Water Behavior with the Use of IoT Flow Meters in University Buildings Using Dynamic Fuzzy Logic for Anomaly Detection

David Vinicio Carrera-Villacrés^{1, 2}; Alex Vinicio Andrango-Defaz¹; Anderson Fernando Lasso-Segovia¹; Cristhian Xavier Tenesaca-Maji¹

¹ Universidad de las Fuerzas Armadas ESPE. Departamento de Ciencias de la Tierra y la Construcción. Av. Gral. Rumiñahui S/N, Sangolquí, 170501, Ecuador

​

² Universidad Central del Ecuador. Facultad de Ingeniería en Geología, Minas, Petróleos y Ambiental (FIGEMPA). Carrera de Ingeniería Ambiental. Av. Universitaria, Quito, 170129, Ecuador

dvcarrera@espe.edu.ec

Background & Motivation

• University campuses present complex and variable water consumption patterns.
• Traditional monitoring systems struggle with noisy and incomplete data.
• IoT and intelligent analytics enable proactive water management.

https://www.libelium.com/es/libeliumworld/quick-report-smart-water-iot-solutions-to-fight-against-climate-change-and-scarcity/#

Research Objectives

• Monitor water consumption behavior in real time.
• Detect subtle anomalies such as leaks and abnormal peaks.
• Generate alerts and operational recommendations.

Study Area – ESPE

Geographical location and campus distribution of monitored buildings.

Study Area Characteristics

• Universidad de las Fuerzas Armadas ESPE – Sangolquí, Ecuador.
• Area: 482,000 m².
• Population: >11,000 users.

METHODOLOGY

• Smart flow meters were installed at different strategic points of the campus to collect accurate data on water consumption and losses.

​

​

• The methodological approach included dynamic and statistical analysis of the data collected, using tools such as thinger.io, Rstudio, Python and Vensim to model the behavior of the water resource and detect anomalous patterns.

​

​

Techniques used include mass balance to determine water losses and time series analysis to study flow behavior at different times of the day.

IoT Monitoring System

• Continuous acquisition of accumulated volume data.
• Wireless transmission to centralized storage.
• Real-time availability of consumption data.

IoT System Architecture

Data flow from sensor to processing and visualization modules.

Mathematical Flow Model

• Flow rate calculated as Q = ΔV / Δt.
• Derived from accumulated volume readings.
• Expressed in L/hour or L/day.

Flow Rate Calculation

Illustration of volume-to-flow conversion over time intervals.

Takagi–Sugeno Fuzzy Logic

• Handles imprecision and variability in consumption data.
• Single input: flow rate.
• Outputs: Flow Status and Alert Level.

Fuzzy Rules

Fuzzy Inference Process

Inference and defuzzification workflow in Takagi–Sugeno systems.

System Architecture

• Fuzzy logic module.
• Anomaly detection module.
• Optimization and recommendation module.

Anomaly Detection

Identification of abnormal peaks, drops, and zero-flow events.

Results Overview

• 92.7% normal and efficient operation.
• 4.1% anomalous behavior detected.
• Critical events identified by time patterns.

Flow vs Expected Flow

Comparison between measured and expected consumption patterns.

Recommendations

• Continue monitoring during normal operation.
• Inspect infrastructure during recurrent anomalies.
• Optimize pumping and maintenance schedules.

Decision Support

This module generates actionable recommendations based on the state of the clear flow and detected anomalies, which are designed for maintenance and management personnel to optimize water use and proactively respond to potential inefficiencies or leaks.

Conclusions

• Fuzzy logic improves interpretation of IoT data.
• System supports proactive and sustainable water management.
• Approach aligns with Smart Water initiatives.

Thank you so much

David Vinicio Carrera Villacrés

dvcarrera@espe.edu.ec

+593984438706

​