The PiCeramic data sheet you linked has some data that is useful. e.g. the PL022.30 has a capacitance of 25nF, and a free resonance>600kHz. The drive voltage is restricted to -20V to +100V.
Driving it brute force would require C*ΔV = I*t, or I = C*ΔV/t.
At 80kHz, one half the period = t = 0.5*(1/80000) = 6.25us
so, the required current from the amplifier is I = 25e-9*100/6.25e-6 = 0.4A
It is possible that you could drive it at less than 100V and still knock loose the bubbles??
It is also possible that by the time you mount the transducer, that will lower its resonant frequency to where you can sweep the drive frequency to match its resonance, and then it will take much less power to drive it...
Note that PiCeramic makes power amplifiers **broken link removed**I understand all of its ratings, except where it says: "Amplifier bandwidth, small signal = 3.5kHz". I would expect that it should have a bandwidth well into the MHz in order to drive your stack at near 80kHz???