Association for Public Policy Analysis & Management

Transportation is one of the biggest sectors for fossil fuel consumption and greenhouse gases (GHG) emissions. Its high dependence (95%) on fossil fuel and infrastructure lock-in makes it the most difficult sector to decarbonization. The rising concern over climate change has dawn more attention on biofuels as a renewable substitute for fossil-based transport fuels. In particular, ethanol produced from lignocellulose is estimated to be a low-carbon fuel that has higher potential than starch-based ethanol in reducing GHG emissions. However, growth in the biofuel industry is dominated by corn ethanol – with rapid cost reduction, corn ethanol production in the U.S. has increased to 15.8 billion gallons in 2018, taking up 10.3% market share. The production of cellulosic ethanol, however, is stagnant and volatile. According to the public Renewable Fuel Standard (RFS) data, cellulosic ethanol production grew from 0.7 million gallons in 2014 to 10.0 million gallons in 2017, but dropped to 8.2 million gallons in 2018. Although RFS provide incentives for cellulosic ethanol, actual production is limited to cost and feedstock supply constrains. Engineering models and experience curves predict that advanced biofuel will have significant cost reduction when considering economies of scale, and the benefits from learning by doing. Assuming fast learning, researchers also suggest larger capacity size for cellulosic ethanol plants. But current data does not find strong evidence for cost reduction. Rather, biorefineries for advanced biofuel are frequently found shuttered, indicating a failure of learning.

This study attempts to understand what factors affect the growth of cellulosic ethanol by considering capacity size, feedstock supply, costs, the underestimation of cost due to optimism, and cost reduction due to learning. A hybrid general equilibrium model was used to predict cellulosic ethanol production under different scenarios that may inhibit or facilitate learning and cost reduction in the future. We model a stereotype biorefinery that produce fuel ethanol from agriculture residues and energy crops with nameplate capacity of 4,400 b/sd and utilization rate of 85%. We assume the four plants of a kind will experience revolutionary learning, and the actual cost will be 120% of planned to count for the tendency of underestimation due to optimism. We assume learning will lower the cost for the 5^th plant and beyond. A range of settings on capacity size, capital cost, labor cost, and learning rate are tested while also considering current policy constrains of obligated volumes and cellulosic waiver credits by RFS.

The result suggests that capacity size and costs are the factors that inhibit production: building biorefineries at a size half of the stereotype plant with halved costs will increase cellulosic ethanol to over 1.5 billion gallons by 2040, and higher learning rate will further increase production to more than 2.5 billion gallons. But the learning effect has no impact if continuing building large capacity ethanol plants. This finding provides policy-makers with new insight that building more of smaller-size biorefineries will allow more plants to learning, and thus reduce cost and increase cellulosic ethanol production.