Gas-solid fluidized beds exhibit characteristic, reappearing patterns. External parameter variations or coupling between internal processes superpose a transient behavior on top of these recurrences. We demonstrate how recurrence CFD (rCFD) can cope with such scenarios by interpolating between multiple, pre-generated databases corresponding to short (2.5 s) time series in different flow regimes. The methodology is applied to a fluidized bed of approximately 80 000 solid particles subject to internal heat production and fluidization with cool gas for a duration of 800 s. The fluid's thermal expansion due to heat transfer ([290 K; 430 K]) and the varying inlet velocities ([0.4 m/s; 0.9 m/s]) cause the desired transiency. We find excellent agreement for particle mean temperatures over time and qualitative similarity for their standard deviation between detailed, established simulation techniques and rCFD with speed-ups of more than two orders of magnitude. Furthermore, we present implementation advancements concerning parallel data exchange and compare all-to-all with one-to-one communication schemes. Analysis of the code's performance points the way for future optimization measures.