In September 2009 I purchased a secondhand GRS train comprising a live steam GWR 2021 and seven wagons. I made this purchase because I have a garden railway with both gauge 1 and gauge 3 tracks, but had no operable G3 stock. I shall not go into how this situation arose, but the GRS train seemed like a perfect solution - instant train! - all I had to do was hand over lots of money. The engine is very attractive, and a British goods train would be a very welcome presence on the railway. Also I would buy myself some time to do the other tasks that we all manage to acquire for ourselves, before building my own G3 train. A perfect plan. A plan which went wrong; but you knew that. Here is the train on the track for the first time, not in steam.
The assembly workmanship turned out to be fair, and sometimes a little less than that. This has turned out to be acceptable except for how it plays sometimes with the main problem, the latter being the subject of these notes.
And so I shall get to the ugly part of my reporting right away. To be nice about it, the engine has turned out to be a pile of junk. And the problems lie mainly in the design and manufacture, some in the assembly and fitting; as I think the reader will discover. It should be noted that my criticisms are of the engine as a miniature live steam engine, not as a model of the prototype. I imagine that the electric powered version is a very nice engine; but an actual steam engine brings realities that cannot be fudged. And the carelessness inherent in the production of this engine reflects on the suppliers, who sell, not give away, the equipment. The market is a small-volume niche, of course; so large engineering efforts are not likely to be feasible. But a poor product that frustrates the market does not seem to be a good idea.
I was mollified to discover that I am not the only one disappointed by this engine. As examples, I list things that other people have written about this (live steam) engine. These comments mostly are from gauge3.org.uk or g3forum.org.uk; my comments are in square brackets.
OK, enough of that odious stuff. However, in my experience there is too much truth in the words above. My first attempt at running the engine was just awful, and so I put in quite a lot of work right away trying to put things right. But I ran into time constraints - which was somewhat bitter given my decision to spend money to get an "instant train" - and had to stop. That was more than a year ago. But now I have found the time to have another go at it. I have had some success, and hope for more. And I have realised that I really owe it to myself and the community to record information, success, and failure. And the method I think will work out best is for me to record things as I go along, diary style, from where I am today. This will not produce a logically correct ordering of topics; but it will get information recorded.
So the rest of what is here is something of a blow-by-blow account. I trust that the content will be interesting enough for the reader to see beyond mistakes, contradictions, and reversals. And I shall be strident in my criticisms since I think that will be healthy in the long run. I am determined to get resolution this time around. The engine is going to run or be scrapped.
The First Good Run|
Yesterday I had the best run ever of the engine; but the run came in two pieces. This image was taken after opening the throttle the first time.
If you look closely at the coupling rod pin on the front axle you can see that it is blurred. As soon as the throttle was cracked the front wheels started to turn, but nothing else happened. It was quite funny really; the engine appeared to be standing on level track with nothing holding back the engine other than its own weight and friction, spinning its powered wheels (The motor drives the front axle only). So, I re-installed the coupling rods, which involved drilling out a bushing. I have no idea why that was necessary, it seems that some selective assembly had been done before and I was correcting some fitting error. On the re-start, the engine ran around the track, climbing and descending 1:33 gradients, running through junctions correctly, and generally behaving itself, for the first time. Hooray! However, it would only do this if I ran it backward; running forward caused a couple of derailments.
So here is where I am currently. The engine is running uphill toward the camera with about 45 psi boiler pressure (I checked the gauge earlier against a large industrial gauge, the small gauge read nearly 50 psi with the large gauge reading 55 psi). The engine has never done nearly this well before; and I shall describe how I got it to this point - but after addressing the wheel spinning and derailments.
After the successful run I checked the position of the centre of gravity both with an empty fuel tank and with a full fuel tank - actually, faked with a cup of water. The balance point runs from about 3/8 inch to about 7/8 inch behind the centre axle with the fuel tank empty and full respectively. The screwdriver stuck in at the right side is holding up the wood for the image shot; the reader will just have to believe me that this is the, highly unstable, balance point.
Now, if we assume that the compensation is working perfectly the weight on the rear two axles is the same. Furthermore, the weight of the engine is supported by the front axle and the compensation pivot. So, if the centre of gravity line of action went through the compensation pivot there would be no weight on the front axle. The reader can see from the image that the centre of gravity is much closer to the compensation pivot than the front axle. This is quite a surprising result; one tends to look at the whole engine and the whole coupled-six wheelbase and judge it very differently. The arithmetic shows that the weight supported by the front axle runs between just over a quarter (empty fuel tank) to a sixth (full) of the weight of the engine. So, with a full fuel tank the weight on the front axle is half what you might guess. And if the compensation is sticking a little, putting more weight on the centre axle than on the rear axle, then it is easy to see how an un-coupled front axle could spin merrily without taking the engine anywhere. At the moment I can lift the front of my engine, on the workbench, and, sometimes, the compensation sticks enough to hold up the front wheels so that there is fresh air underneath them. Thus, the compensation works almost all the time...
What is more, if the engine is running forward and uphill, bearing in mind that the water in the boiler will move to the back of the boiler, what is the weight on the front wheels? I have no idea, but I do speculate that this, coupled with a bit of compensation sticking, might be the cause of derailment. I am not convinced of this idea, but it is something to think about; and there is that, stange sounding, claim by someone else that "It derails if I allow more than 25 p.s.i." I wonder if running backward would stop that problem.
Motor Valve Timing|
I have found what I consider to be a fundamental flaw in the motor design. This flaw can be largely eliminated and evidence that this is an important issue comes from "before and after" bench tests on air, as well as the improvement in track performance under steam. In the bench tests, which were on a chassis that had some tight spots and other unsmoothness, the air pressure required to keep the engine idling went from the 15-18 psi range to a, steadier, 8 psi.
The problem lies in the porting of the motor. A common fundamental design objective in an oscillating engine is that there be instantaneous no-flow at dead centre, but in all other crank positions the cylinder be either consuming supply pressure steam or discharging to exhaust. Satisfying this objective can be achieved by having port diameters (ports usually are round) so that at dead centre the perimeters of the ports are line-to-line; and this, also, is usual. By line-to-line I mean that, if you could see them, the ports would look very similar to OOO, but with the edges of the Os touching. The ports from one side to the other of this simple diagram would be supply, cylinder, exhaust; with supply and exhaust in the fixed portblock and cylinder in the oscillating cylinder block, and clockwise rotation looking at the text diagram.
If the no-flow design objective is not followed by having the ports too small then there is a part of each cycle of the engine where the cylinder is sealed but the piston is moving. This means that the cylinder (with piston) is compressing the exhaust followed by expanding it again. Even if these two actions cancel to some degree they will not be as powerful as using this part of the cycle to exhaust completely and start a new power stroke with fresh supply steam. Furthermore, the steam flow into the cylinder will not be as great as it would be with a larger port; and this will reduce the pressure in, and hence the power from, the cylinder. The extent of the power reduction needs analysis to establish its magnitude; but, intuitively, the ratio of the port diameter to the cylinder diameter will be a major factor.
If the no-flow design objective is not followed by having the ports too large then there will be a part of the cycle where there will be a direct leak from supply to exhaust; which is wasteful, and also cuts down on the length of the power stroke.
The geometry and trigonometry required to design or analyse an oscillating engine are quite straightforward and are all that is required. Steam expansion characteristics, for example, are not significant simply because of the (wasteful) consume-discharge cycle of the oscillating engine. The only important analysis improvement over the past forty years, since geometry and trigonometry have not changed, is the ability to mechanize the calculations with a computer; which permits the glorious, luxurious, ease, of "what if" calculations. However, an evening spent with paper, pencil, and a cup of tea can produce instructive and satisfying results.
In the motor that I have, the ports in the portblock and the cylinder were 0.054 inch diameter (#54 drill bit). My measurements of the hardware and subsequent calculations assuming the instantaneous no-flow criterion give 0.078 inch diameter (#47) as the design target. So I convinced myself that the valve events for the engine were poor. Also, probably the steam flow was constricted excessively since the area ratio of the two port sizes is more than 2.
I note that a common port size for oscillators with cylinder bores in the quarter to three-eighths inch (6-10 mm) range is 0.063 diameter (#52). And I have an 0 gauge engine with a single-cylinder, double-acting, 1/4 inch bore, oscillator, that, in the early days before I put a throttle on it, would run at 8 mph (about 5000 rpm for the motor); and that has 0.078 diameter (#47) ports and passageways.
The first thing that I did to my motor is open up the ports to 0.078 diameter (#47). That is twelve ports, two on each cylinder, eight on the portblock. The objective being to change the valve timing to correspond to the instantaneous no-flow criterion. This was the only thing done between the "before and after" bench tests referred to earlier.
Re-working the ports on the motor must be done carefully. The main problem is that twist drills "grab" so easily in brass under the conditions that pertain here, and they can vandalize an existing bore instantly. It is important not to use a drilling machine or power tool to do this job. Also, to avoid stripping down the block and cylinders it is important to avoid letting swarf fall into the engine parts. I proceeded by holding the workpiece in a bench vice with soft jaws with the portface angled down from the vertical; the idea here being that swarf would fall out-of rather than in-to the workpiece. Then I put a #47 drill bit into a pin vice and tightened it hard, I did not want the drill bit to slip part-way through the operation. I did not use cutting oil or anything similar; but I do not know if that was the best decision. I pushed the pin vice gently to start the cut - as square to the portface as I could - and just as soon as the drill started to bite, around half a turn, I pulled on the pin vice whilst continuing to make the cut. Another couple of turns saw the drill bit come out of the brass, and there was a counterbore about 1/16 inch deep. I checked for swarf in the hole and, in a couple of cases removed some with a dental pick. All the holes had a small volcano crater on the portface after the re-work; but it was easy to touch up with a honing stone - I used oil this time - and finally wash off the portfaces with naptha, alcohol, or something.
The image shows the counterbores in the ports quite nicely. The one on the left appears to be a little deeper than that on the right. This difference is not important so long as the depth is adequate to allow free steam flow. Comparison of the equations of the port area and the cylindrical area of the counterbore gives a depth of a quarter of the port diameter for equality. But the requirement is more complicated than that. My guess is that anything greater than 0.04 (1 mm) is plenty. Definitely it is not worth the risks in taking a second cut simply to even up the depth.
On the workbench, I made big picture decisions about how I should proceed with modifications and configuration of the whole Pannier Tank engine. Amongst other things I decided that the minimal running configuration should include the frame overlay. The reason for this decision is that the overlay provides the horncheeks and I guessed that the lack of these was at least partly the cause of binding of the compensation, leading to derailing. Also decided was that the superstructure, other than the overlay, should not be necessary for running. This decision requires elimination of parts of the superstructure from the runable configuration. In particular, the rear of the engine requires modification since, as designed, the gas tank has to be mounted after most other assembly. In general, the cab area is somewhat jumbled in the design.
I completely re-built the chassis running gear with new spacers, which were missing; and cleaned up the compensation mechanism, including de-burring, chamfering, etc. I made a start on modifying the rear of the engine which meant bolting the gas tank to the top of the overlay. Thus I ended up with my minimal running configuration. At this point another track run was in order; I needed to convince myself that I knew about all the major modifications that need to be made. Assuming that this track test was successful I could change my posture from fault analysis to design development. This would enable me to stretch out the rebuild schedule and just plod along until the job is done. The key items to establish were:
Well, there has been quite a long hiatus in my re-building activities - from April until November, in fact. The first reason is that I ran into trouble with my scalloped wood bridge (see an earlier image): I started to have fairly regular gratuitous uncouplings when trains were run over the scallop crests - this was with gauge 1 trains fitted with Kadee couplers. This track characteristic does not impress visitors and is a precursor of nasty crashes. So, my railway efforts went into replacing the wood bridge with a new steel bridge. Since I was doing a major replacement, also I installed a steam-up siding that I wanted but never thought would happen. The task is finished, but it has been a huge time sink.
The images show the new bridge (20 feet long), the new steam-up bay, and a new junction leading to the bay. Mostly it is all new construction, although I have re-used some of the wood bridge sleepers and all the rail. In the images the bridge itself cannot be seen very well; it is a ladder, constructed from 1 inch by 1/8 inch strip and 1/2 inch square tubing, and the track rests on it with some lateral support, superfluous really I think, which stops the track moving sideways too far. I like to say that the bridge material is hot-rolled slag with steel enhancements; however, it was quite easy to solder with 56% silver solder. The remaining task is some minor gradient smoothing. But it is all good enough for me to get back to the G3 Pannier Tank re-build that I am quite anxious to complete.
There is already a problem with the railer on the end on the siding - it works so well that I like to load trains into this siding; so the siding has become a train-staging area, rather than a steam up bay. And that does not work if there are engines already in place on the siding. I may have to move the railer or make another.
This is mainly show-off on my part. But it is nice to look at; dual gauge junctions are significantly more complex than single gauge; and satisfying when completed successfully 8-)
But then, in September, I ran into the second problem that became another delay for the Pannier Tank rebuild. When the track was first laid, in 2004, it was put into a newly created, and partly raised, walled, garden. This new garden had 18 cubic feet of soil put into the raised, walled area. Now, I understood the need to tamp down new dirt, and I thought that I had done this adequately. Hmmmm. It will suffice to write that, at present, much of the original track bed is at least six inches below the surface. If anyone reading this is thinking of laying new track on new dirt, please let this be a very serious warning - unless you like re-doing jobs until you are sick of them.
So it got to the point where, at the worst sink area, I ran out of width - I was unable to raise the track anymore by stuffing rock underneath it because the stones would fall out of the sides and create a real mess. The effects of the angle of repose, and aesthetic and available space considerations, demanded another solution. Which meant another bridge (but only 6 feet long this time 8-). It all looks much better, and I am glad it is (almost) done. But it took me from September to November. And there is no image because we have a storm going through today and the railway is under a few inches of snow.
I started today, the Second of November. The plan is to get all the technical stuff complete, test the engine, and then, assuming success, take it apart and re-paint it, etc.
Installing the Superheater|
On their engines, Roundhouse run a copper tube down the length of the boiler flue as a superheater. I decided to copy this method; the perceived problem being installation with the need to get a non-kinked bend at the delivery ("smokebox") end with an accessible fitting on it. This task turned out to be quite simple, much to my delight. I soldered a fitting on the cab end; annealed the other end for about four inches; put a tube-bending spring on that end; fed the tube into the flue; pushed and pulled it sideways and out of the chimney flue with a wood spatula and pliers; removed the spring; and soldered another fitting onto the end.
This image shows the steam pipe installation, which is the tube with the union in it. The steam pipe feeds into the top of the motor. The other tube is the exhaust, which comes out of the bottom of the motor. Also shown is a guard that I installed to protect the motor and, primarily, the main drive gears. I plan to insulate the steam pipe.
I stayed at the front, and modified the chimney a little, which has two problems. One problem is that the bottom of the chimney fouls the boiler because not enough vertical room has been allowed for it where it pokes through the false tank superstructure. So the bottom of the chimney is hard against the boiler top once the superstructure is screwed in place; the screws being remote from the chimney area. I may shim up the superstructure when I do the final assembly; I do not think that there is much more that can done.
The other problem is that, as designed, steam oil that is carried up the exhaust tube and into the chimney dribbles down onto the boiler and makes a mess on the boiler. I have wrapped my boiler with ceramic sheet and so I have soggy lagging as well as an oily boiler; see the image. The real plan is that the oil dribbles down the chimney flue and drops down onto your friend's track, making it his problem.
The image shows how I dealt with the oil problem. I soldered a tube into the chimney. This tube extends down into the flue past the top of the exhaust tube, which is configured somewhat like a blast pipe in a smokebox. The hope is that this chimney extension will implement the real plan for spent oil distribution, instead of making a mess of my boiler.
In the image, the chimney is upside down. When installed, the chimney passes through the superstructure seen in the background, and is secured by the large nut that is then on the inside of the superstructure. The nut becomes flush with the bottom of the chimney, and together they rest hard on the top of the vertical boiler flue.
Rebuilding the Cab Region|
There is not a lot to write about. The images show the pipework before and after installing the coal bunker/butane tank. The Roundhouse burner can be seen, also the steam-pipe entering the flue alongside the burner. The regulator handle is temporary since no work has been done on installing radio control.
The next thing to do was run the engine to find out what had worked, and what had not worked. I ran the engine as before, hauling my seven wagon goods train. It was a grey day: not a lot of light; my camera struggles under such conditions. In the image the dummy smokebox door is notable by its absence; but also note that the train is going up a 1:33 gradient.
The train ran as before in many ways; but there were improvements. Here is what I saw.
My engine runs a lot better than it did; it will start a fairly heavy train on a 1:33 uphill gradient; I have not tried it on rice pudding skins. There is more to do, for example I still want to fit radio control. But these notes already address their objective adequately. What follows is a summary of things to consider for anyone possessing one of these engines, especially if it is still yet to be built.
In the text above I identified three areas to investigate: Motor Power, Derailing, and Boiler Steaming. Here I group summary in the same way, adding Miscellaneous as a fourth area.Motor Power
First and foremost: the primary cause of the derailing was not in the engine, but in my track.
However, I do not think that it is going too far to write that the engine is more prone to derailment when going forward than when going backward. And the reason is the axle loading configuration (or weight distribution). I am still intrigued by the comment that "It derails if I allow more than 25 p.s.i. ...".
If the reader would like further information, then I shall be happy to oblige.|
6 November 2011