| Current Research Projects
| Special Project Research | Previous
Research
Previous Research Projects
Development of a Simulation Modeling Discipline for
Manufacturing Systems Design, Analysis and Control
Integration of a Simulation Platform with Manufacturing
Execution System (MES) Software
Integration of Rough Cut Analysis
and Design Tools with Detailed Simulation
Web-Based Manufacturing Simulation
Modeling an Electronics Assembly Production Line
Warehouse Design Visualization
Previous Related Research
Before the establishment of the Virtual Factory Lab, faculty and students
in the School of Industrial and Systems Engineering were active in a variety
of research projects related to virtual factory technology, modeling and
simulation. The Virtual Factory Lab has consolidated the themes behind
these research projects, described below. (Links to be provided soon.)
Object-Oriented Simulation in Manufacturing (OOSIM)
Process-Interaction Manufacturing Simulation (PIMSIM)
Logistical Analysis of a Cluster Tool
Models of Electronic Assembly Devices
Models of an Electronic Assembly Production Line
Computer-Aided Troubleshooting in Electronics Manufacturing
Process Optimization for Printed Circuit Card Assembly
Engineering Workstation for Design of an Automated Guided Vehicle System
Discrete-event simulation is a powerful tool for the design and analysis
of manufacturing system. The tremendous variety of simulation tools
in the commercial software market points to the popularity of simulation.
In fact, existing tools are excellent for developing relatively simple
models of systems which reflect behavior analogous to a network-of-queues
abstraction. At the same time, the variety of tools also points to the
conclusion that there is not a standard way to model a complex system.
This results from the inherent difficulty in using existing tools to
represent complex manufacturing behavior. Thus, simulation projects
are often beset by lengthy model development times, difficulties in
model validation, and difficulties in maintaining models over time.
The purpose of this project is to develop a disciplined approach to
represent manufacturing systems for purposes of detailed modeling. The
fundamental ideas are (i) to develop a manufacturing reference model
which serves as the basis for developing manufacturing-specific modeling
abstractions, (ii) to use principles of modularity and one-to-one mappings
in model development, and (iii) to provide constructs which explicitly
represent the manufacturing control system.
Discrete-event simulation is useful for representing the dynamic physical
behavior of a manufacturing system. However, it is typically the case
in a simulation project that the end result (the simulation model) is
not used when the engineers design the manufacturing control system.
This results from the lack of an understandable mapping of the control
logic used in the simulation model to that used in the actual control
software which will run the factory.
In this project, we seek to integrate a simulation platform with a
commercially available manufacturing execution system. The simulation
represents the physical behavior of the system (material processing
activities, material transport activities), while the MES performs real-time
tracking and control of work-in-process inventory. The fundamental ideas
are (i) to develop an environment which supports concurrent design of
the physical system configuration and control software, (ii) to provide
a "virtual factory platform" to assess system performance
using the MES software which will actually run the factory, and (iii)
to provide a platform which can be used to test the effectiveness of
various MES software packages and to develop ideas for improvement in
MES functionality.
A variety of different software tools exist to aid the manufacturing
engineer design and reconfigure a manufacturing system. These tools
vary in the amount of detail included in the model. At the same time,
there is a trade-off between the amount of detail included and the model
development time. For example, queueing network analysis packages enable
rapid model development and solution, but incorporate many assumptions
which do not reflect actual system behavior (e.g., unlimited buffers).
A detailed simulation model, on the other hand, takes significant time
to develop, but can be potentially as accurate a reflection of system
behavior as the modeler wants. Several questions arise. First, when
is it appropriate to use one modeling approach over another, and can
this question be answered in a quantitative manner? For purposes of
this project, we are more concerned with the question, can these tools
be integrated in such a manner that both can be used at different phases
of a large-scale project and at the same time have consistency between
the model results from each?
The purpose of this project is to develop an integrated environment
which supports queueing network analysis for rough cut design decisions
and detailed simulation modeling for more detailed design and analysis.
The rough cut design decisions, of course, constrain the solution space
of decisions which one may make with more detailed simulation modeling.
And there may be iteration between the two phases.
This project uses a manufacturing reference model to develop a generic
data structure for manufacturing systems. This data structure then serves
as the basis for a queueing network model and for a detailed simulation
model. This project builds on the same basic concept used for the AGVS
Engineering Workstation, developed at the Material Handling Research
Center at Georgia Tech.
Graduate student: Eda Kemahlioglu
Faculty: Douglas A. Bodner and Leon F. McGinnis
The World-Wide Web holds tremendous potential for enabling the rapid
dissemination of information and for facilitating distributed business
decision-making. This type of environment has the potential to improve
the deployment of simulation technology. For example, a simulation model
may reside on a computer at one location, and it can be accessed by
a client machine at another location via the Internet. A web browser
can enable the user at the client location to execute the model and
see its results. Obviously, these
This project is utilizing the World-Wide Web as a vehicle to facilitate
distributed use of simulation models for decision-making. The key activities
include:
- development of a CORBA interface between the client and server applications;
- creation of understanble and intelligent interfaces which aid the
user in understanding, executing and reconfiguring the model and in
understanding output analysis;
- development of the necessary Java programs for the client and server;
- development of modular simulation models which support reconfigurability
for what-if analysis by the user via a web browser;
- studying and addressing network-related performance issues with
respect to model execution.
Electronics assembly is an increasingly important segment of today's
manufacturing systems. As electronic products continue to proliferate,
the importance of this segment will continue to grow. This project seeks
to model an existing assembly line (i) to gain insight into generic
simulation model fragments which can facilitate the model development
process, (ii) develop visualization of process operations to aid in
model validation, and (iii) test various work-in-process control strategies
in terms of overall system performance. The software tools being used
for model development include QUEST and Virtual NC by Deneb Robotics.
This project extends work previously done in modeling an electronics
assembly line.
Graduate students: Prashant Bellur and Mark Insalaco
Faculty: Douglas Bodner, Marc Goetschalckx T. Govindaraj and Leon
McGinnis
Warehouse designers make a number of design decisions such as layout,
equipment and technology selection, and staffing levels. In making these
decisions, they must manipulate a large amount of data on products,
equipment labor and facilities. The purpose of this project is to develop
visualization tools to aid designers in this process. Key activities
include:
- Provide a user interface that facilitates interactive design and
what-if analysis;
- Automate algorithms in a computational form to aid analysis;
- Specify a generic object model for design visualization;
- Develop visualization requirements, evaluate existing tools with
respect to their functionality, and select a tool or tools for use
in the project.
Currently, the project is focusing on the use of Visio™ to aid
with layut design.
|
