Poseidon Tanks: Concept and Design

02 Apr 2021 by Finn in


(Edit 07 Aug 2022)

FAR so far

The FAR test platform has tanks made out of PVC. We chose this material because it’s quick and easy to build a tank out of, and can sustain a pressure of about 8 bar. Additionally, it is relatively cheap and available.

However, these PVC tanks are not flight ready. They weigh almost ten times the flyable weight and cannot sustain the required pressure of about 28 bar.

Requirements for the Poseidon tanks


The rocket must be light, preferably under 1500g. Our simulations have shown that the rocket at 28bar can lift around 7kg. Leaving some room for acceleration, the dry mass should be around 25% of the total rocket mass. This ensures that the rocket can carry enough propellant to reach a decent apogee height (see also: basics of water rockets).

Because there are a lot of other heavy components onboard -for example the solenoid valves- the tanks should not weigh more than about 600g.


The weight requirements leave enough room for a propellant mass of about 4kg. The tanks must be large enough to accommodate that amount of water.


We want to pressure up the rocket as much as possible. The valves can open up to 28bar according to the datasheet, so we’re going to dimension the tank for that pressure. Some factor of safety included. The rocket should also be able to withstand falls and other impacts.


We don’t have access to a lot of expensive tools, so we need to be able to build the tanks in a small workshop. Advanced manufacturing proccesses like friction stir welding for thin aluminium pressure vessels is not even in the realm of possibility for us.


Only a few materials can be considered for a tank of this weight and strength. Steel is to heavy, and using very thin walls is not an option, as they’re prone to buckling.
Aluminum has a much better strength to weigth ratio, but we lack the corresponding manufacturing capabilities.

Composite materials such as glass or carbon fiber composites have an excellent strenght to weight ratio. The fibers are very strong under tension, which is the main load of a pressure vessel. Additionally, a manual lay-up process allows us to manufacture the tanks ourselfs, as hand laminating does not require special/expensive tools and can be done in our workshop.

These tanks would be composite pressure vessels, or CPVs for short. It’s a derivative of COPVs, which are commonly used in the aerospace industry. These have a thin metal liner on the inside, overwrapped by the composite material. We are not implementing a metal liner because we are working with no volatile gases and lower pressures.

Instead, we will use a composite liner as the basis of the CPV. On both ends of this tube are the bulkheads, which have a threaded adapter embedded inside to connect valves and other components.

This is inspired by Air Command Rockets, who is building high powered water rockets out of carbon composites. We’re going to use some of his techniques in the manufacturing process.

Composite material

The base material for the CPV is a carbon fiber sleeve. Because it’s woven continuously, the sleeve has no weak points around its circumference. Pretty advantageous for a pressure vessel, where the greatest stress is in the hoop direction. I’m still undecided on the diameter of the sleeve, because it has to stretch between the largest diameter (the liner) and the smallest diameter (the bulkhead endpiece). Right now, it looks like we’re going to use a 70mm sleeve, but this may change to something like 90mm, because a larger diameter can fit more volume per tank weight – that’s why we ruled out a 60mm diameter. Carbon fiber sleeves with diameters like this usually have roving of 6k and weigh about 40g/m. We’re going to use this sleeve for the liner and bulkheads. Some critical sections like the endpiece connection at the tip of the bulkheads may be reinforced with glass fiber weaves.

The matrix consists of an epoxy polymer, which we mix and apply by hand. After laminating, the composite parts cure for about one day. I’m working on a rotating device, which prevents the epoxy from dripping during curing.


A carbon composite liner (which is basically just a long, thin-walled hollow cylinder) gives the CPV it’s overall shape.

Even though the fibre sleeve we’re using has a diameter of 70mm, the liner will have a diameter of 80mm. That’s because the sleeve can stretch between very small and very large diameters. The sleeve is classified as 70mm because at this diameter the fibres are oriented at -+45° to the axial direction. This orientation of the same angle value for all fibres is called a balanced ply laminate. We’re choosing a larger diameter because of two reasons:

  1. The larger diameter allows for more volume per length, reducing the overall length of the rocket. This way the rocket should be a bit lighter, is easier to control and has less stress on the structural components because of the shorter lever arm.
  2. The optimal ply laminate angle for pressure vessels is 54.7°. This is because the stress in hoop direction is twice that of the stress in axial direction.
    A diameter of 80mm corresponds to an angle of ca. 54°.

As the liner constitutes basically the whole length of the tank, it has to be quite long to accommodate all of the propellant. With about 4kg of water inside, a diameter of 80mm requires a length of about 800mm (subject to change).

For manufacturing, we slide the sleeve over an aluminium tube which gives it its shape. The tube is removed after the lay up process. We’re planning to just laminate by hand, rolling and brushing on the epoxy.


Bulkhead with embedded endpiece

At both ends of the CPV, the bulkheads close off the tank and hold the threaded endpieces. For an even stress distribution the bulkheads are shaped approximately hemisperical. The (larger diameter) cylindrical parts gets glued inside the liner. The carbon fiber sleeve must be able to stretch from the large liner diameter to small outer diameter of the endpiece. To shape the flexible sleeve, we’re going to 3D-print a mold on which we will lay up the sleeve.

Threaded endpiece

Because the tanks are going to be useless if we can’t connect anything to them, the CPVs have threaded adapters embedded at both ends. These endpieces get glued into the bulkheads and are wider and conical on the inside to distribute the pressure more evenly. Aluminium is practically the only choice for the endpieces, because it’s so light. 7075 alloy has a particularly high strength which ensures that our threads wont strip out over time. Most importantly, we can turn these pieces manually on a lathe. The tanks are connected with valves and other pneumatic parts which all use the British Standard Pipe thread. In our case, most of the components use a G1/4 outside thread so we’re cutting a G1/4 thread inside the endpiece.


The CPV which holds the pressurized air – the upper tank of the rocket – is probably under a lot more pressure than the lower water tank. I calculated the resulting stresses with Barlow’s formula, and apparently one layer of the carbon fiber sleeve should be enough to withstand the pressure of roughly 30 bar (with which we’re going to pressurize the CPV) easily.

However, since we never manufactured something like these CPVs ourselves, I don’t really trust the figure of the yield strength it should theoretically have. Because of this and safety, we’re going to add another layer of the carbon sleeve over the whole liner and bulkheads. This should provide enough strength to prevent a RUD of the rocket.

Still working on the specifics and details of the CPVs, more should follow after the FAR high pressure tests.

(Edit: high pressure tests confirmed the thrust simulations. Tests with a modified, convergent-divergent nozzle led to a higher nozzle efficiency coefficient, resulting about 8.7kg of thrust at 25bar).


1 thought on “Poseidon Tanks: Concept and Design”

  1. Pingback: Poseidon - Machining parts for test tanks | Mechanics Development

Leave a Comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.