Virtual machining chapter manufacturing technology handbook

Metadata of the chapter that will be visualized online
Chapter Title Virtual Machining
Copyright Year 2014
Copyright Holder Springer-Verlag London
Corresponding Author Family Name Liu
Particle
Given Name Peiling
Suffix
Organization Singapore Institute of Manufacturing
Technology
Address 71 Nanyang Drive, Singapore,
638075, Singapore
Phone (+65) 6793-8356
Fax (65) 6793-8356
Email plliu@SIMTech.a-star.edu.sg
URL http://www.simtech.a-star.edu.sg/
Author Family Name Zhu
Particle
Given Name Cheng-Feng
Suffix
Organization Singapore Institute of Manufacturing
Technology
Address 71 Nanyang Drive, Singapore,
638075, Singapore
Phone (65) 6793-8336
Fax (65) 6791-6377
Email cfzhu@simtech.a-star.edu.sg
Abstract Virtual machining simulates NC code to discover errors, without a
time consuming trial run or online debugging on real machine tool.
Since machining is a material removal process that will deform the
workpiece geometry with cutting, the traditional rigid geometrical model
could not be used to describe the in-process status of workpiece, which
changes shape continually. The evolution of deformable workpiece
model from the 2D sections to 3D representations revolutionized not
only the machining industry, but also pioneered the digital manufacturing
age with virtual manufacturing. This chapter traces back the history of
CNC simulation, analysis of the different CNC machining models, tested
with application examples, and lists different CNC verification industry
applications for the last 30 years. Working towards a vision of pervasive
modelling and simulation, a unified voxel-based in-process geometry
model for multiple-machining and 3D printing simulations is discussed

with industrial applications of composite material plating simulation.
The virtual machine tool, which includes material removal animation
and machine kinetic movement, can be controlled with a virtual CNC
control panel and equipped with virtual jigs and inspection tools, such as
dial indicator and wiggler, for immersive training of a young machinist.
Towards a competitive sustainable manufacturing future, pervasive
applications of virtual machining are not only technologically possible,
but also make business sense, in this high material and energy cost world.

Virtual Machining
Q1 Peiling Liu* and Cheng-Feng Zhu
Singapore Institute of Manufacturing Technology, Singapore, Singapore
Abstract
Virtual machining simulates NC code to discover errors, without a time consuming trial run or online
debugging on real machine tool. Since machining is a material removal process that will deform the
workpiece geometry with cutting, the traditional rigid geometrical model could not be used to
describe the in-process status of workpiece, which changes shape continually. The evolution of
deformable workpiece model from the 2D sections to 3D representations revolutionized not only the
machining industry, but also pioneered the digital manufacturing age with virtual manufacturing.
This chapter traces back the history of CNC simulation, analysis of the different CNC machining
models, tested with application examples, and lists different CNC verification industry applications
for the last 30 years. Working towards a vision of pervasive modelling and simulation, a unified
voxel-based in-process geometry model for multiple-machining and 3D printing simulations is
discussed with industrial applications of composite material plating simulation. The virtual machine
tool, which includes material removal animation and machine kinetic movement, can be controlled
with a virtual CNC control panel and equipped with virtual jigs and inspection tools, such as dial
indicator and wiggler, for immersive training of a young machinist. Towards a competitive sustain-
able manufacturing future, pervasive applications of virtual machining are not only technologically
possible, but also make business sense, in this high material and energy cost world.
Introduction
Virtual manufacturing is a new and emerging concept to integrate different areas of manufacturing
by using computer technology for creation and execution of virtual models. Virtual manufacturing is
defined as a computer-based system, which consists of evolving models of manufacturing systems
and processes, and is exercised to enhance one or more attributes of the real system. Manufacturing
as a whole is a very complex system consisting of various interacting, interrelated, and
interdependent subsystems and processes. Virtual machining, a small building block in the com-
prehensive virtual manufacturing system invented in the 1960s, pioneered virtual manufacturing
with material removal process visualization long before its coronation in the 1990s.
Machining had been a low productivity manual operation until the invention of numerical control
(NC) in the 1950s, when the hand wheels and levers were replaced by punch tapes control, similar to
telegraphs at that time. These early servomechanisms were rapidly augmented with computers since
the 1960s, the computer numerical control (CNC) machine tools have revolutionized the machining
process and radically changed the manufacturing industry. Complex 3D shapes are relatively as easy
to cut as the plane face, and manual polishing works have been dramatically reduced.
*Email: plliu@SIMTech.a-star.edu.sg
Handbook of Manufacturing Engineering and Technology
DOI 10.1007/978-1-4471-4976-7_16-1
# Springer-Verlag London 2014
Page 1 of 54

In 1958 MIT published its report on the economics of NC. They concluded that the tools were
competitive with human operators, but simply moved the time from the machining to the creation of
the tapes. NC programming became a bottle neck in machining. Automatically programmed tool
(APT) language was developed to generate instructions for NC control during the late 1950s and
early 1960s. It was widely used into the 1970s and is still a standard nowadays. Since APT was
created before graphical user interfaces (GUI) and computer graphics (CG) were available, it relies
on text to specify the geometry and process. Again, this is a highly skilled manual script writing that
slowed NC machining, especially for high volume low mix parts, where a lot of new NC programs
were needed. Computer aided manufacturing (CAM) was developed to quicken and automate this
process, where the shape could be defined by computer aided design (CAD) software. The
CAM/CAD system have proliferated CNC machine tools since the 1980s.
An NC program has thousands of lines of tool movement instructions that may contain errors.
Following these instructions, a CNC machine tool will move blindly, without any check on gauging,
overcut or cutting force. It is impossible to verify code manually, thus virtual machining can virtually
trial run NC code in a computer, to verify NC code and replace trial cut that is time consuming and
dangerous.
Virtual machining, visualization of material removal of various machining processes, is
a geometrical modelling process that realistically simulates the setting up and running of an actual
machining operation. First, the user specifies the stock from which the part will be cut, either by
entering dimensions into the software or importing a CAD model. Then, after parsing NC code with
the selected cutter, NC toolpath backplot can display tooltip trace against design model, highlighting
grammar error and errant movement. NC simulation automatically simulates the motion of the tool
removing material from the stock. The programmer can watch the material removal process and see
details of how each cut changes the shape of the part, which is a deformable in-process geometrical
model. This eliminates having to try to imagine how cuts from the current operation will affect
subsequent operations, which will help to plan for the next operation. NC analysis will compare this
in-process model against the target design and display the remaining stock with a coloured map and
report. Since the NC programming error was haunting machinist from the beginning, this was also
called NC verification, a short name for machine tool numerical control code verification.
Since machining is a material removal process that will deform the workpiece geometry with
cutting, the conventional CAD geometrical model cannot be used to describe the in-process status of
workpiece which changes shape continually. The evolution of deformable workpiece model from
the 2D sections to 3D representations revolutionized not only the machining industry, but also
pioneered digital manufacturing age with virtual manufacturing. Various in-process geometrical
models and their applications are discussed in this chapter. Virtual machine tool, which includes
material removal animation and machine kinetic movement, can be controlled with a virtual CNC
control panel and equipped with virtual jigs and inspection tools, such as dial indicator and wiggler,
for immersive training of a young machinist.
Recently it has expanded from geometrical modelling into process modelling, including the
machine dynamic and FEM cutting simulation, with in-process geometrical model as its foundation.
In geometric modelling, cutter-workpiece engagements are extracted to support force prediction in
process modelling. In process modelling, the physics of the machining process, such as cutting
forces, torque and power, are predicted by integrating the laws of the metal cutting process. Based on
these predictions, process parameters can be optimized for productivity. Methodologies in geometric
modelling for cutter-workpiece engagement extractions require a large number of calculations,
however, the robustness and computational stability of these approaches is a significant challenge,
which will be covered in another two chapters of this handbook.
DOI 10.1007/978-1-4471-4976-7_16-1
Page 2 of 54

Virtual Machining Industry Landscape
Virtual machining has been a vivid academic research theme of virtual manufacturing since the
1990s, together with the low cost PC, internet, virtual reality (VR) and OpenGL. VR has been widely
explored as a means of virtual machining with many academic research prototypes. JAVD 3D is used
to distribute simulation image over the internet. STEP NC simulation is also a part of STEP research.
As there are many review and survey papers on virtual machining academic research available
online (AbdulKadir et al. 2011), this chapter focuses on industrial applications which are commer-
cially available and technologically stable to engineers and machinists.
Virtual machining is so important that the CAD/CAM developers were eager to develop inte-
grated CNC simulation into their system, with great success on native CAD machine tool kinetics
simulation, internal NC toolpath backplot can be displayed as different doted-solid colour lines, and
partial success of a 2D in-process workpiece model, which is always deformable since the geometry
changes with every cutter move. Different from CAD machine frame and toolpath, the 3D in-process
workpiece could not be modelled with conventional CAD B-rep solid modeller, such as ACIS or
ParaSolid. Initially CAD/CAM geometry model was tested to model machining process but failed,
since the in-process geometry of workpiece is deforming but conventional CAD model is
static. How to develop an in-process geometrical model (IPM) that could simulate the deforming
workpiece has been a research challenge, which has attracted great academic research interests since
the 1980s. There is a vivid research theme of a smart machining system (SMS), which builds IPM
into a brain of intelligent machining for CNC machine tool.
Professor Donald Esterling pioneered NC verification, leading the way as a variety of OEM
partners incorporated this technology into their products. N-See™ (Predator VIRTUAL CNC™)
was the first volumetric based solid model NC verification program, raising the bar for speed and
accuracy. While on the engineering faculty of George Washington University, he initiated
a manufacturing program in the mid-1980s funded by a $2 M grant from IBM. Esterling has
received research funding from NASA, NIST, NSF, NATO, US Air Force and the US Army
Research Office. He has been awarded several prestigious and highly competitive Small Business
Innovative Research (SBIR) grants. He has a successful track record of moving research projects
from the lab to commercial applications.
While he worked with McDonnell Douglas Corporation, Occidental Petroleum and MCS as
a CAD/CAM consultant, Jon Prun accumulated abundant experience of computer software, com-
puter graphics, mathematics and digital control technology. Having realized that there was a great
need of digital control simulation technology by manufacturing, he established a company named
CGTech in 1988 to develop a suite of digital control simulation software VERICUT™. One after
another, CGTech partnered with Unigraphics, Dassault Systems and PTC to get support from these
renowned CAD/CAM companies. VERICUT™ was first run in UNIX system computers based on
Sun workstation, and then upgraded to PCs and other workstations such as HP, IBM and DE-
C. Nowadays CGTech has many branches worldwide, and almost all the customers are running
VERICUT on PCs.
There was a time when CAD/CAM components were developed and marketed as an independent
software module, such as ACIS, which was started as object oriented CAM components.
MachineWorks™, which is a spinoff from LightWork UK, pioneered the market for embedded
simulation in the mid-1990s when NC simulation was a “nice-to-have”, by offering the first true
history based solid model simulation, which has been patented worldwide. MachineWorks’ simu-
lation solution allowed CAM software producers to provide integrated simulation as a core
DOI 10.1007/978-1-4471-4976-7_16-1
Page 3 of 54

functionality in their applications. Embedded simulation soon became a “must-have” for main-
stream CAM applications.
However, embedded NC simulation could only discover internal NC toolpath errors, which is in
APT format. This internal NC toolpath is post processed into machine control data (MCD), such as
G/M code. There are much more errors after post processing and human editing, where the
independent machine control code verification package is still a must.
Furthermore, the history based virtual machining will slow down along with the cutting history,
since the method records toolpath into voxel cells so it can zoom without losing resolution. Recently
MachineWorks claimed they have removed history from their model and sped up the simulation
greatly.
ModuleWorks™ from ModuleWorks provides a complete CAD/CAM component solution with
high performance toolpath simulation and NC verification. The toolpath simulation component
supports milling, turning and mill/turn applications with full machine simulation, stock removal
verification and toolpath analysis. ModuleWorks™ toolpath simulation will identify problem areas
such as potential collisions, gouges or over travel and allow correction prior to NC code generation.
It is viewed by many as an essential aspect of the CAM process. NC simulation library provides state
of the art technology and is well proven and in use with many of the leading CAM software solutions
around the world today. The simulation component can be quickly implemented using the easy to
use API. ModuleWorks™ provide full kinematic machine simulation with comprehensive collision
and axis limits checking. A full kinematic machine builder supports mills, lathes, mill/turn, robots
and CMM machines with support for an unlimited number of axes. NC simulation and toolpath
verification component also provides fast, high accuracy verification of stock removal for mill, turn
and combined mill/turn applications. NC simulation also offers a full range of toolpath analysis tools
for many critical aspects of toolpath behaviour such as segment length/type, feedrate and height
allowing toolpath to be refined for optimum finish and quality. NC simulation tools are independent
of the toolpath generation CAD/CAM components and can be used with any toolpath or backplot
NC code.
SIMNC from BinarySpaces produces 3D simulation of complex multi-axis machine tools,
including collision detection, material removal and other non-cutting processes. The SIMNC Core
API, the foundation of the entire product line, is built using the latest software architecture, which
optimizes its use of memory and CPU power – allowing the product to run on-line on a machine tool
control or off-line on a stand-alone PC. This modern architecture supports parallel processing that
maximizes graphics performance on 64-bit and/or multi-core computers. SIMNC includes
a machine tool builder to aid in defining the computer representation of the machine tool. The
SIMNC control emulator allows the simulation engine to run off-line directly from the end user’s
G-code part programs, while the SIMNC part set-up aids the end-user in defining the cutting tools
and fixtures related to each individual CNC program.
Kinetic Simulation of Machine Tool
In advanced virtual machining systems, it is possible to include other elements such as machine
frame, fixtures, clamps and tool holders. These are required for collision detection. In some packages
like VERICUTand NCSIMUL, the entire machine also can be simulated to visualize the kinematics;
this will be particularly very useful during 5-axis machining, where the collision of tool holder and
fixture may destroy machine spindle.
DOI 10.1007/978-1-4471-4976-7_16-1
Page 4 of 54

Machine Tool Model
A machine tooling system is made of rigid parts (frame, jig, fixture and toolholder), wearable cutter,
and deformable workpiece, which is continuously changing and has to be modelled with voxel based
in- process model (IPM). Voxel based IPMs are deformable and expensive in terms of computer
memory, in line with the size of the workpiece. A machine tool frame, including jig and fixture,
could be 10 m long and wide, so it is too expensive to model them in voxel model. During the
machine simulation, only the parts interference is checked with a simple and quick collision check
algorithm, without an expensive material removal calculation.
Generally, the geometrical representation scheme for machine tool is the same as conventional
CAD, so it is possible to simulate the entire machine motions with conventional CAD/CA-
M. Building the machine frame geometrical models, kinetic constrain and control are daunting
tasks that are usually done by CAD/CAM developers. However, the geometrical representation
scheme for material removal is different from conventional CAD, so it is not possible to simulate the
material removal process with conventional CAD/CAM. The material removal process is simulated
outside CAD/CAM with third party applications such as VERICUT. Furthermore, conventional
CAD/CAM systems have their own interactive graphical user interface (GUI) for quick creating and
editing of a geometrical design model and drawing. It is difficult to customize this GUI to manually
operate a virtual CNC machine tool, which functions more like a computer game.
The simulation of material removal on the workpiece and simulation of machine kinematics can
be done in different sessions (CAD/CAM for machine motion and third party applications for
material removal) or simultaneously in a single third party application session, which is more
realistic and good for machinist training.
Since machine tool frame and jigs-fixtures are rigid static geometry, they could be modelled with
conventional CAD and exported as triangular polyhedral mesh by stereo lithography (SLA) format
with file extension of STL, which is quite similar to VRML format. Figure 1 shows the frame model
built up of a 3 axis milling machine, which starts with X axis, Z axis and ends with Y axis.
A comprehensive machine tool system also includes the opaque machine cover, transparent glass
windows, movable doors, operateable handle/lever/pedal, measurement indicators, and virtual CNC
control panel, which is critical for operational training. Figures 2 and 3 depict two models of virtual
machine tools: CNC mill and CNC lathe, that are manually CNC operateable as a real machine.
More than a real machine tool, the virtual machine tool can be zoomed from different angles in
multiple viewports highlighted with translucent colours for enhanced learning effects that are more
like an educational computer game.
Conventional CAD/CAM system is suitable for machine tool simulation with internal NC
toolpath, which is cutter location (CL) data that is generated inside the system, usually in APT
Fig. 1 Virtual machine in X axis, XZ axis and XYZ axis
DOI 10.1007/978-1-4471-4976-7_16-1
Page 5 of 54

format that is an industry standard. However, the post processing converts APT into machine control
data (MCD) with geometry transformation and CNC controller specification, a lot of things could go
wrong at this stage. For example, table plus table and table plus head 5-axis configurations are totally
different. Even with the same configuration, FANUC and Siemens need different codes for com-
pensation. So there is a need to simulate MCD code such as GE FANUC G/M code, Heidenhain G/M
code or Heidenhain conversational code, so the machine tool will move according to real situation.
Reverse post processing will translate MCD back into APT format.
Fig. 2 Virtual CNC mill
Fig. 3 Virtual CNC lathe
DOI 10.1007/978-1-4471-4976-7_16-1
Page 6 of 54

Reverse Post-processing MCD back to APT Toolpath
The reverse post-processor reads, analyses NC program and translates it to internal NC toolpath in
APT format, as described in flowchart Fig. 4. It supports structured programs, variables, cycles and
macro calls for a wide choice of commercially available NC controllers.
Analysis and parsing MCD data, such as G/M code, especially for a manually written program, is
critical to avoid time-consuming debugging on machine tool control panel, when the machine is
running but without production. For most CAM generated G/M code, the grammar error is not
a concern anymore, so it is possible to bypass grammar check and achieve faster NC toolpath
backplot. However, parsing is a critical step for CNC training, where a new trainee may write strange
code and try to run it on machine tool.
CNC control panel is equipped with a keypad for text input. Virtual control panel must parse this
manual input and simulate the action with warning signals. The CNC controller is a fully functional
high level computer language interpreter, so is the virtual CNC controller. There are variables,
formulas, subprogram and mathematic functions within a machine control code script.
Automatically programmed tool (APT) was a high-level computer programming language used to
generate instructions (MCD) for NC machine tool before CAD/CAM revolution. Now CAM
software replaced APT for toolpath generation but still kept it as internal CL data format to express
internal NC toolpath in ASCII text. Most CAM systems can save the internal NC toolpath as APT
format and use a third party post processer (such as ICAM) to generate the machine tool specific
MCD, such as FANUC 16M G/M code. Since APT is the internal NC toolpath format there is no
reverse processing step in reading APT text data file into the internal NC toolpath.
Most virtual machining systems accept cutter path in the form of NC code specific to the CNC
machine or in the form of a generic format APT. If the input is NC code specific to a CNC machine
tool, one would additionally require a machine tool data file that provides its process, kinematic and
syntactic details and cutter data file. An NC reverse processor synthesises them to generate the
corresponding CL file. Note that all the other algorithms further use only APT CL data file.
This process requires that the reverse post processer is properly configured for the particular
syntax of the NC language and the particulars of the CNC control. These configurations are machine
specific. For example, the Fanuc 10A file contains a complete definition of the syntax and
Fig. 4 Reverse post processing flowchart
DOI 10.1007/978-1-4471-4976-7_16-1
Page 7 of 54

conventions used with this control. With multi-axis machines this process also requires that the
machine definition is properly configured for the particular style of multi-axis machine.
The reverse post-processor configuration file defines the relationship between G & M codes in the
NC program and the associated functionality. This is the same process that takes place in the control
itself where each NC code is interpreted before taking effect. G & M codes are identified through
fixed patterns. The reverse-post will identify patterns by comparing NC program contents with the
reverse post’s pattern definitions. When a pattern is identified, a specific functionality is associated to
it, and then the actual output tool motion or NC simulation is generated and expressed with APT
format.
Toolpath Backplot
APTand MCD are text based high computer language scripts that are difficult to visualize against the
part geometry, therefore, the obvious first idea was to couple the NC to a plotter that would trace on
paper the trajectory of the cutting tool. The drawing would immediately reveal an eventual mistake.
Nowadays the screen replaced the plotter but this NC toolpath preview function is still called back-
plotting, the oldest and most popular NC code verification. Only after the simulation shows the
program to be devoid of gross mistakes, the real machine can be used. Toolpath backplot follow the
cutter tip movement and display as doted (G0 fast move) or solid (G1 cutting) lines with different
colours, which could be used to different operations. Good backplot functions could highlight
current position, operation and cutter information. Cutter move animation is also a vivid simulation
of cutter movement. Modern backplot toolkit works together with NC code text editor, so the NC
code editing is visualized instantly, and doubles as an NC code learning tool for students.
The traces of tooltip are imagery lines in space but are useful in visualizing tool movement, so
these traces are also modelled as toolpath, even though in reality there are no such lines. The cutter
and tool holder can move forward and backward with tip on the toolpath, this vivid animation can be
used for NC programming and visual gauging check.
The classic double link list data structure is used for the toolpath model that is modelled in Fig. 5.
The double pointer enables the cutter to move forward or backward without looping through every
Fig. 5 Toolpath data structure
DOI 10.1007/978-1-4471-4976-7_16-1
Page 8 of 54

node. To delete and add one node is easy compared with an array data structure. This data structure
also works for extended Z buffer and extended Z map, where both need to delete and modify
elements in real time.
Material Removal Simulation of Cutting
UNIX based CAD Graphics Workstation introduced a separate graphics card and low cost large CRT
display, where the depth and colour buffers are used to store screen pixels. The painter’s algorithm is
used to draw animation pictures on screen. Virtual machining pioneers extended this depth buffer
with multi value so an animated cutter can move through a screen and paint a “negative” trace, or
erase something, just like cutting the stock from workpiece. This trick is also called extended
Z buffer method that is view angle dependent. The extended Z buffer method runs very fast since it is
directly updating graphic memory without view transformation which is a time consuming algo-
rithm inside the CPU. The extended Z buffer picture is pixel perfect with vivid details, since every
pixel is refreshed with a cutter colour. Some virtual machining systems are still animating material
removal with this technique, with another more precise in-process model as database. However,
directly rendering in-process model is a more popular approach, where the workpiece can be rotated
and zoomed during cutting animation.
This method projects the workpiece and cutter onto the display screen and gets an array of link
lists of an element that has a near Z value and a far value, a stick with the size of pixel. The array
correlates to the size of display window. The cutter image also has near and far Z values that will be
used to compare with the workpiece image. The cutter moves through this array of sticks and cuts
through them in Fig. 6.
If the cutter and the element never touch, there is no change in the extended Z buffer in Fig. 7. The
frame buffer is updated with this background image; this is to erase the previous cutter image at the
previous position. The cutter image is painted on the screen with current position to show a cutter
movement. Exactly a movie trick, the user sees a smooth move of cutter along the toolpath.
However, the developer usually only updates a small area around the previous cutter position to
speed up display speed. Some even used Boolean operation for colour buffer. These algorithms work
Fig. 6 Extended Z buffer cutting animation
DOI 10.1007/978-1-4471-4976-7_16-1
Page 9 of 54

for an old time computer that is slow and expensive. How to show a smooth cutter movement has
been a research theme for a long time.
If the cutter is cutting an inside element that is hidden from the user as shown in Fig. 8, the
extended Z buffer modiﬁes the inside element with the cutter colour. The cutter image is partially
blocked and trimmed by the outside element so the user can see the cutter plunge into the material.
If the cutter is cutting on the near end of an outside element, the user can see the material removal
process. If the cutter is cutting on the far end of an outside element, the user can see the cutter
plunging into the workpiece. If the cutter is cutting both the near and far end of the element as in
Fig. 9, this element will be removed and the user can see the previous hidden inside element, which
becomes the outside element at this moment.
The swept volume of the cutter is critical for material removal animation, where the swept volume
is subtracted from workpiece model continuously. This is a famous computational geometry quest so
there are many academic research works on how to generate this swept volume for 2 axis, 3 axis and
Fig. 7 Air cutting cutter movement animation
Fig. 8 Cutting inside element
DOI 10.1007/978-1-4471-4976-7_16-1
Page 10 of 54

5 axis milling, where the cutter position moves and cutter angle swings. For example, an end mill
linear move in the XY plane can sweep through one box plus a half cylinder, a Boolean operation
between this swept volume and workpiece model subtracts this swept volume from workpiece which
deforms the workpiece model, which will change shape with cutting, so it is a deformable geometry
model.
The extended Z buffer algorithm is good for animation but it is view angle dependent. Once
a simulation starts, the view angle and zoom factor cannot be changed, otherwise the simulation has
to start over from the beginning. The extended Z buffer in Fig. 10 can be saved as a geometry model
and measured against the design model. However, it is precise only in the view direction, so it is
necessary to simulate in different angles for more reliable results.
Workpiece In-process Model
The stock, the machining allowance, is the material to be machined. The workpiece is the target
machined part with stock material on surface, which may be a forged or cast part with the machining
allowance. The initial workpiece is a block produced from forging, casting or rolling process. The
geometry of the workpiece will change after each machining operation. This evolving geometry of
the workpiece is deﬁned as an intermediate or in-process model.
Fig. 9 Remove outside element and expose inside element
Fig. 10 Extended Z buffer in-process model
DOI 10.1007/978-1-4471-4976-7_16-1
Page 11 of 54

The deformable in-process model (IPM) represents the state of the workpiece at each step in the
machining process. It is a 3D geometrical construct that reflects the results of the machining
operations. This model allows the user to visually verify that the machining operations have been
defined accurately and that their sequence is correct. It can be automatically re-generated when there
are changes in the product design, machining parameters or sequence of the operations.
The in-process model is a must for the next step cutting plan. The in-process model is only
a conceptual model for most of the commercial CAM systems, there is no in-process model that can
be output and stored in a database. In a traditional NC programming environment, a significant
amount of time is spent trying to visualize the in-process model through various process stages. The
in-process or evolving model is used in subsequent setups and provides immediate feedback on the
progress being made. Being able to view in-process geometry, while creating toolpath and process
plan, greatly reduces the chances of error in both setup and machining. It also helps in designing
fixtures, positioning clamps and so on.
Host CAD B-rep In-process Model
The B-rep is a typical CAD geometry model. The shape of a part is represented in a point-edge-face
schema. The previous studies showed that NC cutting result could not be modelled in B-rep because
of the complexity of the cutting model. The first choice of IPM should naturally be the geometry
model B-rep used in commercial CAD system. The benefits of using the same geometry model for
CAD as the IPM are obvious. The CAD geometry model is matured and available through CAD
development kit, so there is little need to develop a new geometry model kernel. Sharing a common
geometry model with CAD, the IPM facilitates seamless integration of CAD-CAPP-CAM.
An automatic forging design and manufacture system was developed by the authors in 1986, in
which pre-form forging IPMs were the same as the CAD system CV/MUDUSA running on
VAX-11/750 computer (Liu et al. 1992; Jerard et al. 1989; Stifter 1995). However, the creation of
pre-form forging IPMs took days of calculation and often failed due to Boolean operation failure.
With a great deal of research efforts in the last two decades, the B-rep geometry model has been
improved significantly in terms of Boolean operation stability, but the B-rep based IPMs are still
limited to 2.5-axis milling (Fig. 11). Park reported a prismatic IPM generation method that employed
a polygon extrusion algorithm to sweep a ball-nose cutter (Park et al. 2003).
Host CAD Section In-process Model
Since the integrated B-rep IPMs cannot be created inside a CAD geometry model, a new, ad-hoc
cross-section-wire-frame based approach was proposed in a forging die CAD/CAM system (Liu
et al. 1991). The aim was to use a series of paralleled cross-section drawings to represent 3D shapes.
Figure 12 depicts the cross section representation of a link rod 3D shape.
The cross section IPM is widely used in many commercial CAD/CAM systems. I-DEAS from
SDRC uses water level cross-section as an IPM for generative machining. In a traditional NC
programming environment, a significant amount of time is spent trying to visualize the in-process
stock as it goes through various process stages. With I-DEAS, the wireframe section in-process stock
model can be created for downstream applications such as toolpath generation, process planning,
fixture designing and clamp positioning.
A part can be sectioned along the Z, X and Y axis that is shown in Fig. 13. The Z-axis section is
usually called water level section. For 3-axis milling, the water level section could have many loops,
causing complications in the set operation between sections. X and Y sections are single half loops
and the Z value is unique for every point, thus simplifying the set operation considerably.
DOI 10.1007/978-1-4471-4976-7_16-1
Page 12 of 54

Aworking system for using IPM in pre-forging design is described in (Liu et al. 1992). A drawing
sheet with part sections is ﬁrst created using the BACIS command language of CV/MEDUSA CAD
system. Since there are many sections in a drawing sheet, each section of wire-frame is assigned to
a different layer according to its Y distance, and a certain number of sections can be looped through
layers. Then each cutter section is moved to its cutter location and compared with the part sections.
The overlap between the cutter section and the part section will be removed from the part section.
A real milling IPM is obtained from the collection of the result sections.
Fig. 12 Cross sections of a link rod head 3D shape
Fig. 11 2.5D in process models
DOI 10.1007/978-1-4471-4976-7_16-1
Page 13 of 54

The display of sections is provided by line segments and can be confusing when there are too
many lines, i.e. there is a need to render the IPM as a realistic 3D image. In order to calculate the
surface normal required for rendering, the section wire frame is divided along the X direction by the
same step as that for Y direction. A so-called regulated section is formed to facilitate the calculation
of surface normal and interpolation of points between the sections. A given node in one section is
linked to a node in the next section. A node’s normal can be calculated from the four neighbouring
nodes. Figure 14 shows the regulated section representation, which in fact is called Z map.
Fig. 13 Section representation
Fig. 14 Regulated section representation
DOI 10.1007/978-1-4471-4976-7_16-1
Page 14 of 54

Z Map Based In-process Geometrical Model
The regulated section can also be used to accelerate set operation between cutter section and part
section. Calculation of intersections and trimming between two sections are time consuming and the
re-ordering of the line segments requires more computing time. This can be improved with the
regulated sections, where the line segments are indexed by both cutter section and part section. Only
the line segments with the same index are compared and trimmed, there is no need to trim two line
segments. If all the line segments fall on the regulated nodes, there is no need to trim two line
segments. The set operation can be simplified to the comparison of two Z values, which is very fast
and stable. Hence, the Z map representation of IPM emerges (Jerard et al. 1989).
Classic Z Map
If all the section line segments fall on the nodes, the object surface can be represented by the Z values
of the nodes. A map of Z values represents the object geometry. In computer language terms, the
Z map can be expressed as a two-dimension array Z[i, j], where i represents the index in X direction
and j represents the index in Y direction. The XY position of the Z map can be calculated by i or
j times grid size.
The best analogy for a Z map is a needle bed, where needles are uniformly distributed over the XY
plane of Fig. 15. The tip of every needle touches the object surface that it represents. A milling
simulation can be seen as the tool cutting through the needle bed. These needles can be described in
mathematical terms as Z-axis aligned vectors, passing through grid points on the XYplane. A Z map
representation can be used effectively for surfaces that are visible looking “downwards” on the XY
plane. Since 3-axis milling parts are composed of surfaces visible from the Z direction, they can be
expressed effectively by the Z map representation. With a Z map representation, the machining
process can be simulated by cutting the Z map vectors with the cutter.
Figure 16 shows an example of a 3-axis milling simulation system that was developed by the first
author in 1990. The system used DOS extender for Z map and SVGA for Z map rendering. The GUI
and NC toolpath wireframe display was coded with high C graphics library. The GUI and mouse
control developments were a very hard job and this was not resolved until the arrival of Windows
95 and OpenGL.
The vectors in a Z map have direction and length and are infinitely thin without volume. The top
of each vector, where the Z map and object meet, is just a point having no shape. Only at this point
the Z map and the object meet with each other. Z map models cannot provide accurate object
geometry outside these points. There are many ways of interpolating the geometry between grid
Fig. 15 Needle bed sample of classic Z map model
DOI 10.1007/978-1-4471-4976-7_16-1
Page 15 of 54

points in order to render a Z map model, for example, forming a triangle from three neighbouring
Z values.
It is obvious that the XY resolution of the Z map grid determines the precision of a Z map model.
A finer grid has greater precision but requires increased memory. For a part of 1 m * 1 m, the size of
the Z map is 1,000 Â 1,000 if the precision is 1 mm, but it increases to 2,000 Â 2,000 if the
precision is 0.5 mm. Reducing the model size and achieving suitable precision becomes a critical
issue in a Z map.
One of the solutions is to balance Z precision and XYprecision. An integer array is used to replace
the more common floating array of a Z map, which reduces the Z map size by half. At the same time,
this improves the Boolean operation speed because the comparison of integers is much faster than
the comparison of floats. The memory requirement of a Z map is halved again by compressing the
Z map file section by section, similar to image compression.
Because of the simplicity of its data structure and fast computation time, the Z map model is used
by most commercial CAM software (Jerard et al. 1989; Stifter 1995; Maenga et al. 2003). However,
a Z map cannot approximate vertical wall very well since it always has a slope as shown in Fig. 17.
This is not a problem for forging die design since there are always draft angles in forging parts, but it
is a serious problem for milling parts since profiling nearly always creates vertical walls.
Extended Z Map
Since the precision of the Z map is determined mainly by XY resolution along the vertical walls,
increasing the resolution along these walls while reducing memory is a key issue. Fortuitously, one
important feature of 3-axis milling can be leveraged. Viewing from the top, the vertical walls only
cover a small percentage of the Z direction projection, so it should be possible to use finer resolution
along the vertical walls while maintaining a rough resolution in the planar area. This was the initial
idea for an extended Z map; at least one grid on a Z map is segregated into sub-cells. Only grids
corresponding to intricate features on the surface of an object are assigned sub-cells to improve the
representation of object features. Figure 18a illustrates the plan view of the Z map grid with sub-cells
52 the front sectional view, while Fig. 18b shows the sectional view.
Fig. 16 Z map milling simulation
DOI 10.1007/978-1-4471-4976-7_16-1
Page 16 of 54

The size of the grid can be reduced through using sub-cells, but the precision of the XY dimension
is still limited by the size of sub-cells. For a sub-cell of 0.1 mm, the best precision is 0.1 mm in XY
plane. There is a need to represent XY dimensions precisely. Instead of using vectors in the sub-cells,
the sticks in the sub-cells that have volumes and surface geometry are used. A B-rep surface model
can be represented precisely using a map of B-rep sticks in Fig. 19.
Milling simulation with stick method involves Boolean operation between cutter and stick.
Figure 20 shows different stick shapes after cutting. The experiments with B-rep stick model are
Fig. 17 Classic Z map model with vertical walls
Fig. 18 Extend Z map with sub-cells along vertical walls
DOI 10.1007/978-1-4471-4976-7_16-1
Page 17 of 54

very slow and a huge B-rep model is created. To simplify stick and Boolean operation, a polygon is
used instead of real surface in a stick cell. The data structure of a polygon is much simpler than that of
a B-rep which needs a group of complicated pointers to maintain a double wing data structure.
The real world objects are not always uniform in the XY plane and can be any shape. Nodes are
used to enhance sub-cell precision in object face representations. For example, one edge of the
sub-cell may have two overlapping nodes to represent a vertical face. The nodes of a sub-cell may
not be uniformly distributed over XYplane. Figure 21 depicts an exploded plan view of a portion of
the Z map grid with nodes 54 and illustrates how stick method represents a circular hole and vertical
walls.
Z map has height value that is only suitable for 3-axis machining, where everything can be viewed
from the top. Machine components usually need six sides machining, either with a rotate table or 5-
axis control, which may result in hollow portions in some areas, which cannot be seen from the top.
If one ray is tracing through the hollowed object, there may be more than two intersections. Instead
of one height value, Z map could be extended to multiple values as in Fig. 22. Extended Z buffer is
a special multiple value Z map aligning with screen orientation.
Fig. 19 Stick method
Fig. 20 Different shapes of stick elements after cutting
DOI 10.1007/978-1-4471-4976-7_16-1
Page 18 of 54

Figure 23 shows shop ﬂoor examples of extended Z map IPM based NC simulation and
veriﬁcation that was developed in Singapore Institute of Manufacturing Technology (Liu 2005)
and implemented in precision engineering industry for a decade. The detailed description of the
extended Z map IPM can be found in two patents (Liu et al. 2002).
Voxel Based In-process Geometrical Model
Over the last three decades academic research explored many variations of deformable volumetric
model, such as discrete vector, graft tree, octree or hierarchical space decomposition and ray tracing
method. These inspiring research works contributed to the main stream volumetric in-process model
study, which starts with extended Z buffer material removal animation, enriched with Z map stick,
and ends with extended voxel model. The virtual machining industry learn, enhance, and merge
Fig. 21 Sticks to approximate vertical wall
Fig. 22 Extended Z map to multiple values
DOI 10.1007/978-1-4471-4976-7_16-1
Page 19 of 54

these techniques into their hybrid in-process models, which may use extended Z buffer for material
removal animation, stick for 3 axis mill, voxel for 3–5 axis mill or turn-mill and swept volume for
optimization. However, virtual machining industries seldom publish their internal data structures
and algorithms except for a few patents, only disclosing certain techniques that could be easily
identiﬁed by the export data and user interface.
Octree Hierarchical Space Decomposition
Instead of representing the blank as a collection of sticks in 3 axis NC simulation, it is possible to
represent it as a collection of cubes or spheres or any such cell of the same size. This is called uniform
space decomposition (USD). However, this is a very expensive way of representing solids so it is
limited to medical imaging application. Assuming that a bit is required to denote a cell, for
representing a workpiece of size 1 m with a resolution of 1 mm, more than
1,000 Â 1,000 Â 1,000 GB is required. Obviously this is not practicable. Therefore, methods to
represent an object as a collection of cells of varying sizes were developed, such as hierarchical
space decomposition (HSD) or octree representations in Fig. 24.
An octree is a tree data structure in which each internal node has exactly eight children. Octree is
a HSD representation in which an object is represented by a set of bigger cubes with subdivisions of
eight smaller cubes. This reduces the memory requirement considerably. Each cube is one-eighth of
its parent cube in size and is called an octant. All the octants can be visualized as the nodes of a tree in
which every node has eight branches. An octant can be completely inside or outside the solid, when
there is no need to further divide they become leaf nodes. Only boundary octants are further
subdivided into eight octants. This subdivision continues till the size of the sub octant equals the
required resolution. The total number of octants to be stored in an octree is much less than that of
USD representation, because the boundary octants take part in the subdivision. In practice, the
non-boundary octants memory can be reduced with a compression algorithm. It was found that in the
case of an octree the number of octants needed is nearly proportional to the surface area of the object.
Fig. 23 Extended Z map based milling simulation
DOI 10.1007/978-1-4471-4976-7_16-1
Page 20 of 54

All octree computations are based on integer arithmetic, which means that the analysis algorithms
are fast. Octree algorithms are readily parallel-processes by definition. Memory required by octree
representation is independent of the number of primitives and operations. For a given resolution,
memory required depends only on the surface area of the object. Boolean operations and rendering
display in isometric view are trivially simple since these operations require only tree traversal with
simple exchange of terms. The user is free to choose any desired accuracy (at the cost of speed and
memory). Coarse modelling is a facility unique to HSD. A coarse model of a solid can be produced
and processed quickly to get an order of magnitude estimate of the results. If these are found
favourable, a more accurate refined model can be produced.
However, Octree is an approximate representation and memory requirement increases exponen-
tially with increase in resolution. Instead of using subdivisions of the boundary octants, many
researchers proposed new ways to precisely describe the boundary surface geometry. The boundary
octant is renamed cell since there is no sub octant anymore.
Graft tree added two extra nodes on each edge so a few triangles could be formed to approximate
any polygon mesh. If a mesh node falls inside this cell, an extra node will be recorded. A surface is
subdivided by cells into small pieces and recorded into the cell.
One inspiring invention is the so called machining history based method. Instead of recording
surface into octant, this method records the neighbouring CNC toolpath and the linked cutter into the
cell. The neighbouring toolpath is the piece of toolpath that most likely will cut into the cell. Any
zoom or rotate of the workpiece will trigger a re-calculation of cell geometry and generate a more
detailed extended Z buffer image on screen. This is good for small NC programs. However, the
history of machining grows with the NC code, which could be millions of lines of text.
Voxel as Multiple Layers of Cubic Stick
The term voxel represents a volume element in space decomposition geometrical model schema, just
like the term pixel denotes a picture element in raster graphics. Extended Z map with stick method
could be considered as a simplified and extended one layer voxel model as in Fig. 25.
Figure 26 depicts an example of the voxel model, which could be considered as a many unit height
stick element stacking together and the memory requirements are enormous. There is a need to store
the voxel array in compressed form and use algorithms that will operate directly on the compressed
data, especially when the material is homogenous, where internal voxel could be represented by
boundary voxel extension.
Fig. 24 Octree hierarchical space decomposition
DOI 10.1007/978-1-4471-4976-7_16-1
Page 21 of 54

It is possible to convert the voxel array into some other more compact representation and
reconvert them into voxel when required. Voxelization is the process of converting a 3D object
into a voxel model. Figure 27 shows a test voxelization example.
A voxel-based system should be able to update the display at interactive rates. Current graphics
rendering systems cannot provide a level of rendering performance on voxel models that is
comparable to their polygon-rendering performance. Parallel algorithms and hardware support for
volume rendering are the focus of current research efforts. Only boundary voxel is rendered by
a patented colour list, which effectively avoids expensive ray-casting of huge internal voxels. The
rendering of a voxel model is easily achieved by rendering a points cloud. However, internal voxel
display is not possible with this method and needs more study. Figure 28 shows the rendering of
voxel model with voxel display.
Further analysing the voxel model, it is believed that the voxel-based volume modelling is a very
promising approach to the unified IPM for multiple machining and layered manufacturing simula-
tions. As a natural clone of the layer manufacturing 3D printing technology (Chandru et al. 1995),
the voxel model of an object and the object fabricated using a 3D printing closely resemble each
other since both are made of layers of small cells. Furthermore, voxel based models permit the
designer to analyse the 3D printing object and modify it at the voxel level leading to the design of
custom composites of arbitrary topology. In this paper a simplified voxel-based IPM is proposed to
unite the new 3D printing and traditional machining simulation.
The voxel representation also simplifies the computation of regularized Boolean set operations
and of material removal volumes. By using the material removal rate measured by the number of
removed voxels, the feedrate can be adjusted adaptively to increase machining productivity.
Fig. 25 Cubic stick as one level of voxel model
Fig. 26 Voxel method
DOI 10.1007/978-1-4471-4976-7_16-1
Page 22 of 54

Unified In-process Model of Multi Machining and 3D Printing
During the novel combined 3D printing and multi machining, such as shape deposition manufactur-
ing, a 3D printing part needs to be inserted with an electronic device and milled to a certain shape.
The uniﬁed 3D printing-machining simulation displays the machining process in which the initial
3D printing generated workpiece is incrementally converted into the ﬁnished part. The voxel
Fig. 27 Experiment voxelization
Fig. 28 Rendering model with voxel display
DOI 10.1007/978-1-4471-4976-7_16-1
Page 23 of 54

representation is used to model efficiently the state of the IPM, which is generated by successively
subtracting tool swept volumes from the workpiece (Donggo et al. 2000).
Figure 29 illustrates the framework of the unified voxel-based IPM for 3D printing and multi
machining. The voxel based 3D printing simulation can be achieved by the voxelization of the road
shapes, which are similar to a pipe along the 3D printing toolpath. Boolean addition between the
road shape voxel and the base voxel is fast and stable, independent of the model shape, which is
a critical issue with B-rep. One layer of road shapes would make a B-rep based solid modeller very
slow, since B-rep Boolean operation is dependent on model shape.
Furthermore, proposed unified IPM is a natural voxel mesh model (Nakashima et al. 2002) for so
called image based CAE analysis and this further unified CAD, CAM and CAE.
Current 3D CAD involves only shape data, which consequently poses certain difficulties in
process modelling and simulations aimed at predicting the performance of final products. Kase
introduced voxel CAD, which stores physical attributes together with 3D shape data (Kase
et al. 2003). Voxel CAD allows the sharing of data by different simulations and flexible manufactur-
ing methods.
There are other approaches [17Q2 –19] on unified model of manufacturing processes but none of
them could achieve the uniformity that voxel model could offer. Voxel model could be used in NC
toolpath generation-simulation-optimization, shape design optimization, forming process simula-
tion, and many other manufacturing applications. This will result in a unified volumetric geometry
model for all design and manufacturing processes that would erase the data exchange barrier and
CAE re-meshing problem.
CGTech started from NC verification software and then to NC optimization and simulation
software. During the first 15 years, CGTech has concentrated on removing material, and recently
it started working on adding material. Since most aircrafts now need carbon fibre, the Boeing
787 program asked CGTech to develop the manufacturing and simulation software of composite. So
Fig. 29 Framework of unified IPM for multi machining and 3D printing
DOI 10.1007/978-1-4471-4976-7_16-1
Page 24 of 54

now after 10 years, composite manufacturing and simulation is one of its core businesses. For the
fibre composite, VERICUT can not only simulate, but also can do the fibre placement program.
Therefore, composite manufacturing and simulation has become a new growth for virtual
machining.
VERICUTcomposite simulation (Fig. 30) reads CAD models and NC programs, either from VCP
or other composite layup path-generation applications, and simulates the sequence of NC programs
on a virtual machine. Material is applied to the layup form via NC program instructions in a virtual
CNC simulation environment. The simulated material applied to the form can be measured and
inspected to ensure the NC program follows manufacturing standards and requirements. A report
showing simulation results and statistical information can be automatically created.
AVirtual Machining System Example
Practice is the best way to learn. QuickCNC from Singapore Institute of Manufacturing Technology
is taken as an example to demonstrate the functionality of virtual machining and its process flow. The
system has been successfully applied in industry and training schools for many years to promote
virtual machining technology.
Virtual Machining Process Flow
The graphic user interface (GUI) of QuickCNC is depicted in Fig. 31. The multiple windows can be
viewed from different angles, zoom factors and detail levels. For example, total toolpath and current
toolpath can be separately displayed without workpiece or against design part, with cutter or holder.
The view details are easily controlled with NC toolpath toolbar buttons and hot keys.
The right dialog bar controls toolpath and simulation. The top slide bar interactively controls
simulation speed in run time, the user can slow down cutting animation to watch a certain operation
or get a result without animation. The cutter can move along the toolpath with NC toolpath dialog
Fig. 30 VERICUT composite simulation
DOI 10.1007/978-1-4471-4976-7_16-1
Page 25 of 54

bar. The current operation NC file name and location, cycle time, lowest Z value, cutter information
are updated instantly. The progress bar on top of this dialog bar will show the percentage of
completion and warns user with orange colour and highlights errors with red colour. The error log
will show the kind of error, either too deep, too much, rapid move collision, holder gauging or
overcut, with statistic number. The user can search for error block without reading through the
G code text file, which could be millions of lines long. The user can also move cutter to any node of
the toolpath and get current NC block position, G code, feed and speed, compensation, etc. instantly.
The user also can display only current Z level toolpath and move up/down for water level high speed
cutting.
The dialog bar on the left can analyse workpiece against design model with colour map and cross
section, which are dynamically sliding along XYZ axis with two slide bar control. Colour map range
can be modified with instant remaining stock display. The user can pick any point on the stock and
know which operation, which cutter and which block of the NC code cut the location.
Pan and rotation of the view follow Windows convention of right mouse button and left mouse
button. Dynamic zoom uses centre wheel function. There are four fixed view angle buttons for quick
action.
The virtual machining process flow is depicted in Fig. 27 and summarized as below:
1. The raw material model is created based on the design part model and selected stock.
2. The tool path model is created based on machine control code and selected cutting tools (cutters).
Tool list can be automatically extracted from APT cutter data, or G/M code, where the comment
line could be customized to contain cutter information.
3. Quick display toolpath for identification of geometry errors.
4. Quick simulation or slow cutting animation.
5. After simulation, the workpiece model can be saved and refined with amazing detail. The saved
in-process workpiece model can be reopened as the raw material for the next operation.
The details of the process flow in Fig. 32 and related working principles are described in the
following subsections.
Fig. 31 QuickCNC GUI
DOI 10.1007/978-1-4471-4976-7_16-1
Page 26 of 54

Automatic Creating Raw Material Model
A raw material model can be interactively defined as a box or cylinder as in Fig. 33 or generated from
casting or forging model as in Fig. 34 or design a part model as in Fig. 35.
In Fig. 33, the origin of work coordinate system (WCS) has to be selected, usually at the top centre
or corner, since it is easy to measure with a touch probe. A model resolution - the size of voxel cell,
has to be specified as well. The system usually gives a default value in line with the size of the part,
mostly between 1 mm and 0.1 mm. However, this is not the simulation precision, which is usually
less than 1 mm.
Machining a part from a block could be a great waste if the part and the raw material differ a lot,
sometimes half of the raw material has to be machined and becomes waste. In order to reduce waste
and achieve faster production with near net shape machining, the raw material could be forged or
cast into the final part shape with a few millimetres of machining allowance.
The forging and casting parts, designed with conventional CAD tools, can be exported to the
virtual machining system through a stereo lithography (STL) file which is a triangle polygon mesh in
text or binary format.
As shown in Fig. 34, a raw material model is generated from a casting or forging model, where the
geometry may be a complex surface. The origin of work coordinate system (WCS) usually follows
the part origin.
The so called design part is the target geometry of machining, where the raw material stock model
is the original shape of workpiece. The box envelop of design part can be automatically extracted to
generate a raw material block for quick CNC simulation. However even in shop floor practice it is
difficult to get an exact block of the design part.
Automatic Load NC File and Reverse Post into Toolpath
Machine control data (MCD), such as G/M code, are reverse posted into internal NC toolpath,
usually in APT CL data format. The APT CL data file of commercial CAM system can be directly
read in without the reverse processing step. The grammar errors, such as missing key words, could
Fig. 32 Virtual machining process flow
DOI 10.1007/978-1-4471-4976-7_16-1
Page 27 of 54

be discovered at this stage. Some geometrical errors, such as centre of the circle not-aligned or two
blocks overlapping could be highlighted with colours in toolpath display.
The APT CL data file contains cutter definition, such as cutter diameter/radius/length/angle, so the
cutter information will be automatically loaded without human selection. However, there is no
official cutter definition in G/M NC file, manual cutter selection is a boring task and introduces
another possible human error in NC verification. QuickCNC automated the cutter selection by three
steps:
1. QuickCNC builds a cutter table with company specified cutter names.
2. Customize commercial CAM system to export cutter definition on top of NC file inside comment
lines: (HTC50R4.5 process R).
3. QuickCNC parse comment lines to search for cutter definition.
Awell defined cutter table can standardize the tool room operation and management. Holders and
special cutters can be defined as well. Figures 36 and 37, respectively, illustrate the cutter definition
and tool list, while Fig. 38 depicts automatic cutter search using cutter name. A new cutter will be
created in the cutter table if there is no cutter match.
Fig. 33 Define a raw material stock
DOI 10.1007/978-1-4471-4976-7_16-1
Page 28 of 54

The subprogram can be automatically loaded from the main program in Fig. 39. QuickCNC will
look into the subprogram files and find the linked program number at the head of the file.
Three steps for fully automatic simulation are to automatically create raw material, open all files in
the same folder, search for cutter and subprogram. Now a machinist can complete a quick simulation
with just three mouse clicks to load STL part file to create raw stock, reverse post processing a whole
folder of G code files into toolpath and quick simulates machining operation and automatic verify
NC program.
Fig. 35 Create a block stock from design part model
Fig. 34 Create raw material stock from casting or forging part model
DOI 10.1007/978-1-4471-4976-7_16-1
Page 29 of 54

Quick Toolpath Display
Workpiece, cutter and toolpath can be interactively viewed with zoom and rotate. The cutter moves
along toolpath with a current position in the current operation that is highlighted as the current
toolpath in red. Figure 40 displays all toolpaths in one viewport, which could be overlapping and
confusing. A part could be machined with multiple operations, such as drilling, roughing, semi
finishing and finishing milling, which are in a planed order of different NC toolpath. A colour
scheme is used to distinguish NC toolpath with different colours and highlight the current operation
with red.
Figure 41 displays only current toolpath clearly against part model. For current toolpath, every
node is highlighted with a white dot. The cutter and holder could be shown with solid colour or
wireframe, even with a line or white dot, so the toolpath would not be blocked by cutter shadow.
Figure 42 highlights NC program errors, such as too deep cut, full width cut, cutter fast move into
material, too much cut, cut into machine table, holder gauging, overcut and overlapping blocks.
Fig. 36 Cutter definition
Fig. 37 Tool list dialog box
DOI 10.1007/978-1-4471-4976-7_16-1
Page 30 of 54

In the real world, CNC machine tool follows MCD control step by step. In virtual machining, it is
possible to move forward and backward for easy check up. Figure 43 shows the toolpath control
buttons that can be used to move cutter forward or backward along the toolpath, where the current
NC block information is updated immediately, so the user can check and verify interactively.
Fig. 38 Automatic search for cutter
Fig. 39 Automatic search for sub program
DOI 10.1007/978-1-4471-4976-7_16-1
Page 31 of 54

Cutting Simulation
Quick simulation or slow cutting animation can start and switch at ease, as shown in Fig. 44. The
animation speed could be interactively adjusted with a slide bar that is on the top right of the dialog
bar. The workpiece can be rotated and zoomed at any moment of simulation.
Upon the completion of the simulation, the initial raw material is machined into final shape, an
in-process model is shown in Fig. 45, and a log file is automatically generated. The file records file
name, cutter number and name, diameter and radius, length, minimum length and all types of error.
• C:QuickCNCtraining_example.nc
• Cuttter Number ¼ 67
• Cutter name ¼ D12
Fig. 40 Display all toolpath
Fig. 41 Highlight toolpath
DOI 10.1007/978-1-4471-4976-7_16-1
Page 32 of 54

• Cutter diameter ¼ 12.000
• Corner radius ¼ 0.000
• Error Type 1 Too deep cut ¼ 0
• Error Type 2 Full width cut ¼ 0
• Error Type 3 Rapid G0 cut ¼ 0
• Error Type 4 Too much cut ¼ 0
• Error Type 5 Plunge into table ¼ 0
• Error Type 6 Holder Gouging ¼ 0
• Error Type 7 Overcut part face ¼ 0
• Error Type 8 Minimum distance ¼ 0
• Cutter length ¼ 100.000 mm
Fig. 42 Toolpath error display
Fig. 43 Quick move cutter forward and backward
DOI 10.1007/978-1-4471-4976-7_16-1
Page 33 of 54

• Cutter length can be reduced ¼ 12
• Cutter length should be longer than ¼ 88.000
• Volume of the remaining stock ¼ 71,267
After simulation, the workpiece in-process model can be saved and retrieved later for the next step
of operation.
Fig. 44 Start simulation
Fig. 45 Completed simulation
DOI 10.1007/978-1-4471-4976-7_16-1
Page 34 of 54

Refined Workpiece In-process Model Display
The workpiece in-process model is displayed in a rough mode for quick interactive viewing, such as
zoom and rotates with easy mouse control. Figure 46 shows the reﬁned workpiece display with
amazing detail, such as the remaining stock and scallop height, with a colour map that could be
customized by user.
The mouse cross can be used as a probe to measure XYZ position on workpiece surface. The
cutter name and operation for this position can be displayed instantly on the workpiece dialog bar, as
shown in Fig. 47.
Fig. 46 Zoom to reﬁned details
Fig. 47 Measure workpiece in-process model against part design
DOI 10.1007/978-1-4471-4976-7_16-1
Page 35 of 54

Cross sections of workpiece could be viewed with or without the design part model. The section
position could be controlled with a slide bar for dynamic effects. Figure 48 sectioned the workpiece
in-process model along X axis against design model, with the display control dialog box.
QuickCNC can display CNC errors in graphics as shown in Fig. 49, where conventional log file
could be lengthy and difficult to read. Search for error is easy and quick with two buttons.
Figure 50 uses colour map of the remaining stock to visualize the left over from previous
machining operations. User can define the range of interested area. This technique can also be
used to show spark gaps in EDM machining.
Fig. 48 Section view against STL model
Fig. 49 Highlight cutting errors
DOI 10.1007/978-1-4471-4976-7_16-1
Page 36 of 54

Virtual Training of Machinist and CNC Programmer
As a cutting edge technology of modern manufacturing industry, CNC machining produces essential
inputs for virtually all types of manufacturing products for different applications, including injection
mold, sheet metal die, casting die, jigs and fixtures and other special tools. CNC technology has been
widely used in computer-aided manufacturing (CAM), high speed machining (HSM) and ultra
precision machining (UPM). The extensive use of CNC significantly improves the productivity of
precision engineering but has caused a shortage of skilled technicians or machinists, especially in the
knowledge intensive areas such as HSM and UPM. Training of skilled machinists is therefore
a crucial yet challenging job. A qualified HSM machinist should have good knowledge of machin-
ing, understand the operation of the machine tools, and be able to do planning for machining process.
Traditionally, trainees acquire their operating skills in several years through observation and
reference to the operation manual. After which, they would learn to operate machines for themselves
under the guidance of experienced operators. The acquisition and maintenance of real CNC
machines, the consumptions of real materials in machining, and the set-up and maintenance for
workshops, all contribute substantially to the high cost of conventional CNC training. Cost-effective
and safe CNC training is thus highly desired.
An apprentice will get to know a conventional mill by handling it under controlled conditions, by
machining initially simple parts, always being careful to keep the tool far away from the faceplate.
Accidents happen. An extra turn of the lever and the tool may hit the machine table. Even a broken
cutter and a scratched faceplate in a learning mill is not much of a loss, a CNC machine tool costs
several times more and is more prone to serious accidents. Awrong line of code may punch the main
spindle towards the machine table, provoking a horrendous collision causing serious losses.
Students could be traumatized by the crash and lose interest in this trade, which is facing an
increasing problem of manpower shortage.
With computers becoming more common, the obvious follow up development is software that can
simulate the entire process, dispensing with the real life machine tool altogether. The challenge of
moving from a manual machine tool to a CNC version resides at the programming side, not in
handling the machine. Since both PC and CNC control panels use touch screen, it makes little
Fig. 50 Colour map of the remaining stock
DOI 10.1007/978-1-4471-4976-7_16-1
Page 37 of 54

difference on whether the programming is for a machine simulated in PC or a real life CNC machine
control panel, which itself is a computer, so CNC programming training can be naturally replaced
with virtual machining. While CNC training using real CNC machines is necessary, the use of virtual
reality (VR) technology to support CNC training has been a popular topic in recent years
(Avgoustinov 2000).
Simulation of the entire machining processes for CNC training is significant given its lower cost
and risk-free nature. The drastic decrease of the cost of computer, coupled with the worldwide price
increase in material and machine tools means that virtual CNC training using computerized
modelling and simulation is a cost-effective and sustainable approach to technical and professional
education in manufacturing applications. The virtual CNC training system is developed for simu-
lation of multiple machining processes. It is particularly important in the training of knowledge-
intensive high speed and ultra precision machining. Compared with conventional on-site manual
training or e-learning, the virtual CNC training system greatly increases learning efficiency and
effectiveness of trainees, and improves cost saving in terms of machine and material uses.
Virtual manufacturing is the use of a desktop virtual reality system for the computer-aided design
of components and manufacturing processes. Virtual reality is a computer technology that enables
users to view or ‘immerse’ themselves in an alternate world. Immersion and man–machine interac-
tion is the core of VR technology. VR technology has obvious applications in education and training
where potentially dangerous tasks such as flying or surgery are carried out and also has been used for
many different applications in a variety of industries. This work provided some insight into
reconstructing of virtual machining centre by using PC platform and realized the machining centre
navigation and man–machine interactive operation. In this virtual environment, users can operate the
machining centre and complete a product machining process. Through this virtual platform, users
can obtain knowledge about the structure of machining centre and get familiar with the complex
operation of machining centre before they have the opportunity of manipulating the real machining
tool, which is desirable for practical operation.
Current Status of Virtual CNC Training
Compared to NC simulation applications which are expensive and mature, the virtual CNC training
system is still primitive. The NC simulator developers are not actively trying to provide a training
system because the training software market is logically smaller than production software. More
importantly, NC simulator developers need to revamp the graphics engine or geometry kernel to suit
education game use. None of the leading NC simulators has any CNC training capability. This leaves
the development of virtual CNC training system to machine tool vendors and schools who do not
have expertise to develop a good graphics engine (Garcia-Plaza et al. 2011).
The CNC control vendors developed their own training system. For example, Siemens developed
SinuTrain, which is CNC training software. It runs on PC and is suitable for training purposes and
self-study as it is for writing programs and simulation. It serves for writing and simulating NC
programs on a PC, based on the DIN 66025 programming language as well as the products
ShopMill, ShopTurn and ManualTurn + and language commands for SINUMERIK®
810D,
840D and 840Di controls, all are Siemens products. Programs written with this software can be
used on real machines. A prerequisite is that the SinuTrain software is adapted to the SINUMERIK
control on which the program is to be executed. This adaptation must be carried out by specially
qualified personnel, e.g. from Siemens. It is important to stick to Siemens and the machine-tool
manufacturer’s instructions when adapting the software. No liability is accepted by Siemens if these
requirements are not adhered.
DOI 10.1007/978-1-4471-4976-7_16-1
Page 38 of 54

The vender specific virtual CNC training systems have very good GUIs which have been
customized to the vender’s own CNC controller; some even have a touch tablet that simulates
operation panel. However, the cutting simulation is rough and primitive, despite the sound and chip
flying animation.
VRML can be used as an inexpensive means for simulation of one of the most interesting but also
most time and resource consuming areas of computer aided manufacturing (CAM) – machining of
complex parts.
There has been much research and many publications on virtual CNC training in the last 10 years.
Most of the published graphics engines are based on VRML and Java 3D. They explored internet
based CNC training and remote NC simulation etc., which are futuristic but not practical at this
moment. The computer hardware is very cheap now that there is no need to run a training system
over the internet. Remote graphics over the internet is not necessary. Some of them use flash movies
such as micro media to do animation, which could only be used for pre-fixed scenes.
In VRML, the realization of dynamic material removal during a machining process remains
a problem (Garcia-Plaza et al. 2011). Some commercial software such as Deneb’s virtual NC can
export a VRML animation to describe a machining process. Nevertheless, during the cutting
process, the geometry of a workpiece remains unchanged. The reason is that VRML does not
support set operations among geometric objects such as union, intersection and difference. This
makes it difficult to simulate the change in geometry of a workpiece under cutting.
In layman’s term, the current virtual CNC training systems are educational games that lack the
realistic feeling of machining, which is quite different from realistic simulators, such as the flying
simulators that are used to train pilots.
The next generation of virtual CNC training is to provide knowledge intensive CNC training, for
the future skilled machinist of precision engineering (PE) industry through pervasive physics-
geometrical modelling and simulation of multiple machining processes, especially high speed and
ultra precision machining. To realize this vision, a new in-process model (IPM) that is deformable
and precise is needed.
Virtual CNC Machine Centre
The proposed new in-process model has been used in the virtual CNC training system developed at
Singapore Institute of Manufacturing Technology (SIMTech) for training of CAM programmer and
CNC machinists. The block diagram is given in Fig. 51. Architecture of other virtual machining
systems also will be similar to this; they will differ only in the representation scheme.
Fig. 51 System architecture of a virtual training simulator
DOI 10.1007/978-1-4471-4976-7_16-1
Page 39 of 54

Virtual CNC machine and control panel were developed on Microsoft Windows platform, with
OpenGL as the graphics engine. The system architecture follows Microsoft Application Framework,
with a modeless dialog box as the blue print for control panel.
The CNC control panel in Fig. 52 is different from normal Microsoft Windows Dialog in terms of
user experience, since CNC control panels were developed before the PC age, with CRT display and
hard buttons. In order to simulate the traditional CNC control panel, conventional modeless dialog
has to be customized with special graphics features, even buttons were drawn from the bitmap
image. How to turn the knobs is another problem, the mouse centre wheel was employed to rotate.
The machine frame, door knob, spindle, workpiece, tool change button, probe indicator and work
table are shown in Fig. 53. All the tools on the machine are operable with mouse buttons and centre
wheel.
Fig. 52 Graphics user interface of a virtual training simulator
Fig. 53 Machine frame and ﬁxtures of a virtual training simulator
DOI 10.1007/978-1-4471-4976-7_16-1
Page 40 of 54

Virtual measurement on machine tool can be realized with virtual vise, dial indicator, probe,
wiggler and tool pre-setter in Fig. 54. The virtual instruments reflect the manual operation in the
same way as the real one; the indicator needles rotate just like a real watch; the wiggler vibrates with
a shadow. The user can manually turn the instrument with mouse clicks and centre wheel.
With these virtual tooling components, a trainee can learn how to clamp workpiece on the
machine table virtually with a predefined operation procedure, as shown in Fig. 55.
The trainees can use the system to simulate the milling process and save the “machined” model for
other downstream machining processes. Figure 50 demonstrates the simulation of the remaining
stock and the scallop height. In addition, they can control the simulation speed to see the details at
any angle on the current situation of the machining, which is difficult if not impossible in the real
machine based training.
A set of different machining samples has been provided to demonstrate how the generated tool-
path works with cutter under various cutting parameters with the aid of the virtual controller.
Trainees can learn different setups in a short time using virtual simulator on PC, which significantly
shortens the learning curve compared to the traditional training in a workshop.
While it is dangerous to show the effect of a wrong setup or NC code on the shop floor, the virtual
simulator has graphics and sound developed to synthesize various effects. In particular, the virtual
CNC training system can simulate an accident using a graphic and sound effect when a trainee
Fig. 54 Virtual vise, dial indicator, probe, wiggler and tool pre-setter
Fig. 55 Virtual CNC simulates the clamping of workpiece
DOI 10.1007/978-1-4471-4976-7_16-1
Page 41 of 54

breaks a leg of a work-piece during the virtual machining. Among several beneﬁts safety and
material cost reduction are the direct and major gains from virtual simulation.
Virtual CNC Control Panel
The virtual CNC control panel is a virtual copy of an actual machine control panel. The virtual CNC
panel is integrated with a simulated machine tool. The simulated machine responds to the programs,
commands and inputs of the virtual control panel in the same manner as a real machine. The CNC
emulator and machine tool simulator allow anyone to learn actual CNC automation at any time and
in any place. The new Windows touch screen serves as a good control panel interface. For modern
LCD touch screen control panel, touch screen PC is a natural clone so the emulation is perfect with
the virtual key pad.
The virtual display emulates real CRT display with traditional style of text. The display content
will change according to different control modes, such as actual position, all positions, WCS table,
compensation, etc., as shown in Fig. 56. The virtual CRT display and machine movement is
synchronized without delay.
After the trainee measures the workpiece position on machine table, the data can be easily input to
the virtual control panel by virtual key pad. The data can be retrieved in a later session.
Fig. 56 Virtual CNC control panel display
DOI 10.1007/978-1-4471-4976-7_16-1
Page 42 of 54

Virtual Manual Machining Operation
Virtual CNC training system starts as a power off machine tool. The trainee needs to power on the
machine and release the red emergency stop button. Then the returning to home operation has to be
completed for three axes. Without this step, the next operation will not be accurate and precise.
The door has to be opened before any operation on machine table. The tool table can store the
fixtures. The trainee can pull the tooling between machine table and tool table.
Virtual manual operation is a great challenge for a low cost personal computer. The trainee must
be able to move workpiece, tighten screw, insert a parallel bar, pull a shim through a gap and turn
knobs all the time. The mouse and touch screen are main interactive devices of the PC. After many
trial and errors, the mouse wheel is used to tighten screws and turn knobs, the double clicks on
workpiece are simulated as hammer to shift workpiece for positioning, and the touch screen is used
as control panel key pad, so far this is the most realistic approach.
The different cursors are displayed for easy use. There will be a screw cursor when the mouse
cross is near a screw. A hammer cursor will appear at the boundary of workpiece and indicate
a minor position shift can be achieved with a blow.
A machinist can manually operate virtual CNC training system to mount workpiece on machine
table, clamp it with fixture or vise, align with axis and setup machining origin. Virtual vise and
fixtures must be locked with a screw; otherwise the workpiece position will shift during cutting
simulation. The workpiece shift is animated with vibration and sound.
All the cutters have to be pre-set to the correct length and record this tool length value through the
virtual control panel. Virtual dial indicator, touch probe and cutter pre-setter function as real ones
with two degrees of needles for display. The real time interference checks will feed realistic values
on the display.
Virtual shim could be employed to check gap between the cutter tip and workpiece. The trainees
use mouse to drag the selected shim through the gap and see the difference. If the selected shim is
thicker than the gap, the shim will not pass through the gap. Using different shim and jogging cutter,
the trainees can calculate the correct cutter length and workpiece position.
Virtual wiggler is an even more interesting instrument to align workpiece. It will stop vibrating
only if its outside diameter properly aligns with workpiece walls. Graphics animation of wiggler is
amplified for easy observation.
Safe Training of Machine Operation in a Classroom
The precision engineering Worker Skill Qualification (PE WSQ) Specialist Diploma is a joint
initiative by SIMTech and the Singapore Workforce Development Agency (WDA) to provide
hands-on training to equip future PE professionals in cutting-edge precision machining processing
technologies. This program is conducted through a series of lectures, laboratory demonstrations and
project attachments in selected industrial applications. As most of the training organizations have
limited numbers of CNC machine tools and CNC trainer available, they can install the virtual CNC
training system on their PCs to conduct hands-on training. In this program, the virtual training
laboratory is designed for 40 students to learn CNC. A high speed machining course is conducted for
the WSQ trainees to learn machining using the system. Twelve sets of HSM examples are created
allowing trainees to learn different machining techniques and strategies, one of them is shown in
Fig. 57. Using virtual CNC training can effectively reduce CNC learning curve from typical weeks
long to just one night. Trainees can do self-learning using the same software on their own PCs.
With the financial support from Local Enterprise and Association Development Programme
(LEAD), SPETA has deployed the virtual CNC training system in their classroom, as shown in
Fig. 58. As one of the several critical areas they identified to enhance the capabilities of the PE
DOI 10.1007/978-1-4471-4976-7_16-1
Page 43 of 54

Fig.57VirtualCNCmachiningexampleforWSQcourse
DOI 10.1007/978-1-4471-4976-7_16-1
Page 44 of 54

companies, training of CNC machinists has been paid great attention. In partner with SIMTech,
SPETA uses the system to train CNC machinists – somewhat like the flight simulators to train pilots,
which will significantly trim the training hours on the actual machine.
The system is also used in Institute for Technical Education (ITE) for training computer numeric
control machinists. Significantly reducing the hours and machine resources required, the virtual
CNC training system enables trainees to practice more with various machining requirements within
the same allocated training time. With this additional preparation, trainees would have a shorter
learning cycle when they start working with the companies. Virtual CNC training has been featured
in local TV, radio, and all newspapers. Figure 59 highlighted virtual CNC training in Metal Asia
(MTA) and Singapore Science Festival.
Profiting from Virtual Machining
Virtual machining is not a “nice to have” feature that focuses on trade show demonstrations. There
are practical industrial applications which need virtual machining to complete, which include code
Fig. 58Q8 Virtual CNC training in practice
Fig. 59 Virtual CNC in MTA and science festival
DOI 10.1007/978-1-4471-4976-7_16-1
Page 45 of 54

parsing, toolpath backplot, trial cut replacement, adaptive speed-feedrate optimization and virtual
training.
Just like all virtual manufacturing technologies, virtual machining has been ready for pervasive
industry implementation for a long time. The only barrier was the high computing cost of UNIX
graphics workstations, such as Silicon Graphics, which were beyond the reach of small machining
workshops. With the recent worldwide price surge in materials, energy and machine tool, pervasive
virtual machining is not only technically possible but also makes business sense, since computing
cost is almost zero.
First Part Right
An NC program has thousands of lines of tool movement instructions that may contain errors.
Following these instructions, CNC machine tool will move blindly, without any check on gauging,
overcut or cutting force. It is not possible to verify code manually, so the NC verification software
was developed during the 1980s.
NC simulation features full 3D, solid model, shaded simulation of entire NC machine tools and
material removal. This visualization tool enables programmers and machinists alike to preview
exactly what will happen on the shop floor and check for collisions. Many use NC simulation for
electronic shop floor documentation.
NC verification detects problems in the NC tool path program. It is a powerful visual inspection
tool, which highlights fast feed errors, gouges and potential crashes/collisions. Programmers can
detect and correct problems before prove-out. With NC verification you can virtually eliminate NC
program mistakes, greatly reduce the time spent on prove-outs, and make the move to “lights-out”
machining. The NC simulation program is smart enough to detect problems such as fast feed errors,
gouges and collisions that could potentially scrap the part, break the cutter or crash the machine. Any
error discovered by simulating software allows the programmer to immediately identify the
offending NC program record by mouse-clicking on the error. The problem can therefore be fixed
during the NC coding phase so as to insure an error-free code when it reaches the shop floor.
NC analysis identifies the tool path record responsible for an error. You can quickly verify the
dimensional accuracy of the entire part with a full array of 3D measurement tools. NC analysis
compares the simulated part to the design model so you can be sure the machined part will match the
design intent. NC analysis performs constant gouge checking. Analysis of the “as-cut” part delves
deeper into the verification process. Is the resulting cut part dimensionally accurate? Does it match
the final desired part shape? NC verification software enables the user to zoom in on suspect areas for
in-depth inspection. The part can be rotated and cross-sectioned at any angle to check areas that
would otherwise be impossible to see, such as the intersection of drilled holes. Detailed measure-
ment tools enable the user to verify dimensions such as wall and floor thickness, hole diameters,
corner radii, scallop heights, depth, gaps, distances, angles, volumes, etc. NC simulation software
such as VERICUT®
from CGTech also provides the ability to automatically compare the as-cut part
with the original design. The AUTO-Diff module can embed the CAD design model inside the
stock, automatically comparing the design to the in-process workpiece in order to reveal any
discrepancies such as gouges or excess material not removed by the machining processes.
NC errors could destroy work pieces, even damage machine tool. One NC error could make the
workpiece a waste and take days to rework and eat into profit. In small batch production, there is no
time for trial and error. Especially for high speed machining (HSM), the fast moving and expensive
cutter is very easily broken. The dynamic machining load will greatly affect cutter life, geometry
accuracy and surface finishing.
DOI 10.1007/978-1-4471-4976-7_16-1
Page 46 of 54

The challenges also come from the huge tool path of HSM. A million lines of NC code are
common practice in today’s shop floor. The traditional NC verification is so slow that even HSM
itself is faster than verification. The size of the program combined with a high feed rate makes it
almost impossible to run test simulations prior to cutting metal.
NC verification cannot rely on CPU of faster processing cycles as single-core silicon reaches the
limit of heat dissipation and power consumption. A processor containing multiple cores, leveraging
its ability to execute multiple tasks, offers a higher level of computing power and functionality than
the current generation single-core processor. As this new technology comes to market, software
companies are examining how software will adapt. The current NC simulation models are not
optimized for multi-core computing as some software only run 12 % faster on a dual processor
workstation. How to split the NC simulation between dual-core and graphics card is a new R&D
challenge.
Singapore Institute of Manufacturing Technology (SIMTech) has developed a more efficient
approach based on a patented geometry representation. The system starts with a solid model of the
machined part and quickly simulates and optimizes machining processes. NC code could be
selectively reverse post processed into 3D tool path graphics display and interactively viewed,
edited and optimized. The user can highlight or hide operation, tool path or layer. The user can also
display and edit a certain layer of toolpath. Tool paths and cutting results can be viewed from any
viewpoint and checked automatically. The machined part and the design part are compared for the
remaining stock and over cut. Error-free tool paths are created, eliminating the need for a time-
consuming test cut.
Based on this patented technique, SIMTech developed several practical applications for mould
manufacturers. These include QuickSeeNC, QuickCNC and PartingAdviser, which provide “What
You See is What You Cut” functionality for shop floor machine operators and mould designers. The
technology is suitable for machine tool NC tool path simulation, verification and optimization in the
precision engineering, automotive, aerospace and electronics industries. QuickCNC has been
adopted by several local die and mould makers for its speed and simplicity.
Pervasive Virtual Applications in whole Process Chains
Moving beyond the NC programming department, virtual machining could be used pervasively all
over whole process chains, such as part design, tool design, process planning and scheduling, tool
data management, material, setup, production and quality control, as shown in Fig. 60.
Engineers often need a method of getting a model of the as-manufactured part back into the CAD
system for a variety of reasons. It could be that the required CAD model does not exist but legacy NC
program data to create it does. Frequently the as-cut part contains features (fillets, blends, etc.) not
present in the original CAD design and an accurate and complete model is needed for finite element
modelling or environmental simulation or further engineering analysis. Often, simulating pseudo
NC paths is the fastest and simplest way to create complex offset surface shapes.
Whatever the reason, NC simulation can create either surface or solid model representations of the
simulated machined part. The exported model can be either a surface or solid b-rep model with
geometric shapes (cylinder, cone, plane, torus sweep, etc.) that represent machined features such as
drilled holes, pocket corners and walls, filleted blends and other common manufacturing features.
Very small machined features such as scallops created by ball-end mill contouring of complex
shapes can be collected together into large surface patches representing the nominal feature intended
by the machining operation.
DOI 10.1007/978-1-4471-4976-7_16-1
Page 47 of 54

In-Process Geometry for Manufacturing Engineering
In addition to design engineering’s need for an as-manufactured CAD model, other manufacturing
engineering and planning functions could use the information.
It is difficult to imagine, plan and design all the resources required for subsequent operations
(NC programs, fixtures, custom cutting tools, inspection tools, work handling devices, transfer
methods, etc.) without an accurate representation of the initial material state left by the previous
operation. The accuracy, efficiency and “correctness” of each operation depend on the NC pro-
grammer, tool designer and process planner knowing the material’s initial geometric shape. Until
now, the only way to create an in-process CAD model was with expensive, labour-intensive, error-
prone and inaccurate methods. However, exporting a CAD model of the in-process or as-machined
solid model created automatically from the verification step makes it possible to avoid these time-
consuming activities. Users can create the CAD model at any stage in the machining process.
Simulation for Process Planning, Scheduling, Production and QA
In order for process planners to do their job effectively, it is crucial to know accurate machining
times. This can be easily obtained by simulating the NC program. Times are calculated for every step
in the machining process including the amount of time it takes to change tools, pallets or other
miscellaneous machine actions. This information can be essential to keeping the production floor
operating to its full capacity.
Additionally, the process planners can use an in-process geometry model to create robust
inspection instructions in very little time. Typically, a manufacturing engineer, NC programmer or
process planner manually creates these instructions to tell the machine operator what to measure and
how to document the results. Without an in-process model of the part, manual methods are very
tedious and prone to mistakes. The highly-customizable inspection instructions can be created
automatically. This helps to establish a formal but easy and efficient method to create the necessary
documentation. The software outputs the inspection instructions based on the dimensions of the
simulated cut stock (as-cut semi-finished wall thickness ¼ .1500
, for example). The accurate
in-process geometry is required to automatically generate this type of document, and is only
available by simulating the NC program.
Documentation for Workshop
The latest NC simulation system includes powerful tools for creating custom reports, tailored for
a specific user/department/company’s needs, containing useful process information generated
during the simulation. The automatically generated documents can be used for shop floor or
in-process documentation, NC programming documentation or to capture valuable process infor-
mation generated during the simulation session. Produced in standard HTML or PDF format, the
report layout is highly customizable, including the ability to specify page design, fonts, graphics,
tables, pictures, statistics and user-defined information critical to documenting the CNC machining
process.
Simulating CAM output to view basic workpiece material removal is no longer enough in today’s
competitive global marketplace. It is critical to be operating as efficiently as possible; modern
simulation and optimization software has become a key tool to minimize the cost and time of
production while maintaining or increasing product quality. It has evolved into an important process
that protects and frees up CNC machines, helps to eliminate scraped parts, and creates in-process
information that can be utilized throughout the manufacturing enterprise.
DOI 10.1007/978-1-4471-4976-7_16-1
Page 48 of 54

Adaptive Machining Optimization
CNC optimization automatically determines the best feed rate for each segment of the tool path
based on the machining conditions and amount of material removed. Optimizing NC feed rates
greatly reduces the time it takes to machine parts and improves the quality of surface finish.
After CNC verification and achieving error-free machining, the in-process model could be used to
achieve faster machining, which is based on the calculated material removal rate.
The machining model, simulation and verification processes ensure that the NC programs sent to
the shop are both accurate and efficient. To create the most efficient machining processes possible,
optimization software can determine the best feed rates to use for each cutting operation. Achieving
the best feed rates for each cut in an NC program has always been a goal for NC programmers but has
traditionally been a very difficult task plagued by a number of problems. First, trying to imagine the
cutter contact and cutting conditions or each cut in a large NC program is virtually impossible.
Manually inserting different feed rates for each changing condition is not practical. An incorrect feed
rate estimate can break the cutting tool, damage the fixture or scrap the part.
Typically either a single conservative feed rate is used for an entire machining sequence, as shown
in the left side of Fig. 61, or a higher (i.e., “high speed”) feed rate is used but with a very conservative
machining strategy. Both methods attempt to ensure that the cutter is not overloaded, but at the
expense of very inefficient machining. Both of these strategies result in too slow cutting speeds or
too light removal rates that waste time, increase costs and prematurely wear cutters.
To address this issue, a knowledge-based machining package essentially adds intelligence to the
cutter. During the simulation, the in-process geometrical model knows the exact depth, width and
angle of each cut because the software also knows the exact shape of the in-process material at every
instant of the machining sequence. It knows exactly how much material is removed by each cut
Fig. 60 Virtual applications in whole process chains
DOI 10.1007/978-1-4471-4976-7_16-1
Page 49 of 54

segment, and the exact shape of the cutter contact with the material. With this unique knowledge set,
it determines the best feed rate for each cutting condition encountered, taking into account volume of
material removed, chip load, and machine acceleration and deceleration requirements, as shown in
the right side of Fig. 61. If desired, the software can also divide cuts into smaller segments and vary
the feed rates as needed in order to maintain a consistent chip load or volume removal rate. It then
creates a new NC program, with the same trajectory as the original, but with improved feed rates.
Summary and Looking to the Future
Virtual machining simulates NC code to discover errors, without time consuming trial runs or online
debugging on real machine tool. Working towards a vision of pervasive modelling and simulation,
various deformable in-process geometry models from the 2D sections to 3D representations, from
Z map to unified voxel-based are discussed. A practical system developed based on the deformable
in-process geometry model is taken as an example to demonstrate the application of virtual
machining for NC verification. Virtual machine tool with a virtual CNC control panel and virtual
jigs and inspection tools is introduced for training purpose.
There was a time when the computer was expensive and software was difficult to use, but virtual
machining was still running with profit for high cost aerospace machining. Today the computing cost
is almost zero compared to material and machine centre, so it is time for pervasive virtual machining
application in every sector. An easy to use and low cost virtual machining system will find a wide
market.
Looking forwards, sustainable machining is a great challenge. Towards smart and competitive
sustainable machining, CNC model and simulation will be used to optimize the machining process,
where the raw material could be saved through first part correct technology, the energy could be
saved through cutting speed optimization, and used parts could be saved by remanufacturing.
The simulation of chip formation using the finite-element-method (FEM) predicates the cutting
force and chip thickness, thus saving time in the subsequent machining trials. State of the art today is
the individual simulation of the machining process and the machine performance. In real machining
processes however, these parameters are inter-dependent and influence one another heavily.
Fig. 61 NC optimization
DOI 10.1007/978-1-4471-4976-7_16-1
Page 50 of 54

Integrated simulation, whereby the process-machine interaction is simulated, is therefore a further
key technology for sustainable production in the future.
Tool chatter is the barrier for higher material removal rate and can damage machine tool spindle.
A certain combination of depth of cut and speed can incur self excited vibration of tooling system
and generate cutter marks on the machined surface. Dynamic machining model and simulation can
predicate best cutting speed and depth combination that will cut faster without chattering.
The simulation of machining operations offers the potential to fulfil the ecological, social and
economic requirements of sustainability. For example, the adjustment of the suitable feed rate in the
milling of complex geometries from difficult to machine materials may be optimized through
simulation and thus reduce the machining time by up to 40 %. The resultant reduction in consumed
resources allows a saving of both costs and energy.
The machining stock is the volume difference between the designed part geometry and raw
material geometry. Reducing the machining stock can save raw material. The minimum machining
stock could be achieved through near net shape forming of the raw material, such as casting, forging
and welding.
The most material and energy are wasted in the manufacturing processes. These wastes can be
saved through re-machining of the damaged component, where the damage can be repaired by
welding or thermal spray. For a long time in the aerospace industry, overhaul of jet engines has been
a profitable business worldwide. Nowadays even the automotive industry has started to
re-manufacture many components, especially engines. However, re-machining is a great challenge
for CNC machining since the damaged component geometry is warped. Virtual machining can
simulate this warped component and generate 3D printing and the following cleanup toolpath.
ReferenceQ3 s
Q4 AbdulKadir A et al (2011) Virtual machine tools and virtual machining – a technological review.
Robot Comput Integr Manuf 27:494–508
Avgoustinov N (2000) VRML as means of expressive 4D illustration in CAM education. Future Gen
Comput Syst 17:39–48
Q5 Chandru V, Manohar S, Prakash CE (1995) Voxel-based modeling for layered manufacturing. IEEE
Comput Graph Appl 15(6):42–47
Donggo J, Kwangsoo K, Jungmin J (2000) Voxel-based virtual multi-axis machining. Int J Adv
Manuf Technol 16(10):709–713
Garcia-Plaza E et al (2011) Virtual Machining applied to the teaching of manufacturing technology.
Mater Sci Forum 692(2011):120–127
http://www.engineeringchallenges.org
Jerard RB, Hussaini SZ, Drysdale RL, Schaudt B (1989) Approximate methods for simulation and
verification of numerically controlled machining programs. Vis Comput 5(4):329–348
Q6 Kase K, Teshima Y, Usami S, Ohmori H, Teodosiu C, Makinouchi A (2003) Volume CAD. In:
Proceedings of the 2003 Eurographics/IEEE TVCG workshop on volume graphics VG ‘03
Liu PL (2005) Profiting from dynamic numerical control verification system. Cutting Edge News-
letter of SIMTech, Issue 43. http://www.simtech.a-star.edu.sg/index-publication.html
Q7 Liu PL et al (1991) A new concept integrated CAD/CAM system for complicated die and mold. In:
Advances in computer science application to machinery. International Academic Publisher,
pp 90–95. ISBN 7-8003-154-3/TH.2
DOI 10.1007/978-1-4471-4976-7_16-1
Page 51 of 54

Liu PL et al (1992) 3D Complicated parts design based on the automatic shape generation. Chin
J Mech Eng (English Edition) 5(2):88–92
Liu PL, Lu YQ, AO YT (2002) An object representation method, WO04032001A1
Maenga SR, Baekc N, Shinb SY, Choid BK (2003) A Z-Map update method for linearly moving
tools. Comput Aided Des 35:995–1009
Nakashima T, Ishihara Y, Niiyama K, Makinouchi A (2002) Development of stress analysis system
by X-FEM with voxel-type mesh. In: Proceedings of the ﬁfth world congress on computational
mechanics (WCCM V), Vienna
Park SC, Mukundan G, Gu S, Olling GJ (2003) In-process model generation for the process
planning of a prismatic part. J Adv Manuf Syst 2(2):147–162
Stifter S (1995) Simulation of NC machining based on the dexel model: a critical analysis. Int J Adv
Manuf Technol 10(3):149–157
DOI 10.1007/978-1-4471-4976-7_16-1
Page 52 of 54

Index Terms:
3D printing 24
Extended Z buffer method 9
Extended Z Map 16
In-process model (IPM) 12
Layered manufacturing 22
Machine tool 5
ModuleWorks™ 4
Numerical control (NC) 2, 4, 46, 49
optimization 49
simulation 4, 46
veriﬁcation 2, 46
Octree hierarchical space decomposition 21
QuickCNC GUI 26
Reverse post processing 7
Stick method 17
Uniform space decomposition (USD) 20
VERICUT composite simulation 25
Virtual machining 1
Voxel model 22
Z Map model 16
DOI 10.1007/978-1-4471-4976-7_16-1
Page 53 of 54

Author Queries
Query Refs. Details Required
Q1 Please check if affiliation details are okay.
Q2 Please provide details fo Refs. [17–19] in the reference list.
Q3 Please cite “http://www.engineeringchallenges.org” in text.
Q4 Please provide complete details of “http://www.engineeringchallenges.org”.
Q5 Please check if inserted page range for Chandru et al. (1995) is okay.
Q6 Please provide proceeding location of Kase et al. (2003).
Q7 Please provide publisher location for Liu et al. (1991).
Q8 Figure 58 is found to be poor in quality. Please provide better quality of figures if available.
DOI 10.1007/978-1-4471-4976-7_16-1
Page 54 of 54

Virtual machining chapter manufacturing technology handbook

More Related Content

What's hot (20)

Viewers also liked (15)

Similar to Virtual machining chapter manufacturing technology handbook (20)

More from Liu PeiLing (20)

Recently uploaded (20)

Virtual machining chapter manufacturing technology handbook