Nonlinear adaptive control of a transcritical Organic Rankine Cycle

A new robust weight update for cerebellar model articulation controller adaptive control with application to transcritical organic rankine cycles

“…The modelling of the rest of the components (including condenser, pump and expander) follows standard methods, presented in full in our previous work (Samiuddin et al, 2017). The resulting overall model contains 13 states…”

“…The current work looks at developing a control system for a TORC that can compensate for unknown or variable nonlinearities in the system while remaining robust to disturbances. The method follows from that initially presented in our conference paper (Samiuddin et al, 2017). We use adaptive CMACs, with novel weight updates, in a decentralized control scheme.…”

“…The advantage of the proposed method for ORCs compared with those outlined in the previous paragraph include compensation for nonlinearities, adaptation to changing heat-source temperature, and application to TORC: a combination that should ultimately result in more energy production. Compared with Samiuddin et al (2017), we have more fully described the proposed control, provided theorems/proofs for the control laws, and have added a step-disturbance rejection test. In simulation, the proposed control outperforms PI control in setpoint tracking and disturbance rejection in terms of both control performance and overall energy production.…”

Nonlinear State Estimation and Control of an Organic Rankine Cycle

“…In this work a Rankine Cycle for capturing energy from low-quality heat sources, such as geothermal or industrial waste heat, is modelled and used in a dynamic simulation. This system, called a Transcritical Organic Rankine Cycle (TORC) originally modelled by Samiuddin et al [1] shown in Fig. 1, utilizes three design components to improve the thermal efficiency of the cycle.…”

“…Since most common control systems in industry such as Proportional-Integral (PI) control are derived under the assumption that the system is Linear Time Invariant (LTI) it is likely they will not be able to reliably track process setpoints which are imposed on the system for performance and safety reasons. There have been proposed controls for similar Rankine Cycle systems in literature [1][2][3][4] [5] which have included both Multiple-Input-Multiple-Output (MIMO) and Single-Input-Single-Output (SISO).…”