Introduction:

The objective of this project was to perform analysis on a design of a carbon fiber semimonocoque. The monocoque is both the frame and the main aerodynamic feature that will be
used in competition by the Rose-Hulman Human Powered Vehicle Team in the 2016 competition
season. The analysis will be validated with physical testing and determined if this method is
appropriate for the design of this vehicle.

0 | Page

Model:
The 2016 human powered vehicle must be able to withstand several different loading
scenarios, put forth by the team as expected scenarios to encounter while riding. Several
different models were used to validate the analysis of the full vehicle. The full project schematic
can be seen in figure 1.

Figure 1. Project Schematic

Tensile Model:
The first testing was a tensile test of unidirectional carbon fiber samples. These samples
were carefully manufactured with simple rectangular geometries to ensure simple modeling. 6
samples, seen in figure 2, were tested on a tensile testing machine recording force and
displacement at the head. These samples had an average ultimate strength of 298 ksi with a
standard deviation of 34 ksi. All samples were tested to failure, and failed in a very consistent
1 | Page

way. Looking at figure 3, it is obvious the location of failure on the sample was very close to
where the tabs were located in, shown by the intact center fibers.

Figure 2. Unidirectional tensile samples during testing

Figure 3. Failed Unidirectional tensile samples

2 | Page

These samples each had a test section about 10 in. long and measured a head
displacement of 0.27 in. at failure. However the displacement of the head can be up to 50 times
greater than that of the actual carbon specimeni.
Tensile Finite Element Model:
The geometry was analyzed using shell elements in ANSYS 16.1 ACP Pre-Post processor on the
static structural solver. A zero displacement constrain was put on the end of the model in the
direction of pulling, and a 298 ksi pressure was applied to the free end. A single point was
constrained in the non-testing directions to prevent any chance of the part attempting to move
through space. Analysis found mesh convergence with a 5.0% threshold with 0.1 in. elements.
Failure analysis, seen in figure 4, matched the failure locations of the test specimens supporting
the accuracy of stress, strain, and Langley Research Center (LaRC) failure criteria.

Figure 4. Failure analysis of tensile specimen

It is worth noting that the displacement from the finite element model with failure
loading is 0.091 in., approximately a thirtieth of the measured displacement of the head. This is
however well within the up to 50 times displacement measured at the head. This complication
prevents any validation on the material stiffness from this model.
4 Point bend Model:
Using the layup materials that will be used for the final vehicle, 20, 15 in. rib samples
were made and tested in 4-point bending. The ribs are simply supported over a 8.0 in. span with
a 2 point top load applied across 4.0 in., centered between the simple supports. These ribs were
constructed out of 0.25 in. nomex honeycomb with a 0.0182 in. layer of unidirectional carbon
and a 0.0182 in. layer of weave on either side of it. A test rib can be seen in figure 5. Testing
found an average failure force of 247 lbf. With a standard deviation of 53.1 lbf. Average failure
deformation of these ribs was 0.189 in. with a standard deviation of 0.066 in. Since these samples
3 | Page

were not being tested in the fiber direction, the steel testing rig is assumed to deform a
negligible amount compared to the ribs and the value of head displacement is used for
deformation.

Figure 5. 0.25 in. 4 Point Bend test specimen

4 Point bend Finite Element Model:
The ribs were modeled with a stack up matching that in the previous section. The edges
of the core material were originally chamfered at 45 degrees to match the physical geometry of
the core material in the test samples. This was later removed as it didnt have a significant effect
on the results and is a feature of manufacturing the ribs. These samples were simply supported
in ANSYS and had a 247 lbf load applied across two lines where the 4 point tester would have
rested. The default material properties provided by ANSYS for our manufacturing method and
material was used originally, however this was found to make the ribs much stiffer than testing
showed. This analysis was used to determine the stiffness of our materials, which was found to
be approximately 22% of full material strength. Mesh convergence was found within a 5.0%
threshold with a 0.1 in. mesh sizing. This resulted in a deformation of the loaded area of 0.183
in, only 4.2% away from the physical testing.

4 | Page

Figure 6: Deformation in 4 point bending analysis

Looking at the deformations of this rib, seen in figure 6, originally seemed very peculiar.
The rib has a tendency to return closer to the horizontal plane after it passes the supports.
Looking at a simply supported beam of an isotropic material, figure 7, this is very unlike the
expected deformation. However this was found to be a characteristic of sandwich construction
composites since the core material is unable to carry any significant shear loading through the
materialii.

Figure 7: Simply Supported Isotropic beam

Rollover Protection System Model:
The Human Powered Vehicle Competition mandates that all entered vehicles have a rider
protection system in the event of a roll over. This must be able to support a 600 lbf top load with
less than 0.50 in of deformation at the loading. The human powered vehicle uses a roll bar that
is integrated into the rest of the structure as its primary form of protection. From past testing we
know that our roll bar stack up, seen in figure 8, and geometry, seen in figure 9, is sufficient to
meet this requirement, with only a 0.30 in. deformation under a 600 lbf top load. This test roll
bar was supported with a steel sub frame, similar to that within the actual vehicle, mounted to
steel plates, laid up into the roll bar itself.

5 | Page

Figure 8: Roll bar stack up

Figure 9: Roll bar modeled

Rollover Protection System Finite Element Model:
The Finite element model was built on the geometry for the 2015 Shannon-igans Vehicle
(figure 9). This model used the material properties determined from the 4 point bend testing
and the stack up seen previously in figure 8. This model used non-chamfered honeycomb core,
unlike the physical testing, to make the setup of this and future models easier. Deformation in
the testing rig were considered negligible so the model used fixed supports where the steel
plates held the roll bar to the testing rig. A 600 lbf load was applied along the center line of the
core structure, directed downward. Mesh convergence was found within a 5% threshold with
0.1 in. elements. This resulted in a 0.33 in total deformation of the loading surface, which is
within 10% of the testing, supporting the validity of the modeling methods. The deformed roll
bar was very similar to expected and can been seen in figure 10.

6 | Page

Figure 10: Top load Rollover protection system deformation

2016 Vehicle Model:
The 2016 Human Powered Vehicle was designed in a minimalistic fashion to conserve
weight. Additionally the placement of ribs is based on simple static loading theory, attempting
to maximize material where load is expected to be carried, within the constraints of the rider
geometry and comfort of use. The layout of the ribs can been seen in figure 11. The outer layer of
the vehicle is a single layer of weave oriented vertically and horizontally, as defined in the 4
point bend model. Each rib, excluding the roll bar, has a single layer of unidirectional material
oriented long ways with the rib, as defined in the 4 point bend model. The roll bas uses the
same stack up as defined in the Rollover Protection System Model. Then the entire ribbing
structure is covered in another layer of carbon weave, oriented vertically and horizontally. In this
model draping effects on the fiber direction were ignored. This assumption is relatively safe
since during manufacturing everything possible will be done to maintain the fiber directions
around the ribs of the vehicle, and elsewhere in areas like the nosecone or tail box, the stresses
are not very severe. These critical areas can be seen in figures 12.

7 | Page

Figure 11: Ribbing layout of 2016 vehicle

Figure 12: Critical loading

2016 Vehicle Finite Element Model:
The 2016 Vehicle was modeled using shell elements generated through ANSYS ACP preprocessor. Two different loading scenarios were tested in a static structural solver. The first
8 | Page

loading scenario was of a person riding the vehicle. This involved two remote forces, applied at
the two main contact points of the rider. A 270 lbf remote force was applied to the subframe
mounts, figure 13, where the rider would be pushing hardest on a crank. This value came from
simple testing with one legged leg presses of riders. Another 270 lbf force was applied at the
mounts of the seat, in the direction opposite that of the first in addition to 200 lbf load of the
rider sitting in the vehicle, pointed downwards. The rear wheel of the vehicle was fixed in the
scenario and one of the subframe mounts was restricted from moving in the vertical direction to
prevent the model from attempting to move through space or pivot around the rear wheel.

Figure 13. Subframe Mounting and loading

The original analysis of this loading had very obscure deformations at the mounting
points, which showed the mounts moving separately from each other. Since the structures
between these points are manufactured from steel and for this analysis considered to be rigid,
the analysis did not reflect the desired loading. This was addressed by coupling the mounts of
each piece together. This required that the deformations of every node was the same within
those coupled areas. This yielded much better results where the surfaces all moved as a single
unit seen in figure 14. With a 5.0% mesh convergence threshold the model had a maximum
deformation of less than 20 thou. Preforming failure analysis resulted in the expected, no
surface of the vehicle was remotely close to failure.

9 | Page

Figure 14: Deformation under simple riding

Using what I learned from the simple riding scenario I went to modeling an intense
impact that the vehicle undergoes when it impacts a speed bump as it did during last years
competition. Assuming the front wheel hits a 45 degree incline at 35 mph and the impact lasts
0.1 s, there is approximately a 3800 lbf force direct upwards through the subframe mounts and
into the rider. The seat mounts were set at fixed point and the rear subframe was constrained so
the mounts could not expand or compress.
Results and Discussion:
This intense loading model found a maximum deformation of 0.53 in. in a location that
was only a few inches away from the fixed point of the seat. At this point the small deflections
assumption could be debated however the solver does not report any problems. Looking at the
stress contours in figure 15, the stresses around that location are significantly outside of the
material strength. Failure analysis, seen in figure 16, showed multiple modes of failures in the
area surrounding the seat mounts. The seat of the 2016 vehicle was significantly closer to the
loading than it was last year so the analysis was repeated using the mounting points for the
2015 Shannon-igans, superimposed onto this vehicle, which is known that it could handle the
intended loading scenario repeatedly. The failure analysis of this model, seen in figure 17,
showed significant failures in the same location. The area is significantly reduced but this shows
a clear flaw in the load scenario as it was set up.

10 | P a g e

Figure 15: Principle Direction Stress under speed bump loading

Figure 16: Failure analysis of 2016 vehicle under speed bump loading

11 | P a g e

Figure 17: Failure analysis of 2016 vehicle with modified seat mounts under speed bump
loading
The vehicle as it is currently designed is well within the standard operation parameters
and the rollover protection requirements of the competition. I expect this to perform well as
designed in the competition. The impact loading scenario needs to be reevaluated for accuracy.
It would likely be improved with a transient analysis and more accurately defined loadings.

12 | P a g e

i Value from ASTM D3039 Testing standard

ii Information on composite cantilevers from Dr. Simon Jones