Plant-Level Energy Monitoring with IoT: Submetering & ISO 50001

Plant-level energy monitoring is one of the few IIoT projects with a payback window measured in months, not years. The math is forgiving: industrial energy bills are large, the savings are real, and the equipment to capture them is cheap relative to the bill. The trap is that most deployments stop at “we have data” and never reach “we have savings.”

Why submetering is the right starting point

A typical plant pays for energy at one or two main meters, billed by the utility. The bill tells you what the plant used; it tells you nothing about which equipment used what.

Submetering breaks the bill apart. A current transformer (CT) on each major load — a compressor, a chiller, a production line — measures real-time consumption. Aggregated, the submeter readings explain the main bill.

Once each load is measured, three things become possible:

1. Allocation — which products / cost centres / shifts use which energy
2. Anomaly detection — equipment running when it shouldn’t (HVAC at midnight, compressors during weekends)
3. Targeted improvement — focus energy projects on the actual top consumers, not assumed ones

Without submetering, energy improvement projects target the wrong equipment.

What to submeter

For a typical mid-size plant:

Tier 1 — must-have:

• HVAC system (often 30-40% of plant load)
• Compressed-air system (chronic underestimate, often 20-25%)
• Major motors and motor groups (production lines, conveyors)
• Refrigeration (in food, pharma, electronics)

Tier 2 — high value:

• Lighting (becoming smaller as LED adoption progresses)
• Each major production line individually
• IT loads (data centres, server rooms)

Tier 3 — finer granularity:

• Per-machine submetering
• Per-cell submetering
• Plug loads in offices

Start with Tier 1 — covers ~70-80% of the bill. Tier 2 adds another 15-20%. Tier 3 is a refinement, not a starting point.

Hardware choices

The submetering market is mature. A defensible 2026 stack:

• CTs: Continental Control Systems (Accuenergy, Magnelab) for accuracy class 0.5 or better. Split-core for retrofit, solid-core for new install.
• Meters: Continental Control Systems Wattnode, Veris H8035/H8036, Eaton Power Xpert, Carlo Gavazzi EM340. Modbus RTU or Modbus TCP output.
• Gateway: any Modbus-to-MQTT gateway from our gateway patterns post — Advantech, Moxa, or a custom Linux gateway running Telegraf.

For a 50-meter plant deployment, hardware BOM is typically $15k–$30k. For a 200-meter deployment, $40k–$80k. Commissioning labour is roughly 1x the hardware cost.

What to measure (the metrics that matter)

For each submetered load:

• kW — instantaneous power
• kWh — energy consumed (cumulative + per-time-period)
• kVAR / power factor — for inductive loads (motors), poor power factor is a real cost
• THD (Total Harmonic Distortion) — for sensitive equipment, indicator of power quality
• Demand — peak kW in defined windows; demand charges can be the largest line on the bill

For the plant:

• Total kWh per hour, day, month
• Peak demand per billing period
• Specific energy consumption (kWh per unit produced) — the EnPI in ISO 50001 terms
• Energy cost per unit produced — kWh × utility rate, normalised to production

The last two are the ones operations management cares about. They tie energy to business output.

ISO 50001 compliance

ISO 50001 is the energy management standard. Compliance requires:

• An Energy Policy signed by leadership
• An Energy Review identifying significant energy uses (SEUs) and their drivers
• Baselines and EnPIs (Energy Performance Indicators) — quantitative measures of energy performance
• A monitoring and measurement plan — what gets measured, by what equipment, at what frequency
• Continuous improvement demonstrated through reduction in EnPIs over time

A plant-level submetering deployment delivers most of the data infrastructure needed for ISO 50001. The standard is the framework that makes the data drive savings.

For customers in regulated sectors, ISO 50001 certification can be tied to EU directives (Energy Efficiency Directive), tax incentives, or customer contracts. The IoT deployment is the enabling layer.

The dashboards that work

Three audiences:

Operations dashboard — real-time per-load consumption, with alerts on anomalies. Compressor running with no production demand. HVAC outside scheduled hours. Demand approaching utility threshold.

Energy manager dashboard — daily and weekly trends per load, EnPI tracking, comparison to baseline. Top energy consumers ranked. Improvement project tracking.

Executive dashboard — monthly and quarterly cost trends, ISO 50001 EnPI compliance, project ROI tracking.

The same data feeds all three. The visualisation is what changes.

The 12-month payback path

A defensible plant-level energy project hits payback within 12 months by following this sequence:

Months 1-2: Install submetering on Tier 1 loads. Get raw data flowing.

Months 3-4: Establish baselines. Identify the top 5 anomalies (loads running when they shouldn’t, equipment underperforming). Fix the obvious ones — usually 5-10% savings from process discipline alone.

Months 5-6: Identify EnPIs and start reporting. Engage operations in the conversation.

Months 7-9: Run targeted projects on the top 3 consumers. Common wins: VFDs on motor groups, compressed-air leak surveys, HVAC optimisation, lighting controls.

Months 10-12: Document outcomes. Refresh baselines. Plan next year’s projects.

For a plant with a $1M annual energy bill, 5-10% savings is $50k-100k. The submetering deployment costs $30k-80k. Year 1 payback is realistic; year 2 cumulative savings dominate.

What kills these projects

Two failure modes:

1. Data without ownership. The submeters are installed, the data flows, no one owns reducing consumption. The dashboard becomes wallpaper. ISO 50001 makes ownership explicit; without the standard or equivalent governance, projects drift.

2. Improvement projects without measurement. A team installs VFDs, claims savings, doesn’t measure. Without baseline comparison via the submeters, savings are speculative. Submetering provides the measurement.

What we typically deliver

For a plant-level energy monitoring engagement:

• Submetering deployment plan with prioritised loads, hardware specs, commissioning schedule
• Modbus-to-MQTT gateway integration into the existing IoT pipeline (our Modbus post)
• Time-series storage sized for energy data (lower volume than predictive maintenance, but longer retention)
• Three-tier dashboards (operations, energy manager, executive)
• EnPI calculation engine with baseline comparison
• ISO 50001 documentation templates — energy review, EnPI tracking, monitoring plan
• Training session with the energy manager on how to use the system

If you are looking at plant-level energy monitoring — particularly with ISO 50001 in scope — we have shipped this combination across multiple sites.