# IIoT Energy Optimisation: A Guide for Steelmakers



> Steel mills leak energy in furnace idle time and bad scrap data. Link energy to heats and shifts to cut kWh per tonne, scrap and costs fast.
> 
> **URL:** https://www.gosmarter.ai/blog/iiot-energy-optimisation-guide-for-steelmakers/

**Date:** 2026-07-31
**Author:** BlogSmarter AI

**Categories:** blog

**Tags:** artificial-intelligence, data-strategy, digital-transformation, manufacturing, sustainability

## 



**Steel plants do not lose money on energy in one big bang. They lose it in small, stubborn leaks that sit in bad data.** [Industrial Internet of Things (IIoT)](/hubs/metals-manufacturing-glossary/#iot-and-iiot-industrial-internet-of-things) energy optimisation means you connect meters, sensors and production records. Then you tie energy use to the tonnes, heats and shifts causing the mess.

The pain is simple. **You spot waste too late**, after the furnace idled, the compressor leaked, or the scrap bin filled up.

I would not start with another dashboard. I would start with data you can trust. That is where [GoSmarter](/), built by [Nightingale HQ](https://nightingalehq.ai/), fits. It sits on top of your existing ERP, spreadsheets and email, so there is no rip-and-replace project. It uses AI optical character recognition (OCR) to read clumsy PDF mill certs, then pulls out heat and grade data. It links that paperwork to live production records. That gives steelmakers, production managers and engineers a cleaner heat-level record to set energy, scrap, and rework analysis against — the sensors and meters still do the energy monitoring.

Once meters and sensors are wired in, connecting them to that clean record is what makes the numbers useful:

-   **A live link** between energy use and production output
-   **Faster sight** of furnace drift, idle load and utility waste, once sub-metering is in place
-   **Less typing** from PDF certs and spreadsheets, with [GoSmarter](/) handling the cert side
-   **[Cleaner scrap analysis](/hubs/scrap-waste-yield-optimisation/)** by heat, grade and order
-   **Stronger reporting** for [Carbon Border Adjustment Mechanism](https://taxation-customs.ec.europa.eu/carbon-border-adjustment-mechanism_en) (CBAM) and emissions data

**Old reports tell you what went wrong. Connected data shows you while it is still costing you.** Here's how to fix it.

## Optimising Operations with AI, Edge Connectivity and IIoT - Michael Guilfoyle - ARC Industry Forum 2019

{{< youtube width="480" height="270" layout="responsive" id="242wqD7p4wo" >}}

## Get the Data Right Before You Try to Save Money

Savings start with **data you can trust**, not pretty dashboards. If you can't tie energy use to an asset, a time, and output, you can't cut waste.

### The Sensors, Meters and Connections That Matter Most

Start with **asset-level sub-metering**. A single site meter tells you the bill. That's it. It won't tell you which assets are driving the cost. Fit electricity and gas sub-meters to induction furnaces, casting lines, rolling mills, and finishing equipment. That gives you the detail you need to spot waste and peak demand [\[1\]](https://www.businesswisesolutions.co.uk/metals-manufacturing/).

Add high-frequency sensors to the assets that matter most. Critical furnaces need sensors that show internal conditions, not just total consumption. If all you can see is the final number, you're guessing.

Once you measure the right assets, tie each reading to the exact heat, order, and shift that produced it.

### How to Link Energy Data to Tonnes, Heats, Grades, Shifts and Production Runs

Energy data only matters when you link it to tonnes, heats, grades, shifts, and runs. A spike in kWh tied to a heat number, grade, shift, and scrap mix tells you something useful. It shows whether the product actually needs more energy, or whether that run went off the rails.

Without that link, you get a plant average. Plant averages hide the mess. They hide which grades cost more to make, which shifts run badly, and which routes burn more energy per tonne than they should.

Link energy timestamps to job IDs, heat codes, and order data. Then you can compare energy per tonne across grades and routes. You can also support [Carbon Border Adjustment Mechanism (CBAM) and Environmental, Social and Governance reporting](/blog/lifecycle-carbon-tools-cbam-compliance/) [\[5\]](/hubs/gosmarter-for-metals-operations/)[\[3\]](/tags/sustainability/).

Poor scrap definition skews energy intensity too. A typical electric arc furnace needs around 440 kWh per tonne of steel, against a theoretical minimum closer to 300 kWh per tonne, and well-prepared, well-densified scrap is what closes that gap [\[7\]](https://www.advancedenergy.com/en-us/about/news/blog/improving-scrap-steel-efficiency-with-arc-furnace-optimization-techniques/).

### Where [GoSmarter](/) Helps Clean Up the Data Before Analysis Begins

{{< image src="1a2522ef04568f801ea774469c6af8d8.jpg" alt="GoSmarter dashboard showing mill certificate data linked to production records" >}}

Before you analyse anything, you need clean production data. [GoSmarter](/), built by Nightingale HQ, pulls heat numbers, grades, and chemical compositions from PDF or scanned mill certificates using [AI OCR](/features/mill-certificate-reader/). That means your team can stop typing data out by hand.

It then links certificate data to inventory records and orders. You end up with a cleaner data layer for analysing energy intensity by product and heat [\[5\]](/hubs/gosmarter-for-metals-operations/). The same heat-number record also feeds the [Cutting Optimiser](/features/cutting-optimiser/) and [Metals Manager](/features/metals-manager/), so one clean record powers scrap, inventory and energy analysis together.

With that sorted, the plant can move into the use cases that cut energy fastest.

## The Use Cases That Actually Cut Energy in a Steel Mill

{{< image src="6a6be7c42f6aeb480bfeb109-1785460763634.jpg" alt="Manual vs AI-Driven Steel Mill Energy Optimisation: Key Metrics" >}}

Once your data stops fighting you, the quickest savings usually show up in **furnaces, utilities and scrap**. That's where mills burn cash every shift, often in plain sight.

### How Furnaces Stop Wasting Power When You Have Proper Visibility

Live visibility gives operators something they rarely get from old reporting tools: **a chance to act while the furnace is still running**.

You can tune gas flow as conditions change. You can cut coke use in real time. You stop treating furnace energy like a monthly post-mortem and start treating it like an operating control. That matters, because by the time a report lands in your inbox, the waste has already happened.

The same logic applies to other high-energy furnaces. Link sensor data to production data and the waste shows up fast. You see which jobs, shifts or settings burn more than they should. Then you fix them.

### How Mills and Utilities Lose Money Every Shift

The same visibility also exposes waste in mills and utilities, where small losses quietly repeat every shift.

Put energy sensors across induction furnaces, casting and rolling mills, and you can spot **peak-demand waste** and **mechanical inefficiencies** before they turn into normal practice. In one case, installing 80 Energy Insight sensors uncovered optimisation opportunities worth **GBP780,000** in potential savings and **2,546 tonnes of CO2** savings [\[1\]](https://www.businesswisesolutions.co.uk/metals-manufacturing/).

Off-gas is another lever that too many plants waste. One steel plant redirected **300,000 Nm3/h of blast furnace gas** that had been flared. That cut coal use by about **45% at full load** [\[10\]](https://www.linkedin.com/posts/steel-technology-b2b_steelindustry-energyefficiency-costoptimization-activity-7417099977421635584-LEFe).

### Why Better Scheduling and Lower Scrap Also Cut Energy

Scrap hits energy twice. You waste energy making bad cuts now, then you burn more energy re-melting that material later.

Every tonne of scrap adds avoidable cost and drags down yield [\[8\]](/hubs/cutting-optimiser/). That's why cutting plans matter more than most people admit. [AI vs. spreadsheets](/blog/ai-vs-spreadsheets-smarter-production-planning/) for planning might look cheap, but they often hide a nasty energy bill.

AI-driven cutting plans can reduce long product scrap from an industry average of **5-8%** to **under 2.5%** [\[8\]](/hubs/cutting-optimiser/). In a two-week trial, [Midland Steel](https://midlandsteelreinforcement.com/) applied AI cutting plans across **193 jobs** and **734 tonnes of steel**. That cut scrap by **50%** compared with manual planning [\[8\]](/hubs/cutting-optimiser/). Less scrap means less energy burned on steel that never leaves the yard as finished product.

| The Manual Way | The Automated Way |
| --- | --- |
| Planners use spreadsheets and intuition, often resulting in 5-8% scrap rates [\[8\]](/hubs/cutting-optimiser/) | AI algorithms evaluate billions of combinations to reach under 2.5% scrap rates [\[8\]](/hubs/cutting-optimiser/) |
| Offcuts are often unrecorded, leading to "bone yard" waste and over-ordering [\[9\]](/tags/production-planning/) | Every remnant is automatically flagged, labelled, and tracked in live inventory [\[8\]](/hubs/cutting-optimiser/) |
| Grade substitution errors occur due to manual tracking of mill certificates [\[8\]](/hubs/cutting-optimiser/) | Systems enforce grade compliance as a hard constraint, preventing mixed-grade errors [\[8\]](/hubs/cutting-optimiser/) |

[GoSmarter](/), built by Nightingale HQ, [links mill certificates to live inventory](/hubs/mill-cert-automation/) and orders, so planners work from verified material and avoid grade mix-ups.

Once these use cases are proven, the next step is a controlled rollout with a clear baseline.

## How to Roll This Out Without Disrupting the Plant

Now comes the hard part: sequencing. You need to move from current-state data to a live IIoT programme without knocking production, safety or finance off course.

### Start With a Baseline You Can Stand Behind

Once you've found the main loss drivers, prove the baseline before you touch the plant.

Pull together 12-24 months of historical energy data [\[6\]](https://advantageutilities.com/government-support-energy-intensive-industries). That gives you enough history to split seasonal swings from plain waste. It also lets you set a baseline for specific energy consumption, peak demand, idle losses and utility waste by area. Normalise energy against tonnes per heat, shift and line. Then use the same method across every comparison. If you change the maths halfway through, the baseline turns into fiction.

Once the baseline is in place, rank your quick wins. Start with utilities. Then move to major furnaces. After that, tackle rolling and finishing. Check [British Industry Supercharger](https://assets.publishing.service.gov.uk/media/64fac5879ee0f2000db7c082/energy-bill-policy-statement-energy-intensive-industries.pdf) eligibility early, because it changes pilot payback.

### Pick a Pilot That Pays Back Fast

Once the baseline looks solid, pick the smallest area with the fastest payback.

Utilities are the obvious first pilot. **Automating electricity metering and reconciliation alone can break even within three months** [\[4\]](https://metallurgprom.org/en/news/russia/16746-po-robot-pozvolilo-severstali-optimizirovat-rashody-na-jelektrojenergiju-i-snizit-sebestoimost-gotovoj-produkcii.html).

Then move to your highest-energy furnaces. [Tata Steel](https://www.tatasteel.com/)'s Topscan at Blast Furnace 5 shows what good looks like: radar scans every 10 seconds to give live furnace visibility without stopping production [\[2\]](https://www.newsteelconstruction.com/wp/tata-steel-makes-energy-and-co2-savings-with-digital-technology/). The technology is expected to cut CO2 emissions by at least 50,000 tonnes a year and save millions in coke costs [\[2\]](https://www.newsteelconstruction.com/wp/tata-steel-makes-energy-and-co2-savings-with-digital-technology/).

Assign ownership on day one. Split accountability across Operations, Maintenance, Information Technology / Operational Technology (IT/OT) and Finance. If energy savings live only in the engineering team's spreadsheet, they'll die the moment the next budget round starts.

Use the pilot to prove fast payback in utilities first. Then scale to furnaces once the data and ownership model hold up.

Next, lock down cyber, safety and operator roles before you scale past the pilot.

### Keep Cyber Risk, Safety and Operator Pushback Under Control

A pilot only scales when you sort out Operational Technology (OT) boundaries, safety review and operator workflow from day one.

Keep OT network boundaries clear. Use digital sign-off for automated reporting and reconciliation [\[4\]](https://metallurgprom.org/en/news/russia/16746-po-robot-pozvolilo-severstali-optimizirovat-rashody-na-jelektrojenergiju-i-snizit-sebestoimost-gotovoj-produkcii.html). For closed-loop changes, where the system adjusts a furnace parameter on its own, a Hazard and Operability Study review is mandatory. If you let automated adjustments run without a formal hazard review, you create safety and liability exposure that no energy saving can defend.

If your IT team asks about GoSmarter's own footprint: it connects over a REST API with OAuth 2.0 or Microsoft Entra single sign-on (SSO), keeps your data on UK Azure infrastructure, and never trains its models on your data.

Train shift teams on the dashboard, what each alert means and what action you expect before the recommendation expires.

## What to Measure and Review After Go-Live

Once the pilot goes live, your job changes. You stop installing things and start controlling them. The point now is simple: **check that the savings stick**.

Track **specific energy consumption** in kWh per tonne by line, shift and product route. Then track **GBP per tonne** alongside it, so you can see whether lower energy use is doing anything for your costs. If gas matters on your site, log use in **therms or Nm3 per heat**. That gives furnace teams a number they can act on, not just another line in a report [\[1\]](https://www.businesswisesolutions.co.uk/metals-manufacturing/).

For furnaces, track **heat time** with **tonnes per heat**. Shorter cycles sound good on paper. What matters is whether they push out more steel, not just finish faster.

Stick to measures that show whether **energy, cost and output move together**.

| Metric | Why it matters | Typical improvement |
| --- | --- | --- |
| Specific energy consumption (kWh/t) | Overall energy efficiency by line and shift | Moves toward the theoretical minimum of 300 kWh/t, against a typical 440 kWh/t draw, as scrap density and preheating improve [\[7\]](https://www.advancedenergy.com/en-us/about/news/blog/improving-scrap-steel-efficiency-with-arc-furnace-optimization-techniques/) |
| GBP per tonne | Financial impact of energy changes | Falls in step with specific energy consumption once tariff and demand-charge exposure are tracked alongside it |
| Heat time | Furnace cycle efficiency | Falls as scrap density, charge sequencing and furnace visibility improve |
| Scrap rate | Material and energy waste in cutting and rolling | Up to 50% reduction with AI optimisation [\[3\]](/tags/sustainability/)[\[5\]](/hubs/gosmarter-for-metals-operations/) |
| kg CO2e per tonne | Emissions intensity by product route | Scrap-based (EAF) steel uses around 72% less energy and produces around 58% less CO2 per tonne than steel from virgin ore [\[7\]](https://www.advancedenergy.com/en-us/about/news/blog/improving-scrap-steel-efficiency-with-arc-furnace-optimization-techniques/) |
| Peak demand | Exposure to high-tariff demand charges | Reduce avoidable peaks by shifting flexible loads [\[1\]](https://www.businesswisesolutions.co.uk/metals-manufacturing/)[\[6\]](https://advantageutilities.com/government-support-energy-intensive-industries) |

**Scrap is where cash goes to die.** Every tonne of scrap carries raw material, energy and transport cost you've already paid once. Then you pay again to re-melt it or get rid of it.

Track **kg CO2e per tonne** by route for reporting and comparison [\[3\]](/tags/sustainability/).

### How to Turn Weekly Reviews Into Ongoing Savings

Run a weekly exception review with Operations, Maintenance, Energy Management and Finance. Keep it blunt: **what moved, why did it move, and who fixes it**. Put one site energy manager in charge of the weekly numbers and the review rhythm [\[1\]](https://www.businesswisesolutions.co.uk/metals-manufacturing/)[\[10\]](https://www.linkedin.com/posts/steel-technology-b2b_steelindustry-energyefficiency-costoptimization-activity-7417099977421635584-LEFe).

When you confirm and fix a deviation, reset the baseline. Write down the root cause. Write down the corrective action. Then use that as the benchmark for the next review cycle.

Once a pilot shows a clear break-even case and hits the agreed operating benchmarks, scaling gets a lot easier. Successful pilots have reached break-even in as little as **3 months**. Sites usually justify the next roll-out through steady cuts in **specific energy consumption, heat time, scrap and emissions intensity** [\[4\]](https://metallurgprom.org/en/news/russia/16746-po-robot-pozvolilo-severstali-optimizirovat-rashody-na-jelektrojenergiju-i-snizit-sebestoimost-gotovoj-produkcii.html). If the weekly review keeps showing gains, the next move is a focused pilot on live production data.

## Run a GoSmarter Pilot on Your Next Batch of Certs and Production Data

When weekly reviews keep flagging the same losses, stop writing another report about it. Prove what's going on with the certs and heat records already linked to those losses.

Start with the paperwork you've already got. If your next batch of certs sits in a shared drive, use that.

[GoSmarter's MillCert Reader](/features/mill-certificate-reader/), built by Nightingale HQ, reads [PDF mill certificates](/docs/digitising-mill-certificates/) with AI optical character recognition (OCR) and digitises them in seconds. Then you link that data to heats, batches and production runs. No more squinting at clumsy PDFs or keying the same data in by hand.

The upside is not vague. In a two-week trial, Midland Steel processed 193 jobs and 734 tonnes of steel. It cut scrap by 50% and pushed the rate below 2.5% [\[8\]](/hubs/cutting-optimiser/). **That sort of scrap cut pays back fast.**

## FAQs

{{< faq question="What is IIoT energy optimisation in a steel plant?" >}}
IIoT energy optimisation in a steel plant means you use connected sensors, digital models and live analytics to watch and control energy-hungry jobs like smelting, casting and rolling.

That gives you a clearer view of where energy goes and when peak demand bites. Then you can fine-tune the plant, cut waste, slash electricity and gas costs, and trim carbon emissions without hurting output quality.
{{< /faq >}}

{{< faq question="What data do we need before starting an energy project?" >}}
Before you do anything else, set a **baseline**. Track total energy use across sites, machines, and production lines.

Pull in data from every source that uses or supplies energy. That includes bought-in energy, on-site generation, and process by-products. If you only look at the headline number, you miss the mess underneath.

Granular data shows you:

-   **peak demand**
-   **waste**
-   energy use per unit of output
-   where the money goes
-   the likely cost and emissions impact

This matters because energy bills do not just come from one bad month or one overpriced contract. They come from small losses, hidden spikes, and kit that quietly burns cash all day.
{{< /faq >}}

{{< faq question="How quickly can a steel mill see payback from a pilot?" >}}
A steel mill can often see payback from an **IIoT energy pilot** in under four months.

That is not sales fluff. Automated electricity reconciliation projects have broken even within three months [\[4\]](https://metallurgprom.org/en/news/russia/16746-po-robot-pozvolilo-severstali-optimizirovat-rashody-na-jelektrojenergiju-i-snizit-sebestoimost-gotovoj-produkcii.html), and GoSmarter's own Midland Steel trial cut scrap by 50% in two weeks — less scrap means less energy burned re-melting steel that never leaves the yard.
{{< /faq >}}