Effect of vehicle properties and driving environment on fuel consumption and CO2 emissions of timber trucking based on data from fleet management system. (September 2022)

• Record Type: Journal Article
• Authors: Anttila, Perttu; Nummelin, Tuomas; Väätäinen, Kari; Laitila, Juha; Ala-Ilomäki, Jari; Kilpeläinen, Antti
• Is Part Of: Transportation research interdisciplinary perspectives. Volume 15(2022)
• Publication Date: 2022-09
• Subjects: Log truck -- Fuel economy -- Greenhouse gas emissions -- CAN bus -- Machine learning; Transportation -- Periodicals; 388.05
• DOI: 10.1016/j.trip.2022.100671
• Languages: English
• ISSNs: 2590-1982

Abstract

Highlights: Fuel consumption (l(100 km) −1 ) decreased with increasing transportation distance. Machine learning was applied to model fuel consumption. Driving speed, road gradient, pavement and sinuosity explained consumption. Data resolution too low for the highly varying operating environment. Abstract: This study evaluated fuel consumption and CO2 emissions for 13 typical log trucks in operating conditions in Finland. The effects of season, transportation distance, mass, vehicle and road properties, and weather conditions on fuel consumption for driving were analyzed and modeled. The average fuel consumption and CO2 emission of the 76-t trucks when driving with a load was 0.013 l(t·km) −1 and 30.856 g(t·km) −1 respectively. The consumptions and emissions for the 68- and 76-tonners were found to be at the same level per tonne kilometer due to the overload of the former. The highest consumptions were measured in January (on average 57.5 l(100 km) −1 ), and the lowest in July (on average 48.7 l(100 km) −1 ). Machine learning was applied to predict fuel consumption with the above-mentioned factors for 73, 686 road segments. Based on the developed models, driving speed was the most influential explanatory variable, in addition to road gradient, pavement, and sinuosity. Engine power and truck mass had minor importance. Wind effect was the only significant weather variable. The "big data" approach, as used in this study, enables the collection of a vast amount of data on very varying conditions in log transportation. However, higher resolution data than the fleet management system data used in this study will be needed to construct more accurate models.