Javascript is required

A web-site by Rob Speare


Constructing an original type Merchant Navy class - 'Channel Packet'

Words and pictures by (the late) David Smith.

I am just old enough to remember seeing one of the last steam-hauled 'Golden Arrow' boat trains going through Sevenoaks.  On the front of the train looking very smart with flags and arrows, was a large green flat-sided loco.  At the time, I just thought it was an ordinary loco with an unusual appearance.
                -     Little did I realise what was behind that flat-sided casing !

About 7 or 8 years ago, I was just finishing off my 'Lord Nelson' in 3½"gauge and beginning to think about another project.  I wanted to build an express loco in 5" gauge and being a 'Southern' fan thought about a Bulleid pacific in its original form.

I was made aware that there was a published design for a rebuilt Merchant Navy by Keith Wilson, known as 'Ariel'.

By coincidence at the same time, I was offered some of the drawings for this loco and a few parts from a project that had been started by someone else, and abandoned at an early stage.  I therefore started to investigate the feasibility of using the Ariel design as a basis for an unrebuilt loco.  In hindsight, that was probably the wrong way to approach this project as will be revealed !

The Front Bogie

I paid a visit to M.J. Engineering in Ringwood and looked at all the components they had to offer.  Apart from the centre cylinder casting which I knew would not be suitable, everything else could be used for the unrebuilt loco, so there and then I decided to proceed with the project and came away with laser cut loco frames, more drawings, a box of castings and a large credit card bill !
I was however slightly unnerved when, just as I was going, John Steer (the 'J' in 'MJ') said: "you will need one of these", as he handed me a large bundle of papers titled "errors in Ariel drawings".  It was apparent that 'Ariel' builders had encountered a number of problems with the drawings, and there were probably quite a few scrap bins in workshops around the world full of parts that would not fit (and probably quite a few model engineers cursing their luck !).  At the time, I felt reassured that all these 'bugs' had been ironed out.  Little did I know that they hadn't !

I started with the bogie.  This is one of the few conventional parts of this loco, and is based on the Lord Nelson design.  The parts I had obtained from the aborted project mentioned above were mainly for this component.  I soon realised however that few were going to be useable.  The horns had been fixed to the bogie frames with brass rivets which I felt would not last long.  In addition, the axles proved to be unusable as they had been turned to the wrong dimensions, so I ended up only being able to use the wheels and frames.  Fortunately, the bogie drawings had few errors in them and I was able to complete this component without further difficulty.


At this time, I thought I would order the boiler as there could be a long lead time.  Colin Bushell of CRB Engineering had done a superb job on the boiler for my 'Lord Nelson' so I went back to him.

A few design changes were necessary as the safety valves on the original locos were in front of the dome, but apart from this the 'Ariel' design could be used.  However, it was not long before the gremlins arrived !  Keith Wilson's design showed the gauge glass bushes too low on the backhead.

Colin fortunately realised this before final assembly, and we agreed that he would silver solder a plate over the holes he had cut, and then redrill in the correct place.  The plate would be covered by the backhead cladding so no harm was done.  The boiler was tested to 200 psi and is a testimony to Colin's craftsmanship.

At about this time, I was introduced to John Lloyd from Southampton whose superb model of 21C2 'Union Castle' is featured elsewhere on this site.  John welcomed me to his home on a couple of occasions and allowed me to take numerous photos of his loco which was then nearing completion.  He also lent me various drawings, photos etc, and I am extremely grateful to him for all the help he has given to me.

You can see from the photos of John's loco that it bears the various gunmetal plates showing the loco number etc.  These were only fitted to the first two locos by which time it was realised that they were significantly overweight.  As John had 21C2, I decided to build 21C1 'Channel Packet'.  I contacted Diane Carney who had made the plates for John's loco, and she produced a set for my loco which are of outstanding quality.  She was very pleased that I had contacted her as at that time, she had had to do all the artwork for John's loco as a 'one-off' and was looking for a second customer to spread the costs !

Back to my own loco, I then decided to tackle the main loco frames.  There are several hundred holes to drill and I knew that many on the 'Ariel' drawing were superfluous due to the different design, so I set out to ascertain which holes I could safely drill at this stage.  It then dawned on me that I had a big problem !  I had not appreciated that when rebuilding the original locos, various modifications had been made to the frames.

The first was a cut out for the reversing arm which I was aware of.  The second was that the slope of the frames at the front above the buffer beam had been significantly eased.  The laser cut frames I had bought were of course for the rebuilt loco, so I was faced with the problem of 'putting back' some metal that had been removed on the full size locos.  Fortunately, I had invested in a small oxy-acetylene kit and was able to silver solder some steel fillets back on and recreate the original shape.  I then set the frames on my mill bed and drilled all the holes – there are about two hundred of them and so using the indices on the mill was the only way to do this accurately.

The frame stretchers on the original loco were all very complex steel castings.  The only way to reproduce these would be by fabrication.  I also had to take into account that some of the stretchers in the 'Ariel' design would not be suitable for the unrebuilt loco because of the need to have an oil bath.  There followed several weeks of hard thinking, lots of drawings on the back of various envelopes etc. and eventually some designs emerged.  Construction was no easier – lots of pieces of 1/8" steel all held in jigs whilst being silver-soldered.  I soon found that slight distortion occurred due to the heat so made the stretchers slightly oversize, and then milled them to the finished dimensions – my scrap bin was starting to fill up after a few aborted attempts !


The next job was to tackle the cylinders.  The outside cylinders on the original locos were retained by B.R. at the rebuild, so I used the castings offered by MJ Engineering.  On starting to bore these, I realised that at each end, the casting had become chilled and I was destroying the cutters on my boring bar at regular intervals.  I managed to get round this however by judicious grinding of the bore ends before each pass of the cutter ('Dremel' drills are a great invention !).
I was within 20 'thou' of the final bore before I got through all the chilling, so it was a close call.

The centre cylinder was changed at the rebuild, so I had to think afresh here.  John Lloyd very kindly lent me a pattern but he warned me that the valve chest diameter was too small.  I therefore created a new pattern with thicker valve chests at each end, and also took the opportunity to add suitable lugs for the drain cocks to screw into.

I then took the pattern to MJ Engineering who kindly agreed to get a casting done for me, and was delighted when they said that the pattern was fit for use.  I was even more delighted when on machining it, there was no evidence at all of chilling and no blow holes !

Machining the coupled axles and axleboxes was all fairly straight forward.  The crank axle was built up, pressed together with Loctite and pinned, and runs dead true.  The coupled wheels were relatively easy apart from two areas: the crankpin bosses and the recesses in the 'BFB' wheels.  The crankpin bosses had to be machined on the axis of the crankpin which required off-setting the wheels on the lathe faceplate.  I suddenly realised that my Boxford lathe did not have a big enough gap in the bed !  Fortunately I only needed about an extra 1/16" so by scraping some paint off at the critical point and a small amount of filing, I created enough gap to be able to run the lathe and machine the bosses.  I don't think I have done any damage to the lathe !

The wheels have distinctive recesses cast in them but, on close examination of the originals, there is an undercut at the outer ends that would be difficult to replicate in a casting.  I managed to obtain a 60 degree slot cutter that was short enough to be used to machine the undercuts and performed this operation on a rotary table on the mill bed.

The coupled axle horns are quite complex as the tops are radiused – another jig had to be made and the machining done using the rotary table.

Leaf springs are used for the coupled axles.  I followed the 'Ariel' design here and made up some jigs so that I could crop all the leaves to the correct length.  Assembly was fairly straight forward and I am very pleased with the finished article.  The leaves are cut from phosphor bronze and I used chemical blacking to get them to the right colour.

Trailing Truck.

The final job to complete the rolling chassis was the trailing truck.  On the first two batches of original locos, the basic structure was a massive, complicated steel casting.  On the last batch, a fabrication was used to save weight.  The 'Ariel' design is a hybrid of the two and unsuitable for my purposes.

I therefore set out to construct this from scratch using steel components silver-soldered together.  However, I did utilise part of the gunmetal casting that MJ Engineering had produced for the outer frames – this needed cutting short at the forward end and bending in.  Complicating this were the strengthening ribs at the top and bottom of the casting, so I cut some bronze fillets to fill the gap created and silver soldered the whole lot together.  This component took about 5 months to complete and has been one of the most difficult jobs, so far !

The construction was not helped by the fact that along the way, I encountered two fundamental problems with the 'Ariel' drawings that had not seemingly found their way into the M.J. snagging list.  One was during construction without any wasted effort (just a sigh of relief !); the other was more serious in that the cross section of the rear cross beam was shown 1/8" shorter than it should have been.  I resolved this by silver soldering some 1/16" strip to both sides, but I could have done without the extra work !


Having completed one horrible job, I thought I would tackle another – the smokebox.  The picture of the full size unit shows that apart from a flat front, just about everything else is either angled or curved.  The design changed frequently during the production cycle of the Merchant Navy locos and some drawings of the later units are available.

I could not however find any drawings of the first series, so constructed it simply by reference to the only photo that I know of this.  I treated it like any steel fabrication – tack each component together before finally joining, so that any incorrect assembly can be sorted out.  To try and minimise potential distortion, I used some 5/32" steel plate for the front and rear and 3/32" plate for the curved sides and other parts.  Another horrible job but it fits well now !

Valve Gear.

I have however done some of the design work on the valve gear.  An American, Charlie Dockstader, has a fascinating website ( on which he has developed simulations for just about every type of valve gear I have ever encountered, including the original Bulleid design.  He even has a chart of the dimensions which you can change and see the effect on valve travel and events.  There is a design issue however, in that the original valve gear is in effect a 'miniature gear' and to achieve proper valve travel, there is an 8/3 (267% !) leverage in the steam chest.  Valve travel is therefore significantly affected by any wear in the valve gear.  I therefore plan to increase the valve rod throw by modifying the design using this simulation.


However before getting my mind fully around the valve gear, I thought I would I would build the tender, as at the time I needed some light relief !  Little did I know that there is no such thing in building this loco.

A good friend of mine very kindly offered to draw the tender frames on a CAD program and get them laser cut.  I jumped at this offer and sent him the 'Ariel' drawings.  Big mistake ! 

The cut frames were perfect, but they are for one of the last series of Merchant Navy locos with a 6,000 gallon tender.  The original 'first-series' loco had a 5,000 gallon tender and the frames needed to be 2" shorter.  On telling my friend this, he very generously redid the drawings and I now had some suitable frames.

On further investigation, I realised that there were quite a few other differences between the first-series tenders and those produced for later series.  Apart from the obvious superstructure differences, there were also differences in the braking system and lighting arrangements.

The braking system (like that for the locomotive) is arranged in a clasp fashion, so there were 12 brake shoes and hangers and 6 cross beams together with compensating gear to make; all 12 shoes act on the wheels at the same time.  As far as I can see, the 'Ariel' drawings are correct, but the pull rods need to be shortened to take account of my shorter chassis.  This was a very repetitive process and tested my patience, but eventually the chassis was completed.

A lot of the components on the full-size loco were forged, so reproducing them in model form took some thought !  The brake hangers were all laser cut from 3mm plate and bushes were silver-soldered on each side for the various pins.  The cross beams were cut from 3mm plate and to replicate the cast spigots that fit into the bottom of the brake hangers, I silver-soldered some pins to the ends.  The other complication on these are the parallel plates that are twisted to hold the ends of the pull rods –  I silver-soldered the plates to the beams, turned a spacer to fit between the two plates, located the ends of the plates in the vice and simply turned the beam !  The spacer held the end of plates parallel so all was fine in the end.

It is worth noting at this stage that the handbrake lever on the first-series tenders had a separate cast iron stand, and differs from the later-series tenders which simply had a bracket fixed to the tender front.  There are photos of the full size tenders which show this.

The tender body proved to be a significant challenge as no drawings are available (there may be some at York but they cannot be easily located).  I therefore decided to get every photo of the prototype that I could, and calculate the dimensions from these.  The chassis was fairly easy and followed the 'Ariel' design although it needed to be shortened.

I had originally intended to make all the body out of brass, but soon realised that the curvature of the sides towards the top (and the vertical curvature right at the front) would make this very difficult, so opted to use copper here.

A cardboard template was cut to make sure that the curves at the top edge of the sides looked right, and after a lot of cutting and annealing, I ended up with the right shape. 

I made a steel profile gauge following the shape of the sides and rolled the copper using this as a guide.  It took some time to roll the sides but with the help of the gauge, I ended up with two correctly profiled sides.  The ends and coal bunker were all fitted together using brass angle and then a lot of soldering then took place !  I bought a massive electric soldering iron with a bit made from ¾" copper which was necessary to keep the solder flowing.

One of the biggest challenges in the tender design was to create a removable section to enable me to drive the loco once it is completed.  I wanted to model all the toolboxes and coal hatch fairly accurately, so ended up with a sliding tender cab roof which once removed enabled one of the toolboxes and the coal hatch to lift out as one unit, leaving a toolbox at either side to leave some strength in the front.

One further difference between the first-series tenders and later versions is the water filler.  The first-series tenders had a filler similar to that fitted to Schools class locos which differs to that in the 'Ariel' design (which were fitted to later tenders).


Once the body was completed, I then proceeded to paint it.  I had located the sides to the ends with countersunk screws so these indents (and a few others) were filled with body filler and then rubbed down.  A coat of etch primer was then applied.

Spraying the tender body in Malachite green was fairly easy due to the large smooth surfaces - quite a few coats were applied and rubbed down before a satisfactory finish was reached.  The yellow stripes were painted on top of the green paint afterwards using specialist two-strip automotive lining tape with the right gap in between.  A small hand brush was used to do this.

I wanted to replicate the lighting system fairly accurately so followed photos of the prototype.  Fortunately, some 'lost wax' castings of the lamps were available.  I used clear glass jewellery beads which gave the right curvature to the lenses.

The tender steps were fairly easy to make using pieces of steel strip.  I made some simple jigs to and silver soldered the pieces together.

One final comment on the tender is that as with the loco, there were numerous variations between individual examples.  It is therefore worth getting as many photos as you can of the actual loco that you want to model.

This brings the story up to date on the construction, though there is still a long way to go.