Technote: P-39 Airacobra Exhaust Stack

Technote: P-39 Airacobra Exhaust Stack:

This rather small unassuming item is at first glance a straightforward little model that actually turned out to be a huge headache. I spent several days working on this model which will all be explained in this article. Hopefully, the solutions I found can help you.

The first hurdle was the Bell Drawing 12-614-001… several key dimensions were illegible and a complete end section was non-existent. The first task was to develop what I do know to help determine what I needed to know…that in itself took an inordinate amount of time but eventually managed to get that sorted. In addition to the sections shown on the drawing I needed to include a control sketch to control the dimensions of the eventual loft activity…this was essentially an ellipse with fixed height and variable half-width.

The side/outside lines of the Exhaust Stack are fixed profiles so it will be the inner profile that will change to make sure that the intersections of the 2 pipe exhausts were correctly located in the centre of the element.

You will notice the side profile sketch is separate from the main model sketches, this just makes it easier to see what I am doing as for the most part, it was mainly for reference.

At this stage, everything is quite straightforward as all I had to was loft the 2 pipes by selecting each of the profiles as shown and then trim the surfaces to give me a base model. A small tip: for this to work correctly and ensure the alignment with the external lines and allow for expansion internally it was necessary to loft using the Centreline option…the centerline holds the loft shape between sections normal to the centerline.

This is where everything got crazy. The main body part of this stack has no joint seam and is quite bulbous…so what I had to do was adapt this base model to form the bulbous surface complete with an internal curve.

As you can see in the image on the right the real item shows the bulbous main body part and the generous inside curve. By the way, some may have observed the real item looks shorter than the model…this is a puzzle…some of these exhaust items are indeed short and this is further noted on the Bell drawing as a dotted line! I suspect this may be linked to the engine used but as yet I do not know for sure.

I initially created the short version thinking that this was normal until I found out that most are actually longer versions. That was the start of many frustrations to come as that first batch of models was quickly scrapped and started again. That bulbous bit though is the main problem.

In order to achieve this I had to derive a solution that filled the void between the 2 tube lofts and at the same time provide an internal curve consistent with the real product. What I opted to do was simply trim and then remove the inner surfaces and then blend the remaining void with a surface patch.

My initial effort was to remove the centre section according to the natural divisions along the centre of each tube and the sketch plane. I tried variation after variation on this, adjusting tangency strength and G2 for each side, I even adjusted the dimensions on the control systems to make minor corrections. I eventually ended up with something reasonable and we got it 3d printed. The second image above is the latest model incarnation which I will explain below.

Two immediate issues are quite visible on these printed models…the initial curve between the 2 tubes is far too tight and almost looks as though it has folded. The second is the surface continuity…okay admittedly I was unsure about the short and long versions of this stack so the end extensions were a separate model part…though I should note that the surface should have been tangential and it’s clearly not.

Coming back to the cad model above…the second image shows what needed to be done. In this one you can see the cutout in the main body is an ellipse which gave much better results and a larger smoother curve between the tubes.

What was happening with effort 1 was that choosing the centre of the tubes at 2 and 3 (which by way was a logical choice) was actually restricting the area in which the inner curve could form, resulting in small deviations in surface accounting for the folded look. Essentially by selecting the centre lines I was actually creating self-imposed restrictions. It took me a while to figure that one out and many hours of work.

Further the mere fact that this patch includes sharp angular corners also did not help as the patch stretches to accommodate those corners again leading to imperfections. Finally, I decided to forego that first attempt and ended up using an ellipse profile that extended beyond the centrelines of the tubes. As there were no sharp edges continuity of the surface patch outline was good and ended with a smoother patch surface that was G2 compliant. There is a small additional patch at the end of the stack…there was a natural seam there which I needed to get rid of.

The final model is shown above and is a very much improved surface with good continuity and a generous internal curvature between the tubes. One final point is the mounting plate… which is not dimensioned on this Bell drawings because it is a contractor-supplied item. I searched through many drawings and found something similar…so I cannibalised the dimensions from that drawing to create the mounting plate. It turns out that this was a good effort as it actually aligns with the engine block.

Surface modelling is rarely this complex but occasionally you will come across something equally challenging. If something is not turning out the way you had hoped or expected just check to see if your choices are imposing restrictions on how the surface is created. When you are trimming or splitting surfaces try to minimise sharp edges and instead opt for curved circular solutions.

Update: 7th October 2022:

Just received word that the newly revised Exhaust Stack model has now been 3d printed…just another 11 sets to go. Apparently, the time for printing and ultrasonic cleaning will be about 4.6 days. They look good…check out the awesome curvature on these prints.