A decentralized model of Wind turbine optimization

Abstract

The decoupling of energy prices from fossil fuel is slowly making its way as investment is poured into renewable energy sources. Small Island Developing States are gaining in both stability and cost from this momentum but face threat from the same unsustainable centralization practices. A decentralized framework is proposed for Small Island Developing States aimed at achieving grid stability and in attracting independent financing mechanisms. This framework is applied from a Wind perspective and to ensure replicability on all types of terrains, and the model is analysed through three case studies: high-rise buildings, flat terrains and Gaussian terrains. This study provides a novel framework and a general solution for Wind farming over different terrain layouts including forbidden regions and complex topography.

Keywords

Decentralization energy production Wind farm genetic algorithm complex terrains Small Island Developing States

Introduction

Small Island Developing States (SIDS) tend to share similar sustainable development challenges, including remoteness, growing population, limited resources, vulnerability to climate change and natural disasters, susceptibility to external distress as well as excessive reliance on international trade. SIDS are often bereft in their development process and require special support from the international community (UNFCCC, 2005). This support comes from various technical, financial and administrative forms and thus tends to increase their dependency on other countries. This approach has been severely criticized as the ‘debt-trap’ diplomacy as developed countries look for vulnerable states with strategic locations (Kehoe, 2018; Yue, 2015). To ensure a sustainable model, external dependency on resources has to be reduced.

In the energy sector, SIDS follow the global trend where coal and gas are slowly making their way towards renewable energy (Figure 1), and this has to do with the global urgencies and the commitments of countries to adopt the Sustainable Development Goals (SDGs) and the Paris Agreement on Climate Change action, which essentially aims to phase out fossil fuels by 2100 and 80% by 2050 (Newman, 2017).

Figure 1.

Global energy and renewable energy forecast (Newman, 2017).

While there have been a number of geopolitical implications of the SDGS and the Paris Agreement, where numerous countries are favouring economic prosperity over the global ecology (Cooper, 2018; Gao, 2016; Nong and Siriwardana, 2018). Harari (2018) argued that this has to do with nationalist philosophies which is not relevant in a globalized world. Benedikter et al. (2016) supported this viewpoint and proposed an increased global agreement for all fossil fuel extractions as it ultimately affects the global ecology. Even though, due to geopolitical forces, countries are seen unsupportive of international sustainable agreements, cities and private organizations are seen to lead the climate action by investing in more sustainable businesses (Lee, 2013). This trend further outlines an increase in divestment from fossil fuel energy sectors in favour of renewables (Figure 2) (Grady-Benson and Sarathy, 2015; Healy and Barry, 2017; Newman, 2017; Trinks et al., 2018).

Figure 2.

Investment in renewable energy over coal and gas (Newman, 2017).

While SIDS are on the front line of climate change, they are not seen to lead the renewable energy movement as they lack the contextual capacity and appropriate mechanisms to do so. Not surprisingly, they are also often the last to resort to renewable energy campaign policies. They often have severe land constraints with complex terrains or not enough land space. All these contribute to their very slow rise to adopt renewable energy.

Of equal importance is to encourage a mixture of renewable energy production and so being from decentralized sources. Both are still in infancy stage in SIDS, as traditional investments are geared towards singular production units and western technology has a current focus on solar energy, which may not be applicable to SIDS with hilly terrains supporting high quakes. This study thus seeks to develop a novel framework to encourage decentralization of energy production with a particular focus on Wind energy in SIDS. An adoptable method to evaluate the potential of Wind energy solutions in SIDS with two typologies is enforced: urban setup with high-rise building and on hilly terrains.

The organization of the article is as follows: section ‘The case of Wind energy for SIDS’ elaborates on the case of Wind energy for SIDS. The methodological framework of the technique employed is discussed in section ‘Methodology’ and section ‘Results’ shows the results and analysis with the conclusions in section ‘Conclusion’.

The case of Wind energy for SIDS

As renewable energy is witnessing an increase in its adoption rate, selected energy types are seen to lead the scoreboard. A popularity trend analysis from 2004 to 2018 (Google Trends, 2018) for five renewable energy types, Solar, Wind, Tidal, hydro and Biomass, highlights the popularity of each with a value ranging from 0 to 100 (Figure 3).

Figure 3.

Popularity trends of renewable energy from 2004 to 2018 (Google Trends, 2018).

As the world adopts a paradigm shift in energy production and consumption, much attention is placed on the solar energy, although other forms such as Wind are right behind. In 2017, REN21 (2018) documented that >70% of generated global energy came from renewable sources. Of these, 55% came from solar photovoltaic (PV), 29% from Wind and 11% was contributed by hydropower. The remaining 5% was shared by other renewable sources. The increasing production capacity from solar is pointed by the amount of capital investment that has been dedicated to this form. For instance, projects such as the Kamuthi Solar Power Station, India (Evans, 2018), Longyangxia Dam Solar Park, China (Proctor, 2017), Enel Villanueva PV Plant (Enel, 2018), Mexico and the Alshagaya Renewable Energy Complex, Kuwait (KUNA, 2018). The investments of those amount to $679 million, $920.84 million (Proctor, 2017), $650 million (Enel, 2018) and 592.418 million (KUNA, 2018), respectively (Evans, 2018). Such investments are credited to technological advancement and cost competitiveness of various components such as storage batteries and solar panels. The availability of substantial amount of land in areas that would be considered derelict and the reliability of the solar irradiation provide an advantage to solar energy generation over its counterparts. Similarly, the ability to generate small-scale energy from rooftops and the, what is considered as seamless, integration into modern building technologies have seen their popularity soar over others (International Renewable Energy Agency (IRENA), 2017).

However, just like solar, Wind energy has increasing potential. In particular, in SIDS, it has a competitive advantage in regard to land since it can be incorporated even in hilly terrains that are synonymous to most SIDs. Similarly, it is reported that the bid prices for Wind power have plummeted in recent years, with 2017 being the best year on record (Motyka et al., 2018). Motyka et al. (2018) reported that Wind energy investment has reached parity in a number of countries such as Australia and Germany among others, and this experience can be multiplied through targeted and contextualized policies in other places. This increasing adoption has seen Australia produce >100% of its energy load from Wind in one occasion, while Germany generates >66% from Wind and solar (REN21, 2018). Countries, especially SIDs, with massive Wind have the potential to duplicate the success in Australia and other countries and by doing so, they could positively utilize the hilly terrains, as well as save on space, which could be utilized for other uses.

In addition, Wind energy, like many other renewable energy resources, does not release any atmospheric emissions while being transformed for the generation of electrical power (Borah et al., 2013). Many researchers have been working on various aspects of Wind energy such as Wind Resource Assessment (WRA) (Chowdhury et al., 2013; Dhunny et al., 2016), site selection (including environmental impacts) (Xu et al., 2013), economic benefits (Yang et al., 2015) and Wind farm layout optimization (WFLO) (Borah et al., 2013).

An agenda for enhanced energy diversification and decentralization

While the transition to renewable energy sources is making its way, there is an equal need to encourage more stable energy networks through decentralized models. Allam (2014) supported a decentralized grid on two levels: (a) in terms of smaller production units in diverse geographical locations and (b) in terms of a mixture of renewable energy sources. The author argued that this will provide an increased stability and resilience while also economically gaining from new cost-effective technologies. This is supported by the ideas of Alexander (2002) and the works of Salingaros (2000, 2005) in regards to fractals, and this provides an increased overview of how to achieve resilience and stability in energy grids through decentralized nodes. The underlying idea put forth in this article is that energy networks must be encouraged towards diversification of energy grids in terms of renewable energy selection and that those must also, along the same line, move towards decentralization. This will provide more flexible emergent geometries that gain in strength from an intricate web of connections (Figures 4 and 5).

Figure 4.

Emergent geometries through increased connections.

Figure 5.

Three different connective networks shown separated into layers

A mixture of energy sources is thus encouraged, which will eventually enforce a stronger energy network through smaller scaled production plants. This supports the model of the previous authors (Allam, 2014, 2017, 2018b; Allam and Newman, 2018a; Salingaros, 2000, 2005). The most salient feature of this model is that a fixed geometry is not imposed as opposed to modernist planning ideologies where planners design on article irrespective of contextual constraints. The proposed model also aligns with decentralization principles by SIDS countries such as Mauritius through their Smart regeneration principles (Allam, 2017, 2018b; Allam and Newman, 2018a, 2018b). An application to this idea is made to the Wind energy cluster in the following sections.

Application to the Wind energy cluster

As the applicability of solar energy through various models is widely covered in the literature, this study focuses on the second most popular renewable energy: Wind energy, to bring a contribution to the literature. This is further supported that SIDS harbour hilly terrains, and current solar models support the use of vast surfaces, which tend to be scarce. As such, to ensure a decentralized model supporting small-scale Wind turbines, the principles of decentralization in the form of smaller power plants are supported and explored.

The optimal Wind turbine positioning has interested numerous researchers for decades now. Soft computing techniques such as Monte Carlo (Marmidis et al., 2008), evolutionary calculation (Ogunjuyigbe et al., 2017), GRASP-VNS algorithm (Yin and Wang, 2012), Particle Swarm algorithm (Pookpunt and Ongsakul, 2013) and genetic algorithm (GA) (Emami and Noghreh, 2010; Grady et al., 2005; Mosetti et al., 1994; Song et al., 2014) have been adapted in energy modelling to precisely map the Wind energy conversion systems. The use of weighted aggregation has been reported in various studies, whereas Pareto ranking has also been used in a number of research works (Rehman and Khan, 2016).

The majority of models, especially those involving soft computing techniques, are limited to simple decision criteria that introduce limitations in the decision process. The optimization techniques are generally slow/unreliable for non-flat terrains. The contribution of this work will therefore address the above-mentioned issues by proposing a novel decision model for Wind farm site selection and WFLO using GA. This article presents an algorithm for optimal Wind turbine position selection in SIDS, while maximizing the Wind farm’s power production and taking into consideration the terrain topography, number of Wind turbines, wake effect and the variation of Wind in direction and intensity.

The algorithm is applied for different terrain complexities including flat terrains. For the flat terrains, starting with a high-rise building inside it, to a Gaussian terrain of height 5 m increasing to 80 m, this study covers extensively the difference between a layout containing a constraint and a layout without. Recently, Guirguis et al. (2017) used a Gradient-based multidisciplinary design. They considered the forbidden regions as boundary in their model.

Methodology

The optimization system for planning and erecting a Wind farm is elaborated in this section. A new approach has been introduced for solving the optimum Wind turbine placement in a Wind farm having forbidden areas for both flat and complex terrains. A novel objective function has been designed to cater to (a) a high-rise building inside a flat terrain and (b) slope and height constraints over three-dimensional (3D) Gaussian terrains.

Optimization algorithm

The Wind farm layout problem is a discrete problem whereby applying classical methods to solve is becoming more complicated due to the many variables that are implicated. GAs, which are search-based algorithms, are based on the natural selection and genetics concepts. Hence, making it more adaptable to find the optimal solution for composite problems such as WFLO without the need to reformulate the evaluation of individual solution candidates (Moorthy et al., 2014). The Wind farm layout problem is a discrete phenomenon, which when solved by classical methods becomes more complex. For instance, if we use Computational Fluid Dynamics (CFD) software, it will be required to solve the complex Reynolds-averaged time-dependent equations, which include the assumptions of many parameters. The effect of an important range of computational parameters will have to be investigated for the required CFD software used, and multiple analytical and statistical tests will be required. These will include the grid dependency including their solution of the computational grid, inlet turbulent kinetic energy profile of the atmospheric boundary layer, the different turbulence models and the order of the discretization schemes (Dhunny, 2017). GAs on the other hand are able to economically unearth the optimal solution for different problems. It simply needs information from an objective function. GA operates primarily on the coding of the parameter set rather than the parameters themselves. The focal point of the search is over an entire population rather than a single point (Grady et al., 2005).

The principle of GA is that there is a pool of possible solutions to any prearranged problem. It is used to optimize an objective function. This is done through recombination and mutation that are similar to biological genetics (based on the Darwinian theory of evolution), which hence bring into being new children while the process is continued over many generations. Each objective function is represented as binary strings that are known as chromosomes. Each chromosome has a particular fitness value that signals its quality and then calculates the fitness of an individual based on its genes, which will then provide fitness value for each individual in a population. Population evolves in generations and they consist of the following three operations: selection, crossover and mutation. The main job of selection is to select fit individuals for mating, and crossover creates new offspring, by taking two chromosomes and exchanging their genes whereby from this process two new individuals are created. This process is repeated to generate a certain limit of offsprings for each case. The newly formed progeny then go through mutation. This process changes some of their genes with low probability. Afterwards, a new generation is formed. This process will continue to mate and produce more apt individuals until a certain convergence criterion is set. It should be noted that though GAs are randomized, they perform much better than any random local search (Konfrst, 2004).

Wake and cost modeling

The assumptions that have been made in this study for flat terrains are same as those of the studies by Emami and Noghreh (2010), Grady et al. (2005) and Mosetti et al. (1994). Therefore, the optimization results are expected to be comparable with those studies. The wake decay model employed here (Figure 6) is similar to that which has been developed by Jensen (1983). The assumptions that were applied in this case are as follows:

Momentum is conserved inside the wake.

The near field behind a Wind turbine is neglected for the case of a single wake, making it probable to model the resulting wake as a turbulent one.

The wake radius is equal to that of the turbine, r_r.

While the wake propagates downstream, the radius of the wake r₁ increases linearly in a proportional manner to the distance, x, as shown in Figure 1.

The Wind speed immediately after passing the turbine’s blade is reduced by 1/3 of its initial strength.

Figure 6.

Schematic of wake model.

Using the Betz theory to find the Wind speed immediately behind the rotor, the following expression has been derived to describe the Wind speed downstream of the turbine

u = u_{0} [1 - \frac{2 a}{1 + α (x / r_{1})}]

(1)

where u₀ is the mean Wind speed, a is the axial induction factor, x is the distance downstream of the turbine, r₁ is the downstream rotor radius and α is the entrainment constant. The downstream radius, r₁, is related to the rotor radius, r_r, by equation (2). C_T, the thrust coefficient, is related to the axial induction factor by $C_{T} = 4 a (1 - a)$ , while the entrainment constant, α, is empirically given in equation (3)

r_{1} = r_{r} \sqrt{\frac{1 - a}{1 - 2 a}}

(2)

α = \frac{0.5}{\ln (z / z_{0})}

(3)

where z is the hub height of the Wind turbine and z₀ is the surface roughness.

In the case of multiple wakes, the kinetic energy of the mixed wake can be assumed to be equal to the sum of the kinetic energy deficits and hence the resulting velocity downstream of N turbines can be calculated by equation (4). This equation was used as input into the model to calculate the resulting velocity

u_{i} = u_{0} [1 - \sqrt{\sum_{i = 1}^{N} {(1 - \frac{u}{u_{0}})}^{2}}]

(4)

The extracted Wind power of the turbine is a function of Wind speed, direction, intensity and probability of occurrence. Aspects of the turbine that affect the extracted power are the hub height, rotor diameter and thrust coefficient. Mosetti et al. (1994) presented a power equation for the turbines (equation (5))

P_{t o t a l} = \sum_{i}^{N} 0.3 u_{i}^{2}

(5)

Results

GA has already been given a lot of attention as a powerful optimization method and has been applied successfully in many hard optimization issues as has been demonstrated in previous sections of this article. Our research has attempted to apply the GA method to the Wind farm layout problem but with several improvements and additions to the WFLO. The case is for the optimization of a Wind farm, having some forbidden zone for Wind turbine placement, at any given site with a certain amount of turbine. The case of 3D buildings in a particular region and the case of hilly terrains with different heights are analysed in respect to their potential of power generation. The flowchart in Figure 7 provides an overview on how the model works. It should be noted that in objective functions, we are calculating the velocity of Wind at each turbine considering its wake effects, height of turbines and so on. By using Wind velocity, the calculation of power from that other objective values can be achieved.

Figure 7.

Proposed model for the GA case.

3D building on flat terrain

With scarcity of land and deforestation, to save environment one cannot expect to build Wind farms on only flat regular terrain without any constraint. So, there is a need to study cases where terrains have different shapes and may not be completely empty of constraints, for example, there may be reservoirs, nature reserve and so on. We call these constrains ‘Forbidden region’, in the Wind farm. So, we can still look for the best configurations of Wind turbines in such terrain but by making the GA to circumvent those forbidden regions. This section will be focused solely on these. Below is the add-on to the existing GA flowchart (Figure 8), which the authors implemented to cater for those particular constraints. A snippet of code is provided in Appendix 1 at the end of this article.

Figure 8.

Proposed model for the identification of Wind turbine in an area with a high-rise building.

A building of 25 m length, 25 m width and 60 m height (same height as the Wind turbine of this study) was introduced in the middle of a flat 10 km by 10 km domain. As the height of the building is the same as that of the Wind turbines, there calls for a boundary on the installation of the turbines due to the resulting wake effects from the building. In this context, we have allowed for a distance of 5H (where H is the height of the building) between the turbines and the building. This has been chosen following the work of Franke (2006) which suggested that the distance beyond the building should be at least five times the height of the building. This is to reduce the wake effects due to the building’s shadow. Following the work of Emami and Noghreh (2010), Grady et al. (2005) and Mosetti et al. (1994), we have catered for the wake effects of the Wind turbines; whereby a minimum of 5D (where D is the rotor diameter) is respected so that wake effects do not have great interferences on each other.

In Figure 9, the case of Wind having multiple directions and variable speeds of 0–5, 6–10, 11–15, 16–20 and 21–25 m/s is illustrated. The fraction of occurrence for each angle at each Wind speed is also shown. This Wind farm configuration showcased an efficiency of 80% and a total power output of 24,560 MWh with a Wind turbine of 850 kW at 60 m.a.g.l.

Figure 9.

Optimal configuration for multiple direction and variable Wind speed case in a flat terrain with a 3D building on the middle: (a) variable direction and variable Wind speed and (b) optimum layout.

Gaussian terrains with and without forbidden conditions

The heights of the Gaussian terrain used ranged from 5 to 80 m. It should be noted that a hub height of the Wind turbine was kept at 60 m throughout in this study. So, when the amplitude of the Gaussian reaches 60 m and above, there should not be any Wind turbines behind the hill. This is because the hill will tower over the turbine behind and hence will have barely any Wind activity, thus futile for Wind farming. So, in this constraint analysis, the height constraint too has been added to the codes. For all cases analysed, the direction and intensities are same as Figure 9(a). GA has been run with a population size of 200 and 600 generation. The equations used to generate those terrains are also shown in Appendix 1. A flowchart (Figure 10) is provided for ease of understanding.

Figure 10.

Proposed model for the identification of best Wind turbine layout for hilly terrains.

For this case, the number of turbines has not been kept fixed; rather the optimum number of Wind turbine has been left to be generated by GA automatically. For the Gaussian (Figure 11), the forbidden region was specified to be a slope of less than 0.5°. It is observed that in the case of a slope constraint (displayed by the red crosses), the model finds the next best location for the Wind layout.

Figure 11.

A Gaussian terrain case layout at different heights, with slope and height constraints.

As can be observed in Table 1, the higher the hill, the fewer the turbine numbers, which is due to the height constraint of the hill. So, when the hill’s height reaches a height of 60 m (turbine hub height), the region behind the hill becomes a forbidden region. Therefore, for this domain size considered, the higher the hill, the fewer the turbine numbers when the constraints are considered. On the contrary, when there is no constraint, that is, no height or slope constraint, then the GA will enable the turbines to be at any optimum locations based from its random theory.

Table 1.

Height of Gaussian with power generated.

Height of Gaussian (m)	No of turbines	Without constraint power (kWh)	No of turbines	With slope constraint power (kWh)
5	12	11,000	15	16,300
10	15	16,800	9	7563
40	20	20,813	8	6733
60	16	14,778	5	6242

Discussion

GA is a well-known method in energy modelling but most of the case studies in academic literature discuss flat terrains and considered only wake effects as their constraints. This research article attempted to amend the traditional GA applications to Wind energy to make it cater to different real case scenarios, thus ensuring more practical applications. A 3D building in a flat terrain was first realized, and the power output analysed for different Wind directions and Wind speeds. An interesting feature is that the Wind rose which was used in this study is not of traditional nature by computer-generated data. In fact, a real-life Wind rose from the Island of Mauritius at a 1-year time span was used. The algorithm computed the height of the 3D building as a focal point to calculate the wake effects. The results for the 3D building in the terrain are thus achieved at a very close to reality scenario. The optimum power output is obtained for a terrain of 10 km by 10 km housing a high-rise building.

This article, through the analysis of a Gaussian terrain, shows how hilly terrains can be made to cater for a Wind farm. We have inserted three constraints to support this case, which adds to the novel of this approach:

A Gaussian terrain was generated with varying height for the full analysis.

Wake effects in rejected slopes greater than 0.5°.

Rejected regions behind hills increasing in height were computed.

It is interesting to note that with a hill of 5 m, no interference was caused if our Wind turbines are set at 60 m. But with a hill at 60 m and Wind turbines of hub 60 m, disturbances, interferences and a significant lack of Wind (as the hill will act a barrier) are observed. Hence, this was computed a forbidden region.

Both of those case studies put emphasis on the fact that we do not need flat terrain to build a Wind farm and that it can be done with high efficiency outputs in hilly terrains. The case of Mauritius, being an SIDS, harbouring a complex hilly terrain, perfectly supports this argument.

The analysis performed in this article supports that Wind energy can indeed be expanded on larger regions and its positioning can be made in smaller scale units, while encouraging the sitting of Wind farms in strategic locations to support maximum yield; both in terms of energy output, just as much as economic capital. By doing this, an enhanced energy grid will be achieved with positive impacting on stability and resilience, as elaborated above in this article. By doing this, various terrain types can be used at optimal capacity with a diverse mixture of energy types, thus rendering a more economically resilient model. This will further encourage the rapid adoption of renewable energy. This rapid adoption can be further catalysed by contextualized policies (Allam, 2018a) and by the use of fiscal mechanisms (Allam et al., 2018; Allam and Newman, 2018a, 2018b) as was successfully done in Mauritius.

Conclusion

This study proposes a new energy production framework that encourages decentralization while enforcing network stability in SIDS towards their sustainable pathway to clean energy. One of the most common issues with small islands is the scarcity of lands, which are most frequently hilly terrains. The model is thus applied to Wind energy to capture the potential of various terrains. Policymakers and investors are often afraid to invest on Wind farming for SIDS due to those constraints even though SIDS has high natural resources. A new approach has been introduced for solving the optimum Wind turbine placement in a Wind farm having forbidden areas for both flat and complex terrains. A novel objective function has been designed to cater to (a) a high-rise building inside a flat terrain and (b) slope and height constraints over 3D Gaussian terrains. This study has demonstrated that even though there are constraints on a terrain, a Wind farm can still be erected at optimal efficiency. The novelty of height constraint in a Wind farm has been introduced in this article, this analysis will hopefully change the concept of introducing Wind farming through small-scale energy production in regions of hilly terrains and/or high-rise buildings.

Footnotes

Appendix 1 Acknowledgements

The authors are thankful to the University of Mauritius and to the Tertiary Education Commission for the support of the first author as a Post-Doctoral Fellow and to Prof. Nikos Salingaros for initially sharing ideas on how to create more sustainable and resilient energy grids based on his understanding of complexity theories.

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) received no financial support for the research, authorship and/or publication of this article.

ORCID iD

AZ Dhunny

References

Alexander

(2002) The Process of Creating Life: An Essay on the Art of Building and the Nature of the Universe. Berkeley, CA: Center for Environmental Structure.

Allam

(2014) Only connect – A renewable energy future for small island states. Available at: https://theecologist.org/2014/sep/05/only-connect-renewable-energy-future-small-island-states

Allam

(2017) Building a conceptual framework for smarting an existing city in Mauritius: The case of Port Louis. Journal of Biourbanism 4: 103–121.

Allam

(2018a) Contextualising the smart city for sustainability and inclusivity. New Design Ideas 2: 124–127.

Allam

(2018b) Redefining the Smart City: Culture, Metabolism and Governance. Case Study of Port Louis, Mauritius. Perth, WA, Australia: School of Design and the Built Environment, Curtin University.

Allam

Newman

(2018a) Economically incentivising smart urban regeneration. Case study of Port Louis, Mauritius. Smart Cities 1: 53–74.

Allam

Newman

(2018b) Redefining the smart city: Culture, metabolism and governance. Smart Cities 1: 4–25.

Allam

Dhunny

Siew

, et al. (2018) Towards smart urban regeneration: Findings of an urban footprint survey in Port Louis, Mauritius. Smart Cities 1: 121–133.

Benedikter

Kühne

Benedikter

, et al. (2016) ‘Keep it in the ground’. The Paris Agreement and the renewal of the energy economy: Toward an alternative future for globalized resource policy? Challenge 59: 205–222.

10.

Borah

Roy

Harinarayana

(2013) Optimization in site selection of wind turbine for energy using fuzzy logic system and GIS – A case study for Gujarat. Open Journal of Optimization 2: 116–122.

11.

Chowdhury

Zhang

Messac

, et al. (2013) Optimizing the arrangement and the selection of turbines for wind farms subject to varying wind conditions. Renewable Energy 52: 273–282.

12.

Cooper

(2018) Governing the global climate commons: The political economy of state and local action, after the U.S. flip-flop on the Paris Agreement. Energy Policy 118: 440–454.

13.

Dhunny

(2017) Analysis of Multi-Level Wind Flow Patterns in Urban Areas and Wind Energy Potential in Mauritius Using Computational Fluid Dynamics (CFD). Réduit, Mauritius: Department of Physics, Faculty of Science, University of Mauritius.

14.

Dhunny

Lollchund

Rughooputh

SDDV

(2016) Validation of CFD in highly complex geography and layout optimization of wind turbine placement in Mauritius. Journal of Renewable Energy 101: 1–9.

15.

Emami

Noghreh

(2010) New approach on optimization in placement of wind turbines within wind farm by genetic algorithms. Renewable Energy 35: 1559–1564.

16.

Enel (2018) Enel Green Power Mexico inaugurates Villanueva, largest solar PV plant in the Americas. Available at: https://www.enelgreenpower.com/media/news/d/2018/03/enel-green-power-mexico-inaugurates-villanueva-largest-solar-pv-plant-in-the-americas

17.

Evans

(2018) The world’s biggest solar power plants. Available at: https://www.power-technology.com/features/the-worlds-biggest-solar-power-plants/

18.

Franke

(2006) Recommendations of the COST action C14 on the use of CFD in predicting pedestrian wind environment. In: Proceedings of the 4th international symposium on computational wind engineering, Yokohama, Japan, 16–19 July.

19.

Gao

(2016) China’s response to climate change issues after Paris Climate Change Conference. Advances in Climate Change Research 7: 235–240.

20.

Google Trends (2018) Popularity trends of renewable energy from 2004 to 2018. Available at: https://trends.google.com/trends/explore?date=all&q=solar%20energy,wind%20energy,tidal%20energy,hydroelectric%20energy,biomass%20energy

21.

Grady

Hussaini

Abdullah

(2005) Placement of wind turbines using genetic algorithms. Renewable Energy 30: 259–270.

22.

Grady-Benson

Sarathy

(2015) Fossil fuel divestment in US higher education: Student-led organising for climate justice. Local Environment 21: 661–681.

23.

Guirguis

Romero

Amon

(2017) Gradient-based multidisciplinary design of wind farms with continuous-variable formulations. Applied Energy 197: 279–291.

24.

Harari

(2018) 21 Lessons for the 21st Century. New York: Penguin Random House.

25.

Healy

Barry

(2017) Politicizing energy justice and energy system transitions: Fossil fuel divestment and a ‘just transition’. Energy Policy 108: 451–459.

26.

International Renewable Energy Agency (IRENA) (2017) IRENA Cost and Competitiveness Indicators: Rooftop Solar PV. Abu Dhabi: IRENA.

27.

Jensen

(1983) A note on wind generator interaction. Available at: http://orbit.dtu.dk/files/55857682/ris_m_2411.pdf

28.

Kehoe

(2018) Secret US warning of China ‘debt trap’ on Australia’s doorstep. Financial Review, 14 May 2018, p. 1.

29.

Konfrst

(2004) Parallel genetic algorithms: Advances, computing trends, applications and perspectives. In: Proceedings of the 18th international parallel and distributed processing symposium, Santa Fe, NM, 26–30 April, p. 162. New York: IEEE.

30.

KUNA (2018) KISR: Shagaya project first step toward energy diversification. Available at: https://www.kuna.net.kw/ArticleDetails.aspx?id=2703316&language=en

31.

Lee

(2013) Global cities and transnational climate change networks. Global Environmental Politics 13: 108–127.

32.

Marmidis

Lazarou

Pyrgioti

(2008) Optimal placement of wind turbines in a wind park using Monte Carlo simulation. Renewable Energy 33: 1455–1460.

33.

Moorthy

Deshmukh

Mukherejee

(2014) New approach for placing wind turbines in a wind farm using genetic algorithm. Wind Engineering 38: 633–642.

34.

Mosetti

Poloni

Diviacco

(1994) Optimization of wind turbine positioning in large windfarms by means of a genetic algorithm. Journal of Wind Engineering and Industrial Aerodynamics 51: 105–116.

35.

Motyka

Slaughter

Amon

(2018) Global Renewable Energy Trends: Solar and Wind Move from Mainstream to Preferred (ed Thomas

Bhat

Khan

, et al.). London: Deloitte Development LLC.

36.

Newman

(2017) Decoupling economic growth from fossil fuels. Modern Economy 8: 791–805.

37.

Nong

Siriwardana

(2018) Effects on the U.S. economy of its proposed withdrawal from the Paris Agreement: A quantitative assessment. Energy 159: 621–629.

38.

Ogunjuyigbe

ASO

Ayodele

Bamgboje

(2017) Optimal placement of wind turbines within a wind farm considering multi-directional wind speed using two-stage genetic algorithm. Frontiers in Energy. Epub ahead of print 22 September. DOI: 10.1007/s11708-018-0514-x.

39.

Pookpunt

Ongsakul

(2013) Optimal placement of wind turbines within wind farm using binary particle swarm optimization with time-varying acceleration coefficients. Renewable Energy 55: 266–276.

40.

Proctor

(2017) China’s renewables strategy shines in massive solar park. Power, 12 January 2017, p. 3.

41.

Rehman

Khan

(2016) Fuzzy logic based multi-criteria wind turbine selection strategy – A case study of Qassim, Saudi Arabia. Energies 9: 872.

42.

REN21 (2018) Renewables 2018 Global Status Report. Paris: REN21.

43.

Salingaros

(2000) The structure of pattern languages. Architectural Research Quarterly 4: 149–162.

44.

Salingaros

(2005) Principles of Urban Structure. Amsterdam: Techne Press.

45.

Song

Chen

, et al. (2014) Optimization of wind farm micro-siting for complex terrain using greedy algorithm. Energy 67: 454–459.

46.

Trinks

Scholtens

Mulder

, et al. (2018) Fossil fuel divestment and portfolio performance. Ecological Economics 146: 740–748.

47.

UNFCCC (2005) Climate Change, Small Island Developing States. Bonn: Climate Change Secretariat (UNFCCC).

48.

Yang

, et al. (2013) A Research on Wind Farm Micro-Sitting Optimization in Complex Terrain. Kgs. Lyngby: Technical University of Denmark (DTU), pp. 669–679.

49.

Yang

Zhang

Sun

, et al. (2015) Optimal wind turbines micrositing in onshore wind farms using fuzzy genetic algorithm. Mathematical Problems in Engineering 2015: 324203-1–324203-9.

50.

Yin

P-Y

Wang

T-Y

(2012) A GRASP-VNS algorithm for optimal wind-turbine placement in wind farms. Renewable Energy 48: 489–498.

51.

Yue

(2015) Beyond dependency: The promise of Confucianism in Post-Westphalia International relations. Bandung: Journal of the Global South 2: 4.