Power failure: What really happened and why the grid's size & complexity are a big problem
By Avinash Celestine & Shantanu Nandan Sharma
The massive collapse in India's power grid earlier this week the worst in a decade, has become a bizarre Rorschach test. A power failure, albeit on a huge scale, has become a symbol of something far deeper.
Depending on who you are it came to symbolise everything from the weaknesses and failures of India's power reforms, the moral and policy failings of the eight-year reign of the UPA government, the decline of the India story, or in Team Anna's case, a massive conspiracy (to achieve exactly what was never quite clear).
The grid collapse, apparently, was nothing so trivial as the failure of a system, driven by technology, economics and politics, that was supposed to work in a certain way, but didn't.
The villains of the actual grid collapse were also quickly identified it was a clutch of northern states, most prominent among them being Uttar Pradesh, who were accused of 'overdrawing' power from the grid, leading to its collapse. UP strenuously denied that it overdrew power, but proceeded to sack the head of the UP Power Corporation anyway.
The increasingly desperate hunt for a deeper meaning aside, the biggest problem with this picture was that it didn't quite match the evidence. What exactly happened will only emerge when the enquiry committees complete their reports.
But from the data that's emerged so far, it's becoming increasingly clear that any argument based on a few states overdrawing power from the grid is only a partial explanation at best. UP and Punjab are hardly innocent players in this whole drama, but they aren't necessarily the main players.
In a sector beset with problems, whether generation, regulatory failure or fuel supply, the huge changes in India's power transmission network in the last decade or so, is actually a success story of sorts. But with that growth in transmission capacity, came complexity, and that bought with it new risks something the engineers and bureaucrats responsible for the system clearly recognised.
In a meeting in March 2012, the head of the central electricity authority, a key policymaking body for the sector, noted that the "Indian power system is growing manifold and their complexity is increasing on all fronts". He went on to state that "any incident natural calamity etc, even on single element of this capacity, has the potential to cause a major grid disturbance". This was just one warning. There were others.
A Grid to Rule Them All
Till about the middle of the last decade, India's electrical transmission grid, essentially a system to move power from suppliers (generating plants) to wholesale customers like state electricity boards, was a patchwork. It started off decades ago as a large collection of state-level networks with few if any links among states.
But beginning in the '60s, states in individual regions began to link their electrical networks with each other. This process continued gradually, but it wasn't until 2002, that connections between regions began in earnest.
Between 2002 and 2006, the northern, eastern, western and northeastern grids were all linked up through a set of transmission lines creating a power 'superhighway' across central and eastern India from Agra and Gwalior in the west towards Sasaram in Bihar (see map).
A swathe of India, from Maharashtra and Gujarat to the North East states are now one electrical grid whose crucial artery is this superhighway. The commissioning of the Agra-Gwalior line in 2006, was the final link in closing this mega-loop. It was this Agra-Gwalior link which was to play an important part in the grid collapse.
Why was it important to integrate the grids this way?
Customers for power are widely distributed important urban and industrial centres such as Gujarat, Maharashtra, Mumbai, Delhi for instance, while the fuel for the power they need is highly localised.
The coalfields of Jharkhand and Chhattisgarh are the most critical source, followed by the hydropower rich states in the north, and the northeast. Rather than move the coal to where it's needed, its cheaper to set up generating stations near the coal fields, and move the electricity to customers, which is where the grid comes in handy.
This NEW grid (North East-East-West) is linked with the southern grid, but more loosely. By 2014, the southern states are expected to be as tightly linked with the NEW grid as other states, something which residents of those states might look upon with decidedly mixed feelings now.
Coordinating the movement of electricity between the five grids, and within them, is a hierarchy of what are called load dispatch centres at the state level, at the level of each of the five grids, and finally a national one. The five grid-level dispatch centres, and the national centre are run by Power System Operation Co Ltd (POSOCO), a subsidiary of Power Grid corporation, a central government PSU. The state-level centres, on the other hand, are run by the state governments. This is where the politics comes in.
It Happened One Night
Why doesn't the 'overdrawing by UP and others' theory hold up? Anyone claiming this as the main reason would have to explain why the grid failed on these two days only, given that overdrawing is the norm, by many states throughout the year. Exhibit 1 in the case against overdrawal being the main culprit, is the state of the grid before the first outage in the early hours of July 30. When there's heavy demand on the grid, in relation to the actual supply available, the grid 'frequency' should drop. That didn't happen.
Think of grid frequency as being similar to the water level in a tank with an outlet and an inlet pipe. If more water is being taken out of the tank than is being pumped in, the water level drops. Similarly, if there is more demand in the grid, than there is supply to meet it, the frequency should drop below 50 Hz that's the norm and the standard to which all electrical devices in India, from washing machines, to hair dryers to TVs work.
The grid frequency at around 2 am on July 30 was about 49.68 Hz by the standards of India's electrical networks, that's actually a fairly minor deviation from the norm and something which is seen quite often. And exhibit No. 2 is the total demand in the NEW grid which was being met at the time 74,210 MW, far below India's generating capacity. "Overdrawal by itself would not have triggered this incident," says Bhanu Bhushan, a former member of the Central Electricity Regulatory Commission, the apex-level regulator in the sector.
So what did happen?
It's early days yet, but the possible chain of events began in the western part of the superhighway. The day before the outage, two critical components of the grid were already weak. One of the two circuits linking Agra and Gwalior, and the line between Zerda in Gujarat and Kankroli in Rajasthan were down either due to repair or upgrade work. Between them, these two routes were the main link between the western and northern grids.
With two important links between the western and northern grids down, the remaining circuit between Agra and Gwalior bore the brunt. By July 29, there was over 1,000 MW of power flowing on this one circuit alone, leading to a near collapse of this critical line on that day at 3 pm around 36 hours before the grid actually collapsed.
"The Agra-Gwalior link was the major landmark in the formation of the NEW grid, so removal of part of it meant the grid was already compromised, even before the event," says Puneet Chitkara, principal at Mercados EMI, an energy consulting firm. "With the grid already being weak at that point, any small event could have triggered the collapse."
Given the steadily weakening link between the northern and western grid, how was power to flow from west to north? It was likely that power would have been 'wheeled' through the east a long circuitous route across Madhya Pradesh, Chhattisgarh, Jharkhand, Bihar and Uttar Pradesh, leading to massive loads on the lines there.
This is where the politics begins to seep in. The lines which branch off the grid in different states (called radial lines) are fitted with special equipment which are supposed to disconnect that line from the rest of the grid, when they detect a sharp fall in frequency, signifying high loads. As the line disconnects, the demands on the rest of the grid should fall (since a set of consumers have been cut off), thus helping bring the supply-demand back into balance.
However, many of these radial lines and under frequency relays as they are called, are owned and operated by state government entities. These entities rarely maintain relays properly and are often under political pressure to continue drawing power from the grid, even when the load on the system is heavy.
Incidentally, calls for POSOCO to 'disconnect' errant states are impractical, because the relevant lines it needs to disconnect are crucial to the functioning of the rest of the grid. Disconnecting them would likely take the rest of the grid down as well.
Grid Collapse
With such massive loads, at some point things had to give way. Between 2:33 am and 2:35 am on July 30, a large set of lines between Balia in Bihar, Gorakhpur in eastern UP and Lucknow simply tripped and shut themselves down. In 2009, the Agra-Gwalior line had suffered a similar collapse. In a study of that incident, Power Grid engineers stated: "Outage of only one element in the system might have lead to a collapse...thus we need to revisit...contingency plan for better security in future."
At the same time as the various lines in UP and Bihar were tripping in the early hours of July 30, other components started shutting themselves out or 'islanding' themselves from the problem. The western region would have disconnected itself from the rest of the grid by this time. Then the generating stations started islanding themselves as well to protect against damage to power equipment caused by sharp swings in the grid frequency.
Within two minutes, between 2:33 am and 2:35 am, a vast swathe of north India went dark. It was a classic domino effect a chain of events, each one causing the other. Individual events were not necessarily critically dangerous in themselves, but taken together, they bought system crashing down.
The second outage of July 31 was a consequence of the first. In the hours leading up to it, a large chunk of lines connecting the east and the north from Balia to Patna to Biharshariff were still down. The crucial Ranchi-Sipat link between the eastern and western regions was down. And the Zerda-Kankroli line was still out of action. If the grid was already compromised before the outage on 30th night, it was tottering on the morning and early afternoon of the 31st.
Questions will be asked as to how the grid was allowed to function in such a weak state. Again, a small fault anywhere along the system would have triggered a cascade of trippings. Within the space of a minute, between 1:01 pm and 1:02 pm, 38 links between various parts of the northern, western and eastern grids went down. Most of the generating capacity went too.
The Western Grid could also have suffered a collapse as it islanded itself, the frequency there rose sharply, indicating excess supply of power (since customers in the north were not available) this could have potentially damaged plants in the region. Whether good luck or good management saved the day, is not known.
Better But More Complex
"In the '80s and '90s, these kinds of grid problems were a far more frequent occurrence," Bhushan points out. The NEW grid has helped build and maintain a far more even and steady flow of power across the country from surplus to deficit regions, something that was not possible earlier. Wild swings in frequency, which indicate the extent to which the system is out of balance, are far less common today than they were a decade ago.
Power stations, which earlier had to be very sensitive to shifts in demand within their region over the course of the day, can now export their surpluses to where its needed. Capacity utilisation of plants has improved. However, as Chitkara points out: "The pricing mechanisms and regulatory frameworks need to match up with the physical development. Bigger systems require better regulatory oversight the costs of failure can be huge in a large system," he says.
That's the dilemma. An integrated grid, working in sync, brings huge benefits. Problems are actually fewer. But because its more integrated, when problems do hit, the entire grid can go down. The costs of a single fault, or mistake are far higher.
The government has repeatedly failed to meet its own Plan targets of power production. The shortfalls are due to order delays, slow environmental clearances, rehab issues, litigation, and the non-availability of key resources coal and gas. Coal plants account for 54% of total capacity, hydropower for 22%.
The private sector participation in power production, at 60% in 1947, witnessed a downward trend in the 60s and the 70s, mainly because of policy paradigm. In 1979, the share of state sector was as high as 82% of the total production. But, with big public sector entities such as NHPC, NTPC, PFC coming to the forefront, the Central sector's share has moved up to 31%.
First Private Power Plant: It was a 2 MW diesel plant set up in 1905 at Lahori Gate area in Old Delhi. Under the provisions of Indian Electricity Act 1903, an Englishman managed a license for his company M/s John Fleming to set up the plant. It was later converted into Delhi Electric Supply and Traction Company.
Grid Economics
The system is supposed to incentivise producers to supply more at a time of heavy load. But as soaring fuel costs have boosted the cost of power, this 'incentive effect' has weakened sharply.
Besides the physical infrastructure, the economics of the grid has changed sharply in the past decade. Buyers and sellers declare the power they are likely to draw from the grid, or supply to it, 24 hours in advance to enable engineers plan and distribute loads between different entities.
Delhi for example might announce on Wednesday, that it will draw 2,300 MW in the morning of Thursday, increase that to 3,000 MW by the afternoon, and then gradually wind down in the evening. But if Delhi actually draws more than it said it would, that's what's called 'overdrawing'.
The system attempts to make states pay a price for doing so. When there is heavy demand in the system, and the frequency starts to drop, the cost of each unit of power starts rising automatically, making it more expensive to buy power, or making it more remunerative to supply more to the grid than a supplier earlier committed to.
Alternatively, if a buyer chooses to cut demand at such a time of stress, he gets rewarded by being paid this so-called UI rate. And for sellers who over commit, but underdeliver, the UI rate becomes a penalty they have to pay. The aim is to bring demand and supply back in balance. Its a different matter that some states are in heavy default of their UI dues.
But what regulators discovered a few years ago was that some states began intentionally scheduling more power than they actually needed. At times when they knew the grid was going to be overworked, they would tell the system they needed to draw say 3,500 MW, when they actually only needed 3,000 MW.
Thus they got the credit for being responsible members of the system at a time of stress, while making a tidy sum of money in the process, from the UI benefits they received. Similarly producers had an incentive to announce they would supply far less power than they actually could, then supply more, benefiting in the process.
So the regulator effectively capped the benefits that any player could get from the UI system, by replacing a single rate with a set of slabs. As a result, the per unit UI price has fallen from Rs 6.7 per unit in 2009-10 to Rs 4.09 per unit in 2011-12. Court rulings have also put paid to attempts to raise the UI rate.
But over time, fuel costs and the cost of generating each extra unit of electricity have soared, leading to a scenario where the cost of generation of each unit is more than the UI rate. Put simply, the costs outweigh the returns, leading to a situation where in times of heavy demand, power producers would rather cut production below what they scheduled, and pay the UI penalty, rather than produce more power as the UI system was supposed to incentivise them to do.
Chitkara argues that the shift to a so-called ancillary services market, mooted by POSOCO a few years ago is needed. Ancillary service providers are special producers, who can ramp up power supply very quickly to cope with shortfalls in supply in the grid.