35023 'Holland Afrika Line'
Notes by Bob Gates.
A few years ago, I decided to have a go at building a model of a rebuilt Merchant Navy class pacific.
I had studied Keith Wilson's 'Ariel' design in back issues of 'Model Engineer' magazine and felt that this was within my capabilities.
At the same time, I had the opportunity to buy a second hand set of drawings and castings from a builder who had made little progress.
The package of material I received included a letter that outlined some of the known problems with the Ariel design.
Some of the problems related to dimensional errors on the drawings.
These were fairly obvious on close examination and not atypical of model locomotive designs.
However, other known problems related to the design of the outside brackets which are to the West Country pattern on the Ariel design, rather than the correct Merchant Navy pattern.
At the time, I did not know if this difference was significant.
However, researching this error has led me to review the design of all the brackets and stretchers for the rebuilt Merchant Navy type, and produce CAD drawings to facilitate their fabrication using the tab and slot method.
Although the Merchant Navy and West Country classes are very similar, there are many detailed differences.
In particular, the firebox of the Merchant Navy is almost 1 ft longer than that on the West Country, requiring a bissell truck 1 ft longer, and different positions of brake hangers.
Since a lot of the brackets contain brake supports, all of the external brackets are significantly different between the two classes.
The differences become clear when the 'works' drawings for the rebuilt Merchant Navy are compared with those of the Ariel design.
A further implication of the bigger Merchant Navy firebox is in the design of the bissell truck.
Apart from the truck being longer on the Merchant Navy, the truck design also needs to accommodate much more lateral swing in order for the locomotive to traverse the same minimum radius curve as the West Country.
The Ariel design appears to be a modified West Country with dummy castings on the outside.
However, this approach introduces several major errors.
In scaling up from the West Country design, the Ariel proportions are no longer correct for the Merchant Navy.
A more serious error is that the longitudinal stretchers are shown as parallel on the Ariel design, whereas they should taper inwards towards the rear to allow for the increased swing of the truck.
Actually, the West Country design of stretcher also has a slight taper although this is barely perceptible on the drawings.
The taper on the bissell truck is shown in the adjacent photograph, looking down onto the drag box and bissell truck.
Having determined the extent of the problems with the external brackets on the Ariel design, I decided to examine the design of the internal stretchers and motion bracket.
It appears that the Ariel design was based on a small drawing published in 1945 in the Railway Gazette, from which a plan view could be discerned.
However, detail in the third dimension is lost and the rebuild modifications were not shown.
I have obtained the appropriate drawings from the National Railway Museum and these show the frame stretchers as steel castings of a considerably more complex shape that those on Ariel.
There is significant curvature in the third dimension and a complex support structure for the spring hangers underneath.
During the rebuilds, redundant parts of the frame stretcher between the leading and driving wheels were flame cut away.
Also, extensions were welded on to half of the spring hangers to support revised sanding pipework.
The drag box was also considerably more complex than shown on the Ariel drawings.
The drag box and bissell beam were cast as a single piece rather than as separate components in Ariel.
During the rebuild, part of the drag box supporting the boiler was cut away.
A comparison between the works drawing and the Ariel drawing for the internal motion bracket shows a different pattern of lightening holes, and the boiler support pads centred on the motion bracket frame rather than the locomotive frame.
My challenge was to prepare CAD drawings that would allow a fair representation of these components to be fabricated.
However, it was recognised that the three batches of 10 locomotives had distinct differences.
Additional lightening holes were cut in the castings in the second batch, and the patterns were modified for the third batch.
The first batch of locomotives used a completely different design for the 'frame stretcher (behind driving wheel)'.
The third batch was manufactured with a fabricated bissell truck, although these were exchanged with other class members during overhaul.
All of these changes were reflected in the different weights of the three batches of locomotives.
After rebuild, the first batch weighed 97 tons 16 cwt, the second batch 97 tons 5 cwt, and the third batch 96 tons 1 cwt.
I have decided to build a model of 35023 'Holland Afrika Line' and incorporate the modifications appropriate to the third batch of locomotives.
My preference was for a fabricated bissell truck, and 35023 retained its original fabricated design for most of its life.
Unfortunately, it had its bissell truck replaced with the cast variety during the rebuild.
However, photographic evidence shows that the fabricated variety was reinstated when AWS equipment was installed, probably at the final general overhaul.
Preparation of the CAD drawings was quite a challenge, particularly for those representing the between-frame stretchers.
The brackets made for the rebuilds were easier as the works drawings showed the individual components before fabrication.
However, eventually I had a set of drawings that I was happy with and with as many of the errors that I could identify eliminated.
I wanted to assemble everything using the 'tab and slot' method.
This would ensure that everything held together during brazing.
Arranging a system of tabs and slots that allowed everything to be assembled and would keep the assembly square during fabrication was a further challenge.
Eventually, I sent the drawings off to the water-jet cutters and received a large bag of bits in return.
All of the fixing holes for the bracket and stretchers had been cut with a water-jet, so this ensured that they were accurately located.
Furthermore, the bracket backing plates could be used as a template for accurate drilling of the frame fixing holes.
Brazing of the individual brackets and stretchers was quite straightforward.
I use a conventional gas torch and Tenacity No 5 flux with Easyflow (or equivalent) solder, so that the flux sticks in place during brazing.
Some of the pieces required bending to shape before brazing, sometimes involving complex curves.
Some further errors were identified in my drawings during fabrication.
However, these were generally insignificant and the CAD drawings were subsequently modified to make the correction.
Assembly of each bracket generally took less than a couple of days work, involving several heats to build up the component.
Detailed views of the frames from above and below are shown in the following pictures, from which the general shape of the brackets and stretchers can be discerned.
The brackets and stretchers are quite different from the Ariel design, as can be seen from the first photograph of the frame stretcher which locates behind the driving wheels.
The plan view of it is similar to the general shape used in Ariel.
However, the other view shows the marked curvature in the stretcher, and the structure below supporting the spring hangers.
The bissell truck and bogie are shown in the adjacent photographs.
The Ariel design of bogie is not prototypical because the centre is a large bronze casting without lightening holes.
The original was a steel casting with lightening holes, and I have assembled a set of laser cut parts to represent the original casting.
The bogie includes the modified front stretcher required for fitting the AWS receiver.
From the front, this appears as a flat stretcher with a vee shaped cut-out instead of the round bar stretcher in Ariel.
Driving axle boxes and springs
The springs represent a problem as they would appear to be overly stiff in the Ariel design.
So my springs have been made using a spring steel and polypropylene sandwich, to give the design stiffness whilst retaining sufficient strength.
They have 15 elements similar to the original design, instead of the 20 in Ariel.
Five of the elements were spring steel.
Polypropylene was used for the remainder because the usual tufnol was not available in the required 1.2mm thickness whereas it was in polypropylene, a machineable polymer.
The spring hangers are threaded at their ends to provide adjustment for individual axle loads.
The axle boxes were split and machined from cast iron to the design shown here.
The top of the axle box is curved as this better represents the prototype than the Ariel flat top.
The top of the axlebox has a recess for oil that feeds the horn sides and axle.
The bottom part of the axlebox contains a recess for a felt pad oil retainer.
All journals also have an oil-feed from the axle ends, so lubrication should not be a problem.
The chassis, wheels, bogie and bissell truck have been assembled as shown in the following photograph.
The wheels are the same as the Ariel design except that an undercut has been incorporated in the outer ends of the radial recesses, to better represent the BFB patent design.
The undercut was achieved with a special 60⁰ cutter.
The wheels were pressed onto the axles and keyed for location, as wheels have been known to move on the axles when keyed and Loctited onto the axle.
Notes on machining and assembling the crank axle are given below.
Merchant Navy Crank Wheel-Set Machining
The crank axle was machined first.
This was made of four separate components: the axle, journal and two webs.
Initially, the axle was a single piece to ensue alignment and the four components were Loctited together with a 0.001 in gap.
When the Loctite had gone off, pins threaded 2BA were screwed and Loctited in to hold the assembly secure, and the screw heads filed off.
I then cut out the centre of the axle between the crank webs.
Keys were used to ensure alignment of the wheels to the axles.
The first step was to make a jig shown in the photograph with the keyway button.
The jig was made from gauge plate for its hardness, and has wheel centre, the crank pin centre and the keyway horizontally aligned.
The keyway was initially cut as a half moon using a 1/8 inch milling cutter to ensure axial alignment, and was then opened up using a broach.
Next, the jig was pinned to the wheel centre using the keyway button.
The crank pin centre was then drilled through the hole, and the jig and hole pinned.
The keyway could then be cut in the wheel centre using a broach.
Thus the alignment of the wheel centre, crank pin and keyway was transferred to each wheel in turn.
Subsequently, each crank pin hole was opened out by drilling and reaming.
The eccentric was made and keyed to the crank axle to ensure the correct lag angle for the centre cylinder.
The eccentric was also located by grub screws to ensure that it did not move axially along the axle.
Keyways were cut in the ends of the axles and these were 'quartered' using the indexing feature of the divide head.
This is shown in the photograph.
For the Merchant Navy, the centre cylinder is set 6 degrees steeper than the outside cylinders.
Therefore, the angles between the centre crank and the outside crank pins are unequal.
A digital protractor was used to set the crank web to 126 degrees before the keyway was cut in one end of the axle.
The divide head was then indexed through 120 degrees before cutting the keyway in the opposite end.
The final operation was to press the wheels on to the ends of the axle with a 0.0007 in interference fit.
As the centre of the crank axle had already been cut out, I replaced this with a tight fitting spacer so that the axle did not distort when the wheels were pressed on.
Cylinders and Piston Valves
The cylinders and piston valves have been completed, generally following the Ariel design.
The cylinders were bored between centres so that the bores were accurate to size and parallel, as shown in the following photograph.
A completed cylinder is also shown.
I used countersunk socket cap stainless steel screws to retain the piston covers, because separate covers are required to fit over the top and give a smooth finish.
The piston valve covers are not shown in the Aerial drawings, so a copy of my version, which follows the 'works' profile is shown below:
The piston valves represent a significant departure from the Ariel design, because it was felt that an iron valve in an iron liner was prone to seizure, and had to be made to very tight tolerances in order to achieve a pressure tight sliding seal.
Furthermore, there appears to be no means for adjusting the valve position on the valve rod.
I used a bronze liner with piston valve rings machined from PEEK [material].
These have been machined with a gap so that the ring can spring out against the bore, ensuring a seal.
The sequence of manufacture starts with an oversize blank which has radial slots cut in it giving a dog-leg joint to help sealing.
A special tool is used to machine the compressed ring so that the outer surface is circular and the same diameter as the bore of the valve liner.
The gap is squeezed closed apart from a small expansion allowance.
The special tool compresses the ring axially so that the valve ring does not spring open during machining.
Finally, the internal diameter and outer faces are machined in the lathe.
The completed piston valves and their liners are shown in the adjacent photographs.
The liners are made in two halves because it was felt it was easier to make a Loctite seal to the cylinder with the liner in two halves, rather than in a single piece.
The two halves of the liner abut in the middle, so that the completed piston valve can be pushed from one half of the liner to the other.
In the multi-image picture, the piston valve is shown in its liner.
The piston valve assembly is shown in the middle, and the bobbin without its piston rings is shown at the bottom.
The other photograph shows the piston valve components.
The valve rod has been threaded M3 to allow for adjustment.
The end of the rod is not threaded as it runs in the piston valve tail rod.
Coupling and Connecting Rods
The coupling and connecting rods have been made, generally following the Ariel design.
These have all been machined from solid so require a lot of material removal.
Departures from the Ariel design were to locate the bearings with a grub screw inserted underneath, rather using a taper pin.
This avoids the need for a 'step' in the coupling rods which is non-prototypical.
A boss has been added to the rear of the front coupling rods to allow for the oiling point.
Also the coupling rods have not been fluted as this was only found on the rebuilt West Country pacifics and not on the Merchant Navy Class.
Brakes
The brakes have been largely sorted out by Ian Tiplady (see the write up on Rebuilt MN 'Bibby Line').
The main components have been laser cut, and a selection is shown here.
The bracket at the top left of the photograph is located in front of the leading drivers, and was introduced as part of the rebuild programme.
The bracket was a fabrication and details of the shape have been found from the available works drawings.
The brake adjuster is shown top right.
The Tender
The tender has been completed generally following the Ariel design, although I have adopted the alternative Ariel profile.
That is correct for the rebuild, as the locomotives also received new tender tanks at the same time as rebuilding.
The profile shown in the front view of the Ariel tender is for the original build.
There are a large number of tender variations; apart from the different capacities (5000 gallons, 5100 gallons, and 6000 gallons) there are variations dependant on the type of water treatment fitted at the time to the prototype.
My model has the BR briquette water treatment, and therefore has a small extra filler cap in front of the water filler.
The position of this extra filler varied from locomotive to locomotive.
All of the class received a tender with a cowl over the coal space when rebuilt.
However, on some tenders, the cowl was subsequently cut back, presumably due to damage from coal when coaling.
Cowls were not fitted to the West Country class tenders.
The tender is shown in the following photos.
Details of the opening lockers and and fire-iron tunnels are included.
The vacuum brake cylinders and pistons were machined from solid aluminium bar.
In order to facilitate firing, the whole of the cowl and locker assembly is removable.
The vacuum reservoir assembly is fixed to the tank top and provides a positive location for the cowl assembly.
Mild steel was used for the cowl as it is not part of the water space and does not distort on brazing, whereas brass distorts badly.
Soft soldering was used in the assembly of the tank.
The tank was then sealed using a resin usually sold for sealing vintage motorcycle tanks.
'Petseal Ultra' is used where tanks are thinned by rust as it creeps into any nook and cranny and gives a good seal.
I paint it on all my joints on the inside as I always find they leak somewhere.
The tank top is also removable to access the handpump for maintenance, the top being made from stainless steel for rigidity.
I feel that the work that has been done so far has produced a solid foundation for the model, with most of the errors and inaccuracies in the Ariel design resolved.
The next step is to produce the motion work.