Comparing Commercial Vehicle Fuel Consumption Models using Real-World Data under Calibration Constraints

Abstract

Assessing commercial road vehicle fuel use at a high spatiotemporal resolution helps in understanding underlying usage patterns and informs future interventions toward fuel-efficient freight planning and operations. With the use of global navigation satellite systems in fleet tracking and advancements in driver activity surveys, instantaneous fuel consumption models can calculate fuel use in high resolution using inputs like speed and acceleration derived from GPS data. Given that several models exist, there is a need to compare fuel use estimates from different models, especially under the constraint of limited data for calibration. This study evaluates the accuracy of fuel use estimates from three fuel consumption models (COPERT 4, SIDRA TRIP, and MOVES) applied to 10 diesel commercial road vehicles in Singapore over a standardized drive cycle (NEDC) and real-world activity data derived from GPS traces using the method of space-time path segments and road grade obtained from a digital elevation model. Changes in model performance are examined when supplementary on-board diagnostics (OBD) data and payload information are used. The models gave varying fuel use estimates over the NEDC, especially for heavier vehicles in the sample. When applied to real-world data, SIDRA TRIP was found to be the most accurate for the context studied. SIDRA TRIP’s performance improved consistently when supplemented with OBD and payload information. This comparison approach allows analysts to select the most suitable model for a given context, and take steps toward more sustainable freight transportation.