Sorry for replying a year late. I am faced with similar debate. I know it can be done, wrestling with should we. Tom, great graphic. I have sites that use this and some that don't. Some are high volume production, others are A&D program based where only a few units gets produced.
The argument for two phase and dual revisions schemes is typically focused around slow change processes that cannot support rapid changes in prototype phase. Once you commit to one way or the other, I believe that all parts you develop would follow same path (numerics in development then alpha in production). A competing requirement I have seen is pushing preliminary BOMs to production to support long lead items. This can lead to releasing a known to be incomplete drawing, leading to the schism between Engineering and MFG. Engineers says they were not done yet so don't make it but MFG needs to order material and cannot wait for Engineering drawings to be completed. I do not believe the solution is to throw process out the window.
Another observation is the percentage of items that truly need a two phase lifecycle. I would think that less than 3% of all "designed" items hang around in development, pre-production phase long enough to justify the more complex process. Most will advance straight to production with few changes after initial release. The goal should be to get things right the first time and release as late as possible. So with two-phase, 97% of items would need to be revised again and released at production revisions. So what drives that trigger? How do you know you are ready for production, on a single item or the whole product?
One would think that after the development units are built, there is some gating process that triggers all development items on the BOM to be reviewed, revised and re-release, mostly without any alteration. I see this is that the bill coming due for kicking the can down the road. By allowing rapid changes with little oversight, would we end up with a higher quality design in the end? Something for the academics to figure out. I have seen cases were if left to their own devices, organizations would shy away from doing that large amount of work at the end when in reality, nothing has changed.
I would be interested to see where the break points are. Is two-phase more suited to high volume mfg? Does it lead to bad behavior, more revisions or poorer quality? When you organization will end up selling off the first prototype as a production unit, does this process not make sense? Is there a case to be made that a simpler system is better?
