Abstract
The Canadian SuperCritical Water-cooled Reactor (SCWR) is a once-through pressure tube–type SCWR under development in Canada. It is a multivariable system with strong cross coupling and a high degree of nonlinearity. The outputs are sensitive to disturbances, and the variations in the thermal parameters should be limited to avoid thermal stress to its components. Therefore, designing an adequate control system is challenging. In this paper, robust multivariable feedback control and feedforward control are proposed to design the control system of the Canadian SCWR. Three uncertainty sources are considered: unmodeled uncertainty, linearization uncertainty, and model reduction uncertainty. These uncertainties are evaluated taking into account all aspects affecting the linear dynamic model used in the robust controller synthesis, and the uncertainty bounds are determined to cover the uncertainties. The robust feedback controller is synthesized using the μ-synthesis approach. The feedforward control is added to the robust feedback control to further improve the control performance. It is obtained through disturbance compensation features for a feedforward controller. The control performance of the hybrid control system is evaluated based on the nonlinear simulation by introducing different setpoint changes. The designed control system can stabilize the Canadian SCWR, and the control performance is satisfactory.
Nomenclature
CFB = | = | feedback controller |
CFF = | = | feedforward controller |
d = | = | exogenous inputs |
e = | = | performance output vector |
GTF = | = | transfer function from the steam temperature to the feedwater flow rate |
GTP = | = | transfer function from the steam temperature to the reactor power |
K = | = | robust feedback controller |
k = | = | control valve opening |
Mo = | = | objective function |
m = | = | feedwater flow rate (kg/s) |
P = | = | augmented plant model |
PIM = | = | input-side multiplicative plant |
Po = | = | nominal plant model |
Pes = | = | steam pressure (MPa) |
Pow = | = | normalized reactor power |
s = | = | complex variable in Laplace Transform |
T = | = | steam temperature (°C) |
U = | = | input vector |
u = | = | control inputs |
Wc = | = | control action weighting function |
We = | = | performance weighting function |
Wn = | = | sensor noise weighting function |
Wu = | = | input uncertainty weighting function |
w = | = | input vector of the augmented plant |
wd = | = | reference input signal |
wn = | = | sensor noise |
wu = | = | uncertainty input |
Y = | = | output vector |
y = | = | measured outputs |
z = | = | output vector of the augmented plant |
zc = | = | control action |
ze = | = | tracking error |
zu = | = | uncertainty output |
= | Greek | |
= | uncertainty | |
= | linearization uncertainty | |
= | overall uncertainty | |
= | model reduction uncertainty | |
= | overall uncertainty bound | |
= | unmodeled uncertainty | |
δu = | = | normalized uncertainty bound |
= | structured singular value | |
ρ = | = | control rod reactivity (k) |
= | largest singular value |
Acknowledgments
This project is supported by the National Natural Science Foundation of China (grant 11405126), the China Postdoctoral Science Foundation (grant 2014M552455), and the Fundamental Research Funds for the Central Universities (xjj2014040).