Development and Validation of Predictive Model for Tractor Fuel Consumption in the Gezira Scheme, Sudan

Abstract

Accurate estimation of tractor fuel consumption is essential for effective machinery management, cost control, energy efficiency assessment and environmental impact evaluation in irrigated agricultural systems. The Gezira irrigated scheme in Sudan operates a heterogeneous tractor fleet characterized by wide variations in engine power, age and operational efficiency, resulting in highly variable fuel consumption rates. This study aimed to develop, verify, validate and compare a predictive model for tractor diesel fuel consumption under the specific operational and environmental conditions of the Gezira scheme. Field experiments were conducted using eight tractors with engine powers ranging from 56.0 to 190.3 kW coupled with different implements. Actual fuel consumption (L/h) was measured using an auxiliary fuel tank method. Five regression models; linear, logarithmic, exponential, polynomial and power were developed using tractor engine power as the independent variable. Model performance was evaluated using the coefficient of determination (R²), root mean square error (RMSE) and t-test analysis. Among the tested models, the power model (Y = 0.0679 X¹·²²⁰³) exhibited the best overall performance, combining high explanatory power (R²), low RMSE, statistical robustness and physical interpretability. The model was successfully verified using experimental data and validated using independent datasets collected from Gezira, Al-Rahad, and El Suki irrigated schemes. Results showed no statistically significant differences were observed between predicted and measured fuel consumption values (p > 0.05). Comparative analysis demonstrated that the developed power model outperformed the commonly used linear PTO-based model reported in the literature, particularly at medium and high engine power levels. Sensitivity analysis further confirmed the robustness of the model and highlighted the exponent coefficient as the most influential parameter. The developed model provides a reliable and practical tool for estimating tractor fuel consumption, supporting farm planning, and mechanization management in Sudanese irrigated agriculture.