Dynamic Modelling and Control Design of Pre-combustion Power Cycles

Abstract

Pre-combustion capture is an important CCS technology. Pre-combustion capture of CO2 incurs less of an energy penalty than current post combustion technologies. For this type of reforming to be competitive for power generation, excellent operability and robustness is required. Important for this is a thorough understanding of the system dynamics and a robust control structure design. Therefore the dynamic modelling and control design of two pre-combustion power cycles, i.e., a hydrogen membrane reformer (HMR) based power cycle and an autothermal reforming (ATR) based power cycle, are studied.

The first contribution of this thesis is to develop new mathematical models of the two novel pre-combustion power cycles based on first principles. The precombustion gas power cycle plants consist of reformers and separation units, compressors, gas and steam turbines and a heat recovery system. Analysis of dynamic models at an early stage of design can give valuable information to control structure design as well as to further process design. Dynamic behaviours of the critical reactors as well as the whole plants are investigated based on these models. The simulations are focused on how different process inputs affect the important variables in the system, e.g., SOT, S/C ratio, TIT, GT power, ST power. The simulations show that both the steady state and dynamic behaviour of the plant depend strongly on the flow rates of feed streams. Due to the complexity of the system and the nonlinearities of the dynamic responses, a systematic approach to control structure design is advocated.

The second contribution is to design the control structures of the two power cycles by a systematic approach. To determine the control structure, an economic objective is chosen, the degrees of freedoms and constraints are found, and the possible disturbances are assumed. The controlled variables are selected by using self-optimizing control. The results show that the control structure depends on the CO2 price. Finally, the control structures with well-tuned PI controllers and feedforward controllers are simulated and evaluated. The results show that he designed control structure can result in a stable system and that PI controllers can make the controlled variables converge to desired values. However, some constraints may be violated during the transient period.

The third contribution is to implement MPC controllers with state estimation for the HMR power cycle. The dynamic simulation reveals that constraint violations may be encountered during operation of the HMR power cycle. Therefore, an MPC controller may be used to handle the constraints systematically. Some selected control loops in the control structure of HMR power cycle is replaced with MPC controllers, and the dynamic simulation results are shown. The results show that because of the nonlinearities, the MPC controllers give a better dynamic behaviour than PI with feedforward controllers for given disturbances. PI with feedforward controllers are easy to implement and do not require much information about the system. However, they may give larger overshoot or constraint violation. MPC controllers can overcome these drawbacks and provide a smoother dynamic performance. Hence, for the HMR power cycle studied here, MPC controllers are recommended above PI controllers.

Finally, the last contribution is the study on benchmarking the two power cycles. The dynamic responses and control structures are compared. The dynamic responses of the two systems have lots in common. The main differences are due to different syngas generators, different operating conditions, and process structure. The designed control structures for both systems give rapid response to load changes and exhibit good load-following capabilities. However, the control structure of the ATR power cycle has a lower complexity as compared to the HMR power cycle.