Cameras in Ceramics and Sanitaryware Plants: Kilns, Glaze Lines and Breakage

In a sanitaryware plant, most breakage is not caused where it is discovered. A crack found at post-kiln sorting was usually created hours earlier, during demoulding, fettling or kiln-car setting. Cameras are worth installing only if they can tie the defect back to that moment — and show how long your kilns sat hot and empty.

If you own a unit in Morbi — described as the ceramics capital of India — or in Thangadh, Himmatnagar, Bahadurgarh or Gujarat's newer sanitaryware belt, you already know the number that never sits still: yield. First-grade percentage moves two or three points month to month and nobody can tell you why. The gas bill is metered. The clay is weighed. The labour is on muster. The gap between what you formed and what you invoiced gets one line in the MIS: process loss. This article is about opening that line.

Key points

• Breakage is usually a handling problem before it is a firing problem. Green and dried sanitaryware is at its most fragile exactly when it is handled most — demoulding, fettling, transfer to the dryer, and setting onto the kiln car. A camera over those four points tells you which shift, which trolley and which pair of hands the losses cluster around.
• Post-kiln sorting is a lagging indicator. By the time a piece is graded "second" or scrapped, the cause is many hours old — often more than a full day, depending on your drying and firing cycle. Without timestamped video you are guessing. With it, you walk the clock backwards to the actual event.
• Hot-idle kiln time is real money and it is invisible. A shuttle kiln held at temperature while the next load is assembled burns gas for zero output. At an illustrative ₹3,000/hour of hot-idle gas across your kiln bank, 1.5 idle hours a day over 25 working days is roughly ₹1.1 lakh a month — a number no ERP will show you, because idle time is not a transaction. Substitute your own PNG rate and kiln count before believing it.
• Green-stage scrap is cheap; fired scrap is expensive. A piece broken before the kiln has consumed clay, mould life and labour. The same piece broken after firing has also consumed your single largest variable cost — gas. Any camera project should be scoped to catch losses before the kiln first.
• Glaze line problems show up as rework, not breakage. Spray coverage, booth overspray, and pieces set down on wet glaze are yield killers that grading catches but nobody traces. Video at the booth exit is cheap and diagnostic.
• The law changed. Write your policy against the new one. The Occupational Safety, Health and Working Conditions Code, 2020 came into force on 21 November 2025 and subsumed thirteen central labour statutes, the Factories Act, 1948 among them. Any consultant still quoting Factories Act sections at you is working from a repealed statute.

Where the loss actually hides

Walk a sanitaryware line and count the times a piece is lifted: out of the mould, onto the fettling bench, onto the dryer trolley, off the trolley, onto the glaze conveyor or turntable, off it, onto the kiln car, off the kiln car, onto the sorting table. That is nine manual contacts in the sequence above — count your own line, it will be in that region — each by a person working to a piece-rate or shift target.

Every one of those is a chance to chip a rim, stress a trap, or set a piece down on an uneven surface. The damage often does not show. It shows after firing, as a hairline through the glaze, and it gets recorded as a firing defect because that is where it was seen.

This is the specific thing video does that no other system in your plant does: it gives you a time-indexed record of physical handling. Your ERP records transactions. Your kiln recorder records temperature curves. Nothing records the moment a fitter swung a pan onto a trolley corner.

The four camera positions that pay in a sanitaryware unit

Two or three cameras per position is usually enough. This is not a surveillance grid over the whole plant.

Two things belong in the quote before a single cable is pulled. First, the PoE budget: ceramic plants have long spans between the casting hall, glaze line and kiln bank, and a switch sized only for camera count will brown out when the IR cut-in fires at night. Add up per-camera wattage, add headroom, and check cable runs against the 90 m limit before you accept a layout. Second, the mounting decision: turret (dome-style) cameras on ceiling or truss for the casting and fettling bays where you want a top-down view of set-down surfaces, bullet cameras on wall or column for the kiln door where you want a long horizontal view down the car track. Dusty, humid casting halls punish cheap cabling and cheap housings equally.

A rupee model for a mid-size Morbi unit

These are illustrative inputs. Replace every one with your own before you decide anything — the point is the structure of the calculation, not our numbers.

Assume: 3,000 pieces fired per day, 25 working days, so 75,000 pieces a month.

Now the honest part. A camera system does not recover ₹10.9 lakh. It tells you where a fraction of it lives. If video lets you cut green breakage from 3% to 2.4% and shave 30 minutes a day off hot-idle, that is roughly ₹90,000 + ₹37,500 ≈ ₹1.27 lakh a month — and that is the number to hold the project to. Run it against your actual quote using /cost/does-camera-ai-pay-for-itself-india/.

Note also that the loss value is not your selling price. A broken green pan did not cost you its retail realisation unless you were short against a confirmed order. Vendors who model breakage at MRP are selling you a fantasy.

What the owner should actually receive

Not a video wall. An owner in Morbi does not have twenty minutes a day for playback. What works is a short daily message: yesterday's breakage count by station, the two clips that matter, kiln door-open minutes, and glaze booth running time. That is the format described in /efficiency/daily-shift-report-factory-owner-india/, and it is the only format that survives contact with a working week.

The test of a system is simple: can it answer "why was Tuesday's first-grade 79% when Monday's was 84%" without you watching anything?

Where cameras do not help — and where they can hurt you

Be clear-eyed, because the ceramic sector attracts a lot of vendor optimism.

Cameras cannot see the causes of most firing defects. Body composition, slip rheology, moisture gradient in the dryer, kiln atmosphere, glaze fit — pinholes, crazing, black specks and warpage come from chemistry and thermal profile. A camera sees none of it. If your yield problem is body-glaze fit, buy lab time, not cameras.

Cameras cannot be assumed to grade pieces reliably. Vision-based crack detection on glazed white ceramic under mixed shed lighting is genuinely difficult: the specular surface works against the algorithm. Treat any promise of automated first-grade/second grading as unproven until it runs on your line, on your products, for a month, against your sorters' calls. Do not pay for it up front.

Casting halls are hard on cameras. High humidity, plaster dust, glaze mist. In glaze booth areas, specify sealed housings — IP66 or better — and budget for enclosures and periodic cleaning, or budget for early replacement.

Cameras can cost you your best people. Sanitaryware fettling and casting are skilled trades, and in a dense cluster like Morbi workers can change employer without changing city. If your workers conclude the cameras exist to catch them, your best fettler walks to the unit down the road. Tell people in advance what is recorded, why, and that clips are used to fix stations, not to dock wages — and then keep that promise. Put a written notice up; see /compliance/worker-cctv-privacy-notice-dpdp-india/.

No cameras in washrooms, changing rooms or rest areas. Ever. Not near them, not "pointed at the corridor". This is the fastest way to turn a productivity project into a labour dispute.

The compliance layer, stated correctly

There is no general Indian statute that mandates CCTV in a ceramics plant. Cameras are your commercial choice, not a legal requirement — a point we set out in full at /compliance/osh-code-2020-factory-cctv-india/.

What is live: the OSH Code, 2020 came into force on 21 November 2025 and consolidated thirteen earlier central labour laws including the Factories Act, 1948, so your safety documentation should now be drafted against the Code and the applicable state rules for Gujarat, not against pre-2025 Factories Act sections.

On the data side, the Digital Personal Data Protection Rules were notified on 13 November 2025, and the substantive obligations on data fiduciaries — notice, security safeguards, breach reporting, rights handling — are phased in, becoming fully binding on 13 May 2027. Worker footage is personal data. You have a runway, not an exemption. Use it to fix notice and retention now, while the system is being installed, rather than retrofitting later.

Also relevant to a Morbi exporter: European and Gulf buyers may raise social-audit questions about working conditions and grievance handling. Recorded, retained footage cuts both ways in that conversation — it can support you, and it can be requested. Decide your retention period deliberately, and write it down.

A 90-day way in

Days 1–15. Cameras at demoulding, fettling, glaze booth exit and kiln door only. Record. Change nothing. Do not announce targets.

Days 16–45. Every morning, take the previous day's scrap count from sorting, pick the top three defect types, and walk the video backwards. You are looking for repeat patterns by station and by shift, not for individuals.

Days 46–90. Fix the two clearest patterns — usually a set-down surface, a trolley loading habit, or the kiln loading sequence. Measure green breakage and kiln door-open minutes before and after. Compare the delta to the invoice.

If after 90 days you cannot point at a rupee number, stop. That is a legitimate outcome and a cheaper one than a five-year AMC on a system nobody opens.

FAQ

Does CCTV reduce breakage in a sanitaryware plant on its own? No. Cameras record; they do not handle pieces. The reduction comes from what you change after watching — a set-down surface, a trolley height limit, a loading sequence. A plant that installs cameras and never reviews footage should expect no yield change at all.

Can AI cameras detect cracks in glazed sanitaryware automatically? Treat it as unproven on your line until demonstrated. Glazed white ceramic is highly reflective, and hairline cracks under mixed shed lighting defeat many vision systems. Insist on a trial run against your own sorters' grading, over a full month and your full product mix, before paying for any grading claim.

Is CCTV legally required in a ceramic factory in India? No general statute requires it. The Factories Act, 1948 was subsumed by the OSH Code, 2020, in force from 21 November 2025, and neither creates a blanket CCTV mandate for ceramics. Requirements can still arrive via customer audits, export buyers or specific state rules — check those, not general law.

How many cameras does a mid-size Morbi unit need? Fewer than most quotes assume. Four functional zones — demoulding, fettling, glaze booth exit, and kiln door plus car setting — with two to three cameras each is a working starting point. Add coverage only after the first zones have produced a measurable saving.

What about kiln idle time — can a camera really measure it? Yes, and simply. A camera on the kiln door gives you door-open and door-closed timestamps, which is all you need to compute hot-idle minutes per cycle. Multiply by your own PNG rate for a hot-hold. In many units this is the fastest single payback in the whole installation.

Will workers object to cameras in the casting hall? Some will, and the concern is reasonable in a skilled-labour market like Morbi where fettlers and casters can change employer easily. Announce scope before installation, exclude washrooms and rest areas absolutely, publish a written notice, and use footage to change stations rather than to penalise individuals.