Inferential measurement and control of ballast water treatment system

• Main Author: Pazouki, Kayvan
• Published: University of Newcastle Upon Tyne 2012
• Subjects: 578.62
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.566996

As a result of interaction with the surrounding environment, shipping has become one of the vectors of bio-invasion across the globe. Ballast water is one of the means of bio-invasion from shipping through which microorganisms break through natural barriers and establish in a new location. Shipboard treatment systems are predominately considered as mitigating measures for bio-invasion via a ballast water system. Currently shipboard performance monitoring of ballast water treatment systems, and thus assessment of discharge quality of ballast water as required by the Convention, depends on off-line laboratory assays with long delay analysis. Lack of online measurement sensors to assess the viability of microorganisms after treatment has made monitoring and thus control of ballast water treatment systems difficult. In this study, a methodology was developed, through a mathematical algorithm, to provide an inferential model-based measurement system in order to monitor and thus control non-observable ballast water systems. In the developed inferential measurement the primary output of the treatment system is inferred by using easy to measure secondary output variables and a model relating these two outputs. Data-driven modeling techniques, including Artificial Neural Networks (ANN), were used to develop an estimator for the small scale UV treatment system based on the data obtained from conducted experiments. The results from ANN showed more accuracy in term of Root Mean Squared Error (RMSE) and Linear Correlation Coefficient (LCC) when compared to the other techniques. The same methodology was implemented to a larger scale treatment system comprising micro-filter and UV reactor. A software-based inferential measurement for online monitoring of the treatment system was then developed. Following monitoring, inferential control of the treatment setup was also accomplished using direct inverse control strategy. A software-based “Decision Making Tool” consisted of two intelligent inverse models, which were used to control treatment flow rate and maintain the effective average UV dose. The results from this study showed that software-based estimation of treatment technologies can provide online measurement and control for ballast water system.