|
The MAS had its genesis
in 1996 during a discussion between Paul Wright & Charles Smith,
where it was conceived as a front end for the CyberCut Machining
Service. MAS would act as a filter for users to ensure the part
they wanted to make could be made with CyberCut's tools. The first
incarnation - version 1.0 - was Simon Brown’s masters project.
It was to provide a dynamic
Internet-based engineering reference tool to help an engineer choose
a viable manufacturing process. MAS 1.0 was written in Java to allow
it to run in a web page, providing easy access to those who wish
to study the capabilities and limitations of the various processes.
Once a process is chosen, web-based documentation was available
for each process.
The reference web pages
were created during a class project conducted by Professor Paul
Wright in the fall of 1996.
Click
here to go to the MAS 1.0 website.
However,
various problems were encountered by users:
- Problematic Facets
- Some of the facets used were not flexible or easily applicable.
For example “Life Expectancy” was either single use or infinite,
and it completely ignored loading or stress concentration —
two very important factors in fatigue life. “Level of Automation”
was redundant with batch size, with high batch size corresponding
to high automation.
- Hard Coded Data
- The capability data was fixed in the underlying code, thus
removing the ability to make easy updates to the data. In order
to consider commercial scenarios the MAS must run using data
from a separate database that is remotely accessible.
- Inappropriate
Process Categorization - Drilling, included as a process
in version 1.0, is too specific of an operation. At the conceptual
level, it is more fitting to place drilling into a “general
machining” category. Also, many processes with widely differing
capabilities - such as die casting and sand casting - were lumped
into a single category.
- Single Cost Facet
- The cost is treated as a per part cost. The tooling cost is
completely hidden from the user. Instead, the user is forced
to select batch size first. Thus small batches filtered out
processes with high start-up costs , and large production runs
removed processes with low start up costs. This obscured the
underlying tooling setup cost.
- No Material Selection
Criteria - Designers unfamiliar with engineering materials
are not given any guidance on material selection. This is a
problem, as the final mechanical performance of a component
is strongly driven by material properties.
- Binary Logic
- All of the selection rules are black and white, with each
process being acceptable or inappropriate. Without shades of
gray, it becomes difficult to select the best process if the
list of possible processes is large.
- Obscured Ranking
- There is no convenient way to find out why a process failed.
To find out, the user must back up through the facet specfication
process manually to find out which definition caused the problem.
These problems were used
as the starting point for version 2.0 of the MAS, which was rewritten
from scratch.
|
