Hi Mark -
The "Straight Aways" are as good as it gets, as the longer edges tend to act like "regular" formed edges. The corners, while not accruate, are a pretty good starting place and since the actual configuration of any flat for a draw is a negotiable thing based on thickness, draw height, curvature, inner vs outer shell and tooling factors, this is a reasonable starting point.
Any draw tooling that I have been involved with generally needs a couple iterations of the flat blank before anyone will commit to a "hard tooled" development. Generally, the draw tool will be made and the flat blank will be developed with laser blanks until the desired shape is attained. It is not uncommon for this to take a couple (read as a few, if hacks are involved) iterations for a drawn corner to be developed. Generally, one can radially scribe out lines on a 15 degree (or whatever is needed) radial grid and chart out the points on the drawn piece - after that the next blank is developed with "reverse compensation", charted, formed and compensated again. Eventually one homes in on a smooth development. Essentially, when material is added or taken away, the draw dynamics change and the part acts differently.
The model I posted has a "dreaded" internal draw which is near impossible (as I drew it) to make it one hit. These are a problem as all of the material is in tension, unlike an external draw where the material bunches radially and elongates along the height. These two forces work together to balance each other out, allowing a reasonable chance at success. In any case, the part that one needs to make will invariably involve some development. A large arc (based on thickness) will form almost picture perfect as if it were linear, while a "sharp" radius will have radical distortion. The height of the draw will also influence its final state, a shallow draw will behave more like a "standard" form while a deep draw will not.
I wonder if this is what prohibits SW from making a sheet metal feature that addresses the drawn corner or curved flange. Fear of putting out "bad" geometry, where no good solution is really possible a priori. I think that there must be some element of FEA that could get one pretty close and reduce the number of iterations in development (maybe there is a perfect tool out there). People who develop this stuff usually know better, not to say that some folks out there have a major ammount of experience and can make a very strong estimation for the first hit and be very close.
I know that round blanks for fully drawn shells have a bit of mathematics behind them based on number of hits (reductions), shell geometry and so on, but most of this stuff usually comes down to finer & finer iterations until the part works. Some of this can be translated to the corners here as imperfect 1/4 "shells", but we are no where near "1st pass certainty".
Later,
SMA