Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture http://pib.sagepub.

com/

Effect and control of hatch length on material properties in the direct metal laser sintering process
Y Ning, Y S Wong and J Y H Fuh Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture 2005 219: 15 DOI: 10.1243/095440505X7957 The online version of this article can be found at: http://pib.sagepub.com/content/219/1/15

Published by:
http://www.sagepublications.com

On behalf of:

Institution of Mechanical Engineers

Additional services and information for Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture can be found at: Email Alerts: http://pib.sagepub.com/cgi/alerts Subscriptions: http://pib.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations: http://pib.sagepub.com/content/219/1/15.refs.html

Downloaded from pib.sagepub.com by guest on July 31, 2011

15

Effect and control of hatch length on material properties in the direct metal laser sintering process
Y Ning, Y S Wong*, and J Y H Fuh Department of Mechanical Engineering, National University of Singapore, Singapore The MS was received on 17 June 2004 and was accepted after revision for publication on 31 August 2004. DOI: 10.1243/095440505X7957

Abstract: As one of the advanced rapid prototyping and manufacturing processes, the direct metal laser sintering (DMLS) process gives designers the possibility of building parts of almost any complexity in a wide range of metal materials. In DMLS, the sintered material is anisotropic and heterogeneous, affecting the quality and performance of the built parts. This paper presents a study that focuses on these two material characteristics and proposes a method to control them. Firstly, a series of experiments have been designed and conducted to study the inuence of material anisotropy and heterogeneity on the built part strength. The heterogeneity was analysed and the microstructural changes due to different hatch line lengths were observed using scanning electron microscopy. The relationship between the hatch line length and the part quality is derived on the basis of experimental data collected and an appropriate statistical analysis applied. A genetic algorithm method is also proposed to optimize the hatch direction for obtaining more homogeneous sintered material properties. The experimental results verify that the negative effect of varying hatch lines in the early sintering process can be controlled effectively. Keywords: direct metal laser sintering, material properties, anisotropy, heterogeneity, genetic algorithm

INTRODUCTION

Since the late 1980s, rapid prototyping (RP) technologies have been successfully applied in product development and manufacturing. One of the well-adopted RP technologies is selective laser sintering (SLS). Although SLS has made good progress in the past decade, its further acceptance in the market place will depend on the quality of its built parts. Based on the SLS process the direct metal laser sintering (DMLS) technique is capable of building prototype parts by depositing and melting metal powders layer by layer. It also enables more applications than other SLS materials, e.g. sand and plastic powder. Metal powder can be melted directly to build functional prototypes. As a layered manufacturing process, each layered material is sintered one over another along the part orientation direction. Each two-dimensional layer
*Corresponding author: Department of Mechanical Engineering, National University of Singapore, 9 Engineering Drive 1, Singapore 117576; email: mpewys@nus.edu.sg

can be regarded as the aggregate of parallel hatch vectors. The sintered material is anisotropy because of the hatch direction and part orientation. A few previous investigations have studied and controlled the effect of material anisotropy. Ahn et al. [1] studied the effect of fused deposition modelling (FDM)-built parameters, such as hatch orientation, air gap, bead width, colour, and model temperature on the anisotropic material properties using a design of experiment approach. For stereolithography apparatus (SLA), Corbel et al. [2] tested three different orientations of load on the surface of SLA-built parts for the isotropy of mechanical properties. Andrew and David [3] reported the tensile properties of the effect of layer orientation in SLA. Subramanian et al. [4] discussed the anisotropy of green strength for a part built by SLS, while Gibson and Shi [5] investigated the inuence of hatch direction and orientation on the strength of the SLS process based on thermoplastic polymers. Besides direction dependence, the material property of each layer is also not homogeneous. The various lengths of the hatch lines cause local process variations,
Proc. IMechE Vol. 219 Part B: J. Engineering Manufacture

B10404 # IMechE 2005

Downloaded from pib.sagepub.com by guest on July 31, 2011

16

Y Ning, Y S Wong, and J Y H Fuh

Fig. 1

(a) Spiral path pattern and (b) zigzag path pattern

resulting in heterogeneous material properties. Small variations in process environment, such as the uctuation of ambient temperature and laser power, can also affect the homogeneity of material. The length of hatch line has been found to be one signicant factor that affects the quality of the nal part according to the earlier studies [68]. As the hatch length increases, the time delay between energy pulses increases, thereby lengthening the cooling time and reducing over-sintering [6]. On the other hand, a short hatch length and its corresponding short scanning time results in heterogeneity in the material properties of the part. By applying an appropriate hatch direction, the variation in the lengths of hatch lines can be controlled effectively. However, thus far, other hatch direction search methods only aim to reduce the number of hatch vectors. Rajan et al. [9] proposed an algorithm to select the scan hatch direction by minimizing the hatch vector segments. Qiu et al. [10] chose the path direction with two intelligent features: least number of paths and closest next starting point. No implemented method is yet available to reduce effectively the negative effect brought by the hatch length by optimizing the hatch direction. Therefore, the relationship between the hatch direction and length of the hatch line should be studied. This paper presents a method to reduce the effect of hatch lines by optimizing the hatch direction.

(CAD) model is rst created using computer aided design CAD software, such as AutoCAD and ProEngineer. Then the model is decomposed with a series of parallel layers with a specic thickness along a predened part orientation. After that, for most RP processes such as SLS, SLA, and FDM, each layer is lled with a series of discrete lines. Currently, two popular path patterns are widely used in RP. One is a spiral path pattern that consists of a set of contours parallel to the layer boundary with different offset values. Another is a zigzag path pattern that consists of a series of parallel lines. The two different path patterns are shown in Fig. 1. The study in this paper is based on the latter path pattern because the zigzag path pattern is simpler to implement and more popular. The building of each two-dimensional layer can be regarded as the accumulation of the parallel hatch vectors. In the DMLS process, the laser continuously sinters along the dened hatch vectors to build the layer. Each thin layer of sintered material is successively added over a previous layer. As the laser scans over the material, it melts and fuses together to form the two-dimensional layer. 2.2 Material anisotropy

MATERIAL HETEROGENEITY AND ANISOTROPY OF THE DMLS PROCESS Material accumulation process

2.1

Two critical factors determine the anisotropy; the rst is part orientation. Because the part is accumulated layer by layer along the orientation direction, material properties are much different in the orientation direction from those in the directions perpendicular to the orientation. The material properties in these sliced layers are also different in different directions because of the inuence of another factor, which is the hatch direction. 2.3 Material heterogeneity

Unlike the traditional mechanical material removal process, RP is based on the layered manufacturing principle and is a material addition and building process. Lin et al. [11] dened the principle of layered manufacturing fabrication as a decomposition accumulation process. A solid three-dimensional

Besides the inuence on the material properties due to hatch direction and orientation, the microscopic structure of each layer is uneven. The properties in the local elds lled with different lengths of hatch
B10404 # IMechE 2005

Downloaded from Proc. IMechE Vol. 219 Part B: J. Engineering Manufacture pib.sagepub.com by guest on July 31, 2011

Effect and control of hatch length on the material properties in the DMLS process

17

lines are not uniform. These differences affect the heterogeneity of the material property. The effect of small uctuation of the process environment can be neglected because low variability on a macroscopic scale is referred to as quasi-homogeneous [12]. Effective control of material anisotropy can be achieved by adjusting the orientation and hatch direction. If the built part needs to be operated under a load, the direction with a higher strength should be in the load direction to prolong the life cycle of the built part. However, it will be difcult to control material heterogeneity because the hatch lengths vary greatly for different geometric shapes. The resulting material microscopic structure varies according to the location on a layer. This property is undesirable because it disrupts the homogeneity in each sintered line and is difcult to control.

EFFECTS OF MATERIAL ANISOTROPY AND HETEROGENEITY ON THE PART STRENGTH

To investigate the effect of material anisotropy and heterogeneity on the tensile strength of the part, a set of experiments was designed and implemented. 3.1 Experimental methods

A DMLS machine was used in the experimental study. Based on the SLS technology, it has wider applications because the metal powder can be sintered directly to build functional prototypes. It can fabricate threedimensional near-net-shape metal parts directly in a single process by using a relatively low-power laser

to sinter low-shrinkage steel- or bronze-based metal powders layer by layer. In the following, the experimental results and relevant analyses are based on the DMLS process. The specimens were designed according to the ASTM Standard E 8 [13] for the tensile testing of metallic materials with a 0.235 mm offset [14] for the laser beam compensation. The thickness and reduced sectional width of the specimen are both 6.35 mm. The overall length L is 92 mm. All the dimensions follow the ASTM Standard E 8. To study the inuence of material anisotropy and heterogeneity on the part strength, two groups of experiments were designed. Except for the hatch direction and part orientation, all the other process parameters are set as follows: laser scan speed, 100 mm/s; laser power, 100 W; hatch distance, 0.2 mm; thickness, 0.1 mm. The metal powder system used in this process is a mixture of 60 wt % pure copper powder and 40 wt % pre-alloyed metal powder. The pre-alloyed metal powder, serving as a binder in sintering, is a Cu-based alloy with a melting point of 646 8C. To conduct these experiments, a prototype DMLS system (Fig. 2) was employed. This system consists of a 200 W continuous-wave CO2 ( 10:6 mm) laser. All the parts were built in an ambient atmosphere without preheating by the experimental DMLS system. In the earlier research, different hatch directions usually result in different hatch line lengths that consequently cause signicant change in mechanical strength. However, the direct relationship between the hatch direction variation and the mechanical strength, particularly when the hatch line is kept

Fig. 2 The prototype DMLS system

B10404 # IMechE 2005

Downloaded from pib.sagepub.com by guest on July 31, 2011

Proc. IMechE Vol. 219 Part B: J. Engineering Manufacture

18

Y Ning, Y S Wong, and J Y H Fuh

Table 1 The design of experimental specimens for groups A and B

unchanged, has not been investigated. In these experiments, The rst group A (Table 1) includes three cases with different orientations and hatch directions. In group A, the specimens were built by the DMLS machine directly and the lengths of the hatch lines were different. These were designed to test the different mechanical properties in different orientations and hatch directions with the effect of hatch line lengths (short, A1 and A3; long, A2). The other group, group B, was designed to remove the inuence of material heterogeneity (caused by short hatch lines) on the mechanical properties. In group B, a cuboid was built rst. After the cuboid was built, it was cut to the standard specimen by using the wire cut by electrode discharge machining. Subgroup B1 in group B has the same orientation and hatch direction as another subgroup A1 in group A. The only difference between these two parts is the hatch length in each layer. Unlike the short hatch lines sintered in A1, the hatch line that makes up the layers of B1 is part of a long hatch line. This difference in hatch length is similarly applied to B2 and A3. 3.2 Experimental results and discussion

is higher than the parallel strength (i.e. greater than 25 MPa for subgroups A1, A2, and B1). In these three parts with a build orientation perpendicular to the load direction, the average strength of subgroup A2 is 36.78 MPa. In this case their hatch directions are parallel to the load direction. The other two in subgroups A1 and B1 have a lower average strength of 32.01 and 27.90 MPa (A2 > A1 and A2 > B1 ) when the hatch direction is perpendicular to the load direction. The average strength of those in subgroups A3 and B2 with the part orientation parallel to the load direction was lower (i.e. 19.61 and 19.03 MPa respectively). The part that uses a longer hatch line has reduced part strength when other process parameters

Each of the subgroup sets was built three times by the DMLS system. Using the Instron1 model 8501 dynamic testing system [15], the ultimate tensile strengths  of the test parts were obtained and shown in Fig. 3. When the part orientation is perpendicular to the load direction, the average part strength

Fig. 3 Variations in ultimate tensile strengths for groups A and B

B10404 # IMechE 2005

Downloaded from Proc. IMechE Vol. 219 Part B: J. Engineering Manufacture pib.sagepub.com by guest on July 31, 2011

Effect and control of hatch length on the material properties in the DMLS process

19

Fig. 4 Sintered blocks with different lengths L

are kept unchanged (A1 > B1 and A3 > B2 separately). Two ndings can be summarized from the above results: rstly, although both the orientation and the hatch direction affect the anisotropy, orientation has a greater inuence on the part strength; secondly, the length of the hatch line also affects the part strength. The strength of the built part using short lines is higher than that using long hatch lines.

HETEROGENEITY CAUSED BY DIFFERENT HATCH LENGTHS

microscopy (SEM) (JESL1 JSM-5500 system [16]) and are shown in Fig. 5. In Fig. 5(a), L is the smallest and it can be seen that the interconnected pores are small. With increased L, the interconnected pores become larger. However, when L is sufciently long, there is little change in interconnected pore size, as can be seen in Figs 5(d), (e), and (f). These indicate that the sintered material becomes denser with shorter hatch lines (Table 3). Under that circumstance, the sintered powder does not have sufcient time to cool as it absorbs energy transferred from the neighbouring hatch during the sintering process. The absorbed energy adds to the density of the energy, resulting in a denser structure. 4.2 Variation in percentage shrinkage with the hatch length

In order to study the relationship between the hatch length and the material heterogeneity, blocks (Fig. 4) of different nominal lengths L have been built. L has seven values as listed in Table 2. The height of each block is 10 mm. Three of each block (illustrated by A, B, and C in Fig. 4) were built and the average of the three measured values was taken. The process parameters were a laser power of 100 W, a hatch distance of 0.2 mm, and a layer thickness of 0.1 mm. In each layer, the laser sintered the block in turn; i.e. the laser started sintering block A1 and then moved the beam to the top left corner of A2 and sintered the block, and continued when, nally, C7 was sintered. 4.1 Microstructure of the part built with different hatch lengths

Results of SEM experiments indicate that shorter hatch lines will have a more serious effect on the homogeneity of the part. As a denser part will result in a higher percentage shrinkage, this can be used as a measure of the effect of hatch length on the part property. The percentage shrinkage S of each block can be calculated as

L L0 L

The morphologies of the sintered parts built with different L were observed using scanning electron

The actual length L0 of the block was measured using the Deltronic1 MPC-5 system [17] with an accuracy of 0.001 mm. The percentage shrinkage obtained for each case of nominal length L is based on the mean value calculated from the three built parts and is shown in Table 4.

Table 2 Length settings

1 Size L (mm) W (mm) 3 20 2 6 20 3 10 20 4 18 20 5 30 20 6 50 20 7 70 20

B10404 # IMechE 2005

Downloaded from pib.sagepub.com by guest on July 31, 2011

Proc. IMechE Vol. 219 Part B: J. Engineering Manufacture

20

Y Ning, Y S Wong, and J Y H Fuh

Fig. 5 SEM images of sintered samples using different length of hatch lines: (a) L 3 mm; (b) L 10 mm; (c) L 30 mm; (d) L 50 mm; (e) L 70 mm; (f ) packed powder

Table 4 shows that shorter L results in a greater shrinkage. With increased L, the shrinkage tends towards a steady value (i.e. 0.8 per cent). 4.3 Data tting

As shown in Fig. 5, the shorter hatch lines make the part denser and result in more serious shrinkage.
Table 3

With greater difference between the actual percentage shrinkage with the steady percentage shrinkage, the material caused by the shorter hatch lines becomes less homogeneous. To reduce this negative effect, short hatch lines should be avoided. Because the length of the hatch line affects the material homogeneity, the relationship between the hatch length and the material properties is rst established by a

Fractional density of sintered samples using different lengths of hatch lines

Packed powder L 3 mm 79.6 L 10 mm 75.2 L 30 mm 70.3 L 50 mm 69.1 L 70 mm 68.6

Fractional density (%)

60.5

Downloaded from Proc. IMechE Vol. 219 Part B: J. Engineering Manufacture pib.sagepub.com by guest on July 31, 2011

B10404 # IMechE 2005

Effect and control of hatch length on the material properties in the DMLS process

21

Table 4 The change in percentage shrinkage for various L

L (mm)
3 6 10 18 30 50 70 Shrinkage (%) 9.8 5.2 2.9 1.6 1.1 0.8 0.8

suitable curve tting. Several different curve models were considered. The inverse model curve with the best tting result was nally adopted. The general equation of the inverse model is

S a

b c L L2 28:401 92 1:468 81 L L2

The tted result has been found to be

S 0:204 96

controlled. For a given section of a layer, the hatch length can be signicantly different with respect to the different hatching direction. As it is too complex to identify mathematically the hatch direction because of the variation in the layer geometry, an approach based on the genetic algorithm (GA) is used. GA is a stochastic searching and optimization method based on the metaphors of natural biological evolution [18]. Because the GA does not require derivative information or other auxiliary knowledge and only the objective function and corresponding tness levels inuence the search [18], it is suitable for use to solve this optimization problem. The developed method was implemented using a C programming software package called GAlib from Massachusetts Institute of Technology [19]. GAlib is a set of GA objects and includes tools for searching optimization routines written in any C program. 5.1 Optimization procedure

The tted curve is shown in Fig. 6. Based on the tted curve, the relevant effect F of the hatch length on the heterogeneity of each layer can be dened and estimated as PNtotal i 1 0:204 96 28:401 92=Li 1:468 81=L2 0:8 i F 4 Ntotal where the value 0.8 denotes the steady percentage shrinkage.

MINIMIZING THE EFFECT OF SHORTER HATCH LINES ON THE MATERIAL PROPERTIES BY GENETIC ALGORITHM OPTIMIZATION

By optimizing the hatch direction for each sliced layer, the number of short hatch lines can be

The slicing software has been programmed in Visual C (VC). The developed software can transfer the CAD models in STL format to the layer data required by the RP machine. To implement the optimization procedure, the system software calls the functions in GAlib. A three-dimensional model is rst presliced and translated to two-dimensional layers with their contours recorded. GA is then applied to optimize the hatch direction  for each layer. A real-number genome, which is based on the hatch direction angle , is used. Initially, the value of the genome  is randomly set to a value on the scale of (0, p). The objective function is dened by equation (4) and is used to evaluate the genome. The optimization is to nd the hatch direction  that gives the minimum F value. The value of F determines the tness of the individual value  for

Fig. 6 Inverse model relating percentage shrinkage and hatch length

B10404 # IMechE 2005

Downloaded from pib.sagepub.com by guest on July 31, 2011

Proc. IMechE Vol. 219 Part B: J. Engineering Manufacture

22

Y Ning, Y S Wong, and J Y H Fuh

Fig. 7 The process ow of optimizing the hatch direction with a GA (2-D, two-dimensional)

evolving. A smaller value of F represents a better result. New generations of hatch direction  are derived using the mutation operator based on the tness value F . The number of generations equal to 1000 is used as a stopping criterion. The procedure is depicted in Fig. 7. 5.2 Case study 1

In this case, two rotor blades were built to study the effect of different hatch directions on the nal part quality. The geometric shape of the blade is shown in Fig. 8. Two parts were sliced with the same orientation in the Y direction, but different hatch directions were applied to them. The hatch direction used in the rst part was in the X direction; the second adopts the optimized direction which is in the Z direction (vertical to the hatch direction of the rst part). Both rotor shades were built with identical process parameters settings. The two built parts are shown in Fig. 9.

Fig. 8 The geometric shape of the rotor blade example

B10404 # IMechE 2005

Downloaded from Proc. IMechE Vol. 219 Part B: J. Engineering Manufacture pib.sagepub.com by guest on July 31, 2011

Effect and control of hatch length on the material properties in the DMLS process

23

Fig. 9 Two blades built with different hatch directions: (a) X direction (without optimization); (b) Z direction (with optimization)

The results show that the part [Fig. 9(b)] using the optimized hatch direction can achieve long hatch lines and better surface quality without a warpage phenomenon [Fig. 9(a)]. In the rst part [Fig. 9(a)], the handle of the shade is warped by about 1 mm due to overheating. This is due to the small diameter of the handle (only 5 mm) at each layer. The adjacent short hatch lines were sintered quickly, resulting in fast build-up of heat, and hence overheating the small region. In addition, the surface roughness of the rst case was worse than the second case (Ra 29:46 mm versus 21.71 mm) since the material of the second part was more heterogeneous. 5.3 Case study 2

giving totally 23 layers. The two-dimensional contours lled with hatch lines along the optimized directions are shown in Fig. 10. The top left is the three-dimensional model of the engine carburettor cover. As the gure shows, the hatch directions were signicantly different when the two-dimensional geometric shapes varied. To validate the optimization result of the GA algorithm, the F values based on the optimized direction  and those with a xed direction ( 0) are compared and listed in Table 5. It can be seen that the effect factor F due to the short hatch of each layer can be reduced signicantly. Therefore the effect of the short hatch lines on the quality of nal parts can be reduced by minimizing the number of short hatch lines.

To demonstrate the effectiveness of the aforementioned GA approach, a case based on an engine carburettor cover model with the size of 159.72 mm (width) 130 mm (length) 114.18 mm (height) was investigated. To limit the number of layers for illustration, the thickness of each layer was set at 5 mm,
B10404 # IMechE 2005

CONCLUSIONS AND FUTURE WORK

In DMLS, the hatch length differences cause the built part to be anisotropic and inhomogeneous. From the study of strength variation with the hatch direction
Proc. IMechE Vol. 219 Part B: J. Engineering Manufacture

Downloaded from pib.sagepub.com by guest on July 31, 2011

24

Y Ning, Y S Wong, and J Y H Fuh

Fig. 10

The optimized hatch directions for an engine carburettor cover

and orientation, the effect of the hatch length on the part strength has been identied. The quantitative relationship between the nal quality of the part and the hatch length has been experimentally

determined. The experimental results show that shorter hatch lines can change the homogeneity of the material to a great extent. To reduce this negative effect, a GA-based approach has been proposed to

Table 5 Comparison of the traditional method and the proposed optimization method
Layer 01 02 03 04 05 06 07 08 09 10 11 12 optimized 160 160 99 94 94 94 91 93 91 93 164 146

F
0.028 0.028 0.121 0.096 0.079 0.135 0.134 0.154 0.184 0.254 0.240 0.494

traditional 0 0 0 0 0 0 0 0 0 0 0 0

F
0.313 0.314 0.451 0.349 0.323 0.400 0.590 0.391 0.373 0.516 0.528 0.641

Layer 13 14 15 16 17 18 19 20 21 22 23

optimized 41 173 180 180 180 180 180 6 90 89 91

F
0.508 0.166 0.125 0.066 0.050 0.036 0.030 0.438 0.467 1.152 1.321

traditional 0 0 0 0 0 0 0 0 0 0 0

F
0.547 0.206 0.125 0.066 0.050 0.036 0.030 0.528 0.788 5.743 6.543

Downloaded from Proc. IMechE Vol. 219 Part B: J. Engineering Manufacture pib.sagepub.com by guest on July 31, 2011

B10404 # IMechE 2005

Effect and control of hatch length on the material properties in the DMLS process

25

optimize the hatch direction automatically based on the different two-dimensional layer geometric shapes. The GA-based hatch direction optimization method can be used to improve the part homogeneity effectively. The study on the effects of part orientation and hatch direction can help to improve the part strength effectively by selecting suitable values for these two process parameters. This method can determine an optimized hatch direction that reduces the effect of short hatch lines effectively. However, it still cannot completely prevent short hatch lines, especially when the two-dimensional layer geometry has small independent features.

10

11

12

13

REFERENCES
1 Ahn, S. H., Montero, M., Odell, D., Roundy, S., and Wright, P K. Anisotropic material properties of fused . deposition modeling (FDM) ABS. Rapid Prototyping J., 2002, 8(4), 248257. 2 Corbel, S., Hinczewski, C., and Chartier, T. Mechanical properties of ceramic parts made by stereolithography and sintering process. In Proceedings of the European Conference on Rapid Prototyping and Manufacturing, 1999, pp. 115123. 3 Andrew, C. L., and David, W. R. The effect of layer orientation on the tensile properties of net shape parts fabricated in stereolithography. In Proceedings of the Conference on Solid Freeform Fabrication, 2003, pp. 289300. 4 Subramanian, P. K., Vail, N. K., Barlow, J. W., and Marcu, H. L. Anisotropy in alumina produced by SLS. In Proceedings of the Conference on Solid Freeform Fabrication, 1994, pp. 330338. 5 Gibson, I., and Shi, D. P. Material properties and fabrication parameters in selective laser sintering process. Rapid Prototyping J., 1997, 3(4), 129136. 6 Badrinarayan, B., and Barlow, J. W. Effect of processing parameters in SLS of metalpolymer powders. In Proceedings of the Conference on Solid Freeform Fabrication, 1995, pp. 5563. 7 Beamen, J. J. Solid Freeform Fabrication, A New Direction in Manufacturing, 1997 (Kluwer, Boston, Massachusetts). 8 Richard, H. C. Computer aspects of solid freeform fabrication geometry, process control, and design. In

14

15 16 17 18

19

Proceedings of the Conference on Solid Freeform Fabrication, 1993, pp. 102112. Rajan, V. T., Vijay, S., and Konstantinos, A. T. The optimal zigzag direction for lling a two-dimensional region. Rapid Prototyping J., 2001, 7(5), 231241. Qiu, D., Langrana, N. A., Danforth, S. C., Safari, A., and Jafari, M. Intelligent toolpath for extrusion-based LM process. Rapid Prototyping J., 2001, 7(1), 1823. Lin, F., Sun, W., and Yan, Y. A decomposition accumulation model for layered manufacturing fabrication. Rapid Prototyping J., 2001, 7(1), 2431. Isaac, M. D., and Ori, I. Engineering Mechanics of Composite Materials, 1994 (Oxford University Press, New York). Standard E8-99, Standard test methods for tension testing of metallic materials. In Annual Book of ASTM Standards, 1999 (American Society of Testing and Standards, Philadelphia, Pennsylvania). Tang, Y. X., Loh, H. T., Fuh, Y. H. J., Wong, Y. S., Lu, L., Ning, Y., and Wang, X. H. Accuracy analysis and improvement for direct laser sintering. Presented at the SMA Annual Symposium 2004, Singapore, January 2004. Manual of Instron Model 8501 Dynamic Testing System, 1990 (Instron Corporation, Canton, Massachusetts). Instructions of JSM-5500 Scanning Electron Microscope, 1999 (JESL Corporation, Jeol, Japan). Deltronic MPC-5 User Manual, 1998 (Deltronic Corporation, California). Zalzala, A. M. S., and Fleming, P. J. Genetic Algorithms in Engineering Systems, 1997 (Institution of Electrical Engineers, London). Matthew, W. GAlib: a C library of genetic algorithm components, 1995, http://lancet.mit.edu/ga/.

APPENDIX Notation

a b c i L L0 Ntotal S 

unknown coefcients to be determined number of hatch lines length of the hatch line actual measured value total number of hatch lines in the layer percentage shrinkage ultimate tensile strength

B10404 # IMechE 2005

Downloaded from pib.sagepub.com by guest on July 31, 2011

Proc. IMechE Vol. 219 Part B: J. Engineering Manufacture