Spatial Effects of Wind Generation and Its Implication for Wind Farm Investment Decisions in New Zealand.

AuthorWen, Le

    To achieve a low-emission economy transition, the New Zealand Government aims to lift the share of electricity generated from renewable resources from 80% to 90% by 2025. Electricity generation in New Zealand is hydro-dominated, with around 57% of electricity generated by hydro during 2011-2014. Average electricity percentage generated from thermal sources was 21%, geothermal 15%, wind 5% and cogeneration 3% (ENZ, 2016). New Zealand has 34.5 MW of grid-connected solar power (EA-EMI, 2016). Expansion of hydro capacity is limited. On the other hand, New Zealand has a most favourable wind resource with plants operating at around 45% capacity. Given this potential, it is highly likely that wind power could contribute as much as 20% of electricity if the government's target of 90% is to be achieved.

    New Zealand consists of two main islands: North Island and South Island. The transmission grid contains about 250 nodes. Electricity surplus of one island is transferred to the other island by a high-voltage direct current (HVDC) link. Total installed electricity capacity in New Zealand is approximately 10GW. Currently, both electricity generation and retail are open markets. Transmission and distribution are natural monopolies. Five major generators produce 95% of New Zealand's electricity. Each generator offers generation to Transpower, the Independent System Operator (ISO), in the form of offer stacks. Transpower is a state-owned enterprise and owns the National Grid. It ranks offers in order of price and selects the lowest-cost combination that satisfies demand. Using the SPD (Schedule, Pricing and Dispatch) method, nodal prices on the spot market are calculated every half hour. Notably, SPD can't be used to forecast the regional price effects by adding extra wind capacity.

    Currently, there are 19 wind farms with 689 MW of installed capacity; the majority are located in Waikato, Manawatu, Wellington and Southland. Expansion of wind generated electricity has important implications for electricity supply in New Zealand. First, the total capacity of hydro storage is about 4.9 TWh (EA, 2013), which can only meet about 48 days of national demand (average hourly demand of 4263 MWh), making the electricity system vulnerable to periods of dry weather. Hydro reservoirs play a key role in the indirect storage of electricity, and we would expect increased inter-temporal substitution between generation sources, particularly wind, hydro, and thermal plants, compared to markets with a higher proportion of electricity generated from non-storable sources. Second, because international trade in electricity is not cost effective, the market response to sources of low cost wind generated electricity is conditional on the relative marginal cost of alternative sources, such as hydro, geothermal and gas. Third, given the intermittency of wind generated electricity, climatic conditions are significant in determining the merit-order of generation alternatives entering the market. When wind generation is low, base load capacity is typically hydro/coal/geothermal. When wind generation is high, wind displaces higher cost supply alternatives. When more low cost wind generation is added, this shifts the merit-order curve to the right and pushes out the most expensive generators. This results in the reduction of wholesale electricity price at a given level of demand. Wind generated electricity is likely to be quite variable and may require expensive natural gas backup since this ramps up faster than the alternatives. Consequently, the merit order effect (MOE) of wind generation is relatively larger during periods of peak demand. Fourth, increased wind generation at one grid injection node, contingent on hydro storage and demand, we expect to observe a reduction in the wholesale price at neighbouring connected nodes.

    The impact of wind generation on electricity prices via the MOE has been examined in Ontario, Canada (Rivard and Yatchew, 2016), Germany (Sensfu[beta] et al., 2008), Spain (de Miera et al., 2008) and Denmark (Munksgaard and Morthorst, 2008). However, policies in these countries directly support renewable energy sources (Haas et al., 2008). As no subsidies are offered in New Zealand for the promotion of renewable resources, this provides an ideal opportunity for examining the MOE of wind penetration. (1)

    Spatial models have been extensively used in urban and regional science studies, such as, knowledge and innovation (Anselin et al., 1997; Boschma, 2005; Carlino et al., 2007), cities and clustering (Duranton, 2007; Ellison et al., 2010), and labour and land markets (Faggian and McCann, 2009; Mellander et al., 2011). In a spatial setting, the effect of an explanatory variable change at a particular site affects not only that site but also its neighbours (LeSage and Pace, 2009). In this context, the nodal price in one geographic location is affected by the nodal price in neighbouring locations. By establishing the geographic location of wind farms we estimate the spatial impact of wind-generated electricity at adjacent nodes, controlling for competing sources of electricity. Spatial econometric models can be used to forecast direct and indirect regional price reduction effects and explore the economics of developing wind farms at different locations.

    This study attempts to answer a number of important questions: (1) How does an increase of wind penetration influence the nodal price? (2) Is the MOE larger during the peak demand, and smaller during the off-peak demand? (3) Can we use answers to question (1) to predict the regional price reduction for each node and to further explore where to build wind sites?

    Our study contributes to the literature in several ways. First, the primary contribution is to extend the literature by employing spatial econometric methods to examine the MOE of increased wind penetration and its impact on wholesale price at the grid injection point and neighbouring nodes. We construct three spatial weight matrices, and evaluate different spatial models. Among them, we select a spatial fixed effects bias-corrected (Lee and Yu, 2010) Durbin model (SDM). Second, this is the first study to examine the MOE and the hourly MOE in New Zealand. Third, we apply estimation results to forecast regional wholesale price reduction effects and use these to estimate net financial savings at each node. None of the prior studies have examined regional price effects from wind expansion by considering the issue of local geographic spill-overs between nodal price and wind generation. The evidence affords insight into expanding and integrating wind generation into the electricity system. Transferability of the methodology is not limited. Although New Zealand does not import and export electricity, electricity is imported or exported across nodes. Therefore, in a market which does import or export electricity, we apply a spatial methodology to analyse spillover effects. This analysis can be extended from a cross-region study to a cross-country study. This innovative approach can be applied within an electricity system that is influenced by generation or regulatory factors in neighbouring countries.

    The paper is structured as follows. Section 2 provides a brief overview of the related literature. Section 3 describes data. Section 4 develops the econometric framework, and constructs spatial weight matrices. Section 5 presents the empirical results and simulation, and carries out a robustness check. Section 6 concludes this paper.


    The impact of wind generation on electricity prices via the MOE has been examined widely. Empirical research consistently finds a negative impact of wind generation on electricity price. The extent of MOE varies across countries due to country-specific renewable energy sources (RES) policies, market design, trading opportunities across countries, rules governing the system operator, thermal profiles, transmission constraints, and models applied.

    Rivard and Yatchew (2016) studied the Ontario electricity market when integrating renewables into the electricity system and found a 7 CAD/MWh decrease in the competitive hourly market price due to an increase of wind generation from 500 MW to 1500MW. In Germany, grid operators are required by law to buy electricity generated by specified RES at a guaranteed feed-in tariff (FIT). Electricity supply companies must purchase electricity generated by the RES in advance, which reduces purchases from other sources. Consumers pay for the additional cost of the FIT. This arrangement impacts the MOE. Using an agent-based model Sensfu[beta] et al. (2008) found that raising the renewable capacity by 40% led to a 31% increase in the volume of the MOE. The price effect was similar to the impact of wind energy on market prices in Denmark, where reductions in the order of 12-15% were estimated (Morthorst, 2007). Ketterer (2014) examined the effect of wind generation on the level and volatility of electricity price in Germany based on a GARCH model and found that intermittent wind power reduces the price level but increases its volatility. In their study of Texas Zonal markets Woo et al. (2011) found that a 10% increase in the installed capacity of wind generation reduced price by 2% in the non-Western zones and around 9% in the Western zone, but increased price variance by less than 1% in the non-Western zones and 5% in the Western zone.

    Introducing wind into an electricity system tends to reduce spot prices via the MOE. The impact of MOE is greater when the system approaches its capacity limits (e.g. during peak load). With the reduction of installation and capital cost of photovoltaic (PV), many countries have increased the deployment of PV. Recently, more attention is focussed on studies of the MOE of solar (e.g. McConnell et al., 2013 in Australia, Cludius et al., 2014 in Germany, Clo et al., 2015 in Italy, Welisch el al., 2016 in...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT