State estimation of a stratified storage tank

Abstract

Two approaches for the state estimation of the spatial temperature profile in a stratified storage tank, i.e. the state of the art domestic hot water storage system, are presented. There are a distributed parameter observer as a late lumping and an Unscented Kalman filter (UKF) as an early lumping approach which are investigated with respect to applicability and reconstruction convergence. Both, the distributed parameter observer and the UKF are designed based on a hybrid distributed parameter model of the stratified storage tank. The hybrid model is composed of a finite state automaton interacting with an underlying thermal heat conduction–convection system described by a quasi-linear partial differential equation. The performance of the estimation algorithms is illustrated by simulation studies and measurement data, showing excellent convergence results.