HSR Emissions Paper Was Wrong
Note from Robert: This post is by Clem Tillier, who runs the excellent Caltrain HSR Compatibility Blog.
In late 2009, Mikhail Chester and Arpad Horvath of Berkeley’s Institute for Transportation Studies published a paper entitled “Life-cycle assessment of high-speed rail: the case of California” in the academic journal Environmental Research Letters. This paper received some attention from the media and the blogosphere because it appeared to debunk a popular notion among high-speed rail supporters: that HSR is one of the greenest forms of intercity transportation.
The key finding of the study was that HSR didn’t stack up against cars, planes or regular trains unless the trains were fairly full, as shown in the figure below. (See paper for full caption and explanation)
Leaving aside for the moment the unprecedented notion of high-speed trains being operated at 10% seat occupancy–in reality, service frequency would be dropped to achieve higher loads–the numbers used in the study show that energy consumption, greenhouse gas emissions, and sulfur dioxide emissions are dominated by emissions from “Vehicle Active Operation,” i.e. electrical power consumed for the purpose of actually running trains.
Unfortunately, that is where the study falls apart.
Berkeley’s numbers are undone by a simple unit conversion error committed by a CHSRA consultant. Conversions between metric and imperial units are prone to errors and misunderstandings, most famously in the case of NASA’s $300 million Mars Climate Orbiter mission, which was inadvertently crashed into Mars because of an overlooked conversion between pounds and Newtons. In the case of the high-speed rail study, the CHSRA consultant’s unit conversion error leads to an overestimate of HSR energy consumption by a factor of nearly four–not just in the Berkeley study, but also in the CHSRA’s program level environmental reports.
The energy consumption figure cited in the Berkeley study and its supplementary data is 170 kilowatt-hours per vehicle kilometer traveled, or kWh/VKT, a measure of how much energy a high-speed train consumes on average when traveling one kilometer. This number is correctly converted by Berkeley from a figure of 924,384 BTU/VMT referenced in the energy chapter of the 2008 CHSRA program-level EIR. That chapter in turn references a peer-review study performed for CHSRA by the German firm DE-Consult in 2000, which evaluated the energy consumption of a hypothetical 16-car trainset with a seating capacity of 1200 and a design speed of 385 km/h (240 mph) and an operating speed of 350 km/h (220 mph), essentially a souped-up German ICE3. The DE-Consult study (unavailable online) contains detailed performance simulations for the proposed California system that give the average energy consumption of such a train as 74.2 kWh/VMT, or 46 kWh/VKT (see copy of Annex 4-11). And therein lies the error: CHSRA’s consultant botched the conversion from kilowatt-hours to British Thermal Units, feeding Berkeley a figure of 170 kWh/VKT instead of 46 kWh/VKT.
While the conversion error is no fault of the Berkeley researchers, the least they could have done is cross-checked the number with other papers in their own field of transportation research, such as those referenced here, showing HSR values in the 20 – 30 kWh/VKT range. (DE-Consult’s California figure is higher because the assumed train is very long, with 1200 seats, and operates at higher speeds where aerodynamic drag increases rapidly.) An elementary back-of-the-envelope calculation of the physical quantities involved, such as a train’s maximum power rating, its travel distance, and its timetable, would also have indicated that 170 kWh/VKT was implausibly high.
Using the correct number, the study’s conclusions would be significantly altered, as shown in the modified figure below, where the contribution from Vehicle Active Operation has been proportionally scaled down to the correct DE-Consult number.
The study’s comparison of HSR with other modes (especially cars and aircraft) would show California’s HSR pulling ahead with a significant life-cycle advantage in energy use and greenhouse gas emissions, and with a far lower sensitivity to ridership factors. In view of the enormous errors induced by a single incorrect parameter, it is incumbent on Chester and Horvath to acknowledge this major flaw and to publish a correction in Environmental Research Letters.