This next part was crucial to the model build, but probably tedious reading for many. But without it, the model as seen with a full hull on only three supports would not have been possible, so I’ve included this description for anyone who’s interested.
The model needed a very rigid internal structure to take the loads induced by being supported on only three points. This structure had to continue across the pin joints into the bow and stern sections. Because of the tapering nature of the plan view of the deck, a wide internal structure would be short and not extend much into the bow and stern sections. Conversely, a narrow structure would not offer much torsional stiffness and constrain the space for all the internal systems. So a structure width was chosen that gave a reasonable compromise between the two. This structure width then dictated the spacing between the two aft support points, as the loads had to pass down from the structure width directly into the supports.
Yes, I did use a mixture of metric and imperial dimensioning! There was a reason for this. The actual Meccano structure naturally needed to be thought of in terms of the 1/2 inch hole pitching, whereas to get accurate widths across the hull at each frame station it was easier to measure and scale in mm.
An inverted triangular cross section was chosen for natural stiffness, to run most of the length of the hull. This directs loads down to the keel, which is the backbone of the centre section. However, the aft portion of the structure had to transition from a central keel supported at one central location, to a wider structure for the two aft supports either side of the keel.
The internal structure was clearly a key feature that needed to be right. A failure of the structure – or the need to change it later in the build - would be a disaster. The space frame diagram shows the arrangement originally planned for the centre section (a small change was made after prototyping). It was roughly plotted onto the scaled-up paper blueprints to check it all stayed within the hull profile. The space frame was designed as a determinate structure (i.e. no redundant load paths) for rigidity and ease of calculations.
The bow and stern sections are cantilevered off the centre section, and their loads transfer via the pin joints. But the centre section itself is the heaviest part (full of mechanisms etc.), and its own weight is distributed around the structure (i.e. not concentrated directly above the support points) yet still converge at the 3 support points. For simplicity, this load was divided by 10 and each division added to the centre section upper nodes to represent a distributed load. These are shown as the green vertical arrows.
Lastly, the superstructure sits above the centre section, on 4 discrete points (corresponding to 4 upper structural nodes to direct the loads into the structural members without inducing bending). The superstructure weight was equally spread onto these 4 load points, adding to the distributed centre section loads at L2 and L4.
I wrote a spreadsheet to calculate all the loads for the central structure, based on the initial weight estimates. The output gave the loads in each member, across the pin joints and at each support point. It didn’t particularly matter what the loads were, only whether they were tension, compression, large or small, as this dictated the type of member required. A compression member not appropriately sized would buckle. A subtle change to the aft end was made after prototyping which you can spot in the final structure diagram. This was made to reduce the loads in some members and make it easier to build.
However, the fully assembled ship wasn’t necessarily the worst loading case for any particular member. For instance, during model assembly, if the bow section was attached to the centre section first, then the entire weight of both sections would shift more load onto the single forward support which might be more than when fully assembled. So the calculations were run 4 times using different assembly scenarios. This threw up the fact that some members could change loading from tension to compression and vice-versa depending on the order of assembly.
The bow and stern structures were not designed at this time, nor did they need to be determinate – only robustly constructed – as there were no support points to consider.
The following sketches show the original ideas for the bow and stern structures produced sometime later. The arrow heads on the members indicate whether they are in tension or compression (arrowheads on a member pointing towards each other is the convention to show tension).
A bit of prototyping…..
Nine months after the start of the project I was ready to build something. At the time I didn’t have much good looking restored Meccano, nor have many Meccano nuts and bolts, but I wanted to build the central structure to prove the concept was ‘buildable’. So off the internet I bought thousands of M4 pan head bolts of various lengths, and thin M4 nuts. They have advantages and disadvantages over Meccano bolts. The thread pitch is slightly finer, so for the same tightening torque they grip better and are less prone to undoing under vibration. They are also cheap and available in many lengths. The downsides are the bolt heads and nuts are slightly larger so less able to fit into those tight locations.
However, some standard tools can be used, as well as Meccano spanners opened out to 7mm. In the pictures you can see an array of useful build tools with many ‘modified’ spanners which I suspect is a familiar sight for builders. It's just as well that spanners are plentiful, and it irks me to see them advertised on eBay by chancers at £1 or more plus £1.75 p&p!
A word of caution about buying M4 stuff though. The first lot of M4 bright zinc washers I bought – 20,000 of them – were of inferior quality, as their manufacturing process caused there to be flashing on the O/D on one side, and hole punch flashing on the other side. So either side would scratch the paintwork when used.
To have some usable stock for building, I made a punch which I used on each washer on the hole-flash side, which with several hammer blows flattened out one side of the washer so it could be the ‘component facing’ side (see before/after pictures). I reclaimed a few thousand washers this way before finding a better quality supplier where the flashing only appeared on one side.
Also, the nuts (mild steel bright zinc) were not all threaded properly. Many would spin onto a bolt with a wobble (thread not cut squarely into it) so had to be discarded with an attrition rate of about 25% (that’s a big loss from an order of 13,000). I did find some better quality – more expensive – stainless steel ones, but these were far less useful as they were virtually non-magnetic so could not be used with one of my most useful tools - my nut fitter (see how to make and use this here).
The first version of the structure was built to the original concept using whatever I had available (hence the mix of colours). The first sub-assemblies were for the forward support point and the keel. Some ideas were sketched out for the arrangement of plates and strips for the forward point where many high loads coincide. I noted that some aspects of the design rely on 'selective fit', as only some angle girders accept a strip inside the angle whilst still allowing a bolt through the aligned holes.
This is the assembly taken to one of the club meetings, where I asked members to guess what I might be building. I didn’t reveal the answer until the next meeting.
Ideally, all the structure members should intersect at the theoretical nodes. But with the limitations of Meccano this is not possible, especially with 3D structures. So the joints were made as best I could with the ‘load lines of action’ approximately intersecting at the nodes.
The prototype prompted me to change the aft end of the structure for ease of assembly and reduction of loads in a few members.
There is another structural compromise at the support points. For example the forward support point, section pin joint and structure node should all be coincident. In reality, the pin joint ‘hole’ is one pitch forward of the node, and the actual support point is one pitch aft.