Steam Engine Reviews/Mike Stone

From Open Source Ecology
Jump to: navigation, search

The following comments were sent by email to Mark Norton from Mike Stone.

Sept. 29, 2011 - Mike Stone to Mark Norton

I was browsing the pages about the steam engine design and noticed a mistake. The page Steam Engine Design/Historic defines a unifow engine as one that only admits steam on one side of the piston. That's incorrect. An engine that only admits steam from one side is called a 'single acting' engine. One that admits steam on both sides of the piston is called a 'double acting' engine.

In a uniflow engine, steam only travels in one direction through the cylinder. The opposite is a 'counterflow' engine, where steam is admitted and exhausted through the same port at the top of the cylinder. It's both possible and desirable to build a double acting uniflow engine.

The uniflow engine achieves efficiency by developing a thermal gradient along the wall of the cylinder. That prevents pressure losses from condensation when you're working with saturated (just about to condense) steam. In a counterflow engine the expanded, cool steam absorbs heat from the cylinder wall as it leaves. When the next batch of steam is admitted, it loses heat to the walls and some of it condenses, causing a drop in pressure. As the remaining steam expands, that water re-evaporates due to the lower pressure, removing even more heat from the cylinder walls. When the re-evaporated steam leaves the cylinder, the energy is lost. The uniflow engine avoids those problems by keeping the cylinder hot at the head end, thus preventing condensation. In counterflow engines, the solution is to use superheated steam.. hot enough that it won't condense, even at the exhaust temperature and pressure.

Using bash valves costs you more efficiency than the uniflow design buys you though.

In an efficient steam engine, the compression stage ends with the gas in the clearance (the space between the top of the piston and the top of the cylinder) at the same temperature and pressure as the supply steam. Then the inlet valve opens and the flywheel moves the piston for a certain distance, filling some portion of the cylinder with supply-pressure steam. Then the supply valve cuts off and the rest of the outward stroke extracts energy from the steam by means of expansion. By the time the cylinder gets to the exhaust stage, the cylinder pressure should be the same as the pressure in the exhaust line. The piston pushes the exhaust gas out of the cylinder for some part of the return stroke, then the exhaust valve closes and the remaining gas is compressed to the temperature and pressure of the supply steam.

It's impossible to achieve those admission and exhaust conditions with a bash valve and still do any work.

If the cylinder pressure equals the supply pressure when a bash valve opens, no steam will flow into the cylinder. In fact, since the valve opens before the piston reaches TDC, the piston will push steam out of the cylinder and into the supply line from the moment the valve opens to the moment the piston hits TDC. Steam will flow into the cylinder from TDC until cutoff, but given the nature of a bash valve, the cylinder volume at cutoff is exactly the same as it was at the start of admission.

If the volume and pressure at the start of admission equals the volume and pressure at the moment of cutoff, you have no net flow of steam into the cylinder. The expansion you get on the way to exhaust will exactly equal the work you did compressing the gas on the upstroke, and the cylinder pressure will exactly equal the pressure in the exhaust plenum by the time the piston reaches the exhaust ports. Again, there will be no net flow of steam out of the cylinder. The whole cycle is just an exercise in expanding and recompressing the same volume of gas over and over again.

To get any flow of steam through a cylinder using bash valves, the supply pressure has to be higher than the cylinder pressure at admission, and the cylinder pressure has to be higher than the exhaust pressure at exhaust. If that's the case, you get bursts of sudden expansion, cooling, and condensation at either end of the cycle. Beyond that, the energy from overpressure at exhaust leaves the engine as waste, not as work.

Worse yet, the amount of working steam in your cylinder is proportional to the pressure difference between the cylinder and the supply/exhaust. If the supply pressure is twice the cylinder pressure at admission, your working steam only occupies half the volume of the cylinder. If the supply pressure is three times the cylinder pressure at admission, the working steam will occupy 2/3 of the cylinder. That isn't so bad in itself, but the overpressure at exhaust will rise at exactly the same rate, and all that energy leaves the engine as waste, not as work.

There are also mechanical issues. The pins that open the bash valve have to push against the difference between supply pressure and cylinder pressure in order to open the valves. Since the pins hit the valves before the piston reaches TDC, the relative motion between the pins and the valves will be nonzero, and the force necessary to open the valves will equal the pressure neessary that turns into actual work. In other words, the term 'bash valve' is entirely correct.. you get significant impact between the pins and the valves on every cycle. Then you get an inrush of pressure that hits the top of the piston before the piston reaches TDC. Both of those events will apply sudden braking force to the piston, and send shock through every component between the piston and the flywheel.

That wouldn't be quite so bad in a double acting or multi-cylinder design. The braking forces acting on one cylinder would be balanced by expansion in another cylinder, so at least the shock would stay in the crankshaft. All the flywheel would see is a sudden reduction in the torque spinning it forward, not a sudden burst of torque trying to spin it backward. Even so, you'd have an engine that spends a whole lot of effort pushing energy around internally, and the amount of work you get out of it will still be directly proportional to the amount of energy lost as waste.

With those disadvantages, a counterflow engine running superheated steam (thus eliminating condensation losses) would probably be a lot more efficient than a uniflow basher, and would definitely run more smoothly.

The uniflow engine is a good design overall, but I'd really suggest you find a better valve system. Corliss valves are fairly efficient, easy to make, and easy to control.

If you're going to run at low pressure (HIGHLY advisable if you're designing something to be built and operated by non-engineers), I'd also suggest you add a condenser. A uniflow condensing engine taking superheated steam in at maybe 30psig and exhausting to a 10psi vacuum (about 5psi absolute) would be a pretty efficient device.

The closer you can get to hard vacuum at exhaust, the better your efficiency will be. Steam expanding from 2atm to 1atm (about 30psi absolute to 15psi absolute) doubles in size. Steam expanding from 2atm to a 10psi vacuum (30psi absolute to 5psi absolute) expands to six times its original size. Steam expanding from 2atm to a 12psi vacuum (30psi absolute to 3psi absolute) expands to ten times its original size. Distance traveled equals energy from heat and/or pressure converted to work.

For a small engine, running the exhaust steam into a simple lab aspirator will give you excellent condensation. See Wikipedia Aspirator Pump, [1] or for the really efficient version, see Wikipedia Injector, [2]

It's also worth noting that the efficiency of the uniflow design is proportional to size and inversely proportional to speed. The smaller an engine is, the harder it is to get a thermal gradient that will do any good. The faster the engine runs, the less time any given volume of steam spends in contact with the cylinder walls, and the less heat can transfer between them. For a 4000 HP monster like the the Skinner Unaflow [3] where the pistons are twice as tall as the engineer running them, running at maybe 50-100 rpm [4] (ref: section III).

The thermal efficiency of uniflow pays off. For an engine with maybe 1L of displacement running at 2-4000 rpm.. not so much.

That doesn't mean thermal efficiency is the only advantage to uniflow though. This page in Practical Machinist, [5], quotes a book that says uniflow designs work well for wide variations in load, where ordinary steam engines do their best over a narrow range of loads.


Sept. 28, 2011 - Mike Stone to Mark Norton

mjn: Thank you for pointing these things out to me, Mike. Would it be ok to post this to the OSE wiki?

Certainly. Ditto for this one.

mjn: As an aside, I didn't choose the bash valve engine approach. That was more or less specified by Marcin Jakubowski, the founder of OSE based on his understanding of the White Cliffs solar project. White Cliffs used a converted Lister Diesel engine and bash valves. That design had a condenser that generated a good vacuum on the exhaust side of the engine. It was also a counter-flow design, with valves at top of the engine.

I checked the White Cliffs pages linked from the OSE site, and the design looked like a uniflow.. the cylinder design is essentially the same as the one you have on the OSE page, with bash valves at the top and an exhaust ring toward the bottom. Of course, they also say they saw about 20% efficiency from the engine, which isn't particularly high.

The efficiency is kind of an illusion though. You can keep the efficiency of a bash valve engine relatively high if you keep the overpressure, and thus the output power per stroke, relatively low. It just means that you're using 100 pounds of machine to do what 10 pounds of machine could do with a better valve scheme.

mjn: Since getting involved in this project, I've been looking hard at an electronic valve system with separate valves for steam input and exhaust. The key issues with these is the seal and speed of operation (open/close time). I note that others have overcome these problems and have built steam engines with electronic valve gear. Many of these are patented, however. I am planning on setting up some experiments with a rotating valve controlled by a stepper motor. See [6]. What would be your recommendation on how to proceed?

By all means do the tests. Cars are moving a lot of control to electronics for a reason. Electronics resist many of the problems that plague mechanical systems.. dust, vibration, physical wear, etc. They do have their own problems though, many of which are invisible to the naked eye. Keep an eye on the electromechanical interfaces though.. those tend to combine the worst of both worlds.

Personally, I'd move as much of the control system outside the engine as possible. Most of an engine's complexity comes from trying to cope with variations in the input or output conditions. With good buffers at the input and output, you can make a very simple engine run efficiently. And generally speaking, it's easier to control the environment in a way that gives an efficient engine engine stable operating conditions than it is to make an engine handle varying operating conditions efficiently.

Start with a plain-vanilla engine designed to work at a specific input pressure/temperature, output pressure/temperature, and output torque. Put pressure/temperature sensors in the cylinder so you can confirm that the engine is doing what you expect, but also put pressure/temperature sensors on the input and exhaust lines and some kind of torque meter on the output shaft.

Put pressure/temperature control hardware on your input and output lines, and a motor/generator on your output shaft. Make sure the supply steam is at the right levels as it enters the cylinder, make sure the exhaust is at the right levels as steam leaves the cylinder. Have the generator charge a battery when the demand lies below the engine's preferred output level, and use the motor to convert stored electricity to torque when the demand rises above the engine's preferred output level. Trust the engine to handle everything in the middle, and if it doesn't, replace it. If an engine can't work efficiently under those conditions, no amount of makeup will make a pig look like a pinup girl.

I'd stay strictly mechanical with the engine though, especially for the valve timing. It's fairly easy to make a machine that stays precisely in sync with itself under stable operating conditions.. certainly easier than it is to make an electromechanical system work to the same degree of precision. Besides, giving someone the option to futz with the valve timing while the engine is running creates an almost irresistible temptation to do so, and IMO little good can come of that.

For maximum flexibility, you can design the engine in a way that lets you adjust the valve timing while the engine is shut off, but holds it steady while the engine is running. That will let you use the same engine for different jobs, and in theory could allow you to put engines in series for multiple stages of expansion.

Another benefit to using buffers is that it makes the engine a drop-in component to the system. If you don't like the uniflow, you can disconnect the cables and put in a counterflow engine. Single acting, double acting, crank slider, wobbler, they'll all work with the same basic control system.


Oct. 1, 2011 - Mark Norton to Mike Stone

mike: Put pressure/temperature sensors in the cylinder so you can confirm that the engine is doing what you expect, but also put pressure/temperature sensors on the input and exhaust lines and some kind of torque meter on the output shaft.

This is along the lines that I was thinking, Mike. Some sensors will be needed to provided feedback to the controller for timing, but additional sensors in an experimental rig would give me more insight into what was going on.

mike: I'd stay strictly mechanical with the engine though, especially for the valve timing.

So are you advocating a valve linked to the crankshaft via an eccentric? This is contra to the idea of electronic control. What is your reasoning behind this suggestion?

My wife sorta promised me an Arduino Uno kit for Christmas, so experimentation with electronic valve controls won't happen until late winter, I suspect. I'm pretty busy at the moment anyways.

mike: Single acting, double acting, crank slider, wobbler, they'll all work with the same basic control system.

I can see that if the controls are separate from the engine itself.


Regarding the engine, the current design uses a cast iron pipe with a cylinder sleeve insert. The sleeve was selected from off-the-shelf auto parts. It is of moderate cost, but eliminating it would further reduce the cost to develop the engine, provided extensive re-boring and honing wasn't needed.

The piston head, as currently designed, is turned on a lathe from cast iron. Given the current 4" bore of the engine, that means starting with a pretty large billet. It also means drilling out the interior to reduce weight. One person suggested that cast iron wasn't needed and that another steel alloy might be easier to work with (esp. welding the piston stem in). I think it could also be made shorter - the current head is 4" long - I don't think it needs to be that long. It uses two piston rings, also purchased from auto parts suppliers.

The cylinder size of the current design was determined by two factors: the desire for a decent power output (15-20 hp) and the large supply of quality steam. Well, since then, that supply of steam (SolarFire) has dropped out of the OSE project. Eerik Wissenz (who developed SolarFire) and Marcin Jakubowski had a bit of a falling out (all too common on this project). The current solar steam generator is a trough design and I have some doubts about the amount and quality of steam it can deliver. OSE (Marcin) has put little though or energy into developing a fuel-powered boiler (wood, bio-gas, etc.). Personally, I have some serious concerns about the safety and legal requirements around boilers that might make it difficult to supply an engine that consumes a lot of steam, especially if high pressure and super-heating was needed. It would only take one major boiler explosion to put a stop to further development. As such, my feelings lean towards lower pressure, lower temperature, and lower steam requirements - something that could perhaps be generated by a flash boiler using coiled tubes, similar to the boilers used in old steam cars. I think the TinyTec engine is along these lines.