Forged upsets are some of the most common forgings made. With just two dimensions to protect, the upset is also one of the easiest forgings to make.
Upset forging consists of starting with billet, bar, or alternatively RCS aka "Round Cornered Square". The easiest way to estimate material needed for the upset is to add the finishing allowances, which helps to guarantee the part can be machined from the gaurantee. In most situations, these allowances are 0.250" to 0.500" per side. For instance, if the guarantee needed is 8" Dia x 3" Long, then the forging size would be 8.5" Dia x 3.5".
Once the forging dimensions have been developed, a billet needs to be selected.
For optimum properties (to receive greatest benefit of forging process), you want a forging that has at least 25% work in it. Some companies will also indicate a MHWR or "Minimum Hot Working Ratio", which helps a forger to understand the total amount of work the piece has received since it was cast.
Calculate the total width, which is normally the diameter. For our example this dimension is 8.5" Dia, but we also need the work added. Upset forgings are different than the typical round or block, as we start with a smaller area and forge it to a larger area. These are typically pancake shaped, with the starting length being no longer than the starting width. While longer starting lengths are possible, it adds tremendously to the on die time and the risk for folds and laps.
To add the work amount, we need the area of the finished round and add at least 25% to the area. This would be Ao=Af*0.75. Notice how we are subtracting 25% of the final area of the round? So the pre-forging area we want is 56.74*0.75=42.55in^2. To get back to the radius, we must divide by PI, and then take the square root. SQRT(42.55/PI) = 3.68" Radius, which means we are looking for round or billet about 7.25" Dia, as 7.36" Dia probably doesn't exist. We use this measurement as a maximum Diameter to match up billet or bar to use.
If you wish to calculate the amount of work the piece would see, it is fairly easy to calculate also - using the original area and the final area of the piece. The effective reduction for this piece would be final area minus the original area, all divided by the final area. We could use the other method, but the result would be negative. Using our example it would be ( AfPI*r^2 - AoPI*r^2 ) / AfPI*r^2 = ( 56.74 - 42.55) / 56.74 = 25% Reduction.
We then need to make sure that the length of the starting billet will not be too long. A nice general rule is that the volume before forging is the same as volume after forging. Volume times density provides weight, so the weight is the same as well. It's easier to ignore density whenever possible, one less factor that can go wrong. The volume of the final part is Area x Length, Vf=56.74x3.5=198.59 Vo=198.59 So the length is how many times the area will go into the volume. L=198.59/42.55=4.66" Long, which is not longer than it is wide - all good.
Superalloy upsets usually have demanding applications in nuclear, aerospace, defense environments. This means that once the forging is made, we usually need to test it for properties. Non-critical applications may only need a hardness verification that the material exhibits similar properties to known samples. Critical applications require a full mechanical test (destructive) that may also entail a stress rupture test to qualify the lot. If mechanical testing is required, a sample piece will be needed to submit for testing. Usually the specification that requires the testing indicates where the sample should be from, and the size of the samples needed. I mention this, as the material for a sample usually requires additional material, which can add to the overall length of the part being made - or in some cases another piece long enough to obtain the sample from. Typical samples can be 6" long, while in some cases a sub-sample of 3" is all that is required. ASME samples require at least 8" of material. With an upset forging, sometimes making a longer forging to accomodate the testing is not possible, so the test would be taken radially or in the short tranverse, instead of in the typical longitudinal direction.