تحقیق و توسعه مرتبط و سیاست های سرمایه گذاری به نفوذ بازار CCS از طریق یادگیری دو عامله

Carbon capture and storage (CCS) has the potential to play a major role in the stabilization of anthropogenic greenhouse gases. To develop the capture technology from its current demonstration phase towards commercial maturity, significant funding is directed to CCS, such as the EU’s €4.5 bn NER300 fund. However, we know little about how this funding relates to market diffusion of CCS. This paper addresses that question. We initially review past learning effects from both capacity installations and R&D efforts for a similar technology using the concept of two-factor learning. We apply the obtained learning-by-doing and learning-by-searching rates to CCS in the electricity market model Hector, which simulates 19 European countries hourly until 2040, to understand the impact of learning and associated policies on CCS market diffusion. We evaluate the effectiveness of policies addressing learning-by-doing and learning-by-searching by relating the policy budget to the realized CCS capacity and find that, at lower policy cost, both methods are about equally effective. At higher spending levels, policies promoting learning-by-doing are more effective. Overall, policy effectiveness increases in low CO2 price scenarios, but the CO2 price still remains the key prerequisite for the economic competitiveness, even with major policy support.

prime challenge for the 21st century is the limiting of global warming to 2 °C through the reduction of anthropogenic greenhouse gas emissions, formulated as a central result in the joint Accord of the Copenhagen Conference on Climate Change (2009) and the Cancun (2010). There are several CO2 emission reduction targets addressing this challenge, such as the EU’s commitment to a 20–30% target by 2020 (EU Commission, 2008). As the main contributor to these emissions, the energy sector is especially impacted by this development and significant efforts are made to address this challenge. Besides renewable energy sources, energy efficiency increases, nuclear power generation and others, Carbon capture and storage (CCS) is widely seen as a major opportunity to contribute to CO2 abatement, but at the same time also continue fossil-fuel-based generation. In a context of increasing demand of energy, the measures of improvement of the energy efficiency and the development of the renewable energies are indispensable. However, they are not sufficient to give up fossil energies, which remain the main source of energy in 2030 (IEA, 2010a). The IEA’s World Energy Outlook (WEO) 2010 in its 450 ppm CO2e scenario predicts that 581 GW of CCS capacity will be in operation by 2035 worldwide (IEA, 2010a), and CCS has now been accepted as Clean Development Mechanism by the UN (IEA, 2010b). Whereas the expectations placed on CCS are very high, the capture technology has still not been widely proven at full scale and technological progress has been limited in recent years, with only a few CCS pilot power plants being operational.1 This presents a key obstacle to the anticipated large-scale CCS deployment. The solution lies in technological and managerial learning through extended R&D efforts as well as in the physical construction of an increasing number of (demonstration) CCS power plants. Given the relatively high profile of CCS, we observe the need for additional CCS pilot plants and therefore investments, as foreseen for example in the EU’s plan to have 12 plants operational by 2015 and to provide approx. €4.5 billion (300 m EU ETS certificates) of co-funding for CCS pilot plants in the NER300 fund (EU Commission, 2009a). The United Kingdom (UK),2 Canada,3 and the State of Illinois, USA,4 have similar policies in place. This support is also needed, as stand-alone CCS power plant projects are only commercially viable in specific situations, such as in combination with enhanced oil recovery. Without this support, CCS runs the risk of being trapped in the "Valley of Death", the gap between public and private funding, especially due to the high up-front investment costs (Murphy and Edwards, 2003). A variety of support methods are available, such as R&D grants, subsidies for investment costs and others (Woerdman and Couwenberg, 2009). However, whereas the need to support CCS is accepted and continuously growing, we know little about the dynamics of how to optimally support this technology. This paper addresses these questions concerning R&D effectiveness, funding distribution, and funding level. One method to estimate the relationship between R&D funding and technological improvement is “two-factor learning curves” (2FLC). This approach is based on "technological learning", the phenomenon that the cost of a specific technology decreases along with its cumulative deployment (initially Wright, 1936 and Arrow, 1962), but is extended by the additional consideration of cumulative R&D efforts (Kouvaritakis et al., 2000a, Kouvaritakis et al., 2000b and Jamasb, 2007). From a policy-analysis perspective, traditional learning-by-doing approaches only consider capacity deployment as the driving factor, thereby limiting any policy research to procurement policies that support investments in new capacity. However, policies supporting R&D, although a very popular method, cannot be analyzed with this approach, especially if the technology is at an early development stage. In this paper, we estimate a 2FLC for CCS power plants through analogies, as no empirical data are available because CCS deployment has not yet started. We therefore empirically derive the 2FLC for the SO2 scrubber technology, which is similar to the CO2 scrubber technology used for CCS power plants,5 using cost, patents, and deployment data for the years 1970–2000. To validate the results, we compare them to already-published 2FLC estimates across the energy-generation industry as well as existing one-factor learning estimates for CCS, derived through the same SO2 scrubber analogy (Riahi et al., 2004) or through expert panels (McKinsey, 2008). Based on the 2FLC, we address the question of R&D and investment policy effectiveness for CCS power plants using a modified version of the Hector simulation model (Lohwasser and Madlener, 2009 and Lohwasser and Madlener, 2012). We provide an outlook for the European electricity market, including the diffusion of CCS technology under different policy scenarios until 2040 to explicitly consider the long-term effects of technological learning. The long time horizon is required due to the initial development stage of CCS, as early learning has a strong impact on the future. The analysis explicitly analyzes potential CCS policies considering the two effects mentioned: learning-by-doing (stimulation of deployment through investment cost subsidies) and learning-by-searching (promotion of R&D through grants). The remainder of this paper is structured as follows. In Section 2, we discuss the concept of technological learning which, in Section 3, we then apply to CCS, using our own empirical analysis and comparisons to preexisting one- and two-factor learning curves. Section 4 focuses on the implementation of two-factor learning to the model used to simulate market diffusion as well as the description of the Base Case. In Section 5, we analyze the impact of learning-by-doing and learning-by-searching and, in Section 6, the impact of R&D and investment subsidy policies. Finally, Section 7 concludes with key takeaways and policy recommendations from our analysis.

CCS is a technology with a very large GHG abatement potential, yet it is still far from reaching technological maturity and commercial readiness. To date, significant public and private funding is already dedicated to the technology. However, we know little about how different subsidy schemes relate to the success of the technology in the market, especially with regard to the impact of R&D. A method to relate R&D efforts to the economics of a technology is the concept of two-factor learning, which describes the relationship between the costs of a technology with its cumulative deployment through a learning-by-doing rate, and with cumulative R&D efforts through a learning-by-searching rate. It extends the well-known and intensively-researched notion of technological learning, which just relates cumulative deployment with R&D efforts. To estimate the learning rate for CCS, we estimate the values for FGD based on its empirical R&D, capacity and cost development between 1970 and 2000. FGD uses SO2 scrubbers, which are technologically similar to the CO2 scrubbers deployed in post-combustion CCS, so that this analogy was also already applied to CCS for (one-factor) learning in related work. We find that the learning-by-doing rate is 7.1% and the learning-by-searching rate 6.6%, assuming that R&D patents can be used as a proxy for R&D spending. This effectively means that the cost reduction caused by a doubling of installed capacity is roughly the same as for a doubling of R&D efforts. Other technologies that are currently promoted through subsidies and other funding types, such as solar power and wind power, have learning-by-searching rates that are 2–5 times higher than their learning-by-doing rate, indicating advantages for R&D-driven policies over capacity-addition-driven policies. This conclusion cannot be drawn for CCS, however, as the observed learning rates are very similar, a fact that should be considered when comparing or even applying wind and solar promotion policies to potential CCS promotion policies. The observed values are also in line with expectations drawn from existing publications in the literature. To understand the relevance of technological learning for commercial success, we simulated CCS deployment across 19 European countries with different learning rates with the bottom-up, hourly electricity market model Hector. As a simulation result, we observe only a relatively limited impact of technological learning, with CCS capacity variations of only 2–3 GW by 2040, depending on the learning rate in the high CO2 price scenario CO2-38. This is marginal, considering an overall capacity of 217 GW. Even without learning, CCS capacity is only reduced by 12 GW. The impact of learning is stronger in the low CO2 price scenario CO2-25, but still quite low. Total CCS capacity is, however, almost halved, due to the less favorable market conditions, at only 154 GW by 2040 at the standard 7% learning rate. This effectively means that the learning rate is not the real driver for the market success of CCS. Instead, other factors such as CO2 prices or national plant portfolios, play a far more important role in terms of plant profitability, the driver for investment in our model. A key reason for the relatively low importance of learning is that it only applies to the CCS equipment; the attached coal-fired power plant makes up the majority of the cost, but hardly experiences any learning at all. In an individual sensitivity measurement for each learning rate, a variation of the learning-by-doing rate has a slightly larger impact than the learning-by-searching rate, leading to the conclusion that technological advancement through capacity additions plays a slightly more important role than through R&D. To link specific CCS promotion policies to technological learning, we analyze two policies. One provides R&D funding, addressing learning-by-searching progress. The other provides a subsidy for new CCS plants, reducing the investment costs investors have to pay by a certain percentage. This demand-side policy promotes diffusion and addresses learning-by-doing. At high CO2 prices, both policies only slightly improve the diffusion of CCS technology, and the policy type – i.e., R&D or investment subsidies – plays only a secondary role as their effectiveness is relatively similar, with slight advantages for an R&D-based policy. At lower CO2 prices, the impact of the investigated policies rises and they provide a suitable method for improving CCS diffusion. However, even a massive policy budget cannot compensate for CO2 prices as a key driver for CCS success. Even if €5 billion is spent annually after 2015 on CCS, the technology will not reach the capacity needed to reach commercial readiness of 21–22 GW by 2020, regardless of policy type. In a direct comparison between both policies, their effectiveness is similar at a budget below €0.5 billion p.a., but beyond that, investment subsidies are the more effective policy type. This is due to the logarithmic impact of R&D effort on investment costs, which cannot compensate for the linear reduction of investment costs of the investment cost subsidy. The overall situation is difficult for policy-makers: If CO2 prices are sufficiently high, no diffusion stimulation policies are needed in the first place. If not, opportunities for specific CCS promotion policies exist and do improve the situation, but their impact will never outweigh the unfavorably low CO2 prices, unless extraordinarily high budgets are allocated for CCS. Aggressive GHG reduction policies with high CO2 prices are therefore of prime relevance for CCS. If CCS policies are deployed at relatively low CO2 prices (such as 25 €/t), the impact of R&D and investment subsidy policies on CCS diffusion is about equally effective below €0.5 billion p.a.; beyond that, R&D policy effectiveness stagnates, compared to a continued linear growth for investment subsidies. In summary, we can therefore conclude that both effects on technological learning – R&D and capacity diffusion – are very similar for CCS, suggesting a simultaneous and balanced two-way policy, an insight consistent with the findings in 3 and 5. Given the already high policy budget of over one billion Euros annually across Europe, however, the R&D stimulating portion should be lowered.