We resolve dynamic species transport in a turbulent through-flow reactor by means of recurrence CFD (rCFD). In order to incorporate rCFD into a process control loop, the existing real-time simulation methodology (Pirker and Lichtenegger, 2018) has been further developed significantly. First and foremost, we propose an interpolation methodology which enables the representation of intermediate pseudo-periodic flows in-between two recurrence databases. Interpolated rCFD predictions agree very well with corresponding full CFD results with respect to mean concentration, concentration histograms as well as spatial line profiles and temporal point monitors of concentration. We further apply interpolated rCFD to unsteadily varying through-flow conditions, achieving consistently good predictions. Next, we considered reactive and non-isothermal flow. Still, rCFD predictions agree very well with full CFD simulations with respect to the unsteadily evolving fields of species concentration and temperature. At the same time rCFD simulations run four orders of magnitude faster than corresponding full CFD simulations, yielding faster-than-real-time predictions of the evolving concentration fields. This paves the way towards a rCFD-based process control, such that (1) a recurrence database is selected based on process dynamics before (2) point probe of passive scalars are extrapolated by rCFD to high-resolution field data, which are (3) reduced to descriptive scalars, serving for (4) process actuation. We prove the feasibility of rCFD-based process control for (i) outflow concentration control and for (ii) local hot-spot suppression. (C) 2018 Elsevier Ltd. All rights reserved.