Abstract
This paper investigates the impact of energy policy and the regulatory environment on the deployment of technologies based on renewable energy sources in Germany and presents a model to forecast future renewable energy technology diffusion. Our concept is based on the development of a composite indicator for renewable energy diffusion with a focus on wind energy (onshore) and photovoltaic. The approach has three major components: in-depth, semi-structured interviews with renewable energy developers and energy sector stakeholders to identify the major drivers and barriers (determinants) for renewable energy diffusion in the case study country; an EU-wide, questionnaire-based survey to understand the relevance (weights) of the individual determinants; and an analysis of past renewable energy diffusion patterns resulting in the deduction of a model for short-term renewable energy technology diffusion forecasts. Results demonstrate the substantial impact of economic and non-economic framework conditions on the diffusion of photovoltaic and wind energy in Germany. Furthermore, the use of composite indicators for renewable energy technology diffusion analyses proves to be a promising approach.
Introduction
In 2009, the European Member States (MS) adopted the Renewable Energy Directive, 1 which stipulates that 20% of the total final energy consumption have to come from renewable energy by 2020. On this basis, all EU MS have developed National Renewable Energy Action Plans (NREAPs) 2 and national support schemes for the exploitation of renewable energy sources (RES). Until 2030, an EU-wide share of at least 27% RES is envisaged, 3 and several MS have formulated respective national targets. Germany, for example, envisages an RES share of 30% in gross final energy consumption by 2030.
Against this background, monitoring and forecasting the diffusion of RES technologies is central to being able to examine the attainment of RES targets and – in case that the achievement of these objectives is at risk – to being capable of implementing required changes in the regulatory framework in time. Energy policy-making in general becomes increasingly complex and brings about the need for solid and transparent decision support tools for policy makers.4,5 Especially for quantity-based RES support schemes (i.e. tendering or quota schemes), it is crucial to be able to anticipate future market sizes and scarcity to allow for an adequate policy design (e.g. definition of suitable tender volumes and quota targets). Thus, realistic prognoses of RES market growth for the near future constitute an important tool for monitoring and improving RES policy frameworks. However, a central requirement, in order to be able to provide such accurate short-term market forecasts, is a more comprehensive understanding of the multitude of drivers and barriers for RES deployment and their impact on the resulting RES diffusion process.
A growing number of scientific publications 6 – 14 and European research projects 15 – 20 points out that, apart from primary economic factors (e.g. support levels, technology costs or access to financing), also non-economic factors play a major role for investment decisions in RES technologies. Such non-economic factors comprise, for example, the complexity and duration of administrative processes12,13,16,21– 23 and grid connection procedures,6,12,13,15,16 spatial planning issues8,23– 25 or the design of regulations affecting the access for RES producers to different electricity market segments.14,26
In this paper we present an approach to an integrated assessment of the framework conditions for renewable energy technology diffusion based on the development of a composite indicator (CI) covering economic and non-economic diffusion factors. Based on the indicator we develop a model for short-term diffusion forecasts (up to three years). The approach includes an extensive dialogue with primary decision makers, namely renewable energy technology developers and investors, to ensure that their view of the attractiveness of regulatory environments for renewable energy projects is fully captured by the indicator.
The paper is structured as follows: The section ‘Methodology overview’ provides a brief overview of the methodology. The section titled ‘Composite indicator for RES diffusion’ presents the components of the CI and their relevance for the overall assessment, while the section ‘Derivation of the diffusion model’ introduces the diffusion model. Under section ‘Case study results Germany’ we present the results of applying the approach to the case of Germany, and the section ‘Conclusion and outlook’ presents the conclusion. The Appendix provides a list of notations used in equations in the text.
Methodology overview
In order to derive an objective and reliable tool for monitoring and benchmarking RES diffusion frameworks, the applied methodology should fulfil the following criteria: it should provide a robust understanding of the major determinants for RES diffusion; it should consistently represent the perspective of the concerned stakeholders (namely RES developers and investors) and it should build on a broad empirical basis and transparent data sources. Our approach follows the best practice guidelines compiled in OECD
27
and combines qualitative and quantitative methods. It can be structured into four major steps:
Identification of the major framework factors (determinants) for RES diffusion and selection of suitable indicators and data sources to represent each of them. The development of the conceptual framework for the CI was based on an extensive literature review and expert consultation and is described in detail in (Boie et al.
28
). An overview of the results is given in the section ‘Components of the Composite Indicator’. Assessment of the relative relevance (weight) of each determinant in the diffusion process through stakeholder consultation on EU-level. For this step, contributions from more than 200 RES sector stakeholders across the EU were collected through questionnaires and via an online platform (www.re-frame.eu). The outcome of this step is presented in the section ‘Relative relevance of the CI components’. Derivation of the CI scores on country level. To this end, data were collected from several secondary data sources (see Table 1) and through semi-structured expert interviews in three case study countries (Germany, UK and Spain). The interviews followed a comprehensive interview guideline requesting information about the relevant indicators, their present manifestation as well as the trend over the past three years. A minimum of 10 interviews per case study with an average duration of 1 hour were conducted. Due to volume restrictions of this article, we can only show the results for the German case study (see section ‘Case study results Germany’). Performance of the diffusion analysis based on the previously assessed CI scores and analysis of historical RES diffusion patterns on country level (see methodology in section ‘Diffusion analysis’ and results in section ‘RES diffusion analysis’). Sub-indicators, value ranges and data sources for the composite RES diffusion indicator (CI). Range is based on recommendations given in Cena et al.
16
to lower grid connection time for wind onshore <6 months and the lead time in the worst performing country across the EU which is 33.5 months. Range is based on the spread of grid connection times across EU countries as given in the ‘PV Grid’ database,
30
in the best performing country grid connection permit and connection take 3 weeks, in the worst performing country 50 weeks. Ranges are based on the recommendation given in Cena et al.
16
to lower the share to 1.5 % of the total project cost (incl. hardware cost). The highest reported value across the EU is 5%. Ranges are based on value ranges for development cost (excl. hardware cost) across EU countries as presented in Barth et al.
15
The lowest value reported for commercial and industrial applications in the EU is 2%, the highest value is 98% for commercial applications and 36% for industrial applications Ranges based on the recommendation given in Cena et al.
16
to lower administrative lead times to a maximum of 20 months. The highest reported value across the EU is 58 months. Ranges are based on own interview results. Interviewees considered a duration of <2 months as acceptable and over 12 months as inacceptable. Average value ranges across EU countries presented by PV Grid
30
show a spread between 1 week (for 50 kWp systems) up to 39 weeks (for 2500 kWp systems).
CI for RES diffusion
Components of the CI
The CI is composed of four main determinants: (A) Economic and political framework; (B) Electricity market structure and regulation; (C) Grid infrastructure and regulation; and (D) Administrative procedures for RES-E projects. Each of the determinants is composed of a number of sub-determinants that are again represented by indicators for their quantification. As the indicators have heterogeneous data formats and value ranges, they need to be adjusted to a uniform scale (normalized) to create a comparable data base. To this end, each indicator value is mapped to a score between 0 and 1. Before deriving the overall CI count, the scores for each sub-determinant are added up and again normalized to a range between 0 and 1 implying a maximum CI score of 16 points (for 16 sub-determinants). Table 1 (Table adapted based on Boie et al. 28 ) presents the determinants, indicators, their value ranges and corresponding normalized scores as well as the utilized data sources.
The approach to the definition of the CI framework is described in detail in Boie et al. 28 It is based on a qualitative approach comprising three major steps: a comprehensive review of scientific literature on barriers and drivers for RES deployment to obtain an initial compilation of relevant framework factors; an analysis of the results of interviews and workshops with 70 international stakeholders from the energy and policy sector on the issue of policy design options and framework conditions for RES deployment to consolidate and refine the selection; and conducting a series of moderated expert workshops, each with 6-15 renewable energy experts to discuss the final selection and grouping of the indicator components.
Detailed explanations of each sub-indicator and reasons for its selection are provided in Boie et al. 28
Relative relevance of the CI components
Weights of the CI components.
The weights indicate that all sub‐determinants constitute important factors for RES‐diffusion since no values lower than 5 (“medium relevance”) occur (Note: To ensure that no important determinants were missing in the CI, interviewees during 31 in‐depth follow‐up interviews were asked for their feedback on the approach and whether there are important factors missing from their point of view. The vast majority of interviewees confirmed that the selection of determinants comprehensively covers the relevant decision factors. Also, the short questionnaire included a section for comments and suggestions but none of the respondents suggested additional determinants.). This suggests that each sub‐determinant can become a blocking factor for RES diffusion. Technology differences, although small, exist, e.g. regarding grid access, curtailment, administrative processes and spatial planning, which have a higher weight for wind than for PV (Note: On the basis of the size and the characteristics of the available data sample, especially due to the fact that most stakeholders did not have a clear technology focus but indicated a background in several RES technologies, it is not possible to identify definite technology‐specific differences in the indicator weights. To this end, it would be necessary to conduct additional surveys focusing on more specialized stakeholders. The presented technology‐specific weights are based on subsets of the data with a focus on one technology only. However, for most sub‐determinants the technology differences are not statistically significant. Therefore, the presented differences should be understood as indicative values.). Interestingly, some determinants (e.g. A I) even received higher scores than the remuneration level, which clearly highlights the relevance of such non‐economic factors for RES diffusion. For the calculation of the CI, the weights are used according to equation (6).
Derivation of the RES-diffusion model
Background and modelling approach
The following section presents the derivation of the RES-diffusion model based on the mathematical formulation of technology diffusion processes. According to general diffusion theory and backed by extensive scientific evidence, 32 – 37 the diffusion of new technologies or products in a market frequently follows an s-curve pattern. This implies a steep (nearly exponential) growth rate at the beginning of the diffusion process, followed by linear growth in the mid-term and a saturation of the market (resulting in a relatively slower growth) in the long-term. 37
The mathematical representation of the s-shaped logistic function is given in equation (1) where P(t) refers to the technology penetration over time.
Thereby, the parameter “a” represents the saturation level of the s-curve and can therefore be interpreted as the long-term deployment potential of a technology. The parameter “t0” represents the shift on the time axis and therefore defines the inflection point of the sigmoid. The growth parameter “c” defines the steepness or the growth rate of the curve and thus represents the speed of technology diffusion. Therefore, accelerating or decelerating framework factors for RES diffusion will be reflected in the parameter “c”.
The logistic function shown in equation (1) is the solution of the logistic differential equation given in equation (2), which can again be represented in discrete terms as shown in equation (3).
Consequently, the additional penetration (ΔP) in a year “n+1” is a function of the growth parameter (c) and the long-term potential (a). Assuming undistorted technology diffusion, the growth parameter “c” can be interpreted as the maximum achievable growth if all framework conditions are optimal. Such fully uninhibited diffusion will hardly occur in reality. However, historical RES-diffusion records show periods of classical logistic growth patterns which allow the presumption that during these periods the economic conditions allowed for an attractive rate of return and non-economic barriers were reduced to a minimum (Note: Although certain barriers might be outbalanced by particularly attractive remuneration levels (e.g. high grid connection- or administrative costs or lengthy procedures) other issues cannot be offset by high remuneration. For example, reliability of the support scheme, access to the grid, fair market regulation and availability of sufficient project sites are prerequisites for RES diffusion which cannot be compensated financially. Therefore, we base our model on the assumption that during these periods, non-economic factors did not represent limiting factors for RES diffusion).
Such periods of logistic growth could be observed, for example, for wind energy in Germany between 1990 and 2003 (see Figure 1). Fitting a logistic function to the observed diffusion patterns provides the following values: c = 0.33, t0 = 19. In the period after 2004, however, the growth of the wind energy market in Germany was substantially slower than suggested by the fit of the logistic curve to the earlier observed growth parameter (cf. Figure 1). A possible interpretation of this observation is that several economic and non-economic barriers start to constrain the technology diffusion and diminish the actual growth parameter.
Optimal fit of a logistic curve to the time series of wind energy diffusion in Germany (period of fit 1990–2003).
This phenomenon has already been observed and discussed for a wide range of energy technologies by Lund. 38 Lund shows that the growth parameter “c” often decreases with increasing market penetration. Regarding RES-technologies, one interpretation of this observation is that some constraints and limitations like grid capacity, budget constraints or administrative capacity might become limiting factors once a certain market share of the new technology is reached.
To demonstrate that the growth parameter c is not constant but depends on the point in time or the stage of the diffusion curve, respectively, we have conducted an analysis of c for various fitting intervals. Therefore, we have calculated c based on the historical diffusion of wind onshore and PV in Germany, starting with a fitting period of ten years (1990–2000), which is extended by one year in each subsequent step until the period 1990–2014 is covered. The result of this analysis is shown in Figure 2. Each data point indicates the growth parameter c that has been calculated for the respective period, i.e. covering the observed diffusion from 1990 up to the particular year on the x-axis.
Variation of the growth parameter c with stepwise increase of the fitting period: analysis for historical wind onshore and PV development in Germany starting with the interval 1990–2000, incrementally extended by one year until 1990–2014 is covered.
Depending on the length of the fitting period, we observe a notable variation in the values for c. This phenomenon has previously been shown by Lund 39 who suggested that c(t) might take the form of a power curve (c(t) = a * t-b + c) but could not show clear evidence for the validity of this assumption. However, comparing the varying values for c with the actual deployment of the technologies during the evaluation period (cf. Figure 6 in the Appendix 1) illustrates that when years with high relative growth are included in the fitting period, the c value increases dramatically. This is, for example, the case for PV in 2004 (with > 150% growth compared to 2003) and, to a lesser extent, in 2009/2010 (∼70% growth). For wind onshore the relative capacity additions in the period 2000–2014 are in a more moderate range and more homogenous, thus the c value shows no strong fluctuations.
Based on the above findings, we figure that c is not constant but a function of time c(t). Further, we assume that for each time step of the diffusion process the manifestation of economic and non-economic framework conditions determines the speed of technology deployment and that these framework factors can be captured by a CI. Consequently, we argue that the time-dependent growth parameter cn (i.e. c in a time discrete representation) can be understood as a power function of a time-dependent Composite Indicator (CIn) with an exponent (β) and a constant (α). Thereby, the parameters α and β represent country-specific aspects, which are not covered by the CI and which are assumed to be non-variable (e.g. cultural aspects). These assumptions are reflected in equation (4).
The composite indicator (CIn) explicitly takes account of the temporal changes of the most relevant economic and non-economic framework conditions and their impact on the speed of technology diffusion. As compared to the concept suggested by Lund, 39 the approach proposed here involves substantially higher data requirements based on detailed bottom-up assessments but it also promises a much better representation of the temporal changes of the growth parameter cn.
Based on equation (3) we can estimate a time series of the growth parameter cn using historical penetration data for the technologies:
Figure 3 shows the evolution of the time-dependent growth parameter cn for wind onshore and PV in Germany for the period from 2000 to 2014. In contrast to Figure 2, this graph presents discrete values (cn) for the function c(t) depending on the technology growth in the respective year. In years with a strong growth proportional to the previous year cn shows a sharp increase (e.g. for PV in 2004 with > 150% growth relative to 2003), whereas, cn decreases in periods of declining relative growth.
Evolution of the time dependent discrete growth parameter cn for wind onshore and PV Germany for the period 2000 to 2014.
Based on the calculation of the CI (cf. sections ‘Components of the Composite Indicator’ and ‘Relative relevance of the CI components’) we can then determine the country-specific calibration constants αand β based on a least square fit. When interpreting the composite indicator for a certain year (CIn), we argue that a time delay between the investment decision and the actual installation of RES capacities needs to be considered. This is due to the fact that the barriers measured through bottom-up analysis relate to the moment of the investment decision, whereas the observed growth refers to the date of installation. For wind energy projects we assume a time gap of one year between the investment decision (financial closure of the project) and the date of installation. For PV projects, however, no time gap is assumed, as PV projects typically have much shorter realization time frames than wind projects. Thus it is assumed that they can usually be realized within the same year.
Diffusion analysis
Following the general description of our modelling approach given in the previous section, we are now going to explain the individual stages of the diffusion analysis in more detail. As an overview, the following steps are carried out:
Determination of the saturation level “a” (achievable long-term potential) of a RES technology in a country. Calculation of the time-dependent growth parameter cn. Calculation of the CI based on quantification of sub-determinants and weightings (cf. sections ‘Components of the Composite Indicator’ and ‘Relative relevance of the CI components’). Determination of the calibration parameters αand β based on the assumption of no time delay for PV and a time delay of one year for wind as a typical period between the final investment decision and installation (i.e. for wind onshore the actual growth in 2014 will be calibrated with the CI of the year 2013).
In the following, the individual steps will be discussed in more detail.
Determination of the saturation level a
In our analysis, the saturation level (a) is interpreted as the long-term deployment potential of a technology. Different literature sources exist in this respect (see, e.g. de Vries et al. 39 and Resch et al. 40 and citations therein). For our analysis we use the “Green-X” database of mid-term and long-term RES potentials for the European MS (http://www.green-x.at/) as it is continuously updated and consulted with the MS and based on a broad range of national sources. Therefore, we consider it as appropriate for our purposes. For Germany we base our analysis on a long-term potential of 177 TWh for wind energy and 139 TWh for PV (both until 2050).
Calculation of the time dependent growth parameter cn
Evolution of the time dependent growth parameter cn for wind onshore and PV in Germany.
Calculation of the CI based on the determinants and weightings
The CI is calculated based on a geometric aggregation method (i.e. the product of weighted indicators) of each of the determinant scores and the respective weight (see equation (6)). We apply a geometric approach to derive the overall CI score to take into account that RES diffusion will reduce to nearly zero as soon as the score for one of the main determinants equals zero. This might occur e.g. if the remuneration level is lower than the generation costs or if grid barriers prevent the grid connection of RES projects in the first place. In this case, these factors would become strong barriers for further RES deployment, which could not be offset by high scores for other determinants (as it would be the case with an additive aggregation approach). This method for sub-indicator aggregation is typically used if sub-indicators are not compensatory, meaning that a poor performance with respect to one factor cannot be fully outweighed by a good performance regarding another factor.
27
:p.33,32
Calibration of the residual terms α and ß
The constant calibration factors α and β, which represent those aspects that are not covered by the list of determinants used in this analysis will be determined by calibrating equations (4) and (5) for α and β. Thereby, cn is determined based on the actual diffusion during the years 2012, 2013 and 2014 and 
Case study results Germany
CI scores
Overview over interviewed stakeholders in Germany.
Policy design options and assumed policy risk levels (based on Resch 58 )
Composite indicator scores (weighted) for wind energy and PV in Germany (2012–2014).
It can be seen that most of the framework indicators are stable during the observation period. This shows that the general framework regulating the electricity market (determinants B-I, B-II, B-III) and the grid access (C-I, C-III, C-IV) in Germany is well established and reliable and constitutes a very favourable environment for the development of renewable energies. The legal foundation is provided mainly through the Energy Industry Act (EnWG), 41 which establishes fair competition in the energy market and the Renewable Energy Act (EEG), 42 which represents the basis for the support and market access of RES. The Federal Network Agency, 43 as the regulatory agency, safeguards fair market access and competition.
A positive and stable score can also be observed regarding the revenue risk under the present RES-E support scheme (A-III), as financial support (FIT/ FIP) under the EEG is guaranteed for 20 years 42 and there is a very low revenue risk once an RES-E project became eligible for the support scheme. Also the reliability of the general RES-E strategy and the support scheme itself (A-I) is rated as very high since the EEG has been in place since 2000 44 (with regular amendments in 2004, 45 2009, 46 201247,48 and 2014, 42 and thus provides a reliable basis for RES-E support. Also the overall policy environment in Germany is among the most stable and secure ones in Europe. 49 A slight limitation applies to the score for PV in 2012 as in this year, in addition to the regular amendment of the EEG, 48 an unscheduled reduction of the feed-in tariff was announced, 47 which reduced the attractiveness of the economic framework for PV developers. However, remuneration levels still covered generation costs and allowed for attractive rates of return for investors. Additionally, in 2012 a target range with a deployment cap of 2.5–3.5 MW per year was introduced to allow for a better control of PV development. 47
Also access to finance (A-IV) was evaluated as being very good throughout the whole observation period. This is partly due to the low interest rates for commercial bank loans 50 and the stable and low-risk financial market conditions in Germany51,52 but can also be attributed to the high availability of financial products specifically for RES-E developers. Interviewees consistently stated that German banks are well experienced in financing RES-E projects and that access to capital does not constitute a bottleneck for the development of wind or solar projects.
However, also substantial room for improvement can be observed with regard to both, economic and non-economic parameters. For example, based on interview results and own profitability assessments performed within this study, 53 the remuneration level for RES-E (A-II) becomes a major limiting factor, especially for PV.
The duration of both grid access (C-II) and administrative processes (D-II), as well as the cost (D-I) and complexity (D-III) of administrative procedures, show further room for improvement. Interviewees stated that, especially for wind energy, the cost for administrative procedures might become a limiting factor as requirements related to, e.g. environmental impact assessments and other impact studies, compensatory measures as well as the securities for project dismantling are reported to show a rising trend over the past years. These findings are backed by observations made by Pietrowicz and Quentin 23 :p.39 who analysed onshore wind projects in Germany realised between 2005 and 2014 and recorded a tendency of longer approval periods during this timeframe.
Bottlenecks for both technologies were also identified with regard to the integration of RES in spatial planning (D-IV). Here, interviewees mentioned particularly the exclusion of agricultural areas from the remuneration of PV systems (introduced with the amendment of the EEG in 2009) 46 as problematic. In the case of wind, interviewees mentioned the time delays related to the development of regional spatial development plans as particularly unfavourable and stated that authorities on the regional/communal level sometimes lack the technical background for defining suitable areas for project sites and for processing the project applications. Pietrowicz and Quentin 23 also found that wind projects faced significant delays due to spatial planning issues, particularly when realized in the context of a regional spatial plan. Nevertheless, spatial planning on the regional or communal level was mostly seen as the best solution (as compared to a top-down approach), which should be further developed. However, it was suggested that local authorities should receive more guidelines and support in order to be able to perform this function in a more satisfactory way.
Also the transparency and predictability of the grid development (C-III) did not receive full scores, as the announced grid reinforcement projects in Germany are lagging far behind in their implementation. 54 Although the 2011 Grid Extension Acceleration Act (NABEG) 55 and the 2009 Law on Energy Line Extension (EnLAG) 56 create a favourable and transparent framework for transmission infrastructure development in Germany, it is not clear when the planned transmission capacity will actually be available as many transmission projects face strong public resistance. Therefore, this determinant does not receive full scores.
Based on the findings for CI presented in the previous section, the value of the calibration factors α and β for Germany were determined by a least square fit (see equation (7)) and subject to the constraint given in equation (8). The results are as follows: for wind onshore α = 4.27, β = 12.41; for PV α = 3.85, β = 5.86.
An overview of the historical deployment of wind energy onshore and PV in Germany (1990–2014) and the envisaged future diffusion under the NREAP and the EEG 2014 (until 2020) are presented in Figure 6 in Appendix 1.
RES diffusion analysis
Based on the derivation of the calibration factors α and β, the long-term potential (a) and the CI, the penetration in year n+1 can be derived from the penetration in year n:
For our analysis, we compare the following scenarios with regard to the technology development from 2015 until 2018:
Figure 4 shows the short term diffusion outlook for PV in Germany with regard to the four scenarios defined above. The following observations can be made:
Under BAU-assumptions, growth is expected to lead to 34% exploitation of the long-term potential, electricity generation of 46.5 TWh and an installed capacity of about 51 GW PV in 2018. In this case, the target trajectory of the German NREAP, which foresees a capacity of 44.8 GW PV (35.1 TWh) in 2018,
57
:pp.113–114 would be surpassed. However, the target trajectory based on the EEG 2014,
42
which indicates around 48 GW PV capacity in 2018, would not be met. The assumption of longer administrative procedures leads to a slightly reduced deployment growth after 2014. Generation by PV is diminished by about 1.7 TWh in 2018 as compared to the BAU assumptions. The effect is not very distinct as this determinant received a relatively low weight for PV (cf. Table 2). The assumption of optimized grid development leads to a moderate increase of PV diffusion compared to BAU. This shows that long-term and transparent grid planning procedures are still important for PV although this framework factor does not represent the most relevant constraint for this technology (cf. Table 2). Nevertheless, improved grid development could lead to an increase of PV generation by about 3.1 TWh (54 GW installed capacity) until 2018. Optimized spatial and environmental planning reveals the strongest impact on PV deployment, leading to an increase in PV generation by about 7 TWh (about 59 GW) until 2018, compared to BAU. In particular, the removal of land-use constraints for large-scale PV installations would lead to substantially increased PV diffusion in the frame of our modelling approach. Short-term diffusion outlook for PV in Germany: expected penetration levels (%) and electricity generation (TWh) until 2018 for four scenarios.
Figure 5 shows the short-term diffusion outlook for wind onshore in Germany according to the four scenarios. The results show the following:
Under BAU-assumptions, 48.4% of the long-term potential could be exploited and an expected electricity generation of 86 TWh (52.5 GW installed capacity) could be reached in 2018. Therefore, the indicative target trajectory of the German NREAP, which foresees 68.9 TWh by 2018 (cf. (Federal Government of Germany
57
;pp.113–114)), would be substantially exceeded. Also the target trajectory of the EEG 2014,
42
which indicates installed wind capacities of 48.1 GW in 2018, would be surpassed. Longer administrative procedures lead to significantly reduced technology diffusion after 2014. This is due to the fact that this determinant has a relatively high relevance for wind energy (cf. Table 2) as lengthy administrative procedures constitute a strong risk factor for the realization of investment-intensive wind parks. In this scenario, generation by wind onshore would be reduced by about 14 TWh in 2018, compared to BAU assumptions. The assumption of optimized grid development leads to a substantial increase of wind onshore deployment. This shows that long-term and transparent grid planning procedures are particularly important for wind onshore, both in order to allow the initial connection and to reduce the risk of curtailment during the lifetime of the plant. Improved grid development leads to an increase of wind onshore generation by about 21 TWh until 2018, compared to BAU assumptions. Also in the case of wind onshore, optimized spatial and environmental planning would have the strongest impact on the diffusion path, leading to an increased generation of about 24 TWh until 2018. In particular, overcoming time delays related to the development of regional spatial development plans could lead to substantially accelerated wind onshore diffusion. Short-term diffusion outlook for wind onshore in Germany: Expected penetration levels (%) and electricity generation (TWh) until 2018 for four scenarios.
Conclusion and outlook
We have developed a comprehensive framework for the assessment of economic and non-economic framework factors (determinants) influencing wind onshore and PV diffusion (cf. Boie et al. 28 ). Based on the identification of the major determinants we have developed a CI for RES diffusion which can be applied to cross-country comparisons or benchmarking purposes on EU Member State level.
Based on a large-scale survey among more than 200 RES experts across the EU, we have determined the actual relevance (weight) of each of the determinants in the overall CI. Results emphasize the role of non-economic factors that play a major role for the diffusion of renewable energy technologies, apart from direct economic factors. Particularly important are the stability and reliability of the RES policy framework (median score for relevance of 9 out of 10), as this factor received even higher scores than the actual remuneration level and the revenue risk (both median scores of 8). Also the duration and complexity of administrative and grid connection procedures are highly relevant aspects from the investors' perspective (median scores 6–8) as well as the integration of RES planning with spatial planning. Grid-related aspects received scores between 6 and 8, depending on the RES technology concerned.
Differences in the relevance of the framework factors could be observed between the two RES technologies. For example, the duration and complexity of administrative and grid connection procedures and the integration of RES planning with spatial and environmental planning have a notably higher relevance for onshore wind compared to PV. Also grid access conditions and the regulation for curtailment as well as a transparent grid development were rated higher with respect to wind energy.
The weights of the determinants were implemented in the calculation of the overall CI score which can then be applied to diffusion analysis. We argue that RES technology diffusion over time follows an s-curve pattern (see also literature 32 – 37 ). We developed a model to assess the shape of the diffusion curve based on the analysis of past technology diffusion trends and the CI results.
In a case study for Germany, involving extensive data collection and 11 expert interviews with RES sector stakeholders and project developers, we could show that even slight variations in economic and non-economic framework conditions can lead to significant differences in the future deployment of wind and PV compared to a business as usual (BAU) scenario. For PV, under a BAU-scenario 34% of the long-term deployment potential are exploited until 2018, reaching an installed capacity of about 51 GW. However, in a scenario with optimized grid development or optimized spatial and environmental planning, penetration levels of about 36% and 39%, respectively, can be reached by 2018. A slight change in the duration of administrative procedures only leads to small variations in the 2018 penetration level as this determinant has a relatively low relevance for PV (see Table 2).
For wind energy onshore, BAU-assumptions lead to a further market growth until 48% of the long-term potential are exploited by 2018 (86 TWh). The strongest impact on the projected growth can be observed in a scenario with optimized spatial and environmental planning, which leads to a strong increase of deployment, compared to the BAU scenario. In this case a potential exploitation of 62% and an electricity generation of about 110 TWh will be reached by 2018. These findings correspond well with the interview results, which revealed that spatial planning and the availability of suitable project sites are perceived as one of the major barriers for wind energy diffusion by German RES project developers. Measures in this regard are necessary to further support the diffusion of wind energy in Germany.
Although the presented findings are just preliminary results for one case study, it could be shown that the interplay of various economic and non-economic framework factors has a significant impact on the future growth of RES technologies and that close monitoring of both is required to allow for the optimization of RES policy strategies.
The results of the stakeholder survey and the diffusion analysis have also emphasized the outstanding role of a stable and reliable RES policy framework and the diffusion of best practices especially with regard to the various administrative processes and spatial planning for RES. Regional authorities responsible for RES project authorisation and spatial planning could be further supported by providing best practice guidelines. Also stricter time limits for permit approval were mentioned by many stakeholders as a suitable measure to improve the predictability of the planning procedures and to reduce risks and costs for RES developers.
The methodology presented in this paper contributes to diffusion research by developing a transparent, empirical approach to capture investors' perceptions of major determinants for RES diffusion and by implementing them in a diffusion forecast model. Further, our findings contribute to the debate on RES policy efficiency and the role of non-economic factors for the diffusion of renewable energy technologies.
So far, our approach is limited to a relatively small number of data points, both regarding qualitative data (11 interviews for the German case study) and the quantitative data for the CI, which is only available for three years. This constraint narrows the possibilities for application of econometric analyses. Further analyses with additional data will help to substantiate the modelling approach. Furthermore, case studies in additional countries will be necessary to verify the transferability of the approach and to review the reliability of results for different combinations of characteristics and for extreme cases (this is planned in the frame of further publications). Further research would also be useful regarding the preferences of different investor types as the susceptibility to risks and the preference for certain regulatory settings might vary depending on the stakeholder type. 9
Footnotes
Acknowledgements
The authors gratefully acknowledge the financial support provided by the Intelligent Energy for Europe – Programme for the DIA‐CORE project in the context of which this work has been realised. Also, the authors would like to express they gratitude and appreciation to all interview partners who contributed to the work with their valuable experience and expertise as well as to the anonymous reviewers who helped to improve the quality of this publication through their helpful comments.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: in the frame of the DIA-CORE project which was funded by the Intelligent Energy for Europe - Programme.