Consider the data available. India’s energy-demand is expected to grow 35 -50% by 2030. We have 388 GW of installed capacity, of which 100 GW is renewable energy (and 50 GW is large hydro, not considered “green”). By 2030, we want to increase non-fossil-fuel generation capacity to 500 GW, and the share of non-fossil-fuel sources in generation to 50%. How much thermal capacity does India need to add/phase out?

Articulated thus, it sounds like a simple weighted-average problem. It is not. For one, the difference between capacity (or power) and generation (or energy) would not be obvious instantly. A power plant with a capacity of 100 MW can generate 100 MW of power at any instant. When it does so for an hour, it generates 100 MW-hour of energy. It can do this 90% of the time in a year, requiring the remaining time for maintenance.

A solar  power plant is able to do this for 20% of a year, considering night time and cloudy or foggy conditions. To generate the same amount of energy in a year, solar capacity five times that of thermal capacity is needed. Similarly, for wind, it depends on periods and intensity of wind speeds at different locations, and on availability of water for hydel. So, the increase in generation needed can be met by different combinations of capacities of different types of power.

Capacity planning is done to meet peak demand. Let us consider a peak demand of 2,200 MW, and power plants of 220 MW each. The capacity planner would have planned for 11 such power plants, instead of 10; even if one plant went offline unexpectedly, the grid would still be able to meet the peak demand. This is a rough approximation of the way in which power system experts (Central Electricity Authority, for instance) planned capacity addition.

Their aim was to minimise loss of load probability (LOLP). Of course, all generating units were not of the same size. They identified the largest unit connected to the grid, and planned for another similar sized unit to be on standby. It still did not reduce LOLP to zero, since more than one unit could get impacted, which would require shedding load. Such calculation was done for the expected peak demand.

In India, all of this was theoretical when we did not have sufficient capacity to meet peak demand; and load-shedding kept the grid balanced. In the last decade, India has added sufficient capacity to maintain “spinning reserves”, which can come online if a generating unit fails. Capacity planning has now become more complex. Since power is a concurrent subject, capacity addition (and management) is done by both central and state entities. Power generation has been delicensed.

Balancing the grid on an hour-to-hour basis is done through a detailed set of technical and commercial rules. Capacity addition requirement is calculated by the CEA, considering all of the above.

Now consider additional uncertainties. The time in which solar and wind power can be generated is uncertain, though significant effort is being made to improve day-ahead and week-ahead predictability. But, years-ahead is difficult, more so with increased climate uncertainties. Rate of increase in rooftop solar panels is another uncertainty for the planners.

Storage for upto four hours can address some weather variations, but not for prolonged no-sun/low-wind periods. Several states are separating power lines serving farms from those serving other users so that they can use agricultural load as a tool to match demand with availability of power. But, how soon this can happen, implementation of ‘time-of-day’, and operationalisation of smart-grids are big uncertainties.

Thermal capacity requirement for 2030 needs to be decided in the next 2-3 years. The private sector will find it very difficult to finance thermal power projects; even approvals to final construction, a factor of government efficiency, could take 5-8 years. So, the question needs to be answered without knowing whether we will actually ramp up to add 40-50GW of renewable energy capacity every year, whether storage and hydrogen will become affordable by then, and to what extent transnational grids would have been made functional.

How do you determine LoLP with these uncertainties? Planners can envision various scenarios of the future, and estimate LoLP for each scenario. They can also determine the amount of thermal capacity required to reduce LoLP to a target level, and the associated costs. But who will pick the scenario to bet on? Going wrong in one direction would mean power outages and sharp increases in power prices. Going wrong in the other could mean investment in thermal capacity that is under-utilised, whose cost will need to be borne either by consumers or by tax payers.

Neither of these outcomes makes for good politics. The answer will need to be decided at the political level; the power system managers, weather forecasters, statisticians and sociologists can only give them the what-if scenarios.