Maybe the researchers meant 230 Kwh above the reaction energy of pure carbon and oxygen. That makes the math check out, since 2,485/(2485+230)=91.5%, which could be rounded up to 92%. It would also square up with the estimated costs of $100/ton, since industrial electricity prices can go as low as 0.03 to 0.04 USD.

I wish I could read the source to verify, but unfortunately it's behind an academic publisher paywall.