You are on page 1of 11

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/274309852

Effect of Adaptive Slicing on Surface Integrity in Additive Manufacturing

Conference Paper · August 2014


DOI: 10.1115/DETC2014-35559

CITATIONS READS
16 2,427

3 authors:

Snahungshu Sikder H. A. Kishawy


Sheridan College (Oakville) Ontario Tech University
11 PUBLICATIONS   110 CITATIONS    189 PUBLICATIONS   3,546 CITATIONS   

SEE PROFILE SEE PROFILE

Ahmad Barari
Ontario Tech University
160 PUBLICATIONS   662 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Integrated Inspection System View project

High speed hard machining View project

All content following this page was uploaded by Snahungshu Sikder on 01 April 2015.

The user has requested enhancement of the downloaded file.


Proceedings of the ASME 2014 International Design Engineering Technical Conferences &
Computers and Information in Engineering Conference
IDETC/CIE 2014
August 17-20, 2014, Buffalo, New York, USA

DETC2014-35559

EFFECT OF ADAPTIVE SLICING ON SURFACE INTEGRITY IN ADDITIVE


MANUFACTURING

S Sikder Ahmad Barari H A Kishawy


University of Ontario Institute of University of Ontario Institute of University of Ontario Institute of
Technology Technology Technology
Oshawa, Ontario, Canada Oshawa, Ontario, Canada Oshawa, Ontario, Canada
snahungshu.sikder@uoit.ca ahmad.barari@uoit.ca hossam.kishawy@uoit.ca

ABSTRACT In AM, a part is manufactured using 2D layer by layer


In today’s Additive Manufacturing (AM), a part is typically addition of material from a Computer Aided Design (CAD).
manufactured using layer by layer addition of material from a Traditionally the CAD model is transferred to Rapid
Computer Aided Design (CAD) model. Traditionally the CAD Prototyping (RP) system after converting to Stereo Lithography
model is transferred to RP system after exchanging to Stereo (STL) format which is a triangulated tessellation of the CAD
Lithography (STL) format which is triangulated tessellation of model. Then it is sliced using different slicing algorithms based
the CAD model. Then it is sliced using different slice on the RP machine specification. Alternately, similar to the
algorithms and machine constraints. The inherent uncertainties procedure utilized in this paper. The original CAD model was
in this process have led to development of adaptive direct directly sliced rather than converting it to the STL format to
slicing technique. There are several adaptive slicing techniques reduce uncertainties.
but only few researches have been done to calculate an actual
surface error factor and the cost aspect of the slicing algorithm.
This paper proposes new adaptive algorithm to compute a
surface error factor and to find the cost effective approach for
slicing. The adaptive slicing algorithm dynamically calculates
slice thickness and it is based on the allowable threshold for
surface integrity error to optimize the cost and time. The paper
also provides comparative study of previously developed
adaptive models by the authors based on cusp height and
surface integrity.
Figure 1. Cusp height and staircase effect
INTRODUCTION After finishing the slicing process the information of each
Additive Manufacturing (AM) is a part of layer based non- slice is sent to the RP machine’s material delivery or curing
conventional manufacturing technique where a part is system capable that is capable of tracing out the layer. Each
manufactured using layer by layer addition of material from a layer has an associated thickness which is constant for the
Computer Aided Design (CAD) model. AM is now considered entire layer. Once the current layer is produced, the controlling
to be a new wave of industrial revolution which is commonly computer then sends the information of the next layer to build it
used in various manufacturing and production industries such on top of the existing layers. In this way, the entire object is
as automotive, medical, and aeronautical; to reduce the built layer-by-layer.
production time when the geometric complexity of the part is Slicing of a CAD model is a highly critical stage in AM
relatively high. AM also provides significant opportunity to process chain. Large layer thickness generally incorporates
small business owners from various industries ranging from art, rough surface quality due to the stair-case effect as shown in
food, health, and fashion to manufacturing. Figure 1. Alternatively, utilizing very small thickness of layers

1 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 04/01/2015 Terms of Use: http://asme.org/terms


makes the process time consuming and expensive. These two Though tessellation of CAD data has advantages in
contradictions have led to the development of adaptive slicing reducing computational time and cost, it has also several
technique, in which slice thickness varies according to the disadvantages. Due to first order approximation of the CAD
surface topology and machine constraints. The methodology file, the STL file loses some original CAD topological and
presented and compared in this paper is based on adaptive geometrical data. Moreover, tessellated CAD data carries
slicing. The algorithm dynamically calculates slice thickness uncertainties and defects such as gaps, overlaps, and
considering the local geometric properties of the part’s surface. degenerated faces due to less topological information. STL files
The objective in the current approach is to optimize the sometimes are larger than original CAD files due to a high
prototyping cost and time provides comparative study of degree of redundancy. Subsequent slicing of a STL file
previously developed adaptive models by the authors. generally has higher slicing and chordal error [8].
Several researchers have worked on direct slicing of the
LITERATURE REVIEW CAD model. Among them Jemison and Hacker [9] were first to
Typically, research on the slicing processes for AM can be propose an algorithm for direct slicing of a CAD solid model
divided into two main categories of direct slicing and slicing of with uniform thickness. Zhao and Laperriere [10] also proposed
the tessellated CAD model. Both of these slicing categories can direct slicing of solid models using the same concept of limited
also be subdivided into two major sub-categories of uniform area deviation.
slicing and adaptive slicing. In uniform slicing the layer Kulkarni and Dutta [11] focused on the staircase effect and
thickness of the rapid prototyping process is fixed through the surface quality. They proposed two approaches to calculate the
entire process, whereas in adaptive slicing, layer thickness slice thickness based on a “Maximum Curvature Approach”
varies with the texture properties. As the proposed model is on and a “Minimum Layer Thickness Approach”. Hope et al.[12,
adaptive slicing, the main focus will be on direct adaptive 13] used B-spline surface to define a part and sloping edge
slicing. concept. They claimed that they had achieved higher surface
finish with less build time. They sliced the model by tracing
Slicing of STL CAD file surface contours and cutting vectors. Like Denis et al. [6], Mani
Dolenc and Mäkelä [1] introduced an adaptive slicing
et al[14] introduced the concept of different cusp height for
method for handling the staircase effect of STL CAD files.
different surfaces in direct slicing, using the tessellated CAD
They computed layer thickness at a certain point, based on the
model.
actual cusp height that should fall between users defined
An adaptive slicing procedure with selective hatching was
allowable cusp height limit.
implemented for slicing B-spline surface by Ma and He [15].
Using the same concept of Dolenc and Mäkelä, Sabourin
They first found the extreme point of the CAD model. They
et al. [2] proposed step by step slicing of a tessellated file. They
then sub-divided the model in z direction. These slabs were
first sliced the file with the maximum thickness limit. In the
then subdivided, based on the containment and a pre-specified
next step, they checked whether or not the cusp height of all
cusp height requirement.
slices satisfied the allowable limit of maximum cusp height.
Chen et al. [16] proposed direct slicing for CAD models. In
Those slices that did not satisfy the limit were divided into finer
this method, lines, arcs and Bezier curves were used to describe
layers. Justin and Jan Helge [3] proposed a local adaptive
the section contours, and layer thickness was taken as constant.
slicing technique for parts. Unlike other approaches, they
Similarly, Sun and Lee, [17] developed a user interactive direct
identified the parts and features of the CAD model and sliced
adaptive slicing model where the user input tolerances, desired
them independently.
error limit, and maximum/minimum height. Accordingly, the
Barari and Ahmadian [4] introduced a global adaptive
algorithm calculates the slicing thickness. There was also an
slicing methodology to optimize the layer thicknesses of the
attempt to implement the direct slicing with the Standard for
entire model. However, in that work they considered slicing of
the Exchange of Product model data (STEP) file by Starly et
the tessellated model instead of direct slicing of a CAD solid
al.[18]
model.
Debapriya and Asimava [19] developed a case specific
Sabourin et al.[5] proposed an algorithm to build an
surface–plane intersection algorithm for freeform surface. In
exterior surface using adaptive layer thickness and an interior their algorithm, they performed a unidirectional scan of the
surface with thick material layers for a STL CAD model. The parts and established connectivity of successive points by a
authors claimed their method reduced 50-80% of build time. matching technique. Slicing was done by equidistant parallel
Denis et al.[6] studied the concept of non-uniform cusp height horizontal planes by uniform zero order slicing.
depending on user requirements. Their algorithm allows the Rianmora and Koomsap [20] developed a direct slicing
designer to manually specify maximum cusp height for that applies the image processing technique to determine the
different faces as per their requirements. Pandey et al.,[7] appropriate thickness for sliced layer and recommend slicing
developed a slicing algorithm considering parabolic edge position on a three-dimensional CAD model.
profile for a fused deposition model part. In their model, Sikder et al [21] developed a new adaptive slicing
surface roughness (Ra) value was used instead of cusp height. approach to Control Cusp Height of the CAD model. The
Direct slicing of CAD model developed methodology uses the allowable minimum and

2 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 04/01/2015 Terms of Use: http://asme.org/terms


maximum thickness limits of an RP machine, and the maximum part is converted to IGES format to be exchanged to the
acceptable cusp height of the final part to calculate the number developed computational platform.
and thickness of the layers to directly slice the solid model. The developed computational program reads through the
Sikder et al.[22] studied the controllability of surface multiline 80 Character ASCII IGES file and converts the IGES
roughness based on different adaptive slicing algorithms. In file into executable customized database based on the
their study, they compared the surface derivative approach and geometric features such as Line, Curve, B-spline curves, B-
the allowable cusp height approach to find the effectiveness. spline surface, NURBS, point etc.
Both procedures directly slice the CAD model expressed in Then the data parameters of IGES file are divided into
Standard Initial Graphics Exchange Specification (IGES) three categories of points, curves and surfaces. Maximum Z
format and use the allowable minimum and maximum point (Zmax) and minimum Z point (Zmin) are computed for
thickness limits of the utilized rapid prototyping machine. The slicing algorithm from this data.
first procedure uses allowable slope of the surface in the Z-axis The IGES file is plotted based on the geometric features in
and the other is based on the user defined maximum acceptable the CAD file. If any feature entered to the computational
cusp height to calculate the number and thickness of the layers. system represented by, G. Then G can be expressed as:
Sikder et al.[23] developed a direct slicing that applies G  SN  S p  SR (
adaptive method and surface error function to optimize the 1)
number layers and the surface error due to staircase effect. Where SN represents the NURBS surfaces, SP is Planer
Sikder et al.[24] proposed adaptive slicing algorithm of surfaces, and SR is revaluated and trimmed surface.
CAD models based on a surface error factor. In their study they For SN the cloud of data is computed based on linear
also considered the effect of manufacturing cost of the parts. interpolation. SR is converted to SN. SN, NURBS curves C, and
Their algorithm provides optimized the manufacturing cost and point clouds are computed base on the Equations (2) and (3):
surface error factor for AM. n
Only a few researches have been conducted to calculate the  N i , p ( u ) w i Pi (
actual surface error factor and the cost aspect of the prototyping C (u )  i0
n 2)
process. This paper proposes new adaptive algorithm to 
i0
N i , p (u ) wi
compute a surface error factor and to find the cost effective
approach for slicing to reflect the total manufacturing cost in n m
the process.   N i , p (u ) N j ,q ( v ) wi , j Pi , j
i0 j 0
(
S N (u , v )  n m 3)
EXCHANGING CAD DATA
The work in this paper aims at slicing the original CAD
  N i , p ( u ) N j , q ( v ) wi , j
i0 j 0
models directly and efficiently instead of slicing of the STL
model. Thus the error and uncertainty produced by tessellation Where, p and q present the degree of the curve. n and m are
and other typical disadvantages of the STL file can be avoided. the numbers of control points in the parametric domain. P and
IGES [25] standard format is used as a unified solution that can w correspond to control point and their weights respectively. N
be used to transfer CAD models from any commercial CAD is the basis function for the NURBS. Parameters u and v vary
software to the rapid prototyping computational environment. from 0 to 1
IGES specification is established information structure for For the process of adaptive slicing, first the CAD model is
digital representation and exchange of product definition data. sliced once with maximum layer thickness and once with the
It supports exchanging this data among CAD and CAM minimum layer thickness. Average error for each layer is
Systems. computed for both cases and saved as maximum allowable
Unlike the STL file, the IGES file does not approximate average error Emax-avg and minimum allowable average error
the CAD surfaces during exchange from one system to another. Emin-avg as described in the following Section.
IGES passes the geometric features modelled in a CAD file, Surface Error Factor Calculation
including the primitive features and the freeform NURBS For the consecutive slicing process, the texture error
surfaces and curves. function is computed based on considering the λ numbers of
points belonging to two consecutive layer surfaces. For each
SLICING ALGORITHM pair of points (Pi and Pi-1) in the consecutive slices, the error
In this paper adaptive slicing algorithm is used for AM. factor ep is computed for μ number of different points along the
Central to the problem of slicing CAD model is the profile Curve. For computational purposes in this study the
determination of intersection points between the slicing plane value of μ and λ are taken as 10 and 20.
and the model. A computational platform is developed to
implement the slicing algorithms. The developed algorithm
receives IGES form of the CAD data as input instead of the
STL file. First, a solid or surface mode of multi-feature CAD

3 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 04/01/2015 Terms of Use: http://asme.org/terms


described by using Equations 4 and 5. This εi value is then
compared with Error threshold value Eallowable which is
computed using Equation 10.
a a (10)
Eallowable = E + (1 - )E max- avg
100 min - avg 100

In Equation 10, parameter a is user defined threshold factor


to control the desired error. As described, if εi is greater than
Eallowable then the procedure is repeated with new ti which is
the next possible slicing height based on the machine
Figure 2. Surface Error Calculation specification. Otherwise, height is saved and the computation
The surface error, ep, for Pi - Pi-1 patch is calculated as of the next slice height begins. This process runs till Zj is
follows, where υi and ηi are the Euclidian distances to the smaller than the maximum CAD height in the Z-direction. For
vertical and horizontal plane (Figure 2). the slicing process average surface error was also computed to
compare it with other slicing algorithm to find the efficiency of
 (4) the algorithm.
ep   min(i2 ,i2 )
i 1
SURFACE ERROR VERIFICATION
Surface error function at ith slice εi
 For the purpose of verification and experimental study of
(5)
i   e j the staircase and surface error effects experiments were
j 1 conducted to demonstrate the actual texture error function. The
Where εi is the surcace error function of the individual j-th CAD model and the physical prototype of the test part used in
point number. The total texture error, Et , for the entire part is these experiments are as shown in Figure 4. The length, width,
computed from the following equation. and height of this model which has four sloping surfaces are 40,
30 and 30 mm respectively. As marked in Figure 4a surface 1
 (6)
Et    i has 39.8° slope and surfaces 2, 3, and 4 have 45°, 51.3°, 59°
respectively. The rapid prototyping parts were built using
i 1
3DSYSTEMS’ RPOJET 1500 with 102μm constant slicing
Where κ is the total number of slice. The average texture
thickness.
error function ξ for the slice is calculated from the equation.
 (7)
 i
 i 1

Adaptive Slicing
In order to compute adaptive slicing heights the evaluation
algorithm starts with Zmin and layer thickness range of RP
machine. The algorithm starts by taking Zj as Zmin and the slice
thickness and maximum possible slice thickness tmax as the (a) (b)
current slice thickness, ti. Assume that we have n number of
Figure 3. Texture Error Verification Work piece, (a) CAD
layers with layer thickness ti, where i =1, 2…n. Then, Zj is
Model (b) Rapid Prototyped Model
computed form Equation 8.
j
All the step heights are computed using a three
Zj  t ; j  m
i 0
i
(8) dimensional microscopic roughness measurement and
compared with the actual computational height from the ideal
For i-th Zj height point on the CAD surface number of Pzj surface. Figures 4-7 shows the measured stair case effect data
points are computed considering the intersection between plane using a microscope for the diffrent sloping surfaces. The
SP at Zj and the CAD surface, G. The NURBS intersection measured data for the layers are computed with predicted
points are computed using a subdivision algorithm shown in values of υi and ηi computed by the algorithm showed good
[26] as follows: agreement. As shown in the Table 1 for the four different
sloping surfaces, the computed and measured vertical distance
PZj  S P  G Zj
(9)
υi as described in Figure 2, are within a 5% error limit. This
distances are critical factor in determining the value of texture
For the consecutive slicing the surface error εi is computed error function.
by considering the 20 points on the two consecutive surfaces as

4 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 04/01/2015 Terms of Use: http://asme.org/terms


Table 1: Experimental Verification of Staircase Effects
No of Measured Avg. Computed Avg.
Surfaces Vertical distance vertical distance
(μm) (μm)
1 35.2 34.9
2 65.2 66
3 80.1 81
4 90 90.5

Figure 6. Staircase Effect at Surface -3

Figure 4. Staircase Effect at Surface -1

Figure 7. Staircase Effect at Surface -4

TEST CASES
In order to verify the developed adaptive direct slicing
algorithm, it is applied to four different test CAD models
generated by commercial CAD software. Four models are
shown in Figures 8 to 11. First test model is a biological model
of arm with the minimum height in z direction of 160 mm and
maximum height of 0 mm (Figure 8). The second test model
(Figure 9) is inclined NURBS surface with the minimum height
in z direction is 38.875 mm and maximum height is 74.5mm.
Figure 5. Staircase Effect at Surface -2 The 3rd test model (Figure 10) is a mechanical part with length
in y direction is 120mm which is taken as height Z for the rapid
prototyping.
As shown in Figure11, model four consists of a concave
surface with radius 14 mm with height 20mm in z direction.

5 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 04/01/2015 Terms of Use: http://asme.org/terms


roughness. The cost is computed based on the overall volume
of the CAD drawing, machine run time and the labour cost for
Dimension 1200es. For the test case 1 cost effective breakeven
threshold value is a=40. This value for test cases 2,3 and 4 it is
approximately 38,52 and 55, respectively. The observed
changes of the parameter a can be explained by considering the
complexity of the models. Using breakeven point cost can be
reduced by 20-40% as shown in figure.

Figure 8. CAD Test model-1 Figure 9. CAD Test model-2 Table 2. Average Surface Error Cart vs Cost.

Test Average Layer Error (mm2)


Case
No. Slicing Slicing Slicing
with a=50 with with a=0
a=100

0.00919 0.00881 0.010099


1
Figure 10:CAD Test model-3 Figure 11:CAD Test model-4
0.00821 0.0076 0.0095
2
SLICING PARAMETER
Proposed algorithms are applied on the test model 0.0090443 0.008043 0.0100443
assuming that the RP machine can accept varying layer 3
thickness. The machine thickness limits of Dimensions’ 3D 0.00961 0.00941 0.0099
printer series Dimension 1200es is considered for all the test 4
cases. According to the machine specification the maximum
layer thickness (tmax) is taken as 0.33 mm and minimum layer Secondly the algorithm ran for all the combination of
thickness is (tmin) is taken as 0.25 mm. In Dimension 1200es number of layers for different algorithm. Table 3 compared the
high resolution printing costs $40 per cubic inch of material data of average surface error function with the different
used; low resolution printing costs $30 per cubic inch of algorithms for the optimum layer number proposed by the
material. Standard labor rate is $75/hr, or $100/hr for design algorithm. For all 4 cases, the average surface error function
services. value is reduced by employing the global adaptive algorithm
compared to the cusp height or the constant slice thickness
RESULTS AND DISCUSSION method for a defined layer number. Among test-cases 1-4, for
All the test cases are tested using with the proposed the test-case 2 with inclined NURBS surface, the surface
algorithm and the previously developed algorithms to compare function value reduced significantly compared to the other
the efficiency of the slicing. For comparison, the Cusp Height algorithms. From the table, it can be concluded that the average
Model [21] and constant slice thickness model were used. At surface error can be reduced significantly with the proposed
first Adaptive slicing was implemented for the three test cases algorithm when the complexity of the surface is higher.
varying the thresh hold factor from 0 to 100 to control the For test-case 1, Figure 12 shows the surface error vs. the
Average Surface Error Factor. number of layers compression, Figure 14 shows the slice
Table 1 shows that for the test case-1, the surface error orientation for the optimum layer number of 556. From Figure
factor is increased by 15% for a=50 comparing to using a=100. 14, it can be seen that the slice thickness is varying at the top
Whereas for the test case- 2 the number of layers is decreased and the bottom section of the part and the steady slice thickness
by 15% and the surface error factor is increased by 25%, which at the middle portion of the CAD part. At the top and bottom
is due to the curved geometry of the CAD model. Test case-3 parts the slice thickness is smaller compared to the middle part
shows much improved surface quality with less number of of the model due to the shape. When the number of layers is
layers. In this case the number of layers is reduced by 16% and 556, then the number of layers is reduced by 50% with a
the surface error factor is increased by 50%. As it is expected, decrease in value of texture error function approximately by
by using the developed algorithms when the part’s geometry is 50%.
complex (e.g. highly curved surfaces), using smaller layer For the inclined NURBS surface in test-case 3, the layer
thicknesses results on better surface finish and shorter built number is reduced by approximately 40% with a texture error
time. compromise of 50%, as shown in Figure 15 which shows, due
Figures 13, 16, 19 & 22 show the comparison between the to the curved nature at the top of the part, slice thickness
cost and the evaluated surface error factor as well as the surface constantly decreased.

6 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 04/01/2015 Terms of Use: http://asme.org/terms


Test-model 3 is a complex mechanical shape and sliced 5370
considering the y-axis. Using the proposed algorithm to slice

Average Surface Error(mm2)


5350 0.00981
this model, from Figure 18 and 20 surface error decreased

Cost for 50 pieces ($)


significantly compare to other algorithm. 5330 0.00961

However, for test-case 4, for the concave surface as shown 5310 Cost Curve 0.00941
in Figures 21 and 23 slice thickness was maximum till the 5290 Surface Error

concave surface. After that a sudden drop in slice thickness is 5270


0.00921

observed which then gradually increased till the top of the 5250
0.00901
surface. As shown in Figure 21, surface error and the number of
5230 0.00881
slices are both reduced by 50%. 0 20 40 60 80 100
Thresh-hold factor (α)
Table 3: Surface Error Chart for Different Algorithm
Figure 13. Cost Vs Average Surface Error (Test Model-1)
Test Average surface Error (mm2)
Case
No. Global Cusp Height Constant Number
Adaptive Algorithm[21] Layer of Layers
Algorithm Thickness

1 0.00927 0.00938 0.00946 556

2 0.00857 0.00866 0.00871 125

3 0.00883 0.00889 0.0089 413

4 0.0097 0.0098 0.00985 70

0.0102
Average Surface Error (mm2 )

0.01

0.0098

0.0096 (a) (b) (c)


0.0094 Figure 14. Optimized Layer Orientation with Number of Layers
=125 (a) CAD Model; (b) CAD Model after Slicing (Test-case
0.0092 Adaptive Slicing Algorithm 1)
0.009 Cusp Height Algorithm
0.01
Constant Layer Thickness
0.0088
485 500 515 530 545 560 575 590 605 620 0.0095
Average Surface Error (mm2 )

Number Of Layers
Figure 12. No. of Layers Vs. Average Surface Error for 0.009
Test-case 1.
0.0085

Adaptive Slicing Algorithm
0.008
Cusp Height Algorithm
Constant Layer Thickness
0.0075
108 113 118 123 128 133 138
Number Of Layers
Figure 15. No. of Layers Vs. Average Surface Error for Test-
case 2.

7 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 04/01/2015 Terms of Use: http://asme.org/terms


5750 0.0100443

Average Surface Error(mm2 )


5735
0.0095443
5720

Cost for 50 pieces ($)


5705
Cost Curve 0.0090443
5690 Surface Error
5675 0.0085443
5660

5645 0.0080443
0 20 40 60 80 100
Thresh-hold factor (α)
Figure 16. Cost Vs Average Surface Error (Test Model-2) Figure 19. Cost Vs Average Surface Error (Test Model-3)

Figure 17. Optimized Layer Orientation with Number of Layers


=125 (a)
(a)CAD Model; (b) CAD(b)Model after Slicing(c)
(Test-case
2)
0.0103
Figure 20. Optimized Layer Orientation with Number of Layers
=125 (a) CAD Model; (b) CAD Model after Slicing (Test-case
3)
0.0098
Average Surface Error(mm2 )

0.01

0.0093
Average Surface Error(mm2 )

0.0099

0.0088 0.0098

Adaptive Slicing Algorithm 0.0097
0.0083
Cusp Height Algorithm
0.0096
Constant Layer Thickness
0.0078
0.0095 Adaptive Slicing Algorithm
363 383 403 423 443 463
Number Of Layers Cusp Height Algorithm
0.0094
Figure 18 No. of Layers Vs. Average Surface Error for Test- Constant Layer Thickness
case 3. 0.0093
62 65 68 71 74 77
Number Of Layers

8 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 04/01/2015 Terms of Use: http://asme.org/terms


Figure 21: No. of Layers Vs. Average Surface Error for Test- ACKNOWLEDGMENTS
case 4. Put acknowledgments here. The research support provided
by the Natural Science and Engineering Research Council of
920
Canada (NSERC) is greatly appreciated.

Average Surface Error(mm2 )


910 0.00991
900
Cost for 50 pieces ($)

0.00981 REFERENCES
890
[1] Dolenc, A., and Mäkelä, I., 1994, "Slicing Procedures for
880 Cost Curve 0.00971 Layered Manufacturing Techniques," Computer-Aided Design,
870 Surface Error
26(2), pp. 119-126.
0.00961 [2] Sabourin, E., Houser, S. A., and Bohn, J. H., 1996,
860 "Adaptive Slicing Using Stepwise Uniform Refinement," Rapid
0.00951
850 Prototyping Journal, 2(4), pp. 20.
[3] Justin, T., and Jan Helge, B., 1998, "Local Adaptive
840 0.00941
Slicing," Rapid Prototyping Journal, 4(3), pp. 118.
0 20 40 60 80 100
[4] Barari, A., and Ahmadian, M. T., 1999, "Optimum Slicing
Thresh-hold factor (α)
of Model for Rapid Prototyping," eds., Tehran, Iran, pp. 218-
Figure 22. Cost Vs Average Surface Error (Test Model-4) 228.
[5] Sabourin, E., Houser, S. A., and Bohn, J. H., 1997,
"Accurate Exterior, Fast Interior Layered Manufacturing,"
Rapid Prototyping Journal, 3(2), pp. 44.
[6] Denis, C., Kittinan, U., and Ezat, S., 2000, "Specifying
Non-Uniform Cusp Heights as a Potential Aid for Adaptive
Slicing," Rapid Prototyping Journal, 6(3), pp. 204.
[7] Pandey, P. M., Reddy, N. V., and Dhande, S. G., 2003, "Real
Time Adaptive Slicing for Fused Deposition Modelling,"
International Journal of Machine Tools and Manufacture, 43(1),
pp. 61-71.
(a) (b) [8] Barari, A., 2012, "Profile Tolerance Allocation for Rapid
Prototyping of Sculptured Surfaces in a Direct Slicing Process,"
eds., Chicago, IL, United States, pp.
[9] Jamieson, R., and Hacker, H., 1995, "Direct Slicing of Cad
Models for Rapid Prototyping," Rapid Prototyping Journal,
1(2), pp. 4.
[10] Zhao, Z., and Laperriere, L., 2000, "Adaptive Direct
Slicing of the Solid Model for Rapid Prototyping,"
International Journal of Production
(c) Research, 38(1), pp. 69-83.
[11] Kulkarni, P., and Dutta, D., 1996, "An Accurate Slicing
Figure 23: Optimized Layer Orientation with Number of Layers Procedure for Layered Manufacturing," Computer-Aided
=125 (a) CAD Model; (b) CAD Model after Slicing (Test-case Design, 28(9), pp. 683-697.
4) [12] Hope, R. L., Jacobs, P. A., and Roth, R. N., 1997, "Rapid
Prototyping with Sloping Surfaces," Rapid Prototyping Journal,
CONCLUSION 3(1), pp. 12.
A new adaptive direct slicing algorithm of CAD models based [13] Hope, R. L., Roth, R. N., and Jacobs, P. A., 1997,
on a surface error factor has been developed and implemented. "Adaptive Slicing with Sloping Layer Surfaces," Rapid
Comparing the results of implementation with the common Prototyping Journal, 3(3), pp. 89.
practice for several case studies shows that the proposed [14] Mani, K., Kulkarni, P., and Dutta, D., 1999, "Region-
approach has greater slicing efficiency. Typically by utilizing Based Adaptive Slicing," Computer-Aided Design, 31(5), pp.
this approach of adaptive slicing prototyping cost can be 317-333.
reduced 20-45% compare to the slicing with minimum layer [15] Ma, W., and He, P., 1999, "An Adaptive Slicing and
thickness. Therefore, the developed slicing method provides a Selective Hatching Strategy for Layered Manufacturing,"
reasonable trade-off between the rapid prototyping cost and the Journal of Materials Processing Technology, 89-90(pp. 191-
rapid prototyping accuracy. For the many advantages of direct 197.
slicing outside CAD system, it will be a future solution to the [16] Chen, X., Wang, C., Ye, X., Xiao, Y., and Huang, S., 2001,
slicing process in AM. "Direct Slicing from Powershape Models for Rapid

9 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 04/01/2015 Terms of Use: http://asme.org/terms


Prototyping," The International Journal of Advanced [22] Sikder, S., Barari, A., and Kishawy, A., 2012
Manufacturing Technology, 17(7), pp. 543-547. "Controllability of Surface Roughness by Adaptive Slicing in
[17] Sun, S., Chiang, H., and Lee, M., 2007, "Adaptive Direct Rapid Prototyping Manufacturing," eds., Winnipeg, Manitoba,
Slicing of a Commercial Cad Model for Use in Rapid Canada, pp.
Prototyping," International Journal of Advanced Manufacturing [23] Sikder, S., Barari, A., and Kishawy, A., 2013, "Adaptive
Technology, 34(7/8), pp. 689-701. Slicing Approach to Control Surface Error Factor in Rapid
[18] Starly, B., Lau, A., Sun, W., Lau, W., and Bradbury, T., Prototyping Process," eds., University of Bergamo, Bergamo,
2005, "Direct Slicing of Step Based Nurbs Models for Layered Italy, pp.
Manufacturing," Computer-Aided Design, 37(4), pp. 387-397. [24] Sikder, S., Barari, A., and Kishawy, A., 2013, "Control of
[19] Debapriya, C., and Asimava Roy, C., 2007, "A Semi- Nurbs-Based Surface Error Factor Using a Manufacturing Cost
Analytic Approach for Direct Slicing of Free Form Surfaces for Optimization in Rapid Prototyping Process," eds., Saint
Layered Manufacturing," Rapid Prototyping Journal, 13(4), pp. Petersburg, Russia, pp.
256. [25] Reed, K., 1991, The Initial Graphics Exchange
[20] Rianmora, S., and Koomsap, P., 2010, "Recommended Specification (Iges): Version 5.1, U.S. Department of
Slicing Positions for Adaptive Direct Slicing by Image Commerce, National Inst. of Standards and Technology,
Processing Technique," The International Journal of Advanced [26] Barnhill, R. E., Farin, G., Jordan, M., and Piper, B. R.,
Manufacturing Technology, 46(9), pp. 1021-1033. 1987, "Surface/Surface Intersection," Computer Aided
[21] Sikder, S., Barari, A., and Kishawy, A., 2012, "Adaptive Geometric Design, 4(1–2), pp. 3-16.
Slicing Approach to Control Cusp Height of Rapid Prototyping
Parts," eds., Niagara, Canada, pp.

10 Copyright © 2014 by ASME

DownloadedViewFrom:
publicationhttp://proceedings.asmedigitalcollection.asme.org/
stats on 04/01/2015 Terms of Use: http://asme.org/terms

You might also like