A Repository for my Ideas on Steam Injector Theory and Design
|Andrew Johnston||16/11/2015 23:03:13|
6575 forum posts
Warning: This thread may contain nuts, 'real' engineering, sums and mixed units , as I am building the engines under the imperial regime but use metric for my physics.
There is nothing on my drawings regarding the injectors for my 4" scale traction engines, so I've been playing with some sums to see what size I need.
The engines are compound, double acting, and run at 170pisg, so the injector will need to work at that pressure and generate at least that pressure on the output. It is usual to add a bit, so let's say they'll need to generate 180psig.
Although the engine is a compound we only need to consider the HP cylinder to calculate steam consumption. The HP cylinder diameter is 2", stroke is 3.75" and assuming a cutoff of 80% and a maximum speed of 600rpm we can calculate the volume of steam used per minute. I get:
The specific volume of steam at 170psig is 0.1542m³/kg. Dividing by this we get the mass of steam per minute to be:
Obviously 1kg of steam is generated from 1kg of water and 1 litre of water weighs 1kg. So we need 1.2 litres of water per minute. In real money that's 2.1 pints, or and a drop. It would be helpful if the injector was capable of filling the boiler even with the engine running so let's say we need a 3 pint injector.
If I decide to build a test boiler to play with injectors it will be electrically powered. So how much power do we need to continuously run a 3 pint injector with input steam at 170psig?
Three pints equates to 1.7 litres and hence 1.7kg of water per minute. As a rule of thumb let's assume that the volume of water entering the boiler from the injector is made up of 11 parts water from the tank and 1 part water from the condensed steam at the injector input. So we need one twelfth of the mass to be steam, so 0.142kg/min. The total enthalphy of steam at 170psig is about 2786kJ/kg. So for the steam we will use in the injector per minute we need 396kJ/min, or 6596J/s, or 6.6kW. Seems a lot? But how long does your 3kW kettle take to heat 1 litre of water to 100ºC, let alone the 190ºC associated with steam at 170psig, and also convert the water to steam?
Edited By Andrew Johnston on 16/11/2015 23:04:35
Edited By Andrew Johnston on 16/11/2015 23:04:51
|julian atkins||16/11/2015 23:23:31|
1252 forum posts
i have a very successful (commercial) design for 43 oz per min. happy to sent it to you if you PM me. D A G Brown has a very successful design for 40 oz per minute in his excellent book.
all tried and tested! for steel boilers an additional check valve on the overflow is recommended.
|Bill Starling||17/11/2015 11:29:01|
|98 forum posts|
Julian - please will you elaborate on/explain your comment ' for steel boilers an additional check valve on the overflow is recommended.'
I'm about to fit an injector (Blackgates No. 6, 4 pts/min) to a steel boiler and should have thought a check valve on the overflow would stop it working. Also what difference will the material of the boiler make?
Thanks very much.
Andrew - delighted at the prospect of further demonstration of your erudition.
|julian atkins||17/11/2015 13:36:09|
1252 forum posts
an injector can suck in air via the overflow. this is the chirping sound that is sometimes heard.
you do not want oxygen dissolved in the water in a steel boiler.
it is quite a simple matter to fit a non-return valve on the overflow for injectors used with steel boilers.
others may take the view that water treatment used on a steel boiler avoids the need for the above.
|Bill Starling||17/11/2015 14:07:10|
|98 forum posts|
Thanks very much for prompt, comprehensive and understandable reply. Surely this forum at its best.
|Andrew Johnston||17/11/2015 20:44:02|
6575 forum posts
Bill: I've been thinking about injector theory for some time now. However, at Forncett this year I had a good chat with Robert Bailey (aka 'steam_guy_willy' ) and he kindly showed me his electric powered boiler. Dunno why I didn't think of electrical heating; stupidity I suppose. Anyway, the discussion fired me up, as it should be easier to design/make/run an electric powered boiler than mucking about with liquid or gas firing. Especially as the kitchen has been designated as the test area.
From a practical point of view I am envisioning a simple cylindrical boiler with a couple of large stays, each containing a 1kW heating element. After looking at triacs for control of the heaters I then thought of using an off-the-shelf temperature controller. Much simpler, and as I do electronics for the day job, I don't really want to do it in my spare time. The boiler will need to run up to about 200psig and will only need to produce saturated steam, ie, no superheating needed. Given that I can close the loop on the heater control by measuring the temperature of the steam rather than the pressure, since there is a monotonic relationship between pressure and temperature for saturated steam. I can probably do away with a water gauge too; if the temperature rises too much due to a fault, or no water, then I can just turn off the heaters. A fellow member of this forum has given me a smack round the head and pointed out that temperature controllers are ten-a-penny on Ebay, rather than the expensive ones at Farnell/RS that I was looking at. You can get a controller, thermocouple and SSR (solid state relay) for less than £20. Bargain!
As a further thought, the only hot part of the injector is the steam inlet, and the only high pressure parts are the steam inlet and water outlet. So it could be possible to make the test injector body from a transparent plastic so that you can actually see what is going on in the internals.
You never know, something might even be on display at Forncett next year.
Edited By Andrew Johnston on 17/11/2015 20:44:36
22578 forum posts
Andrew, might be worth looking at the instant boiling water taps that are about and how they heat the water. The units under the sink are quite small and I'm sure a bit of electronic tweaking could up the temperature assuming they have a good factor of safety on the pressure rating.
Out of interest how do your calculations compare with te size of injector suggested by LSM?
Edited By JasonB on 17/11/2015 20:57:39
|Mark C||17/11/2015 21:02:42|
|707 forum posts|
I don't know anything about steam (apart from it is hot and hotter still if you constrain it) but I can tell you that domestic plumbing is generally rated to 5 bar if you didn't know already (i know as my cold mains is often up at +8 bar and had to fit a regulator). As for the instant hot water taps, they might be different but I would imagine the tap is on the cold side?
|Bill Starling||17/11/2015 21:28:37|
|98 forum posts|
The idea of a transparent injector is excellent. I'm sure there are a lot of people, in addition to me, who would love to see 'the works' in action. We'll form an orderly queue at Forncett. (Next Model Engineers' Day will be on 2nd October - never miss an opportunity for a plug!)
|Andrew Johnston||17/11/2015 23:16:28|
6575 forum posts
Interesting, I'll have a look at the water heaters. Although the cartridge heaters I was planning to use are encased in stainless steel they will not see boiler pressure, but will be embedded in large copper stays. The datasheet recommends a H7 fit and also use of their thermal goo, which RS don't sell. Typical, why sell the cartridge heaters if you don't sell the goo needed to use them? It's a bit like Farnell selling expensive high voltage. high current connector shells and crimps, but not the even more expensive crimp tool.
I've no idea what LSM recommend, although I don't recall seeing anything on the drawings. However I do have a priori knowledge. A couple of years ago Terry Baxter ('injectorman' on TT) kindly lent me his bending rolls for my rear wheel rims, so I asked him what injector I needed, and he immediately said 3 pints. Nice to know why he said that!
I didn't know that domestic plumbing was rated to 5 bar. I wonder if there is some sort of regulation in the system as I know that the 4" or so mains water distribution pipes run up to 250psi (17 bar). Some years ago I did all the electronics for a small ultrasonic leak detector that was put into the mains pipe and went with the flow, so to speak, listening for the ultrasonic noise that is indicative of a leak. In order to narrow down the location of a leak the device listened for a series of ground based radios above the pipe and tagged the data each time it heard one. Now the pipes were 1" wall thickness cast iron, so how did the radio work through that? As the textbooks say I'll leave it as an exercise for the reader, for the time being.
I've been looking at the injector steam nozzle this evening. As far as I can see it is essentially a convergent/divergent nozzle similar to those used for impulse steam turbines. Steam turbines are pretty ubiquitous, so there is plenty of literature regarding them and the design of the nozzles. One thing puzzles me though. Most of the diagrams show a relatively short convergent section and a much longer divergent section, which is the opposite of that shown for all the model injectors I've seen. Why? Could be a number of reasons:
1. I've got it wrong and the design of the injector steam cone is not the same as an impulse steam turbine nozzle
2. That's how the first model injectors were made, and everybody has subsequently stayed with it because it works
3. Some other reason I haven't thought of
I've made a note of the next Forncett steam up day in the diary.
|julian atkins||17/11/2015 23:31:10|
1252 forum posts
i know nothing about steam turbines.
the divergent taper on the steam cone ought to be 3 or 4 times the throat diameter to achieve the Venturi effect. taper is 9 degrees.
the convergent taper on the inlet steam side isnt so critical. you will no doubt have read of those who have used just a plain drilled hole here, but all the successful commercial injectors have something like a 13 degree taper here, and DAG Brown explains why in his excellent book. the 13 degrees isnt critical and has more to do with body length and making standard reamers.
i would strongly recommend that you buy DAG Brown's book, plus the book by Ted Crawford. you will find both invaluable, and Ted also covers the larger sizes you are after.
Edited By julian atkins on 17/11/2015 23:33:16
|julian atkins||18/11/2015 00:00:01|
1252 forum posts
i should of course have added Bob Bramson's articles and book to the above. Bob i think has an interest in higher pressure injectors of the larger sizes and ought to be ideal for larger miniature traction engines, though a lot of Bob's published stuff is for medium sized injectors of 26 oz per min suitable for 5"g locos. it would be well worth contacting Bob.
andrew's requirement for a very high pressure miniature injector may well require a bit of re-thinking. it would be instructive to acquire a 2mm Penberthy injector and take it apart etc and carefully measure. not something ive ever done as no one has given me one, and far too big for my requirements! i do however have all of Penberthy's technical stuff and data.
|Gary Wooding||18/11/2015 08:05:25|
|968 forum posts|
Sorry, can you explain that bit please?
|Andrew Johnston||18/11/2015 11:42:55|
6575 forum posts
Cylinder diameter is 2", fixed by the design
Stroke is 3.75", fixed by the design
Cutoff is 80%, this is arbitrary, but is what I will be aiming for when I re-design the valve gear
Maximum engine speed is 600rpm, probably a bit high even for a 4" scale engine, but one has to put a peg in the ground somewhere, and it's what I used when re-designing the water pump
We only need to consider the HP cylinder since the LP cylinder steam comes from the HP cylinder, not from the boiler. Of course we're ignoring use of the simpling valve, but that is only used very short term.
The volume of a cylinder is:
V = pi x radius squared x length
So for one stroke (TDC to BDC or vice versa) the swept volume is:
pi x 1² x 3.75 = 11.78 in³
But with a cutoff of 80% steam only enters the cylinder for 80% of the stroke so the volume we're concerned with is:
11.78 x 0.8 = 9.423 in³
The engine is double acting, so steam fills the swept volume twice per revolution, so:
9.423 x 2 = 18.85 in³
Multiply by 600 (rpm) to get the swept volume per minute:
18.85 x 600 = 11309.73 in³/min
I'm too idle to look up the conversion factor from in³ to m³, so I multiply by 25.4³ and divide by 10^9 to get cubic metres:
11309.73 x 25.4³ / 10^9 = 0.185m³/min
Of course there are all sorts of ifs and buts that can be made against my assumptions, but you have to start somewhere, and the calculations are only a marker to size the injector.
2904 forum posts
Given that water at room temperature naturally contains something like 10 ml of oxygen per litre of water, how do you avoid the oxygen that is there already? Do you pre-boil all the water you put in your boiler? I've never heard of this being done in "real" steam engines although I'm certainly no expert. They must be expected to cope with the oxygen that is released as the water is brought up to temperature and pressure as well as air that is inevitably introduced by the injector surely?
|Chris Gunn||18/11/2015 19:39:33|
|429 forum posts|
Has anyone made a successful lifting injector for 4" and 6" Scale traction engines?
|Andrew Johnston||18/11/2015 19:59:45|
6575 forum posts
Thoughts on the Steam Cone
You may want to grab a wet towel for the head before reading................
The purpose of the steam cone in an injector is to convert the high pressure steam from the boiler into a parallel stream of steam at much lower pressure (ideally below atmospheric) and at very high velocity (supersonic). This is exactly the same as the function of the nozzle in an impulse steam turbine, so I am going to assume that the design process is essentially the same.
The expansion of steam through the nozzle is taken to be adiabatic, and isentropic, ie, no change in entropy but there is a change enthalpy. In essence some of the enthalpy in the incoming steam is converted to kinetic energy in the output steam.
The shape of the nozzle to achieve this will depend upon the properties of the fluid used. The equation of continuity of mass states that the mass flow (kg/sec) equals the cross-sectional area (m² ) times the velocity (m/sec) divided by the specific volume (m³/kg). As stationary saturated steam expands in a nozzle the velocity increases faster than the specific volume so we need a convergent nozzle shape. However, after some critical point the specific volume starts to increase faster than the velocity, so we need a divergent nozzle. That is why the steam cone has a convergent/divergent shape. It turns out that there is a critical output pressure below which there is no further increase in mass flow in a convergent only nozzle. The ratio between this pressure and the input pressure is about 0.58 for saturated steam.
So I think that what this says is that for a convergent nozzle, with a given throat size, as the output pressure drops there is an increase in mass flow, and velocity down to a certain point, after which neither increase with falling pressure. This is not what we want. As far as I understand at the critical pressure the steam velocity at the throat is at the speed of sound in steam at that pressure, ie, Mach 1. To expand the steam further, and increase the velocity into the supersonic region, we need a divergent nozzle.
From a practical point of view it doesn't seem that the actual shape of the convergent part of the nozzle is particularly important. However, the throat area is important, as it is this that determines the maximum mass flow. The shape of the divergent part of the nozzle seems to be more important, although I am not quite sure why at this stage. It may have something to do with creating turbulence if the divergence is too fast? There also seems to be some experimental evidence that a curved divergent nozzle can help in ensuring a parallel output stream.
I don't know about the rest of you but my head hurts now. Next thing is to play around with the equations and put some numbers in. That should improve my understanding, as opposed to the handwaving exercise above.
Edited By Andrew Johnston on 18/11/2015 20:01:18 - Unwanted bl**dy smiley!
Edited By Andrew Johnston on 18/11/2015 20:02:09 - We need a preview facility
Edited By Andrew Johnston on 18/11/2015 20:03:23
22578 forum posts
Andrew I was just looking back at Julia's data from her run up engine hill. She used 80lts of water to cover 16miles at an average 6mph which is something like 30lts/hr
As her 6" has a similar size boiler to your 4" that may be a usefull guide to actual steam used compared to theoretical use. Though I can understand that a higher capacity injector is desirable as you don't want to have it running all the time.
Out of interest what is the amount of steam used by an injector to move say 1lts of water? This would need to be added to total steam usage.
|julian atkins||18/11/2015 21:21:23|
1252 forum posts
a successful automatic re-starting and lifting injector can be made in any size from 4 oz per minute to whatever larger delivery you want!
in fullsize steam loco stuff the GWR 10X injector was used by BR as a standard. it did and still does work very well.
i actually think that some miniature injector designs work much better than fullsize, though, except for Andrew's case, the operating range is not so great.
what makes a succesful lifting injector in miniature is a matter of design detail - those details being well known and understood if you read up on such things. i dont want to hijack Andrew's interesting thread or go off topic, so am happy to supply those details via PM etc.
everything Andrew states in his last post has been well documented for over 100 years about steam cones in injectors. if Andrew doesnt have this info to hand i will gladly supply him with same - he only has to ask!
Edited By julian atkins on 18/11/2015 21:28:50
Edited By julian atkins on 18/11/2015 21:42:03
|Andrew Johnston||18/11/2015 22:06:59|
6575 forum posts
So 30 litres per hour is 0.5l/min or 0.88pts/min; rather less than I am calculating. I am being pessimistic using 80% cutoff, and optimistic with a speed of 600rpm. However, as stated we want the injector to fill the boiler so some over-capacity is no bad thing. Unlike a water pump there is no penalty if the injector is not being used.
One litre of water weighs 1kg. If we assume that for each 12 units of water going into the boiler 11 come from the feedwater and 1 comes from the steam we need 0.083kg of steam for each litre of water injected.
We also need to take into account the behaviour of the injector as the input steam pressure decreases. A quick scan of the literature shows that the output velocity of the steam from the steam cone doesn't vary much with input pressure. However, the mass does, so there is less mass of steam going into the combining cone. If we want to stick with the 1 to 11 ratio that implies less feedwater and hence less water overall going into the boiler with reduced input pressure. No idea if that is correct; my off-the-cuff ideas done while typing have a habit of going pear-shaped because of something I didn't see.
Please login to post a reply.
Want the latest issue of Model Engineer or Model Engineers' Workshop? Use our magazine locator links to find your nearest stockist!
You can contact us by phone, mail or email about the magazines including becoming a contributor, submitting reader's letters or making queries about articles. You can also get in touch about this website, advertising or other general issues.
Click THIS LINK for full contact details.
For subscription issues please see THIS LINK.