Twenty-Five Fire Tenders From Five Districts: What Alwar Says About Estate Fire-Fighting in Your PML

On 16 August 2026, a fire at the Alwar Salvex mustard oil packing factory in the Matsya Industrial Area of Alwar, Rajasthan, destroyed roughly seven lakh litres of mustard oil. The reported loss was put at Rs 25 to 30 crore. The factory owner suspected a short circuit as the cause.

The response is the part that underwriters should read twice. Twenty-five fire tenders drawn from five districts made about 150 trips to the site, and the blaze was brought under control only by 2 pm, after roughly eight hours of fire-fighting. Within the entire industrial area, only two fire tenders were stationed. Everything else had to travel, refill and travel again.

Read that as a risk-engineering statement rather than a news item. A fire that requires 150 tender trips over eight hours is a fire that was never going to be extinguished in the first hour, because the water simply was not there in the first hour. That is not a failure of the fire service. It is a failure of the water supply and the response distance, and both of those are inputs an underwriter is supposed to price before the policy is written, not discover after the loss. The same fire, read for what it says about protecting a plant over a shutdown, is covered in our note on the shutdown-period fire protocol.

Most occupiers on Indian industrial estates have never asked their insurer what Probable Maximum Loss was assumed for their location. The number exists. It sat behind the rate they paid, the terms they were offered and the reinsurance the insurer bought. Alwar is a useful moment to go and ask.

Property risks carry two loss estimates that answer different questions.

Maximum Foreseeable Loss (MFL) is the worst single loss at the location if the protection does not work: detection is late, fixed systems fail or are inadequate, the brigade does not arrive in time, and the fire runs to the limit that the physical construction and separation allow. MFL is protection-discounted. It asks only how far fire can physically travel before something non-combustible stops it.

Probable Maximum Loss (PML) is the worst single loss realistically expected when the protection that is present and reliable does work as intended: the sprinklers operate, the firewalls hold, the hydrant pumps run, and the brigade attends within a useful time. PML is lower than MFL for a genuinely well-protected risk, and the size of the gap is the value of the protection. Both are usually expressed as a percentage of the total value exposed at the location.

The word doing the work in the PML definition is reliable. Every credit in a PML sits on an assumption, and the assumption can be tested. A sprinkler credit assumes a design suited to the hazard and a water supply that will sustain the demand for the design duration. A firewall credit assumes the wall is adequately rated, complete to the underside of the roof, and not breached by conveyors, cable trays or a propped-open door. A brigade credit assumes a tender that arrives while the fire is still small enough to matter.

The mechanics of that derivation, and how a broker challenges a figure that has been set too high, are worked through in detail in PML, MFL and the COPE Study and in the warehouse-specific treatment in MFL vs PML for warehouses.

The convenient PML shortcut on a manufacturing risk is that fire will be confined to one fire division, so the PML becomes the largest single division rather than the whole site. That shortcut works when the fire load is fixed in place: machines, structure, fitted plant.

A packing unit holding seven lakh litres of oil is a different animal, for four reasons.

1. The fire load is the stock, and the stock is mobile fuel. Oil that escapes a ruptured drum, tank or filled container does not respect the fire division the underwriter drew on the plan. It flows under walls, along drains and across floors, and it carries the fire with it. Firewall credit that would be sound for a dry-goods store is much weaker where the commodity is a flowing liquid.
2. Water alone is a poor extinguishing medium on an oil fire. Plain hose streams on burning oil can spread it rather than suppress it. Control usually needs foam, foam-compatible discharge, and enough of both to blanket the surface. Twenty-five tenders shuttling plain water is a cooling and exposure-protection operation, not an extinguishing one.
3. Stock values move a great deal within a year. A packing unit at the start of a mustard crushing season carries a materially different value from the same unit in a lean month. The declared sum insured is often the value on a quiet day.
4. The stock is the business. Destroy the machines and the unit rebuilds. Destroy seven lakh litres of finished and semi-finished oil and the contract book goes with it.

For occupancies like this, the honest PML is closer to the MFL than an underwriter would like, and the difference is only earned back by protection that actually addresses the commodity: bunding and drainage that contain a spill, foam capability on site, separation of bulk storage from the packing hall, and a water supply sized for a long incident rather than a short one.

The same logic runs through the fire ratings on other high-fire-load occupancies. See Fire insurance rate adequacy after de-tariffing for how the post-tariff market prices this class.

Since the fire portfolio was de-tariffed, Indian insurers price fire risk on their own assessment rather than a fixed schedule of rates and discounts. That cuts both ways. There is no longer a tariff table entitling an occupier to a stated discount for installing a system, and there is no ceiling on what good protection can be worth either. What protection buys now is a lower PML, and the lower PML shows up in the rate, in the terms, and in the willingness of the insurer to retain the risk net rather than push it to reinsurers on restrictive terms.

Protection that moves a PML on a stock-heavy estate risk, roughly in order of what it is worth:

• A sustained on-site water supply. Static storage plus pumps sized for the designed demand and duration. This is the credit that Alwar most visibly lacked at the estate level. A brigade with no local water is a fleet of tankers, and tankers deliver in trips, not in flow.
• Automatic suppression suited to the commodity. Sprinklers for the packing hall and stores, foam systems where bulk oil is held. Automatic systems act in the first minutes, which is the only window in which a PML credit is honestly earned. The rating treatment of sprinklers is covered in Fire sprinkler systems and insurance discounts.
• Containment. Bunds, kerbs, ramped doorways and drainage that hold a spill inside one area. On liquid-fuel occupancies, containment is what makes a firewall credit believable.
• Physical separation of bulk storage from process. The single most reliable way to keep a PML below the site total is to make sure the largest concentration of value cannot be reached by a fire starting in the busiest area.
• Detection and a trained on-site response. Early intimation shortens the interval before anything is done at all, which matters most when the brigade is far away.
• Electrical maintenance evidence. The suspected cause at Alwar was a short circuit. Thermography records, load studies and a documented maintenance regime are cheap to produce and directly rebut the "poorly maintained electricals" presumption an underwriter otherwise applies to an older shed.

Each of these has to be evidenced, not asserted. An underwriter cannot give credit for a foam stock nobody has counted or a pump nobody has run under load.

Mutual-aid arrangements between occupiers on the same estate are common in Indian industrial areas, and they are worth real credit when they are real. The Alwar response, 25 tenders from five districts and about 150 trips, is what happens when the arrangement is improvised at the time of the loss rather than agreed before it.

An arrangement an underwriter can price has most of the following:

• A written agreement between named occupiers, not an understanding between plant managers who may both have moved on.
• An inventory of what each member actually brings: tenders, trailer pumps, foam stock, hose, trained crew, and who pays for consumables used.
• A shared water source with known volume and refill rate, and hydrant points that connect to everyone's couplings. Incompatible couplings quietly void the whole arrangement.
• A single point of command for an incident, agreed in advance.
• Joint drills at a stated frequency, with records. A drill log is the difference between an arrangement and a document.
• Estate-level provision that scales to the estate, rather than two tenders for an entire industrial area.

Where the estate provision is thin and cannot be improved, the honest response is not to argue for a credit that is not there. It is to build private protection on site, and to reflect the exposure in the sum insured and in business-interruption indemnity periods instead.

An eight-hour fire is close to a total loss of everything within reach of it. That is where the gaps in a conventionally arranged fire programme open up.

Stock declared at the wrong value. Fire policies on stock in India are commonly written on a declaration basis so that the sum insured can follow a seasonal peak. Where the occupier has instead fixed a single figure at last year's average, the average clause applies and every claim is scaled down in the proportion of the sum insured to the actual value at risk. Underinsurance bites hardest on exactly the fire that destroys the peak. The evidential fight that follows is set out in stock valuation disputes in fire claims.

Buildings and plant insured at book value. Material damage is settled on the basis the policy states. Where reinstatement-value cover has not been taken, or the declared values were never revised, the settlement funds a depreciated asset rather than a replacement. An escalation clause keeps the figures moving through the policy year, and the mechanics are covered in sum insured escalation clauses.

Business interruption sized for a repair, not a rebuild. After a fire of this severity the constraint is not repair time. It is clearance, structural assessment, statutory approvals, plant lead time and the recovery of customers who bought elsewhere in the meantime. Indemnity periods set at six months on a seasonal processing business routinely expire before turnover recovers.

Debris removal, professional fees and firefighting costs. Seven lakh litres of oil, contaminated water and burnt packaging is an expensive site to clear, and clearance is often the first item that has to be paid before anything else can start. These sub-limits are usually set as a small percentage of the sum insured and are rarely revisited.

No cover for the neighbours. On a tight estate, a fire that burns for eight hours damages adjacent units. Third-party property damage claims from neighbouring occupiers fall to public liability cover, which the average estate occupier holds at a limit set years ago.

Whether these gaps are best closed inside a Standard Fire and Special Perils policy or by moving to an all-risks form is a separate decision, discussed in industrial all risks vs the standard fire policy.

Five questions, in order. All of them can be asked in one email.

1. What PML percentage did you apply to this location, and on what basis? The insurer has a figure. Ask for it in writing, with the survey report it came from and the date of that survey. A PML built on a survey from four years ago describes a plant that no longer exists.
2. What did you assume about fire-brigade response and water availability? If the assumption is a station within a few kilometres and a working estate hydrant main, and the reality is two tenders for the whole estate and tankers from the next district, the PML is wrong in the direction that hurts at claim time and the rate is wrong in the direction that flatters the insurer.
3. Which protections were given credit, and what evidence supports each one? Pump test records, foam stock, sprinkler design basis, drill logs, thermography reports. Anything credited without evidence is an argument waiting to happen after a loss.
4. What is the largest single value concentration on site, and can a fire reach all of it? If the answer is yes, the PML is close to the site total and no amount of negotiation changes that. The fix is physical: separation, bunding, and moving bulk storage away from the process.
5. Is the stock sum insured on a declaration basis, and does the declared peak match the season? Then re-run the same question for the indemnity period on business interruption.

A PML is a prediction about how a fire will be fought. Alwar is a reminder that the prediction is only as good as the water supply and the distance behind it.

For an occupier on an estate with thin fire cover, the practical sequence is: get a current risk survey, fix what is cheap to fix (containment, electrical maintenance evidence, drills, hydrant compatibility), evidence what is already there, and then take the whole file to the market at renewal. The PML follows the evidence. In the post-de-tariff market, the rate follows the PML.