Benefits of biofuels in Sweden: A probabilistic re-assessment of the index of new cars’ climate impact

Abstract

The climate impact of new cars in Sweden 2009 has been evaluated by the Swedish Transport Administration. Their report takes into account reduction factors to attribute the positive impact of renewable fuels on CO₂ emissions. The Swedish Transport Administration recommends the public to buy cars that can run on biofuels. Besides acknowledging prevailing uncertainties for many of the input parameters to the index of new cars’ climate impact, reduction factors are based on calculations from point estimates of input parameters. A probabilistic re-assessment of the index is presented to find out the importance of these uncertainties and to assess whether the point estimated recommendation might be misguiding. Probabilistic reduction factors for CO₂ emissions were derived with the same deterministic model proposed by the Swedish Transport Administration, were Bayesian probability distributions or intervals assigned by expert judgements were used to describe uncertainty in the model input parameters. The use of biofuels most likely reduces CO₂ emissions. Probabilistic modelling indicated a CO₂ reduction for E85 as a fuel of 30% (95% credibility interval = 10–52%) in the same order as the 20% given by the Swedish Transport Administration. The best estimate of 28% decrease for gas cars (95% credibility interval = 3–44%) and is lower than the originally proposed reduction of 42%, but still within a similar range. The difference is due to the large extent of optimistic values used in the assessment by the Swedish Transport Administration. The CO₂ emissions from the production of the biofuel had most influence on the model results. We conclude that the recommendation of the Swedish Transport Administration to consumers is still valid after probabilistic recalculation.

Introduction

Carbon dioxide (CO₂) is a greenhouse gas that leads to climate change with major implications for the future of the earth [1], [2]. Road traffic worldwide accounts for 17% of the total CO₂ emissions, while figures vary from country to country (e.g. Germany 18% and Sweden 40%) [3]. The index of new cars’ climate impact [4] was the result of collaboration between the Transport Authority, Environmental Protection Agency, and the Consumer Agency. The report aimed to highlight the climate impact of biofuels as well as car purchasers’ decisions. The authors of the index developed a methodology to calculate a reduction factor to attribute the positive impact of renewable fuels on the climate [4]. The authors acknowledged uncertainty for many input parameters. Still, all calculations were made with point estimates (best estimates) coming from various sources.

Uncertainty can arise in a model when knowledge with regards to input parameters is limited (epistemic uncertainty), or as a consequence of natural variability found in input parameters (aleatory uncertainty) [5], [6]. Furthermore, the model itself can contain stochastic elements or can be uncertain with respect to how suitable it is to describe reality [7], [8].

In risk assessment, uncertainty can be captured by using a worst-case or plausible upper bound point estimate, thus providing a conservative estimate of the risk. The problem with this method is that information is lost and comparisons between different risks are meaningless due to the different levels of uncertainty [5]. Even when using best estimates for the point value, sufficient information about the effects of uncertainty on the results are not provided [6]. Furthermore, “an uncertainty analysis through a quantitative description [of uncertainties by e.g. probabilistic measures] provides better direction for further investigation” [7]. Thus, besides increased complexity to conduct the analysis and to explain the results to the public [9], the general trend leads towards more frequent use of probabilistic risk assessment [6]. The use of point estimates only to enable car purchasers and politicians to make an informed decision is most likely not the best methodology; a quantification of uncertainties might be needed to make a sound decision. Discussions on how to treat epistemic uncertainties are ongoing in the field of risk assessment. Textbooks [5], [10] and journal articles [11], [12], [13], [14], [15], [16], [17], [18] are not conclusive on which of the tools based on probabilities, intervals or probability bounds (or a combination of these) is most appropriate for uncertainty analysis and decision making. Clearly, the mathematical results are different comparing, for example, probabilistic modelling and interval analysis [5], [11]. Bayesian probabilistic modelling has been argued to result in an unjustified sense of precision (too narrow distributions) given the lack of knowledge. On the other hand, probabilistic outputs are more usable than intervals as a tool to base decisions on. In order to cover for these open questions, a probabilistic modelling of the re-assessment of the index was complemented with an interval analysis for comparison.

The aim of this study was to quantify the uncertainty related to the calculation of CO₂ emission reduction factors in the index of the Swedish Transport Administration to determine CO₂ emissions of new cars using the methodology of probabilistic risk assessment. Based on this evaluation, the objectives were to

• Assess the reduction factors of CO₂ emissions for new cars in Sweden accounting for uncertainties in input parameters.

• Identify the most influential input parameters to the reduction factors.

• Verify the appropriateness of the consumer recommendation (from 2009) in the light of identified and characterised sources of uncertainty.