Gujarat Met Coke & Energies

Executive Summary

ThirdEye Data AI proposes to make Gujarat Met Coke & Energies (GMCE) an AI-run plant from the first day of operation. Cameras with AI, sensors on the machines and one plant system will follow every order, every truck, every tonne of coal and every oven, from the moment an order arrives to the moment the coke truck leaves the gate.

1.1 What we are proposing

• A plant run by its orders. Every order sets the coal blend, the coal to buy, and the oven plan, so the plant makes what its buyers ordered.
• AI that sees. Cameras read number plates, count trucks and charges, confirm the right coal goes into the right hopper, record every oven push, check loading, and guard the plant at night with video evidence.
• AI that keeps machines running. Sensors on crushers, conveyors, stampers, and finishing machines feed AI that warns days before a breakdown, alongside a planned maintenance schedule.
• One system, one screen. One plant ERP records every order, weight, test and dispatch. One simple screen shows the plant head everything, live, with forecasts.
• One partner. ThirdEye Data supplies, installs, and supports all hardware and software, with one point of contact.

1.2 One partner for everything

This is the heart of what ThirdEye Data offers GMCE. A digital plant usually needs a camera vendor, a sensor supplier, a network contractor, an ERP company, an AI team and a support agency, and the plant head ends up managing all of them. With ThirdEye Data, GMCE deals with one partner for everything.

Swipe sideways to see the whole figure

• One contract, one contact. One contract and one ThirdEye Data program manager for cameras, sensors, network, AI, ERP and dashboards.
• One team accountable. If something does not work, there is no argument between a camera vendor, a network contractor and a software company. ThirdEye Data owns the fix.
• Built to work together. Every device and system is chosen and designed to work together from day one: one data hub, one login, one screen.
• Faster to start. One team plans, supplies and installs everything, so Phase 1 can be ready in about 12 to 15 days from the order.
• One number to call. Warranties are held by ThirdEye Data, with one helpdesk for every issue and a resident engineer for the first 3 months.
• Grows with you. New cameras, sensors and use cases plug into the same platform as the plant grows.

1.3 How to read this proposal

Section 2 shows how your plant works today. Section 3 explains how met coke is made to order. Section 4 shows where AI fits. Section 5 tells the story of one order, chapter by chapter, and what the AI and software do at each step. Sections 6 and 7 list every use case and the exact solution against each. The rest covers the future process, the plant command centre, hardware, phases and support.

ThirdEye Data’s Understanding of Your Plant

GMCE is restarting a two-line, 328 (approx) oven, stamp-charged met coke plant that has been idle for about 7 years. The machines are in place and being repaired. Start-up is expected in 15 to 30 days, and today there is no CCTV, network or computer system.

2.1 What GMCE asked us for

• Consultant. Walk the plant, understand the process and recommend the full set of use cases.
• Hardware and software partner. Supply, install and run cameras, sensors, network, AI, ERP and dashboards.
• Single point of contact. One accountable team for every device, system and support call.

2.2 Your plant at a glance

Table 1. Plant equipment as observed on our visit. Each line has 4 (approx) units of 41 (approx) ovens.

2.3 How your plant works today

This is the process as we saw it on our visit: coal comes in by truck, is blended, stamped, cooked, cooled, checked, sized and sent out by truck.

Swipe sideways to see the whole figure

Table 2. The 9 steps and what is missing without digital systems.

2.4 The heart of the plant: 328 (approx) ovens

Every oven moves through five states about once a day. 7 (approx) stampers and 3 (approx) finishing machines serve all 328 (approx) ovens, so the schedule is tight. Knowing the state and cooking time of every oven, live, is the most valuable view the plant head can have.

Swipe sideways to see the whole figure

How Met Coke Is Made to Order

A met coke plant makes to order. The buyer's specification decides the coal blend, the blend decides the coke, and the plant proves the result before it dispatches. Demand drives supply.

Swipe sideways to see the whole figure

3.1 Coal grades and blending

Coking coal is bought in grades. Indian coking coal is graded by ash content, and every lot is also tested on arrival for moisture, ash, volatile matter and sulphur.

Table 3. Indian coking coal grades by ash content (Annex B, 1).

No single coal usually matches an order, so plants blend. The mix is worked out so that ash, volatile matter and sulphur land within the target; a volatile matter of about 26 to 29% in the blend is generally considered good for coking (Annex B, 1). Stamp charging, which GMCE uses, presses the blend into a dense cake, which lets plants use lower-cost coking coals and still make strong coke (Annex B, 2).

Swipe sideways to see the whole figure

3.2 What buyers ask for

Table 4. Example low-ash met coke specification from an Indian supplier (Annex B, 3). GMCE's own grade definitions are to be confirmed.

Where AI Fits in Your Plant

ThirdEye Data is an AI company, and AI is at the centre of this proposal. It works alongside sensors and software, and each does a different job. Here is what each one does, in plain words, and where it sits in your plant.

4.1 Four kinds of help

Swipe sideways to see the whole figure

• AI that sees (computer vision). Cameras connected to AI do the watching, reading and counting that people cannot do round the clock.
• AI that predicts. Models that learn from your plant's own data warn of breakdowns, forecast output and check the coal blend before ovens are charged.
• Sensors that measure (IoT). Small devices on machines, hoppers and pipes send readings every few seconds. They are the eyes and ears for the predicting AI.
• Software that remembers (ERP and dashboards). The plant ERP keeps every record in one place, and the command centre shows it simply.

4.2 Where the AI works

Swipe sideways to see the whole figure

The numbers on the picture show where AI works in your plant. In a line or two each:

• 1 · Gate. Cameras read every number plate, count trucks in and out, and match each truck to its order and coal grade.
• 2 · Two weighbridges. Weight goes straight into the system with photos of the truck and plate; AI flags anyone standing on the platform or a truck weighed twice.
• 3 · Coal shed. AI checks that each truck tips into the right hopper for its grade, and heat cameras watch the coal heaps for fires.
• 4 · Crushers and belts. Cameras spot stones, metal and belt damage, and AI warns days before a crusher fails.
• 5 · Stampers. Cameras read the oven number at every charge, count charges, and sound a siren when a person is near a moving stamper.
• 6 · Oven batteries. AI records every push, spots door smoke, under-cooked coke and hot spots on the oven walls.
• 7 · Cooling house. Heat cameras check that no hot spot is left before coke is loaded.
• 8 · Screening and sizing. AI measures coke size by camera on the belt, and the digital lab issues the test certificate.
• 9 · Loading and exit. AI checks the right grade is loaded and the truck is covered, reads the plate, and sends the exit weight to the invoice.
• 10 · Whole plant, 24 × 7. AI raises intrusion and theft alerts with video evidence, checks safety gear, watches for fire and marks attendance by face.

The Story of One Order

The best way to see what the AI does is to follow one order through the plant. A foundry buyer orders 500 tonnes of Grade 2 met coke, 40 mm, for delivery in 10 days. Here is what happens next, chapter by chapter.

Swipe sideways to see the whole figure

5.1 Chapter 1: The order arrives

Swipe sideways to see the whole figure

A foundry buyer orders 500 tonnes of Grade 2 met coke, 40 mm size, for delivery in 10 days. The sales team enters it once in the order book. Within seconds the plant knows what is needed: how many ovens to charge each day and how much coal of each grade.

5.2 Chapter 2: Planning the blend

Swipe sideways to see the whole figure

The blend planner looks at the order's limits for ash, sulphur and strength, and at the lab results of the coal lots in stock. It recommends 25% Coal B and 75% Coal C. Before a single oven is charged, AI forecasts the coke this mix will give and confirms it should meet Grade 2. The lab still tests the coke after it is made; the forecast simply warns early, while the mix can still be changed.

5.3 Chapter 3: Buying the right coal

Swipe sideways to see the whole figure

The plan needs more Coal C than the shed now holds. AI raises a reorder alert days in advance, and purchase books coal from the supplier with the best quality record for that grade. Looking further ahead, AI also forecasts which grades buyers are likely to order in the coming weeks, so the right coal is in stock before the orders arrive.

5.4 Chapter 4: The coal truck reaches the gate

Swipe sideways to see the whole figure

The truck arrives at 7:42 in the morning. The gate camera reads its number plate, AI counts it as coal truck 14 of the day, and the system matches it to the purchase order for Coal C. The barrier opens and the truck's journey through the plant is now being timed.

5.5 Chapter 5: On the weighbridge

Swipe sideways to see the whole figure

On one of the plant's two weighbridges, the weight flows straight into the system with photos of the truck and its number plate. AI checks that nobody is standing on the platform and that the truck is fully on it. The weigh slip prints by itself.

5.6 Chapter 6: Unloading in the shed

Swipe sideways to see the whole figure

A sample goes to the lab and the result is stored against this truck. As the truck tips its load, AI confirms it went into the Coal C hopper. The coal lot is now tracked by grade: from here it can be followed all the way to the buyer's truck. Heat cameras keep watch over the heaps for self-heating.

5.7 Chapter 7: Blending and crushing

Swipe sideways to see the whole figure

The hoppers release coal by weight, and the belt weigher confirms the mix really is 25 : 75. Cameras over the belt spot a piece of metal and stop it before it reaches the crusher. Sensors on the crushers report their vibration and temperature every few seconds.

5.8 Chapter 8: Stamping and charging the ovens

Swipe sideways to see the whole figure

The stamper presses the blend into a cake and charges oven B-117. Its camera reads the oven number, so the oven tracker updates itself and the charge is counted for the shift. When a worker steps onto the track, a siren sounds before the machine moves.

5.9 Chapter 9: Cooking and pushing

Swipe sideways to see the whole figure

Oven B-117 cooks for about 24 hours while the tracker counts down. When it is ready, the finishing machine pushes the coke out and the push is recorded against the oven. AI watches for heavy smoke at push, a sign of under-cooking, and for smoke leaking from doors.

5.10 Chapter 10: Cooling

Swipe sideways to see the whole figure

Water sprays the hot block in the cooling house. Before loading, a heat camera checks for any remaining hot spots, so no glowing coke reaches the trucks or belts.

5.11 Chapter 11: Quality check and sizing

Swipe sideways to see the whole figure

The lab tests the batch and records the results against oven B-117, the blend and the coal lots behind it. The coke is broken to 40 mm and a camera on the belt measures the size mix. The batch passes Grade 2, and a test certificate is ready.

5.12 Chapter 12: Loading and dispatch

Swipe sideways to see the whole figure

The coke is reserved for this order. At the loading bay, AI confirms the truck is loaded from the 40 mm hopper and covered with a tarpaulin. At the exit, the number plate is read again and the truck counted out, the exit weight flows into the challan and invoice, and the buyer receives the dispatch details and test certificate. The order closes.

5.13 Chapter 13: Round the clock: guarding the plant

Swipe sideways to see the whole figure

At 2:14 at night, a person climbs the north boundary wall near the coal shed. AI detects it within seconds and sends the security supervisor a phone alert with a photo and a short video clip. The same cameras check safety gear on the day shift, watch for fire, and mark attendance by face.

5.14 Chapter 14: Keeping machines healthy

Swipe sideways to see the whole figure

Over three weeks, the vibration on crusher 2's bearing slowly rises. AI notices the change long before anyone could hear it and warns maintenance to replace the bearing at the next planned stop. Preventive servicing runs to a schedule, and every job is recorded.

5.15 Chapter 15: The plant head's day

Swipe sideways to see the whole figure

At 8 in the morning the plant head opens one screen: orders, coal by grade, all 328 (approx) ovens, coke made, quality, machine health and alerts. The plant head asks, in Hindi, how much Grade 2 coke went out this week, and gets the answer straight away.

The Complete List of Use Cases

There are 67 use cases in all, in the order of the story. 37 of them use AI directly. Each is described in plain words; the next section gives the exact solution behind each one.

Table 5. All use cases. Type: AI vision = cameras with AI; AI prediction = models that learn from plant data; Sensors = measuring devices; Software = ERP and dashboards.

Our Solution Against Each Use Case

For every use case, this is exactly what ThirdEye provides: the AI or software, the hardware, and the phase in which it goes live.

Table 6. Solution, hardware and phase for every use case. Phase 1: 25 use cases before start-up; Phase 2: 23; Phase 3: 11; Phase 4: 8.

The Future Process: Where Every Solution Plugs In

In the future plant, the order plans every step, and each step gets the AI, sensors and software that close its gaps. Everything lands in one plant data hub that feeds the command centre and the alerts.

Swipe sideways to see the whole figure

The Plant Command Centre and ERP

The plant head asked for something simple. The command centre shows the whole plant on one screen: a few large numbers, colour only where action is needed, and a tap to open the detail behind any number.

9.1 What the plant head sees, live

The next page shows the plant head's main screen. It refreshes by itself every few seconds from the cameras, sensors, weighbridges and tablets, so nobody has to type numbers in or phone around for them.

• Six big numbers at the top. Coke made against plan, coke dispatched, ovens cooking, orders on time, days of coal left and today's alerts.
• Demand against supply. For every open order, by coke grade and size: how much is dispatched, how much is made and in stock, and how much is still to make. AI marks any order at risk of being late.
• Coal coming in. Coal in stock by grade with the days it will last, an AI warning when a grade will run short, trucks in and out read by number plate, and the status of both weighbridges.
• Bharat zone and Gujarat zone. Every one of the 164 (approx) ovens on each line, coloured by state: empty, charging, cooking, ready to push or overdue. Charges and pushes against plan, average cooking time, and the status of each stamper.
• Alerts with evidence. Each alert comes with a photo or video clip as evidence and also goes to the phone of the person responsible.
• Flow, machine health and questions. Today's flow from coal received to coke dispatched, the health of the main machines, and a box where the plant head can ask a question in English or Hindi.

Swipe sideways to see the whole figure

9.2 The ERP behind the screen

Every number on the screen comes from one plant ERP. These are its modules.

Table 7. ERP modules. Accounts data is passed to GMCE's accounting software so nothing is entered twice.

How It All Fits Together

Cameras, sensors, the 2 weighbridges and tablets in every section feed one platform in an on-site control room. The ERP, the command centre and the alerts all read the same data.

Swipe sideways to see the whole figure

• Runs on site. AI runs on servers in the plant, so gate, weighbridge and security alerts keep working if the internet drops.
• Built for a coke plant. Every device is rated for dust, heat and steam; fibre is used over long runs near heavy motors.
• Secure. Cameras, sensors and office systems sit on separate networks, with role-based logins and a record of every change.
• Ready to grow. New cameras, sensors or use cases plug into the same platform without redesign.

Hardware We Supply

The full scope is about 95 cameras, about 65 machine-health sensors and an on-site control room. GMCE's 2 existing weighbridges are kept and connected. Quantities are indicative until a 2-day placement survey.

11.1 Cameras and vision

11.2 Sensors and instrumentation

11.3 Network, compute and control room

Table 8. Indicative hardware. All cameras will be certified under India's current CCTV security requirements.

Implementation Roadmap

Phase 1 puts the essentials in place before start-up. The rest follows over about six months, each phase starting once GMCE signs off the one before. No fixed date is set: start-up is expected in 15 to 30 days, and Phase 1 needs about 12 to 15 days from order to readiness.

Swipe sideways to see the whole figure

Table 9. Phases and gates.

One Partner, End to End

GMCE deals with one ThirdEye program manager for everything: hardware, software, installation partners, warranties and support.

• Resident engineer on site for the first 3 months after start-up.
• Helpdesk: critical issues (weighbridge, gate or ERP down): response within 1 hour, on site within 4 hours.
• Annual maintenance contract from the end of warranty: device checks, cleaning in dusty areas, spares and updates.
• Monthly review with the plant head on alerts, AI accuracy and the next use cases to switch on.

Why ThirdEye Data

• AI first. Computer vision, predictive maintenance and data platforms are our core work, not add-ons to a hardware sale.
• Proven in the field. We have built AI counting and inspection systems for warehouses and factories, and predictive maintenance solutions for industrial equipment.
• One team for everything. Cameras, sensors, network, AI, ERP and dashboards come from one team, so nothing falls between vendors.
• Built for your plant. Our AI models are trained and tuned on GMCE's own cameras and data, and keep improving as the plant runs.
• Simple to use. Plain-language screens, training in Hindi and English, and an AI assistant that answers questions in either language.

Assumptions and Information Needed

15.1 Assumptions

• Equipment counts are as observed on our visit; small differences do not change the design.
• Start-up is expected in 15 to 30 days; Phase 1 is ready for start-up if the order is placed about 12 to 15 days before it.
• GMCE provides power at each cabinet location and an air-conditioned control room.
• Mounting sensors on machines does not affect their warranties; we confirm this machine by machine during the survey.
• AI prediction needs about 3 months of plant data; early-warning alerts work before that.
• Pricing is not part of this document; a commercial proposal follows once scope is agreed.

15.2 Information we need from GMCE

• Make and model of the 2 weighbridges and any weighbridge software
• GMCE's coke grade definitions and key buyers' specifications
• Coal grades and suppliers GMCE plans to buy, and the lab tests run on each lot
• Number of coal hoppers in the shed
• Make and model of the stampers, finishing machines and crushers, and any control panels we can read
• Existing oven temperature instruments, if any
• Whether the plant has a waste-heat or power section
• Accounting software in use and the invoicing process
• Shift pattern, headcount per shift and contract labour agencies
• Site drawing and boundary length
• The confirmed start-up date and the name of GMCE's plant coordinator

Next Steps

Five steps take GMCE from this proposal to Phase 1 ready at start-up.

1. Walk GMCE's leadership and plant head through this proposal, agree the Phase 1 scope and confirm the start-up date.
2. ThirdEye runs a 2-day placement survey and signs off the camera, sensor, fibre and control room layout.
3. ThirdEye sends the commercial proposal; GMCE places the Phase 1 order.
4. About 10 days of Phase 1 installation and ERP setup, while GMCE shares master data.
5. Phase 1 live and tested before plant start-up; Phase 2 starts with operations.

Words Used in This Proposal

Sources

1. Coke (fuel), Wikipedia: coking coal criteria, blend volatile matter, Indian coking coal grades by ash: https://en.wikipedia.org/wiki/Coke_(fuel)

2. India's Shift to Stamp Charging, Commodity Insights (March 2025): https://www.commodityinsights.com.au/news-article/indias-shift-to-stamp-charging

3. Low Ash Metallurgical Coke specification, METSIL: https://metsil.in/lam-coke.php

4. Coke Oven waste heat recovery, Thermal Systems: https://thermalindia.com/application/coke-oven-cdq/

Plant facts come from ThirdEye's site visit to GMCE in November 2026.