Comparative analysis of fuel level monitoring systems and justification of a hybrid microprocessor-based system for river vessels

Abstract

A comparative analysis of four types of fuel level monitoring systems (float, capacitive, ultrasonic, hydrostatic) was carried out under real operating conditions on the river vessel OTA-900 (Project 758B) in the Siberian region. The study aims to justify the transition from outdated float systems to a modern hybrid microprocessor system based on Arduino, combining capacitive and hydrostatic sensors. During a 30-day experiment with a daily fuel consumption of 1200 liters, key accuracy metrics were measured: Root Mean Square Error (RMSE) and Mean Absolute Error (MAE). Results showed that float sensors have the highest error (RMSE ≈ 7.1%), while capacitive (RMSE ≈ 1.5%) and hydrostatic (RMSE ≈ 2.0%) sensors demonstrate high accuracy and stability. The hybrid system provides fault tolerance, automatic cross-verification of readings with a 3% alarm threshold, and adaptation to extreme conditions, as confirmed by the absence of false alarms for these sensors, unlike ultrasonic and float types. Implementing the proposed system increases measurement accuracy fivefold, enables continuous monitoring, and provides automatic alarms, which is critically important for navigation safety in remote regions.