I, too have done quite a bit of cast iron welding over the years.
Exhaust manifolds and grate bars IME are the worst to weld. Manifolds tend to be thin and large, and both often act barely metallic. I guess that oxygen penetrates along the graphite flakes of hot cast iron, and a casting with enough "thermal history" is as much iron oxide as iron. The only way I have managed to weld some of these has been oxyacetylene, with flux, where there is no limit to how much you can "puddle".
Brazing, or "braze-welding" as it is sometimes called to distinguish it from capillary brazing of well-fitted parts also works well, and if done well will make a joint stronger than the parent iron. In my experience, a fresh break in cast iron cannot be brazed, and I suppose this is because CI usually breaks through the graphite flakes, thus the fracture surface is largely non-metallic. I vee the break out, as if for fusion welding, before brazing. If the iron properly "tins" or is wetted by the bronze, I see no need for mechanical interlock. The actual interface between the braze and the iron is a diffusion zone, and as strong as the metal. Nor do I use embedded pins or studs within a CI braze or weld job. IMO, all these do is reduce the loadbearing cross-section.
Usually these days I prefer to use E-NiFE55 SMAW (stick electrode) rod, DCRP. On good quality iron it is the quickest method and gives very good results. I weld 1/2" - 2" or so, depending on the weight , complexity, and temperature of the job, and peen vigorously while the bead is still hot, and skip around, as Joe M. describes.. Cheaper than 99% Ni, and I do not see how composition of the welding rod can affect hardness/machinability of the HAZ of the parent iron. Only slow cooling can do that.
I always use pre-heat, maintained through the process, and very slow cool.. Several reasons. First it removes moisture and oil. Second it minimizes thermal gradients, thus thermal stress, and cracking. Third, it anneals everything (if hot enough), which provides all the ductility the materials are capable of. Fourth, it slows the cooling, minimizing "quench" and formation of hard martensite and cementite, which reduces brittleness, reduces cracking tendency, and promotes machinability.
A manifold or cylinder-head I will pre-heat and weld in a furnace---which may be merely a charcoal or wood fire enclosed with random sheet metal. Yes, one gets pretty well cooked, reaching in through the smallest possible holes in the furnace, to weld and turn the casting. Often I clamp the prepared casting to appropriate steel "strongbacks", whether angle irons or plates, to maintain alignment as I weld and turn the workpiece over to weld from all sides. Everything heats and cools together.
A small, simple part not likely to suffer from thermal stress I will preheat only locally with a torch, to prevent quenching.
I welded water-jacket cracks on a Cummins 855 engine, in the truck, a few years ago, so of course I could not preheat into the red, but I used a big soft propane flame to get the whole side of the engine up to 300-400 deg F. Even such preheat, well below transformation temperature, acts as an austemper for the HAZ and provides significantly increased ductility.
A customer misused the 3-point hitch on a small John Deere and pulled the lugs right out of the rear-axle housing. Again, only mild local pre-heat was practical, some of the nicest cast iron I have ever welded.
Repaired a backhoe outrigger once, presumably Austempered Ductile Iron, that twisted through 45 degrees before it cracked most of the way through its U-section. After veeing-out the cracks, I clamped one end to my large I-beam workbench, supported the other end on a jack-stand and and clamped a 6-foot lever with anvil hanging on the end. to the other. Built sheet-metal and glass-blanket furnace around it. Lots of propane. As it heated, over several hours the twist sagged out of it. I removed the anvil when it was straight, and welded it up. Who says "creep" is always a bad thing?
I once made the mistake of offerring to repair a cracked cast-iron sectional domestic heating boiler. IIRC four of the sections were cracked, not one as I had thought. I chased cracks for hours, and even after getting them all, I had to fill it with salt water overnight to rust the pinholes shut...but it went back into service successfully.
Joe M.'s suggestion of steel reinforcement is a good one. Oliver crawler final drive (bull gear) housing split when a roller went AWOL from roller bearing and passed through gear mesh. After welding it back together, I made a hoop of 1/4" x 1 1/2 flat iron. a little too small to go around the rim of the housing, heated it up to a forging heat and drove it over the housing. Housing will never again see tensile stress.
Yes, cast iron can be welded as good as new...But if it broke in the first place, sometimes that means it was not strong enough when it was new.