Model predictive hydrodynamic regulation of microflows

We study the problem of hydrodynamic regulation of a flow stream in a microfluidic geometry using model predictive control (MPC). Via bit-accurate emulations we mimic, on a 32-bit microcontroller, the optimization operations associated with MPC. Additionally, we use finite element method simulations to solve the coupled partial differential equation problem that describes the flow and diffusive transport in the examined microfluidic geometry. This combination of simulations and hardware emulations allows us to examine the feasibility of a real-time embedded MPC chip in closed loop with a microsystem, and provide estimates of the performance.