Could you just equate the heat transfer rate as q_r = (T_s1 - T_inf)/R_tot , then solve for T_s1? Adding the thermal resistances would be conduction from r1 to r2 + conduction from r2 to r3 + convection from r3 to the bulk. I did it this way and my final equation for T_s1 was dependent on T_inf, the geometry of the pipe and various radii, heat transfer rate, and the coefficients of heat transfer. Did you do it that way to get rid of the heat transfer rate term from the final equation, if so why?
r2/r1 is correct since you're focusing on the temp difference of Ts1 and Ts2, not the outer radius of the pipe. If you're talking about Rcond, ins. then yes, I think he made a mistake