DATTS

Decision-Centric Thinking in Steel: Making the Invisible Visible

Back to series

The most dangerous decisions in steel plants are not the complex ones.

They are the invisible decisions.

The ones that:

  • Are never written down
  • Never appear in procedures
  • Never show up in post-mortems
  • And yet, quietly decide millions of dollars every day

They live inside a person.

Which Decisions Depend on One Person’s Presence?


Ask a simple question on the shop floor:

“If this person doesn’t come to work tomorrow, which decisions stop happening?”


In many steel plants, the answer is uncomfortable.

  • The charging sequence “just feels right” to one furnace operator
  • The rolling order reshuffle happens because one planner senses trouble
  • The casting speed reduction happens because a senior metallurgist notices a pattern

No alarm goes off. No rule is broken. No system logs a decision.

And yet – the plant’s fate quietly changes.

The Most Dangerous Decisions Are Not Complex

Complex decisions attract attention:

  • Large capex approvals
  • Monthly production planning
  • Major grade change strategies

They are reviewed, simulated, and debated.

But steel plants are not lost by monthly decisions.

They are lost by thousands of small, invisible, moment-to-moment decisions, such as:

  • “Charge this slab now, not that one.”
  • “Wait five more minutes before rolling.”
  • “Let’s not stop — it will probably be fine.”

Each looks harmless. Together, they shape:

  • Yield
  • Energy intensity
  • Quality losses
  • Safety risk

Real-Life Scenarios (And Why They Are Scary)

Scenario 1: The Furnace Whisperer

One senior operator knows exactly when a slab is “ready enough.”

The MES says it needs 8 more minutes. He overrides it.

Most days, he’s right.

Until the day he isn’t.

That day:

  • Overheated slabs wait during a mill slowdown
  • Scale loss increases
  • Fuel consumption spikes
  • Surface defects appear downstream

No system flags the decision. The report shows only higher scrap.

The decision disappeared with the shift.

Scenario 2: The Planner Who Sees Trouble Coming

A planner notices that:

  • Casting delays are creeping up
  • The hot mill campaign looks tight

She quietly reshuffles the rolling sequence. No formal reschedule. No system approval.

The shift survives.

Then she goes on leave.

Two days later:

  • WIP piles up
  • Cooling beds saturate
  • Urgent orders miss delivery


Everyone asks: “Why didn’t the system warn us?”

The truth: The system never knew a decision was being made.

Scenario 3: The Unspoken Energy Trade-Off

A supervisor allows higher reheating temperatures “just to be safe.”

No KPI breaks. No alarm triggers.

But over weeks:

  • Fuel per ton creeps up
  • CO₂ intensity rises
  • Furnace refractory life shortens

Energy audits find symptoms. Not the decision.

The Core Problem: We Don’t Treat Decisions as Assets

Steel plants track:

  • Heats
  • Slabs
  • Coils
  • Tons
  • Minutes


But not decisions.

Decisions are:

  • Implicit
  • Experience-based
  • Person-bound
  • Lost during shift changes

This is not a people problem.

It’s a decision visibility problem.

A Decision-Centric Future for Steel

Imagine a different plant.

A plant where:

  • Every critical operational decision is recognized as a decision
  • The system knows why a slab was charged early
  • Trade-offs are made explicit: energy vs throughput vs quality
  • Decisions survive people, shifts, and experience gaps


This does not mean removing humans.

It means augmenting them.

How We Make It Better (Futuristic, But Practical)

1. Make Decisions First-Class Objects

Just like heats and orders, decisions should have:

  • Context
  • Alternatives
  • Risk
  • Outcome

2. Use AI to Surface the Invisible

AI’s real value is not automation.

It is:

  • Highlighting when a decision is being made
  • Showing what could happen if we choose differently
  • Learning from outcomes over time

3. Shift from Rule Enforcement to Decision Support

Rules say:

“You cannot do this.”

AI-supported systems ask:

“If you do this, here’s what is likely to happen.”

That is a fundamentally different conversation.

4. Preserve Expertise Without Depending on Presence

The goal is not to replace the furnace whisperer.

The goal is to:

  • Capture their intuition
  • Test it against data
  • Make it available at 2 a.m. on a bad day

Final Thought

The future steel plant will not be defined by:

  • Faster mills
  • Bigger furnaces
  • More dashboards


It will be defined by:

How well it understands, supports, and preserves its decisions.


Because the most expensive failures are not caused by bad decisions.

They are caused by invisible ones.

Disclaimer: The views expressed in this article are personal in nature, for informational purposes only, and do not constitute professional, operational, or financial advice for specific manufacturing facilities.