Profiled Superabrasive Grinding Wheels for the Machining of a Nickel Based Superalloy

D.K. Aspinwall1 (2), S.L. Soo1, D.T. Curtis1, A.L. Mantle2 (3)
1
Machining Research Group, Dept. of Mechanical & Manufacturing Engineering, University of Birmingham,
Birmingham, UK
2
Manufacturing Technology, Rolls-Royce plc, Derby, UK

Abstract
Machining data are presented for small diameter, profiled (fir tree root form), single layer/electroplated CBN
(B46, B76 and 91) and diamond (D46) grinding wheels, when cutting Udimet 720. Spindles operating at
60,000 and 90,000rpm were employed, with a synthetic polyalphaolefine (PAO) oil based fluid in a down
grinding mode on single sided specimens. Operating parameters were selected to reflect finishing conditions.
Measured tool wear was lower for CBN grit as compared to diamond however workpiece roughness was
lower with Ra approaching 0.75μm when using D46 wheels. Higher rotational speed produced lower grinding
wheel wear. No workpiece burning was observed irrespective of grit type at the conditions tested.

Keywords:
Grinding, Nickel, Cubic boron nitride (CBN)

1 INTRODUCTION workpiece scoring / smearing etc. Furthermore, the use of

Despite the extensive research and development
associated with blisk (bladed disk) manufacture and
repair over the last two decades, the great majority of
aeroengines, whether for commercial or military use,
employ detachable rotating blades. These are typically
located and held in position on a series of disks using
either a fir tree or dovetail blade root design. The rotor
assembly for the engine compressor and turbine, will,
depending on size, contain many hundreds of blades of
varying size and shape with similar variation in the scale
of the root attachment. Figure 1 shows sample blade root
mounting slots, the fir tree design being used where blade
loading is high [1]. Figure 2 shows a selection of blade
root configurations.
10 mm 10 mm 10 mm
Figure 1: Sample root mounting slot profiles.
10 mm 10 mm 10 mm
Figure 2: Sample blade root configurations.
The machining of axial / oblique root mounting slots in
individual disks is almost universally accomplished using
broaching, in some cases following a preliminary milling
operation. The process is mature and consequently there
have been few recent publications outlining equipment,
technology or machinability, although commercial
considerations are an additional factor. Broaching is able
to provide complex slots of appropriate geometrical size
(within ~ 10μm), quality (workpiece Ra ~ ≤ 2μm) and
integrity, with roughing and finishing possible within one
stroke. Broach design, accuracy and finish are however
critical factors in maintaining acceptable performance and
minimising tooth breakage, chip jamming and associated
different tooth / tool geometries at different positions on
the broach produces variations in chip thickness and
hence differences in wear rate, thus providing scope for
process monitoring [2,3]. Major shortcomings of the
process include the high capital cost and large size of
machine tools together with their inflexibility, costly
tooling, lengthy setup, validation and changeover times,
high cutting forces (up to 10,000N) [2] and relatively long
manufacturing times, with broaching speeds as low as
2m/min quoted in heat resistant alloys.
Feasible options for alternative processes to broaching for
disk slot manufacture are limited, although significant
changes have occurred in recent years in the way the
equivalent root form is machined on the compressor or
turbine blade. Here, broaching has largely been
supplanted by creep feed grinding (CFG), involving the
use of multi-axis, flexible machining centres equipped with
porous conventional abrasive wheels, semicontinuous
dressing and a high pressure / high flow rate fluid delivery
system, aptly termed VIPER (Very Impressive
Performance Extreme Removal) [4]. Solid carbide profiled
milling cutters employing either a full or partial form, are
also used in blade root manufacture and similar or
indexible insert tooling is used in some machining of disk
slots.
The re-entrant nature of disk slots appears to preclude
the use of conventional grinding wheel arrangements for
finishing the form. However, recently published details
involving VIPER machining techniques [5], suggest that a
modified arrangement is feasible at least for larger fir tree
roots, although details are scant. A more versatile
grinding arrangement with scope for machining a wide
range of slot forms and sizes, involves the use of small
diameter (≤ Ø25mm) single layer, electroplated
superabrasive grinding wheels (sometimes referred to as
grinding pins or points) employing either a full or partial
form, see Figure 3. By necessity, rotational speeds are
high, typically > 50k rpm in order to achieve even
moderate peripheral cutting speeds, which will vary
according to the shape of the blade fixing slot. High speed
spindles of adequate power and stiffness are therefore
required as is the machining of a preliminary or initial slot
using either CFG / high efficiency deep grinding (HEDG)

Annals of the CIRP Vol. 56/1/2007 -335- doi:10.1016/j.cirp.2007.05.077

or electrical discharge wire machining (EDWM). With the
former, the slot cross section may typically resemble a
stepped pyramid or quadralateral with the widest feature
at the periphery of the disk. In the case of EDWM, a
shape more closely mirroring that of the root form,
provides the possibility to obtain a more uniform wear
pattern on the grinding wheel. A flow chart showing
alternative manufacturing sequences is given in [6].
Superabrasive
grits
Profiled
grinding wheels
Plain grinding
wheels
10 mm
Figure 3: Electroplated superabrasive grinding wheels.
Root slot grinding reported in [6-9] essentially
demonstrated the feasibility of the approach, using plain
cylindrical diamond and CBN wheels and subsequently
full form wheels, when machining two key nickel based
superalloys; Inconel 718 and Udimet 720. Taguchi
fractional factorial experiments were used to identify key
operating factors and levels, with testing undertaken
predominantly at 60k rpm, giving grinding speeds up to
45m/s. International interest in the approach may be
gauged from the recent spate of patents [10, 11] dealing
with preliminary / pre-cursor slot production, full form slot
and root machining. Such documentation provides
methodology and boundary parameters but not their
effect on performance. The following experimental work,
which utilises a commercial fir tree root design (RR
Trent), was undertaken to explore the effect of
superabrasive type (diamond and CBN), grit size and
cutting parameters on grinding wheel wear, workpiece
surface roughness and cutting forces.
2 EXPERIMENTAL WORK
2.1 Equipment, test procedures and operating
parameters
Laboratory testing was undertaken on two high speed
machining (HSM) centres. The first was a 20k rpm unit
equipped with two separate retrofit high speed spindles,
one rated at 3kW operating at up to 60k rpm and another
rated at 4.5kW able to operate at up to 90k rpm, together
with ancillary cooling, lubrication and control equipment.
Mounting of the spindles necessitated the manufacture of
two manifold units due to the different arrangement of
supply connections. The second HSM centre was a linear
motor machine which utilised its’ own 60k rpm spindle
(3kW) and was equipped with laser tool setting.
Modifications were made to both machines to incorporate
high pressure, high flow rate fluid supply, mist extraction
and CO2 fire suppression systems for use when
employing a mineral oil cutting fluid. This was a synthetic
PAO supplied via a twin filter 83 bar, 140 l/min pumping
system. Twin laminar flow nozzles were used with
pressure monitoring behind the nozzle outlet. Although
not detailed here, comparative CBN grinding tests using a
water based cutting fluid and the PAO mineral oil,
suggested a significantly improved life with the latter, in
accordance with the consensus view and published work
by Brinksmeier et. al. [12] and Gift et. al. [13]. Figure 4
shows the general arrangement on the first high speed
machine when using the 90k rpm spindle while Figure 5
shows the table arrangement on the second linear motor
machine.
-336-
90k spindle
in manifold
Vibration
Linear flow damping system
nozzle
Figure 4: 90k rpm retrofit spindle and experimental setup.
Workpiece blocks typically 15mm and 36mm thick were
mounted in purpose made clamping units fitted to a
vibration damping palletisation (VDP) system. Force
measurement involved a piezoelectric 3-component force
dynamometer, associated charge amplifiers and analysis
software on a desktop computer.
Profiled
grinding wheel
Figure 5: Experimental setup on 60k rpm linear motor
unit.
Tool inspection was principally undertaken using a
shadowgraph set to 10x magnification which was fitted
with an X-Y digital readout. Tool wear was monitored
using graphite replicas (3mm thick), which were cut at
predetermined times throughout the tests and measured
on the shadowgraph (diametric wear results detailed
which were averaged from measurements between all fir
tree turning points / lobes). Workpiece surface roughness
and grinding wheel roughness were assessed using
stylus profilometry equipped with 3D analysis software.
When measuring grinding wheel roughness (lobe side), a
synthetic rubber replicating compound was employed to
produce a negative replica after which a fast curing
polymerising resin was used to produce positive, hard
replicas suitable for stylus measurement. Cut off was
0.8mm. A digital camera arrangement connected to a
laptop computer was used for recording low magnification
micrographs, while a scanning electron microscope
(SEM) was used for analysis of grinding wheel /
superabrasive grit wear.
Figure 6 details the single sided slot configuration used
throughout the present work, a machined graphite replica
and a sample B91 (preconditioned) grinding wheel.
Testing involved diamond; D46 and CBN; B46, B76 and
B91 grit sizes respectively.
(a) (b) (c)
lobe
Figure 6: (a) Single sided slot, (b) machined graphite
replica, (c) sample B91 grinding wheel.
All tests employed down grinding, a feed rate of 2m/min
and 20μm depth of cut per pass. Peripheral cutting speed
varied according to the spindle rotational velocity and
shape of the fir tree root form, but ranged up to ~ 90m/s.
Both B91 and B76 grinding wheels were subject to a
preconditioning or pre wearing regime by the wheel
manufacturers which was intended to truncate the wear
response and reduce workpiece roughness. Typically,
fluid pressure at the nozzle was ~ 3bar with a combined
flow rate of ~ 135l/m.
2.2 Results and discussion
Differences in diamond and CBN abrasive wear and
associated workpiece roughness are shown in Figure 7
for D46 and B46 products when operating at 60k rpm on
the second HSM centre. Statistical analysis (two sample
t-Test) suggests there is a significant difference in means
Data when operating with the 90k rpm spindle for B46,
B76 and B91 grinding wheels are shown in Figures 9 and
10 which details wheel wear and workpiece Ra
respectively. The effect of ‘preconditioning’ for the B76
and B91 wheels is clearly visible when compared with the
unconditioned B46 wheel, which displays an initial step
change in wear levels. Additionally, the workpiece Ra for
the B76 and B91 display a far shallower response than for
the B46. The disposition of the Ra plots is as expected
with finer grit producing a lower surface roughness. Trials
3
with B76 and B91 were discontinued at ~3500mm as the
required workpiece surface roughness was not achieved.
0.05

Wheel wear (mm)

at the 5% level, with the CBN grit providing a lower wear
rate. The results for workpiece surface roughness B46 B76 B91
(measured at a point corresponding with the lobe side)
between the different grit types provided similar trend
results, with diamond (D46) producing the lowest
roughness, despite higher Sa values for the diamond
grinding wheel at the cessation of the tests (start and
0.00
finish Sa results for D46=4.85 / 3.58μm and B46=4.95 /
2.78μm). In view of the fact that the operation was 0 3000 6000
essentially plunge grinding with no vertical oscillation, it is Material removal (mm 3)
likely that the result relates in part to replication Figure 9: Effect of grit size and preconditioning on
inaccuracies (synthetic rubber compound accurate to grinding wheel wear at 90k rpm.
0.1μm, for the hard polymerising resin, 1μm is given).
B46 B76 B91
D46 wear B46 wear roughness, Ra (μm)
Workpiece surface
D46 surface roughness B46 surface roughness
0.10 3.0 2.0
roughness, Ra (μm)
Wheel wear (mm)

Workpiece surface

2.0
0.05
1.0
1.0
0 3000 6000
0.00 0.0 Material removal (mm 3)
0 4000 8000
Material removal (mm 3) Figure 10: Effect of grit size and preconditioning on
workpiece Ra at 90k rpm.
Figure 7: Wheel wear and workpiece Ra at 60k rpm. Figure 11 shows the uniform distribution of abrasive grit
Figure 8 gives 3D topographic maps (curvature on a worn (~ 3500mm3 material removed) B91 grinding
compensated) taken from the hard replicas of a B46 wheel at the indicated position, with details of the same
grinding wheel when new and worn (~7500mm3 material area at higher magnification when new and worn, no
removed). Although mean grit size and concentration special cleaning was undertaken following use. No
were approximately the same, the morphology of the obvious loading or grit blunting is evident although there
diamond and CBN abrasives was different, with the appears to be some grain pullout as indicated.
former having a more ‘blocky’ shape. No significant
loading of the wheels was observed, neither was there SEM view
any obvious discolouration / burning of the workpiece Worn B91
surface.

48.29μm

500μm
5mm
2mm (a)
New B91 Worn B91

30.62μm

5mm 100μm 100μm

2mm (b)
Figure 8: 3D topographic maps of (a) new B46 and (b) Figure 11: SEM micrographs of grit distribution and wear.
worn B46. Maximum normal cutting force when operating at 60k rpm
on the first HSM machine using B46 wheels was ~ 45N.

-337-
Total spindle power levels recorded on the second
machine at 60k rpm for both B46 and D46 wheels were
1.03kW (stable) and 1.03 to 1.59kW (over the duration of
the test) respectively. The actual grinding power
component within these figures was calculated to be 94W
and 94 to 656W respectively, with corresponding specific
grinding energies of 6.54J/mm3 for the CBN and up to
45.79J/mm3 for the diamond wheel. The very low values
of force and power reflect the nature of the operation
(finishing).
A comparison of wear performance for B46 grinding
wheels when operating at 60k and 90k rpm is illustrated in
Figure 12. Lower wear was observed at 90k rpm as a
result of the smaller undeformed chip thickness and the
anticipated effects on forces and power. There are
however practical considerations relating to the use of
such high speeds in any potential industrial application
(spindle life, reliability, tool fabrication / balancing etc.).
Wheel wear (mm)
0.05
B46 (90,000rpm) B46 (60,000rpm)
0.00
0 4000
Material removal (mm 3)
Figure 12: Effect of cutting speed on wheel wear.
Figure 13 details entry burr formation in the lobe region of
the workpiece when operating at 90k rpm for worn (~
3500mm3 material removed) B46 and B91 wheels. The
preconditioning of the B91 is considered to be the main
factor controlling burr size, which was up to 4mm long.
Similar burrs were observed with the preconditioned B76
wheels.
B46 B91
Figure 13: Entry burrs at 90k rpm for B46 and B91.
The testing raised a number of concerns relating to the
suitability of machine tool configurations, fluid supply etc.
The different spindle arrangements necessitated different
tool adaptors which produced unavoidable variations in
tool overhang and runout, the latter ranging from 2μm to
10μm. Such aspects will impact directly on slot accuracy,
wheel wear and workpiece roughness.
3 CONCLUSIONS
Lower absolute wear values were recorded for CBN
rather than diamond abrasive, however it was not
possible to assess the influence of grit morphology. Wear
rate following preconditioning was significantly reduced
however the ‘pretreatment’ had an effect on burr
formation which was evident to a greater extent. Lower
workpiece surface roughness was obtained using
diamond superabrasive, most likely as a result of the
‘blocky’ morphology of the diamond grit and possibly the
higher wear rate. The use of higher rotational speeds and
hence cutting speed produced lower grinding wheel wear.
Under the conditions tested, no obvious wheel loading
was observed, however there was wholesale grit pullout.
No workpiece burning or surface discolouration was
evident in any of the tests undertaken.
4 ACKNOWLEDGMENTS
We wish to thank Rolls-Royce plc., Saint-Gobain
Abrasives (Chris Davis & Andreas Grimm) and Matsuura
Machinery (David Edwards) for the provision of funding,
tooling and technical support. Additional thanks go to
Faemat & Dick Langley of High Gain Technology for
technical support in relation to high speed spindle
installation and use, together with Peter Lampitt at
System3R for provision of the vibration damped
workpiece palette system. Finally we are indebted to Dr.
John Webster of Cool Grind and staff at Pumps &
Equipment (Warwick) Ltd. for technical support / advice in
relation to fluid supply and in particular laminar flow
nozzle technology.
5 REFERENCES
[1] Huenecke, K., 1997, Jet Engines, Fundamentals of
Theory, Design and Operation, Airlife Publishing
Ltd., Shrewsbury, England.
[2] Budak, E., 2001, Broaching process monitoring,
Proc. of the 3rd Int. Conf. on Metal Cutting and High
Speed Machining, Metz, Vol. 1 oral presentations:
251-260.
[3] Axinte, D., Boud, F., Penny, J., Gindy, N., 2005,
Broaching of Ti-6-4 – Detection of workpiece surface
8000 anomalies on dovetail slots through process
monitoring, Annals of the CIRP, 54/1:87-90.
[4] Hill, C.P.R., Watkins, J.R., Ray, C., Ray, S., 2004,
Method and apparatus for grinding, EU Patent EP
0924028B1.
[5] Venables, M., 2006, It’s a grind, IET Manufacturing
Engineer, 85/4:42-45.
[6] Aspinwall, D.K., Dewes, R.C., Burrows, J.M., Bailey,
M., Soo, S.L., Cook, M., 2003, The use of diamond
and cubic boron nitride grinding points for the
machining of nickel-based superalloys, Proc. of
INTERTECH 2003, Vancouver:1-13.
[7] Burrows, J., Dewes, R.C., Aspinwall, D.K., 2002,
Grinding of Inconel 718 and Udimet 720 using
superabrasive grinding points mounted on a high
speed machining centre, Proc. of 3rd Int. MATADOR
Conf., Manchester:447-452.
[8] Soo, S.L., Ng, E.G., Dewes, R.C., Aspinwall, D.K.,
Burrows, J.M., 2002, Point grinding of nickel-based
superalloys, Industrial Diamond Review, 62/2:109-
116.
[9] Aspinwall, D.K., Dewes, R.C., Burrows, J.M., Paul,
M.A., 2001, Hybrid high speed machining (HSM):
System design and experimental results for Grinding
/ HSM and EDM / HSM, Annals of CIRP, 50/1:145-
148.
[10] Packman, A.P., Vaillette, B.D., 2004, Process for
machining axial blade slots in turbine disks for jet
engines, United States Patent Application
Publication, US 2004/0064944 A1.
[11] Mantel, B.M., 2006, Slot machining, United States
Patent, US 7007382 B2.
[12] Brinksmeier, E., Heinzel, C., Wittmann, M., 1999,
Friction, cooling and lubrication in grinding, Annals
of the CIRP, 48/2:581-598.
[13] Gift Jr., F.C., Misiolek, W.Z., Force, E., 2004, Fluid
performance study for groove grinding a nickel-
based superalloy using electroplated cubic boron
nitride (CBN) grinding wheels, Trans of ASME
Journal of Manufacturing Science and Engineering,
126/3:451-458.
-338-