By Earl J. Ritchie, Lecturer, Department of Construction Management
Everyone talks about wind and solar power becoming cost competitive, but the cost will rise as its share of generation increases and we have to pay more to integrate it into the electrical system. How much it will rise remains the subject of debate.
The cost of electricity from wind and solar energy, as well as other variable sources, has two components: the cost of generation and the cost of integration into the electrical system. As discussed in an earlier post, integration costs are expected to increase disproportionately as the share of wind and solar increases, potentially offsetting the decreasing cost of generation.
The cost of generation alone is fairly well defined. There is some disagreement about the likely extent of future cost reduction but the ranges are relatively narrow. The Bloomberg New Energy Finance estimates of about $40-$50 per megawatt-hour (MWh) are typical.
Source: Bloomberg 2016
As shown below, except for utility scale solar, the rate of cost reduction has slowed in recent years, so estimates for future reductions in wind power and rooftop solar costs may be optimistic. These are levelized costs, estimates of the actual cost of generation. They do not include integration costs and may differ from reported auction costs, which are affected by market conditions and subsidies.
Source: Lazard 2016
The IPCC estimate
As addressed in Section 7.8.2 of the IPCC’s fifth Assessment Report, there are three components of integration cost: (1) balancing costs (originating from the required flexibility to maintain a balance between supply and demand), (2) capacity adequacy costs (due to the need to ensure operation even at peak times of the residual load), and (3) transmission and distribution costs.
The IPCC does not give specific costs at high penetration levels. Their ranges for levels of 20% to 30% penetration are $1-$7 for balancing, $0-$10 for capacity adequacy, and $0-$15 for transmission and distribution. Total range is $1-32.
Even at these levels the integration costs are significant. At an estimated future generation cost of $45, the middle of the IPCC range of integration costs adds 37%. It is generally recognized that the integration cost of variable renewable energy (VRE) penetration above 30% will be higher but is difficult to estimate.
The complexities of integration
Dealing with intermittency must be managed at a continuum of time scales from milliseconds to years. There are costs associated with all timeframes; however, published analyses focus primarily on the longer intervals of balancing and adequacy.
Source: World Bank 2015
Various measures to manage this variation – storage, source mix, overcapacity, demand management, etc. – have differing costs, advantages and disadvantages which can be traded off. This results in a complex situation in which the optimum solution is typically not obvious.
Estimates of integration cost at higher levels vary so widely that it is almost impossible to generalize. Local conditions and design choices significantly affect cost. As a study by the Danish Association of Engineers put it “the design of future 100% renewable energy systems is a very complex process.” An almost infinite number of possible combinations of sources is possible depending upon location, anticipated demand, degree of decarbonization and emphasis on economics.
How future costs are estimated
Both optimization and cost forecasting are done with mathematical models. Significant differences may result from the model used. Some characteristics and weaknesses of the three main classes of model are shown below.
Source: Ueckerdt 2015
Limitations of the models mean that not all aspects of the system can be incorporated in any one model. This may result in overestimates or underestimates. In addition, published studies frequently consider only one aspect, such as the addition of wind power alone.
The limitations and possible sources of error in these studies are normally well understood by the authors, and explained in the original articles. Such caveats rarely reach popular articles quoting the results. There is also deliberate or subconscious bias in the choice of parameters due to the prejudices of the authors.
The variation in estimates
The result of these factors is considerable variation in cost estimates, even when similar systems are being analyzed. Two examples demonstrate the range:
The first estimate below is a model of adding wind energy to an existing grid similar to the European grid. It does not consider externalities, such as renewables mandates, but does include a carbon tax of 20 Euros per ton of CO2. The upper dashed line shows short term costs, and the solid black line long term.
The model shows integration cost equal to generation cost at 40% penetration. That is, the cost doubles. It does not consider possible storage or extending the grid to optimize the system.
Source: Ueckerdt, et al. 2013
A 2016 US study by Lantz, et al., showed a mix of about 42% variable renewable energy to have a net present value cost $59 billion higher than an economically optimized scenario. They did not give a per kilowatt-hour cost, but modeled a modest 3% increase in retail electricity cost in 2050. The authors comment that the cost may be understated because of lack of detail in the model.
Source: Modified from Lantz, et al. 2016
Further examples include the widely publicized papers by DeLucchi and Jacobson, which estimate transmission and storage costs as $20/MWh for 100% variable renewables, and the 2012 NREL study, based on somewhat dated costs, which estimates up to $54/MWh over a fossil fuel dominated scenario for 90% renewables (48% wind and solar). Published scenarios are hotly debated.
The headline cost in such studies cannot be taken at face value. In addition to variances due to choice of model, such obvious influences as assumed fossil fuel prices and future cost reductions in generation methods must be weighed in assessing the estimates. As might be expected, proponents of a particular technology will frequently make assumptions favorable to their preferred energy source.
Other renewables and the social cost of carbon
Some issues not discussed in detail here include the other variable renewables, wave and tide; the dispatchable renewables, hydroelectric, geothermal, and biomass; and the social cost of carbon.
Wave and tide are expected to contribute only a small fraction of future electricity generation. They may be complementary to other forms of variable renewable energy.
Hydroelectric and geothermal can be highly desirable as low carbon, low-cost and dispatchable. Very high renewables penetration has already occurred in areas where these resources are abundant. New Zealand is above 80%; Norway and Iceland are over 90%.
Electricity generated from biomass is dispatchable but creates greenhouse gases at the site of generation. The extent to which this is offset by land use changes and carbon storage of the fuel crops depends upon the generation technology, the type of fuel crop and management of the crop. Estimates of offset are controversial but most calculate net reduction in greenhouse gases compared to fossil fuel generation.
The social cost of carbon (SCC) is not the focus of this article, which concentrates on the actual cost of generation. SCC is speculative, with typically quoted numbers from about $5 per ton of CO2 to $100, although extremes can exceed $1,000. The US government’s 5th percentile to 95th percentile range of the cost in 2020 is from zero to about $180. Obviously, the inclusion of any positive SCC will shift economic analysis toward low carbon sources.
Little effect in the short run
Wind and solar intermittency are not likely to be very costly in the near-term, say to 2030, because most scenarios do not have them reaching high penetration levels by that time. For example, wind and solar are 15% of electricity generation in the Reference Case of the EIA’s 2016 Annual Energy Outlook.
Even the highly publicized German Energiewende (Energy Transformation) has wind and solar currently at 21%, below the level of potential significant cost increase. Intermittency is still being handled by fossil fuels, dispatchable renewables, and exports. Germany’s target for 2030 is 33%.
Source: Burger 2017
Local areas with more ambitious goals will be an interesting test. California has a goal of 50% of retail electricity sales from renewables by 2030. A 2014 analysis by the consulting firm E3 modeled reaching this goal with 43% wind and solar. The report said “This is a much higher penetration of wind and solar energy than has ever been achieved anywhere in the world.” Capital costs under various scenarios ranged from $89 billion to $128 billion in 2012 dollars, with electricity rates increasing between 15% and 30% solely due to the renewables standard. An additional 40% would be due to infrastructure replacement and other factors. The report further says “overgeneration and other integration challenges have a substantial impact of (sic) the total costs for the 50% RPS scenarios.”
Will intermittency costs limit high penetration?
It is clear that there is a cost to managing intermittency and this cost will likely be greater than the decrease in generation cost itself. Actual experience suggests that this cost will be higher than is envisioned in the more optimistic scenarios.
However, cost is not the only consideration. High cost generation may have value where the cost of alternative sources is higher or the match to demand is good. Carbon taxes and renewables mandates will increase the share of renewables, regardless of the underlying economics.
Predictions of whether costs associated with increasing share of variable renewables will outweigh future cost reductions depend upon expectations of both, as well as future costs of storage and other means of dealing with intermittency, all of which are speculative. Storage costs are a topic for another day.