So I've been screwing around with CEARUN (https://cearun.grc.nasa.gov/), which is the chemical equilibrium with applications program created by NASA. I've been using it to calculate theoretical ISP's for rocket engines using RP-1 and LOX for a variety of LOX to RP-1 mixture ratios. I've been using the 165:1 nozzle expansion ratio that the Merlin 1D vacuum uses. If we use LOX to RP-1 ratio of 2.327 (about what the Falcon 9 uses based on tank mass numbers then we get a theoretical vacuum ISP of 370.5 seconds. If we use a LOX to RP-1 ratio of 2.9 we get a theoretical vaccum ISP of 381.5 seconds. Note these are all at a chamber pressure of 9.3 MPA, which is what the merlin's use.
In reality I know the Merlin 1D vacuum has an ISP of 348 seconds, which is 93.93% the theoretical ISP. So it follows that by using this mixture ratio we'd get around 358.3 seconds of ISP after efficiency losses. So my questions pretty simple; does anyone know why spacex doesn't use a higher LOX to RP-1 mixture ratio? Doing so would seem to see a 3% increase in ISP. This is nothing to sneeze at, as that would correspond to a ~ 6% increase in orbital energy that the upper stage could achieve.