Author ORCID Identifier

Date Available


Year of Publication


Degree Name

Master of Science (MS)

Document Type

Master's Thesis




Civil Engineering

First Advisor

Dr. Lindell Ormsbee


The main objective of this thesis is to develop a working digital twin for a small water system in central Kentucky which will serve as a general format for other similar systems in the region wishing to implement digital twins for operator support. While the benefit of having a calibrated hydraulic and water quality model is widely understood, small distribution systems tend to not have the same financial and economic means to properly support these tools. Creation of a digital twin using this methodology provides a means for operators to predict pressure, flows, chlorine residuals, and total trihalomethane (TTHM) concentrations within their system with little to no cost and maintenance.

The application is developed using the MATLAB app development toolkit and is then linked with the EPANET hydraulic and water quality engine via the EPANET-MATLAB toolkit. The application provides simple user inputs such as initial tank levels, pump scheduling, demand scenarios, and mapping capabilities for results.

Reliability of the digital twin output is rooted in the extended period simulation (EPS) calibration steps which ensure the variation of demands both spatially and temporally accurately reflect conditions seen in the system. Both the Box-Complex (multi pressure zone systems) and the bisection method (two zone systems) were used in the processing of tank telemetry and meter data to create representative demand factors.

The creation of a useful digital twin is highly reliant on both the programming capability of the developer and familiarity with the many nuances of hydraulic and water quality calibration which are necessary foundations upon which accurate predictions of key parameters are accomplished. While outputs given in the MATLAB interface are simple, accurate, and robust against failure, there is much to be desired by way of interactive mapping. Python offers a broader range of available libraries capable of supporting mapping which will make inputting parameters and viewing results much simpler for operators. Additionally, the tools provided in this digital twin use historical data for hydraulic calibration (demand factors) and testing which are useful based solely on operator understanding of which demand scenarios in the past will most accurately reflect what they will see in the present. Extension of these historical patterns into forecasted demands using machine learning or time series analysis will greatly improve the usefulness of the model and overall operator experience.

Digital Object Identifier (DOI)

Funding Information

Robert A. and Maywin S. Lauderdale Graduate Fellowship