Skip to content
Confidential

Site Project Catalog — Cleveland Works

Purpose: Groups Cleveland's 27 initiatives into actionable projects for the plant manager. Each site project bundles related initiatives, sizes the local opportunity, and references the corporate project it feeds into.

Audience: John Stubna (Plant Manager), Cleveland site leadership, internal team

Last updated: 2026-04-16 (consistency pass: corporate project mappings verified against ch5/ch8)


Project Summary

ID Project Horizon Corporate Bundled Initiatives Value ($/yr) Champion Status
CLV-P01 Ops-Maintenance Data Integration H1 PRJ-01 CLV-01 $2-5M Jamie Betts validated
CLV-P02 Predictive Maintenance Platform H1→H2 PRJ-03 CLV-12, CLV-22, CLV-23, CLV-09 $3-12M Paul Aaron Dash validated
CLV-P03 Maintenance Workflow & Copilot H1 PRJ-06 CLV-07, CLV-08, CLV-24, CLV-25 $3-11M Dan Hartman validated
CLV-P04 Process Risk & Cobble Prediction H2 PRJ-05 CLV-04, CLV-11, CLV-17, CLV-18, CLV-20 $6-23M Dan Hartman validated
CLV-P05 Quality & Yield Traceability H2→H3 PRJ-04 CLV-06, CLV-14, CLV-19 $2-5M + enabler Stephen Palmer partial
CLV-P06 Production Scheduling & S&IOP H3 PRJ-02 CLV-02, CLV-03, CLV-15, CLV-26 $20-65M Andrew Mullen validated
CLV-P07 Intra-Plant Logistics H2 PRJ-07 CLV-10, CLV-16 TBD Shipping team identified
CLV-P08 Caster Chemistry Optimization H2 PRJ-08 CLV-05 $2-8M TBD identified
CLV-P09 Utility Management H2 new CLV-21 $7-15M Paul Aaron Dash validated
CLV-P10 Environmental Compliance H1 new CLV-13 $0.5-1M Andrew Mullen identified
CLV-P11 Enterprise Data Platform Strategy H2→H3 enabler CLV-27 Enabler Andrew Mullen identified

Project Cards


CLV-P01: Ops-Maintenance Data Integration

Field Value
Horizon H1: Bridge the Gap
Corporate project PRJ-01 — Ops-Maintenance Data Integration
Status validated
Champion(s) Jamie Betts (1SP Maint), Dan Hartman (HSM), Paul Aaron Dash

Local problem statement:

"Our operational side, it's all captured. The maintenance side, it's all captured. But we need to be able to integrate them." Operations tracks delays in SQL reports by area. Maintenance tracks work orders in Tabware by asset hierarchy. A 14-minute delay logged by operations has no matching work order. Different naming conventions, different systems, same events — no one can have a data-driven conversation about root causes.

Bundled initiatives: - CLV-01: Ops-Maintenance data integration — semantic matching layer linking SQL delay reports to Tabware work orders

Systems involved: Tabware (CMMS, ~95% hierarchy), operational SQL delay reports (custom-built, web dashboards), Axiom (ERP)

Value estimate: $2-5M/yr (reduced misattribution, faster root cause, targeted maintenance spend) Confidence: Medium-High — validated by 7+ stakeholders across 5 transcripts Implementation approach: Map operational delay categories to Tabware asset hierarchy (2-4 weeks) → build semantic matching logic (4-6 weeks) → unified dashboard (2-4 weeks). Both datasets exist and are ~95% structured. Cleveland is the natural entry site — strongest signal.

Dependencies: None — this is the foundation everything else builds on. Feeds CLV-P02 (PdM needs attribution data) and CLV-P04 (cobble prediction needs ops-maint linkage).

Palmer readout alignment: - Scalability: 4 sites (every integrated plant has this gap) / Quick-ROI: yes / Palmer named: no (but meets all criteria)


CLV-P02: Predictive Maintenance Platform

Field Value
Horizon H1→H2
Corporate project PRJ-03 — Predictive Maintenance Platform
Status validated
Champion(s) Paul Aaron Dash, Phil Thorman, Brian Thompson, Jamie Betts, John Messi

Local problem statement:

Vibration and condition monitoring is fragmented. AssetWatch at powerhouses only. Cloud vibration sensors on 1SP cranes that plant IT didn't even know existed. BOF bag house runs 4 ID fans — "it works until it doesn't, and then we all freak out." Scrubbing system degrades predictably over a 3-week cycle and "generates a shitload of data" that gets "a bit load of attention." Multiple high-value assets generate data nobody watches.

Bundled initiatives: - CLV-12: PdM platform (multi-asset) — consolidate fragmented condition monitoring, ML anomaly detection - CLV-22: BOF bag house predictive monitoring — PRIMARY PdM PoV target. 4 ID fans, rich sensor data, environmental compliance - CLV-23: BOF scrubbing system predictive monitoring — SECONDARY target. Predictable 3-week degradation curve - CLV-09: Spare parts inventory intelligence — predictive spares management feeds into PdM planning

Systems involved: Pi historian (10yr), AssetWatch (powerhouses), Viz (route vibration), cloud vibration (1SP cranes, Starnet?), Wonderware (powerhouses), Emerson (iron producing)

Value estimate: $3-12M/yr (PdM PoV through full expansion) Confidence: Medium — proven at powerhouses, scaling question. Bag house + scrubbing have richest data = highest PoV confidence. Implementation approach: PdM PoV on bag house (primary) + scrubbing (secondary) + Crane 300 (tertiary) — 8 weeks. Then expand to charging/teeming cranes, HSM work rolls, line shaft tables. Separate SOW governs the PoV phase.

Dependencies: CLV-P01 (ops-maint integration provides attribution data for maintenance events). PdM PoV is a separate project/SOW — implementation vehicle for first assets.

Palmer readout alignment: - Scalability: 4 sites (every site has critical assets) / Quick-ROI: yes (PoV in 8 weeks) / Palmer named: no directly, but PdM is core to value story


CLV-P03: Maintenance Workflow & Copilot

Field Value
Horizon H1: Bridge the Gap
Corporate project PRJ-06 — Maintenance Workflow Digitization
Status validated
Champion(s) Dan Hartman (HSM), Paul Aaron Dash, John Stubna

Local problem statement:

Green technicians go to repairs without knowing the history. Paper PM sheets come back crumpled and thrown away. "Where's my Ask Jeeves?" Everything above $500 goes to the plant manager — a $300 power supply takes 2 months to order. Employees buying inventory apps with their own money. Purchasing silently cancels requisitions after 60 days. Caster segment tracking lives in one contractor's Excel spreadsheet.

Bundled initiatives: - CLV-07: Maintenance co-pilot — voice-first mobile assistant for technicians (talk → structured WO) - CLV-08: Procurement decision tree / auto-approval — risk-stratified approval to cut weeks→hours - CLV-24: Caster segment lifecycle tracking — digitize Excel→system, link to campaign tonnage + procurement - CLV-25: Critical spares identification — accelerate Brian Thompson's 500-part effort with AI-assisted item master population

Systems involved: Tabware (CMMS), Axiom (ERP/procurement), custom HVAC tablet app, caster segment Excel

Value estimate: $3-11M/yr (copilot $0.5-2M + procurement $1-3M + segments $1-3M + spares $1-3M) Confidence: Medium-High — deepest visceral pain across every conversation Implementation approach: Copilot MVP on 1 asset class (3-4 months) → procurement rules-based fast-track (parallel) → segment digitization (builds on existing Brian Thompson effort) → critical spares AI-accelerated identification. Must solve Wi-Fi dead zones for copilot.

Dependencies: None significant for copilot. Procurement reform requires corporate buy-in (policy change, not just tech). CLV-25 feeds CLV-09 (spare parts intelligence).

Palmer readout alignment: - Scalability: 4 sites (same corporate procurement policies everywhere) / Quick-ROI: yes / Palmer named: knowledge capture theme aligns


CLV-P04: Process Risk & Cobble Prediction

Field Value
Horizon H2: Build the Foundation
Corporate project PRJ-05 — Cobble & Process Risk Prediction
Status validated
Champion(s) Dan Hartman (HSM), John Stubna, Chad Helms (Iron Producing)

Local problem statement:

Cobbles destroy drive spindles "multiple times," cause major downtime, and are increasing as experienced operators retire. "Some people just know the right call to make" — but those people are leaving. New operators override safety checks and cause cobbles. Context-dependent alarming is too simplistic — vibration depends on product, grade, speed. BF thermal management is single-expert dependent.

Bundled initiatives: - CLV-04: Cobble prediction & prevention — ML model per slab (temperature, chemistry, mill setup, operator) - CLV-11: Operator decision support (BF/HSM) — AI-assisted guidance, prevent dangerous overrides, context-aware alarms - CLV-17: BOF endpoint prediction — reduce reblows through ML (seed) - CLV-18: Caster breakout prediction — augment breakout detection with ML (seed) - CLV-20: BF thermal state prediction — predict Si/temperature 30-60 min ahead (seed)

Systems involved: L2 (HSM, BF, BOF, caster), Pi historian, operational delay system, Global Gauges (vision — 2 HSM stands), Emerson (BF C5/C6)

Value estimate: $6-23M/yr (cobble $3-10M + operator support $1-5M + BOF $0.5-2M + breakout $1-3M + BF thermal $1-3M) Confidence: Medium — cobble prediction and operator support validated, seed items need further scoping Implementation approach: Cobble prediction model first (6-9 months, needs L2 integration) → operator decision support → BOF/BF/caster models as data matures. Context-aware alarming is an early ML win within the operator support track.

Dependencies: CLV-P01 (attribution data for cobble root cause). L2 data access at all stages. Pi historian depth.

Palmer readout alignment: - Scalability: 3 sites with HSM, 4 with BF / Quick-ROI: cobble prediction can be bounded / Palmer named: BF stove optimization (related to CLV-20/CLV-11 pattern)


CLV-P05: Quality & Yield Traceability

Field Value
Horizon H2→H3
Corporate project PRJ-04 — Through-Process Quality & Yield
Status partial
Champion(s) Stephen Palmer (quality/vision AI interest), TBD (quality/metallurgy)

Local problem statement:

No end-to-end genealogy linking heat→slab→coil→finished product. Customer claims traced manually over days/weeks. 90-day slabs get remelted — a symptom of planning-demand misalignment. SIS exists but accuracy unknown. Slab cut optimization already saving $3M/month (Phil Thorman), so the biggest yield win is captured — remaining opportunity is in traceability and through-process quality.

Bundled initiatives: - CLV-06: 90-day slab remelting reduction — analytics on slab aging, reallocation recommendations - CLV-14: Through-process traceability (heat→coil) — automated genealogy linking L2 across process steps - CLV-19: Surface defect detection (CNN on SIS) — reduce SIS false positives (seed)

Systems involved: Janus (slab WMS), Axiom (ERP/demand), L2 across all stages, SIS cameras

Value estimate: $2-5M/yr direct (slab remelting + defect detection) + enabler value (traceability unlocks $13-27M in yield improvement) Confidence: Medium — slab cut already captured the biggest win. Traceability is foundational but hard to size. Implementation approach: Slab aging dashboard (2-3 months, quick win) → automated genealogy (6-12 months, strategic) → SIS defect detection CNN (6-9 months). Cleveland is NOT the primary site for through-process quality — Middletown's finishing chain makes it the better candidate.

Dependencies: CLV-P06 (scheduling) for systemic slab aging fix. Data architecture for cross-stage L2 linkage.

Palmer readout alignment: - Scalability: 4 sites / Quick-ROI: slab aging dashboard is bounded / Palmer named: surface inspection is his explicit priority (but Middletown is the stronger site for this)


CLV-P06: Production Scheduling & S&IOP

Field Value
Horizon H3: Predict & Optimize
Corporate project PRJ-02 — Production Scheduling & S&IOP
Status validated
Champion(s) Andrew Mullen, Terry Fedor, John Stubna

Local problem statement:

"1% improvement = tens of millions." 5-6K orders/week with no optimization logic. No formal S&IOP — commercial, operations, and capacity are in separate silos. Roll changes account for ~50% of HSM operational delays. No demand forecasting (firm orders only). Dynamic pricing doesn't factor in capacity consumption — "we'll take heavy line pipe orders which destroy our equipment." In September (4th crew), the constraint shifts from 1SP to iron producing or HSM — no dynamic constraint identification.

Bundled initiatives: - CLV-02: Cross-stage scheduling / S&IOP — integrated business planning + AI scheduling optimization - CLV-03: Roll change sequencing optimization — minimize HSM changeovers - CLV-15: Dynamic pricing by capacity consumption — profitability scoring per order - CLV-26: Raw materials cost optimization — burden/alloy mix optimization + supply chain risk

Systems involved: Axiom (ERP), L2 (process control), Tabware (maintenance windows), scheduling tools (unknown — likely Excel)

Value estimate: $20-65M/yr (scheduling $10-30M + roll changes $5-15M + pricing $5-20M + raw materials TBD) Confidence: Low-Medium — massive scope, validated as real pain but needs decomposition Implementation approach: Phase 1: cross-stage visibility dashboard (3-4 months) → Phase 2: roll change sequencing (6-9 months) → Phase 3: full scheduling optimizer (12-18 months) → Phase 4: S&IOP process + AI planning. Highest value, longest horizon.

Dependencies: CLV-P01 (ops-maint integration), CLV-P05 (traceability), enterprise data architecture.

Palmer readout alignment: - Scalability: 4 sites / Quick-ROI: NO — this is H3, long horizon / Palmer named: no. This is the strategic bet, not the entry point.


CLV-P07: Intra-Plant Logistics

Field Value
Horizon H2: Build the Foundation
Corporate project PRJ-07 — Intra-Plant Logistics Optimization
Status identified
Champion(s) Facilities division manager, shipping team

Local problem statement:

Coils touched 3-4 times before shipping (target: 1-2). Rail/truck mode switching adds unnecessary moves. 600+ coil increase in shipping volume recently. Slab yard is LIFO instead of FIFO. 33 bottle cars are a hard constraint across the river.

Bundled initiatives: - CLV-10: Intra-plant coil movement optimization — warehouse-style placement, reduce crane touches - CLV-16: Rail car inventory visibility — real-time tracking with class 1 railroads (NS, CSX)

Systems involved: Custom SQL coil tracking, shipping systems, external railroad data (NS, CSX — PDFs/Excel today)

Value estimate: TBD — need movement data to calculate crane hours saved + throughput impact Confidence: Low — need data Implementation approach: Analyze coil movement data → storage allocation optimization → rail car visibility integration. Rail car component depends on external cooperation (railroads).

Dependencies: None internally. Rail car visibility requires railroad API/data feed access.

Palmer readout alignment: - Scalability: 4 sites / Quick-ROI: analysis phase is fast / Palmer named: YES — coil logistics is his #1 cross-site priority


CLV-P08: Caster Chemistry Optimization

Field Value
Horizon H2: Build the Foundation
Corporate project PRJ-08 — Caster Chemistry Optimization
Status identified
Champion(s) TBD (caster/metallurgy team)

Local problem statement:

"We've got a Prime Metals model. We don't even use. We spent millions on it." Chemistry transitions in continuous casting generate scrap when the cut point between chemistries is wrong. The model they bought doesn't get used — need to understand why (adoption? accuracy? integration?).

Bundled initiatives: - CLV-05: Caster chemistry transition optimization — predict optimal cut points, sequence chemistries, recover in-between grades

Systems involved: L2 (caster), Prime Metals model (status unknown), Axiom (grades/orders)

Value estimate: $2-8M/yr Confidence: Medium Implementation approach: Phase 1: understand Prime Metals failure (adoption vs. accuracy vs. integration). Phase 2: build chemistry transition optimizer leveraging existing infrastructure if possible. Company-wide problem.

Dependencies: CLV-P05 (traceability — need heat→slab chemistry lineage).

Palmer readout alignment: - Scalability: 3 sites with casters / Quick-ROI: depends on Prime Metals investigation / Palmer named: no


CLV-P09: Utility Management

Field Value
Horizon H2: Build the Foundation
Corporate project new (site-specific, may validate at other sites)
Status validated
Champion(s) Paul Aaron Dash (20-year champion), Dan Hartman

Local problem statement:

No utility management system at Cleveland. Cannot track water, gas, O2, N2, or compressed air by area. Compressed air alone = $7M/yr savings opportunity identified in engineering study. Nitrogen pressure drops affect 1SP operations with no diagnostic capability. Paul Aaron Dash has worked on this for 20+ years with budget constraints — he is the last utility engineer. Descale water system: 5 pumps, 12 headers, 18 variables manually charted weekly.

Bundled initiatives: - CLV-21: Utility management system — AI on metering infrastructure for energy/utility optimization

Systems involved: Wonderware (powerhouses, water treatment), Emerson (new water meters being installed), manual charting (descale system)

Value estimate: $7-15M/yr ($7M compressed air + water + gas + nitrogen) Confidence: Medium — compressed air backed by engineering study, others need scoping Implementation approach: Build AI layer on metering infrastructure as water meters (Emerson) come online. Anomaly detection for leaks and pressure drops. Automated reporting replacing manual charting. 6-12 months.

Dependencies: Metering infrastructure completion (water meters in progress).

Palmer readout alignment: - Scalability: TBD — need to validate at other sites / Quick-ROI: compressed air already scoped / Palmer named: no


CLV-P10: Environmental Compliance

Field Value
Horizon H1: Bridge the Gap
Corporate project new (site-specific)
Status identified
Champion(s) Andrew Mullen

Local problem statement:

Environmental compliance (SDS tracking, refrigerant reporting) is manual and paper-based. Tied to employee bonuses. Exceedances at Burns Harbor already this year.

Bundled initiatives: - CLV-13: Environmental compliance automation — automated SDS tracking, purchase-to-storage-location linkage

Systems involved: Manual pencil-and-paper tracking

Value estimate: $0.5-1M/yr Confidence: Low — needs scoping Implementation approach: Rules-based automation. Quick Win.

Dependencies: None.

Palmer readout alignment: - Scalability: 4 sites / Quick-ROI: yes / Palmer named: no


CLV-P11: Enterprise Data Platform Strategy

Field Value
Horizon H2→H3
Corporate project enabler (not a standalone project)
Status identified
Champion(s) Andrew Mullen

Note: This is NOT a standalone project — it is the emergent outcome of doing the work. Every H1 project connects data sources; every H2 project adds new streams. By H3, the foundation exists. Databricks, Snowflake, and Microsoft Fabric are all under evaluation. CEO wants enterprise-wide data platform. Our recommendation: build bottom-up through project delivery, not top-down through a data lake initiative.

Bundled initiatives: - CLV-27: Enterprise data platform (cloud strategy)

Value estimate: Enabler (unlocks $50-100M+ across all projects) Confidence: Low — early exploratory Implementation approach: Progressive data foundation through project delivery. Each project ingests new sources. Platform decision (Databricks/Snowflake/Fabric) can happen in parallel without blocking project work.


Corporate Project Cross-Reference

Corporate Project Site Projects Validation Strength
PRJ-01: Ops-Maint Integration CLV-P01 Strong — strongest signal, every stakeholder
PRJ-02: Production Scheduling CLV-P06 Strong — massive value, validated but H3
PRJ-03: PdM Platform CLV-P02 Strong — multi-asset, PoV targets selected
PRJ-04: Quality & Yield CLV-P05 Partial — slab cut already captured biggest win
PRJ-05: Cobble & Process Risk CLV-P04 Strong — cobbles + operator support + BF
PRJ-06: Maint Workflow CLV-P03 Strong — deepest visceral pain
PRJ-07: Logistics CLV-P07 Partial — need movement data
PRJ-08: Caster Chemistry CLV-P08 Partial — Prime Metals investigation needed

Site-Specific Notes

  • 1SP is THE constraint: 28 heats/day target, all casters booked through May. Every missed production unit = lost revenue.
  • 70/30 reactive/planned maintenance. Zero close-the-loop. 2001 ISG restructuring merged planner + reliability engineer + area manager into one role — structural root cause for broken information loops.
  • Union: USW. Generally cooperative but change management matters.
  • Systems landscape: Tabware (CMMS), Pi (historian, 10yr steel producing), Wonderware (powerhouses), Emerson (iron producing), Axiom (ERP), Janus (slab WMS), L2 (process control), Viz (route vibration), cloud vibration on cranes.
  • Addressable value: $26-69M/yr across all projects at Cleveland alone.
  • PdM PoV targets (revised Day 5): BOF bag house (primary), scrubbing system (secondary), Crane 300 (tertiary).
  • Key relationship: John Stubna (plant manager) has "full support of corporate" and young leadership team open to change.