A UPCA-based monitoring and fault detection approach for reverse osmosis desalination plants

A perspective on reverse osmosis water desalination: Quest for sustainability

“…In recent years there have also been mounting efforts to develop effective methods of plant fault detection and isolation so as to avoid plant failures and reduce maintenance costs. Effective self‐adaptive RO feed pretreatment that can handle temporal variability of raw feed water quality and production demand is also likely to lead to significant reduction in the use of feed treatment chemicals while reducing the frequency of membrane cleaning and replacement.…”

Fault Detection and Isolation in a Spiral-Wound Reverse Osmosis (RO) Desalination Plant

“…In recent years, model-based FDI methods have been applied in RO water desalination. ,− For example, actuator FDI integrated with a fault-tolerant-control (FTC) strategy was reported for a single-membrane-unit RO desalination process without pre-treatment or post-treatment. The approach relied on model-based feedback control laws making use of the fundamental RO transport equations and was successfully applied to operation with varying levels of feed salinity fluctuations.…”

“…Once a faulty sensor is detected and isolated, data reconciliation or rectification is necessary in order to estimate the sensor true reading values for the faulty data trace. The above approach was demonstrated for RO desalination plants for detection of sensor along with other faults in the plant , as the basis for model-based fault tolerant RO plant control . Faults (e.g., in system hydraulics, pumps, sensors, and actuators) were induced in a computerized model of the RO plant (treatment capacity of 500 L/h of feed with a conductivity value of 800 μS/cm, producing 250 L/h of permeate with a conductivity of 7 μS/cm).…”

“…Statistical techniques such as principal component analysis (PCA), based on the analysis of process history, are commonly used for fault detection and monitoring. For example, monitoring and fault detection was carried out applying a PCA-based scheme and an unfolded PCA (U-PCA) using an RO desalination plant simulation (∼1.7 m 3 /h capacity with sand and cartridge filters for RO feed pretreatment) that operated with cyclical cleaning phases . The approach was tested by simulating plant operation using an object-oriented and dynamic simulation tool (EcosimPro).…”

“…For example, monitoring and fault detection was carried out applying a PCA-based scheme and an unfolded PCA (U-PCA) using an RO desalination plant simulation (∼1.7 m 3 /h capacity with sand and cartridge filters for RO feed pretreatment) that operated with cyclical cleaning phases. 16 The approach was tested by simulating plant operation using an object-oriented and dynamic simulation tool (EcosimPro 19 ). Three different U-PCA models were developed (for the two filters and for the RO membranes).…”