Why the Lights Come Back On After 20 Minutes

There is a green metal box somewhere at the end of most residential streets in Kerala. You have probably walked past it a thousand times without thinking about it. It steps down the 11,000-volt supply coming from the upstream substation into the 240-volt supply that reaches your home. It contains copper windings, a steel core, and several hundred liters of mineral oil. It has no moving parts and makes no sound.

This summer, that box has been working harder than it was ever designed to work. And when it hits its limit, it cuts the power to protect itself. Not because an engineer in a control room decided to. Because a relay inside the transformer did the math and tripped the breaker.

The 20 minutes of darkness that follow is not energy conservation. It is the transformer cooling down. Once it is cool enough, the relay permits a reconnection. If the same households immediately turn their ACs back on, the cycle starts again.

This is the physical mechanism behind the blackouts that have made Kerala ungovernable for the past two summers. Understanding it requires starting with what happens to voltage when a transformer is overloaded, and why modern appliances make the problem substantially worse.

---

What happens to voltage under load

The Kerala Electricity Supply Code mandates that KSEB maintain a domestic supply voltage of 240V with a permissible variation of plus or minus 6%. The minimum acceptable voltage at a consumer's socket is 225.6V.

The reality in many Kerala neighborhoods during the summer night peak is a line voltage of 140V to 180V. This is not a billing fiction or a measurement artifact. It is a direct, unavoidable consequence of overcrowded distribution lines.

The physics is straightforward. Every cable and conductor has electrical resistance. When current flows through a resistance, it drops voltage. The relationship is:

$V_{drop} \approx I \times R_{line}$

Where $I$ is the current flowing through the cable and $R_{line}$ is the cable's resistance. The low-tension service cables running from the neighborhood transformer to each home were sized based on declared loads. When hundreds of undeclared ACs and EV chargers activate simultaneously at 11 PM, the current $I$ surging through those cables is two to three times the design figure. The voltage drop becomes enormous. By the time electricity reaches the consumer's socket, a substantial fraction of it has been lost as heat in the cables themselves.

On top of this, during periods of extreme grid stress, KSEB substation engineers sometimes deliberately lower the tap settings on primary transformers to restrict total power flow and prevent a cascading collapse upstream. This protects the high-voltage transmission network but lowers the starting voltage for the neighborhood distribution system, compounding the consumer's low-voltage experience.

---

The vicious cycle: constant power loads

A low-voltage supply would be a nuisance but a manageable one if the appliances connected to it simply consumed less power when voltage dropped. A resistive water heater works this way. Lower voltage means lower current, which means less heating. The appliance underperforms but the grid sees a reduced demand.

Air conditioner compressors and modern EV chargers do not work this way. They are what electrical engineers call constant power loads.

Real power in an AC circuit is given by:

$P = V \times I \times \cos\theta$

Where $P$ is real power in watts, $V$ is voltage, $I$ is current, and $\cos\theta$ is the power factor. An AC compressor motor is designed to deliver a specific thermodynamic output: it needs to remove a certain amount of heat from the room to maintain the set temperature. If the voltage drops and the power delivered decreases, the control circuitry and motor characteristics respond by drawing more current to compensate and maintain the same power output.

The numbers here are not trivial.

Assumes constant 2 kW load and unity power factor for illustration. Real-world induction motor behavior is more complex but directionally identical.

A 53% increase in current drawn through the transformer's windings and the distribution cables feeds directly into the transformer's internal heating. And the heating is not linear.

---

Joule's Law and quadratic destruction

Heat generated in an electrical conductor is governed by Joule's Law:

$P_{loss} = I^2 \times R$

The key word is squared. A 50% increase in current does not produce a 50% increase in heat. It produces a $(1.5)^2 = 2.25$ times increase, or a 125% increase above baseline. A 53% increase in current produces $(1.53)^2 = 2.34$ times the heat, or 134% above baseline.

Inside the neighborhood distribution transformer, this $I^2R$ heat is being generated in the copper windings continuously during peak load hours. The transformer's passive cooling system, which relies on convection currents in the surrounding mineral oil to carry heat from the windings to external cooling fins and radiate it into the ambient air, is designed for normal operating loads. It is not designed for a sustained 230% heating event happening simultaneously during a hot summer night.

---

What happens inside a failing transformer

A standard KSEB distribution transformer is filled with highly refined mineral oil. The oil serves two purposes: electrical insulation between the copper windings, and thermal management through passive convection. Under normal operating conditions, oil temperature stays well within safe limits, typically below 65 degrees Celsius above ambient.

Under sustained overload, this passive cooling is overwhelmed. As winding temperatures spike, the oil heats to temperatures far beyond its design envelope.

The degradation sequence is as follows.

At elevated temperatures, the cellulose paper that wraps and insulates the copper windings undergoes thermal decomposition. This process is exponential: the rate of degradation roughly doubles for every 6 to 10 degree Celsius rise above the design temperature. A transformer rated for 30 years of service life at normal temperatures loses years of operational life for every hour it operates at 140 degrees Celsius.

As oil temperatures continue to rise, the mineral oil begins to oxidize. Oxidized oil is darker, more viscous, and electrically less insulating than fresh oil. It also forms conductive sludge that settles on winding surfaces, impeding the very heat transfer it was supposed to facilitate.

If the thermal runaway is not interrupted, the degraded insulation eventually fails dielectrically. The voltage difference between adjacent windings drives current through what was previously an insulating gap. This arcing generates intense, localized heat. At this point, the mineral oil can ignite. The transformer explosively fails.

Between February and mid-April 2024 alone, KSEB reported 578 distribution transformer failures attributable to peak-hour overload. That is roughly five blown transformers every single day, for more than three months. Each replacement involves capital expenditure, dangerous live-line work, coordination with constrained supply chains for 250 kVA and 500 kVA units, and substantial skilled labor time. KSEB's ability to replace failed transformers faster than the next ones fail is a genuine operational constraint that the utility was struggling to meet.

---

The ANSI 49 relay and why it trips

Modern KSEB distribution transformers and feeder lines are equipped with protection relays that are specifically designed to prevent the catastrophic failure sequence described above. The relevant standard is ANSI Device Number 49, the Thermal Overload Relay.

An ANSI 49 relay does not behave like a simple fuse. It does not trip the moment a current limit is crossed. Instead, it runs a continuous mathematical model.

The relay monitors the current flowing through the transformer. Using integrated algorithms, it calculates the integral of current squared over time:

$\text{Thermal State} \approx \int I^2 \, dt$

This computation generates a real-time "thermal replica," a mathematical estimate of the transformer's actual internal hot-spot temperature. The relay does not know the temperature directly; it infers it from the electrical signature.

As the evening load ramps up and the current spikes due to undeclared ACs running at depressed voltages, the relay's thermal replica climbs. When the model predicts that the internal winding temperature is approaching the critical threshold at which irreversible insulation damage begins, typically programmed around 140 degrees Celsius for distribution transformers, the relay sends a trip command. The primary circuit breaker opens. The transformer is electrically disconnected.

The neighborhood goes dark.

---

Why it is always 20 minutes

The specific duration of these blackouts, consistently reported by Kerala consumers as between 15 and 30 minutes, is the most direct empirical evidence that the events are driven by autonomous thermal logic rather than human decision-making.

Once the ANSI 49 relay trips the circuit breaker, it enters a lockout state. The relay calculates a mandatory cooling period during which it will absolutely not permit the breaker to be closed again, regardless of any commands from operators or automatic reclosing systems. It will only permit reconnection once its thermal replica model calculates that the transformer's internal temperature has fallen to a safe operating baseline.

This cooling process is governed by the transformer's thermal time constant, typically denoted $\tau_c$. The decay is exponential:

$T(t) = T_{ambient} + (T_{trip} - T_{ambient}) \cdot e^{-t/\tau_c}$

Where $T_{trip}$ is the temperature at the moment of tripping and $T_{ambient}$ is the surrounding air temperature. During a hot Kerala summer night, ambient temperature itself is elevated, which slows the cooling curve significantly.

Standard thermal protection relay engineering literature documents the cooling lockout as typically lasting 15 to 30 minutes after an over-temperature trip. The consistency with which Kerala consumers report exactly this duration is not a coincidence. It is the thermal time constant of the transformers at the end of their streets.

When the breaker recloses and power is restored, if the same households immediately turn their ACs back on at the same time, the thermal accumulation restarts from an already-elevated baseline. A second trip follows within minutes. This is why some neighborhoods report multiple sequential blackouts in a single night.

---

The role of automatic circuit reclosers

Many suburban and rural KSEB feeder lines use Automatic Circuit Reclosers, which are smart pole-mounted switchgear units designed to handle transient faults, like a palm frond brushing a live wire, by tripping and reclosing automatically without requiring a linesman to intervene.

Under normal conditions, an ACR trip lasts a few seconds. The fault clears. The ACR recondenses and power is restored. Nobody even notices.

Under sustained thermal overload, the ACR's behavior is different. It progresses through its programmed safety sequence: trip, wait, reclose, detect the overload is still present, trip again. Once it has exhausted this sequence without the fault clearing, it enters a final lockout state. In lockout, the ACR requires either the expiration of a long pre-programmed thermal reset timer, or a KSEB linesman physically traveling to the pole to manually reset the unit after confirming the load has dropped.

Given that KSEB section offices are severely understaffed during night shifts, and the number of tripped units across the network on any given night is substantial, manual reset response times of 20 to 30 minutes are entirely realistic. This adds a second mechanism that produces outages in exactly the window consumers are reporting, independent of the relay physics.

---

What actually needs to change

The crisis is real, it is compounding, and the interventions required operate at two separate levels that need to be addressed simultaneously.

At the procurement level: The financial hole created by the DBFOO cancellations has to be filled with new, legally unassailable long-term contracts. Continued dependence on the short-term exchange at ₹8 to ₹12 per unit guarantees operational losses that crowd out capital investment indefinitely. KSEB cannot modernize the distribution network while hemorrhaging ₹15 crore a day in emergency procurement.

At the metering level: The connected load affidavit system is obsolete. It relies on voluntary, self-reported data that consumers have a clear financial incentive to understate. Advanced Metering Infrastructure, which records real-time peak load data rather than relying on decade-old declarations, gives KSEB actual intelligence about what each transformer is serving. Predictive sizing becomes possible. Reactive catastrophic replacements become avoidable.

At the tariff structure level: The regulatory friction and financial penalty for transitioning from single-phase to three-phase supply needs to be substantially reduced. The current structure creates a situation where the individually rational choice for every consumer, to stay undeclared on a single-phase supply, produces a collectively catastrophic outcome for the neighborhood grid. Subsidizing the transition and simplifying the upgrade process shifts the incentive structure.

At the infrastructure level: Available CAPEX needs to go immediately toward replacing aged 100 kVA and 160 kVA transformers in high-density urban and suburban corridors with 250 kVA and 500 kVA units. The areas to prioritize are not hard to identify: sustained nighttime voltage sags below 200V are a reliable signal that the local transformer is thermally overloaded. Night after night of brownout data from even basic metering points to exactly which transformers are at risk.

---

What is and is not happening

I want to state this as plainly as I can, because the public conversation has been genuinely confused about it.

The 20-minute blackouts are not KSEB operators conserving energy. They are distribution transformers executing autonomous thermal protection logic to avoid catching fire. The grid cannot be instructed to do this at 6 AM instead of midnight. It responds to physical stress, and the physical stress peaks when thousands of undeclared ACs run simultaneously on hot nights.

The broader crisis is two separate failures that have intersected at the worst possible moment. A macro-level regulatory and judicial process that stripped the state of 465 MW of cheap baseload power and crippled the utility's finances. And a micro-level infrastructure failure driven by millions of consumers quietly tripling their electrical load while formally declaring nothing.

Neither failure is simple. Neither has a quick fix. The contracts are gone and will not return under the same terms. The hidden load has been accumulating for a decade and will not surface overnight even with amnesty schemes.

What can change, if the state commits to it, is the information quality that KSEB operates on and the regulatory structure that determines whether consumers have a financial reason to tell the truth about what they are running. Right now they do not. Until that changes, the transformer at the end of your street will keep doing the math and tripping the breaker at 11:30 PM, and the lights will keep coming back on exactly 20 minutes later.