In Korea’s air pollutant inventory, construction machinery is a major emission source in the non-road sector. Since 2004, the Korean government has introduced and reinforced emission regulations to reduce the air pollutants emitted from their diesel engines. Since the engine dynamometer test method used in emission regulations has limitations in reflecting emission characteristics under the diverse working conditions of construction machinery, it is necessary to examine the effectiveness of emission regulations and the validity of the emission factors applied as inputs to the air pollutants inventory. This could be done by evaluating engine operation and emission characteristics under real-world working conditions. In this study, 14 units were selected among the excavators, wheel loaders, and forklifts that represent approximately 90% of the registered construction machines in Korea. They were equipped with a portable emission measurement system (PEMS) to measure gaseous emissions and collect engine data under various real-world working conditions. With the reinforcement of emission regulations for the construction machinery from K-tier3 to K-tier4 in Korea, exhaust after-treatment technologies, such as selective catalytic reduction and diesel oxidation catalyst, were applied. Real world NOx were reduced by approximately 83%, and THC 77% and CO by 73%, respectively. Real world NOx + THC of the K-tier3 machines exceeded the laboratory emission limit, but the K-tier4 machines considerably improved despite some differences. The emission factors applied to the air pollutant inventory have been developed using the engine dynamometer test method, but they were considerably underestimated compared with emissions under real-world working conditions. The difference was even larger for the K-tier4 machines. In this study, the possibility of developing emission factor equations that use the engine load factor as a parameter was confirmed by using the engine work 1 g/kW·h segment moving averaging window (MAW) method.

Contemporary global policy, driven by concern for the environment, imposes increasingly stringent goals for reducing CO2 emissions. Therefore, there is an urgent need to develop effective models that allow an accurate prediction of CO2 emissions from vehicles powered by alternative fuels such as CNG. This article presents the process of creating one of the first models of CO2 emission for a vehicle powered by CNG. Emission modelling is based on data obtained from chassis dynamometer tests and road tests using the portable emission measurement system (PEMS). CO2 emission modelling was conducted in Python programming language using the Optuna framework for the XGBoost technique. The models obtained were validated, with indicators R2 0.9 and RMSE 0.49 for data from chassis dynamometer tests, and R2 0.7, RMSE 0.71 for road test data. This work has the potential to be used by transportation decision makers involved in environmental analyses and policymaking for urban areas.