Sie sind auf Seite 1von 6

th

19 International Conference on Production Research

COMPUTER AIDED GEOMETRIC MODELING AND 5-AXIS MILLING OF A SCREW


PROPELLER IN A SINGLE SETUP: A CASE STUDY
1
2
3
4
5
A.C. Munar , E.L.J. Bohez , M. Singh , T. Lin , S.S. Makhanov
1,2,3
Industrial Systems Engineering Department, Asian Institute of Technology (AIT), Pathumthani, Thailand
4
Industrial Engineering Department, Konkuk University, Seoul, Korea
5
Sirindhorn International Institute of Technology (SIIT), Thammasat University, Pathumthani, Thailand

Abstract
This paper presents a geometric modeling method and 5-axis CNC machining algorithm for the manufacture
of screw propellers in one single setup. A novel approach for 5-axis roughing is developed and implemented
in addition to further streamline finishing sequences. The toolpath is generated by dividing the propeller
model into milling regions viz. the front & rear blade faces, the leading & trailing blade edges and the lateral
hub surfaces between adjacent blades. Cutting tools for each region are then selected along with the
appropriate tool orientation for 5-axis flank milling. The CL-data is acquired using UnigraphicsTM and
translated into NC code using a postprocessor for the Maho MH600E milling machine. The viability of the
proposed method is verified by virtual machining on VericutTM and actual machining on Maho MH600E.
Keywords:
5-Axis CNC machining, CAD/CAM, sculptured surface machining.

1 INTRODUCTION
Sculptured surface machining has significantly developed
ever since its inception in the 1950s under the historic
project called Automatic Programmed Tool Language
(APT). The term sculptured has earned popularity in
machining as NC programmers have gained more control
of the cutting tool thus resembling the movement of an
artists chisel. Machining of free-form surfaces called for
advanced CNC multi-axis machines which have a higher
degree of flexibility and precision than conventional 3-axis
types. Its implementation also demanded even more
sophisticated CAD/CAM systems to ease designer work in
modeling and programming. CAM technology has assisted
designers in selecting cutting parameters in addition to
preparing NC data based on the required design surface
tolerance. The selection of cutting variables involves
specifying cutting tools that are geometrically compatible
with the design surface as well as choosing the appropriate
milling technique. Countless research has been devoted to
harness the full potential of multi-axis NC machining in
both hardware and software aspects [1].
Manufacturing parts with complex geometry requires
flexible methods of CNC programming and machining
especially when the design part covers an area of several
meters such as gas turbine blades and marine propellers.
Among the numerous advantages of 5-axis machining, the
three most significant are: reduced process time due to
higher material removal rates, reduced setup time for
intricate prismatic parts and improved surface quality thus
minimizing the time required for subsequent finishing [2,3].
The inherent ability of 5-axis machines to position the tool
and workpiece at any given relative point and angle allows
them to produce the design part using several approaches
[2,4]] that which is evidently a shortcoming of 3-axis
machines. In contrast to their predecessors, 5-axis
machines have considerable advantage in terms of
accessibility and productivity. For example, the effect of
employing 5-axis machines in the manufacture of die molds
has resulted in 10-20 times more than the efficiency set by
3-axis machines [5,6]. Moreover, parts with irregular
shapes such as turbo impellers can be machined using a
single setup since areas previously inaccessible to 3-axis
machines are made workable with added degress of
freedom although under certain constraints [7,8].

Apart from their benefits in sculptured surface machining,


5-axis machines have also introduced both computational
and functional difficulties. First, current CAM systems still
do not provide adequate support for toolpath generation
and verification such that designers still rely on iterative
methods [8,9,10,11]. Apparently there are still huge
numbers of research concerning the effective control of
scallop heights based on tool geometry and positioning.
Second, considering the rigorous task of developing
complicated algorithms for interference and collision
detection in addition to position correction [12], 5-axis is
prone to machining errors of which many are classified as
NC programming related [5]. 5-axis operations can be
categorized as either point milling or flank milling [8]. In
conventional point milling, material is removed using the tip
of the tool. Although the process can be applied to
machine any complex surface, the main drawback in using
point milling is that is it time-consuming and the milled
surface would require polishing in order to remove scallops
[13]. The process of flank milling on the other hand
removes material using the side of the tool, which then
leads to higher machining efficiency and to a great extent
eliminates the presence of surface scallops [14]. Yet it has
disadvantages involving large overcuts and undercuts with
increased chances of cutter interference and collision.
Flank milling can be further classified as either ruled
milling or skive cut [5]. Ruled milling refers to the
machining of flat ruled surfaces or the more convoluted
hyperbolic paraboloid surfaces both of which are bound by
two guide strings. Common applications of ruled-milling
include the manufacture of fan, compressor and impeller
blade surfaces. The major drawback of ruled milling
includes relatively large deflections when slender tools are
employed as well as gouging for the case of concave or
sharp cornered features. While the cut also mills with the
side of the cutter, it is preferred for convex surfaces such
as the leading and trailing edges of airfoils found in gas
turbine blades. Screw propellers have been the primary
products of 5-axis CNC machining since the beginning.
With their visibly complex geometry, the manufacture of
propellers presented NC programmers the difficulty of
guiding the tool through narrow areas between adjacent
blade surfaces without causing gouging or interference.
Research on 5-axis machining of propellers however have

mostly focused on the semi-finishing and finishing


sequences [15,16]. Roughing is still widely performed on
3-axis machines for two reasons mainly, cost-effectiveness
as well as high material removal rate [6,8]. The motivation
of this research is to deviate from such common practice
where roughing would be performed straight-away in 5-axis
mode thus further complimenting the procedure with a
significant reduction in setup time and overall machining
time.
2 GEOMETRIC MODEL OF SCREW PROPELLER
The geometry of a propeller is generally derived from the
following parameters: chord length, pitch, camber, skew,
rake and the profile thickness [17]. From such dimensional
and other non-dimensional parameters, the efficiency and
aero-hydrodynamic performance of the propeller is
estimated depending on its specific application. Since the
main focus of the work on propeller modeling is not
intended to support hydrodynamic testing such as in a
cavitation tunnel therefore greater emphasis has been laid
on geometric modeling and subsequent 5-axis machining.
For this reason a more simplified approach is adopted to
generate the 3D propeller model with UnigraphicsTM. The
suggested method would first take into account the airfoil
coordinates, overall diameter, mean pitch and pitch ratio.
Thereafter the specific pitch angle and profile thickness
distributions of the selected propeller class can be applied
by forming the airfoils at each local propeller radii using a
series of affine transforms [18,19]. The surface for the
blade was then created employing the airfoils as section
strings. To complete the procedure, the blend surface was
generated about the root section to make a fillet. After the
construction of a single blade, it was duplicated into 4
copies, which was then rotated about the hub centerline at
72 interval to make a total of 5 blades. The completed 3D
model of the screw propeller along with the resulting
mechanical drawing is depicted in Figure 1.The proposed
modeling approach is summarized in Figure 2.

(a)

(b)
Figure 1: Propeller mechanical drawing and CAD model

Figure 2: Modeling of screw propeller blade.

th

19 International Conference on Production Research

5-AXIS MACHINING OF SCREW PROPELLER

3.1 Process Planning and Setup


A four pronged process plan was followed to mill the screw
propeller [4]. The strategy started off with the geometric
identification of the part surfaces which classified them into
either convex, concave or saddle. Following is the grouping
of the identified surfaces into milling regions depending on
their curvature properties. Then the maximum allowable
tool diameter was determined for each region after which
the milling direction is selected. Consequently the drive
surface to machine the collective milling regions are
created. In this case the minimum distance between two
adjacent blades determined the maximum tool diameter.
The flat end mill was used due to its wider range of
effective cutting radius in contrast to ball nose cutters. The
setup of the blank part on the MH600E along with algorithm
for collision avoidance is shown in Figure 3.

Figure 3: Gouging avoidance for 5-axis milling of propeller


3.2 Toolpath Planning and Generation
Prior to the toolpath generation for the roughing phase, a
preform geometry of the modeled propeller was first
constructed. The geometry coined the bounding boxes is
a crude approximation of the propeller which is composed
of ruled surfaces that envelop each blade to form a
polygon. The concept of using such preform geometry is
similar to that of a cast propeller at the near net shape
stage which would have to undergo finishing so that the
assigned tolerance is achieved [20]. Since the preform
geometry is relatively less complex than the final part
geometry, a more straightforward machining strategy for
the roughing phase can be employed as a result. The
blade polygon as shown in Figure 4 was divided into three
milling regions namely front face, leading side and back
face. With these regions, the drive surfaces were assigned
accordingly. The roughing toolpath for each milling region
of the bounding boxes geometry is depicted in Figure 5.

(a)

(b)

Figure 4: Bounding boxes perform geometry

Figure 5: Bounding box roughing toolpath generation


The number of tool pass for each segment of the bounding
box is 30. Once generated for a single bounding box, each
toolpath was duplicated for the remaining four consecutive
boxes. Hence a total of 15 toolpaths were generated to
form three independent CL files namely front, leading and
back which would then lead to that illustrated in Figure 6.

(a)

(b)

Figure 6: Bounding boxes roughing toolpaths


For the semi-finishing and finishing phases, toolpaths were
generated to machine the blade contour and the blade
faces. The contour is divided into two parts: trailing edge
and leading edge. Since the selected method of machining
is ruled-milling, a ruled surface is constructed along the
contour, which was used as the drive surface. However,
ruled-milling is accessible starting from any segment of the
trailing edge up to only the upper half of the leading edge.
Due to the problem of overlapping areas between adjacent
blades, ruled milling would result into gouging if used
further down the leading edge. To address the problem,
skive-cut combined with a long tapered tool is thus chosen
to machine the difficult segment. The toolpaths used to mill

the contour of a single blade are summarized in Figure 7.


The total number of tool passes for the trailing and leading
edges are 5 and 20 respectively.

(a)
(b)
Figure 9: Blade and tool collisions at 5 and 50 tilt angles
Similar trials to determine the permissible range of the tilt
angle was conducted for the blade back face. Presumably
the tilt angle would range from 0 to 50 for an interference
free toolpath based on the trials done for the front face.
However it was ascertained that the primary constraint in
this case is the maximum travel limit of the swivel axis and
not the level of gouging avoidance. At a tilt angle of 2 the
postprocessed CL-data contained some lines where the Baxis rotation angle is greater than the machine maximum
105. As the tilt angle is increased so does the instances of
lines with invalid values for the B-angle. If on the other
hand ruled-milling is used that is with 0 swarf tilt angle,
gouging would result when the cutter approaches the root
section. Since the major constraint is given by the machine
swivel axis, it was decided that a 2 tilt angle such that the
G-code would be subsequently corrected by hand.
The toolpaths employed to mill the front and the back faces
of the blade with no interference is shown in Figure 10.
Figure 7: Blade contouring toolpath generation
The duplication of the generated toolpaths for a single
blade would lead to that illustrated in Figure 8 below.

(a)

(b)

Figure 8: Combined contouring toolpaths.


The procedure for semi-finishing and finishing is reserved
for the front and back faces of the blades. Likewise to the
roughing phase, flank milling with a tilted tool is used for
the reason that it facilitates control of the tool axis and
avoids probable instances of tool collision with adjacent
surfaces. In fact the operation somewhat emulates point
milling where the difference mainly lies in the formation of
the cusps which are in effect triangular rather than elliptical
since the tool used is a flat end mill. Different tilt angles
were tested for each blade face depending on the level of
gouging avoidance and the maximum travel limit of the
swivel axis.
The front face which has a concave surface was suggested
to have a tilt angle sufficient to avoid tool interference with
the blade tip and with the surface of the consecutive blade
while at the same time ensuring no rear-cutter surface
gouging. Conversely, the back face can have a smaller tilt
angle given that it is a convex surface.
Simulation have shown that the allowable tilt angle for the
front face range within 5 to 50 relative to the surface
normal vectors. It is found that a tilt angle less than or
equal to 5 would eventually result into a collision between
the tool and the blade tip as the tool approaches the root
section. Furthermore the tool would interfere with the back
face of the adjacent blade if the assigned tilt angle were
greater than or equal to 50. These two cases are depicted
in Figure 9.

Figure 10: Blade faces finishing toolpath generation


The generated toolpaths for the front and back blade faces
were consequently instanced about the propeller centerline
at 72 rotation angle resulting to that shown in Figure 11.
Note that the intol and outol for finishing is 0.01 mm.

(a)

(b)

Figure 11: Combined front and back faces toolpaths

th

19 International Conference on Production Research

The next and final toolpath was generated to mill residual


material between blades about the lateral hub surface as
depicted in Figure 12 below. An auxiliary drive geometry
representing the area located between consecutive blades
was developed at the onset. The tapered ball-nosed end
mill was again used given its long reach and small milling
diameter hence allowing for material in the narrow area to
be removed without causing gouging or interference.

Table 1: Bounding boxes roughing performance


Milling Region
Front Faces
Leading Sides
Back Faces
Total

Material Removed
3

531909 mm
38357 mm3
502813 mm3
1073079 mm3

Cutting Time
87.81 min
11.29 min
84.86 min
184.48 min

Figure 14: Contouring verification


Figure 12: Hub lateral face finishing toolpath
3.3 Toolpath Verification and Actual Machining
TM
After postprocessing the CL data from Unigraphics , the
TM
acquired G-codes were imported into Vericut for initial
verification prior to actual machining. The parameters used
in the simulation are given as follows: XYZ traverse speed
= 6m/min; A-axis angular speed = 235/sec; B-axis angular
speed = 162/sec; and Feedrate = 300 mm/min. The initial
5
3
part stock volume is 14.777x10 mm . The outcome of both
virtual and physical verification of each toolpath is shown
in Figure 13 to Figure 16. The total time incurred and net
volume of material removed for each machining phase is
summarized in Table 1 to Table 4.

Table 2: Contouring performance


Milling Region
Trailing Edges
Leading Edges
Total

Material Removed
3

65126 mm
68005 mm3
133131 mm3

Cutting Time
22.56 min
145.62 min
168.18 min

Figure 15: Propeller blades semi-finishing verification


Table 3: Propeller blades semi-finishing performance
Milling Region

Figure 13: Bounding boxes roughing verification

Concave Faces
Convex Faces
Total

Material Removed
3

204055 mm
79188 mm3
283243 mm3

Cutting Time
706.55 min
212.65 min
919.20 min

[5]

[6]

[7]
Figure 16: Lateral hub finishing verification
Table 4: Propeller blades and hub finishing performance
[8]
Milling Region

Material Removed

Cutting Time

Concave Faces
Convex Faces
Hub Faces
Total

14250 mm3
37060 mm3
22766 mm3
74076 mm3

1213.4 min
212.65 min
199.17 min
1625.2 min

Results from the simulation show that the entire roughing


phase took a minimum total of 3 hours where an estimated
10.7x105mm3 of material was removed or 70% of the initial
part stock volume. Altogether it took 48 hours to transform
a cylindrical blank part into a propeller with 5 blades. A total
of 15.64x105 mm3 of material was milled during the
roughing and finishing sequences.
4 CONCLUSION
The proposed modeling and 5-axis machining approach for
marine propellers has highlighted the advantage of having
the roughing phase performed in 5-axis rather than in the
conventional 3-axis mode. A huge reduction in setup and
machining lead time was achieved by having a more
unified approach such that roughing and finishing were
conducted in a single setup. Moreover a higher material
removal rate was achieved for the roughing phase by using
a preform bounding boxes geometry combined with 5-axis
flank milling. Further research in the presented method
would include the optimization of toolpath parameters in
order to address surface quality and tool wear rate.
5 ACKNOWLEDGMENTS
This research work was funded by AIT and SIIT. The
authors are greatly appreciative to AIT CIM Laboratory
supervisor Somchai Taopanich for his valuable expertise in
machining and also to Suradash Chungpaiboonpatana for
his assistance in editing this paper.
6 REFERENCES
[1] Dragomatz, D., Mann, S., A Classified Bibliography of
Literature on NC Toolpath Generation, Computer
Aided Design, 29(3), 1997, 239-247.
[2] Bohez, E.L.J., Five-axis Milling Machine Tool
Kinematic Chain Design and Analysis, International
Journal of Machine Tools & Manufacture, 42, 2002,
505-520.
[3] Lai, X.D., Zhou, Y.F., Zhou, J., Peng, F.Y., Yan, S.J.,
Geometrical Error Analysis and Control for 5-Axis
Machining of Large Sculpture Surfaces, International
Journal of Advanced Manufacturing Technology, 21,
2003, 110-118.
[4] Lauwers, B., Kruth, J.P., Dejonghe, P., An Operation
Planning System for Multi-Axis Milling of Sculptured
Surfaces, International Journal of Advanced
Manufacturing Technology, 17, 2001, 799-804.

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20]

Choi, B.K., Jerard, R.B., Sculptured Surface


Machining Theory and Applications, Kluwer Academic
Publishers, Dordrecht, 1998.
Gray, P., Bedi, S., Ismaili, F., Rao, N., Morphy, G.,
Comparison of 5-Axis and 3-Axis Finish Machining of
Hydro Forming Die Inserts, International Journal of
Advanced Manufacturing Technology, 17, 2001, 562569.
Chen, S.L., Wang, W.T., Computer Aided
Manufacturing
Technologies
for
Centrifugal
Compressor Impellers, Journal of Materials
Processing Technology, 115, 2004, 284-293.
Yilmaz, O., Noble, D., Gindy, N.N.Z., Gao, J., A Study
of Turbomachinery Components Machining and
Repairing Methodologies, International Journal of
Aircraft Engineering and Aerospace Technology,
77(6), 2005, 455-466.
Bliko, I., Kowerich, S. and Paulik, P., Experiences
from a Quantum Leap Improvement in Turbine
Manufacturing. In Machining Impossible Shapes, ed.
Gustav J. Olling, Byoung K. Choi and Robert B.
Jerard, Kluwer Academic Publishers, Boston, 1999,
8-23.
Bohez, E.L.J., Senadhera, S.D.R, Pole, K., Duflou,
J.R., Tar, T., A Geometric Modeling and Five-Axis
Machining Algorithm for Centrifugal Impellers, Journal
of Manufacturing Systems 16(6), 1997, 422-436.
Lai, X.D., Zhou, Y.F., Zhou, J., Peng, F.Y., Yan, S.J.,
Geometrical Error Analysis and Control for 5-Axis
Machining of Large Sculpture Surfaces, International
Journal of Advanced Manufacturing Technology, 21,
2003, 110-118.
Bohez, E.L.,J., Minh, N.T.H., Kiatsrithanakorn, B.,
Natasukon, P., Ruei-Yun, H., Son, L.T., The Stencil
Buffer Sweep Plane Algorithm for 5-Axis CNC Tool
Path Verification, Computer Aided Design, 35(12),
2003, 1129-1142.
Gong, H., Cao, L.X., and Liu, J. 2005. Improved
Positioning of Cylindrical Cutter for Flank Milling
Ruled Surfaces. Computer-Aided Design 37:12051213.
Bedi, S., Mann, S., Menzel, C., Flank Milling with Flat
End Cutters, Computer Aided Design, 35, 2003, 293300.
Elber, G., Fish, R., 1994, 5-Axis Freeform Surface
Milling using Piecewise Ruled Surface Approximation,
Technical Paper, Israel Institute of Technology, Haifa.
Youn, J.W., Jun, Y., Park, S., Interference-free
Toolpath Generation in Five-Axis Machining of a
Marine Propeller, International Journal of Production
Research, 41(18), 2003, 4383-4402.
Van Beek, T., Technology Guidelines for Efficient
Design and Operation of Ship Propulsors, Marine
News, Wrtsill Propulsion, Netherlands BV, 2004.
Cortes, J.L., 2004, Do-It-Yourself Project, Designing
and drawing a classical wood propeller (in
Solidworks), http://www.nmine.com/propeller.htm.
Kouh, J.S., Han, C.H., Chen, Y.J., 2004, A New
Geometric Modeling Method for Marine Propellers, 9th
Symposium on Practical Design of Ships and Other
Floating Structures.
Gao, J., Chen, X., Zhang, D., Yilmaz, O., Gindy, N.,
Adaptive Restoration of Complex Geometry Parts
Through Reverse Engineering Application, Advances
in Engineering Software, 2006.

Das könnte Ihnen auch gefallen