One thing I noticed today going back through Watts' posts is that his ballista was getting close to 100 m/s of velocity with 422 gram bolts, which are fairly heavy for his machine. Changing that would require rejiggering a bunch of the formulas, so I'm not advocating for a change, just putting data out there for review.
However, I would like to discuss crew sizes, since Demetrius' 78kg stone-throwers apparently had a crew of only nine, although that was in a siege tower that may have had pulleys to provide them with mechanical advantage in drawing back the arms. Per steps 3, 6, and 7, a 78 kilogram stone will have 97,500 joules of energy, require strength 1625 and a crew of 233 to achieve Reload 5. It would need a crew of 47 to get Reload 25.
One solution might be extension, so a tenth of the crew has Reload 50 and a twentieth has Reload 75. I'd be OK with this, since it suggests a ~4 minute reload period for a heavy torsion engine (in fact, even that might be too fast).
The other option would be to drop crew sizes, though I don't know what they "should" be, since there's little information I've found on anything else except the smallest scorpios and the onager.
There should probably be a maximum crew based on some sort of reasonable measure, since 233 people working on a single siege engine, even if it's out in the open, seems improbable. And, on the flip side, doubling crew on small torsion pieces might reduce Reload to 3 or 4, since the light scorpios seems to fit into Crew 1 using the design sequence, but had 2 crew and fired 3 to 4 times per minute, which is faster than Reload 5.
You might want to download and look at the following two sources.
https://archive.org/details/Book_of_the_Crossbow_The_by_Sir_Ralph_Payne-Galloway
https://archive.org/details/cu31924102030891
Payne-Galloway actually built crossbows and projectile engines
I have a dead tree edition of Payne-Gallwey. There are some accuracy issues here and there with him (his onager in particular; De Reffye's version is more likely to be correct). His ballista probably existed at an early development stage, but would have been dangerous to the artillerist - if the arms aren't equally torsioned (due to one skein getting damp or having a few strands snap), the bolt can jump the track. On an Orsova-type ballista, a jumper can keyhole through the large gap in the frame. On a Gallwey-type ballista, the jumper can hit the large inner stanchions and become shrapnel ricocheting back into the artillerists' faces.
I ran some designs using bolts and found that the weight of the bolt is roughly equal to the weight of the stone.
For example, the 36" and 521 grams bolt is 0.914 meters and 0.546 kg.
From Vitruvius, the "washer hole" would be 0.914/9 = 0.102 meters diameter.
The "washer hole" for a 0.546 kg stone would be 0.106 meters diameter.
So the bolt throwers just use very small stone throwers, or the larger stone throwers just throw very large javelins.
Using Vitruvius' equations relating both stones and bolts to the "washer" diameter, we can calculate the length of Vitruvius' Bolt for any given weight as:
Length of Bolt = 1.169 x (Wa)^0.33 = Meters
That's very useful to know, and makes the design sequence easier. I'm torn on how to handle the Dam/Pen relationship for bolts versus stones. They're low enough velocity that it feels like bolts should have better Pen (and maybe Rng?) but less Dam than stones from the same size machine. Or maybe they just scale up in Dam less, so that they're more efficient at the smaller size but become less effective at large sizes. I'm mostly brainstorming here without solutions, but it feels like there needs to be a reason the heavy machines are stone-throwers and not giant bolt-throwers.
