INTERNATIONALLY KNOWN MAKER OF HIGH-PRECISION
TOOLS USES FEA TO SHORTEN DESIGN PHASE
Harold Lawson, Jr., Project Engineer
Moore Special Tool Co., Inc.
Bridgeport, Connecticut
| Harold Lawson, Jr., of Moore Special Tool Co., Inc. |
Moore Special Tool Co., Inc., of Bridgeport, Connecticut, is
an internationally-known manufacturer of high-precision Jig Grinders,
Aspheric Generators, Universal Coordinate Measuring Machines,
and Metrology products.
After Moore began using the CADKEY 3-D design and drafting software
for some of their engineering work, they looked for other computer-aided
engineering tools, most notably finite element analysis (FEA).
"While we were looking for FEA software, one of our major concerns
was how the FEA programs would interact with the CAD software
that we had already invested in," recalls Harold Lawson of Moore.
"We needed a system that would be CADKEY-compatible. Not only
did Algor's FEA System have, at the time we purchased it, individual
modules available that could convert our CADKEY files, but their
CAD program, Superdraw II, had Import and Export commands incorporated
into it that let us transfer CADKEY files directly to and from
their system. They also offer training courses throughout the
US. This was another definite plus, since it meant that we could
learn how to use the software quickly and begin applying it to
Moore's designs right away."
Equally important for Moore in their decision to use Algor's
FEA System was the fact that Algor's computer analysis is so much
faster than the physical testing that Moore had been using.
As Lawson explains, "Before we began using Algor's FEA, the method
of testing a design was to go through the whole process of fabricating
the part and trying it out - but that still didn't tell us how
manufacturing tolerances would affect it. That is, I wouldn't
know what would happen if I ran plus or minus 10 or 15 percent
unless I were to go through the trouble of making parts that were
10 to 15 percent off the actual specifications and test them,
too. Unfortunately, that requires the use of a lot of personnel
and manufacturing time." Application of FEA at Moore will mean
a substantial time savings in their prototype cycle by enabling
them to come much closer to the optimal design the first time
around.
"Since we have very close tolerances, prototype testing created
a real bottleneck in the design process. Whereas most manufacturers
are not worried about millionths of an inch, we are."
The Problem
One part that Lawson has analyzed with Algor is a classic shrink-fit
problem involving a plain, thin ring gear having an inner diameter
of approximately six inches, and measuring 13/16 inch high with
a cross-section of 0.180 inch. This gear is shrink-fitted 1/3
of the way down from the top of a thick, slightly cone-shaped
cylinder. The contact pressure between the inside of the gear
and the outside of the cylinder has to be high enough to maintain
a friction torque; but if the interference fit is too high, the
gear and cylinder will deform. Moreover, the higher the interference
fit, the more the cylinder ID will deform. Moor's design goal
was to minimize the deformation of the inside cylinder while still
producing the required torque transmission between the cylinder
and gear. Lawson also wanted to know the tolerances and how they
would affect the torque transmission capacity of the gear.
"We could have just looked up this shrink fit in the ASME Handbook
and used their rules. But there was a problem: in the case when
a thin ring gear is being shrink-fitted onto a thick, hollow shaft,
the Handbook doesn't provide a way to determine the minimum amount
of interference fit required; there really isn't one," Lawson
relates. He could have attempted to figure it out using the torque
and contact pressure of the two surfaces, but this still left
yet another question: how much would the inside of the thick cylinder
deform due to the pressure from the ring gear? Says Lawson, "The
answer to a question like this is critical to us - is it 20 or
40 millionths of an inch?"
"Furthermore," he continues, "how much material do I have to
leave so that the toolmakers (who hand-fit and hand-assemble these
parts) can do an accurate job?" Moore uses CNC griders, lathes
and milling machines to manufacture their parts; but for hand-fitted
parts in general, a hand-lapping and hand-scraping process is
employed to remove material from two adjoining surfaces to create
a perfect match. The materials from which the gear and cylinder
will be fabricated are two types of steel: 4140 and 4150. Both
these substances withstand Moore's heat-treating very well.
FEA Analysis
The Algor FEA System was an indispensable tool that Lawson used
to help him design the shrink-fitted gear. He began by creating
the outline geometry of the model in Superdraw II which, he says,
"does everything you want it to do." Although it was only necessary
to model a "pie-section" of the gear/cylinder interface due to
its symmetry about the center axis and the uniformity of the loading
conditions, Lawson chose to model the full 360 degrees for documentation
and display purposes. To create the full model, Lawson had a choice
of generating the mesh using either Superdraw II or Algor's mesh
generators. He chose the latter method, linking the outline of
the model through MSHGEN and into RADGENBR, which automatically
generated 64 "pie-sections" of four elements each, creating a
total of 256 brick elements.
With AEdit, Algor's preprocessing editor, Lawson set boundary
conditions on the model and applied loads. He constrained several
nodes in the direction of the Z-axis to hold the gear fixed in
space. "You have to hold at least one node totally fixed, or two
of the nodes translationally fixed," he explains. "Otherwise,
the model is free to move around anywhere in space. Then, when
you apply the loads, they won't affect the geometry of the model
- they'll just be pushing it around in space." A uniform internal
pressure of 281 psi was also applied to each of the element faces
on the inner diameter of the gear with AEdit.
"Calculating the resulting displacements from the applied loads
was easy, using Algor's processor," comments Lawson. "All I had
to do was enter the file name and specify which options I wanted
the processor to use. Their program did the rest, using the model
that I had specified." The results of Algor's analysis indicated
that a radial displacement of approximately 0.0008 inch had occurred.
Knowing this, Lawson could then work backwards. When the displacement
of 0.0008 inch is reached, the contact pressure required to hold
the gear in place would be produced.
To verify the model, checking boundary conditions and ensuring
that displacements had occurred, Lawson used CPLOT. To view the
original and displaced models in 3-D after verification, he used
TDraw.
Lawson's analysis confirmed that Moore could use this gear on
the cylinder and that it would handle the 90 foot-pounds of torque
to which it would be subjected. "That torque will be transmitted
by the outside ring to the inner ring, and won't slip," he points
out. The Algor program also determined that the stresses, according
to the von Mises criterion, were 2500 psi - very low for the type
4140 steel that would be used. It turned out that the interference
fit was 0.0015 inch. Lawson says, "I spoke with the designer,
who said the 0.0015 inch interference fit was within safety and
manufacturing factors - and, in fact, was quite good. This means
that we could relax manufacturing tolerances from those that had
originally been anticipated, so they wouldn't be ridiculously
high."
| FEA was used to analyze the fit between the gear and
cylinder pictured. |
Future of FEA at Moore
"In the future, when we have more expertise in using Algor's
FEA System, we plan to use larger models created with CADKEY and
then converted into Algor's format through SuperDraw II's Files:Import
command," Lawson relates. "In particular, I would like to use
Algor to analyze displacements from castings, some of which are
simply due to gravity. We would like to predict the bowing due
to gravity or nonuniform loading of rolling members on the castings
under consideration."
According to Lawson, another area where Moore plans to apply
FEA is their 3-point support method for the Jig Grinders, Aspheric
Generators, and Universal Coordinate Measuring Machines. "Three
points of support are sufficient to prevent rocking," he notes.
"However if the support points do not form an equilateral triangle
and the loads are not evenly distributed, determining the best
configuration can be tricky. We'd like to use Algor's FEA to help
us with these support problems."
Lawson concludes, "After using Algor on the ring gear and finding
out not only how much it shortened the design phase, but also
how much more information it could provide for us, we intend to
use FEA as a tool whenever possible."
|