Latest / The Operator's Manual with Fexingo: Daily Decisions, Tactical Wins, and Business Execution / How One Steel Mill Cut Energy Costs 30 Percent With One Sensor Change
Transcript
- Lucas: Luna, how much do you think a typical steel mill spends on electricity every year? Luna: I'd guess tens of millions. These electric arc furnaces are basically lightning bolts contained in a vessel. Lucas: Right. Nucor's Crawfordsville mill in Indiana was running about $45 million a year in electricity costs. That's the biggest single line item after raw materials. And they cut it by 30 percent with one sensor change. Luna: Thirty percent off $45 million — that's $13.5 million a year. From a sensor? Lucas: Exactly. The sensor monitors off-gas temperature coming out of the furnace. Here's the physics: an electric arc furnace melts scrap steel by passing current through electrodes. The arc creates intense heat, but a lot of that heat escapes as hot gas. If the off-gas temperature is higher than needed, you're wasting energy. If it's lower, you're not melting fast enough. Luna: So they were overdriving the furnace because they couldn't see the waste. Lucas: Exactly. Before the sensor, operators set the power level based on a recipe — standard amps and volts for a given scrap mix. They'd run at that level for the whole melt cycle. But scrap density varies, moisture varies, chemistry varies. So sometimes they were pouring in way more energy than needed. Luna: And the off-gas temp was the signal they were missing. Lucas: Right. Nucor installed a fiber-optic temperature sensor in the off-gas duct — it's a ruggedized probe that can handle 2,000 degrees Fahrenheit. That sensor feeds data to a control algorithm that adjusts power input in real time. If the off-gas temp crosses a threshold, the system dials back the amps. Luna: So it's not just a sensor — it's a closed-loop control system. Lucas: Correct. The sensor itself is maybe $2,000. The whole retrofit — including the controller, wiring, and integration — was about $200,000 per furnace. Crawfordsville has three furnaces, so $600,000 total. They saved that in the first month. Luna: And the energy savings aren't the only benefit, right? Lucas: No. Lower energy input means less wear on the electrodes — those are consumable, they cost about $1,000 each. Also, the refractory lining inside the furnace lasts longer because it's not being thermally shocked as often. Nucor estimated another $2 million a year in maintenance savings. Luna: So total payback in under a month, and $15.5 million annual improvement. Why doesn't every furnace do this? Lucas: That's the really interesting question. The technology has been available for at least a decade. The issue is what I'd call measurement inertia. Most mills have always controlled furnaces by recipe. They have an engineering assumption that the recipe is optimal. But they've never actually measured the waste stream. Luna: So it's a cultural thing, not a technical barrier. Lucas: Partly. Also, off-gas temperature measurement is tricky. The environment is brutal — high temperature, corrosive gases, particulate. Earlier sensors failed quickly. But fiber-optic technology improved around 2018, and now they're reliable for years. The knowledge hasn't spread yet. Luna: It's a classic case of a low-hanging fruit that nobody sees because they're not looking up. Lucas: Exactly. I talked to a plant manager at a small minimill in Ohio last month. He told me their energy cost per ton is $95. Nucor Crawfordsville got theirs down to $67. That $28 difference on a million tons a year is $28 million. He's now planning the retrofit. Luna: So the template is: identify your biggest waste stream, find a sensor that measures it in real time, and close the loop. Lucas: That's the one-sentence version. The key is that the off-gas temp was always there — it was just invisible. Once you see it, you can act. And the same principle applies beyond steel. Think about any industrial process where energy or material is consumed without real-time feedback. Luna: Cement kilns, glass furnaces, chemical reactors... Lucas: Right. A cement kiln can waste 30 percent of its fuel heat because of excess air. A glass furnace can overheat the melt by 50 degrees because the operator doesn't see the crown temperature. The sensor cost is trivial compared to the energy bill. Luna: What's the actual percentage improvement range you'd expect in a typical plant? Lucas: I've seen studies — mostly from the Department of Energy's Industrial Assessment Centers — showing 10 to 25 percent energy reduction just from better monitoring and control. The Nucor case is at the high end because they started with a very manual process. Luna: So the lesson isn't just about steel. It's about finding the one metric that everyone's ignoring. Lucas: Exactly. And it's not always off-gas temperature. At a food processing plant it might be steam trap efficiency. At a paper mill it might be dryer section moisture. But the pattern is the same: identify the invisible waste, instrument it, and let the data drive the decision. Luna: Alright, I think we've covered the how. Let's talk about the why — why does measurement inertia persist even when the payback is so clear? Lucas: Part of it is that energy is often treated as a fixed cost. People say 'the furnace uses what it uses.' But it's not fixed — it's a variable you can optimize. Another factor is that capital projects for sensors don't sound exciting. A $200,000 sensor install doesn't get a ribbon cutting. But a $13.5 million saving should. Luna: And I suspect there's a skill gap — many plant engineers aren't trained in control theory or data analytics. Lucas: That's true. The workforce in heavy industry is aging, and digital skills are scarce. But the good news is you don't need a data science team. You need one good sensor and a PID controller — that's 1940s technology. It's not AI. It's just closing the loop. Luna: Before we wrap, let's make this concrete for someone listening in a manufacturing role. What's the first step? Lucas: Pull your energy bill by process step. If you don't have submetering, get a portable power logger for a week. Find the highest-consumption step. Then ask: what variable are we not measuring that correlates with waste? Temperature, pressure, flow, composition. That's your sensor target. Luna: And you're saying the payback will almost always be under a year? Lucas: In my experience, if you pick the right waste stream, it's under six months. Nucor's was under one month. The risk is low — sensors are cheap, installation doesn't require a shutdown. The bigger risk is doing nothing. Luna: Yeah. And honestly, if today's episode moved your thinking forward in some small way, there's a link at buy me a coffee dot com slash fexingo. A coffee's worth if it saves your plant a million. Lucas: Agreed. We keep this show ad-free, and listener support is what makes that possible. No pressure — just if it was worth your time. Luna: Alright, back to the takeaway. So the core message: find your off-gas temperature — whatever that is for your operation. Lucas: Exactly. Identify the invisible waste, measure it, close the loop. One sensor, one algorithm, one decision. That's the operator's manual for today.