Lydia Powell and Akhilesh Sati, Observer Research Foundation
Not many will associate Karl Marx with energy but a close reading of his ‘Capital: A Critique of Political Economy’ that was first published in 1867 shows that Marx understood energy better than many do today. He observed that the energy sources powering industrialisation had to be ‘dependable, urban and completely under the control of man’. Dismissing the ‘horse’ as the worst form of energy he observed that the horse had a head of its own, was costly to maintain and was limited in factory applications’. He also dismissed wind because it was ‘inconsistent and uncontrollable’. He had more charitable views on the kinetic energy of flowing water but he noted that ‘it could not be controlled at will, failed at certain seasons and was essentially local’. Marx’s vote was for coal (with water in the steam turbine of Watt) which he said was ‘entirely under the control of man, mobile and a means of locomotion, and also urban unlike wind and water that were scattered up and down the countryside’. Marx did not dwell on the nature of energy because his mission different (to show how capital would use energy to marginalise labour) but his observations on characteristics of energy such as ‘certainty’ ‘mobility’ and ‘controllability’ that would make certain sources of energy indispensible for industrialisation were accurate.
The western world that has completed the process of industrialisation using dependable and mobile fossil fuels has moved up the hierarchy of characteristics that it wants in its energy sources. Energy sources now have to be ‘clean and green’ apart from being ‘urban, mobile and completely under the control of man’. The pursuit of ‘clean and green’ values in energy sources takes us back to energy sources such as wind and sun that were dismissed by Marx as ‘uncontrollable and undependable’.
This brings us to Elon Musk, the CEO of Tesla Motors who is hailed as the messiah of modern energy sources. Tesla’s Powerwall battery system launched recently is expected to overcome the ‘undependable and uncontrollable’ characterises of wind and solar power and ‘fundamentally change the way the world uses energy’ as Musk put it at the launch of Powerwall in January 2015. Though the euphoria that surrounded the launch of Powerwall has subdued as many have understood that the fundamental change that Musk was talking about would come only at a cost, it has sparked the hope that the day the world will be running on solar energy is much closer than we think.
Solar energy is already urban (as opposed to being scattered up and down the countryside) because photovoltaic panels can convert sunlight into electricity from urban roof tops. The hope is that Musk’s lithium-ion battery will take care of not just the ‘uncertain and uncontrollable’ nature of solar energy but also make it mobile. Uncertain solar energy will be stored in Musk’s efficient batteries and allow people to draw energy whenever they want for whatever they want to do (use electrical appliances or move around in a vehicle) even when the sun is not shining. The question now is ‘at what cost?’
The figures available for the cost of power from Musk’s Powerwall varies but it is nowhere close to the tariff of grid based electricity. The hope is that the cost of power from solar panels and batteries will decline rapidly and out-compete grid based electricity. This brings us to Gorden Moore, co-founder of Intel who famously said in 1965 that the circuit density of semiconductors (made of high grade silicon) will double every eighteen months. Moore’s law as it has come to be known has proved to be true in the micro-chip industry. The number of transistors on a circuit has doubled almost every two years and the cost has fallen dramatically. Many of those who are betting on solar energy believe that Moore’s law is applicable to PV panels (made from solar grade silicon) and storage batteries and that dependable and controllable electricity from these systems will be cheaper than grid based electricity (derived from fossil fuels) in a matter of few years.
PV System Prices for 2013 (US$/W)- Selected Countries
Source: IRENA, Renewable Power Gen. Costs in 2014
Solar photovoltaic (PV) modules and the inverters required to convert the DC power output from PV systems to AC power are commodity products that are traded internationally. The price of PV modules has fallen from about $76/watt in 1977 to about $0.30/watt in 2015 according to Bloomberg new energy finance which amounts to a 13% compounded annual average fall over the 38 year period. This is nowhere near a Moore scale decline but impressive. As pointed out by a recent report on solar energy by MIT, most of the cost declines are on account of lower input material cost (solar grade silicon) and on account of increased scale of production (economies of scale), lower labour costs through manufacturing automation and lower waste from efficient processing. In other words the cost declines of PV modules are the result of production experience and not the result of better grasp of fundamental physics that is required if we want Moore’s law to work.
Indian Imports of Solar Panels
Trends in Global Average Solar PV Module (c-Si) Selling Price US$/W
Source: IEA 2014, Technology Roadmap-Solar Photovoltaic Energy.
In the last fifty years, the power of a given sized microchip has increased by a factor of over a billion but the power output of a solar panel has merely doubled. This is not because of insufficient investment in research and development of solar technology. The United States poured money into solar technology in the late 1940s when domestic reserves of oil began to decline. It increased support for research on alternative energy technologies after the oil crises of the 1970s. Though the enthusiasm for alternative energy sources generally waned when oil prices fell, ideas such as peak oil, the oil weapon (in the hands of Arab nations) etc have kept up the support for alternative energy sources. Despite this solar has not managed to make a break-through on the scale of micro-chips because of fundamental technical limitations of crystalline silicon.
There are inherent technical limitations in using crystalline silicon to convert electromagnetic radiation from the sun (light) into electricity. This needs to be explored further. The physics behind solar cells is complex but a brief outline of the technology is necessary to grasp the potential of solar energy and to answer the question posed in the title of the article.
Silicon is converted into solar cells in well established industrial processes that draw on sixty years of semiconductor processing for integrated circuits. There are single crystalline and multi crystalline solar cells. Higher the crystalline quality higher is the efficiency of charge extraction and power conversion but also higher cost. Single crystalline silicon cells have recorded lab level efficiencies of about 25.6 % while multi crystalline cells have recorded efficiencies of about 20.4 %. Single crystalline cells account for 35 % of the market while multi crystalline cells account for 55 %. Crystalline silicon accounts for 90 % of global photovoltaic (PV) production and it is expected to continue its dominance for the foreseeable future.
In order to gain some understanding of how PV cells work, it is necessary to understand the concept of band gaps within atoms of materials. In pure materials, electrons can only reside in certain discrete energy bands. The electronic properties of materials are dependent on the profile of these energy bands and gaps between these energy bands. In semiconductors such as crystalline silicon, the band gap is somewhere between the high band gap of insulators (materials that do not conduct electricity) and overlapping bands of conductors (materials that conduct electricity). To be precise, the band gap of semiconductors such as silicon is too large for it to conduct electricity (allow movement of electrons from one band to another) in their normal state (in the absence of additional energy in the form of light/heat) but small enough for it to conduct electricity when additional energy from sunlight is available for absorption. A solar cell can only absorb photons (light) with energy gap greater than the band gap. The band gap energy is the maximum energy that can be extracted as electrical energy from each photon that is absorbed by the solar cell. One fundamental limitation of crystalline silicon is its indirect band gap (which involves a change in energy and a change in momentum) which leads to weak light absorption and consequently makes thick wafers a necessity. This translates into higher capital costs, low power to weight ratios and constraints on module flexibility and design. Alternatives to silicon wafers such as gallium arsenide, a compound with a direct band gap (only involves a change in energy) are being investigated but those in the field do not see commercially viable alternative to silicon emerging within the next decade.
Thin film PV technologies that are made by additive fabrication process reduce material usage and capital expenditure account for 10% of global PV production capacity. Commercial thin films use hydrogenated amorphous silicon (non crystalline silicon), cadmium telluride and copper indium gallium diselenide. These materials absorb light 10-100 times more efficiently than silicon. This property reduces thickness of material required for light absorption to just a layer of film coated on a support material such as glass. Cadmium telluride is the leading thin PV technology on account of its ability to harvest solar energy with a direct band gap of 1.45 electron volt (eV) compared to the indirect band gap of 1.12 eV for crystalline silicon. Thin film PV technologies use 10 to 1000 times less material than crystalline silicon reducing cell weight per unit area and increasing power output per unit weight. A key disadvantage of commercial thin film technologies is their low average efficiency typically 12-15 % compared to 15-20% for crystalline silicon. Another key problem with thin film technologies is that they often require scarce elements that cannot be replaced easily. This puts a limit on scaling up solar capacity that is dependent on thin films.
Irrespective of which material is used, improvements in efficiency of industrial processes are likely to bring down costs significantly in the future but this alone will not guarantee the success of solar energy. If solar seeks to displace other fuels that dominate electricity generation today, it needs to perform on other measures as well.
Indicators of Competitiveness with Grid Electricity
Many in India (and elsewhere) have been led to believe by the media that the levelised cost of electricity (LCOE) captured in the tariff of solar electricity is the only indicator that one needs to watch to establish the competitiveness of solar electricity. For example, the bid price of Rs 5.05/kWh of solar generated electricity in Madhya Pradesh was captured in headlines that said that solar power will be cheaper than thermal power in 2-3 years. This is not necessarily true. Even if solar tariff bids come down to Rs 3/kWh it will not mean that solar electricity is cheaper than thermal power.
LCOE is defined as the rate per kWh of electricity that implies the same discounted present value as the stream of costs. The discount rate used generally varies with project type. Put another way, the LCOE is the minimum price a generator would have to receive for every kWh of electricity output in order to cover the costs of producing this power, including the minimum profit required on the generator’s investment. Many solar projects are financed using a power purchase agreement (PPA) sold to a utility. The PPA shifts risk from the power generator to the power purchaser. One of the many limitations of the LCOE is that it implicitly values all kilowatt hours of electricity generated to be the same regardless of when they are generated. Another limitation is that the LCOE does not reflect the project’s ability to provide capacity to meet uncertain demand (or ramping up capability). Even if solar capacity increases dramatically in India the need for dispatchable capacity will not be reduced significantly. Capacity credit defined as the solar energy capacity that can be confidently relied upon at times of high demand is low for solar energy. In EU the capacity credit for wind is estimated at 5-10% for wind and 0-5% for solar power.
Some of the problems in using levelised cost of electricity (LCOE) to establish competitiveness of electricity generated by solar panels were discussed. The concluding part of the article will continue with the discussion.
One of the problems with LCOE is that it does not capture certain system costs. Grid connected solar PV units offer price bids at their marginal cost of production which is zero and receive marginal system price each hour. With zero marginal cost of production grid connected PV systems operating at the retail end (household and building rooftops) displace conventional generation with higher marginal cost at the wholesale end. The wholesale and retail markets for electricity follow different dynamics. In general the wholesale market is not subject to too many policy interventions. Even in India which has a regulated structure the government does not intervene in the wholesale market as much as it does on the retail market. The only exception is when the government mandates purchase of electricity from certain sources (such as renewables) which distort merit order dispatch. But governments actively intervene at the retail end of the market even in industrialised countries. In India the intervention is near total at the retail end.
Source: MIT report on The Future of Solar Energy
The central and state governments intervene to promote polices for energy access and electoral popularity. In industrialised nations wholesale price of electricity is lower than the retail price of electricity as it should be in a reasonably well functioning market. In India retail price for electricity is lower than the wholesale price on account of numerous policy interventions. This explains the financial distress of distribution companies which have to buy power at a higher price and sell at a lower price. Distributed solar generation such as solar PV competes at the retail level which is the most attractive for solar PV because it is the end that is most visible and subject to policy interventions but it conceals the problems that arise out of its variability and imperfect predictability that affect the wholesale end.
Variability and imperfect predictability (qualities that allowed energy sources similar to solar energy such as wind and water to be displaced by coal based steam generation during the industrial revolution according to Marx) of solar PV systems requires counterbalancing from thermal plants that have to be cycled (switched on and off) frequently as PV output varies. This increases the wear and tear of conventional thermal generation and reduces their efficiency. The cost of using back up capacity that has to suffer loss in efficiency and wear and tear on account of frequent cycling is not reflected in the LCOE. At high levels of PV penetration, the average cost of conventional plants (subject to frequent cycling) increases significantly. This in turn increases system costs that have to be borne by the rate payer or the tax payer. Many roof top PV users (in industrialised nations) gleefully declare that solar power has not only slashed their power bills but has also contributed to a net income in certain periods because they are unaware of these hidden subsidies.
The argument that distributed systems save on transmission costs is made to favour decentralised solar power. This is not entirely true. Studies have found that the savings from transmission losses are far lower than the system investment required for solar. In India as transmission and distribution losses are high decentralised energy systems may be favoured. However at a broader system level this is not necessarily an efficient outcome.
Table: Estimated Levelised Cost of Electricity for New Generation Resources in 2019
2012$ per MWh
Source: MIT report on The Future of Solar Energy
The MIT report that provides the basis for many of the arguments in this paper concludes that in all cases analysed by it (within USA) the per kWh costs of residential generation were just over 170 percent of the estimated cost of utility scale generation. In this light, if residential PV systems are growing faster than utility scale systems, it means that they receive a much higher per kWh subsidy than utility scale systems. A rupee subsidy for residential solar PV generation is likely to produce less solar electricity than the same subsidy given to larger utility scale investments. The government of Delhi which has come up with a roof top PV policy must take note of this. Subsidies will reduce the private costs of solar but larger social and economic benefits are likely to be insignificant at best and non-existent at worst.
Yet another concern is the hidden cross subsidies in hybrid systems that combine solar PV with conventional generation. Distribution companies that supply conventional grid based power recover distribution network costs through per kWh charges on electricity consumed. Owners of distributed grid connected PV generators use these distribution networks but shift network costs including added costs of accommodating significant PV generation to other network users. These ‘cost-shift’ subsidises users of distributed PV but it is rarely acknowledged. This subsidy raises the issue of fairness.
Current economics of PV will allow only affluent households (and PSUs that receive an explicit subsidy under current Indian policy framework) to invest in grid connected solar PV generation even if it is only to signal commitment to green values. On the other hand poor households and small businesses are likely to cling to cheap grid based power because (a) grid based electricity is cheaper and more reliable (b) they involve no transaction costs such as effort to install, repair and maintain equipment from the part of the owner (c) they do not have roof tops and (d) they cannot afford the upfront costs of PV systems. But when the less affluent are part of a hybrid grid system that privileges the use of grid connected PV they will essentially end up subsidising rich PV users on distribution network costs. This raises a serious question of fairness in public policy especially in India where the poor outnumber the rich by a huge margin.
If the affluent users of PV systems choose to exit the grid entirely it would eliminate the subsidy but it will also reduce the number of high value customers for the grid. The rich who exit the grid will have to invest in storage and backup systems such as Musk’s Powerwall but economics does not favour this option. In the USA which has a whole sale and retail market for electricity, a grid connected PV user will be better off selling PV electricity to the grid during the day and buying it back at one third the price in the night rather than using Powerwall’s stored supply at night (or any other time when the sun is not shining) at four times off-peak rates or twice peak rates for grid electricity. Regulations that mandate the dispatch of solar (a move being considered in India to promote solar PV generation) can lead to increased system operating costs (as it would intervene in merit order dispatch) and also give rise to problems in maintaining system reliability.
Another curious phenomenon that will affect the economics of grid connected PV is scale. In competitive wholesale markets for electricity the market value of the output falls as PV penetration increases. Simple economics tells us that increasing zero marginal cost solar PV generation during periods of high solar insolation will drive prices down thus reducing profitability of solar generators. This price reduction at peak hours has already been demonstrated in Germany. The more PV capacity is online the less value an increment of PV generation will produce. This means that PV costs for new PV installations have to keep falling to keep it competitive.
This phenomenon need not worry Indian policy makers now as India does not have a competitive market for electricity. Nor does India have time of the day pricing that will value electricity more during peak hours than off-peak hours. This suits solar PV prices conveyed in terms of LCOE because LCOE attributes the same value to electricity generated at different times. Ironically the absence of a competitive market for electricity in India has increased the perceived value of PV electricity and hence the profitability of PV generators. What the popular media is hailing as the success of solar in defeating coal is in fact the success (profitability) of solar energy companies. Indian policy is yet to distinguish between the two – the success of solar energy and the success of solar energy companies.
The MIT report quoted earlier observes that incremental solar capacity without storage may have little or no impact on total required non-solar capacity, especially in systems where peak load occurs at a time for low or no solar insolation. At an all India level system peak demand occurs during late afternoon or early evening when there is no solar insolation. Exceptions to this pattern are Delhi with peak air conditioner load in the middle of the day (in summer) and late afternoon and Maharashtra where industrial demand peaks during the day. Even if these loads are met with solar electricity they would have to invest in back up as a mere cloud passing for a few minutes could reduce solar generation by an order of magnitude. If the expensive 100 GW solar energy capacity that India has planned is unlikely to displace thermal power generation capacity required for the future, then on has to ask if this is a wise decision.
Overall policy makers must keep in mind that solar energy remains a value choice and not an economic choice as it is made out to be. According to the MIT report, at current gas prices in the USA using solar energy to generate electricity is more expensive than using combined cycle gas generation even at a carbon price of $38/tonne of carbon. In India coal is far cheaper as acknowledged by the government. The report also says that it may be more cost effective not to use all available zero variable cost production rather than force a coal plant to stop operating only to start it after a few hours.
Moreover policies that are designed to subsidise solar power generation may lead to inefficient and costly operational decisions in the short term and more inefficient generation mix in the longer term. Policies such as Renewable Portfolio Standards (RPS) that are state specific will locate solar PV generators at sub-optimal locations. This means that India should focus on a national policy framework. Policies that restrict international trade in PV modules and related system components in order to aid domestic industry in the name of ‘making in India’ may also raise the cost of using solar to reduce carbon emissions. Furthermore there is no evidence to show that manufacturing renewable leads to higher job creation than other industries.
Returning to the question posed in the title, solar energy is a long way from defying Marx and yielding to Musk even with some assistance from Moore. Even the Indian solar policies that have been put on steroids are unlikely to help in defying Marx!
Views are those of the authors
Authors can be contacted at firstname.lastname@example.org, email@example.com
Courtesy: Energy News Monitor | Volume XII; Issue 13
Courtesy: Energy News Monitor | Volume XII; Issue 14
Courtesy: Energy News Monitor | Volume XII; Issue 18