AI-driven, Demand-rated, Reconfigurable Pipeline Design and Drinking Water Distribution System

 by  G. A. Shanmugha Sundaram  -  Department of Electronics and Communication Engineering, Amrita School of Engineering, Coimbatore, Amrita Vishwa Vidyapeetham, India and  Center for Computational Engineering and Networking (CEN), Amrita School of Engineering, Coimbatore, Amrita Vishwa Vidyapeetham, India

Statement of the Challenge

In the large cities worldwide, a typical population is in the range of 1-5 million humans, as per the limits set for its administrative boundaries with a municipal corporation as its fundamental unit of goverance administered by a dedicated authority [1, 2]. Several such urban conclaves constitute a sprawling city and metropolitian area [3]. There is a crucial need for assured, demand-rated, smart potable pipeline water supply in facilities for inhabitation which may be individual residences and commercial facilities. The current practice is heavily human intensive, and has the attended element with all its fallacies. An assorted distributions of variables, that include pipeline dimensions, nodal valves, check dams, reservoirs and overhead water tanks along the path of water flow from the catchment area (main reservoir), could be identified as impact parameters for a reliable supply of clean water to the end user [4]. Such parameters need a thorough software-based modeling to be conducted [5, 6], that would incorporate variables like pressure-heads, valve mechanisms, tolerance limits to the pipes and valves, wear and fatigue elements in the pipelines and valves, water utilization patterns (seasonal and diurnal), wastage due to leaks enroute [7], water quality (that factors in pollutants of various types), weather patterns over recurrent and transient time scales, geographic locations of prime sources for water from rainwater-dependent regions, snow and glacial meltwater thaw patterns, and global warming.

Technological Innovation

The smart solution advanced for the proposed work considers each of these variables, in order to optimize them in a unique combination of factors that sets the peak efficiency metric in the water distribution system, on a geographic scale of a typical city municipal corporation. Optimal performance of the system derives the benefits of data driven algorithms such as machine learning (ML) and deep learning (DL), thereby leveraging the concepts of artificial intelligence (AI).

As a typical case, several types of valves that are deployed at various levels of water distribution are set to operate in a dual mode, whereby, while still having the existing arrangement in place, a suitable valve actuation mechanism is also rigged in. A common valve that is employed along the pipeline is the ubiquitous multi-turn rotary valve. Using a rotary encoder, the precise number of turns about the valve axle determines the exact volume of water that moves across this point in the pipeline. The encoder module has the attached linear actuator that bring about a realization in the water flow pattern. The convolutional neural networks (CNN) algorithm is of an evolved ML/DL class, used to set the functional attributes for the valves, that takes specific cues during its extensive training phase wherein it had been applied on pre-compiled data in the various domains. The CNN algorithm, that has established an apriori demand pattern in the given region of water usage, now determines the appropriate granularity in terms of the requisite number of layers of classification and trend estimation when applied on the test or real-time data. This gets further translated into the equivalent inputs for the encoder, and the corresponding number of turns of the rotary valve that regulates the quantity of water flow to the region of interest.

Import and Potential

The critical information component of the communication data packet is sourced from a GIS database on water utilization over variable levels of granularity, and translated into the appropriate release mechanism in the the valve functionality. This shall enable a measured quantity of water flow past the valves for a specific duration, to a regions whose demand profiling has been well characterized, in terms of a wider spectrum of temporal domain analyses. Use of a data driven approach, to be implemented in ML or AI, renders a highly dynamic flow pattern, while the element of reconfiguration rationalizes the total stock meant for efficient distribution.

Anticipated risks and mitigation strategies

A GSM-based communication solution would also offer the necessary data security during every stage of this process, deployment of which shall attend to another crucial element that denies malicious intrusions from disrupting the system's functionality. The entire system is intended to be self-driven, when once implemented on its full scale, thereby ensuring a reliable and safe civic water supply to the city dwellers.

Conclusions and future work

The proposed solution would offer the least amount of disruption to the existing arrangement, that is significantly human driven and highly subjective, by also including smart augmentation technologies that have reconfigurability and scalability attributes, while keeping the developmental and operational costs to a minimum, given its efficiency improvements and human-optional aspects.The scheme would thereby enable equitable availability and measured utilization of a precious commodity, such as water, given its ability to make objective decisions that control and rationalize the flow pattern and distribution based on a computation-intensive multivariate data analytics process..

Acknowledgements

The author wishes to profoundly thank the two reviewers of the original manuscript who had generously offered their wise comments and critical inputs in a manner that has positively refined the contents and presentation.

 

References

1. World Urbanization Prospects: The 2007 Revision Population Database; United Nations, 2008; URL: http://esa.un.org/unup/index.asp?panel=6 (accessed on 15 November 2018).

2. SOWC-2012-DEFINITIONS; UNICEF, 2012; URL:https://www.unicef.org/sowc2012/statistics.php (accessed on 15 November 2018).

3. Squires, G. (Ed.), “Urban Sprawl: Causes, Consequences, & Policy Responses”. The Urban Institute Press (2002).

4. Araujo, L.S., Ramos, H., Coelho, S.T., "Pressure control for leakage minimisation in water distribution systems management"; Water Resources Management: Springer Nature Switzerland AG, 20(1) (2006) 133-149.

5. Chandapillai, J., Sudheer, K.P., Saseendran, S., "Design of Water Distribution Network for Equitable Supply"; Water Resources Management: Springer Nature Switzerland AG, 26(2) (2012) 391-406.

6. Creaco, E., Franchini, M., "A new algorithm for real-time pressure control in water distribution networks"; Water Science and Technology - Water Supply: IWA Publishing UK,

   13(4) (2013) 875-882.

7. Xu, Q., Chen, Q., Ma, J., Blanckaert, K., Wan, Z., "Water saving and energy reduction through pressure management in urban water distribution networks", Water Resources

   Management: Springer Nature Switzerland AG, 28(11) (2014) 3715-3726.

8. Machell, J., Mounce, S.R., Boxall, J.B., "Online modelling of water distribution systems: A UK case study"; Drinking Water Engineering and Science: Copernicus Publications UK, 3(1) (2010) 21-27. 

 Author

Dr. G A Shanmugha Sundaram (IEEE Membership #: 92232641) is Research Faculty with the Center for Computational Engineering and Networking, Department of ECE, Amrita Vishwa Vidyapeetham, Coimbatore, India. He has a proven record of research, spanning nearly 20 years, conducted across a crosssection of engineering disciplines that all involve array topologies, in areas as diverse as radio astronomy, vehicular traffic, and distributed sensors. He is currently the PI of an academic research project funded by National Instruments Inc. USA. In the area of RADAR systems design. He had earlier been research scientist in the Square Kilometer Array Working Group on Design & Simulations, at ASTRON, Netherlands, EU. He obtained his PhD in 2004 from the Indian Institute of Science, Bangalore, India, in the areas of Solar Astrophysics and RF front end instrumentation design for radio astronomy.

19 December 2019