A key step to develop an AI model for coagulation process is to know HMW framework


The coagulation process has been extensively researched to improve its efficiency and applications. In particular, the Water Treatment Engineering System (WTES) where this process plays a major role in the removal of colloid particles and natural organic matter from water environments. However, recent studies have shown that physical, chemical and numerical simulations are among the main methods for studying coagulation processes. These coagulation processes include: preparing improved coagulants, optimizing velocity gradient (G), and mechanical stirring time (t). Such processes help reduce operational costs. Additionally, the development of a numerical simulation system to determine precise doses of chemical reagents.

More recent research focused on physical and chemical processes than on numerical simulation in recent years. In this regard, the development of the WTES demands simulation technology to be enhanced. So far, it is rather unfortunate that the variations of parameters such as temperature, pH value, organic matter concentration, metal salt concentration, water flow rate, and turbidity have hampered the progress of the coagulation phenomena. These challenges present a great obstacle in the practical applications in real-time treatment applications. To address this, scientists from the Hunan University of Science and Technology: Professor Guoncheng Zhu, Dr. Nana Xiong, Dr. Chuang Wang, Dr. Zhongwu Li, in collaboration with Professor Andrew Hursthouse from the University of the West of Scotland, developed a new HMW framework to derive ANN model for optimization of aquatic dissolved organic matter removal by coagulation. The original research article is now published in the journal Chemosphere.

In their approach, the authors divided the factors affecting coagulation into three parts, namely: hydraulic condition (H), metal salt (M), and black–groundwater matrix (W). To realize this, researchers investigated the feasibility of the framework to determine the treatment efficiency through mathematical models. It is noteworthy to mention that no coagulation model has not been proposed through such a framework to date. Therefore, the construction of the framework and the model had not been addressed during water treatment before this.

The authors found out that non–linear artificial neural network (ANN) model was a better fit to the experimental data than the linear partial least squares (PLS) model. To be specific, the ANN model could explain 76% of the total variations while the PLS could only explain 71%. Moreover, the experiment showed that the interactions between the HMW framework components were not a simple linear relationship. Finally, the authors concluded that the HMW framework was a new classification of factors affecting coagulation, therefore leading to a better understanding of the coagulation process and sensitivity to influencing variables. According to the characteristics of machine learning, a large number of data collected in the water treatment engineering system is far more abundant than the data in this paper. Using HMW framework can better capture the internal relationship of data, so as to support the longer-term development of HMW framework.

As a typical case, the authors believe that the HMW framework can guide the construction of feed-forward system of dosing in drinking water plants, and achieve the purpose of calculating dosing quantity through the accurate signals provided by the feedback system. The construction of feedforward system can be carried out in the framework of HMW, where M component will be an important response output. In short, the concept of HMW framework is very important, which is the theoretical basis to guide the construction of water treatment model.

In Summary, the study presented a novel approach developed to investigate the application of the HMW framework constructed ANN model for analysis of aquatic dissolved organic matter removal in polyaluminium chloride coagulation. Through their experiment, researchers were able to classify several factors influencing coagulation within the HMW framework. Remarkably, the authors found out that the ANN model was effective in creating a good relationship between the framework and the coagulation efficiency. In a statement to Advances in Engineering, the lead author, Professor Guoncheng Zhu, explained that the presented HMW framework can provide WTES with more optimization measures, thereby expanding their applications in monitoring the coagulation process in WTES.

A key step to develop an AI model for coagulation process is to know HMW framework - Advances in Engineering

About the author

Dr. Guocheng Zhu received Ph.D. degree in Municipal Engineering from Chongqing University, China. His postdoctoral work is to treat harmful substances through effective nanotechnology, remedy pollution of industrial wastewater, and reduce treatment cost for enterprises to the greatest extent. At present, he is working in Hunan University of Science and Technology (HUST). His interests focus on water pollution remediation and monitoring. He has won many honors, including doctoral student academic newcomer award of Ministry of Education of China (MEC), natural science and technology award of Hunan Province, natural science and technology award of Chongqing City, and excellent scientific research achievement award of Ministry of Education of China (MEC). He has published over 60 papers and obtains more than 15 national patents. Since 2009, one of his interests is introducing the application of artificial intelligence technology into water treatment engineering system. Until now, he is conducting research on neural network application to monitoring of drinking water plant based on HMW framework and further research on deep leaning application is also developed.

About the author

Professor Andrew Hursthouse is an environmental geochemist based at the University of the West of Scotland, UK. He has a BSc (Hon) degree in Geochemistry from the University of Reading and a PhD in Environmental Radioactivity from the University of Glasgow. He undertook postdoc research on actinide geochemistry associated with nuclear waste processing facilities, developing analytical tools and applying them to study environmental transport processes. Subsequently initiating research applying these principles in studies of the impact of wastes and development of management systems, resource recovery strategies and exposure assessment. Fundamentally focused on understanding behaviour of potentially harmful elements and compounds in urban and rural environments in both terrestrial and aquatic systems.

He has extensive experience working with industry, treating wastes and problem solving production processes and is involved a number of knowledge transfer programmes. He held a High End Expert Scholarship (100 talents) from Hunan Provincial Government at Hunan University of Science & Technology, China (2016-2021).


Guocheng Zhu, Nana Xiong, Chuang Wang, Zhongwu Li, Andrew S. Hursthouse. Application of a new HMW framework derived ANN model for optimization of aquatic dissolved organic matter removal by coagulation. Chemosphere: volume 262 (2021) 127723

Go To Chemosphere

Check Also

A simple approach for adding thixotropy to an elasto-visco-plastic rheological model to facilitate structural interrogation of human blood - Advances in Engineering

A simple approach for adding thixotropy to an elasto-visco-plastic rheological model to facilitate structural interrogation of human blood