I went the other direction in terms of powder burn rates and it seems to be working. Shooting 90 VLDs out of a 30" 7-twist barrel, I initially determined velocities for ~ 6% reduced loads (based on Quickload predictions) of IMR4320, Varget, H4895, H322, and Benchmark powders. I did this mainly to generate some fired cases so I could determine my case volume and to establish a baseline for adjusting burn rate factors in Quickload to better estimate what sort of velocity and barrel times these powders would generate in more "optimal" loads.
From the initial testing, IMR4320, Varget, and Benchmark did not look as though they would give the velocities I was looking for without compression and/or pressure issues. I should point out that most of my other loads for .223 and .308 have ended up very close to the optimized barrel times as reported by Chris Long (
Optimal Barrel Time Concept). In the beginning, I was not aware of his theory and was not consciously trying match any specific barrel time, the optimized loads I came up with just happened to be end up there. In any event, it is a parameter that I now try to pay closer attention to during load development.
If you look at his table for optimal barrel timing, you will see that there are "nodes" for different barrel lengths. For a 30" barrel, the optimized barrel time is 1.3684 ms. Depending on the powder, hitting this node will likely put you somewhere in the 2810 fps to 2870 fps range. Reaching the next higher node would require ridiculous velocity and pressure, and is pretty much impossible for a standard .223 case with any powder. In my hands, Varget wasn't optimal. I could reach the 1.3684 ms node using it, but the velocity only ended up at around 2820 fps, which was pretty disappointing. Using Quickload estimates, H4350 may well suffer from the same issue. In all fairness, I have not actually tried it and it is possible that the preset burn rate factor in Quickload for H4350 is way off from the actual value in a .223. So it might work...only way to know for sure is to try it.
For these reasons, H4895 has been the best choice IMO, which is what I suspected well before doing any actual load development. H322 might also work, but I haven't had the time to pursue anything with it as yet. The biggest potential advantage to H322 (if it will work) is that the grains are very fine and it can be weighed to extremely close tolerances. For me, accurate and precise powder measurement is absolutely key in obtaining low ES/SD for a small cartridge like the .223. I'm currently running 24.4 gr of H4895 and getting velocity in the range of 2860 fps with pretty decent ES. More importantly, this load is shooting extremely well so far and the pressure estimates are well under max. During initial development, I ran into some issues of primer pockets loosening up noticeably after only one firing when I went above 25.0 gr of H4895. Not surprising as the velocity was well above 2900 fps. In any event, I think my brass life will be pretty decent where it is now.
All in all, you can try a bunch of different powders and probably find a recipe that will work for several of them. The only problem with that is pushing a 90 gr bullet from a standard .223 case is really stretching boundaries for that cartridge. At that point, every little thing can make a big difference. As an example, I'm getting ~40 fps more velocity by using H4895 as compared to hitting the same sweet spot with Varget. That will be worth something at 1000 yd. Maybe not much, but something. I'm also sorting bullets and pointing them. When you're on the edge of what a cartridge/load can do, every little bit helps. The other consideration (IMO probably one of the most important) is seating depth. Regardless of the powder you choose, you will have to find the best seating depth for your specific rifle. Hopefully, it will be one where you'll have some headroom as the lands erode, so you don't have to re-measure your chamber and adjust your seating depth every couple hundred rounds. In my rifle, I'm seeing approximately .002" erosion per hundred rounds, which is pretty typical. Just something else to think about.