I'm looking at the Apollo CSM Propulsion Systems. These were bi-propellant hypergolic systems. However, for some reason each type of propellant was divided into a storage and sump tank.

For example, it says:

Sector 2 (70°) contained the service propulsion system (SPS) oxidizer sump tank, so called because it directly fed the engine and was kept continuously filled by a separate storage tank, until the latter was empty.

Same thing for the fuel tanks, one storage and one sump. That is four tanks total for the main engine.

The question is why? Why divide tanks like this?

Was it needed for some kind of pressure regulation? I don't see how. The total prop mass is given as 18.41 t, and the sump tanks appear to be the exact same size as the storage tanks. So if the sump couldn't be used when the previous tank was empty, that is wasting 50% of of the props or about 9.2 tons as unusable... That sounds like a huge, impossibly wasteful amount of unusable props.

A quick Tsiolkovsky equation check confirms. Total mass 24.52 t, only using 9.2 t of fuel with a Ve of 3.08 km/s, so dv would be 3.08*(24.52/15.32) = 1.449 km/s. The article says that the Apollo CSM dv was 2.8 km/s which is almost 2x bigger, and keep in mind there was supposed to be an Apollo LEM attached to the CSM for about half of its burns anyway. So all this to say, no I don't think the props in the sump tank becomes unusable if the storage tank goes empty.

And also I don't think any sump tanks exist for the RCS systems despite using the same bi-propellant pressure-fed style (AFAIK).

Edit: This is wrong. The ACS system in the article says:

Each cluster of thrusters had its own independent primary fuel (MMH) tank containing 69.1 pounds (31.3 kg), secondary fuel tank containing 45.2 pounds (20.5 kg), primary oxidizer tank containing 137.0 pounds (62.1 kg), and secondary oxidizer tank containing 89.2 pounds (40.5 kg). The fuel and oxidizer tanks were pressurised by a single liquid helium tank containing 1.35 pounds (0.61 kg).[23] Back flow was prevented by a series of check valves, and back flow and ullage requirements were resolved by containing the fuel and oxidizer in Teflon bladders which separated the propellants from the helium pressurant.

So each ACS cluster also splits up the fuel and oxidizer into two tanks each or four total. I don't know if they were considered "sump" or not or have the same reasons.

Could it have something to do with ullage? I don't see how. In Apollo spacecraft, ullage was provided by the RCS, which doesn't have a sump tank.

Could it have something to do with restrictive geometry? Looking at the diagrams, I'd have to say no. The sump tank appears to always be right next to the storage tank. I see no reason why all that volume cannot be collectivized into a single tank in the same location.

Could it have something to do with redundancy/safety? Again, looking at the diagrams, I'd have to say no. If the storage and sump are directly adjacent, then any kind of tank explosion is almost certainly gonna disable the other one too.

Edit 2: The Apollo LEM Descent Stage also had 4 large tanks instead of 2.

DPS propellant mass: 18,000 lb (8,200 kg) stored in four 67.3-cubic-foot (1.906 m3) propellant tanks

This appears to be a deliberate system design choice for all the hardware, but I cannot understand why.

So, why was Apollo designed this way? Why use sump tanks and storage tanks instead of one big tank?

  • $\begingroup$ @OrganicMarble the sump tanks were of a slightly different design, having propellant retention features at the outlet that the storage tanks lacked Any reason why a single larger tank couldn't have those features? And if ullage is provided by RCS, then what retention features are they and why are they needed? $\endgroup$
    – DrZ214
    Commented Aug 24, 2022 at 15:05
  • $\begingroup$ @OrganicMarble This is spawning more questions. If the sump tank is full then why wouldn't you need a settling maneuver? The storage tank could be a blob of liquid anywhere too and would not flow to refill the sump. Larger single tanks wouldn't have fit in the service module design. Can you explain how or why with a good diagram? In some diagrams I can see walls dividing sectors but they appear totally unnecessary to me. Just remove the wall and fit a 2x size tank in 2 sectors. $\endgroup$
    – DrZ214
    Commented Aug 24, 2022 at 15:47
  • $\begingroup$ @OrganicMarble Why are sump tanks needed? If there's a structural reason why they went with two tanks instead of one, the answer belongs here. $\endgroup$
    – DrZ214
    Commented Aug 24, 2022 at 21:58
  • 1
    $\begingroup$ @OrganicMarble What do you know about it, compared to the people who designed it? Nothing. You're kidding, right? That's exactly why I'm asking the question. I don't know why they used two sump tanks. I don't see any reason why structural/ullage/pressure regulation would restrict it. But I can see they did the same thing (4 instead of 2 tanks) on the RCS, and also the LEM Descent Stage. That's why I'm very curious and hunting for a reason, and asking about it. Can you at least refer to a page or chapter in the pdf? I read thru a lot, lots of good data but can't find explanations as to why. $\endgroup$
    – DrZ214
    Commented Aug 25, 2022 at 12:00

2 Answers 2


I'm not sure if a definite answer can be given by anyone except the original design engineers, but here's an educated guess.

The Apollo Lunar Surface Journal website hosts two excellent documents:

  1. "NASA Apollo Command Module News Reference, North American Aviation, 1968", of which a scanned copy is found here. In particular, section 7, " Service Module Overview", is relevant.
  2. "Apollo Operations Handbook, Block II Spacecraft, Volume 1", wich can be found here. The relevant section is "Service Propulsion System (SPS)", which starts on page 2.4-1.

First to address is why two tanks were used, instead of one big one. From the News Reference:

The service module is a relatively simple structure consisting of a center section or tunnel surrounded by six pie-shaped sectors. The basic structural components are forward and aft (upper and lower) bulkheads, six radial beams, four sector honeycomb panels, four reaction con­ trol system honeycomb panels, an aft heat shield, and a fairing.

(source: "NASA Apollo Command Module News Reference")

A few pages further the following diagram is shown:

SM general arrangement

(source: ibid)

Here the beams are indicated as "radial web beam", and it can be seen that the beam extends along the full height of the center tunnel and radially outwards to the outside perimeter. The document explains:

The radial beams are made of solid aluminum alloy which has been machined and chem-milled (metal removed by chemical action) to thicknesses varying between 2 inches and 0.018 inch, thus mak­ ing a lightweight, efficient structure.

(source: ibid)

The six compartments separated by "walls" (the radial web beams) are a structural design choice. From the above quote it can be concluded that they tried to make these beams as thin as possible, but leaving them out to make a bigger section was apparently not an option. Hence, the need to divide up the fuel and oxidizer storage over two tanks.

In this top-view image it can be seen that the tanks completely fill up their respective sections (they have to be cylindrical, because they're pressurized):

Service module sectors

(source: "Apollo Operations Handbook, Block II Spacecraft, Volume 1" (Figure 2.4-3))

This should answer as to why there were two tanks instead of one. Why one was named "sump" tank, I can only guess: a "sump" tank is is typically a tank that collects a fluid (e.g. engine oil, water in a basement, etc.) and then that liquid is pumped from there to somewhere else, i.e. a collection tank. Looking at the following functional diagram, that name makes sense:

Excerpt from SPS functional flow diagram

(source: ibid (excerpt from Figure 2.4-1))

Fuel/oxidizer is expelled from the storage tank by pressurized helium, then collected in the second (sump) tank, and then fed from there to the engine:

The total propellant supply is contained within four similar tanks; an oxidizer storage tank, oxidizer sump tank, fuel storage tank, and fuel sump tank [...] The storage and sump tanks for each propellant system are connected in series by a single transfer line. The regulated helium enters the fuel and oxidizer storage tank, pressurizing the storage tank propellants, and forces the propellant to an outlet in the storage tank which is directed through a transfer line into the respective sump tank standpipe pressurizing the propellants in the sump tank. The propellant in the sump tank is directed to the exit end into a propellant retention reservoir. [...] The propellants flow from the propellant sump tank, through their respective plumping [...] to the engine injector.

(source: ibid)

I'm assuming that the sump tank needed to have a slightly bigger diameter to accommodate the retention reservoirs, engine feed lines, etc., which would explain why the sectors of the SM were not identically sized (i.e. 65/65 degrees instead of 70/60 degrees), but this is speculation.

Similarly for the LM: the descent engine in the center precluded large tanks, hence the split among to similar-sized tanks on opposite sides of the engine to maintain the center of mass as much as possible.

  • $\begingroup$ The "Apollo Operations Handbook" is a wonderful document with tons of details diagrams of every subsystem in the CSM. Worth to browse through! $\endgroup$
    – Ludo
    Commented Aug 26, 2022 at 21:18
  • 1
    $\begingroup$ Great answer! Those radial beams support the command module and escape system during the multi-g ascent boost but then get carried all the way to the moon and (mostly) back, so you can bet a lot of effort was expended to minimize the weight of that design while maintaining the desired factor of safety. $\endgroup$ Commented Aug 26, 2022 at 22:19
  • 1
    $\begingroup$ but leaving them out to make a bigger section was apparently not an option. This is probably the key then. I did not think about Saturn 5 launch loads on the CSM at all, (but correct me if I'm wrong, the escape system only pulls the CM, not SM/LM). This also reminded me that the CSM was originally designed for direct ascent, engine was 2x more powerful than necessary. I guess that doesn't matter in the face of Saturn 5 launch loads anyway. $\endgroup$
    – DrZ214
    Commented Aug 26, 2022 at 23:36
  • $\begingroup$ And you know, via the topdown diagram, i can see now the limiting geometry is the radius between center and outer walls. Those tanks barely fit behind the outer wall. This brings up one more thing tho: The center section has 2 helium tanks, but how much of the engine sticks up into it? Cant find good diagram/pics of that. If needed i can do another question about this and structural loads, thanks for the info so far. $\endgroup$
    – DrZ214
    Commented Aug 26, 2022 at 23:38
  • 1
    $\begingroup$ "collected in the second (sump) tank, and then pumped from there to the engine" - I don't think pumped is quite the correct verb since AIUI the LEM motors were purely pressure-fed by the helium (and the engine diagram doesn't show pumps). If that's correct then maybe just "fed from there to the engine"? $\endgroup$ Commented Aug 26, 2022 at 23:56

This answer is almost 2 years on, and it's only conjecture on my part.
That said, what I've read in numerous accounts of other systems leads me to think it worth suggesting the following.


  • Sector 2 (70°) contained the service propulsion system (SPS) oxidizer sump tank, so called because it directly fed the engine and was kept continuously filled by a separate storage tank, until the latter was empty.

The key factor here may be "kept continuously filled".
When a motor must restart in 'weightless' conditions the restart is often preceded by an ullage burn using thrusters to settle the propellants in the 'bottom' of the tanks. Pumping vapour is not a good idea.

When a craft operates at a mix of low thrust, sudden thrust decreases or change of direction or attitude, situations may arise where propellant presence at the exit point cannot be guaranteed with an adequately high degree of confidence. In such cases, having a sump tank that is guaranteed full, or always so close to full that not ingesting vapour can be guaranteed, would make overall system design far less demanding. ie no need to always monitor all factors and guarantee safe pumping, or prevent pumping if unsafe.
Having the latter occur when eg on final on Lunar approach would be a very bad idea.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.