This document summarizes a senior design project that tested the effects of different grain geometries, propellant formulations, and vacuum time on solid fuel rocket propellant characteristics. The project tested divergent vs straight nozzles, flat vs O-ring vs conical grain geometries, baseline vs oximide vs guanidine nitrate formulations, and the effect of varying vacuum time on propellant density. Key findings were that divergent nozzles and O-ring grain geometry yielded the highest specific impulse. Propellant formulations containing guanidine nitrate performed better than oximide formulations. Increasing vacuum time did not significantly affect propellant density.

1. Propellant Characterization
Senior Design Project
Summer 2014
MENG 491
Dr. Martin Weiser
Jesse Hutson, Sean McCoy, Spencer Scott

2. 2
Abstract:
Our research focused on grain geometry effects on solid-fuel rocket propellant burn
characteristics, propellant characterization, and the effects of vacuum time on propellant
density as a percentage of theoretical. We found that divergent nozzles yielded approximately
5.22% higher specific impulse (ISP) when compared to straight nozzles. Three grain geometries
were considered; flat, O-Ring, and conical. Flat geometry delivered marginally higher ISP’s than
the other two but burnt through the liner significantly. Conical and O-Ring geometry yielded
comparable ISP’s, so we elected to use O-Ring geometry to minimize machining of the grains.
We found that our baseline propellant formulations had higher ISP’s than either the oximide or
guanidine nitrate. The guanidine nitrate formulations had higher ISP’s than the oximide
batches. Batches with a designation of one (GN1 for example) consistently yielded higher ISP’s
than the equivalent batches with designations of two or three.
All of our batches except for OX1 were characterized by low densities (below 91.1% of
theoretical). The ISP values of all of our motors were between 70% and 75% of the theoretical
ISP. There was no connection between low density batches and low ISP’s, as our batch with an
ISP closest to theoretical was one of our lowest density batches. Guanidine nitrate and oximide
are compounds that are supposed to slow the burn down and yield a less progressive pressure
curve. However, we found that neither guanidine nor oximide slowed down the propellant to a
measureable degree. We found no discernible change in density as vacuum time is increased.

3. 3
Introduction:
In the world of model rocketry, the quest for an efficient long burn propellant is ongoing. We
will be working on several formulations using a solid composite based fuel. Composite fuel is
generally more stable than other types of solid propellant and is therefore safer to handle. Solid
composite motors are identified by their binder, in this case hydroxyl terminated polybutadiene
resin (HTLO). The oxidizer will be ammonium perchlorate and the fuel will be aluminum and
HTLO. There are several other additives that give the propellant desirable characteristics. For
this study we will mainly be focused on the effect of oximide (OX) and guanidine nitrate (GN).
By mixing different formulas in both the OX and GN series we hope to determine which additive
is the most effective by both increasing time of burn and specific impulse.
We will initially complete a previous year’s project on graingeometry. Because of the
inconclusive results of the previous study, we propose repeating the study with a larger initial
batch. The original question of whether changes in the internal geometry affect the specific
impulse and burn time will be answered by investigating four geometry types. The four types
consist of flat grain contact surface with a flat nozzle, conical grain contact surface with a flat
nozzle, O-ring between the grains with a flat nozzle, and flat between the grains with a
divergent nozzle. Using MOSLB as the baseline for this test we will establish which geometry
gives the highest specific impulse. In addition to completing additional work, we will gain
experience mixing, manufacturing, and testing before attempting the propellant

4. 4
characterization portion of the study. We should have a better understanding of which grain
geometry we should carry into the other rocket-based studies.
We will then conduct analysis on propellant characterization. This will be our contribution to
the growing body of research produced by the EWU Engineering department. Our study will
focus on two additives exclusively; oximide (OX) and guanidine nitrate (GN). We will
manufacture eight different formulas for this portion of the study. We will manufacture and
test two baseline formulations and three variations each of oximide and guanidine. Individual
formulation data can then be compared such that the most efficient additive for our uses can
be determined.
Our last project focus will be to determine if there are any effects of vacuum time duration on
the grain density as a percentage of theoretical. During our training on proper mix procedures,
we allowed the propellant to be kept under vacuum for one week. The resultant density was
extremely high, above 95% of theoretical. Higher densities are deemed to be desirable in the
industry, so we elected to make an investigation into the density effects of vacuum time
variations, prior to adding the curative, a part of our project.

5. 5
Methods:
Propellant Synthesis
Dr. Martin Weiser, an experienced materials engineer, our project direct, and a rocketry
enthusiast, provided us with standardized propellant recipes. The recipes are adjustable by
grain diameter, number of sticks, length of sticks, and expected waste. For all of the testing, we
used standard 38mm two grain motors. We adjusted the recipes for 38mm, one percent waste,
and eight inch sticks. Each stick yields four 1.9 inch grains, so we cast four sticks of each
baseline and three sticks of each additional formulation. This gave us eight tests per baseline
and six tests per additional formulation.
The recipe reads much like that of any baking recipe. We used a household KitchenAid stand
mixer to combine our chemicals and a digital scale accurate to 0.1 grams with an uncertainty of
0.05 grams. We combined all of our liquid ingredients except for the E744 rubber curative in
the bowl of the mixer and then started mixing. The dry ingredients were weighed and added
separately as the mixer was running. Great care was taken to ensure the 325 mesh aluminum
powder was completely integrated into the liquids before adding the ammonium perchlorate.
Failing to do so can initiate a spontaneous and extremely exothermic reaction. The chemicals
were left to mix for one hour with a scrape down of the bowl and paddle every fifteen minutes.

6. 6
The bowl was then removed from the mixer, capped with a vacuum tight lid, and placed under
a 22-25 psig vacuum for an additional hour. This is the time that was augmented for our
propellant density testing, which is addressed later. The bowl was then scraped down, the
curative was added, and the ingredients were mixed for fifteen minutes with scrape downs
every five minutes. A single drop of PDMS was added at the beginning of this mixing session to
smooth the mixture. The bowl was then capped with the vacuum lid and placed under vacuum
for an addition fifteen minutes. During this time, we used the integrated stirring bar on the lid
to agitate the mixture and remove as much air as possible. The mixture was removed from the
vacuum after the conclusion of the fifteen minutes and the process of packing the sticks began.
We used coated liners that we had prepared earlier to
house the propellant. The propellant was then rolled into
balls roughly 0.75 inches in diameter, being careful not to
introduce air pockets, and placed in the liner. We used a
wooden dowel that was slightly smaller than the liner to
tamp the propellant before placing another ball in the liner.
The process was continued until the liner was full. The sticks
were then set aside to cure for three weeks.
Figure 1.1
Three sticksof propellantsetaside
for curing.

7. 7
Grain Preparation
The cured sticks of propellant
needed to be cut and a bore hole
drilled before they were ready for
testing. We used a miter box with
a blade guide to ensure the cuts
were even and vertical. We
placed a stop at two inches and
cut the grains. The measured
length of the grains after cutting is between 1.9 and 2.0 inches. There was a slight loss due to
the width of the saw blade. The dust is highly flammable and was collected to be incinerated
later. A hand saw was used to keep the dust in one place and to ensure minimal heat
production. After the grains were cut, we lightly sanded the ends to remove any excess paper
or propellant.
The bore hole needed to be centered on the grain so that when it
burns, there is an equal amount of propellant radially. We machined
an aluminum collar that held a drill collet centered on the grain.
Using a 5/16” bit, we created the bore hole. It was important to drill
slowly to collect the waste propellant and create an even bore.
Figure 2.1
Miter box withstopfor cuttinggrains.
Figure 2.2
Drill collarfor5/16” bore.

8. 8
Flipping the grain over and drilling again removed any propellant burs that had remained in the
bore hole.
The grains were then measured, weighed, and catalogued for testing. Each stick was given a
letter designation and each grain was given a number location in the stick. Using the length and
area of the face, we calculated volume of each grain. We divided the masses by the volumes
and calculated a mean density for each batch. We paired grains to yield similar masses for each
motor.

9. 9
Motor Assembly
Two grains were placed inside a protective liner with an O-ring
between them. The liner and grains were then placed inside of the
aluminum housing, roughly in the middle. The chosen nozzle was
seated against the grains in one end of the housing. We made sure
that there was a satisfactory O-ring between the nozzle and housing
to prevent gases from escaping around the nozzle. A steel washer,
sized just small enough to fit into the housing, was placed against
the nozzle and a snap ring was used to keep them all in place.
On the other side of the housing we inserted a cap
with a grease-filled brass tube to transfer the
pressure to the pressure transducer. Grease is used
Figure 3.1
Motor assemblylayout. Figure 3.2
Graphite nozzle withO-ring.
Figure 3.3
Aluminumcapandbrass pressure coupling.

10. 10
to reduce the amount of compressible volume added by the pressure transducer. It is
important to refill the grease in the tube after each test as some of it will melt and run out.
Another O-ring was used around the cap to prevent gases from escaping. A snap ring was used
to hold this side in place.
We alternated between two sets of
housings, rings, and nozzles to allow
us to test without a significant down
time between ignitions. The housings
heated up significantly so it was
important to allow them time to cool
before disassembling them. The
capped end was removed first and
then a dowel was used to push the
nozzle out of the other side. The
motors were housed in milled aluminum cases with graphite nozzles. The nozzles were
interchangeable, so the aluminum cases and nozzles could be reused for multiple tests.
Between tests, the cases and nozzles were cleaned of any slag deposits.
Milledaluminumhousingwithsnapringgroove.
Figure 3.4

11. 11
Test Procedure
The test rig was placed in an open field with
wet canvas mats and a metal diffuser to
handle the jet of hot gases. Though the rig is
heavy enough to not lift off the ground, we
staked it down at each corner to prevent any
unexpected movement. We kept a few feet
between the ignition cable and the transducer
cable to ensure there would be no
interference in our data. Our ignition box was
set roughly 25 feet from the test rig on a
rolling cart and the computer interface
another 20 feet behind that.
The assembled motor was placed in the rig and secured in place.
We then connected the transducer to the brass fitting on the
motor. After checking that the power was off and grounding the
leads, we placed the igniter in the motor. The igniter sits roughly
0.25 inches from the top of the motor to ensure ignition on all
sides. We wrapped each lead of the igniter around an alligator
Figure 4.1
Testrig preparedforignition.
Figure 4.2
Motor mountedinrig.

12. 12
clip for a good connection. It was necessary to clean the
alligator clips occasionally due to corrosion. The test was
treated as live at this point, the area was cleared, and we
prepared for ignition.
Our ignition box had two safeties. The first was a break in the in
the circuit that can be restored by inserting a plug into an outlet
built into the box. The second was a switch that toggles the
power on or off. When both were set to allow the circuit to
complete, we started recording data. Roughly ten seconds after
starting the recording we press the ignition button. After the
motor finished burning, the safeties were engaged and the
recording was stopped and saved. We were able to export our
data directly into Excel for processing later.
Figure 4.4
Ignitionbox.
Figure 4.3
Wrappedignitionalligatorclips.

13. 13
Grain Interface Geometry
Our first test group focused on the geometry between the two grains inside the motor. The
overall objective was to create even burn area within the motor. This would create even
pressure and a constant thrust until the brief burn out. We looked at specific impulse to
determine the best geometry for our purposes.
The first geometry we tested was our baseline flat or butt joint. This geometry is when the
grains are seated directly against each other with no gap in between. This geometry is what is
used in commercial grade motors. The only preparation that is required is a light sanding to
make sure the ends of each grain are flat.
Figure 5.1
AutoCADcomparisonof geometries.

14. 14
The second geometry we tested was using an O-ring spacer between the grains. The O-ring was
1/16 inches thick and was just small enough to fit inside the protective liner. The physical
geometry of the grain was the same as the flat geometry. There was also enough space in the
motor housing to accommodate the extra length.
Our third geometry was a conical gap between the grains. The gap was created by sanding a
two degree chamfer on one side of both grains. The chamfered ends were placed together in
the liner to create a partial gap between the grains. The mass of these grains is generally lower
because we had to sand some of the propellant to create the geometry, but we divided mass
out of the final total impulse values to accurately compare the performance of the geometries.
Our final geometry test was a divergent cone on the nozzle. This used the same flat geometry
between the grains and only changed the graphite nozzles. Divergent cones are used in typical
commercial grade rocket motors. The rest of the tests were conducted with a bore of constant
diameter through the nozzle. It is quite a bit easier to create nozzles that have a straight bore
instead of a divergent cone, so we were testing to see if it had an impact of performance.

15. 15
Propellant Characterization
The primary idea behind our experiments in propellant formulas is to eliminate filler ingredients
in favor of chemicals that will add to the energy of the reactions but still serve to slow down the
overall burn rate. We generally used calcium carbonate and ammonium chloride to slow the
burn rate, but elected to replace these chemicals with test batches containing oximide and test
batches containing guanidine nitrate.
We used two baseline formulas that had no chemicals to slow the burn rate, three formulas
with oximide, and three formulas with guanidine nitrate. The first and third formulas in oximide
and guanidine nitrate contained 83.12 percent solids and the second formulas contained 80.33
percent solids. There reactant ratios varied slightly for each batch of oximide and each batch of
guanidine nitrate.
Our first and third batches of oximide turned out satisfactory, but the second batch was about
the consistency of molasses. The second batch was not liquid enough to pour and too liquid to
tamp. We ended up with grains that were full of voids and imperfections. We mixed the batch a
second time to eliminate anomalies in the mix, but we ended up with the same result. We
elected to not use this formula in testing and not mix the second Guanidine Nitrate batch due
to the same liquid to solid ratio in the formulas.

16. 16
Our first batch of Guanidine Nitrate, referred to as GN, did not combine as expected. Our
supply of the chemical GN was in the form of small pellets. We expected them to dissolve
during the mixing process, but that was not the case. The GN did not dissolve at all and the
batch had to be mixed a second time. We used a mortar and pestle to grind the GN under a
fume hood to a 200 micron mesh size. It integrated well and the batches were usable.

17. 17
Vacuum Propellant Density:
To test our hypothesis on time under vacuum being linked to propellant density, we mixed our
standard formula for EWU-12 and divided it in to three equal parts. The first test sample was
mixed with no vacuum time prior to adding the curative. The second test sample was subjected
to vacuum for one hour before adding the curative. The third test sample was subjected to
vacuum for one week before adding the curative. The grains were left to cure for three weeks
before cutting, drilling, and calculating their densities.
Igniter Production:
For all of our tests, we used igniters that we manufactured. The
wires are twisted pairs from standard network cabling. We used a
fairly standard method of wrapping the pair with a much smaller
gauge wire to produce heat through high resistance. For the
ignition material we used a mixture of Red Dot smokeless powder
and boron potassium nitrate. The igniter was left to dry and then
coated in nitrocellulose lacquer to prevent cracking.
Figure 6.1
Standardigniterfortestburns.

18. 18
Results:
Grain and Nozzle Geometry
There are three distinct grain interface geometries that we tested; flat, conical, and flat with an
O-ring spacer (herein referred to as O-ring). Flat geometry has the two motors flat against one
another with no gap between. Conical geometry has the contacting grain ends chamfered at
two degrees. O-ring geometry has the grain ends flat with an O-ring spacer between them.
Different grain geometries affect the burn characteristics of a motor. These characteristics
include the pressure curve, the maximum pressure amplitude, and the likelihood of burn
through of the liner concentric to the grains.
We also tested the effect of nozzle geometry on the pressure characteristics of the motor;
straight versus divergent. A straight nozzle is just that; a hole drilled straight through the
graphite cylinder to allow gas to escape. A divergent nozzle has a cone drilled into the outer
end of the graphite. The latter nozzle type allows lower turbulence in the exhaust gases than a
straight nozzle.

19. 19
Table 7.1
Rows highlighted in orange are EWU3, our first tests. Rows highlighted in purple are EWU1, 2,
and 6.
In addition to being dependent upon the burn characteristics of the propellant in question, the
total impulse yielded by the motor is directly proportional to the total propellant mass. A motor
can still have a very high total impulse even with an inefficient powder as long as it has
sufficient mass. A more effective way of comparing different propellants and geometries is to
analyze the ISP (specific impulse). The ISP divides out the mass so that we get the total impulse
per unit mass.
We tested EWU3 early on using the same nozzle size for each test. The purpose of these early
tests was to determine which geometry was optimal. The EWU3 data set shows the flat grain

20. 20
geometry having 1.27% and 2.43% higher ISP, on average, than conical and O-ring geometries
respectively. Not only did flat grain geometry deliver only marginally higher ISP’s than conical or
O-ring, flat geometry is characterized by significantly higher likelihood of burn-through of the
liner. If a motor burns through the liner, it may send a hot jet of exhaust gases out through the
side of the metal casing, ruining the rocket and the test. Considering the higher risk of burn-
through and the insignificant performance benefits, we decided after testing EWU3 that we
would eliminate flat grain geometry from further analysis.
We found that when flat grain interface geometry was combined with a divergent nozzle for
EWU3, a consistent 5.22% increase in ISP was delivered over the flat grain geometry with a
straight nozzle. We determined this result to be significant enough to warrant using a divergent
nozzle for all of our subsequent tests.
We elected to compare our results using several Student’s t-tests to determine whether our
samples were from statistically different populations. We used a two-tail test with assumed
unequal variances and an alpha of 0.05. Nearly every comparison yielded a result that indicated
the sample groups were not separate. The only two tests that yielded results of separate
populations involved the divergent conical geometry. When compared to the flat conical and
the rest of the total population of alternative geometries, the divergent conical was shown to
be a separate population. Samples of the T-tests are located in the Appendix, Tables 11.6 and
11.7.

21. 21
We ran several tests with conical geometry and divergent nozzles for batches EWU1, 2, and 6.
The ISP values for EWU6 were 3.53% higher for O-ring geometry than for conical and were the
same for EWU1 and EWU2 using the conical geometry. Our EWU3 tests showed the ISP’s being
1.17% higher for conical than for O-ring. In testing multiple batches, there was no consistent
advantage of O-ring over conical geometry, or vice-versa. O-ring geometry involves less
machining of the grain than conical does, enhancing the capability for high precision. So, we
decided to use O-ring grain geometry with divergent nozzles for all of our subsequent tests.
Propellant Characterization
Figure 8.1 Figure 8.2
When a motor reaches maximum pressure quickly, it is operating at the highest level of thrust
possible over the longest period of time. This condition maximizes total impulse, also
maximizing the height achieved by a rocket motor. If a motor is progressive, gradually building

22. 22
pressure up to a maximum value and decreasing almost immediately thereafter, the total
amount of time spent at that maximum pressure is minimal. The resultant total impulse will be
correspondingly less with a progressive burn than if it had reached maximum pressure very
quickly. We want our motors to burn such that their pressure curves appear like figure 8.1
rather than the progressive as shown in figure 8.2.
All pressure curves for the baseline propellant formulations had a degree of progressive nature,
with most being very progressive. For a majority of the pressure curves, the peak pressure was
only reached for a short time before the pressure decreased rapidly back to atmospheric.
The Guanidine batches were characterized by this same progressive nature. The GN1 and GN1b
batches were significantly more progressive than the GN3 batch. Only a single divergent O-ring
motor out of the GN1 and GN1b batches combined did not exhibit highly progressive behavior.
By contrast, of the six GN3 motors tested, only a single one had a very progressive burn while
the other five were more moderately progressive. Through all of our GN tests, the GN1 and
GN1b especially, the burn was extremely fast compared to the other batches.
Our Oximide batches were also progressive, but less so than the GN or the baselines. A majority
of the Oximide tests exhibited moderately progressive burns, with only a minority exhibiting
the same extremely progressive curves found throughout most of the GN and baseline tests.

23. 23
Guanidine and Oximide are each designed to moderate the burn of a propellant, making them
theoretically slower and less progressive than a baseline (which is without any moderating
compound). However, the progressive nature of the Oximide and GN batches indicates that the
propellant burn was not being slowed down effectively by either compound. The slightly less
progressive nature of Oximide when compared to GN may be indicative of a higher
effectiveness of the former in slowing down the burn of a propellant.

24. 24
Baseline Propellant Formulations
Table 9.1
Average ISP before and after outliers were removed. Standard deviation values are also
included for each batch, helping gauge the relative consistency of the batch. The densities and
ISP’s relative to theoretical are also included.
The average ISP for GN1a and GN1b was 137.9, 6.22% higher than for GN3. The average ISP for
OX1 was 2.96% higher than for OX3. This seems to indicate that a higher amount of Guanidine
or Oximide as found in the batches with a designation of three decreases the overall efficiency
of the propellant.
The average ISP’s for the individual Guanidine batches are higher than for those using Oximide.
GN1 yielded a 5.81% higher ISP than OX1 and GN3 gave a 2.55% higher ISP than OX3.

25. 25
The baseline propellant batches were characterized by higher ISP’s than either GN or OX. BL1
provided a 6.58% and 12.77% higher ISP than GN1 and OX1 respectively. Additionally, BL1 had a
2.50% ISP advantage when compared to BL2. The two primary differences between BL1 and BL2
are the presense of lamp black and a slight increase in Ammonium Perchlorate in the latter.
One universal trend in our results are that the batches with a designation of one are higher
than the equivalent batches with designations of two or three. The baseline propellant batches
yielded significant increases in ISP versus Guanidine or Oximide-based batches.
Only a single batch (OX1) that we cast and tested had a density above 91.1% of the theoretical
prediction. We were advised by Dr. Weiser that he would be apprehensive about flying motors
with less than 90% of theoretical density. The obvious concern with low densities is the
possibility of voids within the dried propellant. However, our consistently low densities
combined with relatively small standard deviations perhaps warrant further analysis into the
theoretical calculations of propellant densities.
We posited early on that low densities could perhaps be linked with low ISP values. However,
we found no such link between the two properties. OX1 was our single high density batch, with
94.6% of theoretical density. This batch yielded an ISP of 71% of the theoretical value. BL2A by
comparison was one of our lowest density batches and gave an ISP of 75% of theoretical (our

26. 26
highest value). The inexplicably low ISP’s from our tests when compared to theoretical
predictions indicate that there could indeed be a flaw in the program’s calculation of ISP.
Table 9.2
Equations for burn rate versus average pressure and ISP versus Kn for each batch.
A desirable characteristic of a rocket motor is a burn rate that is independent of the average
pressure. Different nozzle sizes directly affect the pressures achieved by a given motor. If a
motor changes its burn rate significantly when operating at different pressures, then one can
only use a single nozzle size if a specific burn time is desired. By contrast, if the burn rate
remains constant regardless of the average pressure, the nozzle size can be changed to
whatever best fits the situation without worry about changing the burn time.

27. 27
The burn rate can be expressed as;
𝐵𝑢𝑟𝑛 𝑅𝑎𝑡𝑒 = 𝐴 ∗ 𝑃𝑎𝑣𝑒𝑟𝑎𝑔𝑒
𝑏
Where A and b are constants.
Since we are looking for a burn rate that is independent of 𝑃𝑎𝑣𝑒𝑟𝑎𝑔𝑒 , we want A and b to deliver
as flat a curve as possible. From figures 9.3-9.5, it appears that the flattest curve is OX3, with
OX1 and GN3 closely following. These propellants seemto deliver the most desirable burn rate
characteristics.
Figure 9.3

28. 28
Figure 9.4
Figure 9.5
Burn rate versus average pressure curves for all tested propellants.

29. 29
Vacuum Density Effects
Table 10.1
Grain specifications for the vacuum density batches. The grains with designation “a-“ were not
placed under vacuum before adding curative. Designation “b-“ was given to grains subjected to
vacuum for one hour, while designation “c-“ was for grains under vacuum for one week.
For our vacuum density analysis, we elected to make 12 total grains; four for each vacuum
time. We discovered that densities of 91.54% of theoretical were yielded with no vacuum.
When vacuum time was increased to one hour, densities decreased to 90.80%. With one week
under vacuum, densities rebounded to 91.38% of theoretical.

30. 30
The extreme spread of this data was only 0.74%, with the no-vacuum batch having the highest
density. The no-vacuum batch yielded densities 0.81% above the one-hour batch. With the
densities of each batch being so close regardless of vacuum time, we could discern no
appreciable effect of increased vacuum time.

31. 31
Conclusion:
Through our testing, we have come to realize that the majority of our hypotheses were
incorrect. We had expected the flat grain geometry to be a clear leader. It is used in all of our
commercial rocket motors and seemed the most logical for keeping a constant burn rate. As it
turns out, flat geometry had a tendency to burn through the side of the liner between the
grains. The O-ring spacer and conical geometries were slightly less efficient, but had
significantly less burn-through. The ISP’s for conical and O-Ring geometries were virtually the
same. Since conical geometry involves more machining of the grains, perhaps leading to added
inconsistencies, we elected to use O-Ring geometry. Forming grains was a very dirty process,
often leading to a slight coating of propellant on the outside of the inner liner. This could have
had an effect on the burn-through characteristics, but this has not been tested.
The Student’s t-tests indicated that our test groups primarily came from the same statistical
population. To clarify the results, we would need to run another set of tests to increase the
number of points in our data set. After verifying that our test groups were actually members of
the same statistical population, we could choose the easiest group to manufacture as our
primary test geometry. Alternatively, it could have been helpful to run T-tests for the same
groups with safety ratings based on the burn-through characteristics of each geometry. Our
rudimentary observations of liner burn through during testing indicated that there may be
differences between the test groups based upon this parameter.

32. 32
The guanidine nitrate and oximide additions did not have the predicted effect on burn rate or
specific impulse. The pressure curves became generally progressive and the burn rates seemed
to be comparable to the burn rates of our baseline formulas. We are unsure as to why the
chemicals did not react as predicted. It appeared that a higher proportion of the guanidine
nitrate tests were characterized by extremely progressive burns. The oximide appeared to be
slightly less progressive in its burning nature. This suggests that although still ineffective, the
oximide is more effective than guanidine nitrate at slowing the burn of a propellant down. We
noted that the baseline propellant formulations delivered the highest average ISP values, with
guanidine nitrate being second, and oximide having the lowest ISP’s. It was interesting to note
that guanidine nitrate does not dissolve during the mixing process. This could indicate that we
would need to use an aqueous solution to properly integrate this in to the formula.
All of our batches yielded very low densities; only OX1 had a density above 91.1% of the
theoretical value. All of our tests had ISP’s between 70% and 75% of the predicted value. We
posited that these low ISP’s could be linked with our low densities. However, we found that the
batch that yielded the ISP that was closest to the theoretical was also one of our lowest density
batches. It does not appear that low density is linked with our low ISP’s. The amount of time
spent under vacuum before the curative was added had no apparent impact on the density of
the grains. Our highest density during this test came from the batch with no time under
vacuum. It was not a great difference, but it was certainly unexpected.

33. 33
A propellant that burns with a constant burn rate, regardless of average pressure, is most
desirable. A propellant whose burn rate does not vary with pressure can be used with
predictable performance with varying nozzle sizes. This allows a rocketeer to have a single
motor that can be used with a wide variety of conditions. If we graph the burn rate versus the
average pressure for a given propellant, the most desirable contour will be a flat line parallel to
the pressure axis. Of the propellants that we tested, it appeared that OX3 delivered the flattest
curve, with OX1 and GN3 following closely behind.

34. 34
Appendix: Sample Data
Table 11.1
Output data for each set of grain calculations.
Table 11.2
Input data for each set of grain calculations.

35. 35
Table 11.3
Table of time dependent data (Pressure and thrust).

36. 36
Table 11.4
Raw data collected by transducer. The machine collected pressure data in volts; we added a
pressure calculation (and normalized it) in line with our calibration equation relating pressure
to volts.

37. 37
Table 11.5
Summary of data collected from each batch; grains used, motor weight (g), % of theoretical
grain density, Kn, burn time, average pressure (psig), burn rate (in/s), impulse, and ISP.

38. 38
Table 11.6
Table 11.7
Tables 11.6 and 11.7 are the results of the Student’s t-tests for our divergent conical geometry
when compared to flat conical and the total remaining alternative geometries.
Divergent Conical vs Conical
t-Test: Two-Sample Assuming Unequal Variances
Alpha = .05
Variable 1 Variable 2
Mean 137.7782847 146.8748027
Variance 15.08291596 49.51783448
Observations 4 7
Hypothesized Mean Difference 0
df 9
t Stat -2.762271123
P(T<=t) one-tail 0.01101509
t Critical one-tail 1.833112933
P(T<=t) two-tail 0.02203018
t Critical two-tail 2.262157163
Divergent Conical vs Alternative Geometries
t-Test: Two-Sample Assuming Unequal Variances
Alpha = .05
Variable 1 Variable 2
Mean 137.7782847 146.8149534
Variance 15.08291596 53.89282905
Observations 4 25
Hypothesized Mean Difference 0
df 7
t Stat -3.712028784
P(T<=t) one-tail 0.003767338
t Critical one-tail 1.894578605
P(T<=t) two-tail 0.007534676
t Critical two-tail 2.364624252

39. 39
Figure 11.1
Figure 11.2
Figure 11.3
Figures 11.1-11.3 are plots of ISP versus Kn for each formulation.