Back to
Air Command Rockets


Find Data

Pressure Chambers




Test Procedures

Test Equipment



psi to bar

mm to inches

feet to meters

grams to ounces

Request Experiment

Current Requests

Submit Experimental Data



About Labs

081004-001 Calibrated Static thrust experiments
Date 4th October 2008
Tested by PK and GK.
Location Sydney, Australia
Test Procedure Thrust Measurement
References (see below)
Related N/A


  • To accurately measure the static thrust produced by a water rocket.
  • To compare water only and foam thrust curves and their corresponding performance.
  • To test what choking effect a Robinson coupling has on performance.
  • To measure what effect foaming agent concentration has on performance.

Test Stand Experiments

Below is a mini index to make it easier to jump over the boring sections. For further reading of other people's experiments and theory see the reference section. Water rocket (and pyro) thrust experiments have been performed by many dedicated individuals and teams over the years and we have drawn on their knowledge, analysis and their results in helping us perform these experiments.

Disclaimer: While all the experiments were produced with care we acknowledge that there may be errors and that the interpretation might also need further work. We explain our measurement process below, and provide all the raw and processed data for those interested in performing further analysis and draw their own conclusions.

IMPORTANT NOTE: These thrust curves are for static tests and somewhat differ to actual flight thrust curves. However, for relative comparisons they are still valid.

Experiment Background

Before we delve into the actual results it's important to understand how accurate the results are and how we calibrate the test stand. The full details of our test stand can be found here.


Since the load cell amplifier's gain can be adjusted, the amplitude of the measured thrust curve is only a relative value. In order to get absolute readings and be able to compare them against those taken at a different gain settings we need to calibrate the readings each time we change the gain. The gain, however, can also drift slightly with temperature so in order to get the most accurate results we calibrate each test.

After each firing and while the data logger is still recording we suspend a 1Kg load from the nozzle of the empty rocket for a period of time. This records a 9.8N step difference in the waveform. As this is constant all the other values on the waveform can be scaled relative to this difference. See the software section for more details.

Waveform showing the recorded 1Kg weight at the end of the test.

Although we know the weight of the water, we can't really use the difference between the empty and full rocket because while the rocket is full, the release mechanism and air supply hose also hang off the nozzle. (ie. they weigh something) Since the hose is resting on the ground it can give different readings each time.

We tested the linearity of the load cell by applying known weights and plotting them against the reading as recorded by the data logger. The linearity was within 1% over the range we tested.

Accuracy and Repeatability of Measurements

In order to minimize errors in measurement due to variances in pressure and water amounts, we run multiple tests with identical setups and then compare and average the results.

We use a pressure regulator with a fine adjustment that allows us to repeat the pressures quite accurately. This is, however, likely to be have the highest error.

We measure the water amount going into the rocket on a digital scale rather than sighting along the meniscus and this gives us an accuracy of around 1mL.  Some amount of water in the form of water droplets will always remain in the container, in the funnel and in the rocket after each test. Over multiple tests, the amount of water in the droplets remains more or less constant.

An example of three test runs is shown below. The curves show almost identical timing and shape. The total impulse for these curves is within 0.01 % of the average. NOTE: The "ringing" seen in waveform 3 was caused by the ricocheting release head hitting the side of the bottle. This did not alter the thrust.

Close correlation between three separate runs

Turning raw data into thrust measurement 

The thrust we measure here we are calling "Static effective thrust". 'Static' because the rocket is not moving and hence any change in thrust due to acceleration is not taken into account. 'Effective' because the thrust we measure only takes into account the thrust that causes upward acceleration of a rocket having zero mass. The curve also incorporates the weight of water as a negative thrust component due to the need to orient the rocket vertically. If we determine the rate of water loss we can add that component back to the thrust curve.

The weight change can be simulated and added to the actual data, however, in most instances this is not necessary if we are doing relative comparisons between designs.

In the calculations below we calculate the total impulse as the area under the curve and above the horizontal "zero thrust" line.  To include the weight of the rocket one needs to draw a horizontal line above the zero thrust line representing the weight. The area above this new line and below the curve represents the total amount of thrust contributing to the vertical acceleration of the rocket. If the entire thrust curve falls below this line then the rocket will never leave the ground.

Other Notes

We use coloured water in the experiments for two reasons:
a) For more contrast when reviewing video, and;
b) Each water quantity has a different colour code to make the test data easier to verify in video and pictures.

The following diagram describes the different bottle and water configurations used by the experiments.

Bottle Configuration

Test Result Summary

The table below gives a summary of all the static tests performed on the day. The experiments will refer to this table.

The raw and processed data is available here. This archive includes the full recording and the 1Kg calibration measurements with each test.

Test # Capacity
(psi / bar)
Av. Thrust
Total Impulse
1 5450 1000 0 9 110 / 7.6 28.74 45.86 Bottle configuration B, normal water config.
2 5450 1000 0 9 110 / 7.6 28.66 45.86 Bottle configuration B, normal water config.
3 5450 1000 0 9 110 / 7.6 28.68 45.89 Bottle configuration B, normal water config.
4 5450 1000 0 9 110 / 7.6 27.47 46.48 Bottle configuration A, normal water config.
5 5450 1000 0 9 110 / 7.6 27.62 46.84 Bottle configuration A, normal water config.
6 5450 1000 0 9 110 / 7.6 27.93 46.55 Bottle configuration A, normal water config.
7 5450 1000 0 9 110 / 7.6 27.45 46.67 Bottle configuration A, normal water config.
8 5450 1000 0 9 110 / 7.6 24.33 49.57 Bottle configuration A, jet foaming water config
9 5450 1000 0 9 110 / 7.6 24.15 50.82 Bottle configuration A, jet foaming water config
10 5450 990 10 9 110 / 7.6 21.54 56.01 Bottle configuration A, jet foaming water config
11 5450 990 10 9 110 / 7.6 20.04 52.03 Bottle configuration A, jet foaming water config
12 5450 990 10 7 110 / 7.6 13.84 55.36 Bottle configuration A, jet foaming water config
13 5450 990 10 7 110 / 7.6 13.17 54.45 Bottle configuration A, jet foaming water config
14 5450 700 300 7 110 / 7.6 11.68 54.53 Bottle configuration A, jet foaming water config
15 4600 1190 10 7 110 / 7.6 13.05 47.57 Bottle configuration C, jet foaming water config. Bottom bottle ~40% full
16 4600 1190 10 7 110 / 7.6 15.83 41.17 Bottle configuration C, jet foaming water config. Bottom bottle ~95% full
17 5450 2250 0 7 110 / 7.6 12.27 50.97 Bottle configuration A, jet foaming water config, fizzy lemonade used as liquid

Table 1 - Test Result Summary

Here is a video with examples of the experiments:


Top of the test stand with load cell amplifier on top and load cell connected to the top of the rocket.
The laptop and data logger set up nearby. Pressure for the tests was provided by a scuba tank.
A video camera recorded each test. A plastic bag protects the electronics from the rain.
Colour coded water allows us to easily identify the water amount in images and videos.
A foam test under way. The bricks and towel at the bottom deflect the foam away from us.
Wires, hoses and strings everywhere.
Testing a reduced foam mixing chamber to try to reduce the amount of residue foam left in a rocket.
Comparing bottle configuration A on the left and configuration B on the right. Blow-through is quite evident in config A.
Testing varying concentrations of foam. 30% foam on the left, 1% on the right.
Testing reduced mixing chamber with small and large air pocket in the chamber.


Experiment #1 - Robinson Couplings


For this experiment we wanted to see what effect a Robinson coupling has on the performance of a rocket. We were looking at two aspects here.

  1. The blow through effect allowing some pressure to escape before all the water is ejected; and
  2. The internal chocking effect of the coupling.


In order to make an accurate comparison we constructed a rocket with two 2.1L spliced bottles connected together with a 22mm ID tornado coupling. Then we connected a 1.25L bottle with an 8mm Robinson coupling at one end. When used with the Robinson coupling at the bottom (Configuration A), pretty much all the air had to pass through it. When the rocket was turned upside down (Configuration B) then most of the air remained in the tornado coupled sections and therefore the chocking effect was drastically reduced. This was to ensure that the volume of the rocket was identical for both experiments.


The three tests with Configuration A showed very close correlation in timing an amplitude. The total impulse for the three runs was on average 45.87 Ns with the three tests within 0.01 % of the average. The average thrust of the three runs was 28.69 N with all within 0.02% of the average.

NOTE: The "ringing" seen in test 3 was due to the release head impacting the rocket momentarily shortly after release, however, the thrust measurement was not affected. The timing of the event was confirmed on video analysis.

Thrust Curve for Test 1,2 and 3

Configuration B showed slightly more variation in the water phase thrust curve, but this can be attributed to the turbulence generated in the water due to the blow through effect.  The total impulse on average was 46.64 Ns all within 0.05% of the average. The average thrust for the 4 runs was 27.61 N all within 0.1% of the average.

Thrust Curve for Test 4,5,6 and 7

Test #5 and Test #6 show slight dips in the water phase part of the curve. On video analysis these were traced to bubbles exiting the nozzle during the blow through effect.

It is interesting to note that although there was slightly more variance in the thrust curve shape for Configurarion A, the Total impulse was consistently 0.77 Ns or 1.7% higher than Configuration B. The average thrust, however, was 1.1 N or 3.8% lower.

Comparison of the best two curves from each configuration

Comparison of all 7 tests

Bottom line

  • Although the difference between Configurations A and B is measurable, and contrary to what was expected, it is not significant enough at this pressure.
  • Follow on: -This  needs to be tested with higher pressures and larger nozzle.

Experiment #2 - Jet Foaming with and without foam


This experiment was designed to see what effect of a small amount of foaming agent dissolved in the water has on the thrust curve.


The rocket was set up as in Configuration A and water was distributed in such a way that some amount was in the upper bottle and some in the bottom bottle. (Jet Foaming configuration)

Two runs were performed with water only (test #8 and #9), and two runs with 1% concentration of bubble bath solution (test #10 and #11). By weight the liquid was equal in all four tests.


Thrust curves for test 8 and 9

The thrust curves for water-only showed a significant deviation from those in seen in Experiment 1. Of note were the large spikes (loss of thrust) seen in the water phase. On video analysis this was traced back to the blow through effect with large bubbles exiting the nozzle. After the water ran out in the top bottle the water settled more in the lower bottle and the thrust was more conventional during the air-pulse.

The difference in timing of the large negative spike is due to the differences in the water levels in the upper bottle. The more water there was in the upper bottle the later the spike occurs.

The total impulse average of the two tests was 50.2Ns or 8% higher than Experiment 1. The average of the average thrusts was 24.24 N. which was ~12% lower than Experiment 1.

Thrust curves for test 10 and 11

Adding 1% bubble bath concentration to the water (10mL of water was replaced with 10mL of bubble bath) in tests #10 an #11 had a significant effect on the shape of the thrust curve. The thrust curve was more even and produced a longer burn. The blow-through spikes were also quite evident. The average total impulse was 54 Ns and the averaged average thrust was 20.75Ns.

Comparison of no foaming agent vs. 1% concentration

The thrust produced by the rocket with 1% concentration had a total impulse of 7% higher than water alone. At the same time the 1% concentration produced a 14% lower average thrust.

Bottom Line

  • The total impulse increased significantly for the foaming agent and the average thrust at the same time also decreased.
  • The foaming agent even in low concentrations has a significant effect on thrust duration.
  • Follow on: - Need to test with higher pressures.

Experiment #3 - Jet Foaming with 7mm vs 9mm nozzles


The main aim of this experiment was to see how much of a difference there is between the 7mm and 9mm straight through nozzles.

This was considered important as the 8mm Robinson coupling chokes the flow of water out of the 9mm nozzle. With a 7mm nozzle the chocking happens at the nozzle.


Bottle configuration A was used for this comparison. Except for the nozzle diameter all other rocket parameters were identical for all four runs. Test #10 and #11 used a 9mm nozzle and tests #12 and #13 used a 7mm nozzle.


Thrust curves for test 10 and 11 - 9mm nozzle

Thrust curves for test 12 and 13 - 7mm nozzle

When we look at the 7mm nozzle, there was a significant increase in the duration of the thrust as would be expected having the mixture move through a 40% smaller cross sectional area.

The average total impulse for the 9mm nozzle was 54 Ns and the averaged average thrust was 20.75Ns. While for the 7mm nozzle the averaged total impulse was 54.9 Ns and the averaged average thrust was 13.5 N.

Comparison of 9mm vs. 7mm nozzle

Bottom line

  • The total impulse for the 7mm nozzle was only 2% higher, while the average thrust was 35%  lower. The total burn time was increased from 2.6 seconds to 4.1 seconds or by 37%.

Experiment #4 - Foaming agent concentration effect on Jet Foaming


This experiment set out to see if changing the concentration of foaming agent has any significant effect on the thrust. The main driver for this experiment was Anti-Gravity Research's record flight where they used 30% detergent in their water. Although they used a different detergent and a different foam generation technique.


We performed only one run(#14) of the high concentration test mostly due to the amount of detergent flowing into our garden. We compared this to tests #12 and #13 as they only used 10ml of foaming agent.

We placed 300g of the bubble bath solution in a container and filled the rest with water until the scale read 1000g. This gave us the same mass of liquid in the rocket. We measured by weight as the bubble bath has a different density.


Comparison of 30% concentration of foaming agent vs. 1%

The higher concentration appeared to generate a much more homogeneous foam solution and as a result created a much smoother decaying curve. The total impulse for the higher concentration was 54.53Ns and the average thrust was 11.68N the lowest out of all the tests. This was also the longest burn of all the tests at 4.6 seconds.

While the total impulse was within 1% of the lower concentration, the average thrust was 13% lower. The burn duration was also 13% longer.

Bottom line

  • There was enough of a measurable difference to warrant further investigation on higher foaming agent concentrations.

Experiment #5 - Jet Foaming with different sized air pockets


From previous experiments we have learned that a non trivial amount of weight in the form of residual foam remains in the rocket when all the air is gone. In order to minimise the amount of foam remaining in the rocket one approach we are considering is to use a small mixing chamber as the lowest bottle in the stack. This would allow foam to only form in the small volume of the bottle and hence lower the amount of foam left in the lowest bottle.

Because of the small volume we wanted to see if it's better to have a small pocket of air in the lower bottle or a larger one.


We set the rocket up as Configuration C and let water fully drain into the lowest bottle. We also used 1.2L of water for this test with 1% concentration of foam. We were going to do the 95% test first but found that during the normal filling rate about 60% of the water was pushed up to the upper bottle. The problem was that once pressurised it would not drain back. So we tested that one first.

For the second test we had to fill really slowly and it took about 8 minutes to fill the rocket with air without pushing the water into the upper bottle.

The idea is that in a real rocket there would be a narrow launch tube that fits through the Robinson coupling and emerges above the water line. This would allow the rocket to be pressurised without creating any foam in the upper bottles, and only creating foam in the lower bottle once launched.


With the lower bottle being mostly full it effectively worked as a single full bottle and as a result the thrust curve looked more like a water only one with not much foam being produced.

The test with the much larger air pocket generated much more foam and as a result the thrust curve was more like normal foam. The total impulse for the 40% water (test #15) was 47.57Ns and the average thrust was 13.05N.

Comparison of mixing chamber air pocket size - 40% vs. 95% water in the lower bottle

The total impulse for the 95% water (test #16) was 41.17Ns and the average thrust was 15.83N. This means that the larger air pocket gave a 15% higher total impulse while lowering the average thrust by ~17%.

It was interesting to note that there was no significant blow through spike evident in either of these tests.

Bottom line

  • The air pocket size is very important for generating more foam.

Experiment #6 - 2L of Fizzy lemonade used in Jet Foaming


This was more of a fun experiment to do at the end of the day, but it has to be one of the most common questions we get asked if we have put soda water into our rockets to see if they fly higher. Actually we needed the empty bottle to use in a new rocket, and it was either dump the lemonade down the drain or through a nozzle. As Damo later pointed out we should be now worried about more ants in the garden.


We used Configuration A for the bottle setup, and since there were 2L of lemonade we allowed it to drain into the lowest bottle and the rest went into the upper bottle. One of the biggest problems was that as the lemonade was poured through the funnel quite a bit of CO2 was released and while the rocket was pressurised even more was liberated so by the end it was difficult to say how much actually stayed dissolved. The total impulse was 50.97Ns and the average thrust was 12.27N. It is difficult to compare this to any of the other tests as a different volume of water was used.

Thrust curve for 2L of fizzy lemonade

Bottom Line

  • Needs to be performed again but without filling through the water. Most likely through a fill tube above the water line.

  • The lemonade seemed to remove the residual foam from the bottle really well.

  • Needs to be compared to 2L of normal water to see what difference there is.


  • When only air passes through a relatively narrow Robinson coupling there is no significant effect on thrust when both the nozzle and coupling diameters are similar.
  • There is a large momentary loss of thrust during jet foaming when the water runs out in the top bottle. The spike is more pronounced in jet foaming most likely because the density of the liquid/foam in the bottom bottle is lower than just regular water. The air/water mixture can punch through it more easily.
  • The large blow-through spikes seen in a lot of the tests were directly attributable to air bubbles exiting the nozzle when video of the tests was analyzed and explain the foam thrust anomaly identified previously when we lacked the sample rate to clearly identify it.
  • When we compare the total impulse of water alone (test #2) and foam (test #10) we see that in effect foam had 19.2%! more "energy". This is a very significant number. At the same time the average thrust was  25% less. This will require further testing at higher pressures and at "optimised" water levels. How the extra impulse translates to actual altitude is a little more complex but is highly dependant on weight of the rocket and drag. More on this in future updates.
  • There is a long bump in the middle to later part of the foam thrust curves that suggests perhaps some optimal combination of foam/air mixture where the thrust does not decay as one would expect. This may be due to water settling out of the foam. Foam no longer being efficiently generated perhaps? This will need further investigation.
  • When using a reduced mixing chamber size for jet foaming it is important to have a large air pocket in the small bottle.

A comparison of all 17 tests done on the day



Copyright 2006-2010 Air Command Water Rockets

Total page hits since 1 Aug 2006: