Exhaust Elbow Repair

WOW! I thought Bill was good at fixing things and "making things work".
That is just incredible to me......ppat

Hell yes...I would be interested. Lots of pictures and good explainations, please. Thanks.....ppat

My dad owned a good sized tool and die shop, where I started working at 16. It's so refreshing to see a good loe Bridbeport getting the job done. I love to have one of my own, but too much tronical junk taking up the place.

Great job Bill, it's great to see a craftsman getting her done.

Well, you asked for it. This is how "Bill" figured out how to fix the taper on the botched elbow:

When we started our journey you'll recall that the gasket on one elbow had failed because somebody agressively cleaning old gasket material off had dished the face of the elbow so there was insufficient clamping force in the middle of the flange.

Our quick fix was using a belt sander to even out the face of the elbow. BTW, this was the kind of belt sander used to surface cylinder heads, not the woodworking kind used to turn a minor repair into a total refinishing job. Although in the end it worked out the same.

The final fix was more complex because we no longer had an original machined surface to reference to. Also, since the elbow is a casting, and a compound curve, we can't reference to a non-machined surface either.

Here's where the explanation gets tricky; imagine a box with all sides exactly the same size, a perfect cube. For this cube to be perfect all sides must be exactly 90 degrees from any adjoining side. They also have to be exactly parallel to their opposite sides. Now lets remove all the sides except the bottom and the left side, still maintaining their perfect relationship. We can still observe if that relationship stays correct because they have a common edge so if one side tilts or twists the edge will not align or the 90 deg angle will change.

In the words of Don Henley, "Are you with me so far?"

Now lets move the left side of the cube to the left about six inches, up four or five inches, and forward maybe four inches. It still has to have that perfect relationship, but now we have no way of comparing it to the other surface. This is where we improvised.

Enter, the other elbow. A mirror image of the elbow in question, and one with known-good surfaces. So if we clamp them face to face we can now establish one part of the relationship since there's only one "bad" surface on the botched elbow. But there's still the question of how you locate the surface on the other axis i.e. making sure they're not twisted across the mating faces. So we went with the only other machined surface available. The faces of the ports where the cooling jumper elbows go. Using various machinist's parallel blocks which are dimensionally correct steel bars and a boatload of clamps and vise grips we attached the whole mess to a granite surface plate. Don't laugh:



It took a long time to get that whole mess together so that everything was flat and true and solidly clamped. In the picture the nearer elbow is the good one, and the big C-clamps are holding the parallels on the bosses for the elbows.

Here's what it looked like from the "front". Where the flanges are clamped together is the left side of our imaginary cube, and the cube's bottom is the flange facing down on the right:



Now that we had our relationships established and a reference surface under the questionable flange it was "easy" to go around with feeler gauges and measure for high spots. What we discovered was that the weight of the elbow had caused it to lean over on the belt sander which resulted in removing .020" more material from one side than the other. There was also a small taper in the other plane, but only a few thousandths. I know it doesn't look like much in the picture (if you can see anything, .020" gap is on the left side), nor sound like much, but when you're dealing with an already less than perfect component (remember there's at least two re-designs) and a system that heats, expands, cools and shrinks pretty dramatically and frequently there isn't a lot of room for "close enough".



So now there's real numbers and locations to work with so "all you gotta do is" fixture it in the Bridgeport and shim it so the high spot is actually high. Then machine it until it's all flat which brings back that relationship between the surfaces. You'll note in addition to writing the dimension on the part I drew myself a line to reference which way the taper actually went. You'd be surprised how easy it is to get confused since the part faces the other way on the machine. This picture shows about one-third of the way into the surfacing passes, and you can see that the material is only coming off one side. With each pass gradually the shiny area spreads to cover the entire face:



After a bunch of passes we finally end up with an elbow that is ready to run. While CAT has upgraded and up-rated these elbows over time my engines remain the same 260 HP, not the 435 HP Scatterpillars that most of these upgrades were probably required for. If the original ones lasted since 1982 I'm sure these will hold up for a few more years and I have over $800.00 more in-pocket than I would have if I'd gone down to the dealer and bent...uh... I mean spent, for new ones.

Last pic, the port engine with elbows installed. These particular ones are the elbows from the first post in this thread, but the others look about the same. Yes, I know the ER looks like the boat sank. I'm only on step two of Float, Go, Look Good.



I hope that's all clear for everybody, it was the best explanation I could think of.