February 21, 2018

Propulsion Part 1

Steam first went to sea in the early years of the 19th century. At first, it was used by navies primarily to tow ships in and out of harbor, being immune to wind and resistant to tides, both serious problems for sailing ships. Later, steamships were used extensively for minor roles, such as dispatch boats. In the 1830s, the first steam warships were built, such as HMS Gorgon. However, the use of paddle wheels meant that they could only mount a limited broadside armament, so only small ships received them. Larger ships retained sail exclusively until the advent of the screw propeller. Besides clean broadsides, screws also allowed the engines to be entirely below the waterline, where they were protected from enemy fire.

HMS Gorgon

The British and French began modifying existing warships, both frigates and line-of-battle ships, in the mid-1840s. The resulting "blockships" were intended to use their steam only sparingly, as the inefficient steam engines gave them only limited endurance. As the engines improved, so did the importance of steam in ship design, and by the time HMS Warrior was designed, her steam plant was as important as her sails. But before we examine them it more detail, we need to look at the basics of a steam plant.

The basic steam cycle, as implemented on Iowa1

There are four main processes in the steam cycle: generation, expansion, condensation, and feed. Generation is where water under pressure is turned into steam by heating it up in a boiler. In the expansion phase, energy is taken out of the steam by the engines, lowering temperature and pressure. This is easiest to think of in terms of a simple reciprocating engine, where the steam pushes on the piston, but the same physics apply in a turbine. Condensation turns the steam back into a liquid, known as condensate. This creates a partial vacuum, which was actually the method by which the first steam engines worked, rather than using pressure above atmospheric. Finally, the condensate is pumped back into the boilers as feedwater.

A box boiler

Warrior's steam plant was typical of the ships of the day. She had 10 box boilers, producing steam at about 22 psi. The boilers were coal-fired and fueled by stokers who shoveled the coal in by hand. The heating was done using fire tubes, the combustion gasses passing in many small tubes through the water that filled most of the boiler.

Warrior's trunk engine today

The engine was a two-cylinder trunk engine made by John Penn and Sons. The trunk engine was an ingenious method of giving high power in a relatively small package. The trunk was a hollow tube through the center of the cylinder, wide enough to give the connecting rod room to oscillate, and essentially cutting out the center of the cylinder. While this may seem bizarre, it allowed the connecting rod to be mounted at the cylinder head instead of behind it, giving a longer rod and smoother movement. The each cylinder had a diameter of 112.1"2 and a stroke of 48". The maximum speed was 56 rpm, and the whole setup produced about 5270 hp.

A diagram of a Trunk Engine

Once the steam had been expanded, a jet condenser was used to cool it. This sprayed seawater into a box on the opposite side of the crankshaft, cooling down the steam. This caused problems as the resulting mix of fresh and salt water was returned to the boilers. When water turns to steam, it leaves behind anything dissolved in it, so running a boiler on salt water makes it scale rapidly. In the 19th century, boilers had to be replaced every few years, an expensive and difficult job.

Warrior under way

Of course, the point of turning a shaft is to turn the propeller on the end. Warrior had a single screw 24'6" in diameter, coupled directly to the shaft. On trials she made as much as 14.3 knots, making her the fastest warship in the world at the time. However, early steam engines were very inefficient,3 and thus Warrior was also designed to perform well under sail. This included a mechanism that allowed the propeller to be disconnected from the shaft and hoisted up out of the water. The 35 tons of propeller and frame required 600 men to haul it up, and the funnel also retracted to reduce wind resistance. Needless to say, this was not a popular feature, and most later ships just had a fixed propeller.

The boiler room on Warrior

Over the next few decades, sail declined in importance and then went away altogether, as engines became both more powerful and more efficient. The way this happened neatly illustrates the interlocking nature of progress. Warrior expanded her steam from the 22 psi all the way to near-zero in the course of a single stroke. This was inefficient and caused rapid wear from varying loads on the machinery. The obvious answer was to expand the steam multiple times in separate cylinders, but that was not particularly practical with the low steam pressures then in use. Higher steam pressure also meant that the machinery itself could be smaller. However, the boiler scaling problems got worse and worse at higher pressures, so the jet condensers were going to need to be replaced by a system that did not let salt into the feedwater.

A surface condenser

Eventually, the surface condenser was developed. It worked rather like a boiler in reverse, passing the condensing steam through tubes in a box filled with seawater. Early efforts were thwarted by rapid corrosion of the tubes, a problem not fully solved until the 1930s. Improved metallurgy also allowed higher steam pressure, and in 1875, Dreadnought and Alexandra were fitted with the first compound engines aboard British battleships.4 On these ships, the engines were mounted vertically, as their thick belts were seen as sufficient protection against enemy fire. The engine on each of Alexandra's two shafts consisted of a 70" high-pressure cylinder exhausting into two 90" low-pressure cylinders, the whole assembly running on 60 psi steam produced by cylindrical boilers. The total installed power was 8,610 hp, and the ship was capable of 15.1 kts. On 680 tons of coal, she could steam 3,800 miles at 7.5 kts, although her performance under sail was poor at best, as steam had clearly become the dominant power source.

Cross-section of HMS Alexandra, with boilers and engines visible

The Admiral class introduced another innovation, forced draft.5 Instead of simply relying on the natural circulation of hot gasses through the boilers, air was forced into the boiler room by fans, allowing more coal to be burned in the boilers. At least in the early years, forced draft was only used when really high speed was needed, and it was not popular with crews, due to the extra effort and wear on the machinery, as well as the need to seal the stokehold. In the Admirals it was also mostly useless because the engines were not able to absorb the extra steam.

A Triple-expansion steam engine

The use of the compound engine allowed sails to be discarded, but even greater economy was desired. The triple-expansion engine was first used on HMS Victoria. In this engine, the steam was expanded three times, in cylinders of 42", 62", and 96". The total power of 7,500 hp under natural draft gave a speed of 15.3 kts, while forced draft could give 14,000 hp and 17.2 kts. Victoria carried 1,000 tons of coal, and could make 7,000 miles at 10 kts, a massive improvement over Alexandra.6

A Cylindrical boiler

In the 1890s, naval engineering continued to evolve rapidly, spurred on by the development of the torpedo boat, and then the destroyer. These small, fast vessels demanded ever-increasing levels of power and lower weight. As a result, they were almost always the first to introduce new innovations. However, this is ultimately a battleship blog, and so we'll focus on the improvements in naval engineering as applied to capital ships.

A Belleville boiler

The next innovation was the water-tube boiler. First introduced on the Canopus class, this reversed the position of the fire and water in the boilers. Water-tube boilers were lighter, could be throttled faster, and were smaller for the same power. The fact that the waterside was smaller meant that steam pressure could rise even more.7 However, the first water-tube boilers, the Bellevilles, proved difficult in service. This was a common problem with new and more sophisticated equipment, which was often operated by men used to older, simpler, and more forgiving gear. The resulting "Battle of the Boilers" saw other types, such as the Yarrow enter service.

Boiler room of USS Illinois (BB-7)

The machinery spaces of a warship in the late 19th century were hellish. They were hot, extremely loud, dirty, and full of vibrating machinery. Steam leaks were a constant threat. The best way to search for a steam leak is with a piece of wood. When pieces start going missing, you've found the leak. The alternative is to notice when body parts are cut off by the high-pressure steam, which is considered a poor second for obvious reasons.

A Yarrow boiler, showing the steam tubes on the left and the downcomers8 on the right

The Canopus class had a total power of 13,500 hp propelling the ship at 18.25 kts from two sets of 3-cylinder triple expansion engines on two shafts, and 20 Belleville boilers operating at 300 psi. She could carry up to 1,800 tons of coal, although 900 tons was the normal value. This gave a range of 5,230 nm at 10 kts, and 2,590 nm at 16.5 kts.

All of the ships described here ran on coal and used reciprocating engines. Next time, we'll look at the dramatic change in the first decade and a half of the 20th century, when this setup was swept away and replaced by oil and turbines.

For more details on early marine engines, see this book.

1 Thanks to Jim Pobog for the diagram.

2 Once the trunk was subtracted, the effective diameter was 104.6".

3 Warrior burned about 3.5 to 5 lbs/hp/hr, and had a coal capacity of 850 tons. At 11 kts, she could steam about 2,100 nm.

4 As with my early design history, I'm sticking with British practice here due to source availability.

5 For some reason, the British insist on spelling this as "draught". I assume this is due to their weird "u" obsession.

6 Some later pre-dreadnoughts had four-cylinder triple-expansion engines, which had two low-pressure cylinders to avoid an impractically large single low-pressure cylinder.

7 The obvious question is why they weren't introduced sooner. David W at SSC has offered an excellent explanation. Basically, steel starts to lose strength at about 300°C, and coal and oil burn at 2000°C, so the steel has to be kept cool by the water. If a tube in a water-tube boiler goes dry for some reason, it softens or melts, then breaks, and the boiler loses pressure. A fire-tube boiler is guaranteed not to have that happen, because it's a big box of water with fire going through it. As they were able to avoid clogging and other mechanical issues, water-tube boilers became practical.

8 Tubes that return water from the steam drum (top) to the lower water drum


  1. February 21, 2018ADifferentAnonymous said...

    Fascinating stuff!

    So with the blockships and later sail/steam hybrids, was it basically steam for combat, sail for transportation?

  2. February 21, 2018bean said...

    It varied. For the blockships, emphatically so. As the steam engines got more efficient, sails were less useful and received less design emphasis, so they got even less useful. Alexandra, for instance, was very good under steam but quite poor under sail. Sails lasted longest in cruisers intended for distant stations far away from coaling facilities. Battleships that would operate around bases got rid of them fairly quickly.

Comments from SlateStarCodex:

  • bean says:

    Today, Naval Gazing begins a look at battleship engineering, starting with the introduction of steam power.

    • cassander says:

      I’ve never understood fire tube boilers. I assume there’s some good reason for them, but water tube just seems to make so much more sense to me. Do you have any idea why firetube boilers were a thing? Lower pressures, maybe?

      • bean says:

        I’m not really sure either, and for the same reasons. Water-tube just makes so much more sense. I may have to poke around a bit on that.

        • bean says:

          I did some digging, and the answer is hard to find. People were apparently experimenting with them as far back as the 1840s (maybe earlier), but they just couldn’t make them reliable for some reason. I’d suspect tube failures, maybe due to chemistry issues with the small volume of water. Apparently ship pressures were well below locomotive pressures until they got the condensers sorted out.

          • johan_larson says:

            It may also be related to how boilers developed. First the boiler was a tank with an inlet and an outlet, with the fire outside/under it. Then a flue was added, to let the fire gases come in closer contact with the tank. Add multiple flues, and you have a fire-tube boiler.

            More here:

          • bean says:

            Yes, but people were doing serious experiments with water tubes at least 50 years before they came into naval use. Initial lock-in can only last so long. I think David W’s explanation below is correct.

      • SamChevre says:

        One immediately-obvious advantage to me is strength: flue gas is not under pressure, while steam is–and bursting strength is a lot lower than crush strength for a cylinder. So a fire tube can be much thinner-walled than a steam tube.

        • John Schilling says:

          Cylinders under compression fail by buckling, and the buckling strength of thin-walled cylinders is very small. As can be verified with any convenient empty soda can.

        • bean says:

          Not really. One of the big advantages of water tubes is that the volume that has to be pressurized is a lot smaller, which makes it stronger/lighter.

        • SamChevre says:

          OK–bad guess on my part. Looking up strength for PVC pipe, the pressure differentials are slightly lower if the interior pressure is the higher pressure.

      • David W says:

        Steel loses strength starting around 300 Celsius, and the loss gets severe by 500 C. Flame temperatures of both coal and oil is well over 2000 C under ideal conditions, but even a non-ideal flame will be well above the softening temperature of steel. This is the basic technology of a blacksmith. In order to keep your steel boiler cooled below the flame temperature you’re working with, you need to keep any steel that is exposed to flame, wetted with water. The water picks up the heat fast enough to keep the steel temperature down. Heat transfer rate of boiling is much greater than heat transfer rate from gasses, which helps.

        In a water tube boiler, this means careful design and control of your water circulation pumps and distributors, to keep an even flow rate of excess water through every single tube, matched to your firing rate. If even one tube is allowed to dry up (that is, fill with steam instead of water), it will rapidly heat, soften and burst and then you’ve lost pressure on your whole boiler. This gets even harder when you don’t have precise control of your water quality, so that tubes and your distributor may be accumulating deposits in unwelcome places, altering your flow patterns.

        A flame tube boiler, on the other hand, can protect itself from the softening effect of excess heat by…keeping the flame on the bottom of the vessel and enough of a water level inside. So long as the flue gasses lose enough of their heat to the water before getting above the waterline, you’re safe. You only need to maintain water level, not flow rate and distribution. Gravity and natural circulation can keep water moving past the flame tubes. Much easier to achieve, which means it makes sense as the earlier design.

        • bean says:

          That makes a lot of sense, particularly when combined with the iffy metallurgy of the time. “Unreliability” didn’t seem a fully satisfactory answer, but that does. Thank you very much. I’ve added a footnote with my summary of your explanation and a link to your comment.

        • Andrew Hunter says:

          This seems to make sense, except: modern (WW2) steam turbines operated with steam well into your danger zone. My copy of Naval Engineering is at home and I’m in Wenatchee, but I want to say 1200psi (which online engineering tables round off to ~600F?) is in play here?

          How does your explanation square with their steam piping not melting?

          • bean says:

            Actually, that’s well below the temperatures involved, because they were using superheated steam. Iowa’s system was nominally 600 psi and 850 F, which is very much in the danger zone.

            I believe the answer is that they simply designed around it. The loss of strength is serious, yes, but it just means you need more steel. The real problem is if you have a section that gets well above its design temperature, but that can be quite a bit above the point strength loss sets in.

          • David W says:

            Bean’s pretty much got it. Loss of strength is very much a curve, rather than an instantaneous thing. Steel won’t melt until 1300-1500 C, but it loses strength along the way. You correct for that with thicker tubes – and metallurgy, and design. Which is a combination of requiring additional knowledge, and requiring additional money and weight. If you check the link I provided, you’ll see a statement that “Strength loss for steel is generally accepted to begin at about 300°C and increases rapidly after 400°C. By 550°C steel retains approximately 60% of its room temperature yield strength”. You’ll also see that they very quickly stop talking about ‘steel’ and start talking about specific alloys (BS EN 10025 grade S275 steel, for example).

            But, if you really need 550°C tolerance, you can get there, you just need to provide twice the steel to do the job. And very solid controls to ensure you don’t ever see 570°C, not even when the captain is on the phone demanding a little more power.

            Some alloys are better than others at heat resistance. As you combine better materials science with better understanding of heat transfer and flame behavior, you can push into higher regimes. Current state of the art is around 1100 F (600 C) and 3800 psig simultaneously. The better you understand the heat generation and transfer, the more you can control the steel temperature to ensure no hot spots. Steam power has been continuously improving ever since it was invented. Still – no one can get to 2000°C, adiabatic flame temperature of most hydrocarbons. It’s all about the design making sure that the heat keeps moving at a rate where your steel stays within its specifications.

            I’m not a boiler engineer, but I’ve gotten boilers quoted for plants I’ve been designing. When I specify something stupid, the typical response these days is ‘are you sure you really need that? We can do it, but it’ll cost you.’ Typically by crossing the line from a lower pressure class of pipe to a higher class, by asking for my steam superheated just a touch more than I should.

            The other failure mode is to ask a boiler to be able to run at too much of variation in demand. You can’t just cut the fuel and water both to 25% nor increase it to 150%, you’ll end up with hot spots.

            Anyway, back to the original point. Water tube boilers are definitely possible, and even preferred once your engineering is sophisticated enough. It’s just that the design is less fault-tolerant, so you’d better have solid engineering, solid metallurgy, quality control, good maintenance, and well-trained operators.

          • Lillian says:

            And suddenly i understand why it was so difficult to make functional and economic water tube locomotive.

          • David W says:

            Hmm, I just realized – when I said ‘the strength of steel declines above 300 C’, I meant as a function of increasing temperature. Perhaps you interpreted that as a function of time instead?

            There’s a range of temperatures where you can simply design around the decreased strength of steel, it’s just that you’d better know what temperature you’ll be dealing with.

    • bean says:

      My comments on Strike Warfare (or why the A-10 and the Gripen aren’t a complete solution) have been revised and reposted.

      • FLWAB says:

        Loved that article bean. I have a question, though I don’t know if you could help me at all: my grandfather worked for Boeing as an engineer from 1950 to 1985. I know for sure that he worked on the B-52 (the picture in your post made me think of it) and the Minuteman missile, among other things. However whenever I try to find any documentation about his role on those projects I don’t find much (he has a fairly unique last name, which helps in searching). So far I’ve found a couple official documents about the Minuteman that had him listed (among dozens of others) as someone the document was distributed to, and an abstract for a presentation he and others gave on the feasibility of transporting nuclear missiles via railroad. I would love to know more about his role, but I’m a novice at this kind of research. How do you go about researching the designers on projects like this? Should I file a FOIA request or something? Would his work at Boeing even be eligible for that sort of thing?

        Not sure if you have any advice on this, but either way I enjoy your blog and how well researched it is.

  • Leave a comment

    All comments are reviewed before being displayed.

    Name (required):

    E-mail (required, will not be published):


    You can use Markdown in comments!

    Enter value: Captcha