Steam Age meets the Digital Age: Open Source Steam Engine

No, this is not the steam engine that we’re working on…

First off I should introduce myself, I’m Nick and I’ve been watching OSE’s efforts for the last year or so and thinking along tangential lines for a while longer. After graduating in mechanical engineering at UW-Madison and working my previous two years as an engineer at a coal power plant I came to the conclusion: a very fundamental change in the way we power our society is in order. So I have left the job with a personal commitment to try to work towards solutions that make sense for future generations to come.

If you are reading this you probably have similar sentiments – I’m excited about the possibilities of open collaboration towards local sustainable technology. I’ve been to Factor e Farm a couple of times in the past year and have pitched in on the recent compressed earth brick building. After seeing the success of the CEB press and getting a feel for the advantages of open collaboration I decided that working on developing a solar power generation system with OSE would be a good outlet for my commitment towards sustainable energy solutions.

So for the last month or so my mind has been stuck on a subject considered by most to be locked in the history books for good: steam engines. Family, friends and concerned acquaintances wonder if I’ve popped a gasket myself in devoting perfectly good time to this research . Please allow me to share my thoughts on this subject and why I think it is worthwhile to revive it and put some modern touches on it. We are developing a steam engine as the heat engine of choice for the solar power system, and we are aiming to convert our open source tractor to steam power as well.
My motives for pursuing this research are spurred on by a multitude of criteria such as: (1), finding an engine simple enough to be locally manufactured and serviced, (2), ability to take both solar thermal and biomass thermal heat inputs, and (3), operate with low maintenance requirements and costs. Engine building is no small task and is fraught with little details that are easily missed on paper but are a real world show stopper.

For example the Stirling engine looks great on paper, it uses air as a working medium, it can accept high temperature heat inputs – and therefore has a cycle that should get very close to ideal efficiencies. Realizing this in a reliable engine, however, has been very difficult. Low specific heat capacities of air mean that the engine size will be larger than a steam engine and this in turn requires much larger cylinder sizes. To counter this the cylinders usually operate with a pressurized gas, but this in turn requires perfect sealing of the cylinder, all of which puts this engine out of the reach of all but the most sophisticated machine shops.

Other engine options that can take multiple heat inputs – such as the steam turbine – are also full of difficult manufacturing and engineering barriers such as precise machining and high speed dynamic balancing.

So perhaps it should be obvious that the low speed simple steam engine was the main engine to usher in the industrial age up to the 1920’s. So does it have a place in our history now? Well that is the question we intend to investigate and answer in our steam engine project.

The pros to this engine are:

  • Fuel Flexibility: ability to take either direct solar thermal energy or indirect solar energy in the form of combusted biomass. Given that cellulose is both the most plentiful form of biomass energy and that it doesn’t have to compete with food production – we need to consider developing engines around this fuel. Since refining cellulosic materials to ethanol or diesel are not yet proven – external combustion engines like the steam engine are more attractive.
  • Simplicity and Durability: With combustion occurring externally at a steady pace the engine cylinders do not experience the shock of internal fuel combustion. Additionally since full power can be obtained at very low engine speeds the life of the engine is typically orders of magnitudes higher than IC engines.
  • Emissions: With either non-polluting solar energy or biomass as the power source this engine is inherently clean. Since combustion of biomass occurs externally without large pressures thermal Nox is not formed, which would be problem in an IC engine.


  • Efficiency: The large con is that the steam engine is historically of lower efficiency than an IC engine. This however isn’t always true and I think it need not be true in the future. Higher steam temperaturess raise the engine efficiencies and many late model compounding steam engines achieved >20% overall efficiency which is on par with a spark IC, but still half of a good diesel cycle. For stationary power applications running the engine in cogeneration mode were hot water and house heat are produced from the exhaust will eliminate this concern. In mobile applications it should be noted that a spark IC efficiencies will vary from 11% -13% at idle to 25%-28% at full load. The steam engine will remain constant through load variations and so a simple steam engine achieving 19% efficiency will in practice equal a typical gas engine in efficiency.If down the road the steam engine is adopted and developed, higher boiler temperatures and pressures and cylinder compounding could bring efficiencies up to the level of sub-critical Rankine plants, 35-37%. Give the low fuel refining requirements for steam engines, this may give steam power the edge over the slightly more efficient diesel.
  • Steam Danger: Working with steam successfully has required over 150 years of trial and error to master. Luckily we can stand atop that knowledge as we move forward. The flash steam generator – the boiler type of choice – has no danger of explosion.

Adding it all up it seems the steam engine deserves a modern small-scale attempt. So one option to move forward with the project is to simply dig through some old engineering books and duplicate a simple steam engine for a solar thermal boiler tube or biomass-fired flash steam generator.

After considering this option initially, I realized that there appears to be some mechanical baggage that we don’t need in the digital age. Control of yester-year’s steam engine was done mechanically with a variety of steam valves and a centrifugal governor. These parts make a simple steam engine much more complicated than just a simple cylinder with an oscillating piston. Given the availability of low cost, open source control boards, optical sensors, and solenoid steam valves – why not try to improve?

Steam can be injected simply by using electronically-controlled valves – much like modern fuel injection. Down the road – steam temperatures and pressures can also be incorporated into the control board to add safety features like automatic dumping of steam if boiler temperatures and pressures go beyond design conditions.

To me, it sounds attractive enough to pursue.  So I have decided to put my money where my mouth is and fund the first prototype. I have purchased the required valving and electronics and am finalizing my mechanical crank plan. I plan to post as I hit some project milestones and aim to have something running on steam in a few months.  If you want to offer some info or comment on the project please post it under the steam engine construction set on the wiki, I’m also available via email at raaumn at


  1. Jeremy

    I think has “software PLC’s” which might be able to be used for the programmable logic controllers and it has inputs/outputs for the sensors/valves, if feasible for this application at some point. It even seems to have a GUI layout for planning the logic.

  2. Nick

    I’ll have to study the linux plc softwar more,but then this would require the use of a computer at al times for the engine, which may be OK for beta testing, but probably not practical as a finished product. I ordered an open source arduino for $30 that I hope will do the job.

  3. Richard Schulte

    Man, you really know what you are talking about! 19% efficiencies are pretty damn good, and yea, with computer modelling, new advances in materials, and whatnot, we could definately realize a vastly more efficient steam engine than they could have imagined around the turn of the century. steam engines used to be very widely used, I have a family picture of my great grandpa running a steam powered tractor on his land after WWI. He also had a shingle mill he operated with steam power, turning saw log into tens of thousands of shingles a year.

    I especially appreciate the safety of the engine your working on as well. Often, many folks will dismiss steam as dangerous technology… but of course, this is just another thing we have advanced ourselves beyond.

    Finally, applying an arduino to control a flash steam injection system. will be one of the coolest applications of an arduino I have yet heard of. keep up the good work.

  4. Lucas read somewhere in twitter – helps?

  5. Lucas

    re IW&FMS, google for george chan and dream farm and jeff buderer of one village foundation

  6. Geoff

    I applaud the effort to get steam working for humanity again, but I do wonder how you are defining “complexity” here? Steam valves and mechanical governors vs an array of electronic devices such as optical sensors and valves? Which is going to be more accessible technology given target markets? Which is going to be the more robust solution?

  7. Nick

    Your point is certainly a good one. There are a couple of good reasons to try to produce an electronically valved steam engine. Precise control and variation of steam cutoff ratio should allow for steam engine efficiencies to be otpimized. Secondly incorporating a microcontroller will allow for automation of the flash steam montoube boiler.

    I think industry has sucessfully answered the question about electronic controls and sensors in general concerning robustness. Availability also does not seem to be a problem as microcontroller price and availablity continues to increase. However should any serious roadblocks develop with the electronic solution we could just build a mechancical slide valve on existing cylinder and proceed with old fashioned mechanics.

  8. Harmon Seaver

    Some really neat ideas here! One of the things you mentioned — densification of biomass for fuel is something I, and a whole lot of other people around the globe, have been interested in for a long time. The big problem always comes down to the fact that it takes a lot of pressure to make good fuels, like wood pellets. And those machines are really expensive, even small ones, even made in China or India. So I’m wondering if the CEB machine could be modified to densify biomass, at least enough to make decent briquettes. Sure, you probably couldn’t develop the pressures of a wood pellet screw press, which is enough to actually liquify the lignens which then glue the sawdust together into a pellet, but certainly it could do a lot better than the current hand operated presses that turn soaked (retted) biomass or paper into a briquette like people are using in the 3rd world today.

  9. Brian Korsedal

    The arduino is great, but it’s expensive and slow. You might want to look into an FPGA based system. You could probably fit this application on the smallest Spartan-3NA. The benifts would be extremely fast processing and very reliable timing. With a microprocessor based system (unless your running a special OS) it’s hard to get reliable timing.

    With an FPGA you could specify some event to happen X number of clock cycles (at perhaps a 100 MHz clock) after some other event. This gives you 100% reliability and real fine grain control of timing.

    They are starting to use FPGA’s in cars a lot too.

    oh, and FPGA’s can be cheap. You can probably get this FPGA for ~$5-$7 US.

    If you want to know more, write me at brian dot korsedal at gmail dot com.


  10. Tinkerer

    Dear Brian,

    I doubt that one can fit or even need any kind of “operating system” in such a small microcontroller as in Arduino, which is ATmega168 as far as I can see from
    Some specs here: Devices&func=displayDev&objectid=78

    Instead, one compiles a small C-coded loop handling all the necessary monitoring & control, with a cross-compiler in one’s PC,
    and just uploads that to the Flash-memory of the microcontroller,
    which the loop automatically started anytime the microcontroller is reset.

    Somehow I guess the latency wouldn’t be too much for controlling the steam-engine accurately enough. But in any case, applying FPGA’s to steam engines is a nice, wacky idea in itself. (But the devil might be in the soldering.)

  11. Nick

    Brian and Tinekerer,
    Thanks for the great input, I must admit having only a background in mechanical engineering I’m quite a bit out of my element in anything electronic, but I’ll learn.

    So it sounds like a field programmable gate array would be the cheapest and fastest method of control, but with the disadvantage of having to solder inputs and outputs to the FPGA?

    I have thought about the fact that the Arduino might not be able to execute commands fast enough, but I’m looking at probably nothing more than 25hz as the engine itself will runnning at 2.5 hz. Like Tinkerer says it will just be running a loop that is checking the pulse output of a optical sensor, if the pulses fall too low the valve open duration will be increased until the desired engine speed is met. I’ve already ordered the arduino so I guess I’ll find out if it will work soon enough.

  12. Tinkerer


    with that kind of RPM’s (two and halfs in second? 150 RPM?)
    FPGA will surely be an extreme overkill! (Man, these kind
    of devices are used for on-the-fly manipulation of video
    signals!) But beware of us electronic buffs, as we always
    recommend our favourite toys for all possible applications!

    (E.g. a FPGA is overkill even to my project here:
    although, of course, programming FPGA is much more intellectually
    rewarding than programming a microcontroller!)

  13. Tinkerer

    This probably goes outside of THIS topic, but is pertinent
    to the general idea of more resiliency and less centralism.
    So, how far we are from a low/medium tech, small-scale manufacturing
    of integrated circuits/or other components with a similar function?

    It seems that the state-of-the-art what one person can do in a workshop
    is this:
    and even that seems to be bordering on a miracle, at least according
    to many of the commentators.

    It seems that Parallax Inc (that I referred to),
    founded by Chip Gracey, had somewhat small-scale beginnings:
    and see especially:

    But this is not quite at the farm scale yet…

    (Instead of silicon technology ever reaching that point,
    I would bet on some kind of a biocomputer, say
    e.g. a programmable nut that grows in the genetically
    modified walnut tree.)

  14. Tinkerer

    I beg a pardon, but my monologue continues:

    If we think about different ways of controlling a steam engine, we might have a series of technological complexity like this:

    1) a mechanical centrifugal governor
    2) some electromechanical system based on electromagnetic relays & solenoid valves
    (hmm… How one implements a continuous negative feedback loop with these?)
    3) some electromechanical system based on vacuum tubes & solenoid valves (not recommended!)
    4) some electromechanical system based on transistors/ICs/microcontrollers/FPGAs & solenoid valves

    Only options up to the point 2 (or maybe 3) can be implemented with
    the components that one could in principle manufacture by himself,
    without relying on large centralized high-tech manufacturing.
    (Here I’m conveniently ignoring where all the required tools are made and how.)

    It would be quite easy to make relays and solenoids. Hmm…
    (I’m sorry for the speculative contents of these last two comments.
    You can move them to some other section, “wacky ideas”, if you wish.)

  15. Nick

    I guess one of my assumptions on this open source project is that processors and solid state circuitry is going to around and easily acessible so why not use it? However the point is certainly debatable, especially since we are after very high degrees of self reliance in general at OSE.

    If I was forced to do away with the processor I’d head straight for option number 1 since it is simple and proven. I believe that option number 2 could also work if one was a clever electrical engineer. It would be easy to make the solenoid valve and easy to time it open at TDC, however getting to open for the correct duration would be a trick. You’d need to make a circuit that would lag a propotinal amount to speed so that at slow speed the valve would be slower to open than at high speed.

  16. Marcin

    Tinkerer, good comments. Can you point us to any information that could lead to open source solenoids? The reason that Nick gave up the idea of a solenoid valve is that an off-shelf solenoid valve that would suit the purpose cost $300 for one. That was prohibitive as far as cost. So opensourcing a solenoid valve can allow for a low cost, higher performance steam engine.

    Solenoids are nothing big, but the devil’s in the details – regarding the actual fabrication procedure.

  17. Nick

    Very interesting links, i actually picked up a book on building your own vacum tube using light filaments, hand held pumps and a glass jar. I think we really need that sort of ground level tinkering and development so that new approaches and methods are made that hopefully give us all access to technology from the ground up.

    I haven’t actually given up using an off the shelf solenoid valve, I bought this a 1/4″ 8263 from Asco It just had a response time of 4-10ms to open and another 4-10ms to close. Which gives me a worst case of 5hz cycle time which might be able to make the engine run but not optimally. Since I’m looking to run the engine at 2.5hz this valve would be basically opening and closing as fast as possible if it is on the oter limits of the valves cycle time. I could run the engine very slowly though at say 50 rpm. Also I may be able to put a bigger solenoid and bigger spring on this valve and get a much faster response time.

    The other valve I found was skinner valve, I have a pdf manual where they state a little faster response time for a direct acting valve called the skinner A-10.
    Response Time
    • AC–Approximately 4-8 milliseconds to open or

    This still is not good enough to really optimize and for an extra $150 definitely not worth it. After having the asco solenoid taken apart I can say it should be to hard to wire it up with a different solenoid and spring. If my intial testing doesn’t work I’ll try that route, any other ideas?

  18. Tinkerer


    Sorry, I realize that I was shooting bulls with my comment about solenoid valves.
    I thought just generally, that if somebody can make vacuum tubes by himself,
    then making solenoid valves should be “easy” as well. But now remembered
    that these are not just for controlling small-scale aquarium pipes,
    but hot steam with some pressure. But probably not impossible, either.


    Now I would add to the option 2 maybe a potentiometer and some capacitors.
    No, I didn’t consider it as a realistic alternative, any more than
    using vacuum tubes. Just tracing history of electronic “control” components.

    Also, one could use analog/digital IC’s, but with no processor.

    BEAM design philosophy for robotics has developed some very interesting
    hacks and ideas for various kind of negative feedback controls.
    See e.g. (good!)

    (BTW, you don’t necessarily need to purchase the optical sensors.
    Every computer mouse is full of IR-leds & sensors that could be possibly

    But still, I recommend a programmable microcontroller instead of
    special analog circuitry, as if one microcontroller is no more
    manufactured, it is easy to program the same thing on a slightly
    different one.

    And yes, I also believe that processors and solid state circuitry are
    going to be around. But sometimes it’s just useful to extrapolate to
    extremely unlikely scenarios, and see what would then be possible.

    Maybe you need a new page/category: Electronic components from bottom up,
    with links to various sites, what is possible to achieve with workshop-level
    technology. (At least resistors are easy, eh…)

  19. Tinkerer

    (This comment was lost yesterday into some abyss of blogosphere.
    So posting it here again, if it’s of any value. I have removed
    the smileys from all the tongue-in-cheek comments, as Mozilla
    converts them to those hideous yellow pill-faces, so you have
    to guess when I’m not serious.)


    More about the electronic components…

    I guess Brian meant Spartan-3AN, not NA (which is
    probably “not available” at Xilinx for the foreseeable future.)

    I gather, that it is XC3S50AN that is the smallest
    FPGA in the Spartan-3AN family.
    (The document
    gives some idea what one can do even with a FPGA this “small”).

    It is available only in TQG144 package. I.e. one of like these
    but with 36 leads on each side, instead of just 20.
    Here is a drawing:

    The distance from the middle of lead to the middle of next lead
    at the edges (the “lead pitch”) is 0.5 mm (that is, 0.019685 inches).
    Remember that this is for the technology that should be possible to
    maintain with a tiller’s hands.

    So, my advice, stay away from surface-mounted components,
    and stick to the microcontrollers which are done in huge quantities,
    and are available in DIP-package:

    It seems that the smallest microcontroller from Atmel
    is this:
    (or maybe ATTiny12)
    The price shouldn’t be too much, e.g. ATtiny45 here with $2.10 each:

    Microchip’s PIC-series is another hugely popular
    family of microcontrollers in hobbyist projects:

    And if you want to control eight steam engines
    simultaneously, maybe you can do that with “eight-cog”
    (the CNC-controller mentioned is a more realistic application, yes…)

    So, I hope this helps. And I think Arduino was a good
    choice for the prototype experimentation, so you don’t
    need to waste your time on making your own PCB’s and all the
    assorted soldering…

    One thing more: How near the boiler the electronics will be
    in the “final production version”? E.g., would you need components
    with ceramic packaking?

  20. Hugh Lippincott

    I’m not clear about your steam cycle. I see no condenser.
    W/O a condenser you sacrifice a lot of efficiency as you lose all the power from 212F down to atmosphere ~ 80-100F; but you gain the complexity of operating with a partial vacuum. Typically the condenser has to be so big that it is only possible in stationary applications.
    W/o a condenser you probably are running your water only once through so you need a big investment in cleaning up your water or you scale your boiler quickly.

  21. Jonathan Herz

    I read what you wrote about your engine ideas with interest. I agree with you that the possibility for integrating electronic valve timing with a traditional steam engine. When I looked into the most obvious solution, solenoid valves such as you suggested, I found that the requirements for the solenoid are not trivial. If you pursue this, you will want to look closely at the performance of the solenoid, both in terms of duty cycle and actuation time. The results suggest that you need a very powerful, very fast solenoid with nearly 100% duty cycle. This is not inexpensive. I tried reducing the weight of the valve to improve response time by actuating a very light pilot valve, but the problem is that the iron in the solenoid plunger has a lot of mass. You can increase the power of the solenoid by increasing the magnetic field, but greater the magnetic field, the longer you need to apply a given current to generate the field, thus the slower the response time.
    The problem is not insoluble. I know that the automotive industry is looking at solenoid valving, which is an even more demanding application. But they are looking at a highly engineered solution.
    I do seem to spend a lot of time dishing out bad news in alternative energy circles, but I also wanted to let you know that if you wish to pursue your steam engine (and I hope you do), you will need greatly reduce your efficiency expectation below the 19% you quoted. I think you are about four times too optimistic. The sort of engine that you sketched can be expected to produce about 5% efficiency assuming a reasonably efficient boiler. You quoted the Skinner unaflow engine at 19% peak efficiency. Taking that as a benchmark, here are the ways in which an engine such as yours falls short. Ignoring boiler efficiency as outside the scope of this discussion:
    1)The Skinner used pressures around 800 psi, much higher than the 100-200 that is common for most smaller setups.
    2)It operated at several hundred degrees of superheat, which would be difficult to achieve without expensive controls and high alloy superheater tubes.
    3)It was Unaflow and condensing. Unaflow is a great way to go, but if you are not operating condensing at a low condenser pressure, unaflow operation requires complex auxiliary exhaust valve which opens only after the unaflow ports are opened. In any case, a non-condensing engine will be less efficient than a condensing engine. One advantage that marine engines had is that they were sitting in a body of water and could thus easily cool their condensers and keep them very low pressure. If you don’t use unaflow valving then compounding becomes VERY important for efficiency.
    4) your engine is smaller and therefore you will have less mechanical and thermal efficiency.
    5) You do not jacket the cylinder and head with high pressure steam. Again, more condensation losses.
    6) Unless you have a very constant load, you cannot obtain good efficiencies without variable cutoff (varying the expansion ratio to accommodate different loads). One of the advantages of electronic valving is that you can more easily implement variable cutoff, but your electronics would be more complex and you would need some form of sensing for the electronics to determine the right cutoff.

    Those are the major efficiency factors that you would want to consider if you want to achieve better than 5% efficiency. As you can see, there is, as always, a tradeoff between efficiency and manufacturability.

    Good luck with your work. I look forward to hearing more about it in the future.

    P.S. This comment has been floating around my hard drive for a long time. I first started writing it when I read your initial posting. I see that my comments about the performance requirements of solenoid valves have already been discussed. Have you found something satisfactory?

    Jonathan Herz

  22. Marcin

    Jonathan, yes, your points are well taken on the efficiency – we have a long way to go – but none of the obstacles are insurmountable. Nick is not at Factor e Farm any more, so the project is not receiving much attention at present. We’d like to find someone who can lead further development. Are you in a position to do further design work?

  23. Stanislav

    Hello, quite interesting ideas.

    I was busy in auto sport, and so called NOS solenoid would be better for presented task. It has very light-weight plunger/small operating times.

    Also do you consider boiler-less steam engine ? something like heating the head of engine with lots of area, and injecting water into it at TDC… (like flash boiler mounted inside in engine..)

    Up to controllers. Simple VEMS DIY ECU (engine management unit) with old, slow 8 bit at mega processor is fully capable of ignition, injection calculations for 8 cylinder engine…

    And what have been done on engine oiling ? for it is the most difficult part… Or to use less temperature and go for oil-less cylinder with ptfe(teflon)/graphite cylinder

    Sincerely, Stanislav

  24. Stanislav

    PS. about efficiency

    this –

    (simple uniflow engine made from diesel engine, with bash valve with ZERO electronic go for 22 % in efficincy..)

    So it won’t be so bad at all..

  25. Arthur

    Here is an intresting site on flash tube steam engines for model planes and boats

  26. sandra742

    Hi! I was surfing and found your blog post… nice! I love your blog. 🙂 Cheers! Sandra. R.

  27. stu

    Great stuff. I don’t want to tease you with patented ideas but I think this proves out what you’re saying:

  28. Links for January 8th to January 9th |

    […] Factor E Farm Weblog » Blog Archive » Steam Age meets the Digital Age: Open Source Steam… – Adding it all up it seems the steam engine deserves a modern small-scale attempt. So one option to move forward with the project is to simply dig through some old engineering books and duplicate a simple steam engine for a solar thermal boiler tube or biomass-fired flash steam generator. […]