Five Explosions in Seven Months: Why Insurers Are Re-Rating India's Secondary Steel Sector

Indian secondary steel has always been underwritten as a severity class. A molten-metal event is violent, it kills people, and it takes a melting shop out for months. What changed through 2026 is that the same class started producing losses at a pace that no longer reads as bad luck.

The dated sequence, drawn from an August 2026 analysis published by SMM on news.metal.com and from wire reporting, runs like this:

1. 22 January 2026, Baloda Bazar, Chhattisgarh. An explosion at a sponge iron plant killed six workers and injured five.
2. 6 April 2026, Ramgarh district, Jharkhand. A furnace explosion at a private steel plant killed three workers and severely injured nine.
3. 8 June 2026, Visakhapatnam Steel Plant. A ladle explosion killed eight workers initially, with the toll rising to ten by 14 June.
4. 9 August 2026, Ramgarh. A furnace explosion at a private steel plant injured at least nine workers.
5. 21 August 2026, Ramgarh district. A further explosion at a plant killed three engineers, reported by Xinhua English on 22 August.

Five events in seven months, on the public record, with a combined death toll above twenty. Three of them in Ramgarh district, two of those twelve days apart in August. That last detail is the one underwriters keep returning to, because a district-level repeat says something about how a cluster of plants is being run.

The commodity-market read on this cluster is what makes it an underwriting problem rather than a run of headlines. The same SMM analysis sets the losses against three capacity numbers.

Over three years, Indian steel capacity grew by 40.9 million tonnes while output rose by 25.85 million tonnes. Capacity was added faster than the market absorbed it. Utilisation fell over the same period from 80.4 percent to 77.2 percent. And the fastest-growing slice of that new capacity was induction furnace capacity, up 16 percent year on year.

Read those three numbers together and a risk picture assembles itself:

• New capacity means new plant commissioned by operators who may be building their second or third unit, not their tenth, with commissioning-phase crews and unfamiliar equipment.
• Falling utilisation means margin compression across the sector, which is exactly the condition under which discretionary spend gets deferred. Refractory relining cycles stretch. Cooling-water instrumentation replacement slips. Safety headcount is not backfilled.
• The induction furnace route growing fastest concentrates the growth in the melting technology with the sharpest molten-metal-and-water exposure, and in units that are typically smaller and less heavily engineered than an integrated blast furnace operation.

The three combine badly. Deferred maintenance on a coreless induction furnace does not degrade gracefully. A cooling coil running past its inspection interval leaks water into a bath at roughly 1,500 degrees Celsius, the water flashes to steam at over a thousand times its liquid volume, and the melt is ejected across the shop floor. The same physics applies to a wet or sealed charge, and to a ladle with moisture trapped in its lining.

Property rating for a steel melting shop was already among the higher industrial rates in the Indian market. What follows is where a class carrying this loss record and this capacity data normally lands at renewal, rather than a published market position. What is moving in 2026 is less the headline rate than the structure around it.

The rate. Expect hardening on melting-shop and sponge-iron occupancies, applied unevenly. A well-instrumented plant with a clean five-year record and documented furnace controls will see a modest increase. A plant in an affected district, or one with a molten-metal near-miss on file, will see something steeper, and in a few cases will be told capacity is not available at any rate until specified improvements are made.

The deductible. This is where most of the movement is. The deductible on molten-metal and furnace-related perils is being pushed up faster than the rate, often as a separate and materially higher figure than the general property deductible. The logic is straightforward: insurers are willing to keep writing the catastrophic furnace loss but not the recurring medium-sized one, and a separate molten-metal deductible sorts the two.

Capacity and co-insurance. Lead insurers on large secondary steel accounts are cutting line size, which pushes more accounts into co-insurance structures and lengthens placements that used to close in three weeks to six.

What is not changing. The perils remain insurable. An explosion causing physical damage is a standard peril under the fire insurance section, and sudden mechanical or electrical failure of the furnace, its converters and its capacitor banks sits in machinery breakdown. There is no market-wide exclusion of molten-metal risk in India, and a broker told otherwise should test that assertion. What is being restricted is terms, sub-limits and warranties, not the grant.

Downtime. Business interruption terms are getting more scrutiny than material damage, particularly on the indemnity period. A furnace body, a transformer or a bespoke ladle carries a re-fabrication lead time that a twelve-month indemnity period frequently does not cover. Underwriters are asking to see the spares and lead-time analysis that justifies the period chosen, and declining to write a period that the buyer cannot evidence.

Risk-improvement recommendations (RIRs) at a steel plant used to be a discussion about pricing credits. On melting-shop occupancies in 2026 a subset of them has moved from the credit column to the entry column: without them, the account is either not offered or is offered with the molten-metal peril carved out.

The items being treated as conditions rather than suggestions:

• Dry-charge discipline, documented. A written and audited procedure covering charge material preheating or drying, exclusion of sealed containers, and rejection of visibly wet scrap. Underwriters want the audit records, not the procedure document.
• Cooling-system leak detection on every furnace. Continuous monitoring of cooling-water flow, temperature differential and conductivity, with an automatic power-cut interlock on a leak signal. A manual gauge that an operator is meant to watch does not satisfy this.
• Refractory and lining management with hard limits. A recorded heat count per lining with a mandatory relining threshold, plus lining thickness measurement, applied to ladles as well as furnaces. The Visakhapatnam ladle event has made ladle lining records a specific question on survey.
• Furnace pit and drainage design. An emergency drain pit sized for a full furnace charge, kept dry, with no water services routed above or adjacent to the melt path.
• Separation of the melt floor from occupied areas. Distance, barriers and controlled access, so a breakout does not put a control room, rest area or contractor gathering point in the ejection path. Casualty counts on molten-metal events often turn on who was standing nearby rather than on the size of the event.

When a survey produces RIRs, ask the surveyor which are conditions precedent to the molten-metal peril and which are pricing credits, and get the answer in writing. In a hardening market the two arrive in a single list, and a buyer who treats a condition as a credit finds out the difference at claim stage.

Completion evidence matters as much as completion. Dated photographs, calibration certificates for the leak-detection instrumentation, signed audit records for charge inspection and lining measurement. A verbal assurance that a recommendation is closed carries little weight on a class this loss-affected.

The structural change that plant owners find least intuitive is that a single unit's clean loss record no longer stands on its own.

Group aggregation. Where a promoter group runs several plants, insurers are increasingly pricing the account on the group's combined experience rather than treating each unit as a separate risk. A group with one unit that had a fatal event and three that did not will find all four re-rated. Management practice, maintenance budgeting and safety culture are set at group level, so a loss at one unit is evidence about the others. Where units sit with different insurers, that view is assembled during the survey through questions about common management, shared maintenance contractors and centralised procurement.

District and cluster reads. The three Ramgarh events, two of them twelve days apart in August, have made district-level concentration a live rating factor. When a cluster produces repeat losses, underwriters begin asking whether the plants share contractor pools, scrap suppliers, or a local labour market with the same training gaps, and whether the state factory inspectorate has the capacity to enforce in that cluster. A plant in an affected district carries some of that inference whether or not it has had a loss.

What a plant owner can do. The inference is rebuttable, and rebutting it is worth real money at renewal:

1. Present the group's safety governance explicitly. Board-level safety reporting, a group standard applied across units, internal cross-audits between units.
2. Show that a loss at a sister unit produced a documented corrective action that was implemented at every unit, with dates. This is the single most effective response to group aggregation, because it demonstrates that the group treats a loss as systemic rather than local.
3. Separate yourself from the district read with specifics. Named contractor vetting, in-house rather than contract labour on the melt floor, scrap supplier qualification with radiation and sealed-container screening.
4. Bring your own near-miss data. Presenting a near-miss register with investigated entries reads as control, and its absence reads as a plant that either does not record near misses or does not investigate them.

Every event in the 2026 sequence killed or injured workers. On a class where the human loss is the defining feature, the property programme is not the whole answer.

Employee compensation. Workers compensation and employers liability sums insured need checking against actual wage rolls including contract labour, which is where most melt-floor exposure sits and where cover is most often deficient. A policy scheduled on direct employees only can leave the largest group on the floor uncovered.

Public liability. The Public Liability Insurance Act cover attaches to plants handling hazardous substances, and a plant that has lapsed on it finds the gap at the worst moment.

Directors and officers. A fatal accident brings personal exposure for the occupier and the factory manager under the Factories Act, and increasingly for directors where systemic maintenance failure is alleged. Defence costs on a prosecution running two or three years are material even where the finding is eventually favourable.

The reporting trap. Underwriters are asking, on renewal proposals, for disclosure of fatal and serious-injury incidents whether or not a claim was made. A plant that settled an incident directly with the family and did not notify the insurer at the time now has a disclosure decision to make. Non-disclosure of a material fact on a proposal is the fastest route to a repudiated claim on the next loss, and a fatality at a melting shop is material by any reading. Disclose it, explain the corrective action taken, and price the consequence.

The practical sequence for a plant owner or broker approaching renewal on secondary steel capacity, in the order it should happen:

1. Start ninety days out, not thirty. Reduced line sizes mean more markets and longer placement. A late submission on this class in 2026 gets whatever terms are left.
2. Commission your own risk survey first. A pre-renewal engineering survey lets you close the obvious RIRs on your own timetable and present a plant that has already acted, instead of one being told what to fix.
3. Rebuild the sums insured. Check that plant and machinery values reflect current replacement cost so the average clause does not cut a claim, and that the sum insured on stock tracks billet and scrap values, which have moved.
4. Rebuild the indemnity period. Get written lead times from the furnace, transformer and ladle suppliers, and set the business interruption period on that evidence. Bring the correspondence to the underwriter.
5. Assemble the safety file. Cooling-leak detection calibration records, lining heat counts, charge inspection audits, near-miss register, training records for melt-floor crew including contract labour, and corrective actions from any group incident.
6. Decide the deductible trade deliberately. A higher molten-metal deductible in exchange for a workable rate and full peril coverage is usually the better trade than a lower deductible with a sub-limit on molten-metal damage. Model both against your own loss history before the market presents the choice.

The wordings are where this class is won or lost. Two insurers can quote within a few percent of each other on rate while differing sharply on how they draft the molten-metal peril, whether furnace refractory sits inside or outside the machinery-breakdown grant, what the cooling-system warranty requires, and how the indemnity period responds to a bespoke re-fabrication. Sarvada puts insurer wordings side by side so a broker or risk manager can read those clauses against each other on a steel account and argue the terms that decide a claim rather than the headline premium. Secondary steel operators and their brokers can Request Access to put that comparison to work.

For the process hazards behind this, the steel industry risk profile and the coal-based sponge iron plant profile work through the kiln, melting shop and downtime exposures in detail.