I am designing an induction heating circuit using a half bridge, series resonant circuit. Wildly simplified schematic follows (will use IGBTs instead of MOSFETs for the final design, but they were not available in the schematic editor):
simulate this circuit – Schematic created using CircuitLab
L1 will be the induction heating coil, using a spiral geometry similar to this:
The basic theory is explained in lots of materials around the web such as this old ST application note.
Given a value for L1, and a desired switching frequency \$f\$ (say 20-30 kHz), one can calculate a value for C using $$ f = \frac{1}{2 \pi \sqrt{L_1 C}} $$
From there we can take \$C_2 = C_3 = C/2\$. The actual switching frequency should be safely above resonance so as to ensure the circuit works in the inductive area.
Of course, all of this assumes that a value of \$L_1\$ is given, and this is the part where I am stuck. I've been searching the internet as well as academic papers, but so far I haven't found a design procedure detailing how to select \$L_1\$ so as to achieve the desired heating power.
In principle I could just build an inductor of the desired physical size (say 20 cm of diameter), measure it with an LCR meter, and then select \$C_2\$ and \$C_3\$ according to the procedure above. However, say I build this circuit and it doesn't achieve the desired heating power; then what should I do next? Increase the physical size of the inductor? Increase/decrease inductance (with a corresponding adjustment in the capacitor to maintain the switching frequency constant)?
In summary: how should I go about actually designing/engineering the induction heating coil, rather than just applying blind trial and error?