Summarizing: For a given amount of chassis flex (however much there is), your simulations suggest that the two factory bars decrease strut tower deflection by ~50%. Adding the Grimmspeed bar would decrease tower deflection by ~90%.
Questions:
1) Tell us about your model. FEA can be a mixed bag, and it all depends on how you set things up. Be proud of your hard work, and show it off!
2) How much strut tower deflection have you guys seen during testing?
RCE recorded 4-5 mm of deflection on the 2004 STI in a 1g skidpad session, per this discussion:
LINK
The FEA results are directly proportional to the chassis stiffness you chose, so that's why I'm curious about it.
Alot of methods have been discussed over the years, without enough actual results.
Deflection: RCE's draw a mark method (kinda ghetto). Linear potentiometer (can be noisy).
Stress: Shear pins in a strut brace (only tells max compression force). Strain gages on a strut brace (can be noisy).
While I appreciate the "stiffer is always better" argument, if we can identify a legitimate deficiency and address it properly... your products will fly off the shelf.