Study of RP&M and Implementation of the Software For General Part Fabrication

Transcription

1 Study of RP&M and Implementation of the Software For General Part Fabrication Submitted to Committee Members Dr. Kai-Hsiung Chang Dr. Bor.Z Jang Dr. Wen-Chen Hu By Qiang Yao In Partial Fulfillment of the Requirement for the Degree of Master of Computer Science and Software Engineering September 21, 2000

2 ACKNOWLEDGMENT I would like to thank Dr. Bor.Z Jang for his supervision, guidance and financial support. Without his help, I can not finish my graduate study. Especially when I was in hard time, he gave me encouragement, nice understanding, and kind help. Thank you very much! I will remember your kind help all my life! I would also like to thank my advisor Dr.Chang for his suggestion and valuable time, thank Dr.Hu to be my advisory committee member and his friendly help. Thanks are also given to others for their support and proofreading. I

3 PAPER ABSTRACT STUDY OF RP&M AND IMPLEMENTATION OF A SOFTWARE FOR GENERAL PART FABRICATION Qiang Yao Master of Computer Science and Software Engineering 51 Typed Pages Directed by Dr. Bor. Z Jang, Kai-Hsiung Chang Rapid Prototyping and Manufacturing(RP&M) is a new technology in manufacturing industry. In this report, the RP&M technology features, principles, and applied methods are introduced. Specifically this paper addresses some software issues related to a new RP&M process for the fabrication of continuous fiber reinforced composite objects. Useful algorithm and software have been developed for handling a commonly used RP&M solid interface data format called Common Layer Interface (CLI). The software developed is capable of reading CLI files, organizing these files and then converting them to appropriate CNC file for controlling the operations of a layer manufacturing machine. Fiber composite samples have been successfully fabricated by using this machine. II

4 TABLE OF CONTENTS Chapter 1. Rapid Prototyping and Manufacturing Technology Introduction Technology Features of RP&M The Principles of RP&M CAD Modeling Slicing along the Z Direction Layer Information Processing Layer Forming and Stacking Post-processing Simple Graphic Expression of RP&M More Common Methods of RP&M Stereolithography Laser Fusion Lamination Extrusion Ink-Jet Printing The RP&M Systems and the Approach Adopted in our Lab The RP Software 15 Chapter 2. Implementation Background Objectives of the Project CAD Modeling Data Requirement and Representation.18 III

5 2.3.2 Selection of CAD Systems Solid Modeling Slicing CLI File CLI File Structure Header Information Geometry Information Geometric Commands Data Processing General Method Design for Layer Manufacturing Mathematics Expression of Polylines Program design Representation of Points Representation of Contour Data Point Storage Structure Design G-code Flowchart of the RP program One Layer G-code The Sample Part Fabrication Conclusions References IV

6 LIST OF FIGURES Fig.1 Fig.2 Fig.3 Fig.4 Fig.5 Fig.6 Fig.7 Fig.8 Fig.9 Fig.10 Fig.11 Fig.12 Fig.13 Fig.14 Fig.15 Fig.16 Fig.17 Fig.18 Fig.19 Fig.20 Fig.21 Slicing /Stacking Procedure...3 General RP&M Process.6 Stacking Process.6 RP Enabling Technologies.7 Laser Stereolithography Laser Fusion Lamination Extruding Freeform Shapes. 12 Ink-jet printing systems RP Procedures 18 3-D Solid Model.. 22 Debug Content of CLI Debug Content of CLI Content of Data File Content of Data File External and Internal Contour Polines of a Contour.33 Parallel Lines Tool Paths Data Storage Structure.. 40 The Fabricated Part.. 50 V

7 Chapter 1. Rapid Prototyping and Manufacturing Technology 1.1 Introduction During the last decade a new type of technology called Rapid Prototyping and Manufacturing (RP&M) has gained popularity worldwide in manufacturing industry. RP&M is an integrated technology of mechanical engineering, CAD, numerical control, laser technology and material science and technology. It is one of the most important breakthroughs of recent technological developments in product design and manufacturing. This technology has developed at a tremendous rate and brought a profound impact on the conventional technology in manufacturing industry. 1.2 Technology Features of RP&M Rapid Prototyping refers to the fabrication of physical parts in a layer-by-layer manner, directly from three-dimensional CAD models. Traditionally, prototype models or parts have been built by experienced craftsmen using hand tools and/or machine tools. As the CAD technology has matured, emphasis has been focused on improving the fabrication process by a way of controlling the part machining process with data from the CAD model. In essence, efforts have been made to automate the functions of the experienced craftsmen. By adopting Rapid Prototyping (RP) technology, we can build prototype parts quickly and without the need for expensive and hard tools. Using this technology, we can produce parts and tools flexibly without worrying about the part shape complexity. The key idea of this new technology is 1

8 based on slicing a 3-D CAD model into thin cross-sectional layers. Physical layers are then stacked up layer-by-layer until a 3-D part has been fabricated. The main advantages of RP&M technology as compared to the traditional manufacturing methods are: 1.) We can use it to build parts with an arbitrarily complex 3-D geometry. 2.) We can quickly build prototype parts without the need for expensive and special tools. 3.) We can uses a generic fabrication machine, and do not require part-specific fixturing or tooling. 3.) We can make process planning automatic, it requires minimal or no human intervention to operate during the part construction. Another important aspect of RP&M is that, by using RP&M, we can realize the objectives of feature-based design, manufacturability analysis, simulation, computational prototyping, and virtual and physical prototyping. These technologies can not only give us a better product design, improve part quality and dramatically reduce time to the market, but also substantially lower product development costs and greatly improve overall productivity. This is very important for a country that wants to take a lead position in a competitive global economy. 1.3 The Principles of RP&M The basic manufacturing principle of the RP&M technology is slicing and stacking. The constructing of a three dimensional part can be seen as the stacking of twodimensional layers. The information about the layers can be obtained from a three dimensional CAD model by slicing. The manufacturing process involves integrating 2

9 the data with the shaping process parameters to form the machine-controlling codes. The codes are then used to control the material handing devices and tools to form a physical component, from point to line, line to layer and layer to volume. The following chart (Fig.1)[6] shows the basic procedure of the slicing/ stacking process. CAD Modeling Slicing Process Computer Z-direction Slicing Layer Information Processing Layer Forming and Stacking N All layers Completed? Stacking Process RP Machine Y Post processing Fig.1 Slicing/Stacking Procedure 3

10 1.3.1 CAD Modeling The three-dimensional shape of a prototype or a part can be designed using a CAD software package (e.g. AutoCAD, PROE Magic,etc). A CAD model then can be converted into a data pattern, e.g., STL format. Most advanced CAD systems provide the function to realize this kind of conversion. STL stands for Standard Transformation Language. STL files are generated by dividing the surfaces of CAD solid models or curve models. In these STL files, the original CAD models are approximated with polyhedrons which are totally represented by triangular patches. There are only three vertex coordinate values of each triangular patch and its normal vector in a STL file. Because of its simple structure, it is very easy to process STL files. The CAD-generated STL file is a three dimensional representation of a CAD model and is becoming a de facto standard of input data for rapid prototyping system for its simplicity, portability, and easy interpretation Slicing along the Z Direction According to individual specific requirements, the geometry of each prototype or part is divided into layers. The process of using planes which are parallel to the xy plane to intersect a three dimensional CAD model, after STL format conversion, is called slicing. The region between adjacent slices is referred to as a layer. The contour ( geometry information which represents the boundaries of solid material ) of each layer with any Z values can be obtained through slicing. All the layer contour information is stored in a data file which we called CLI (Common Layer Interface) file. CLI file is a universal format for the input of geometry data to rapid prototyping manufacturing system in which layer manufacturing technologies are applied. Slicing 4

11 is one of the key procedures in the whole treatment, the result of slicing has great influence on the next procedure Layer Information Processing In this procedure, the geometrical information in a CLI file will be processed and finally converted into NC code which controls the machine tool to fabricate the layer. This procedure involves a large mount of data processing and data format transformation. The processing result directly influences the surface accuracy and precision of the required part. The quality of designed algorithm for the layer information processing is the key factor and plays a very important role in the part fabrication Layer Forming and Stacking After the intersection information is converted into the NC code, we can use it to control the operation of RP equipment to fabricate every layer of the part. The two dimensional layers will be manufactured in order and connected together. The manufacturing and connection will continue until the three-dimensional object (physical model or prototype) is made Post-processing This process is generally necessary for the prototypes/parts generated by the RP &M system. This process includes model finishing, deep solidification, painting, coloring and cleaning. It is to meet the requirements of quality and precision Simple Graphic Expression of RP&M For better understanding of the RP&M principle, a few pictures are given to illustrate the main processes and technology characteristics of RP&M. They are shown in Figs 2, 3, and 4[4]. 5

12 Fig.2 General Process

13 Fig. 3. Stacking Process 6

14 Fig.4 RP Enabling Technologies 1.5. Methods of RP&M The main methods used in RP&M technology include stereolithography, laser fusion, lamination, extrusion, and ink-jet printing. Each method has its own shaping strategies and deposition/fusion processes. It is very useful and helpful for us to learn these methods. Only through a good understanding of how these methods work can we learn the RP&M technology in depth, carry out our study smoothly and apply the RP&M technology successfully Stereolithography 7

15 A stereolithography RP&M system builds a part by using light to project the photocurable resin and make it solidify to form a layer. This process is repeated until the required shape is constructed. There are two basic approaches, laser stereolithography and photomasking. The laser stereotolithography approach (depicted in Fig.5) is currently the most widely used RP technology. It was first commercialized by the 3D System, Inc. Laser stereolithography can manufacture parts which are made of acrylic or epoxy material. The liquid photocurable polymer is put in a vat and selectively solidified by a scanning laser beam. There is an elevator platform on which the part is built up. The solidified layer is lowered into the vat through a distance of the layer thickness by the platform incrementally. To build each layer, a laser beam is controlled across the surface to draw a cross-sectional pattern in the x-y plane to form a solid layer. The platform is guided to lower into the vat and then the laser beam begins to draw the next layer. Each drawn layer is put on the previous layer. These procedures are repeated until the complete part is built up. Because the photopolymers are relatively viscous, if we simply lower the elevator by a small distance of the layer thickness down into the vat, we can not gurantee the liquid to uniformly recoat the upper surface of the part in a timely fashion. To overcome this problem, we need a recoating mechanism. In this mechanism, the elevator is first lowered several millimeters so that the liquid entirely flows over the current upper surface of the part. The elevator is then raised to the desired height and a wiper arm traverses the surface to quickly level the excess viscous material. 8

16 Fig.5 Laser Stereolithography[4] Laser Fusion Due to the variety of material, the realization method of RP technology is greatly different. In the laser fusion method, the laser is used to selectively fuse powdered material. The new fused material is adhered to the last one in a layer-by-layer manner, the required shape is built up in the end. The "selective laser sintering" approach is depicted in Fig.6. In this illustrated working procedure, a layer of powdered material is spread out and leveled over the top surface of the growing shape. A CO2 laser beam is guided to scan and fuse the powdered material in those areas defined by the geometry of the cross-section. We can apply a variety of fusion mechanisms to join the powder, such as melting, surface bonding, sintering aid, polymer coating, etc. 9

17 Fig.6 Laser Fusion[4] The material that is not fused remains there and can be used as the support structure. After each layer is deposited, the formed shape is lowered down a layer thickness distance by the elevator platform, and then the next layer of powder is spread out. When the required shape is completely constructed, the part can be separated from the loose supporting structure. There are several types of material that can be used in this method. These materials are plastics, waxes, low-melting-temperature metal alloys, polymer coated metals and ceramics. Many parts of complex shapes have been made using the laser fusion method. 10

18 1.5.3 Lamination This method is called Laminated Object Manufacturing(LOM). LOM builds each layer with paper or plastic (see Fig. 7a). As its name implies, the materials themselves are laminated. All sheets of paper or plastic are pre-coated with a thermally activated adhesive and are glued to previous layer with a heated roller. Each layer outline of the part cross-section is cut by a laser beam. Due to the fact that only the layer contour is scanned, LOM spends a relatively small amount of time to build up a part compared to other RP methods. This is especially true when large parts are built. Fig.7 Lamination systems[4] 11

19 1.5.4 Extrusion This approach is called Fused Deposition Modeling (FDM). The working mechanism of FDM can be summarized as extrution freeform shaping. In FDM, a continuous filament of a solid thermoplastic polymer, wax or mental is delivered as a wire into the extrusion head and heated into a liquid state. The liquid material is extruded through a resistively heated nozzle (Fig. 8, top left) and deposited on the platform. Each layer of the part is formed by a way of relative movement between nozzle and platform. This relative motions is controlled by a computer based on the part geometry. The material is to be heated slightly above its flow point, so the extruded material solidifies relatively quickly(only about a few seconds) after it exits the nozzle. Fig.8 Extruding freeform shapes[4] 12

20 1.5.5 Ink-Jet Printing This method has taken advantage of ink-jet printing technology to build layers of a 3-D shape (see Fig. 9.a). It is also called Three-dimensional Printing (3DP). In 3DP, different powdered materials can be used to build parts. The part is constructed in a bin which is fitted with a piston. During manufacturing, the powdered material is provided from a hopper which is above the bin and the powder is spread and leveled by a roller. After that, an ink-jet printing head scans the powder surface Fig.9 Ink-jet printing systems[4] 13

21 and selectively injects a binder into the powder surface area which is defined by the geometry of the cross-section. The binder joins the powder together to form the layer. After the layer is formed, the piston incrementally lowers the shape into the bin and the powders which are not bound become the support material. When the part is completely built up, the "green" structure is fired. The part is then removed from the unbound powder The RP&M Systems and the Approach Adopted in Our Lab From the introduction above, we could know that each method fabricates a part in a layer-by-layer manner. Since the selected materials are different, the manufacturing system, shaping process and software are also totally different. In our lab, the RP&M system consists of a FLASHCUT CNC, which is a powerful CNC control system for Microsoft Windows operating system, a CNC Mini-Mill, which is a desktop Manufacturing System and can be controlled by the G-code to perform milling, drilling and boring. The spindle of the CNC Mini-Mill is capable of moving materials in relation to the spindle in a 3-dimensional and simultaneous motion. These machines are all controlled by a PC-based computer. The RP&M approach we adopted is similar to the Fused Deposition Modeling (FDM). But we use fiber reinforced composites as the material to build parts. The material of the filament is made of fiber and polymer. Because we use a new type of materials, the shaping process has its own distinct features. For example, during the fabrication, the material supply system does not need a special device to provide the propelling force. The obvious and direct benefit of it is the product cost could be decreased. The filament is delivered through a nozzle and heated simultaneously. 14

22 After the filament towpreg heated, the resin on the filament will be melted, but the fiber in the filament will not. When the filament comes out of the nozzle, we make the resin solidify quickly, so the filament could be pasted and solidified on the platform. The relative movement between the nozzle and platform draws the filament continuously to move and the filament is deposited on the platform to form the layers of the shape. The most important thing is how to control the movement between the nozzle and the platform to form the required layer. The relative movement between them is not arbitrary. It is controlled by a CNC code, and the CNC code should be automatically generated by a specific RP software based on the geometry of the part CAD model. 1.7 The RP Software At present, all the companies that could offer the commercialized RP technology and equipments are all developing their own RP software. However, these software are designed especially for specific equipments, materials and processes. There is no any established software for our new RP process, so a new RP software for fiber reinforced composite part fabricating should be developed in this project, and applied in our RP manufacturing system. 15

23 Chapter 2. The implementation 2.1 Background Fiber reinforced composite has great stiffness, high strength, damage tolerance, fatigue resistance, corrosion resistance and other superior properties. However, currently available RP&M technologies can not build up parts with continuous fiber composite. We have discovered that Fused Deposition Modeling (FDM) could be modified and integrated with textile structure-forming operations to produce parts based on an essentially layer-by-layer manner. The parts made of continuous fiber reinforced composite with RP&M technology will also have superior structural integrity. We are now trying some new processes to build up certain composite parts. If the new processes are proved to be successful, they would likely become competitive, preferred or possibly the only approaches to fabricate some composite parts with complex geometric shape. FDM and Shape Deposition Manufacturing(SDM) are two relatively flexible and versatile approaches. But both FDM and SDM typically involve building a prototype part with a neat material, such as a resin or metal without any fiber reinforcement. No prior art has been reported on building continuous fiber reinforced parts using a FDM- or SDMbased technology. In general, currently available RP&M technologies are not involved in fabricating continuous fiber composite parts and current composite processing techniques are not capable of producing parts of a complex geometry or producing parts of a specified geometry directly from a CAD three dimensional model. Accordingly, it is desirable to develop a new process that can be used to fabricate continuous fiber 16

24 reinforced composite parts of high structural integrity and complex geometry. It is further desirable that the process also has the capability of producing a three-dimensional object automatically in response to the computer-aided design of the object. 2.2 Objective of the Project In our new RP process, the main procedures are : CAD Modeling 3-D Part Slicing CLI Data Processing A RP software automatically generates G-code. The generated G-code controls CNC machine to fabricate a part. Fig10. RP Procedures The CAD modeling and 3-D part slicing are done via a selected CAD software system. The other tasks will be implemented in this project. Therefore, the objective of this project are: 1.) Design and implement a RP software which processes the sliced CLI data and generates the G-code. 2.) Use the generated G-code to fabricate the sample part, and test the RP software. This project is a new attempt. There is no other RP software to utilize or learn from. As well known, it is not a once-for-all task to develop this RP software. It might involve a lot of complicated issues, which are not limited to programming- related problems. For example, it involves the problems of material, mechanics, apparatus, NC system, process plan and control, and so on. However, we believe, as time progresses, we can achieve the objective after a series of experiments and hard work. 17

25 2.3 CAD Modeling The first step in the RP&M process is virtually identical for all systems, it involves the generation of three-dimensional computer-aided design model of a part. This is done by a CAD system. Before a specific CAD system is selected and a part 3-D CAD model is generated, the requirements and representation of the part geometry data should be known. These requirements and representation must be guaranteed by the selected CAD system Data Requirement and Representation When a part is going to be built, a natural question is: "How should the part that we desire to construct be represented?". Even with a perfect RP&M system, the shape of the part can only be as accurate as the data represents. So the quality and accuracy of the input data are every important. RP&M systems are highly dependent on their electronic input data. Simply speaking, they are three-dimensional duplicating machines. These systems take an electronic description of a three-dimensional part and reproduce the description into a solid part. If the description is inadequate, the generated part will also be inadequate. The geometric descriptions required for current RP&M systems are provided by CAD systems. They must guarantee the nonambiguous data descriptions of geometry of the part. The model geometric data must facilitate the generation of closed paths and differentiation between "inside" and "outside" of the part. Representation methods used to describe CAD geometry vary from one system to another. Therefore, a standard interface to convey geometric descriptions from various CAD packages to rapid prototyping systems is needed. This interface is the STL format. Currently, the STL files have become the de facto standard for data input into all types 18

26 of RP&M systems. After the CAD solid model is generated, the CAD system should be able to transform the geometry data of 3-D model into the STL format. In this project, the STL format of the CAD model also needs to be sliced and the data of each sliced layer is stored in a file with the CLI format. The final CAD-Generated CLI file is input of the RP software that should be designed Selection of CAD System RP&M systems receive their data from the CAD system. Because the output of a RP&M system is a specific part, the best representation of the part comes from those CAD systems that utilize three dimensional solid modeling approaches. Considering the data requirements mentioned in section 2.3.1, the Magics RP 4.31 CAD system was finally selected to model a solid object. Magics RP 4.31 CAD system can not only represent a part with three-dimensional sold modeling method, but also automatically generate machine ready STL files and slice the STL files into a CLI file. Rapid prototyping begins with STL files to produce the prototypes or parts. Although most 3- D CAD systems can output STL files, there are frequently many problems in the STL file. For example, if a certain vertex is a common vertex of N triangular patches, it will be recorded N times. As a result, this may cause verbose data. When there are many triangular patches, the STL file may be too large and occupy too much resource of the computer. It will also greatly increase the time spent in data processing. STL file is generated through dividing the surface of the original CAD models into many small triangular patches. An accurate STL file must ensure that every triangular patch abutment with only other three triangular patches. Furthermore, triangular patches are only permitted to intersect at their common vertexes and edges. However, due to some 19

27 reasons of CAD models or factitious considerations, errors are often brought about when STL files are generated. These are mainly caused by the loss of topological information. This issue will greatly influence the accuracy of geometry description. With Magics RP 4.31, these corrupted STL files can be fixed in a simple, fast and effective way. Magics RP 4.31 can automatically fix some errors and the remaining errors can be manually edited. Also with Magics RP 4.31 slice preview function, if the contour fixer closes gaps in contours and if the slice output is correct can be checked out. Under certain circumstance, it is very advantageous to modify the design directly from the STL information. This can save a lot of time compared to adapting a design from the original CAD data, because data conversion problems can be avoided Solid Modeling After a CAD system was selected, the solid model of the experimental part can be designed with it. As its name implies, solid modeling is a method that represents an object as a solid part. These parts are typically shown as some shaded images on the computer screen. Solid modeling is an advanced design technology that offers numerous advantages over traditional design methods. The use of solid modeling has resulted in many benefits for our design. It improves our visual and analytic capability and provides required inputs for rapid prototyping system. It allows us to visualize and understand the part's actual appearance and function within an assembly by clearly displaying the critical relationship of part layouts, interference, and clearances. It helps us detect the design flaws before hardware fabrication. It increases our ability to calculate mass properties. It also helps us realize design optimization, and decrease the cost and cycle time associated with prototype part fabrication. With the help of Magics RP system, an experimental 3-D 20

28 solid model was designed (Fig.11). Since the project is now in experimental stage, the 3D model should not be designed too complicated or too simple. It must possess the general characteristics of a part. Therefore the designed software will fit the requirements for general usage. Fig. 11 A Part 3-D Solid Model 21

29 2.3.4 Slicing When the part 3-D model design was finished, the part 3-D model STL-information was sliced by the Magic slicing program. If there were open contours, Magic's contour fixer automatically would close or fixe them. The slice program converted 3-D model STL information into two-dimensional cross section data and stored this data into a CLI file. These cross sections can be sliced from any of the CAD x-,y-,or z-axis. The slice axis, by definition, is then perpendicular to the planes created by these cross sections. In this project, Z-axis was used as the slice axis. When the CAD model was sliced, a layer thickness (for example 0.5 mm) should be assigned to the slice program. These cross sections were derived at increments of this layer thickness during the slicing process. In actual production, the part layer thickness was generated by the stepping of the elevator in the same increments. The RP&M system converted the two-dimensional cross sections into the three dimensional layers of the tangible part. After slicing, it was found that the contents of the generated CLI file consisted of ASCII data and binary geometry data, the geometry data includes each layer information, contour information, point coordinates of the polyline on each layer. As shown in Fig.12, and 13. In order to get each required layer data, and make it easy for next step, the binary geometry data must be converted into ASCII data. Because the quantity of geometry data in CLI file was very large and stored in binary format, in order to guarantee the conversion of the binary data to ASCII data to be correct, the CLI file structure and requirements should be known first. 22

30 Fig.12 Debug the Content of CLI Fig.13 Debug the Content of CLI 23

31 2.4. CLI File CLI File Structure In a CLI file, the contents are divided into two sections, a HEAD-section in ASCII data format and a GEOMETRY-section in binary data format. A start and an end markers mark each section. Various markers mark the CLI file structure. 1. $$HEADERSTART: It starts the HEADER-section. 2. $$HEADEREND: It ends the HEADER-section. 3. $$GEOMETRYSTART: It starts the GEOMETRY-section 4. $$ GEOMETRYEND: It ends the GEOMETRY-section and is interpreted as the end of data Header Information 1.$$BINARY: It indicates the data in the GEOMETRY-section is binary 2.$$UNITS/u: Parameter u is REAL, it indicates the unit of the coordinate in mm. 3.$$VERSION/v: v divided by 100 gives the version number. 4.$$DATE/d: Parameter d is interpreted in the sequence DDMMYY. 4.$$LABEL/id,text: id indicates the number of models in one file, text is an ASCII string that gives some comments on the part Geometry Information It uses commands to describe each layer information. All commands have the following general form: Command Index, p1, p2,..pn. There is no separator between Command 24

32 Index and parameters, nor within the parameter-section. Command Index(CI) is a number(unsigned integer) indicating the command according to the command list shown in Table 1. Table 1. Binary CLI File Command Index CI Command Format Parameter 127 Long CI, Z Z: Real Layer 128 Short CI, Z Z: Unsigned Integer 129 Short CI,id,dir,n,p1x, p1y pnx, pny id, dir, n, p1x, p1y.pnx, pny : Unsigned Integer Polyline Long id, dir, n : Integer CI,id,dir,n,p1x, p1y pnx, pny Short CI,id,n,p1x, p1y pnx, pny p1x,p1y.pnx,pny : Real id, dir, n, p1x, p1y.pnx, pny : Unsigned Integer Hatch Long id, n : Long Integer 132 CI,id,n,p1x, p1y pnx, pny p1x,p1y.pnx,pny : Real 25

33 Geometric Commands 1.) CI 127, z This command marks the start of a layer with upper surface at height (z*units[mm]). All layers must be sorted in ascending order with respect to z. Thickness of the layer is given by the difference between z values of the current and previous layers. A thickness for the first(lowest) layer can be specified by including a "zero-layer" with a given z value but no polyline. Parameter z : REAL. 2.) CI 128, z This command marks the start of a layer with upper surface at height(z*units[mm]). All layers must be sorted in ascending order with respect to z. Parameter z: unsigned INTEGER. 3.) CI 129, id, dir, p1 pn. This command marks the start of a polyline. Parameters: id : unsigned INTEGER. It is an identifier to allow more than one model information in one file. dir : It shows the orientation of the polyline when viewing in the negative z direction. 0:clockwise 26

34 1:counter clockwise 2:open line n: unsigned INTEGER. It tells the number of points. p1x pny: Coordinates of points 1 n. 4.) CI 130, id, dir, n, p1 pn. This command marks the start of a polyline. Parameters: id: Long INTEGER. It is an identifier to allow more than one model information in one file. dir: Long INTEGER. It shows the orientation of the polyline when viewing in the negative z-direction. 0: clockwise 1:counter clockwise 2:open line n : p1x pny: Long INTEGER. It tells the number of points. REAL. Coordinates of points 1..n. 5.) CI 131, id, n, p1x, p1y, pnx, pny. This command marks the start of the hatches. Parameters: id: unsigned INTEGER. It is an identifier to allow more than one model information in one file. 27

35 n : unsigned INTEGER. It tells the number of the hatches(n*4 = number of coordinates). p1sx..pney: unsigned INTEGER. Coordinates of hatches 1..n 6.) CI 132, id, n, p1x, p1y, pnx, pny. This command marks the start of the hatches. Parameters: id: Long INTEGER. It is an identifier to allow more than one model information in one file. n : Long INTEGER. It tells the number of the hatches(n*4 = number of coordinates). p1sx..pney: REAL. Coordinates of hatches 1..n. 28

36 2.5 Data Processing Having known the sliced CLI file format, a program was written in C language. When executing it, the data in the sliced CLI file can be read by the program according to the value of Command Index. Binary data to ASCII data translation was the first task of the RP software design. During the conversion, a large number of data would be dealt with. It was not a simple one to one data conversion. It required not only to translate the data correctly, but also arrange them into a kind of format by which geometric information and point coordinates of the polyline on each layer can be easily identified, obtained and processed. To satisfy this requirement, the geometric information of each layer has been placed at the head of file. "x=" and "y=" symbols were used to distinguish x and y values of a point and let each point coordinate value account for one line in the data file. This arrangement was easy for reading information about each point on the polyline and processing these data points. For the beginning of the file are ASSCII characters, the first part information was read in the form of character. After this work was done, the space, linefeed character and end of line character must be skipped, so the exact beginning position of binary geometric information could be determined. If one bit was missed, the whole data would not be correct. The contents of binary CLI file were debugged as shown in Figs.12 and 13. The data length (bytes) of each kind of data was also calculated, if the data is of INTEGER, 4 bytes are read at a time. If it is a REAL or Float, 4 bytes are read at a time. If it is an unsigned short, 4 bytes are read at a time. For each character data, 1 byte is read at a time. After the translation, part of the layer data is shown in Figs.14 and 15. In Fig.14, it is found that n=737, which means at this layer (z=0), there are 737 points. If the total number of layers of the part is 1000, then there are totally 737x2x1000 = values. 29

37 How to use this data to fabricate each layer, How to store this large number of values effectively and search them correctly are big problems.the key factor is to design a general algorithm for layer manufacturing and a good data structure that can not only represent this large amount of data and allow easy access, but also require a minimum memory. Fig.14 Content of the Data File 30

38 Fig.15 Content of the Data File 31

39 2.6. General Method Design for Layer Manufacturing One obvious characteristic of traditional manufacturing method is cutting some material from the original ingot to get the required shape. However, the method adopted in RP manufacturing is stacking the material in layer-by-layer manner to form the required shape. So, there arises the question, every part has their unique shape, how is each layer formed and in what way? The requirement for the RP software is that it should be applied to fabricate general part. Therefore, a general method must be designed based on the generated CLI layer information. After slicing and data conversion, layer geometric information of the part was obtained. This information included data point coordinates of each layer polyline and contour property parameters. Contours represent the boundaries of a solid model within a layer and are defined by polylines. They are classified into external contours and internal contours (see Fig.16 ). For correct interpretation, each contour must be closed and must not intersect itself or with another contour. This requirement is guaranteed by the Magic RP 4.31 system. The relations among contour, polyline and data point coordinates are: each contour consists of polylines, each polyline consists of data points and these points had some relationship. But it is not enough to have this information, this information only can be used to trace out the outline of each layer. The solid part still can not be fabricated! For example, given one layer data, the points on the external and internal contour have been obtained, see Fig.17. These points are not enough to form a required layer. As we know, a line consists of points, a plane consists of lines, and a volume consists of planes. If we can make use of these obtained points to generate some other points that can be used to generate some lines to form the required layer then 32

40 External contour Internal contours Fig.16 External and Internal Contour C A B D E L K F G J H I Fig. 17 Polylines of A Contour 33

41 we can build up the part in layer-by-layer manner. For example, if a circular plane is expected to be generated, it could be realized through the way shown in Fig.18 y x Fig.18 Parallel Lines In this manner, a number of parallel lines (parallel to x-direction) are used to stack upon along some direction (y-direction) until the required layer shape is formed. But there go the questions, how could these parallel lines be generated? What are their mathematics expressions? If the geometric shape of a layer is complicated, for example, there might be different internal contours on the plane. What order should be used to stack them to guarantee the correct geometric shape? The basic ideas are: 1. Several special lines that are parallel to the X-axis could be used to intersect the external and internal polylines (segment on the contours) and to get the intersection points. 2. These lines are supposed to have certain width. 3. The increment value ( Y) along the Y-axis direction is taken as the line width. The chosen value of Y is very important, it would directly influence the shape and surface precision of the part. Its value range is also limited by the equipment and the composite fiber. 34

42 4. These lines are stacked along some direction (Y-direction) to form a plane. The starting point is the point whose y-coordinate value is the minimum. The end point is the point whose y-coordinate value is the maximum. 2.7 Mathematics Expression of Polylines There are 3 kinds of polyline in a contour. They are vertical polyline, parallel polyline and general polyline. According to mathematics principle, given any two points, a line can be defined. Suppose the first given point of a polyline is (Xs,Ys), and the second is (Xe,Ye), (see Fig.19). The slope, k, of a polyline can be calculated as: k=(ys-ye)/xs-xe (1) The polyline equation can be expressed as: Y-Ys=k(X-Xs) (2) For a vertical polyline, the equation would be: X=Xs (3) For a parallel polyline, the equation would be: Y=Ys (4) The parallel lines that are parallel to the x-axis can be expressed as: Y=Ymin + N* Y (5) Where: N: 0~(Ymax-Ymin)/ Y 2.8 Tool Path Generation Tool paths are those lines which lead a machine tool to trace out a layer geometry of the required part (see Fig.19). From the analysis in section 2.6, we know that the tool paths are those lines defined by intersect points of parallel lines and ploylines. Before intersect points are computed, whether a polyline intersect a parallel line or not should be judged 36

43 first. The intersect condition is the y-coordinate value of a parallel line lies between the y- coordinate values of the start point and the end point of a polyline. That means if a parallel line intersects polylines then the condition: Y-start <= Y-parallel <= Y-end, or Y- end <= Y-parallel <=Y-start is true. So the intersect points between a parallel line and polylines can be computed according to the equations (2), (3), (4) or (5). Y (Xe,Ye) Y-max (Xs,Ys) Y Y Y-min O Fig.19 Tool Paths 36

44 Program Design So far, the general layer manufacturing method design and tool paths generation have been completed. Next, it is the task to implement the RP software based on the ideas above. The software development environment is the Microsoft Visual C studio, the language adopted is C. Main topics are: data point representation, contour representation, point storage data structure design, program flowchart and G-code generation Representation of Points The basic unit which will be dealt with is the data point. How to represent these points, their relationship and corresponding contours they belong to is very important in the program design. The best way is to use the composite data type to define a point, this composite data type should include a point's x-coordinate, y-coordinates and some information about its next point. C language provides this kind of data type 'structure'. So the structure data type is used to represent a data point. The structure data type is defined as follow: struct point{ unsigned short x; unsigned short y; struct point *next; } Representation of Contours When the data type for a contour is defined, it is hoped that it could not only distinguish external and internal contours, but also specify the relationship between 37

45 a contour and its points. So an integer variable "dir" is used to identify each layer contour. When dir=1, it is an external contour, dir=0, it is an internal contour. The data type of contour is also composite. It includes one integer variable (dir) and two pointer variables: "nextc" and "nextp". "nextc" points to contour structure data type and "nextp" points to point structure data type. The current contour pointer variable (nextc) points to the next contour that is adjacent to it. The current point pointer variable points to the next point that is on the same contour. By this way, the relation among contours and points is set up. The contour data type is defined as follow: struct contour{ int dir; struct contour *nextc; struct point *nextp; } Data Point Storage Structure Design The number of points on the whole contour of a part is very big. Consider the generated tool path points, the total number of points is even larger. If they are not arranged well, they can not be identified or accessed correctly, and processed effectively. Also, different parts may have different shapes. This means the shapes of contours and the number of data points vary from part to part. Therefore, a data structure must be designed and implemented. It must not only dynamically store the randomly generated data points, but also clearly provide the relationship among the layer, contour, and points. On the other hand, the space of computer memory is limited. When the data points are stored, it must be guaranteed that all data points could be accessed and the capacity of memory is not exceeded. 38

46 Consider the fact that the processed data points will not be used any more, they should not be kept all the time. The whole data can be processed in the unit of a layer. After the data of a layer has been processed, the occupied memory should be released for the next layer. By this way, the required maximum memory is no more than the number of the points on a layer. The dynamically generated points could be stored in a dynamic linkedlist structure. After some study, a dynamic linked-list was designed to store the random generated data. Its structure is shown in Fig G-Code G-Code is a standard Numeral Control language. Most CNC machine supports this language. It provides some commands which can be used to control the machine tool movements. The main commands are: G00 G01 G02 G03 G04 G17 G18 G19 G20 G21 G28 G29 Rapid Tool Positioning Linear Interpolated Cutting Move Clockwise Circular Cutting Move (XY Plane) Counter Clockwise Circular Cutting Move (XY Plane) Dwell XY Plane Selection XZ Plane Selection YZ Plane Selection Inch Units Metric Units Return to Reference Point Return from Reference Point 39

47 Dynamically Generated Contour 1 Contour 2 Contour 3 Contour n Point 1 Point 1 Point 1 Point 1 Point 2 Point 2 Point 2 Point 2 Dynamically Generated Point 3 Point 4 Point 3 Point 4 Point 3 Point 4 Point 3 Point 4 Point n Point n Point n Point n Fig.20 Data Structure 40

48 G43 G44 G70 G71 G90 G91 M00 M02 M06 M30 M98 M99 F Tool Length Compensation (Plus) Tool Length Compensation (Minus) Inch Units Metric Unit Absolute Positioning Mode Incremental Positioning Mode Program pause End of Program Tool Change End of Program (Reset) Subroutine Call Return From Subroutine Federate The final output results of the RP software are the G-code commands combined with the coordinates of the tool path point. Because the layers are fabricated in term of lines, so the most used commands are G00, G01, and F.

49 2.9.5 Flowchart of the RP software Start Read data from CLI file, generate ASCII data file ASCII Data processing 1 End of file?? N Y Stop? Read a layer data 2 Generate a contour node and put the geometry information of the contour into the node. Add the node to the dynamic linked-list. 3 Generate a point node and put the coordinate value into the node. Add the point node to the contour node. 42

50 N End of point? Y N End of layer? Y Calculate Ymin and Ymax Y=Ymin 4 Y=Y+ Y 5 Get a contour 6 Get the point on the contour 43

51 N If a parallel line intersects a polyline? Y Calculate the tool path data point 6 N Search all points on contour? Y 5 N Search all contour? 7 Y Read tool path data Point Logic analysis of the position between two adjacent points. 44

52 First point dir = 1? second point dir = 0 Y Generate G-Code: G01 X1,Y1,X2,Y2 N First point dir = 0? second point dir = 0? N Generate G-Code: G01 X1,Y1,X2,Y2 Y Both points on Ythe same contour? N Generate G-Code: G01 X1,Y1,X2,Y2 Y Generate G-Code: G00 X1Y1,X2Y2 7 N End of the tool path pata point? Y 1 45

53 2.9.6 One Layer G-code G00 X Y Z F3.0 G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y

54 G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y

55 G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y G01 X Y

56 2.9.7 The Sample Part Fabrication In our lab, a CNC Mini-Mill/2 desktop manufacturing machine is used to fabricate a part. It is controlled by a FlashCutCNC software, which is a powerful and Windows-based program. With FlashCutCNC, the following functions are provided: 1. Visualize and verify the tool path generated from a G-code file. 2. Watch the current position of the machine tool as it moves. 3. See the current position of the machine tool in either machine, program, relative,or distance-to-go coordinates. 4. Create, edit and display a G-code program. 5. Move the machine tool in any four modes: Jog,Point, G-code, or home. To see if the RP software is correct, the G-code file is read into the FlashCutCNC to control the machine tool to fabricate the designed 3-D part. The fabricated sample part also can be viewed under the FlashCutCNC simulation environment (see Fig.21). The final result showed that it can build the required 3-D part. However, the software still needs improvement and optimization, Nevertheless, the main function has been realized! 49

57 Fig.21 the Fabricated Part 50