Bias of Averages in Life-Cycle Footprinting of Infrastructure: Truck and Bus Case Studies

• Published in: Environmental science & technology 2014-11, Vol.48 (22), p.13045-13052
• Main Authors: Taptich, Michael N, Horvath, Arpad
• Format: Article
• Language: English
• Subjects: Air Pollutants - analysis; Applied sciences; Bias; Buses; Climatology. Bioclimatology. Climate change; Earth, ocean, space; Emissions; Environmental Monitoring; Exact sciences and technology; External geophysics; Ground, air and sea transportation, marine construction; Humans; Infrastructure; Life cycles; Meteorology; Motor Vehicles; Road transportation and traffic; San Francisco; Trucks; United States; United States Environmental Protection Agency; Vehicle Emissions - analysis
• ISSN: 0013-936X; 1520-5851
• DOI: 10.1021/es503356c

The life-cycle output (e.g., level of service) of infrastructure systems heavily influences their normalized environmental footprint. Many studies and tools calculate emission factors based on average productivity; however, the performance of these systems varies over time and space. We evaluate the appropriate use of emission factors based on average levels of service by comparing them to those reflecting a distribution of system outputs. For the provision of truck and bus services where fuel economy is assumed constant over levels of service, emission factor estimation biases, described by Jensen’s inequality, always result in larger-than-expected environmental impacts (3%–400%) and depend strongly on the variability and skew of truck payloads and bus ridership. Well-to-wheel greenhouse gas emission factors for diesel trucks in California range from 87 to 1,500 g of CO2 equivalents per ton-km, depending on the size and type of trucks and the services performed. Along a bus route in San Francisco, well-to-wheel emission factors ranged between 53 and 940 g of CO2 equivalents per passenger-km. The use of biased emission factors can have profound effects on various policy decisions. If average emission rates must be used, reflecting a distribution of productivity can reduce emission factor biases.