Process Response Parsing to Improve the Operation of Multistage Irrigation Canals

Canal controllability is a pressing and significant topic that addresses unreliability/uncertainty in water delivery, the efficiency thereof, and the modernization of irrigation systems. We here present a quantitative model of multistage canal controllability (MCCQM) that estimates controllable performance with consideration of the coupling effect of canal reach (CECR). We design an optimization method of the quadratic performance estimation (QPEOM) that identifies the coupling effect and, thus, whole-system controllability (CCA). A linear quadratic control algorithm (LQA) was applied to the Nongchang canal (NCC) data to verify the reliability of our MCCQM. The effects of the flow-water level weight ratio (η) on the convergence time (Tss) during canal control were explored. The NCC response rates (RRs) were on the order of reaches 3>4>1>2. A faster RR indicates that a certain canal reach is more rapidly transformed from the dynamic phase to the steady-state phase and is thus less able to cope with external perturbations in turnout flow variations. The time constant (Tc) ranged from 5.93–10.17 h. The spatial characteristics of the NCC reach controllabilities (the Scon values) were on the order of reaches 2>1>4>3. The slower the RR and the larger the Tc are, the better is the Scon, possibly attributable to the buffering effect of the online canal reach reservoir. The CECR featured conduction of shallow water waves; thus, the least controllable reach dominated whole-system controllability. In that reach, the relatively poorly controlled turnout compromised the efficiency and adequacy of neighboring turnouts in the same reach. The NCC was relatively controllable (total CCA 0.35; weight contributions by reaches 1 to 4 of 0.135, 0.166, 0.422, and 0.277, respectively), implying that the LQA was appropriate. The algorithm performed well during 5 days of field experiments; the control performances of reaches 1 to 4 were in line with the Scon results. The benefits of the control algorithm have marginal effects. A reduction in the water level deviation was associated with an increased steady-state error (σ). The convergence time (Tss) fell as the flow-water level weight ratio (η) decreased. However, this increases the control σ value. The η yielded by our proposed QPEOM significantly reduced Tss with an acceptable water level deviation and met the σ constraints (η=1.28, Tss=15.9 h, maximum water level deviation [MAE]=10.5 cm, σ=0.05 m for NCC reach 3, and a Tss reduction of 44.21%). We thus present a model that reliably assesses multistage canal controllability with consideration of canal coupling. We show a practical method for deriving an acceptable η that considers the Tss and σ and can effectively balance these parameters while meeting the control objectives. Our work will aid efficient water resource utilization and irrigation system operation and thus enhance clean production and sustainable agricultural development in irrigation areas.