The key is to stop welding it. With the typical weld job, done by guys who have no real metallurgic understanding, you will crack again and again and again, ad nauseum. The metal gets annealed after welding. The only time the metal is structurally sound, is when it leaves the factory. Even most of the aftermarket people dont know how to weld a header. Case in point. I bought some JBA headers for my cobra 3 or 4 years ago. The port entrances were really ugly, so I got my die grinder and put a stone on it to start shaping it. It seemed to just fall into the metal like butter. I wanted to confirm that the metal was soft, so i stuck a burr in the die grinder (the same for porting), and I was able to shape all the runners with total ease with this bit. I shouldnt have been able to, had they known what they were doing when they made these headers. That metal and those welds are supposed to be glass hard, and you should only be able to shape them with a stone. These were the creme de la creme of cobra headers, but I dont know why, given the fact there were so many reports of these headers cracking. After shaping those ports, I understood why it was happening to so many. The steel had been annealed from improper welding. Ive had quite a bit of metallurgic experience through my job, so I can say that the reason this is happening, is because while welding, the steel is being taken to a temperature at, or higher than it was when it was originally formed at the mill. Then they are likely quenching the metal (dumping it in water or oil) while its at this temperature, and this permanently locks the molecules into a position that gives the steel these soft characteristics. Say that in order to harden a steel, you take it to 1750 degrees, let it soak, and then immediately quench it. The steel is now glass hard, but very brittle. (just using these terms to make it understandable) The harder it is, the more abrasion resistant it tends to be as well. The next step is tempering. You heat it back up to a temperature, less than the temperature it was when you hardened it. The tempering takes away a little hardness, but alot of the brittleness. Say that in this step, you temper it at 300 degrees, then quench it. Now its lost a little hardness (not much), and some brittleness. Instead of tempering at 300, say you temper at 900. Now it has a nice balance between hardness and brittleness. You're now in the tempering area of most springs. They are hard (takes dikes or a stone to cut), yet you can flex it up to a certain point without it yielding, or permantently deforming, like a coat hanger would. Now say instead of 900, you tempered at 1200. Youve now lost alot of the hardness and abrasion resistance, but it is very flexible, yet very prone to yielding. The higher the tempering temperature, the closer you are bringing the steel back to its pre-hardening state. Brittleness doesnt really exist. Its just a way of describing the range of flexibility without permanent deformation, yielding. You take a 3 foot long, 1/4" diameter steel rod, with a rockwell hardness of 68, which is very hard, yet very "brittle", and hang a 1 pound weight off the end of it. You measure how far the rod would bend from the weight. Now get the same bar, with a rockwell of 40 and do the same thing. It will bend the exact same distance as the hard steel would have. The temper would decide how much the metal can flex before it gives, either by shattering or cracking, or by permanently deforming, which the softer metal would do. The hard metal has very little range in which it can bend and snap back to its original state. Try to bend it say, 4 inches, it will break. The soft, annealed metal...try to bend it 4 inches, it will permanently deform. Get a steel that has a medium temper. Bend it 4 inches, and it snaps back to position. The temper will determine how far the metal can be stressed before it gives by either cracking or bending. After welding the header improperly, there are spots all over the header that have different hardnesses/tempers. You torque that header down, and the hard spots will crack, and the soft spots will stretch and deform. It will usually crack where the soft, annealed area meets the glass hard area. The quality control procedures and process control are very strict with OEM products, usually. They would never let such a header leave the factory. Now say that after 18 years worth of heat cycles on a factory kick ass header, the temper is going to drift into an area where it finally cracks. Its unavoidable. Especially if people like to open their hood all the time when the headers are still smoking hot. Every time you do that, you are altering the temper a little bit. Welding it will only make it worse. You need a new header. Whew that was long winded! Sorry!
I have welded several of them on my car and friends cars with my mig welder and all have been ok. I just clean them up and weld them no preheat or bolting to the head. I also never put gaskets between the manifold and the head, I check them after I weld, and if they are warped I surface them just enough to make them true. Hope this is of some help. If we were closer I would try to give a hand. I don't know if it would be worth trying to repair one that has been welded on already. Maybe someone else can give some better insight.
