@Mr. Green, that does help, and I do see more of what you're saying.
What people sometimes misunderstand - but you clearly get it - is that only the dollar amount of a Roth conversion is spendable before age 59.5; any earnings on the Roth conversion inside the Roth goes into the earnings bucket, which as you properly note is not available prior to 59.5 without taxes and penalties.
There are several differences between your situation and mine (and maybe ender's):
1. It sounds like your plan calls for converting most of your tIRAs before age 55 or so. I'm already 54.5 and am nowhere near done.
2. You have a lifetime spending plan and appear to want to spend most of your assets, maybe preferentially before age 70 or so. None of those three things are true of me.
3. You appear to weight somewhat heavily access to spendable funds between now and 60. I don't.
4. You retired earlier than I did and have done more optimization work than I did at your age.
5. I'm guessing you don't have much in taxable, or it's mostly going to pay your taxes on your conversions. I have some extra taxable, so don't need to worry much about my Roth ladder being big. Also, I'm closer to 59.5 so I have more flexibility than you do at the moment.
6. I'm single, and I'm sure that my spending, tIRA, Roth IRA, taxable, and SS situation is different from yours (and everyone else on the planet).
If you did deplete your tIRA by conversions and your Roth basis before 59.5, instead of drawing on the Roth earnings it could make sense to not convert the last few years in your plan and just spend from the tIRA instead. You'd still have EWP plus income taxes, but you'd be paying the income taxes anyway; that would leave the Roth earnings to be tax free after 59.5.
I sort of wonder what iORP would say about your optimization. Have you tried it? If so, how does it's plan compare to yours? If different, have you examined those differences? These sorts of optimization problems can be tricky to code right unless you have the theoretical underpinning - I know just enough to know I couldn't do it properly myself, and I think the iORP author does know enough to do it right.
When you say "the model fails" in your last sentence, is that because you run out of tIRA funds to convert, or some other reason?