Professional Documents
Culture Documents
To cite this article: K. Siva Kumar & G. Paulraj (2012) Geometric error control of
workpiece during drilling through optimisation of fixture parameter using a genetic
algorithm, International Journal of Production Research, 50:12, 3450-3469, DOI:
10.1080/00207543.2011.588616
Download by: [Orta Dogu Teknik Universitesi] Date: 15 February 2016, At: 20:58
International Journal of Production Research
Vol. 50, No. 12, 15 June 2012, 3450–3469
Fixtures are an essential component of manufacturing and production. They are used to accurately position a
component or workpiece within a machine-tool coordinate system. During material removal, the function of
Downloaded by [Orta Dogu Teknik Universitesi] at 20:58 15 February 2016
the fixture is to ensure the stability of the workpiece which is constrained to meet the design requirements. The
fixture performs this function by providing accurate and repeatable location of the component with respect to
the machine tool, and by resisting motions, deflections, and distortions of the workpiece under the influence
of the cutting forces. Therefore, the focus of this paper is to develop a methodology that analyses and
optimises the fixture parameter configurations using a genetic algorithm (GA) with an ANSYS parametric
design language (APDL) of a finite element analysis that minimises the geometric dimensional tolerance errors
of the final component dimensions. Here, the workpiece is located and constrained in a fixture, as it undergoes
material removal owing to the insertion of a drill bit. The deformations and distortions of the workpiece are
modelled at discrete intervals throughout the drilling process. Three case studies were used to illustrate the
application of the proposed approach. The first explains the methodology used in this research. The second
and third explain the control of deformation and geometric error of the hole in sequential and simultaneous
drilling operations by optimising the fixture parameters using GA. This work minimises the deformation of
the workpiece using a GA–ANSYS integrated tool. Then, the same layouts given by the integrated tool are
used in the experimental setup, and it is found that the errors found in the perpendicularity and circularity of
the drilled holes are minimal when the above geometric dimensional tolerances were checked using a
coordinate measuring machine.
Keywords: deformation; finite element analysis; geometric error
1. Introduction
This paper presents the fixture layout design for machining applications for drilling operations. During the
machining of a workpiece, it should be appropriately located and constrained. Machining fixtures are designed to
constrain a deformable workpiece that is uncovered using the significant machining loads. The fixture ensures that
the dimensions and shape of the machined component are within the required tolerances. Nixon (1971) reported that
approximately 40% of component rejects are due to dimensioning errors that are attributed to poor fixturing design.
Supports and locators are used to reduce the error caused by elastic deformation of the workpiece. The optimisation
of support, locator, and clamp locations is a critical problem in minimising the geometric error in workpiece
Õ
machining. A GA-based approach is developed to optimise fixture layout by integrating ANSYS running in APDL
code to compute the objective function values for each generation.
2. Literature review
Fixture design algorithms that use rigid-body models for the workpiece have been reported (Shirinzadeh 1971, Liu
and Strong 1993, Brost and Goldberg 1994, Varma and Tasch 1995). Here, rigid-body kinematics under various
geometrical constraints is used for designing fixtures that prevent workpiece motions. However, this approach fails
to account for the elastic deformations of the workpiece, as it is exposed to significant machining loads. In addition
to the rigid-body kinematics approach, more recent studies (Liu and Strong 1993, Brost and Goldberg 1994,
Hershkovitz et al. 1996) have addressed fixture optimisation via the minimisation of an objective function mainly
based on the reaction forces at the locators and supports. The objective of these studies was to select the clamping
locations associated with statically admissible minimal reaction forces. In those studies, it was hypothesised that
minimising the maximum normal reaction force led to fixturing designs which also minimise the workpiece
deformations in the neighbourhood of regions undergoing machining. Brost and Goldberg (1994) formulated
optimal fixturing designs that minimise the reaction forces at the locators. In this formulation, the workpiece is
modelled as a rigid body, and the deformations induced by the clamping and machining forces are not considered.
Zhang et al. (1995) compared the optimal fixturing designs obtained through rigid-body analysis to fixtures
calculated by elastically deformable workpiece models. Their findings demonstrate that the two formulations result
in substantially different optimal fixturing layouts. Lee and Haynes (1987) and Menassa and DeVries (1991)
employed FE models to, respectively, obtain fixtures that minimise the work done by the clamping and machining
forces, and minimise the displacements of selected nodes of the FE mesh. Pong (1994) introduced frictional contact
elements to the FE models and incorporated formulations that minimise the displacements of selected FE nodes as
well. De Meter (1998) applied optimisation techniques to fixturing design based on contact region loading, where
the workpiece is treated as a rigid body. Here, the selected objective function is the summation of the magnitudes of
the loads in the contact regions. Krishnakumar and Melkote (2000), on the other hand, used genetic optimisation
Downloaded by [Orta Dogu Teknik Universitesi] at 20:58 15 February 2016
algorithms to determine the layout of fixtures in which the deformation of the machined surface owing to clamping
and machining forces is minimised over the entire tool path. Liao et al. (1998) optimised fixture layouts by
minimising workpiece deflections owing to clamping forces; nevertheless, the effects of machining forces are not
accounted for. In this model, the deformations of the workpiece and locators are accounted for, and the interface
between the locator and the workpiece is modelled as a surface contact pair. 3-2-l fixturing scenarios are considered,
and FEA is used to solve the deformation field. The deformations of selected nodes constitute the objective function.
In addressing fixturing problems of deformable sheet metal that is neither prismatic nor solid, an N-2-l fixture
design has been proposed by Cai et al. (1996). They report algorithms for finding the best N locating points such
that total deformation of a sheet metal is minimised. These algorithms are based on the N-2-l fixturing scenarios.
Here, FEA models and nonlinear programming are utilised to obtain the optimal fixture layout. Sayeed and De
Meter (1999a) introduced a Mixed Integer Programming (MIP) model for determining the optimal locations of
locator buttons and their opposing clamps for minimising the effect of static workpiece deformations on machined
feature geometric error. This model utilizes a finite element analysis of the workpiece, yet its solution is size-sensitive
to the square of the number of machining response points considered. A technique that makes the MIP model size
strictly proportional to the number of machining response points and alleviates this severe limitation is reported in
Sayeed and De Meter (1999b). Another fixture optimisation algorithm was suggested by De Meter (1998), who
presented fast support layout optimisation (FSLO) models. FSLO determines support locations that minimise the
maximum displacement-to-tolerance ratio of a set of workpiece features subject to machining loads. The
applications of FEM and optimisation techniques have been reported by researchers who minimised workpiece
deformations. Lee and Haynes (1987) first employed the FEM technique to compute the objective function which
was defined as the maximum work done by the clamping and machining forces. Here, the deformation index was the
maximum stress applied on the workpiece. Such studies emphasize the consequence and importance of part
deformation and stress with respect to the necessary number of fixturing elements; however, the methodologies
utilised do not contain a systematic method to optimise the fixture configuration. In the optimum design of fixture
layout outlined by Menassa and DeVries (1991), the optimisation scheme was subdivided into two modules: FEM
and optimisation. In these studies, the objective function was selected as the displacement of surface nodes of the
unmachined workpiece. Pong (1994) utilised the ANSYS software coupled with optimisation algorithms to further
enhance the methodology initiated by Menassa and DeVries (1991) by taking into account the elastic behaviour of
the fixturing components. The objective function used by Pong (1994) also consisted of the displacements of selected
surface nodes. In reality, workpieces are deformable, and therefore the magnitude and directions of the
deformations are a result of the fixture configuration and the action of the cutting and clamping forces. These
deformations clearly affect the dimensions of the machined workpiece and result in part dimensions that deviate
from their nominal values and possibly exceed the allowable tolerances. Moreover, as precision requirements for
parts produced are increased, the effects of workpiece deformations on the machined dimensions are no longer
negligible. Fixtures designed in traditional ways may no longer have the capability of producing high-precision
parts. The considerations of workpiece deformations and the development of mathematical models that predict the
dimensions of the machined surfaces become necessary for fixture layout design for precision machining. The most
realistic and logical approach to minimise the effects of workpiece deformations on workpiece dimensions is to
develop mathematical models which have the capability of predicting the geometry of the machined surfaces upon
3452 K. Siva Kumar and G. Paulraj
completion of the machining process. Wardak et al. (2001a) focused in their paper scientific methods on capturing
the shape and geometry of machined surfaces generated through drilling operations and design of fixtures that
maintain dimensional errors within allowable tolerances. Wardak, et al. (2001b) studied five fixture layouts obtained
under different drilling conditions. The drilling studies include single, sequential, and simultaneous (gang) drilling.
These studies demonstrate the known fact that more frequent setups and longer machining time periods result in
higher workpiece accuracies. The simulations also suggest that the proposed optimal fixturing model developed in
this study can be used to command greater control over the drilling process, resulting in elevated drilling quality.
Kim et al. (2001, 2005) and Kim and Ramulu (2004) reported in their work that the cylindricity (or roundness) of
the drilled holes worsened as the cutting speed and feed rate increased. In this paper, geometric errors of the drilled
hole generated in single and multispindle operations have been studied, and the geometric errors have been reduced
by optimising the fixture parameter using the optimisation techniques of the genetic algorithm (GA).
3. Problem description
There is no direct analytical relationship between the machining error and the fixture layout. The finite element
Downloaded by [Orta Dogu Teknik Universitesi] at 20:58 15 February 2016
method can be used to compute the machining error owing to workpiece elastic deformation, and the GA can be
applied to reduce the error by optimising fixture parameters. Since the GA deals with only the design variables and
objective function value for a particular fixture layout, no gradient or auxiliary information is needed (Nixon 1971).
Locators and clamping positions are more influenced in the workpiece deformation/deflection (Siva Kumar et al.
2010). In this paper, GA is applied for optimisation of locator and clamping positions in 3D deformable plate
fixture layout problems. The workpiece side surfaces are used for locating and clamping. Six locators and two
clamps are to be used here. Three locators are positioned on the primary surface; two locators are positioned in one
of the perpendicular surfaces, and one locator is positioned in the tertiary surface of the fixture system. Two clamps
are used to hold the workpiece against the locators. This paper develops the methodology that analyses and
optimises the fixture configurations using GA with APDL for a finite element analysis that minimise the errors of
the final component dimensions. Finally, comparisons between the geometric tolerances of the holes drilled by
sequential and simultaneous drilling operations are obtained using CMM.
4. Procedure for GA
Figure 1 shows the structure of a GA. In GA a set of initial population is generated randomly. The basic steps
involved in the GA are shown below:
(1) initial population: generate random population of chromosomes;
(2) fitness: evaluate the fitness of each chromosome in the population;
(3) test: if the end condition is satisfied, stop, and return the best solution in current population;
(4) new population: create a new population by repeating the previous steps until the new population is
complete.
Establishing the GA parameters is crucial in an optimisation problem because there are no guidelines (Rajendran
and Vijayarangan 2001). The GA contains several operators, e.g. reproduction, crossover, mutation, etc.
Downloaded by [Orta Dogu Teknik Universitesi] at 20:58 15 February 2016
4.1 Reproduction
The reproduction operator allows individual strings to be copied for possible inclusion in the next generation. After
assessing the fitness value for each string in the initial population, only a few strings with a high fitness value are
considered in the reproduction. There are many different types of reproduction operators, which are proportional
selection, tournament selection, ranking selection, etc. In this study, tournament selection is chosen, since it has a
better convergence and computational time than any other reproduction operator (Deb 1999). In tournament
selection, two individuals are chosen from the population at random. Then, the string which has the best fitness
value is selected. This procedure is continued until the size of the reproduction population is equal to the size of the
population.
4.2 Crossover
Crossover is the next operation in the GA. This operation partially exchanges information between any two selected
individuals. Crossover selects genes from a parent and creates new offspring (children). Like the reproduction
operator, there exist a number of crossover operators in GA. In a single-point crossover operator which is used in
this paper, both strings are cut at an arbitrary place, and the right-side portions of both strings are swapped among
themselves to create two new strings, as illustrated in Figure 2.
In order to carry out the crossover operation, two individuals are selected from the population at random. Then,
a random number in the range of {0, 1} is generated. If this random number is less than the probability of crossover,
these individuals are subjected to crossover, otherwise they are copied to a new population as they are. Also, the
crossover point is selected at random. The probability of crossover (Pc) is selected generally between 0.6 and 0.9.
4.3 Mutation
This is the process of randomly modifying the string with small probability. The mutation operator changes 1–0 and
vice versa with a small probability of mutation (Pm). The need for mutation is to keep diversity in the population
(Deb 1999). This is to prevent convergence of all solutions in population into a local optimum of solved problem.
Figure 3 illustrates the mutation operation at the seventh bit position. In order to determine whether a chromosome
is to be subjected to mutation, a random number in the range of {0, 1} is generated. If this random number is less
than the probability of mutation, the selected chromosome will be mutated. The probability of mutation should be
selected as very low, as a high mutation will destroy fit chromosomes and degenerate the GA into a random walk.
Pm should be selected between 0.02 and 0.06 (Deb 1998).
(5) replace: use new generated population for a further run of the algorithm;
(6) loop: go to step 2.
3454 K. Siva Kumar and G. Paulraj
GENERATE INITIAL
POPULATION
FITNESS VALUES
EVAULATE THE
FITNESS OF EACH
INDIVIDUAL
LOCATORS AND
CLAMP ING POSITION
YES ARE
BEST
INDIVIDUAL OPTIMIZATION
CRITERIA MET? APDL
CONTROL FILE
NO
ANSYS
Downloaded by [Orta Dogu Teknik Universitesi] at 20:58 15 February 2016
SELECTION
RECOMBINATION/ DISPLACEMENT
CROSSOVER
MUTATION FITNESS
5. Methodology
The flow chart in Figure 3 is a summary of the fixture design analysis methodology developed and used in this work.
In summary, a workpiece geometry is created by ANSYS or by any other solid modelling software which is exported
to ANSYS. The workpiece model is meshed and initial boundary conditions are applied. Then the model is loaded
according to the machining conditions and it is solved to optimise the fixture parameter (position of locators and
clamps, and clamping forces) for minimising workpiece deformation. These activities are done in the control file.
Then fitness value is sent to the optimisation tool and optimised values obtained. The optimised layout given by the
integrated tool (GA-FEA) is used in the experimental set-up.
will provide the shape and the dimensions of the hole subsequent to machining. In the deformed state, these points
form a perfect cylinder while coming into contact with the drill bit as the drill bit penetrates the workpiece. It is
assumed that the drill bit is rigid. Assuming that the nodes are displaced from their original locations after the
element deforms from coordinates xi, yi to x0i , y0i , there exists a line containing points p0 and q0 on the drill-bit outer
surface, and these points are displaced from positions p and q. Assuming that the curve containing p and q becomes
a straight line containing p0 and q0 , one needs to determine p and q from the knowledge of the coordinates of p0 and
q0 . The coordinates of p0 and q0 are known quantities with respect to a global reference frame. Figure 5 shows an
eight-noded hexahedral element.
1 Workpiece Al 6061
2 Young’s modulus, E 70000 N/mm2
3 Poison ratio, 0.33
4 Yield strength 270 N/mm2
5 Density 2700 kg/m3
6 Static coefficient of friction 0.375
Downloaded by [Orta Dogu Teknik Universitesi] at 20:58 15 February 2016
5.5.1 Locators
To model a rigid locator with the contact area, multiple nodes are fixed within the contact area. A local
coordinate system (LCS) is created on the workpiece surface at the centre of the locator contact area. All three
International Journal of Production Research 3457
FtiY Fti
Cutting tool perimeter
Node i
FtiX
Y q
r
translational dof of each of the nodes are constrained. This model assumes rigid constraints, but in reality locators
are elastic.
Downloaded by [Orta Dogu Teknik Universitesi] at 20:58 15 February 2016
5.5.2 Clamps
The clamps are used to fully constrain the workpiece once it is located. It is common to use multiple clamps and
clamping forces that are generally constant for each clamp. The clamping force (Fcl) is applied through a clamping
mechanism, which moves the plunger that comes into contact with workpiece. Friction is an important factor in
clamping and in locating. In order to model friction, a comprehensive 3D model of the entire workpiece–fixture
system is required, with contact and target surfaces defined at the fixture–workpiece contact areas. The clamping
forces are modelled in ANSYS as surface pressures on contact areas selected within a circular area on the workpiece
surface for a hydraulic clamp. Both clamps are modelled within a circular contact area.
5.6 Loading
The purpose of this research is not to accurately model the machining process, but to apply the torque and forces
that are transferred through the workpiece in machining and to determine the reactions at the boundary conditions
of the workpiece. The forces in a drilling operation include a torque (T) to generate tool rotation, shear force (V)
created by tool rotation at the cutting edge contact for chip removal, and an axial load (P) owing to feeding. The
forces in drilling are time- and position-dependent and oscillatory owing to cutter rotation, since the cutting edge of
the tool is not in constant contact with the workpiece at a particular location. The cutting force increases
monotonically during tool entry and then approaches a steady state. Fluctuations in the cutting force are due to
cutting-tool tooth contact area distribution during rotation. In this study, the torque and thrust forces in feeding are
applied at steady-state loads, since the initial tool entry is not considered. In previous FEA fixture design research,
loads were applied in a steady state. An initial attempt to model the distributed loading using a number of point
loads applied at key points was unsuccessful, owing to limitations in ANSYS. The model involved placing key
points on a local coordinate system created on the machining surface of the workpiece. The key points were located
at exact R, , and Z positions on the cutting-tool perimeter. At each key point, forces were applied to model a
drilling operation. The torque was modelled with tangential forces placed at the outer radius of the cutting-tool
contact area. The tangential couple forces were decomposed into global X and Y components. The axial load was
modelled by applying forces at each key point in the global Z direction. The reason for the failure of this model is
that the key points created on the workpiece surface are geometric entities and are not part of the finite element
mesh, i.e. key points are not nodes. Owing to this limitation in ANSYS, the point load model was modified to apply
loads at existing nodes on the workpiece surface. Figure 7 shows the modified load model for drilling. Notice that
node i is slightly offset from the cutting tool perimeter. Because a node may not exist in the exact location specified
by R, , and Z, the node closest to that location in the local coordinate system is selected, and forces are applied as
point loads with global X, Y, Z components. The user may minimise the distance between a specified coordinate
location and an existing node by increasing the mesh density. The nodes are selected at equivalent intervals on or
near the cutting tool perimeter. At each selected node, global X and Y components of the tangential couple force
(Fti) and axial load component (Fci) are applied. The applied torque is equal to the sum of the tangential forces
multiplied by the cutting tool radius(r). FtiX and FtiY are the global X and Y components of the tangential
3458 K. Siva Kumar and G. Paulraj
force (Fti). Fci is equal to the total axial load (Fc), divided by the number of nodes over which it is applied. A
simplified model entails the use of a single point force normal to the surface of the workpiece to model the cutting
tool axial load and a couple to model the applied torque. A study was conducted to determine whether multiple
point forces applied along the cutting tool perimeter are actually necessary to model the axial load and assess the
validity of the simplified model.
In this work, the GA implemented with a FEM tool, following the static equilibrium condition for each time step
ti, is written in Equation (1):
½K uðti Þ ¼ rðti Þ , for i ¼ 1, 2, . . . , T, ð1Þ
where: K is the structural stiffness matrix; u(ti) is the structural displacement vector; r(ti) is the instantaneous force
vector; and i is the ith time step.
The instantaneous force and displacement vectors may be written as
2 3
r1 ðt1 Þ r1 ðt2 Þ . . . r1 ðtT Þ
6 r ðt Þ r ðt Þ . . . r ðt Þ 7
6 2 1 2 2 2 T 7
Downloaded by [Orta Dogu Teknik Universitesi] at 20:58 15 February 2016
½R ¼ 6 6 . . . . .. 7 7 ð2Þ
4 . .. .. . 5
rn ðt1 Þ rn ðt2 Þ . . . rn ðtT Þ
2 3
u1 ðt1 Þ u1 ðt2 Þ . . . u1 ðtT Þ
6 u ðt Þ u2 ðt2 Þ . . . u2 ðtT Þ 7
6 2 1 7
½U ¼ 6
6 .. .. .. .. 7 7, ð3Þ
4 . . . . 5
un ðt1 Þ un ðt2 Þ . . . un ðtT Þ
where n is the total acquisition point quantity.
Rewriting Equation (1):
½K ½U ¼ ½R: ð4Þ
The dof that match the acquisition points present known displacements [U ]a, where: ½Ub
fuðt1 Þgb fuðt2 Þgb . . . fuðtT Þgb , and T is the total
quantity of time steps. The unknown displacements are represented
by [U]b, where½Ua fuðt1 Þga fuðt2 Þga . . . fuðtT Þga . The equilibrium equations can be divided as follows:
½K aa ½K ab ½Ua ½Ra
¼ : ð5Þ
½K ba ½K bb ½Ub ½Rb
Considering that all external forces [R]a are null:
½K aa ½K ab ½Ua ½0a
¼ : ð6Þ
½K ba ½K bb ½Ub ½Rb
Taking the first equation set:
½K aa ½Ua ¼ ½K ab ½Ub ¼ ½0: ð7Þ
So:
½K aa ½Ua ¼ ½K ab ½Ub ¼ R a : ð8Þ
The displacement matrix [U]a is calculated by an algorithm for static solution case based on Saturnino (2004).
Locators
Screw
clamp
Base plate
Downloaded by [Orta Dogu Teknik Universitesi] at 20:58 15 February 2016
deflection can be reduced by optimisation by using one of the intelligent optimisation techniques (GA). This is
mainly useful when thin castings have to be machined. The second example is a comprehensive one, in which a single
spindle drill is used to drill four holes in sequence. Figure 8 shows the experimental fixture setup with a sample
workpiece.
B C P
0.5
ns
C « 0.1 A
P P
P-P O P- P’ O PO
P’ O P’ O
B PO P’ O
A CS P CS P
CS P na na
CS P
Figure 9. Position and perpendicularity of a hole. na
it has two data and one target of the hole. Figure 8(a) shows the true position and perpendicularity of the hole
(Rong et al. 2001).
Downloaded by [Orta Dogu Teknik Universitesi] at 20:58 15 February 2016
tpos
i ¼ P~i Po
~ n~a ð9Þ
TPOS
maxtpos
i
, ð10Þ
2
~ ¼ theoretical exact location point on the axis; P~i ¼ any arbitrary point
where: n~a ¼ unit vector of the datum plane; Po
pos
on the substitute axis; ti ¼ position value of the point P~i ; and TPOS ¼ required position tolerance of the hole
(actually the axis).
Tperp
maxtperp
i
, ð12Þ
2
where: P~0 o ¼ location point on the substitute axis; tperp
i ¼ perpendicularity value of point P~i ; Tperp ¼ required
perpendicularity tolerance of the hole.
Tperp
maxtperp
i
, ð15Þ
2
where: n~a ¼ unit vector in the normal direction of the primary datum plane; n~s ¼ unit vector in the normal direction
of the theoretical exact plane; P~i ¼ any arbitrary point on the substitute plane; P~0 o ¼ centre point of the substitute
axis; tperp
i ¼ perpendicularity value of point P~i ; Tperp ¼ required perpendicularity tolerance of the plane.
International Journal of Production Research 3461
Sample
s1 no. Description Details
s2
1 Diameter of drill 19.05 mm
2 Material of the drill HSS
3 Material of the workpiece Al-7075
4 Cutting speed (V) 1.16 to 1.67 m/s
m2 5 Spindle speed (S) 1500 rpm
6 Feed 0.2 mm/tooth
Figure 13. Perpendicularity error of the hole axis. 7 Horse power 1
holes (Wang 2002). For a set of m critical points, a direct evaluation of these variations is to use the sum of their
squared magnitudes, such that:
Downloaded by [Orta Dogu Teknik Universitesi] at 20:58 15 February 2016
" #
X
m I Rj
A¼ : ð19Þ
j¼1
Rj R2j
The positional error of a hole can be described as a directional pointwise error. If the plane perpendicular to the axis
of the hole is represented with two orthogonal unit vectors m1 and m2 at the hole centre sj, then the positional error
of the hole can be described as:
2
2
d2j ¼ ðm1 ÞTj sj þ ðm2 ÞTj sj : ð20Þ
IfPmultiple holes are to be machined in a single fixture setup, it is reasonable to use the sum of the positional errors
ð j d2j Þ of all the holes for a representation of their overall error.
Another example is about the perpendicularity of a hole to a surface shown in Figure 13. According to the
standards, the perpendicularity form error is being evaluated based on the following calculation of the hole centre
point deviations at its end positions s1 and s2:
t ¼ mT1 s1 mT2 s2 : ð21Þ
8. Case study 1
A 3D case study of a component used in this research has been taken from Wardak (1999) for describing the
methodology. A rectangular plate with lx ¼ 38.1 mm (1.5 in), Lx ¼ 76.2 mm (3 in), ly ¼ 50.8 mm (2 in),
Ly ¼ 101.6 mm (4 in), and h ¼ 6.35 mm (0.25 in) is drilled using a drill bit of 19.05 mm (0.75 in) diameter. The
machining detail of case study 1 simulation is given in Table 3. This problem under consideration is solved using
the ANSYS software. The magnitude of the maximum deformation 0.1493 was significant as per Wardak (1999).
The optimum positions of the locators and clamps are given in Table 4. Table 5 shows the simulation GA
parameters for case study 1. Figures 14 and 15 show the convergence of GA for fixture layout and the relationship
International Journal of Production Research 3463
Table 4. Optimal position of fixture Table 5. Simulation GA parameters for
element layout (case study 1). case study 1.
0.018 GA Von_stress
0.016 200
0.014
0.012
Von_stress
0.01 150
0.008
0.006
100
0.004
0.002
0 50
0 10 20 30 40 50 60
Number of generation
0
Figure 14. Convergence of GA for the fixture layout of case 0 10 20 30 40 50 60
study 1. Number of generation
between the von Mises stress and number of generations of this case-study problem. Figures 16 and 17 show the
workpiece deformation and von Mises stress in optimal fixture configuration.
9. Case study 2
This example is a practical simple problem where a fixture is designed to drill four holes simultaneously in a thin
square aluminium workpiece (i.e. 100 mm 100 mm 6.35 mm). Figure 18 shows the simultaneous drilling
experimental fixture setup. The fixture design process is to determine the positions of the fixture elements (locators
and clamps) and the reference data to machine and the workpiece. The machining detail of this simulation is given in
Table 6. The positions of these fixture elements along with machining loads created by boundary conditions to
the finite element model are optimised to minimise workpiece deflection based on the FEM simulation. The
constraints on this FEM model are the six fixed displacement constraints on the locators’ location, the two
clamping-point locations, and the four load cases. The optimum positions of the locators and clamps are shown in
Table 7. Table 8 shows the simulation GA parameters for case study 2 and 3. Figures 19 and 20 show the workpiece
deformation and von Mises stress in optimal fixture configuration. Figures 21 and 22 show the convergence of GA
for fixture layout and the relationship between the von Mises stress and number of generations in this case-study
problem.
3464 K. Siva Kumar and G. Paulraj
Downloaded by [Orta Dogu Teknik Universitesi] at 20:58 15 February 2016
Sample
no. Description Details
0.03
Von_stress
150
0.029 100
0.028 50
0
0.027 0 10 20 30 40 50 60
Generation
0.026
0 10 20 30 40 50 60 Figure 22. Von Mises stress vs generation of case
Generation nos study 2.
Fixture Co-ordinates
element (x,y) mm
L1 62.05, 3.11
L2 35.30, 3.13
L3 72.52, 3.01
L4 27.28, 80.88
L5 65.46, 74.91
L6 50.51, 27.05
C1 42.25, 3.04
C2 30.18,3.00
F1 500 N
F2 500 N
Downloaded by [Orta Dogu Teknik Universitesi] at 20:58 15 February 2016
Figure 24. Deformation of the optimum fixture configuration (case study 3).
Figure 25. Von Mises stress of the optimum fixture configuration (case study 3).
International Journal of Production Research 3467
Convergence of GA for fixture layout
0.034
GA
0.033
Objective function
0.032
0.031
0.03
0.029
0.028
0 5 10 15 20 25 30 35 40 45 50 55
Generation
Von_stress vs generation
300
250
200
Von_stress
150
100
Von_stress
50
0
0 10 20 30 40 50 60
Generation
Table 11. CMM measurement for a multispindle drilled hole Table 12. CMM measurement for a single-spindle drilled
work piece. hole work piece.
Downloaded by [Orta Dogu Teknik Universitesi] at 20:58 15 February 2016
Perpendicular Perpendicular
Hole No Cylindricity Diameter angle Hole no Cylindricity Diameter angle
Single-spindle
Geometrical Multispindle drilled drilled
analysis work piece work piece
each support location is compared with the yield stress of the workpiece material, to ensure that the material does
not exhibit plastic deformation during machining. The von Mises stress is treated as a state variable and is not
allowed to exceed the workpiece material yield strength.
Acknowledgements
The authors would like to thank Dr Kalyanmoy Deb, KanGal, IIT, Kanpur, for the GA code used in this study.
References
Brost, R. and Goldberg, K., 1994. A complete algorithm for synthesizing modular fixtures for polygonal parts. In: Proceedings of
IEEE Robotics and Automation, San Diego, CA, pp.535–542.
Cai, W., Hu, S.J., and Yuan, J., 1996. Deformable sheet metal fixturing: principles, algorithms, and simulation. ASM Journal of
Manufacturing Science and Engineering, 118 (3), 318–324.
International Journal of Production Research 3469
De Meter, E.C., 1998. Fast support layout optimization. International Journal of Machine Tools and Manufacture, 38 (10–11),
1221–1239.
Deb, K., 1998. Genetic algorithm in search and optimization: the technique and applications. In: Proceedings of International
Workshop on Soft Computing and Intelligent Systems, Calcutta, India, 58–87.
Deb, K., 1999. An introduction to genetic algorithms. Sadhana Journal, 24 (4–5), 293–315.
Deng, H., 2006, Analysis and synthesis of fixturing dynamic stability in machining accounting for material removal effect. Thesis
(PhD), Georgia Institute of Technology Atlanta.
Hershkovitz, M., Tasch, U., and Teboulle, M., 1996. Towards a mathematical model of the human grasping quality sense.
Journal of Robotic Systems, 13 (13), 25–34.
Kim, D. and Ramulu, M., 2004. Drilling process optimization for graphite/bismaleimide-titanium alloys stacks. Composite
Structures, 63 (1), 101–114.
Kim, D., Ramulu, M., and Garbini, J., 2001. Hole quality in drilling of graphite/bismalemidetitanium stacks. In: International
SAMPE Technical Conference, 33, 1315–1326.
Kim, D., Ramulu, M., and Pedersen, W., 2005. Machinability of titanium/graphite hybrid composites in drilling. Transactions of
NAMRI/SME, 33 (1), 445–452.
Krishnakumar, K. and Melkote, S.N., 2000. Machining fixture layout optimization. International Journal of Machine Tools and
Downloaded by [Orta Dogu Teknik Universitesi] at 20:58 15 February 2016