Tuesday, April 28, 2015

Can India leapfrog into decentralised energy?

India woke up to telecommunications through the reforms of the late 1990s: the power of DOT was curtailed, VSNL was privatised, private and foreign companies were permitted, new methods of working were permitted. At the time, wired lines were mainstream and wireless communications was novel. However, setting up wire lines in India is very hard. India leapfrogged, and jumped into the mobile revolution for both voice and data. The concept of not having a land line at home was exotic in the US when it was normal in India. In similar fashion, India was an early adopter of electronic order matching for financial trading, and of second generation pension reforms: these things became mainstream in the world after they were done in India.

Could similar leapfrogging take place in the field of electricity? An important milestone in this story will come about with the announcement by Tesla Motors on Thursday the 30th of April, 2015.

The problem of electricity, worldwide

Electricity consumption fluctuates quite a bit within the day. More electricity is purchased when establishments are open (i.e. daytime), when it's too hot or too cold, and when humans are awake in the dark. The electricity system has to adjust its production to ensure that instantaneous consumption equals instantaneous generation.

If producers are inflexible and consumers are inflexible then generation will not equal consumption. The puzzle lies in creating mechanisms through which both sides adjust to the problems of the other in a way that minimises costs at a system level.

For producers, it is not easy to continually modify production to cater to changing demand. The two most important technologies -- coal and nuclear -- are most efficient in large scale plants which run round the clock. It may take as much as a day to switch off, or switch on, a plant. These plants are used to produce the base load': the amount of electricity that is required in the deep of the night. Other technologies and modified plant designs are required to achieve flexibility of production within the day. This flexibility comes at a cost. Suppose the lowest demand of the day is $L$ and the highest is $H$. For the electricity system as a whole, a given level of average production is costlier when $H/L$ is higher. The cheapest electricity system is one where $H/L=1$; this runs base load all the time.

Matters have been made more complicated by renewables. Solar energy is only available when it's light, while peak demand of the day is generally in the late evening. Electricity generation from windmills is variable. Further, the planning and despatch management of the grid is made complicated when there is small scale production taking place at thousands of locations, as opposed to the few big generation plants of the old days.

There are thus a large number of decisions: how to produce, how much to produce and when, how much to consume and when. Economic efficiency is achieved by putting a market in between buyers and sellers, where the price of spot electricity continuously fluctuates. The electricity industry, organised around this price, becomes a self-organising system where a large number of players make uncoordinated decisions about how much and when to consume, how to produce, and how much and when to produce. The price in this market is the summary statistic of the problem of electricity, worldwide' as articulated above. Here's an example (source) of the key patterns, from PJM Interconnect, the biggest power market of the world:

 Figure 1: Demand and price of electricity at the PJM Interconnection

The orange line shows consumption. This was lowest on Saturday night at around 70 GW. It peaked in the evening of Thursday at around 160 GW. This was $H/L> 2$! This gave huge fluctuations in the price, which is the blue line in the graph above. Base load production has no flexibility and was probably configured at 70 GW. When demand was 70 GW, the price was near zero, given the inelasticity of base load production. The price went all the way up to 450 \$/MWh at the Thursday peak. From the viewpoint of both consumers and producers, these massive price fluctuations beg the question: How can we do things differently in order to fare better? The question for consumers is: How can purchase of electricity from the grid be moved from peak time to off-peak time? The question for producers is: How can more production be achieved at peak time? Unique features in India All this is true of electricity worldwide. Turning to India, there are two key differences. The first issue is that ubiquitous and reliable electricity from the grid has not been achieved. The mains power supply in India is unreliable. The euphemism `intermittent supply' is used in describing the electricity supplied by the grid in India. Households and firms are incurring significant expenses in dealing with intermittent supply (example). Intermittent power imposes costs including batteries, inverters, down time, burned out equipment, diesel generators, diesel, etc. Diesel generation seems to come at a cost of \$0.45/kWh. When power can be purchased from the grid, it isn't cheap, as a few buyers are cross-subsidising many others.

In large parts of India, the grid has just not been built out. There are numerous places where it would be very costly to scale out the conventional grid. There are places in India where calculations show that a large diesel generator in a village has strengths over the centralised system. There are small towns in Uttar Pradesh where private persons have illegally installed large generators and are selling electricity through the (non-functioning) grid, in connivance with the local utility staff.

Global discussions of energy systems talk about base load and peak load. In India, the existing generation capacity is not adequate even at base load! The apparent $H/L$ in the data is wrong; demand at the peak is much greater than $H$ -- we just get power cuts. Every little addition to capacity helps. There has been a large scale policy failure on the main energy system. Perhaps more decentralised solutions can help solve problems by being more immune to the mistakes of policy makers.

The second interesting difference is high insolation with high predictability of sunlight.

Compare the insolation in Europe:

 Figure 2: Insolation in Europe

against that in India:

 Figure 3: Insolation in India

(source for these maps). Note that the deep blue for the European map is 800 kWh/m$^2$ while for the Indian map, the same deep blue is 1250 kWh/m$^2$. Arunachal Pradesh and Sikkim get more sunlight than Scotland.

Innovations in renewables

Substantial technological progress is taking place in wind and in solar photovoltaics (SPV).

Wind energy is enjoying incremental gains through maturation of engineering, and also the gains from real time reconfiguration of systems using cheap CPUs and statistical analysis of historical data from sensors.

The price of crystalline silicon PV cells has dropped from \$77/watt in 1977 to \$0.77/watt in 2013: this is a decline at 13% per year, or a halving each 5 years, for 36 years. This is giving a huge surge in installed capacity (albeit a highly subsidised surge in most places).

For decades, renewables have been a part of science fiction. Now, for the first time, massive scale renewable generation has started happening. The present pace of installation is, indeed, the child of subsidy programs, but the calculations now yield reasonable values even without subsidies. If and when the world gets going with some kind of carbon taxation, that will generate a new government-induced push in favour of renewables, which could replace the existing subsidies in terms of reshaping incentives.

Innovations in storage

Electricity generation using renewables is variable (wind) or peaks at the wrong times (solar). In addition, wind and solar production is naturally distributed; it is not amenable to a single 100 acre facility that makes 2000 MW. These problems hamper the use of renewables in the traditional centralised grid architecture. These problems would be solved if only we could have distributed storage.

What would a world with low cost storage look like? Imagine a group of houses who put PV on their roofs and run one or two small windmills. Imagine that these sources feed a local storage system. The renewable generation would take place all through the day. When electricity prices on the grid are at their intra-day peak, electricity would be drawn from the storage system.

For the centralised system, the cost of delivering electricity at a certain $(x,y,t)$ can be quite high: perhaps households at certain $(x,y,t)$ can sell electricity back to the grid.

This is the best of all worlds for everyone. The grid would get a reduced $H/L$ ratio and would be able to do what the grid does best -- highly efficient large-scale base load technologies. The grid would be able to deliver electricity to remote customers at lower cost. Consumers would be better off, as payments for expensive peak load electricity would be reduced.

This scenario requires low cost storage. For many years, we were stuck on the problem of storage. In recent years, important breakthroughs have come in scaling up lithium-ion batteries, which were traditionally very expensive and only used in portable electronics. Lithium Ion batteries have 2.3 times the storage per unit volume, and 3.1 times the storage per unit mass, when compared with the lead acid batteries being used with inverters in India today.

Tesla Motors is an American car company. They have established a very large scale contract with Panasonic to buy Lithium Ion batteries. Nobody quite knows, but their internal cost for Lithium Ion batteries is estimated to be between \$200/kWh and \$400/kWh. On Thursday (30 April 2015), they are likely to announce a 10 kWh battery for use in homes. It's cost is likely to between \$2000 and \$4000 for the battery part, yielding a somewhat higher price as there will also be a non-battery part. (It is not yet certain that the part they announce will be 10 kWh. There are many stories which suggest this will cost \$13,000, which are likely to be wrong). A 10 kWh battery can run for 10 hours at a load of 1000 Watts. Note that Tesla is only pushing innovations in manufacturing; they are not improving battery technology. Many others are on the chase for better battery technology. Stupendous progress has happened with batteries in the last 20 years. Only two years ago, this price/performance was quite out of reach. It is a whole new game, to get a Lithium-Ion battery at between \$200 to \$400 per kWh. Suddenly, all sorts of design possibilities open up. Further, this is only the beginning. Experts in this field in the US believe that when Lithium Ion batteries are below \$150/kWh, they will be fully ready for applications in the electricity industry in the US. These experts believe this number will be reached in 5 to 10 years.

The rise of storage links up to the rise of electric cars in two ways. First, electric cars are driving up demand for lithium-ion batteries and giving economies of scale in that industry. Second, a home which has an electric car has that battery! The present technology in electric cars -- Tesla's Model S -- has a 85 kWh battery, which is good capacity when compared with the requirements of a home.

Renewables have generated excitement among science geeks for a long time, but have disappointed in terms of their real world impact. Scientific progress in renewables, and in batteries, are coming together to the point of real world impact.

Storage is one method for coping with the intermittent generation from renewables. The other method is to make demand more flexible. As an example, a smart water heater or a smart air conditioner could do more when electricity is cheap, and vice versa. This would make consumption more price elastic.

Leapfrogging in India?

The Indian environment with expensive and intermittent electricity from the grid is an ideal environment for renewables + batteries.

Distributed generation and distributed storage are seen as ambitious cutting edge technology in (say) Germany. Perhaps the natural use case for this is in India. In Germany, the grid works -- there is no problem with achieving high availability. In Germany, there isn't that much sun. In India, every customer of electricity suffers increased costs in getting up to high availability, and there is plentiful sunlight.

A weird thing that we do in India is to charge high prices for the biggest customers of electricity. For these customers, roof-top PV systems are already cheaper. Problems in the fuel supply have given a steep rise in base load prices, and have pushed the shift to renewables.