This document summarizes an experiment conducted to determine the significant parameters affecting the shear strength of hot-air welds between polypropylene rod material and twintex composite material. The experiment tested different welding temperatures, feed rates, and fan speeds using a designed experiment approach. Samples were tested and data was analyzed to make recommendations for further testing.

1. Final Report: Hot-air Welding Twintex
Objective: Find significant parameters affecting shear strength when hot-
air welding twintex using polypropylene rod material.
Austin Scheffer
Winter Quarter 2014
ITVD Independant Study
Introduction:
This is the final report for the independent study project conducted this quarter on hot-air
welding twintex. Here is the proposal description as noted in the independent study proposal.
This project will be intended to be a self-guided exploration and application of
designing experiments. Polypropylene hot-air welding was used extensively on the

2. fabrication of the hybrid bus chassis. While there are many sources for tensile strength
of polypropylene to polypropylene hot-air welds, no data found on hot-air welding
polypropylene to Twintex, a fiberglass/polypropylene composite. Using textbooks Design
and Analysis of Experiments and Engineering Statistics as well as student and advisor
help, an experiment design will be selected and implemented to gain understanding of
the hot-air weld strength of polypropylene to twintex.
The final report includes all previous components of the independent study and is compiled into
one document for easier understanding of process and reporting.
Here is an outline of the following documentation:
1. One to two page research paper on how to conduct a properly designed experiment.
2. One to two page paper on conducting a properly designed experiment of hot-air welding
of twintex to polypropylene written in proper lab format.
3. Testing results
4. Data analysis and recommendation for further testing.
DOE Explained
Design of Experiments is an experimentation method used to provide data analysis in an
extremely efficient manner.
DOE applies the use of statistical analysis to gain more information from testing data in fewer
testing samples, saving time and money.
Traditional methods of testing evolve around the concept of OFAT (one factor at a time). In the
article “Eight Keys to a Successful DOE” written by Mark J. Anderson and Sheri L. Kraber,

3. observe DOE and OFAT to be really a difference of “parallel processing” vs. “serial processing”.
The two-level factorial design takes advantage of multiple variables being changed at the same
time, while also still being able to pull out key findings and connections between variables
individually.
DOE styles of experimentation require much more planning up front, but this planning can
produce much more efficient results in the long run if done correctly.
Various steps to proper DOE include:
1. Deciding on a Specific Objective
a. What do you want to learn?
b. Three different types:
i. What factors are important? (Ex. does this factor have an effect on
surface quality?)
ii. What are the relative effects of certain factors? (Ex. what happens if I
change this variable?)
iii. What is the best combination of factor levels? (Ex. what are the optimum
settings for the desired results?)
2. Quantifying Results
a. To have reliable data you must quantify your results. Make a list of the factors
that are involved with the process you are testing, then decide how they each
can be reliably quantified so that the process can be repeatable by others.
3. Randomizing Run Order
a. This will eliminate test bias, something uncontrolled that may affect results of the
study. Some examples are tool wear, ambient temperature, and changes in raw
material.
4. Keep Testing Materials and Procedures Consistent (Block Known Variations)
a. Using the same material for every sample, and creating consistent procedure
guidelines are imperative to achieving consistent testing results. Theoretically
everything but the factors you have chosen should be held the same throughout
the test.
5. Identify Aliased Effects (if any)
a. When choosing to do a fractional factorial design instead of a full factorial design,
some of the effects will be combined, according to an article written by Penn
State University. When trying to cut certain interactions between factors “out” in a
sense you aren’t really taking them out, you’re just not measuring them. This is
important when analyzing the data because some interactions may need more
research to know what is actually happening, since a full factorial design wasn’t

4. chosen.
6. Planning Ahead
a. This is probably one of the most important parts of designing experiments. I have
learned that the whole point really is to plan ahead. All of these other criteria
need to happen BEFORE you conduct your experiment, thus requires planning
ahead. How are you going to eliminate outside factors? How are you going to
keep your tests consistent? These and many more questions need to be
answered ahead of time.
7. Replicate to Dampen Noise
a. The more times you replicate your experiment the better your results will be. This
becomes a judgment call on the engineer’s part. If the results are highly critical to
be within a certain standard of accuracy, then more replications will need to take
place. Replications can be done with each sample group within one test, and the
whole test can be replicated multiple times with different people to achieve even
more accurate results.
8. Confirm Findings With More Testing
a. Testing is never finished, it just reaches a satisfactory level for the ones who
need the information. More testing can always be done to better understand the
process and interactions taking place within the test.
DesignedExperimentSelectionand Implementation:
First all variable factors of the hot-air welding of twintex were brainstormed and
analyzed.
List of Factors/Inputs:
● Feed Rate/Weld Speed (inches/sec)
○ How fast you move welder along plays a key role in how strong the weld is. As
the welder is pulling the polypropylene rod through the tip and heating it up to just
over the plastic deformation temperature.
● Weld Temperature (degrees F)
○ As the welder is pulling the rod material through the tip it is being heated up. The
temperature at which the tip is set it is then very important to the strength of the
weld. The temperature also controls how hot the air is while being blown onto the

5. parent material to heat up the area before welding occurs.
● Downforce/Pressure (lbs)
○ While moving the welding gun over the affected area there will be some amount
of pressure downward onto the weld, pressing the rod material into the melted
parent material. If pressure is too high then the melted rod material will be
pressed out of the weld area. If the pressure is too low then the rod material may
not bond well with the parent material.
● Welding Angle
○ Looking from a side view of the weld line, the angle at which the welding tip
interacts with the weld line may be a factor in how well the parent material is
heated up before the rod material is introduced.
● Type of Material (how it was processed)
○ Different types and weights of Twintex may have varying ratios of polypropylene
to fiberglass. Different manufacturing methods (hot-rolled or vacuum-pressed)
may influence where the polypropylene is located in the layers of fiberglass.
● Fan Speed (low/med/high)
○ How fast the hot air moving out of the welder and onto the parent material may
affect the strength of the weld. If the speed were too high, the melted
polypropylene on the surface may be blown off to the side. If the fan speed were
too low, the material may not be heated up as fast as it needs to be to make a
proper weld.
Next, factors were chosen to conduct experimentation upon. Due to time and budgetary
constraints not all factors could be included in testing. The variable factors chosen were
chosen due to their ability to be controlled easily and because they were hypothesized,
after consultation with colleagues, to have the most potential effect on the shear strength
of the weld:
● Fan Speed
○ This variable is easily controlled by the settings on the hot-air welder itself. The
temperature can be changed and adjusted easily and reliably without much
effort. Fan speed is also very important because it is transferring the heat from
the welder to the parent material.
● Temperature
○ Temperature can also be easily controlled by the settings on the hot-air welder
itself.
● Feed Rate
○ With some set-up, feed rate could be controlled by using the shopbot’s x and y
axis motors, which can be set to precise jog and move speeds which are needed
for cutting different materials. The only challenge would be securing the welder to

6. the motor in some way that the welder would be unhindered while welding, yet
still stay attached securely so as to not slip while conducting tests.
For each factor, parameters were chosen for highs and lows in a multi-factorial designed
experiment.
● weld temp = 400 deg. to 500 deg.
○ test high temp at lowest feed rate and highest fan rate to make sure it doesn’t
degrade
○ hottest weld achieves higher weld penetration
● feed rate = 6 in per min. 9 in. per min
○ polypropylene rod material just straight polypropylene, no insulative material
added like in twintex, so it will melt at a lower temp.
● fan speed
○ med/high
○ still undecided if I should do med. and high, but low and high will make greater
variance, and will produce easier distinguished results
Test Procedure
1. Cut out individual cards, using white pre-processed twintex (get processing info from
Will)
a. 1 long strips 4” wide
b. cut squares out of strips, 4” x 8”
c. clean off all edges of the twintex so that the edges are clean of debri
i. this will allow maximum proximity during testing when you place the two
cards together to weld
2. Cut strips of polypropylene weld rod (flat), 14” long (need extra for feeding through
nozzle)
a. strips do not need to be an exact length, just around 14”
b. slice strips with an exacto knife once firmly across top, then bend the rod until it
snaps.
3. Once cards and rod are cut, place two cards on the shopbot table as shown in Figure 1.
a. 1. so that weld path where the two cards meet is more or less parallel with either
the x or y axis of the shopbot.
b. 2. place the cards together to make an 8” x 8” square.
c. 3. be sure to place the edges as close together as possible. There should be less
than or equal to a 1mm gap between the two cards along the weld path.

7. 4. Secure cards to the table as shown in Figure 1.
5. Clamp a 2”x14”x.25” piece of steel to the shopbot apparatus.
a. clamp with enough pressure to support the hot-air welder at the end of the piece
of steel.
b. dimensions need not be accurate, just make sure the piece of steel is parallel to
the plane that you will be welding your cards on.
6. Secure the hot-air welder to the end of the piece of steel as shown in Figure 2 and
Figure 3.
7. Perform test runs in randomized order.
8. Using the table saw, cut cards into .75” wide strips for tensile testing.
a. Discard the first and last two strips.
i. This is done so only the strips with the greatest weld consistency are
used in the test.
b. Keep all the strips from each card together and label with the appropriate group
number.
c. Clean off all strips’ edges so debri is not brought into lab testing area.
9. Using and MTS machine, measure and record the shear strength of the individual strips
in each group.
a. Record each strips measurement, then average the strips peak load numbers for
each group to use when analyzing data
10. Using excel, create an ANNOVA table, running regression simulation using the statistical
analysis toolpack.
11. Analyze your data, pulling out significant findings and worthy comparisons.
Next I have included a table to show my runs (2^3=8) and the high and low factors
I went with after preliminary testing. The column farthest on the right shows
which order the run was performed (1st, 2nd, 3rd…)
Run Temp. (480/530) Feed Rate (6/9in
per min)
Fan Speed
(med/high)
Order of
run
1 -1 -1 -1 1
2 1 -1 -1 7
3 -1 1 -1 2
4 -1 -1 1 6
5 1 1 1 4
6 -1 1 1 3
7 1 -1 1 5

8. 8 1 1 -1 8
Data Analysis
For the first ever DOE experiment conducted, results were seen as very favorable. Some values
to take note of are (please refer to Figure 4):
● The adjusted r-squared value is decently close to 1, which means that there is a factor/s
affecting the data.
● Significance F is also close to zero, which means the data collected is reliable and can
be trusted for analysis.
● The P-values of Temperature, Fan Speed, and Feed Rate were all at or below .05
(highlighted in green), which means each of those factors are significant.
● Next, looking at the coefficients (the column on the far left hand side in Figure 4) for
each of those factors we can see how much each of those factors are affected when
they are changed.
○ The coefficients are then multiplied by two for the actual effect, which in this case
is measured in lb-force.
○ We can see that the most significant factor was feed rate, with a change of 53.5
lb-force (highlighted purple in Figure 4).
○ temperature and interaction ABC (all factors high or low) were about 20% lower
in magnitude in the 40lb-force area.
○ fan speed had the lowest of the significant interactions, with about 40% in
magnitude less interaction in comparison to feed rate.
Generalanalysis seems to indicate:
● Better welds are produced at lower feed rates, higher fan speeds,
and higher temperatures.
○ This makes sense empirically, but it is also good to have solid
data to back up those numbers.
Recommendationsfor Further Testing
For further testing from a manufacturing point of view, more testing could be done to
see how fast welds can be produced while not compromising shear strength. One could
optimize three different feed rates, finding the best fan speed and temperature settings
for each of those feed rates, and then compare the shear strength data to see if the
feed rate has an effect. If feed rate doesn’t affect weld strength, faster welds could be

9. produced, saving a company more cost in the long run.
Further testing could also be done with downforce, welding angle, and type of
processed twintex. From visual observation of the surface of the twintex used in this
study, micro-sized surface voids were seen. These could be gaps of polypropylene
along the surface, which could be a factor in the weld strength if there is less
polypropylene for the weld rod to attach to. When comparing to vacuum-bagged
processed twintex in-house, those same micro-surface voids are not seen. Testing the
two materials may lead to findings that could support this hypothesis.
Different surface preparation may also be a factor worth testing. If more of the fiberglass
strands were exposed to the rod material while welding, there may be an increase of
strength, providing the rod material still adheres well since there will be less
polypropylene.
Figure 1: Test Card Securement and Alignment

10. Figure 2: Welder Securement
Figure 3: Welding Test Set-up

11. Bibliography
Anderson, Mark J., and Sheri L. Kraber. "Eight Keys to Successful DOE." Eight Keys to
Successful DOE. Quality Digest, 1999. Web. 18 Mar. 2014.
Anderson, Mark J. "Trimming the FAT out of Experimental Methods." Stat-Ease Inc, n.d.
Web. 10 Mar. 2014.
"Fractional Factorial Designs Involve Aliasing of Effects. What Does This Mean? | The
Methodology Center." The Methodology Center. Penn State University, n.d. Web. 18
Mar. 2014.
~Special Thanks to~
Chris Brown
John Gower
Hybrid Bus Team