Recreating an exhaust manifold for my Honda S800 - CAD attempt

At some point, I may need to resume work on my S800, given that this was being described as a "retirement project" - and I'm not officially retired.

My recollection of the exhaust manifold was that it was a bit on the rusty side, so I've been half heartedly looking into the practicalities of fabricating one. 

Pipe bending malarky:

When I was in Canada, I acquired a pipe bender with a view to being able to make the required bends. However, unless you use a mandrel bender to form the bends, they tend to collapse. The tighter the bend, the more they collapse. 

This is what I won in the auction: No formers, just the basic machine:

I figured out what was needed:

And made up 3 different sizes. They need to be over 90 degrees each if you want to do a 90 degree bend, so you can get 3 per round blank. This was quite a job, given their size and the (necessarily) interrupted and heavy cuts.

Here's one of them in  its completed form:

And in action:

Not bad - but without an internal mandrel, there's some collapsing, which is noticeable as an oval shape to the cross section of the pipe.

Where to start again:

Here's the downstream section of the exhaust - the one that mates with the manifold side. This ("4 into 2") section is much worse than the manifold side in terms of rust. This is a photo taken in Canada, around Dec 2012:

But initially, let's look at the upstream / manifold side of the connection flange that joins them:

Here's the flange that mates with a similar flange on the downstream section, using M8 bolts:

These pipes have an OD of around 32mm, although that varies depending where you measure them. In practice, the flange plate is the best,

Let's model that up in Fusion. Note the shallow counterbore that accommodates some funny asbestos(?) filled o-rings that seal the manifold against the head:

Create a 2D drawing:

It's got rather a lot of dimensions but I put those in so I can check the dimensions against the actual manifold itself, eg the bolt holes, port bores etc. If I make one of these, it will be on the Shizuoka, so a dimensioned drawing isn't really needed anyway.

The assembled engine isn't easy to access for measurement purposes but I have a shitty old head that was cannibalised for spares, so I will use that as a space model.

If I've got it right, a 3D printed flange should fit. It's too big to fit in my printer but the manifold is actually in 2 sections so that it fits over the oil pump drive shaft, so this part of the manifold is easy to print.

Looks OK

I wouldn't describe this as a perfect fit. Some of that is due to the rough tolerances found on engine castings. The ports aren't circular in places and this isn't a deliberate feature from what I can see. 

The solution to that is to match the ports to the manifold flange using a die grinder. But that can wait until later.

And as for the mating flange plate:

Initially, I assumed it was largely symmetrical.

But it's clearly not symmetrical. Bollocks. They are all over the place.

I think the answer to this may be to set it up in the Shizuoka and use the Renishaw probe to determine the hole centres and their positions from a chosen datum. Luckily the overall length of the manifold is only about 39cm, so I should be able to drop the knee far enough to probe the flange with its surface horizontal.

Conclusion:

• Looking at the manifold, it actually looks pretty solid. I'm struggling to convince myself it's worth recreating the entire manifold from scratch. I can't even see any potential to increase the pipe diameters, as the manifold flange barely accommodates the existing (32mm OD) headers. Let's leave that as it is.
• The downstream section clearly isn't long for this world. I'll probably have a go at replicating it. That will require me to flash one of those flange plates up.
• The last painful / most accurate way to replicate the flange plate would be to probe it on the Shizuoka. Trying to do this with digital callipers sounds like a hiding to nothing.

The biggest task is probably going to be locating that downstream section. It's clearly in a box somewhere in the workshop but I haven't stumbled across it recently.....