General Definitions
- Industrial Engineering
- "The branch of engineering that is concerned with the efficient
production of industrial goods as affected by elements such as plant and procedural
design, the management of materials and energy, and the integration of workers within the
overall system" -- The American Heritage Dictionary of the English Language, Third
Edition. Industrial engineers try to make systems run more efficiently.
- Operations Research
- "The use of quantitative models to analyze and predict the
behavior of systems that are influenced by human decisions" -- Introduction to
Operations Research, by J.G. Ecker and M. Kupferschmid, John Wiley & Sons, 1988.
Operations research tools are frequently used by industrial engineers, and include
queueing and simulation models.
- Queueing Model
- A mathematical representation of a system characterized by customers
(or jobs) waiting in a queue for service. Queue is the British term for a waiting line
(e.g., 'we waited in the queue at the post office for 20 minutes today.'). Queueing models
can be used to estimate the long term performance of manufacturing or communication
systems, without employing full-scale simulation models.
- Simulation
- "The process of designing a model of a real system and conducting
experiments with this model for the purpose of understanding the behavior of the system
and/or evaluating various strategies for the operation of the system" -- Introduction
to Simulation Using SIMAN, by C. D. Pegden, R. E. Shannon and R. P. Sadowski, McGraw-Hill,
1990. A simulation model is a representation of a real system.
- Semiconductors
- "Elements in the periodic table that have the property
of high conductivity at high temperatures and nearly zero conductivity (property of an
insulator) at low termperatures. The most prominent of these semiconductor materials are
silicon and germanium, which are located in the fourth column of the periodic table"
-- The Microprocessor: A Biography, by Michael S. Malone, Springer-Verlag, 1996/. For more
details (in non-technical form) see the Lexicon of Semiconductor Terms
provided by Harris.
- Wafer Fabrication
- A wafer is a piece of silicon used to fabricate semiconductor
chips. Wafer fabrication is a complex process by which circuits are layered onto silicon
wafers through several repeated sequences of operations. When completed and tested, wafers
are broken down into individual chips.
- Capacity Planning
- Capacity planning, in its broadest sense, encompasses all decisions
about what products a company can and should produce, and what facilities will be required
to produce them. Due to the tremendous capital burden carried by semiconductor
manufacturers, these decisions often prove to be pivotal. Semiconductor capacity planning
includes long-range business decisions and shorter-term more tactical decisions.
Queueing Models for Manufacturing
Semiconductor wafer
fabrication facilities (fabs) are extremely complex and expensive systems, and are the
heart of a highly profitable sector of the economy. Managing them, and understanding their
dynamics, is a challenging endeavor. This is because the factory environment, in addition
to being complex, is highly variable. There are often many products, each of which has
several hundred processing steps, with dozens of different kinds of processing equipment.
The industry also suffers from notoriously unreliable equipment, and frequent product mix
changes.
A particular problem for semiconductor manufacturers is planning
capacity in this unreliable environment. Static models are commonly used (primarily
spreadsheets). A typical spreadsheet model calculates either the maximum amount that a fab
can produce for a given mix and toolset, or the correct toolset that will produce a
pre-specified mix and volume. Spreadsheet models generally include historical loss factors
to account for time that the equipment is unavailable due to failures, setups, operator
delays, rework, and other capacity loss factors. However, these loss factors often become
inaccurate when the product mix changes. Also, static models do not include dynamic
elements such as resource contention, and cannot predict cycle times or work-in-process.
Simulation models are sometimes used as a parallel effort to
estimate cycle times. However, these estimates are not often integrated directly into the
capacity planning process. This is a problem, because cycle time is often the most
important performance measure for a fab. Also, because the simulation models are large and
detailed, they typically require long run times (on the order of an hour per replication
for a long simulation of a large fab). This discourages their use for what-if analysis and
optimization schemes.
Queueing models offer a potential compromise between
spreadsheets and simulation. Because they yield analytical solutions, they are typically
much faster than simulation models (on the order of 2 minutes to analyze a large fab).
They also capture some of the dynamic elements missing from spreadsheet models. My
research is in understanding the applicability of queueing models to semiconductor fabs,
and developing improved models in certain areas.
Integrated Simulation, Capacity and Cost Modeling
"Cost analysis is often treated as a separate task from capacity
and simulation modeling activities. Several important factory-level
cost performance measures, however, rely on detailed capacity analysis calculations. In
fact, much of the difficult groundwork for making these calculations has already been
completed in existing capacity and simulation analysis tools. Frank Chance and I (in an article for the WWK newsletter) argue that these activities fit quite
naturally together, and give a more complete picture of factory performance than isolated
analyses. Also, capacity and simulation analysis tools are increasingly used to support
strategic and tactical business decisions. These decisions are most often framed in terms
of their impact on the bottom line. Therefore, effective decision-support tools must speak
not only in terms of capacity and cycle time, but also in terms of dollars. We believe
that making decisions in an integrated cost, capacity, and simulation analysis framework
offers several advantages over isolated analyses."
Understanding and Improving Capacity for Semiconductor Facilities.
"Planning capacity accurately is critical in today's highly
competitive semiconductor industry. Equipment and
facility costs are rising, integrated circuit technologies are changing rapidly, and
customers are demanding ever faster chips. And they want them delivered on time, at lower
costs. With most equipment costing several million dollars each, capacity planning
decisions have an immediate impact on the bottom line. Understanding capacity is critical
to maintaining profitability over time. Miscalculation of a fab's capacity results in
serious penalties. If a factory is planned at a lower level than its actual capacity, the
result is lost revenues, increased cost of goods sold, and a potential for market share
loss. On the other hand, significant negative consequences can stem from overloading a
factory. These outcomes include long cycle times, missed delivery dates, excessive
inventory, and possibly lower yields."
|
