Tracepro Tutorial Tom Brokaw

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Coupling FEA Analysis and Solid Model Ray Tracing to look at

Focal Plane Deformations


By Tom Brokaw
OPTI 521 Tutorial

Model of a 5 foot diameter solar collector (F1 parabola), a 1 ft diameter receiver is located 55 from the base of the collector (red)

Abstract
The purpose of this tutorial is to demonstrate a method for analyzing optical systems that are subjected
to real world loading conditions. To do this, two different software packages are used together. First,
the finite element analysis (FEA) tools within SolidWorks are used to load a model. Then TracePro will
be used to look at how rays will propagate thought the loaded system. In this way, the changes to the
system performance under the loaded conditions can be observed. There are two key aspects that
make this type of analysis possible. First is the solid model ray tracing that is available in TracePro.
Second is the ability to build a new solid model based on the results of an FEA analysis within
SolidWorks. It should be noted that this is a new feature within SolidWorks, and did not work as well as
expected. It is probably too early to be using the methods described here for serious optical analysis.
However, because it is a new feature, it is my hope that they will continue to improve on its
functionality, and that this will soon become a powerful tool for designers. Due to its potential
applications, I believe it is worth studying so that as the tools improve, they can be utilized.

Background
Solid model ray tracing is based on the interactions of solids and sources. This is different from
conventional surface based ray tracing in that you are dealing with actual solids. In addition, most solid
based ray tracing is non-sequential. This means you dont have to define an optical order; rays just keep
going until they encounter a surface to interact with. For this tutorial, TracePro 5.0 by Lambda Research
was used to trace rays through the system. In TracePro solids are built, or imported in the ASCII (*.sat)
format. Once the solid objects are in the model, individual properties are applied to the solids and can
also be applied to the corresponding surfaces. Sources are then defined, and rays are traced through
the system. Whenever a ray encounters a surface it can be transmitted, reflected, or scattered (both
forward and backward). The distribution of the energy in the ray and how it acts is based on the surface
and solid properties at the interaction. TracePro uses a modified Monte Carlo analysis to converge on a
solution for the system.
Finite element analysis (FEA) is becoming increasingly popular and accepted as a method of testing
equipment at the design phase to understand performance of the end product. There are many
potential problems with using FEA in this way. Primarily, there are the problems associated with bad
initial conditions, and loading / constraining the model. Addressing these problems is beyond the scope
of this tutorial; it is assumed that the designer will be able to constrain and load the model in a way that
accurately represents the conditions that are being modeled. Once a loaded model converges at a
solution, the designer has access to the models Deformed Geometry. This is a representation of how
the model will look and behave under the loads that have been applied (the default display will usually
have a large scale factor applied to it so that small changes can easily be seen). In many optical designs
this deformed geometry could alter the system performance. For example, the expansion of a plastic
lens due to a temperature change could change the power of the lens, or add aberrations. Likewise, the
self weight deformation of a mirror could alter the overall system performance. There is a new option
within SolidWorks that will build a new solid body based on the deformed shape. This new deformed
body becomes a new configuration in the existing model.
It is possible to build an entire optical assembly in SolidWorks. This assembly can then be imported into
TracePro. Once in TracePro, the properties of the materials and the surfaces can be defined and sources
can be added to the model. Rays are then launched from the sources and propagate through the model.
This provides a good analysis of what is happening in the system, and also a very good visual
representation of how the light is behaving in the system. Taking this to the next level, some or all of
the components of the system can be analyzed using the FEA tools in SolidWorks. New solid bodies can
then be created based on the deformed shape that resulted from the analysis. These new bodies can
then be imported into TracePro, and the same material properties and sources can be applied. The raytrace can then be rerun with the new shapes, and the results compared to the unloaded case.
Method
In order to illustrate this capability, I have modeled a simple F1 parabola that will be used as a solar
collector. The solar energy field would be very well suited to this type of analysis. Most solar energy
applications use non-imaging optics, and are therefore only concerned with the large first order
problems. This type of analysis could easily show how the light was missing a receiver, or any hot spots

that resulted from light concentration. This field is also very concerned with cost, and less expensive
materials are commonly desirable for light collection.. These materials are often very susceptible
suscep
to
large deformations under normal loading conditions, and novel supports may be required.
In this example, a 5 foot diameter dish was created
created out of parabolic segments. Each
E
segment has a 60
focal length, and comprises about 60 degrees of the total
tota dish. The segments are made out of 1/8 thick
mirrored acrylic. The six segments are supported by ribs with the same parabolic shape located at the
seam where two segments meet. A one foot diameter receiver was place
placed on the optical axis 55 inches
from the back of the collector. This type of collector would be placed on a two axis solar tracker so the
sun was modeled as directly on the optical axis.
axis. The following results were obtained in TracePro for this
assembly (in the illustration, only about 1% of the total traced rays are being displayed).

As expected, all of the rays striking the dish are focused onto the receiver. The next picture shows an
irradiance map for the front surface of the receiver. In this case the sun was modeled with an initial
irradiance of 1 kW/m^2.

Again, as expected, the concentrated energy is fairly evenly distributed on the receiver. It is reasonable
to assume that the 1/8 inch acrylic would deform a considerable amount if it was only supported at the
edges. This deformation
formation would have an effect on the mirrors
mirror s performance as a solar collector. The first
step to modeling this is to look at the self weight deflection of one of the segments. As a side note, this
was a fairly labor intensive analysis. For any bent geometry,
ge
you need at least three elements
throughoutt the thickness. On the outside of the bend you need an element in tension, and on the inside
there is an element in compression.
ion. There needs to be at least one element in between for the
transition. Thiss makes for a very large number of elements
elements on one segment since the surface area was
much larger than the thickness. However, I was able to get the FEA static analysis in SolidWorks to
converge and
nd provide the following result:

Once the result was obtained you can right-click on the


Results in the feature tree and select Create Body from
Deformed Shape as shown on the right. The new part is then
created as a configuration of the part that you are working on.
You can then use this configuration in any assembly, or export it,
just like the original.
In this case the model was deformed by 250m, this should be
more than enough to see an effect in the raytrace. However
when an assembly was built based on this deformed geometry,
the results from the raytrace were the same. This is due to an
unfortunate current limitation of this tool. When SolidWorks
builds the new body, there is some curve fitting that it performs.
If the error is very small it gets smoothed out in the new body.
Hopefully this will become an option in future versions so that
you can save bodies that will maintain small errors.
As another side note, I was initially unsure where the
smoothing was occurring: whether it was in SolidWorks, or
when the model was imported into TracePro. I was able to use

some of the evaluation tools in SolidWorks to look at the deformed bodies and see if the deformation
was really coming through. The zebra stripes analysis in particular was very useful for this. With the
zebra stripes turned on, you can rotate the model and look for changes in the curvature. Using this, I
could easily see that the deformed shapes were coming out smooth.
For the sake of this tutorial, and to see if this method would ever work, I added a force to the front
surface of the segment and reran the analysis to see if I could get a deformed body to save. I added
100lb force to the front surface and obtained the following result.

Now at last, with 2.5cm of deformation in the middle of the segment, the error stayed with the model
when the deformed shape was built. A new assembly was built using the deformed configuration of the
segment, and imported into TracePro. On the next page is a picture of the ray trace, and an irradiance
map that came out of the analysis. As you can see now some of the rays are missing the receiver, and
there are some hot spots from the concentrated light.

Conclusion
The current limitation that was discovered in this tool was discouraging; however, it is a new tool and
hopefully they will continue to improve on it so that it will be useful for this type of optical analysis. The
results are important, but the visual nature of the raytrace and results would also be very useful. The
ability to see the rays and what went wrong would be very helpful especially when trying get approval
for a design from people with little or no optical training. I believe there is a lot of potential for this type
of analysis, and it is something worth keeping an eye on as future upgrade are released.

You might also like