Fluctuations in the wind energy supply do not impair acceptance of wind farms

Abstract

With the growing contribution of wind power to the electric energy supply, fluctuation in available wind energy might increasingly become part of public debate. Fluctuation influences energy prices and might lead to acceptance problems. As a first step to gauge potential communication strategies, we investigated in an experimental study if acceptance is influenced by being made aware of fluctuation. Participants either answered questions on acceptance before or after judging fluctuation and predictability for 30 days of wind energy input. To test for potentially biased cognition, we additionally checked if judgments were affected by whether or not it was transparent that the graphs showed wind energy input. Mean acceptance did not differ between the groups. Directing attention to the fluctuation of power generated by wind does not have an impact on acceptance. Additionally, presence or absence of labels on the graphs and thus knowing that the graphs depicted wind energy supply, did not affect fluctuation and predictability ratings.

Introduction

Global warming is strongly determined by the increased concentration of greenhouse gasses in the atmosphere. The natural greenhouse effect keeps earth warm by trapping the heat from the sun and makes the planet habitable for many different species (IPCC, 1990). However, the elevated concentration of greenhouse gasses, due to human activities, facilitate excessive absorption of solar energy and leads to the earth’s surface overheating (Forster, et al., 2008; IPCC, 1990). Anthropogenic climate change, induced by greenhouse gasses, can affect our environment and human well-being in many ways. For instance, the increase of global annual mean temperature led to agricultural drought and flooding (Arnell et al., 2019), negatively affected biodiversity (Nunez et al., 2019), and elevated risk of infectious diseases (Butler, 2018) as well as health impairments (Patz et al., 2014). Anthropogenic climate change will also harm the global economy in the long term, especially in poor countries (Tol, 2018). Phasing out of fossil fuel consumption and consequently reducing the atmospheric concentration of greenhouse gasses can attenuate the global temperature increase (IPCC, 2014). Global CO₂ emissions should reach zero by 2060–2080 in order to adhere to the Paris Climate Agreement (Clémençon, 2016). The major contributor to the increase of annual greenhouse gas emissions between 2000 and 2010 was energy supply, its share made up 47% (IPCC, 2014). Subsequently, renewable energy sources, along with the energy efficiency, are considered to be a key component in addressing this issue, as they have the potential to fulfil the required reduction of emissions (Gielen et al., 2019). In 2018, more than 26% of global electricity was generated by renewable energy and the number of countries relying on renewables is increasing gradually (REN21, 2019).

The crucial role of renewable energy in the reduction of greenhouse gasses has become increasingly apparent (see Gielen et al., 2019 for a review). However, the practical deployment of renewable energy technologies, especially wind energy plants, faces low public acceptance (Devine-Wright, 2007). Therefore, the factors determining this low acceptance should be examined carefully in order to facilitate further development of wind energy. Wide implementation of renewable energy systems strongly depends on public acceptance (Devine-Wright, 2007). Generally, public attitudes toward renewable energy are mainly positive (Liu et al., 2013; Ntanos et al., 2018; Stokes & Warshaw, 2017). According to Eurobarometer (2019), about 92% of respondents rate the implementation of renewable energy systems as important and think that the portion of renewable energy used should increase. Yet, public attitudes and acceptance are not the same for different forms of renewable energy. Citizens have a more positive attitude towards solar energy than to wind energy (Liebe & Dobers, 2019). Despite the acceptance of wind farms in the general sense and acknowledgment of its crucial role in the process of climate change mitigation, the practical deployment of wind plants still faces considerable public resistance, especially, on the local community level (Olson-Hazboun, 2018; Sharpton et al., 2020). This paradoxical attitude is often described by Not In My Backyard (NIMBY) phenomenon. According to NIMBY, some citizens, who generally have positive attitudes toward wind power plants, resist the installation in their vicinity (Larson & Krannich, 2016). However, there is also evidence that almost 75% of U.S. citizens supported the deployment of a wind farm within 25 miles from their home (Ansolabehere & Konisky, 2009).

Apart from the territorial proximity of wind plants, acceptance of wind energy is affected by various factors including early participation of locals in the planning and implementation processes, perceived fairness of decision-making process and its results, expected benefits, trust in policymakers, as well as perceived risks and disadvantages (Aitken, 2010; Bell et al., 2005; Jones & Eiser, 2009; Sonnberger & Ruddat, 2016). Negative effects, associated with wind energy plants, may also shed light on low public acceptance. One of the main reasons for resistance to wind plants is the visibility of wind turbines from the home (Ladenburg, 2014; Pedersen et al., 2009). Strong visual intrusion of wind plants may decrease property values significantly (Jensen et al., 2018; Sunak & Madlener, 2016). Another reason for resistance is noise produced by the turbines (Pedersen et al., 2009). Some scientists argue that low-frequency noise and infrasound coming from the turbines can be harmful to health, whereas many others do not share this view (see Knopper et al., 2014 for a review). Potentially, wind farms can also pose a threat to biodiversity by increasing bird fatalities (Panarella, 2014; Rollins et al., 2012).

Alongside the local effects of wind plants, there are other potentially important factors influencing public acceptance. The reliability of wind energy, for instance, is one of them. Reliability is mainly determined by two characteristics: (a) whether energy generation is consistent over time without large fluctuation and (b) whether it is highly predictable (Rohrig et al., 2013). With a larger contribution of wind energy, fluctuation might have a more dominant effect on the availability of electricity in the grid. Fluctuating availability and low predictability will be mirrored by fluctuation in energy prices at the level of stock exchange. It is therefore conceivable that the successful implementation of wind energy at a larger scale endangers acceptance because fluctuation in supply becomes a relevant factor. Fluctuation in the wind electricity supply due to the high variations of wind speed are quite common (Shafiullah et al., 2013; Watson, 2014). This fact makes wind energy less predictable. One of the main challenges the wide implementation of wind plants faces is fluctuation in energy generation as well as energy demand. There are times, when the amount of generated power and demand may not match—for instance, an increase in seasonal demand for electricity and a decrease in wind speed. Such fluctuation and mismatched demand–supply patterns lead to price variations (Badyda & Dylik, 2017) and can even cause the negative price spikes, for instance, when supply exceeds demand significantly (Brandstätt et al., 2011). To compensate for gaps in wind energy generation, countries fill the demand by using coal-fired power plants (OIST, 2016) or other sources of renewable electricity as a backup (Vivas et al., 2018). Nevertheless, compensating for fluctuation through battery storage and energy control systems seems increasingly conceivable (Katoh et al., 2014).

Although lay people have their opinions and views regarding climate change and renewable energy, they may have limited knowledge and information about the topic (Assefa & Frostell, 2007). Thus, public opposition toward renewable energy and especially toward wind energy may be characterized by a deficiency in knowledge and information regarding the pros and cons of renewable energy sources. In a study by Bidwell (2016), participants who gained more information about wind energy showed higher support for it, compared to the other participants. Therefore, providing additional information about wind energy supply could lead to the reconsideration of attitudes and acceptance. The present study aimed to investigate whether bringing fluctuation of wind energy supply to mind would reduce acceptance. If this was the case, communication strategies highlighting the potential for flexible energy usage by industry and consumers might be warranted in order to prevent stances regarding fluctuation and predictability from threatening acceptance. In order to make sure that participants processed data on fluctuation and predictability, we had them rate these characteristics for 30 data graphs charting the generated wind energy (30 days, from 0:00 to 24:00 with 5 h intervals; see Fig. 1A for labelled groups). We had an additional group of participants (see Fig. 1B) rate the graphs in an unlabelled variant. This allowed us to check whether stances towards wind energy would bias the rating of predictability and fluctuation (cf. Lewandowsky et al., 2016).

Fig. 1

Examples of the graphs presented in each group

Methods

The present study allowed for two different tests with respect to acceptance of wind energy. Group 1 (labelled-pre) reported on wind energy acceptance after rating predictability and fluctuation for 30 data graphs (see Fig. 2). Group 2 (labelled-post) ran through the online experiment in reverse order. By comparing the acceptance of wind energy between these two groups, we could test whether bringing fluctuation of wind energy to mind would have a negative impact on acceptance. Furthermore, we were able to compare the rated predictability and fluctuation for the graphs depending on whether they were labelled or not. This should allow detecting biases in judgments that are driven by the stance towards the domain (cf. Lewandowsky et al., 2016). Participants were randomly assigned to one of the three groups (n1 = 71, n2 = 74, n3 = 73). Groups 1 and 2 received 30 labelled graphs one at a time. Design, content, and layout of graphs were identical and each graph was followed by fluctuation and predictability ratings in both groups. The only difference between these groups was the position of acceptance items. Participants in group 1 received the acceptance items after presenting all 30 graphs. Group 2 received the same acceptance items right before seeing the graphs. All nine items were presented on the same page and the order of the items was fixed in both groups. Group 3 (unlabelled) only received the unlabelled graphs and did not receive the acceptance items. Thus, participants in this group had to make fluctuation and predictability ratings without knowing the domain of the presented graphs. The amount of graphs shown was equal in all three groups. They all depicted the same content, which was the veridical data about the wind energy supply (in Megawatts) and all of them were followed by the same items (fluctuation and predictability ratings).

Fig. 2

Design and procedure of the experiment

Participants

Overall, we recruited 252 participants. Thirty-four were excluded from further analysis. Exclusion criteria were low proficiency in German language, self-reported lack of seriousness, and very short duration (< 300 s) of participation. Data from 218 subjects (74% female) were included in the final analysis (age M = 34.30, SD = 11.90, range 19–73 years).¹ Participants were either native German speakers (85%) or fluent in German (15%). Students received one course credit for their participation in the online survey. The study was approved by the ethics committee of FernUniversität in Hagen.

Material and procedure

The present experiment was conducted entirely online. Participants were randomly allocated to the three experimental conditions. They received the instruction that they would be presented with different graphs to rate predictability and fluctuation. While the instruction for two of our three conditions clarified that the graphs were about daily wind energy production (and the graphs were labelled accordingly), the domain was not disclosed to the third group and the graphs were not labelled. After a short introduction, participants were to give informed consent and answer some demographic questions (age, gender, and proficiency in the German language). Participants received 30 graphs (either labelled or unlabelled) showing fluctuation in wind energy feed-in on daily basis in September 2019 (TenneT TSO, 2019). All graphs were created in Microsoft Excel. If labelled, X-axes of the graph depicted the amount of produced Megawatts and Y-axes depicted the time of the day with 5 h intervals (0:00, 5:00, 10:00, 15:00, 20:00, see Additional file 1).

In the unlabelled condition, graphs were presented without any information about the content of data or axes. Each graph was followed by two items. In the first one, participants had to rate the fluctuation (the fluctuation in the course of the day is very large) of the trend-line on the graph from 1 (strongly disagree) to 5 (strongly agree). In the second one, participants had to rate the predictability of trend fluctuation (the fluctuation in the course of the day is well predictable) from 1 (strongly disagree) to 5 (strongly agree). Participants in the labelled conditions also received nine items measuring the acceptance of wind energy (see Table 1). They were modified version of acceptance items previously used by other studies (Diller et al., 2017; Sonnberger & Ruddat, 2016). Participants had to rate each item statement from 1 (strongly disagree) to 5 (strongly agree). Between the labelled conditions, we varied, whether the acceptance items were provided before or after the task of rating predictability and fluctuation of the 30 wind energy supply graphs. In this task, graphs were presented one-by-one in randomized order. The time participants had to complete the online experiment was not limited. At the end, participants were asked to give final informed consent, declare whether their participation was serious or not, leave comments, share their experience and opinion about the experiment.

Table 1 Acceptance items presented in group 1 (after the graphs) and group 2 (before the graphs)

Results

Table 2 Descriptive statistics for dependent variables

Acceptance of wind energy

As main issue of the analysis, we tested whether prior exposure to data graphs conveying information on fluctuation and (lack of) predictability of wind energy supply would affect acceptance ratings. An independent samples t-test revealed no significant difference in mean acceptance of the group that had reported on acceptance after vs. prior to rating fluctuation and predictability of wind energy supply (group 1 vs. group 2), t(143) = − 0.86, p = .390, d = − 0.14 (see Table 2). According to the Bayesian independent samples t-test (van Doorn et al., 2020), the null hypothesis predicts the data almost four times better (BF₀₁ = 3.98) than the alternative hypothesis (BF₁₀ = 0.25). Thus, there was substantial evidence that bringing fluctuation in wind energy supply to mind did not affect acceptance. We exploratively analysed the distribution of the group means on the single item level (see Fig. 3) to check whether the some acceptance items would show ceiling- or floor-effects. As this was not the case, we could rule out that the robustness of acceptance with respect to information on fluctuation and predictability was due to such a measurement problem.

Fig. 3

Means of acceptance items within each group. Ascending order of items is based on the mean ratings in the group 1. Error bars depict the standard errors in rating of each individual item in one specific group

Fluctuation

We checked how participants processed the data graphs in order to (1) test whether they indeed had attended the material which implemented the experimental manipulation and to (2) check whether labelling the data as related to wind energy influenced judgments. In a first step, we checked whether fluctuation ratings were affected by potential stances towards wind energy. For this, we compared fluctuation ratings across the experimental groups with labelled vs. unlabelled data graphs. Univariate ANOVA showed no significant effect of graph format on fluctuation ratings, F(2, 215) = 1.46, p = .235, $\eta_{p}^{2}$ = .013 (see Table 2). Bayesian ANOVA revealed that predictive strength of the null hypothesis is almost six times higher (BF₀₁ = 5.78) compared to the alternative hypothesis (BF₁₀ = 0.17).

Predictability

An equivalent analysis was performed on predictability ratings. According to the results of univariate ANOVA, groups differed significantly regarding their ratings of predictability of energy supply on daily basis, F(2, 215) = 3.12, p = .046, $\eta_{p}^{2}$ = .028 (see Table 2). As a post hoc t-test (corrected with Games–Howell because the assumption of homogeneity of variances was violated) showed, this effect was carried by the difference between group 1 and group 2 (p = .041). Group 3 (the unlabelled group) did not differ from group 1 or group 2 significantly (ps ≥ .395), which means that predictability ratings were not significantly influenced by the presence or absence of labels on the graphs. Despite the significant main effect revealed by univariate ANOVA, according to the Bayesian ANOVA, data are about 1.3 times more likely to occur under the null hypothesis (BF₀₁ = 1.34) than under the alternative hypothesis (BF₁₀ = 0.75).

Consistency in the ratings

In a final step, we analysed how consistent the ratings across participants were (see Fig. 4). High consistency across participants would suggest that participants worked on the data graphs attentively and judged them based on similar features. To explore consistency, we sorted the mean ratings of fluctuation and predictability for the 30 graphs. Lack of consistency across participants would present a shallow slope (lack of substantial rating differences among the graph that on average was rated lowest vs. highest). If participants were consistent, there should be differences spanning more than a rating point among the low-rated and top-rated graphs. Accordingly, this analysis served as a manipulation check securing that the experimental manipulation of bringing participants in contact with data graphs conveying information on predictability and fluctuation of wind energy supply had indeed been achieved. Figure 4 suggests that the manipulation was successful. The group means of lower-rated vs. higher-rated data graphs spanned a large part of the scale implying that participants showed high consistency in ratings based on processing the data graphs.

Fig. 4

Consistency of fluctuation and predictability ratings across the participants for each individual graph. Error bars depict the standard errors in the ratings of each individual graph in one specific group

Discussion

With the growing share of wind power in the electric energy supply, fluctuation in wind energy supply might increasingly become part of public debate as they influence energy prices. Here we explored whether this might lead to acceptance problems that should be taken into account by developing appropriate communication strategies. Our results suggest that acceptance of wind energy is robust and not perturbed by salient information on fluctuation of wind energy supply. Participants’ acceptance of wind energy plants was not compromised by making information on fluctuation in wind energy salient. Furthermore, mean scores of fluctuation and predictability ratings were not different for groups working on labelled data graphs vs. participants working on unlabelled data graphs (not knowing that the data was about wind energy supply). Thus, potential stances towards wind energy did not affect judgment of fluctuation and predictability (cf. Lewandowski et al., 2016).

Our study contrasts work suggesting that reliability, predictability, and stability of wind energy supply are considered to be important elements of wind energy acceptance (Rohrig et al., 2013; Schubert et al., 2014). Different from prior work we experimentally manipulated the availability of information on predictability and fluctuation. The divergence in results might hint that in other studies reliability, predictability, and stability were used as arguments by citizens to support a negative stance towards wind energy (rather than this actually being the basis of their position). Many studies emphasize that the resistance toward wind energy plants mainly exists on the local level (Hammami et al., 2018; Olson-Hazboun, 2018; Sharpton et al., 2020). For instance, according to the studies, distance from wind turbines, locals’ engagement in the planning and implementation process, perceived benefits, equity, fairness (see Ellis & Ferraro, 2016 for a review), and community co-ownership (Musall & Kuik, 2011) can effectively attenuate resistance. Acceptance of wind farms is further determined by socio-political, environmental, and personal attitudes (Huijts et al., 2012; Kempton et al., 2005). Thus, factors affecting the acceptance of wind power technologies are manifold.

According to Assefa and Frostell (2007), lay people are not well equipped with relevant information about climate change or renewable energy sources. They constantly seek the balanced information (Dütschke & Wesche, 2015). This underlines the importance of studies checking which aspects (do not) lead to acceptance problems. Moreover, the fact that enhancing knowledge about renewable energy can improve acceptance (Liu et al., 2013) may be indirect evidence that uncertainties about this innovative technology underlie the local resistance. Citizens might not feel affected by fluctuation in the supply as other sources of electricity can compensate (Vivas et al., 2018; World Nuclear Association, 2020). Future studies might test whether we can also count on robust levels of acceptance when the costs of this strategy are made transparent. Importantly, fluctuation in supply can place strong incentives on consumers and industry to flexibly shift the use of energy towards peaks in supply. Thus, charting the economic and ecologic potential of flexible energy usage might substantially increase the acceptance of renewable energy sources.

The current work used an experimental approach optimized to secure that differences between conditions (or lack of differences) could be attributed to the impact (or lack of impact) of the independent variable (cf. Shadish et al., 2002): participants were randomly assigned in a blind way. We secured (see Fig. 4) that they attentively processed the data graph material so that we can be sure that they took notice of the high variability in wind energy. Thus, this manipulation check secured that the independent variable was varied successfully. Together with the high statistical power (large sample), this makes us quite confident to conclude that varying information on fluctuation in wind energy has indeed no effect on acceptance. Furthermore, the generalizability of the findings is likely high given that we tested an age- heterogeneous sample of students most of whom are working. The current work can serve as a basis for addressing questions on the acceptance of energy change in studies using panel data from representative samples including more information on the economic and sociodemographic background of the participants.

Conclusion

Overall, results of the present study suggest that showing how fluctuating the wind energy supply can be, does not affect whether or not participants accept wind energy farms. Thus, fluctuation in wind energy supply can be directly addressed without compromising acceptance. As the successful implementation of wind energy systems, which is one of the most promising sources of renewable energy, highly depends on public acceptance (Devine-Wright, 2007), it is crucial for climate change mitigation to investigate factors facilitating acceptance of wind energy.