01
Connect
Incomer and feeder meters · SCADA/EMS exports · equipment states · shift and job context · MGVCL bill
Stamped Energy
ITC Nadiad
ITC Packaging & Printing, Nadiad · Technical brief
Stamped Intelligence
How we help you turn plant signals into clear actions with owners
A short walkthrough of how Stamped works with meters, SCADA, shift data, and MGVCL bills: name the owner, spell out the action, attach the evidence, and check whether it paid off.
How we turn plant signals into owned actions, with proof when they work.
Load decisions
Equipment ML
Assigned actions
Verified outcomes
Load decisions
Equipment ML
Verified outcomes
Works alongside your Industry 4.0 stack. Read-only. No PLC or line-control writes.
stamped.work
The question after monitoring
You already see the plant. Who should act, and on what?
Nadiad already has monitoring, job sequencing, shift planning, and TPM. We are not rebuilding that. This brief is so your team can see how we get from a signal to a named next step, before any commitment.
Not more dashboards. A named next step someone can actually run.
What ITC already has
- Line and equipment states
- Production sequence and shift context
- OEE, downtime, and fault information
- Meter and utility data where connected
- TPM ownership and review routines
What Stamped adds
- One time-aligned view of load, process, schedule, and tariff
- A recommended action, not only an alert
- An alternative when the first action conflicts with production
- A named owner, expected impact, and due window
- Evidence showing whether the action worked
Decision and closure
Less time debating what the chart means. More time running the action and checking the bill or the curve.
Read-only on the existing plant stack
We sit on what you already have: meters, SCADA, bills, shifts
02
Build plant context
Line up signals in time and map them to how load actually flows: consumer → feeder → line → utility → equipment.
→03
Decide
Load and tariff models flag better windows. Equipment models flag drift against comparable runs at Nadiad.
→04
Assign
Each finding becomes a prescription: what to do, why, who owns it, effort, expected impact, and timing.
→05
Verify and improve
Compare the post-action signal to the locked baseline. Track what was possible vs what actually landed.
→Read-only OT · Existing systems remain systems of record · No automatic line-control changes
Pillar 1 · Load & Energy Efficiency Intelligence
Work the full load picture, not only the MD spike
Most plants react when the bill lands. We watch load against the shift plan and MGVCL ToD through the month, and call out moves that are realistic for production.
Practical load moves tied to shift plan and ToD, not only end-of-month MD firefighting.
Sequence
When the schedule allows, stagger big starts and preheats so you do not stack demand in one window.
Reduce
Call out compressors running unloaded, machines left on with no output, and utilities holding load off-shift.
Shift
Move flexible work into cheaper ToD or solar hours when gravure, offset, or utilities can still deliver.
Correct
Catch PF drift, wide pressure or temperature bands, and repeat MD hits before they show up on the bill.
Examples at Nadiad: offset makeready · gravure dryer warm-up · compressor and chiller staging · HVAC and extraction · MGVCL ToD windows
Production, quality, and safety stay first. If a lower-energy option breaks the job plan, we do not push it.
Pillar 2 · Prescriptive Equipment Intelligence
Each machine judged against its own normal, not the plant average
After we connect, models learn how each asset runs at Nadiad by shift and job. Slow drift on one compressor does not get buried in a site-wide kW number.
Per-asset baselines. Drift shows up before the monthly review.
Baseline model
Learns the expected operating band by time, shift, process state, and production context.
Drift and anomaly model
Flags sustained changes in specific power, duty cycle, startup shape, current/load behaviour, or trip frequency.
Context check
Separates production-driven change from a likely equipment or operating issue.
Prescription and verification
Recommends inspect, clean, or tune; then checks whether the curve recovered after close-out.
Typical signals
- Compressor specific-power drift at comparable header demand
- Chiller or cooling efficiency drift under similar ambient/load conditions
- Dryer fan or exhaust duty changing for matched jobs
- Repeated starts, trips, or longer ramp behaviour on connected assets
Earlier than a fixed alarm threshold. We do not claim bearing life or RUL without the right condition-monitoring data.
Example prescriptions · Equipment drift & load
A prescription tied to what is happening on the floor
Illustrative examples · not Nadiad measurementsHow a finding becomes the next-best plant decision
ML spots the issue. Then we turn it into something people can act on.
Models flag drift or tariff overlap. The agentic layer adds job window, shift state, and maintenance constraints, then drafts an action the right person can run.
Findings get context and a feasible owner, not a raw alert.
ML finding↓
Plant + production + tariff context↓
Agentic decision layer↓
Best feasible prescription + alternative
1
Detect
Baseline, drift, attribution, and tariff models produce a structured finding with evidence and confidence.
2
Add context
Read the relevant job or batch window, shift state, tariff, asset relationship, and approved maintenance knowledge.
3
Propose
Draft what to change or inspect, expected impact, owner, effort, and how you will verify it.
4
Adapt
If the first option clashes with production, suggest the next feasible window instead of sending something unusable.
5
Learn
Followed, deferred, and rejected actions improve ranking and plant preferences with human review.
Grounded in plant data and approved operating knowledge. Not a generic chatbot, and no autonomous setpoint writes.
Compressor inspection was suggested for Tuesday morning, but standby air capacity that day is too low to isolate safely. Stamped flags that and proposes Thursday 2-4 pm instead, after Job 447 closes.
Getting the action to the right person
The right action reaches the right owner, with evidence attached
At a plant Nadiad's size, one energy finding often crosses utilities, production, maintenance, and shift supervision before anyone acts. Stamped routes the prescription to whoever can actually run it.
One action card to the right owner (utilities, gravure supervisor, or maintenance), with evidence attached.
WhatsApp for assigned actions · dashboard for queue and evidence · export for management review
Workers see a short card. Leadership sees queue, constraints, expected value, and what actually landed.
A finding becomes a meeting chain: utilities → production → maintenance → shift supervisor. Four handoffs before anyone on the floor gets a clear ask.
- Map the asset to the owner (e.g. compressor → utilities + maintenance).
- Check the job / ToD window so the ask is feasible this shift.
- Pack why, impact, and evidence on one card.
- Send that card on WhatsApp (or the dashboard queue) to that owner.
Owners still decide. We remove the chase: no round-robin emails, no “who owns this?” thread. Leadership still sees the full queue and outcomes.
10:42
5G
‹
SE
Stamped Energy
online
⌕⋮
Today · assigned action · 1/2
Rx · High
Inspect Compressor 2 inlet filter and unload valve
Why: Specific power 14% above eight-week baseline for nine days
Impact: ₹45-70k/mo if curve recovers [illustrative]
Owner: Utilities lead + maintenance
Effort: ~2 hrs · isolation/permit dependent
Verify: Compare post-service kW/Air for 72 operating hours
Due: Next approved low-load window
10:42
Stamped Energy
Inbox clear for this shift
+
Message
◉
Evidence before credit
Savings count only after the signal actually moves
Issuing a prescription or marking it done is not a saving. We lock the baseline before action, watch the post-action period, and record value only when the agreed evidence clears.
Baseline locked first. Credit only when post-action evidence clears.
Baseline
Expected performance under comparable production, shift, and operating conditions.
Action
What changed, who approved it, when it was executed, and which assets were affected.
Outcome
Actual energy, demand, or equipment-performance change after adjustment for relevant context.
Ledger
Potential versus realised ₹, kWh, and reliability outcome, with source lineage and status.
Confirmation
Utility bill lines can provide secondary confirmation when the billing period closes.
| Prescription | Type | Value | Status |
|---|---|---|---|
| Compressor inlet filter | Potential | ₹55k/mo | In review |
| Gravure warm-up shift | Realised | ₹41k/mo | Verified |
| Chiller staging | Deferred | Pending | Production hold |
Prescription history · calculation method · source tags · baseline version · post-action curve · exceptions · EnMS / utility review support
Models propose. Plant execution and measured evidence determine what counts.
Designed for brownfield plants
Fits a brownfield plant: read-only, no control writes
Integration
- OPC-UA, Modbus, MQTT, API, or scheduled file export, depending on existing access
- Optional edge-buffered path when direct cloud connectivity is not permitted
- Start with meter + bill; add feeder, SCADA, and production context only where useful
Operating boundaries
- Read-only on OT in the Proof Run
- No PLC, machine, furnace, chiller, or utility setpoint writes
- Existing EMS, MES, production, and maintenance systems remain systems of record
- Human approval for maintenance, schedule, capex, or high-impact actions
Data governance
- Minimum data scope agreed before connection
- Role-based access and action audit trail
- Retention, hosting, and plant-network design agreed with ITC IT/security before deployment
- Source lineage retained for every prescription
Stamped is not an EMS replacement, MES, CMMS, vibration-monitoring system, solar/EPC service, or autonomous plant-control system.
Any first engagement should stay small enough to review technically and useful enough to produce one decision worth verifying.
A conversation, not a commitment
Happy to walk your team through how this would work at Nadiad
This brief is so you can see the approach. If it looks useful, we can sit with utilities and Industry 4.0 and map one realistic starting point. No rollout ask in that meeting.
See the approach first. Then a short working session if useful.
What we'd cover in a working session
- How energy opportunities are assigned at Nadiad today
- Which meter, SCADA, or bill data you already trust
- One utility or load area that would be a fair first look
- Who would own a prescription if one landed on the floor
What we would not ask for first
No PLC writes
Read-only on OT. Your control stack stays as-is.
No full rollout
We are not asking to wire the whole plant in a first conversation.
No long RFP
Start with a clear working session. A scoped pilot only if both sides see fit.
Keep it practical
Focus on one consumer, one utility system, and one production context if you go further later.
Suggested next step
A 30-45 minute working session with the right utilities / Industry 4.0 contacts. We walk the brief, answer questions, and leave with a clear yes / no / later on whether a scoped pilot is worth discussing.
Read-only OT
Working session first
No commitment required