Choice 1 — Elmeasure LG 6400: Class 1 default, Class 0.5S / 0.2S optional. Explicit sliding-demand feature (W/VA/VAR). Preferred here — Selec/Secure aren't well serviced by electricians on this site, Elmeasure is.
Option 2 — Selec MFM384-C: Class 1 fixed (no 0.5S/0.2S option). Same family already used at Meter 3 — standardizes spares, Modbus register map, and commissioning profile across the plant.
Both support configurable sliding demand (Elmeasure: DIP + sliding width; Selec: Modbus regs 40035 Duration + 40036 Length) — see Bill Analysis §6 for why the sliding-window method matters and what still needs confirming with TNPDCL before commissioning.
Site-installed CT today: 1600/5A toroidal (IS 2705, 15VA, accuracy class not legible on the plate) — marginal against the transformer's 1,666.76A secondary FLC (already at/above rated current at full load); needs replacing. Unaffected by the meter choice above.
Replacement — 2 options, revised to 2500/5A for real headroom: A — Rishabh RISH split-core CT (2500/5A): Rishabh's split-core datasheet doesn't confirm a Class 0.5S variant at this rating — verify the exact model and burden with Rishabh before ordering if 0.5S accuracy is required. B — Selec SCCT split-core, 2500/5A, VA15, CL 0.5, CE: same confirmed family as the 2000/5A unit — verify the 2500/5A variant carries the same class before ordering.
Demand (kVA) is read via Modbus by the IoT Gateway/EMS platform, which raises the pre-alarm in software before the 950 kVA contract limit — neither candidate meter fires a native threshold alarm on its own.
Sits on the transformer secondary — sees only the EB path, not the genset. Used as the "EB present/absent" half of the software-based source-tagging logic (see Meter 2 and Genset).
Panel-door flush mount, 92×92mm cutout
380 kVA Volvo genset, new since the last site visit. Powers the furnace only during an EB outage, via a manual changeover realized as two interlocked 630A breakers in the Master Distribution Panel (not a separate lever).
Local panel has analog V/Hz/A gauges only — no kWh/energy meter, and none is planned. Confirmed as acceptable: genset energy is captured indirectly through Meter 2 (furnace feeder), whichever source is active.
Not ordered today. Genset energy is currently captured indirectly through Meter 2 (furnace feeder), whichever source is active — see the Genset node for how EB-vs-genset source tagging works without a dedicated meter.
This node exists to make the fallback path concrete: if the software-based source-tagging inference (Meter 1 vs Meter 2 correlation + voltage-THD cross-check) doesn't hold up against real switchover events, add a dedicated meter here rather than continuing to trust the inference.
If added: same candidate family as Meter 1 (Elmeasure LG6400 / Selec MFM384-C) for spares and commissioning consistency, with its own CT sized to the 380 kVA genset's rated current (≈528A at 415V) — CT model TBD.
Revisit after the furnace+genset upgrade and after the source-tagging inference has been validated against real switchover events.
Schneider PowerLogic iEM3565 (A9MEM3565) — mV inputs accept Rogowski signal directly
Sensor: revised to a 1000A-class flexible Rogowski (was METSECTR0600U/600A) — see sizing note below. 2.4m shielded lead.
Why Rogowski: induction furnace harmonics saturate iron-core CTs; the air-core coil can't saturate
Site visit confirmed the furnace's two incomer breakers (EB-path / Genset-path) live in the Master Distribution Panel, not at the furnace — placing this meter there instead of at the furnace itself is the plan, provided both breakers converge to one common furnace feeder inside the panel (open item — needs on-site confirmation of the internal busbar).
35mm DIN-rail mount, away from radiant heat
Selec Controls MFM384-C (MFM384-C-CE)
Sensor: Selec SPCT-100/60-400/5, Class 1, 5VA — sized 400/5A or 600/5A depending on the auxiliary board's main breaker
Panel-door flush mount, 92×92mm cutout
A — Optris CTratio 2M / CSvision R1M (two-colour/ratio): 550–3500°C, robust to smoke, dust, slag and dirty optics. Higher upfront cost.
B — Tempsens A450C+ (two-colour, Indian-made): Udaipur manufacturer — local service and shorter lead time, typically lower cost. Confirmed range 600–2500°C (A450C+ variants; the A250C+/A450C+ family together spans 475–2500°C) — comfortably covers the 1610°C tap.
Sighted directly on the furnace bath — independent of Meter 2, which only measures the electrical feed.
A portable handheld IR thermometer is already in use at the ladle station today — the proposal is to reuse this existing unit rather than procure a new one. The ladle is portable, so a fixed sensor doesn't work here regardless; pointing/dipping stays manual, and the lower accuracy that comes with manual handheld use is acceptable at the ladle (vs. the fixed furnace sensor).
Because the ladle itself is portable, connecting this sensor into the PLC/SCADA system isn't plausible — it has no record/capture output, so the operator reads the value and logs it by hand. Not wired to the RS-485 bus, not captured in the EMS.
Current, nameplate-confirmed: 350kW, PF 0.91, 600A max input @415V
Planned upgrade: 500kW → estimated FLC ≈ 764A @415V/0.91 PF. Existing 630A furnace breakers and the 600A Rogowski coil are undersized against this — both flagged for resizing (see Meter 2). Vendor's post-upgrade nameplate needed to confirm final numbers.
Tap temperature 1610°C · ladle preheat 900°C (existing process control points)
Meter 2 (electrical) + both temperature sensors (thermal) instrument this stage independently.
Together they cover more than just kWh/tonne: cycle time and temperature profile per melt let you derive melt efficiency and audit rejection/quality KPIs against energy and thermal signature, not energy alone.
Phase 3 (planned): this load — and the ladle heater — will be switchable via the EMS Platform → IoT Gateway → relay/actuator control path. Relay/actuator hardware not yet installed.
Candidate for peak-hour load shedding once individually metered.
Stage 12 (Heat Treatment) furnace source is still unconfirmed — may or may not sit on this feeder.
One gateway, one RS-485 Modbus RTU bus, multi-dropping Meters 1–3 by slave ID — RTU already supports multiple devices per bus, so a second RS-485 port isn't needed for this device count. Meter 4 (genset, deferred) is provisioned for on the same bus if it's ever added.
Ladle reading is manual with the candidate device having no record/capture output
Uplinks over Ethernet to the nearest plant LAN switch, which reaches the server room / internet — no new switch hardware planned; a spare port + cable run at the Master Distribution Panel end still needs confirming.
Forwards the aggregated data to the cloud SaaS EMS platform (Phase 1). Repointing to an on-premise server later is a config change on the gateway, not a rewiring job.
Phase 3 (planned): Control loop with EMS/SCADA platform to furnace/ladle relays.
Phase 1: hosted as a cloud SaaS platform — fastest path to dashboards/alerts, no on-premise server needed to start.
Later (optional, once data-processing needs grow): migrate to an on-premise EMS server — the IoT Gateway is simply repointed to the new host; no change to field wiring or metering.
Hosts the demand pre-alarm, SEC tracking, and dashboards. Current scope is monitoring only.
Phase 3 (planned): extended to full SCADA functionality — issuing furnace/ladle heater switching commands back through the IoT Gateway to relay/actuator hardware at each load (blueprint Sections 11–12 control logic) — control loop not yet implemented.
- Confirm the Master Distribution Panel's internal busbar: do the EB-path and Genset-path furnace breakers converge to one common furnace feeder inside the panel (needed for a single Meter 2 CT to cover both sources), or does the changeover happen only at the furnace's own local panel?
- Existing incomer CT is 1600/5A (toroidal, IS 2705, 15VA) — marginal against the transformer's 1,666.76A FLC already; plan calls for replacing it with a 2500/5A, Class 0.5S unit.
- Furnace upgrade to 500kW (from the nameplate-confirmed 350kW) raises estimated FLC to ≈764A — the existing 600A Rogowski coil and 630A breakers need resizing (revised to ~1000A coil / ~800A breaker frame); confirm against the vendor's post-upgrade nameplate once available.
- Genset-vs-mains source tagging is software-inferred (Meter 1 vs Meter 2 correlation + voltage THD), not a hardware limit switch — validate against real switchover events; Meter 4 (dedicated genset meter, deferred) is the fallback if inference proves unreliable.
- Confirm a spare Ethernet port and cable run distance from the Master Distribution Panel to the nearest LAN switch, for the IoT Gateway's uplink.
- Phase 2: Meter 3's CT count and rating depend on a detailed study of the auxiliary distribution board — not yet done.
- Confirm TNPDCL's exact demand integration period and sliding sub-interval for this connection, and program Meter 1 (whichever candidate is ordered) to match or exceed that granularity — see Bill Analysis §6 for why a coarser window would understate demand relative to the utility's own bill.
- Commissioning exercise: calibrate every installed meter (Meters 1–3, and Meter 4 if it's ever added) against a portable reference-grade calibration meter before go-live — confirms accuracy class in practice rather than just on the datasheet, and gives a documented baseline for periodic re-checks.