

Testing for Reverse Engineer

OPE

Component 12 Existing Designs

ASE

Development Finite Element

Pneumatic 2 Analysis (FEA)
Laser
DATA B

Hot Film Product CFD Preprocessor

Development "Virtual
Thermal Preprocessing STRESSPREP
Structural System Laboratory": System Thrust, AXISTRESS
Time CFD Studies Leakage Balance and Pushbuttton CFD
Dependent Supporting the
11a Physical Lab Turbine Cooling
11b 3 6
ION AL

Computational Fluid
Rapid Prototyping Dynamics (CFD)
Stereo Lithography Life Design Optimization
Sintered Casting Cores Pushbuttton CFD 3D FEA for
Evaluation
R AT

Numerical Machining 10 Stress and

Ruled Surfaces Vibration Analysis
Strip Milling
OP E

4
Arbitrary Surface 9

Computer Aided Pushbutton CFD

Drafting (CAD) Postprocessing
and Solid Modeling (SM) 8 7a
Solid Edge

D
AutoCAD Mechanical Desktop
Pro-Engineer
SDRC I-DEAS
Rotor Dynamics
5
FEA
C
Catia, and Others Postprocessing
System Issues RotorLab Campbell Diagram
Gears
Interference Diagram
Motors Seals
Generators Life Evaluation
Bearings
Cooling
Critical Speeds
Unbalance Response
Forced Response
7b

Component & OP E Design Procedure &

R AT ION ASE
System Optimization
AL DATA B Parameter Limits

ENGINEOUS RULES

Figure 2. The integrated turbomachinery agile engineering design and technology system.
The arrows indicate the flow of information (Japikse 1999[2]).

3
 The resulting preliminary design is shown on a provisional basis in Figure 1, as mentioned. This conceptual
figure shows roughly the type of layout (to scale) that could be developed into a successful turbopump. However,
further modifications are expected. Considering Figure 1, several notes can be offered. The two-stage turbine, as
shown on the right-hand side, does not, as yet, illustrate the location of nozzle guide vanes. These would be added
in a later stage of design and development. Additionally, the direction of flow through the turbine has not yet been
set. This would be chosen as details of thrust balancing are evaluated later. Appropriate seals on the backface of the
pump impeller and additional seals around the turbine rotors would be considered as the overall thrust balancing is
achieved. This detailed leakage path calculation and force balancing would be done later with a separate analysis,
suggested at the center of Figure 2. This is a tedious exercise, but not particularly difficult in practice; there is no
risk in this area when the work is carefully executed, and, hence the details were saved for later. The precise
location of the mechanical face seal may very well be modified at a later time, but the initial configuration offers a
very sensible ‘strawman’ for review and analysis. A continuous, low level of leakage from the pump impeller
backface through the bearings and back into the pump inlet (not shown) is anticipated in order to maintain
appropriate bearing temperatures and acceptable life.

The housing layout is strictly preliminary. Modifications to the pump diffuser configuration on the left are
certainly to be expected and the size of the housing components will be adapted with some design synergy
developed between the pump and turbine housing components. It is possible that a single casting may be employed
for the inner housing (although the part line above the pump side bearing suggests a two-piece construction). The
overall configuration resembles a high-performance turbocharger. Manufacturing techniques have been widely
developed for nearly one century for the economic production of turbochargers and much good insight enters from
this industry. The only significant difference is the existence of a two-stage turbine rather than a single-stage turbine
which is common for the turbocharger application. All other elements may be found in turbocharger systems.

CETI has worked very closely with MSFC and with Boothroyd Dewhurst, Inc. (BDI) to develop a preliminary
costing system for rocket turbopumps. Substantial progress has been made and correlated data have been
thoroughly exploited in this investigation. Details of the preliminary design studies are now given in the subsequent
sections.

PUMP DESIGN

The design of the LOX pump was a straightforward process. CETI presently has under development several
high-performance rocket turbopump configurations that have evolved from a careful business relationship with one
of the leading suppliers to NASA MSFC and to the Phillips Laboratory, USAF (also a major SBIR source for CETI
system development). These designs incorporate a single-piece inducer and pump which can achieve a suction
specific speed on the order of 30,000. Full design head can easily be made in a single stage. CETI has designed
such stages and taken water-rig test data to confirm these design levels. There are no significant hydrodynamic or
structural problems associated with the pumps which have not been addressed in past design and laboratory
experimental evaluations. Several alternative diffuser options are available; for the purpose of the preliminary
study, a cascade or airfoil type diffuser was utilized. However, in the final design study, channel diffusers, conical
diffusers, vaneless diffusers, and other types of compact diffusers will be considered.

The pump design process begins with finding the optimum meanline design specifications using PUMPAL3
(Section 1, Figure 2). An extensive database is available to guide the sensible application of PUMPAL and this was
utilized to establish the expected impeller exit radius, impeller exit tip depth, and appropriate blade angles.
Likewise, the diffuser and volute parameters were laid out using PUMPAL. Following the meanline analysis, the
CCAD4 program (Section 2, Figure 2) was used to lay out a preliminary impeller geometry. While this geometry
is far from optimized and complete, it closely resembles previous impellers. This follows naturally, since design
information from previous designs was imported into this particular evaluation for purposes of expediency. It is
expected that the final design will differ in meaningful, but comparatively small, ways. All design calculations were
made using real fluid properties as the PUMPAL calculations used a compressible evaluation of the liquid oxygen.

3
PUMPAL is a registered trademark of Concepts ETI, Inc.
4
CCAD is a trademark of Concepts ETI, Inc.

4
 Table 2 summarizes the important design parameters for the LOX pump. The basic size parameters, critical
flow angles, design head, and efficiency conditions are displayed. A reasonable estimate, at the point of this scoping
study, is that the efficiency of the pump is approximately 74%, +2, -3. A broad band of uncertainty is projected at
this point, pending final design optimization. It is anticipated that efficiencies in the higher range cited may very
well result from the final optimization study. The design is considered to be advanced, but not high risk.

Table 2. LOX pump design specifications and results (preliminary).

Specification
p00 50 psia
T00 183°
m 78.2 lbm/s
N 11,905 rpm
NPSH 17.1 ft
pexit 585 psia
ηgoal 75%
Preliminary Design Results
r1t 1.745 in.
β1bt 8.40°
r2 2.680 in.
b2 0.283 in.
β2b 32°
r5 4.50 in.
b5 0.241 in.
ηcalc 0.75

TURBINE AERODYNAMIC DESIGN

Preliminary design studies were undertaken for turbines to drive the LOX pump. These studies were limited to
a basic scoping and proof of concept, and further effort would be required for design optimization and detail design.
The operating speed and power output were defined by the pump requirements. The inlet conditions to the LOX
pump turbine were defined by the analysis of the complete turbopump cycle although in practice, because the two
turbines operate in series, the inlet temperature and pressure for the LOX turbine depends on the power and
expansion ratio of the fuel turbine. However, the design concept for the LOX turbine is not particularly sensitive to
changes in these parameters. The working fluid is products of oxygen and hydrogen combustion in the gas
generator, and comprises gaseous hydrogen and steam. The operating conditions are summarized in Table 3.

Table 3. LOX turbine operating conditions.

Mass flow rate (lb/s)

(44% H2, 56% H2O) 0.914

Inlet total pressure (psia) 75

Inlet total temperature (R) 1420

Shaft speed (rpm) 11905

Power (hp) 217*

* Pump power + 2% bearing loss assumed.

5
 The turbine design approach was to select a range of values of stage loading and flow coefficients. These are
fundamental non-dimensional parameters which describe the power output per unit flow rate, and the flow rate
itself, respectively, for the turbine. The maximum blade speed was dependent on the rotor and disk material limits,
and to determine that the aerodynamic and mechanical designs were carried out in parallel with iteration between
them. Multistage concepts were also investigated because of the high powers required in both applications. With
these parameters the basic velocity triangles at inlet and exit of each blade row could be calculated, and from that
preliminary estimates of the key blade geometric parameters, such as mean radius, blade height, and blade angle,
were made. Candidate designs were then set up and analyzed in CETI’s axial turbine meanline analysis program
AXIAL5 (Section 1, Figure 2), which permits the performance of a single or multistage axial turbine to be
calculated. Once a satisfactory design was achieved, the results were transferred into AXCAD6 (Section 2,
Figure 2), where the actual blade shapes were set out. No optimization of the blade profiles was attempted, but
experience was used as a guide to provide suitable forms. The output of AXCAD was used for structural evaluation
(Sections 6 and 7, Figure 2).

Several design concepts were considered for the LOX pump turbine. The requirements of output power, blade speed,
and shaft speed for this application were such that it was possible to design a single-stage turbine, albeit a very highly
loaded one. The final design, summarized in Table 4, is a supersonic impulse turbine with a high level of exit swirl. A
limited amount of partial admission (0.82 of the total number of blade passages are open and flowing, the remainder
are closed) was required to keep the blade height and the passage aspect ratios within reasonable limits. The stator
exit Mach number is 1.306. The design of nozzles to achieve such a high Mach number is a complex task that would
almost certainly require some experimental development. Furthermore, such nozzles do not operate well at
conditions other than the design point.

Table 4. Summary of LOX turbine design concepts.

Descrip- prtt prts Efficiency Efficiency Stator Rotor exit Exit Partial Mean Inlet
tion tt ts exit relative swirl admis- radius blade
Mach no. Mach no. angle sion (in.) height
(in.)
LOX Single- 3.50 3.96 0.238 0.220 1.306 0.551 -57.5 0.82 5.804 0.392
Pump stage 8
Turbine super-
sonic
impulse
turbine
Two-stage 1.67 1.69 0.534 0.521 S1: 0.840 R1: 0.244 -10.4 0.79 5.790 0.434
subsonic 3 S2: 0.706 R2: 0.212
impulse
turbine

The principal limitation of this design, however, is the low efficiency and the corresponding large expansion
ratio that is required. The overall turbopump cycle analysis shows a gas generator delivery pressure of 394 psia. The
fuel pump turbine inlet pressure was assumed to be 384 psia, allowing 10 psia for piping losses. For a fuel pump
turbine expansion ratio of 5.5, which should be achievable with some exhaust diffusion, the LOX turbine inlet
pressure is 70 psia. With a LOX turbine total-to-static pressure ratio of 3.97 shown in Table 4, the LOX turbine
exhaust pressure is only 18 psia, or possibly as high as 20 psia with some exhaust diffusion. The acceptable limit for
this pressure is uncertain at present, but an early turbopump cycle analysis showed a LOX turbine exhaust pressure
of 30 psia, which is considerably larger than achievable with the single-stage turbine. Even if this expansion ratio
proves to be acceptable, there are still considerable design difficulties associated with the single-stage concept,
because of the high supersonic flow and the high blade loading, and, therefore, an alternative was sought.

The two-stage turbine, also shown in Table 4, represents a somewhat more conservative design, but also one
which can be considered to have a much higher probability of success. The expansion ratio is much lower, at 1.67
total-to-total and 1.69 total-to-static, and this corresponds to an exhaust pressure of approximately 43 psia. The
Mach numbers are well below sonic in all of the blade rows, and the exit swirl angle, at -10°, is quite small. In this

5
AXIAL is a trademark of Concepts ETI, Inc.
6
AXCAD is a trademark of Concepts ETI, Inc.

6
case, the rotors of the two stages are identical in section, and so too are the stators of the two stages. All of the
blades are two-dimensional. Although the two-stage concept will be heavier than the single-stage, these design
measures will at least keep down the design and manufacturing costs, and overall it can be considered to be a much
more feasible design solution. The best compromise between the conflicting requirements of aerodynamic
performance, structural integrity, size, and manufacturing cost appears to be a two-stage turbine. The oxidizer pump
turbine is not particularly highly loaded, but is still a moderately challenging design. The design efficiency is
estimated at ηT = 0.53 ±.05. Although low, it will be quite acceptable in the intended cycle.

The turbine design is intended only to demonstrate that the concept is feasible and to provide first-order
estimates of performance. The designs are not optimized and it is clear that further development would be required
so that the performance estimates can be refined, and before detail design is undertaken. It is important that the
design and development process should include experimental testing of prototype turbines. Because of the inevitable
uncertainties in the design of such highly loaded machines, experimental testing is necessary to ensure that the
design goals are met.

STRUCTURAL AND ROTODYNAMIC EVALUATION

With its impact on bearings, seals, and rotor bore stress, the initial step in the structural evaluation of the LOX
pump was sizing the shaft. The main consideration was carrying the torque from the turbine, including the stress
concentrations at the shaft shoulders. Using a torque of 94 ft-lb, and a design stress of 150 ksi, the minimum shaft
diameter will be roughly 0.39 in. for fillets in torsion. The rotor engagement could be a simple clamped stackup, an
area where CETI has considerable experience. In addition to a clamp load, either splines or a polygon drive could
be used, but the high stress concentrations would rule out a simple keyway. Although these limits do not present
any immediate problems, the steady-state and transient thermal effects on the shaft strength will need to be
evaluated in detail.

The final bearing selection for this pump will be an easy task, as this falls well within the capability of
preloaded angular contact ball bearings. The turbopump layouts show that the bearing bore diameter will likely fall
in the 20 - 25 mm range, putting the DN values in the 250,000 – 300,000 range. Such low DN values will allow
some latitude in setting the preload and choosing the component materials. The bearing races will most likely be
either 52100 or 440C alloy races, with steel or silicon nitride balls.

It is assumed that the bearings would be required to run without lubrication, and would be cooled with a
metered flow of the pumpage. To assure zero leakage, this will require a mechanical seal, possibly backed up with a
gas-buffered labyrinth, between the bearing and the turbine. The mechanical seal will most likely be a face seal,
although a gas-buffered segmented seal could also work in this application. The face seal package would utilize a
carbon seal and a stainless steel mating ring. The secondary seal would be either a metal bellows or a polymer, such
as Vespel. The surface speeds for the seal will be on the order of 100 fps, which is well below the 450 fps working
limit of typical face seals. Another limit of such seals is pV, where the upper limit is generally in the 500k-1M psi-
fpm range. In this case, the pV limit translates into a pressure differential upwards of 100 psi across the seal face.
Since the thrust loads have not yet been balanced, there is still work to do before the seal configuration is finalized.

The pump rotor stress analysis, see Platt and Marscher, 1993[3], was conducted with a combination of 2D and
3D FEA models using blade shape information from CCAD linked through OLE to the STRESSPREP7 model
generation and post-processing code. Given the low tip speed, the stress and deflection on the pump is well below
the material yield. Within reasonable limits, this will allow the selection of the most economical materials and
manufacturing processes. The blade and disk natural frequencies are very high compared to running speed-related
excitations and are not likely to present problems.

The turbine stress analysis was done in a similar fashion – making 3D FEA models with AXCAD blade shape
information linked to AXISTRESS8 for model generation and post-processing. Only the first turbine stage was
analyzed since the two stages are very similar in terms of blade shape, tip speed, etc. As with the pump stress
analysis, the results for the turbine reflect the relatively low tip speed. The results, with only minimal disk shape
iterations, show roughly 30 ksi at the disk bore and 20 ksi at the blade root. While the blade and disk natural

7
STRESSPREP is a trademark of Concepts ETI, Inc.
8
AXISTRESS is a trademark of Concepts ETI, Inc.

7
frequencies are expected to change somewhat as the final geometry is defined, the preliminary check shows no areas
for concern.

The rotor dynamics was investigated using the RotorLab9 software package, and critical speed maps and
forced response predictions were made for a variety of turbopump layouts. The results show that the LOX
turbopump will run sub-critical with a good margin of safety relative to the first critical speed. This margin will
give some latitude in the final design, since small changes in the layout can be accommodated and the shaft size
could always be increased if needed because of the already conservative DN and pV values for the bearings and
seals. The final analytical results will vary depending on the bearing preload and stiffness, added mass and damping
effects from the liquid, and contributions from the labyrinth seals.

DFMA ANALYSIS AND PROJECT COSTS

The last principal step of the feasibility assessment for the LOX and LH2 rocket turbopumps was the
preparation of a proper analysis (albeit on a preliminary basis) in order to ascertain the expected unit cost in both
low numbers and high volume (50 per year) production. In order to do this, a typical turbopump was laid out and
every part identified. Figure 3 shows the identification of every piece in the LOX turbopump. Additionally, the
heat shield and the associated bolting ring was assumed, which is not shown in this figure. Otherwise, a very
reasonable (preliminary) assessment of all components is presented. The components identified in Figure 3 are
listed in Table 5. A prior example of a turbopump and a turbocharger DFMA study is given by Gauthier, et al.,
(2000[4]).

Figure 3. LOX Turbopump component identification.

Also used for the LH2 study.

9
RotorLab is a trademark of Concepts ETI, Inc.

8
 Table 5. LOX pump component listing.

HOUSINGS PRE-ASSEMBLY MACHINING OPERATIONS

H1 – pump front
1. Pump stationary cover (H1)
H2 – pump rear
2. Diffuser ring 6” diameter
H3 – bearing
3. Housing H2 (2x’s 8” diff plate)
H4 – stator 1
4. Bearing house H3 (3x’s 8” diff plate)
H5 – stator 2
5. Nozzle faces
H6 – stator 3
H4
H7 – retain ring (use cover cost 4” size)
H5
SHAFT H6 + exit flange
1 – main rotor 6. Impeller bores and facing and tip dia.
BEARINGS Trimming (2 x’s)
B1 - Ball bearings (2) 25% increase due to hardness in above
SPACERS
Cleaning 2 x 40 hours (initial)
S1 + support
S2
S3 Assembly 2 x 40 hours
S4 + support
SPRINGS
1 set (12 pieces) Inspection 40 hours
SEAL CARTRIDGE + support
1 primary seal set
Special capital equipment amortized
LABYRINTH SEALS Balance, machine centers, inspection
L1 - piece, turbine side equipment
IMPELLERS
1 – pump impeller (machined) Miscellaneous (place holder – forgot what?)
2 – turbine impeller (2 pieces) cast
O-RINGS Balancing (4x’s) use 5” size
O1 – inlet and outlet flanges (4) Tooling, special fixtures
O2 – volute flanges
O3 – bearing housing MATERIALS
O4 – seal housing (1) Pump imp. 100 in3
O5 – seal housing (2)
Shaft 25 in3
O6 – turbine housing (1)
Labyrinth 25 in3
O7 – turbine housing (2)
H3 ring 25 in3
O8 – turbine housing (3)
Spacers 4x’s 2 in3
NUT
N1 - turbine/shaft
BOLTS W/TAB LOCKS
BR1 - inlet flange (12 pieces)
BR2 - volute flange (12 pieces)
BR3 - inner locking ring (12 pieces)
BR4 - seal assembly (8 pieces)
BR5 - heat shield (12 pieces)
BR6 - turbine inlet (12 pieces)
BR7 - turbine housing (24 pieces)
BR8 - pump exit flange (12 pieces)
BR9 - turbine exit flange (12 pieces)
HEAT SHIELD
1 – sheet metal – flat

9
 As presented previously, (Japikse, et al., 2000[5]; Gauthier, et al., 2000[4]), CETI has already established
manufacturing costs of each individual component in rocket turbopumps under contract for NASA MSFC. Further
refinement of these charts is ongoing at the present time, but they are sufficiently accurate to provide a very sensible
preliminary assessment. Additionally, based on data being acquired for the DFMA analysis for MSFC, reasonable
estimates of assembly time, clean room operation, and cleaning have been made. The charts of component costs
include the cost for making a first item, the cost for making a second item, the cost for making a fifth item, then the
20th item and then the 100th item. This database has been checked, partially, with other manufacturers so as to begin
the process of detailed validation. The specific listing of individual costs is proprietary information and not included
here.

Based on the analysis conducted, the probability is high that the LOX pump can be built for approximately
$100,000, probably for slightly less.

It is important to justify the costs above from a practical perspective. Compared with current rocket turbopump
manufacturing costs, these costs may seem overly competitive (i.e., too low). There are important factors to
consider. First, all the costing is based on reliable estimates representing parts of the type to be used in these actual
projects. Virtually all these parts, in some form or another, have all ready been manufactured at CETI. The CETI
database goes back over 18 years of prototype component manufacturing and has been thoroughly reviewed and
incorporated in the DFMA database. All parts which are required for these turbopumps can either be manufactured
at CETI or procured from outside shops. The pricing assumes that the lowest cost source of quality parts will be
used. In other words, the CETI internal shop (small but extremely comprehensive) will bid against outside vendors
and vice versa. All outside vendors understand that any part requested from CETI can, in fact, be also manufactured
within CETI. Consequently, CETI holds an advantage against a late delivery of parts, so-called “acts of God or acts
of war,” which would upset a flow of components. At the same time, this approach utilizes good cost savings from
outside vendors.

Early in this report, a reference was made to turbocharger design and development. Turbochargers of
approximately the same size and comparatively similar complexity, are manufactured for large off-road automotive
applications, small marine applications, aviation applications, and very small locomotive applications. Mass-
produced common truck turbochargers generally cost on the order of $350 to $1,000. Small marine turbochargers
will cost anywhere from a minimum of $2,000 or $3,000 up to $10,000. Turbochargers of this size and complexity,
but used for the light (piston engine driven) aircraft industry, are priced up to $25,000 per copy (but 40% of this
price is purely insurance costs). The biggest differences between the rocket turbopump application and the
turbocharger application are the two-stage turbine which is required for these turbopumps, the much higher
production level of the turbocharger, and the great difficulty brought about by rapid thermal transients during start-
up of the rocket turbopump. It may be noted, however, that the production levels for aircraft turbochargers are
roughly on the same order of magnitude as the production levels for these rocket turbopumps. Additionally,
production levels for some of the marine turbochargers are approximately the same. The dimensional accuracy of
the turbochargers is approximately the same as that required with these rocket turbopumps in many key areas,
particularly in the shaft and bearing system and the balancing requirements. The higher prices quoted herein for the
rocket turbopump reflect the differences mentioned plus the need to assure traceable manufacturing and assembly
methods and the requirements for extreme cleanliness.

SUMMARY

An appropriate feasibility study has been conducted of a preliminary design layout for a liquid oxygen rocket
turbopump. No insurmountable technical problems have been identified. To achieve a practical design
configuration, the turbopump is designed so that it roughly resembles a modern turbocharger. By following
turbocharger design and manufacturing practices, as closely as feasible, reasonable costs can be achieved. Pump
efficiencies in the range of 74%, +2, -3, are considered realistic; turbine efficiencies are on the order of 53% for the
LOX turbopump. These efficiencies are ±5 points at the present time. The purpose of a subsequent design
optimization study would be to simplify the design further, to reduce the cost further (if possible), and to increase
efficiency wherever possible. The LOX turbopump operates well below its first shaft critical speed.

10
 It is recognized that the cycle specifications will change. The parameters of this study can be readily adapted to
new configurations. However, it is not anticipated that any fundamental hurdles will be raised by subsequent
modifications in the operating cycle.

Subsequent analysis will concentrate on further steady-state operating conditions and the full design
optimization required for eventual product release. However, comprehensive transient analysis must also be
conducted to look at questions of heat soak and thermal response of the entire system. No assessment of these
transient problems has been conducted and it is expected that some tailoring of the internal leakage/cooling system
will be required. This is a common step in the design of a rocket turbopump. Due to the simplicity of the present
design, it is unlikely that any insurmountable problems would be found on the transient basis, given sufficient time
and resources to carefully conduct appropriate analyses and adapt the design accordingly.

It is presently concluded that the LOX turbopump can be manufactured for a cost of approximately $100,000
per copy in lots of 50 per year for a private launch business. Nominal assumptions have been made for clean-room
requirements, and for documentation according to standard requirements of any quality, engineered product process.
No allowance has been made for government/military level reviews, documentation, or traceability. If additional
burden is introduced in this area, some nominal increases in estimated costs are to be expected.

REFERENCES

1] Japikse, D., Baines, N. C., Platt, M., “Design Study for a Low-Cost LH2 Turbopump,” to be presented at the
Joint Army-Navy-NASA-Air Force (JANNAF) Propulsion Meeting, November 13 – 17, 2000.

2] Japikse, D., “Restructuring of Engineering Design and the Role of DFMA,” Proceedings at the Committee for
the Advancement of Competitive Manufacturing, University of Rhode Island, March 24, 1999.

3] Platt, M., and Marscher, W.D., “Rapid FEA of Impeller Stress and Vibration in the Agile Engineering Process,”
ASME Rotating Machinery Conference and Exposition, Somerset, NJ, November 10-12, 1993.

4] Gauthier, B., Dewhurst, P., Japikse, D. “Application of Design for Manufacture and Assembly Methodologies
to Complex Aerospace Products,” to be presented at the 36th AIAA/ASME/SAE/ASEE Joint Propulsion
Conference and Exhibit, July 16-19, 2000.

5] Japikse, D., “Precision Turbomachinery Products in the Agile Engineering Era,” Keynote address presented at
The Pacific Center of Thermal-Fluids Engineering, Maui, Hawaii, March 26-30, 2000.

11