I love the settling chamber idea. I use a 2 level sump system with a settling tank. Some constructive critisim - Instead of 8 settling chambers it would probably be easier to build and maintian if you dedicate the top to settling/mech and the bottom to bio duties. The top will handle the big bits and a filter sock will polish out the little bits. Outside of the filter socks you wouldn't do any mech filtration in the lower sump.
The water level is a bit high on the bottom sump (the top sump will act like a display tank and always stay at the same level). If the bottom is full of water you wont get the benefits of having the media exposed to air. You can have fully submerged media but at that point you might as well go with a canister. Additionally, with the bottom sump only 1/3 full you'll have room for overflow on power outages. The problem with the current design, the water running off at the fitting by the green arrow, is when the power comes on you could run your pump dry. Both running the pump dry and having room in the sump for overflow will depend on how much overflows from the tank at power outage. If you keep that emergency drain you could use it for a drip system, just move it to where you want the water level in the sump. Not only will you be water change free, but in the event of power outage and water going down the drain you'll be adding water to replace it once the powers back.
I think baffles in W/D's were developed to remove micro bubbles. In that case you wont need them in the upper sump as bubbles will be added again when draining to the lower sump. In the lower sump you'd only need them in the first part of the sump. Micro bubbles are not as big of a problem in fresh water so a lot of DIY designs don't have them.
The bad news on the settling idea - typically that depends on dwell time and gravity but different baffle stratigies may improve it. With a 3500gph the water is going to flow through those sumps like a mountian stream in a rainstorm carrying all the debris with it. I'm sure some will get caught but it wont be very effective. The design would probably work well on a smaller tank with much less gph. I'd consider building one big unit instead of two smaller ones, 4'x1.5'x4'. But then again if you have that much room to work with I'd think about building a parabolic screen.
