Confidential

Financial Analysis — Burns Harbor¶

Mission: Bottom-up value and cost analysis for every Burns Harbor initiative. Each card defines the formula and variables — the math structure, not necessarily the final number. Numbers are tagged by source: workshop-confirmed, estimated, or needs-corporate. This is the audit trail: every dollar in the business package traces back to a card here.

Methodology: Value per §3 of consolidation-plan.md (5-step: metric → current state → target → delta → sensitivity). Cost per §4 (effort-based: team composition × duration × rates).

Site context: Burns Harbor is CLF's largest integrated steel mill (~5M t/yr, ~4,039 employees). Longest process chain in CLF: coal blend → coke → sinter → BF → BOF → caster → HSM → plate/cold mill. GM's #1 = shipping velocity. Prior AI failures (caster plugging 2.5 yrs, cobble prediction 6 mo). BF thermal model is a positive counter-example. Coke plant knowledge cliff is the most acute in CLF. Cloud bandwidth is a blocker. GE rolling model source code access. Indiana Harbor cross-site validation from T5.

Last updated: 2026-04-16 (appendix preparation — no content changes, audit trail preserved as constructed)

How to Read This File¶

Initiative cards define the value formula and cost structure for each of Burns Harbor's 57 initiatives. They are grouped by parent site project (BH-P01..BH-P17).

Three card types: - Type A — Anchored: Formula defined, key variables have workshop-confirmed values. Math can be partially computed now. - Type B — Structured: Formula defined, but most variables need corporate data to populate. The math is built, the inputs are TBD. - Type C — Absorbed: No standalone formula. Value captured in parent project (enabler, seed, or subsumed scope).

Project roll-ups aggregate initiative cards into site project totals.

Corporate Inquiry Table (end of file) collects all needs-corporate variables into one table for IE to take to Cleveland-Cliffs.

BH-P01: Coil Velocity & Shipping Intelligence¶

BH-34: Coil Velocity & Shipping Intelligence¶

Card Type: A — Anchored Corporate Project: PRJ-07

Value Analysis¶

Value Types: Throughput gain + Working capital reduction + Cost avoidance Value Formula:

(additional_tons_shipped_per_day × margin_per_ton × 365)
+ (coil_cycle_time_reduction_days × inventory_tons × carrying_cost_per_ton_per_day)
+ (reprocessing_events_avoided_per_month × handling_steps_per_event × cost_per_handling_step × 12)

Variable	Value	Source	Status
current_shipping_tons_per_day	10,000-11,000	Paul/Sam: "10,000 tons/day shipping target, recently 11,000"	workshop-confirmed
additional_tons_shipped_per_day	[TBD]	Gap between current and optimized throughput	needs-corporate
margin_per_ton	[TBD]	Product mix average	needs-corporate
coil_cycle_time_reduction_days	[TBD]	Current avg cycle time (birth → ship) vs target	needs-corporate
inventory_tons	100K-140K range	Paul: "<100K = flowing, 120K = slowing, 135-140K = plant stops"	workshop-confirmed
carrying_cost_per_ton_per_day	[TBD]	Working capital × cost of capital / 365	needs-corporate
reprocessing_events_per_month	[TBD]	QMS flags → manual review → reroute frequency	needs-corporate
handling_steps_per_event	4-5+	Senior ops: "4-5 additional handling steps" per misrouted coil	workshop-confirmed
cost_per_handling_step	[TBD]	Crane time + labor + damage risk	needs-corporate

Workshop-Sourced Range: $10-25M/yr Confidence: High — GM's #1 priority, 6-person team articulates problem precisely, all-time shipping records prove execution capability Key Quotes: "If we can ship more, we can make more." — Paul. "Every time we handle double, triple, quadruple — risk goes up." — Sam. "Sometimes it takes a month to dig out."

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, domain specialist, PM	[TBD]	Map 60+ criteria that knock coil off shortest route
Data engineering	Data engineer (senior)	[TBD]	IMS + Genesis + QMS + MES integration
ML/AI development	ML engineer, optimization specialist	[TBD]	Coil routing optimizer, inventory constraint model
Application/UX	Frontend dev	[TBD]	Shipping intelligence dashboard, crane operator guidance
Infrastructure	Moderate	[TBD]	Real-time data feeds from IMS/Genesis, on-prem (cloud bandwidth constraint)
Change management	—	[TBD]	High — 6-person team workflow change, cross-functional. 25%.

BH-35: Automated Quality Disposition at Coil Birth¶

Card Type: A — Anchored Corporate Project: PRJ-04

Value Analysis¶

Value Types: Throughput gain + Cost avoidance Value Formula:

(coils_flagged_per_day × auto_disposition_rate × handling_steps_saved × cost_per_handling_step)
+ (disposition_delay_hours_saved × coils_per_day × margin_per_coil × velocity_factor)

Variable	Value	Source	Status
coils_flagged_per_day	[TBD]	QMS flag volume	needs-corporate
auto_disposition_rate	80%	Senior ops: "80% could be programmed in"	workshop-confirmed
handling_steps_saved	4-5	Per misrouted coil	workshop-confirmed
cost_per_handling_step	[TBD]	Crane time + labor + damage risk	needs-corporate
disposition_delay_hours	~24 hrs	Manual review happens next day	workshop-confirmed
coils_per_day	[TBD]	Derived from 220K+ tons/month ÷ avg coil weight	needs-corporate
margin_per_coil	[TBD]	Avg coil weight × margin/ton	needs-corporate
velocity_factor	[TBD]	Revenue impact of 24hr faster disposition	needs-corporate

Workshop-Sourced Range: $5-12M/yr Confidence: High — senior ops quantified the 80% rule, data already collected (temperature maps, gauge data, chemistry) Key Quote: "AI doesn't even have to collect the data because we are collecting it already. It has to just read it, apply the customer filter criteria and say yay or nay."

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, data scientist, PM	[TBD]	Customer tolerance table codification with quality group
Data engineering	Data engineer	[TBD]	QMS + IBA + chemistry integration
ML/AI development	Data scientist	[TBD]	Decision table engine (Phase 1), ML classification (Phase 2)
Application/UX	Frontend dev	[TBD]	Real-time disposition display at coiler
Infrastructure	Minimal	[TBD]	Data already collected — just need inference at coil birth
Change management	—	[TBD]	Moderate — quality group must trust AI disposition. 20%.

BH-38: Coil Field OCR & Computer Vision¶

Card Type: C — Absorbed Corporate Project: PRJ-07 Reason: OCR cameras already installed. Value captured in BH-P01 coil velocity roll-up. Additive safety and tracking improvement. Value Contribution: $1-3M/yr — safety (remove people from coil fields) + inventory accuracy. Absorbed into BH-P01 total. Cost Contribution: Camera integration with Genesis system — bounded data engineering within BH-P01 scope.

BH-17: HSM Scheduling Optimization¶

Card Type: A — Anchored Corporate Project: PRJ-02

Value Analysis¶

Value Types: Throughput gain + Working capital reduction Value Formula:

(warehouse_clogging_days_per_month × tons_blocked_per_day × margin_per_ton)
+ (future_product_tons_per_month × carrying_cost_per_ton × avg_hold_days)
+ (schedule_alignment_improvement_% × monthly_shipping_tons × margin_per_ton)

Variable	Value	Source	Status
warehouse_clogging_days_per_month	[TBD]	Days above 135K ton threshold	needs-corporate
tons_blocked_per_day	[TBD]	When above threshold, shipping reduction	needs-corporate
future_product_tons_per_month	[TBD]	Tons rolled >3 weeks ahead of ship date	needs-corporate
carrying_cost_per_ton	[TBD]	Working capital cost	needs-corporate
avg_hold_days	[TBD]	Average days "future" product sits before shipping	needs-corporate
schedule_alignment_improvement_%	10-20%	Conservative estimate	estimated
monthly_shipping_tons	220,000+	Paul/Sam: established throughput	workshop-confirmed
margin_per_ton	[TBD]	Product mix average	needs-corporate

Workshop-Sourced Range: $5-15M/yr Confidence: Medium — confirmed acute pain from Paul ("They run future and fill up shipping with future"), but quantifying the gap requires production scheduling data Key Quote: "They run future and we fill up a whole number one shipping with future and it's just sitting and now I'm out of room." — Paul

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, domain specialist, PM	[TBD]	Schedule-to-ship alignment mapping
Data engineering	Data engineer	[TBD]	L-scheduler + MES + shipping data integration
ML/AI development	Optimization specialist	[TBD]	Schedule optimizer with inventory constraints
Application/UX	Frontend dev	[TBD]	Real-time warehouse capacity feedback to scheduling
Infrastructure	Moderate	[TBD]	MES integration, real-time data feeds
Change management	—	[TBD]	High — scheduling is politically sensitive. 25%.

BH-P02: BOF/Caster Chemistry Optimization¶

BH-41: BOF Off-Chemistry Analysis (Carbon + Sulfur)¶

Card Type: A — Anchored (MOST DATA-READY PROJECT AT BURNS HARBOR) Corporate Project: PRJ-08

Value Analysis¶

Value Types: Cost avoidance + Yield improvement Value Formula:

off_chemistry_rate_reduction_% × heats_per_year × tons_per_heat × exposure_per_ton
+ operator_deviation_capture_value (model improvement from successful deviations)

Variable	Value	Source	Status
current_off_chemistry_rate	5%	Dave: "5% of heats are off chemistry"	workshop-confirmed
carbon_sulfur_contribution	3% of 5% (60% of off-heats)	Dave: "carbon and sulfur contribute 3% of the 5%"	workshop-confirmed
target_off_chemistry_rate	[TBD]	Industry benchmark vs current	needs-corporate
heats_per_year	[TBD]	3 BOFs × heats/day × 365	needs-corporate
tons_per_heat	300	Dave: "300 tons"	workshop-confirmed
exposure_per_ton	[TBD]	Dave: "$1M" per off-chemistry heat (300 tons) ≈ $3,333/ton	workshop-confirmed
model_error_fraction	50% of carbon misses	Dave: "half of carbon misses attributed to model errors"	workshop-confirmed
operator_deviation_magnitude	70-100 lbs	Dave: "operators deviate by 70-100 lbs and get better results"	workshop-confirmed
sql_data_history	2001-present	PA group: "could go back to 2001"	workshop-confirmed

Workshop-Sourced Range: $5-15M/yr Confidence: High — Dave is the clearest champion at any site, PA confirmed data is "sitting readily available — pretty kind of ready to feed into an LLM," SQL data since 2001, in-house L2 models (no vendor lock-in) Key Quotes: "If I solve carbon or sulfur, 80% of my problem goes away." — Dave. "Start with something super simple. See if we have a proof of concept." — Dave. "These are the things that are important to the company and should be easy to do." — Dave.

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, data scientist, PM	[TBD]	SQL schema analysis, man/machine/process categorization
Data engineering	Data engineer	[TBD]	Doug Fortner's SQL Server extraction (needs read-replica)
ML/AI development	Data scientist, ML engineer	[TBD]	Statistical analysis (Phase 1), pattern model (Phase 2)
Application/UX	Frontend dev	[TBD]	Off-chemistry root cause dashboard for Dave's morning review
Infrastructure	Minimal	[TBD]	On-prem SQL, read-replica needed (BH-P17)
Change management	—	[TBD]	Low — Dave is the champion AND the user. 10%.

BH-15: Caster Chemistry Transition Optimization¶

Card Type: A — Anchored Corporate Project: PRJ-08

Value Analysis¶

Value Types: Cost avoidance + Yield improvement Value Formula:

chemistry_transitions_per_day × off_spec_tons_per_transition × margin_per_ton × reduction_%
+ grade_transition_time_saved_minutes × transitions_per_day × production_value_per_minute

Variable	Value	Source	Status
chemistry_transitions_per_day	[TBD]	2 casters × transitions per campaign	needs-corporate
off_spec_tons_per_transition	[TBD]	Quality records — transition zone yield loss	needs-corporate
margin_per_ton	[TBD]	Product mix average	needs-corporate
reduction_%	30-50%	Optimized cut points + sequencing	estimated
end_tap_to_open_window	75 min	Dave: "very tight" vs MDT 130-140 min	workshop-confirmed
grade_complexity	"a lot more complex grades, a lot more chemistry changes" than MDT	Dave	workshop-confirmed
production_value_per_minute	[TBD]	Derived from throughput	needs-corporate

Workshop-Sourced Range: $2-8M/yr Confidence: Medium-High — Dave validated, in-house L2 models allow rapid adaptation, but transition data needed Key Quote: "If I'm off chemistry, that's 300 tons — million dollars." — Dave

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, domain specialist	[TBD]	L2 model audit, transition pattern analysis
Data engineering	Data engineer	[TBD]	L2 caster + chemistry data extraction
ML/AI development	ML engineer	[TBD]	Cut point optimization, sequence optimizer
Application/UX	Frontend dev	[TBD]	Caster operator guidance display
Infrastructure	Moderate	[TBD]	Real-time inference at caster
Change management	—	[TBD]	Low — in-house L2 models, operators already familiar. 15%.

BH-42: Caster Plugging/Clogging Prediction¶

Card Type: A — Anchored Corporate Project: PRJ-08

Value Analysis¶

Value Types: Cost avoidance + Throughput gain Value Formula:

plugging_events_per_year × (production_loss_minutes × production_value_per_minute + tundish_cost)
× prevention_rate

Variable	Value	Source	Status
plugging_events_per_year	~25 YTD (annualized ~50+)	Dave: "25 plugging events this year" (as of ~Mar)	workshop-confirmed
production_loss_minutes_per_event	80	Dave: "80 minutes production loss"	workshop-confirmed
production_value_per_minute	[TBD]	Caster throughput × margin	needs-corporate
tundish_cost_per_event	$40,000	Dave: "$40K tundish cost"	workshop-confirmed
prevention_rate	20-30%	Conservative — prior ArcelorMittal attempt failed after 2.5 years	estimated
clogging_factor_data	Exists — live monitoring	Isabelle: temp + gate position monitoring	workshop-confirmed

Workshop-Sourced Range: $2-5M/yr Confidence: Medium — live clogging factor data is strong, but prior ArcelorMittal AI failure (2.5 years, zero results) on this exact problem at BH demands caution. Frame as predictive analytics (flying earlier), not real-time control. Key Quote: "Better to fly than plug — $40K to fly vs. 80 min + cascading disruption for plug." "These have plugged steel shops since the beginning of time." — Dave

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Data scientist, PM	[TBD]	Forensic review of prior ArcelorMittal failure FIRST
Data engineering	Data engineer	[TBD]	Clogging factor history + grade + tundish sequence extraction
ML/AI development	ML engineer, data scientist	[TBD]	Pattern analysis by grade/sequence (NOT real-time control)
Application/UX	Frontend dev	[TBD]	Early warning display for caster operators
Infrastructure	Minimal	[TBD]	On-prem, existing clogging factor feed
Change management	—	[TBD]	Medium — prior failure creates skepticism. 20%. Must demonstrate what's different this time.

BH-12: BOF Endpoint Prediction¶

Card Type: C — Absorbed Corporate Project: PRJ-05 / PRJ-08 Reason: R&D already building at MDT using Copilot. Scalable to BH's 3 BOFs (highest opportunity) but not field-validated at BH. Value Contribution: $2-5M/yr estimated. Absorbed into BH-P02 roll-up. Cross-site with MDT R&D work. Cost Contribution: One ML model within BH-P02 scope, leveraging MDT R&D foundation.

BH-P03: Coke Plant Operations & Battery Vision¶

BH-46: Battery Vision — Coke Plant Integrated Ops Dashboard¶

Card Type: A — Anchored Corporate Project: new (BH-unique)

Value Analysis¶

Value Types: Efficiency gain + Risk mitigation Value Formula:

manager_data_consumption_time_saved_hours_per_day × labor_rate × 365
+ delay_response_speedup_minutes × delays_per_month × production_value_per_minute × 12
+ maintenance_planning_improvement_% × annual_coke_maintenance_cost

Variable	Value	Source	Status
manager_data_consumption_time	"12 different places to consume it"	Coke Plant Div Mgr	workshop-confirmed
labor_rate (manager)	[TBD]	Loaded rate	needs-corporate
delay_response_speedup_minutes	[TBD]	From dashboard vs current manual lookup	needs-corporate
delays_per_month	[TBD]	Coke plant delay database	needs-corporate
production_value_per_minute	[TBD]	Coke throughput × downstream BF value	needs-corporate
annual_coke_maintenance_cost	[TBD]	Battery maintenance spend	needs-corporate
maintenance_planning_improvement_%	10-20%	Conservative with integrated visibility	estimated

Workshop-Sourced Range: $1-3M/yr Confidence: High — Division Manager has detailed requirements, Bill Barker is the delivery engine via iFix, data streams exist Key Quote: "I want it integrated in a way that's consumable by all of my people." "It's so disjointed right now. You gotta go 12 different places to consume it."

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, PM	[TBD]	Map 12+ data sources to unified view
Data engineering	Data engineer	[TBD]	iFix integration, thermal map data, push scheduling
ML/AI development	Minimal	[TBD]	Phase 1 is dashboarding; ML in Phase 2 (predictive)
Application/UX	Frontend dev	[TBD]	Delivered through iFix (universal plant access)
Infrastructure	Minimal	[TBD]	iFix already deployed — Bill Barker can deliver "in hours/days"
Change management	—	[TBD]	Low — Division Manager is the champion. 10%.

BH-18: Coke Plant Optimization (Push Timing, Temperature, Quality)¶

Card Type: B — Structured Corporate Project: new (BH-unique)

Value Analysis¶

Value Types: Energy efficiency + Quality improvement + Environmental compliance Value Formula:

(heating_uniformity_improvement_% × annual_energy_cost)
+ (green_push_prevention_events × cost_per_green_push)
+ (coke_quality_improvement_impact_on_BF × BF_annual_throughput_value)

Variable	Value	Source	Status
annual_energy_cost (coke plant)	[TBD]	Gas consumption for 164 ovens × 19hr cycles	needs-corporate
heating_uniformity_improvement_%	[TBD]	Current variability vs target	needs-corporate
green_push_events_per_year	[TBD]	Push records — undercooked pushes	needs-corporate
cost_per_green_push	[TBD]	Quality loss + reprocessing + environmental exposure	needs-corporate
coke_quality_variability	[TBD]	CSR/CRI standard deviation	needs-corporate
BF_coke_rate_sensitivity	[TBD]	Tons coke/ton hot metal × improvement %	needs-corporate
pyrometer_coverage	2 of 4 pushers operational	PA group: "some are very good and some are very bad"	workshop-confirmed

Workshop-Sourced Range: $3-8M/yr Confidence: Medium — pyrometer data quality varies, heating control is manual ("pipe wrench + nozzle"), but the value chain (coke → sinter → BF) amplifies any quality improvement Key Quote: "I don't know what's going on inside of that wall between charge time and push time."

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, domain specialist, PM	[TBD]	Pyrometer data audit, push cycle mapping
Data engineering	Data engineer	[TBD]	Pyrometer + push schedule + coke quality integration
ML/AI development	ML engineer, data scientist	[TBD]	Temperature prediction model, push optimization
Application/UX	Frontend dev	[TBD]	Push guidance display via iFix
Infrastructure	Moderate	[TBD]	Pyrometer transmitter replacement (hardware dependency)
Change management	—	[TBD]	Medium — manual heating culture change. 20%.

BH-48: Coke Plant Delay Classification & Root Cause Analytics (NLP)¶

Card Type: B — Structured Corporate Project: PRJ-01

Value Analysis¶

Value Types: Efficiency gain + Cost avoidance Value Formula:

top_delay_category_elimination_hours × production_value_per_hour × 12
+ root_cause_identification_speedup × investigations_per_month × labor_rate

Variable	Value	Source	Status
delay_entries_per_year	[TBD]	Historical freeform delay records	needs-corporate
top_delay_categories	[TBD]	NLP-derived Pareto from freeform text	needs-corporate
production_value_per_hour (coke)	[TBD]	Coke throughput × downstream value	needs-corporate
investigations_per_month	[TBD]	Mike Zamuta's current workload	needs-corporate
labor_rate (manager)	[TBD]	Loaded rate	needs-corporate

Workshop-Sourced Range: $0.5-1M/yr Confidence: Medium — freeform text makes NLP the right approach, Mike Zamuta can validate immediately Key Quote: "I can't say show me all my Larry car delays for the past three years." — Mike Zamuta

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Data scientist, PM	[TBD]	Delay text corpus analysis
Data engineering	Data engineer	[TBD]	Delay database extraction
ML/AI development	NLP engineer	[TBD]	Text classification, entity extraction, Pareto generation
Application/UX	Frontend dev	[TBD]	Delay analytics dashboard (via iFix or Power BI)
Infrastructure	Minimal	[TBD]	On-prem NLP inference
Change management	—	[TBD]	Low — Mike Zamuta is the user and validator. 10%.

BH-47: Coal Blend Optimization Model¶

Card Type: B — Structured Corporate Project: new (BH-unique)

Value Analysis¶

Value Types: Cost avoidance + Quality improvement + Environmental compliance Value Formula:

(coal_cost_optimization_% × annual_coal_spend)
+ (coke_quality_improvement × BF_productivity_gain_per_%_quality)
+ (sulfur_compliance_margin_improvement × environmental_risk_value)

Variable	Value	Source	Status
annual_coal_spend	[TBD]	8 coal types × volume × price	needs-corporate
coal_cost_optimization_%	[TBD]	Blend optimization potential	needs-corporate
coke_quality_metrics (VM, sulfur, reflectance, contraction)	[TBD]	Lab analysis history	needs-corporate
BF_productivity_gain_per_%_quality	[TBD]	BF model sensitivity to coke quality	needs-corporate
sulfur_compliance_margin	[TBD]	No desulfurization facility — low-sulfur coal is mandatory	needs-corporate
PhD_model_recovery_status	"Nobody seems to know where they're at"	Coke Plant Div Mgr	workshop-confirmed

Workshop-Sourced Range: $2-5M/yr Confidence: Low-Medium — PhD models lost ~1.5 years ago, domain expertise scattered. Tom Zenzian (corporate coal buyer) is key recovery contact. Scalable to 4 CLF coke plants. Key Quote: "The PhDs who built coal blend models were let go and their models are lost. Nobody seems to know where they're at."

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, domain specialist, PM	[TBD]	Lost model forensics, coal blend parameter mapping
Data engineering	Data engineer	[TBD]	Lab data + coal procurement + coke quality integration
ML/AI development	Data scientist, ML engineer	[TBD]	Multi-objective optimization (cost × quality × sulfur)
Application/UX	Frontend dev	[TBD]	Blend recommendation dashboard
Infrastructure	Minimal	[TBD]	—
Change management	—	[TBD]	Medium — requires Tom Zenzian partnership. 20%.

BH-P04: Plate Mill Shipping Intelligence¶

BH-43: Plate Shipping Hit List Automation¶

Card Type: A — Anchored (PROVING GROUND) Corporate Project: PRJ-07

Value Analysis¶

Value Types: Efficiency gain + Throughput gain Value Formula:

(meeting_hours_saved_per_week × labor_rate × participants × 52)
+ (faster_order_completion_days × orders_per_month × margin_per_order × 12)
+ (partial_car_combinations_per_month × savings_per_combination × 12)

Variable	Value	Source	Status
meeting_hours_saved_per_week	[TBD]	Current meeting hours → automated hit list	needs-corporate
participants	4-5	Dave: "4-5 people live and die by this"	workshop-confirmed
labor_rate	[TBD]	Manager loaded rate	needs-corporate
action_items_per_meeting	10-15	Dave: "10-15 action items"	workshop-confirmed
faster_order_completion_days	[TBD]	Current vs automated execution speed	needs-corporate
orders_per_month (plate)	[TBD]	Plate shipping volume	needs-corporate
margin_per_order	[TBD]	Plate product margins	needs-corporate
partial_car_combinations_per_month	[TBD]	Partial rail cars combinable	needs-corporate
savings_per_combination	[TBD]	Rail car cost + scheduling	needs-corporate
OTIF_improvement_%	[TBD]	0% unless 100% OTIF — binary metric	needs-corporate

Workshop-Sourced Range: $1-3M/yr Confidence: High — Dave built the data infrastructure (Power BI since 2015-16), team confirmed "achievable in a few weeks," most self-contained project at Burns Harbor Key Quotes: "That should be a process, not a meeting." — Dave. "Put three losers together and make one win." — Dave. "That's achievable in a few weeks." — Team. "Our goal would not be to try to change the underlying business system. Build on top." — Dave.

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, PM	[TBD]	Codify hit list logic from Dave's team
Data engineering	Data engineer	[TBD]	Power BI + IBM mainframe integration
ML/AI development	Minimal	[TBD]	Rules engine Phase 1; ML optimization Phase 2
Application/UX	Frontend dev	[TBD]	Automated daily hit list report
Infrastructure	Minimal	[TBD]	Build on existing Power BI infrastructure
Change management	—	[TBD]	Minimal — Dave IS the change agent. 10%.

Note: BH-P04 is the stepping stone to BH-P01. Plate is smaller, simpler, Dave built the data infrastructure. Prove the model here, then scale to hot strip's 220K tons/month.

BH-20: Plate Mill Scheduling & Quality Prediction¶

Card Type: B — Structured Corporate Project: new (BH-unique)

Value Analysis¶

Value Types: Throughput gain + Quality improvement Value Formula:

plate_rework_reduction_% × rework_tons_per_year × rework_cost_per_ton
+ scheduling_improvement_% × plate_throughput_tons_per_year × margin_per_ton

Variable	Value	Source	Status
plate_rework_tons_per_year	[TBD]	Plate quality records	needs-corporate
rework_cost_per_ton	[TBD]	Reprocessing + scheduling disruption	needs-corporate
plate_rework_reduction_%	15-25%	Conservative estimate	estimated
plate_throughput_tons_per_year	[TBD]	Plate mill production data	needs-corporate
scheduling_improvement_%	5-10%	Conservative estimate	estimated
margin_per_ton (plate)	[TBD]	Plate product margins (generally higher than flat-rolled)	needs-corporate

Workshop-Sourced Range: $3-10M/yr Confidence: Medium — Dave understands the business deeply, but plate scheduling is "way more complex" and SAP tried 14 years + $20M to build a plate business system and failed Key Quote: "Our goal would not be to try to change the underlying business system. Build on top." — Dave

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, domain specialist, PM	[TBD]	Plate scheduling pattern analysis
Data engineering	Data engineer	[TBD]	MES + Power BI + quality data integration
ML/AI development	ML engineer	[TBD]	Quality prediction, scheduling optimization
Application/UX	Frontend dev	[TBD]	Integrated scheduling + quality dashboard
Infrastructure	Moderate	[TBD]	MES integration (12 years to develop)
Change management	—	[TBD]	Medium — SAP failure is institutional memory. 20%.

BH-P05: Ops-Maintenance Data Integration¶

BH-01: Ops-Maintenance Data Integration¶

Card Type: A — Anchored Corporate Project: PRJ-01

Value Analysis¶

Value Types: Throughput gain + Efficiency gain Value Formula:

misattributed_delay_hours_per_month × production_value_per_hour × attribution_correction_rate
+ root_cause_resolution_speedup_hours × incidents_per_month × labor_cost_per_hour
+ repeat_failure_reduction_% × repeat_failure_annual_cost

Variable	Value	Source	Status
misattributed_delay_hours_per_month	[TBD]	Cross-ref ops delay reports vs Tabware WOs	needs-corporate
production_value_per_hour	[TBD]	BH throughput × margin/ton (~5M t/yr)	needs-corporate
attribution_correction_rate	50-70%	Conservative	estimated
root_cause_resolution_speedup_hours	[TBD]	Current mean time to diagnose vs target	needs-corporate
incidents_per_month	[TBD]	Tabware work order volume	needs-corporate
labor_cost_per_hour	[TBD]	Blended maintenance tech rate	needs-corporate
repeat_failure_reduction_%	15-25%	Industry benchmark for closed-loop maintenance	estimated
repeat_failure_annual_cost	[TBD]	Frequency × cost per event from delay reports	needs-corporate

Workshop-Sourced Range: $2-5M/yr per site (BH + IH = 2 effective sites) Confidence: High — 5th consecutive site validation. IH = worst communication breakdown documented. BH BF area = best-practice counter-example. Key Quote (IH): "They are terrible at just talking to each other." "We do not close the loop." "Horse blinders on — they're trying to manage their area with their 8 guys."

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, PM	[TBD]	Map delay categories → Tabware hierarchy (BH + IH)
Data engineering	Data engineer	[TBD]	IBA + data warehouse + Tabware integration
ML/AI development	ML engineer	[TBD]	Semantic matching layer (NLP), delay pattern recognition
Application/UX	Frontend dev	[TBD]	Unified delay-to-work-order dashboard
Infrastructure	Minimal	[TBD]	On-prem, existing data sources
Change management	—	[TBD]	High — IH: no Wi-Fi, 4+ radio channels, cultural. BH: moderate. 25%.

BH-36: HSM Delay Analysis & Pattern Recognition¶

Card Type: B — Structured Corporate Project: PRJ-01

Value Analysis¶

Value Types: Efficiency gain + Cost avoidance Value Formula:

top_repeating_delay_hours_per_year × production_value_per_hour × elimination_rate
+ maintenance_focus_improvement × avoided_unplanned_downtime_hours × production_value_per_hour

Variable	Value	Source	Status
top_repeating_delay_hours_per_year	[TBD]	IBA + data warehouse delay records	needs-corporate
production_value_per_hour (HSM)	[TBD]	HSM throughput × margin	needs-corporate
elimination_rate	20-30%	Targeted maintenance on top delays	estimated
maintenance_focus_improvement	[TBD]	Time saved from prioritized repair focus	needs-corporate
avoided_unplanned_downtime_hours	[TBD]	Derived from targeted maintenance	needs-corporate

Workshop-Sourced Range: $2-5M/yr Confidence: Medium — Miles B's explicit ask, data exists (IBA + data warehouse + decades of history) Key Quote: "How do we get better at finding out what delays keep repeating, how do we focus our maintenance team on what's important?" — Miles B

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Data scientist, PM	[TBD]	Delay code taxonomy, IBA signal correlation
Data engineering	Data engineer	[TBD]	Data warehouse + IBA extraction
ML/AI development	Data scientist	[TBD]	Pattern recognition, Pareto analysis
Application/UX	Frontend dev	[TBD]	Top-10 delay dashboard, maintenance focus recommendations
Infrastructure	Minimal	[TBD]	On-prem data warehouse
Change management	—	[TBD]	Low — Miles B is champion. 10%.

BH-21: Root Cause Analysis Platform¶

Card Type: C — Absorbed Corporate Project: PRJ-01 adjacent Reason: Foundation enabler — RCA capability feeds ops-maint integration. BF engineers captured live cascading failure during session (lake water → BF shutdown → twier hit → transfer pump → belt issues). Value captured in BH-P05 roll-up. Value Contribution: $1-4M/yr — absorbed into BH-P05. Cost Contribution: ML/NLP model within BH-P05 scope.

BH-22: Cross-Site Caster Reliability Analytics¶

Card Type: C — Absorbed Corporate Project: PRJ-01 adjacent Reason: Seed — R&D weekly cross-site caster meetings already exist (Matt). MDT is the benchmark. Not field-validated at BH beyond cross-site reference. Value Contribution: $1-3M/yr cross-site — absorbed into BH-P05 roll-up. Cost Contribution: Analytics layer on existing cross-site data.

BH-P06: Maintenance Workflow & Inventory Intelligence¶

BH-03: Procurement Automation (Conversational Front-End)¶

Card Type: A — Anchored Corporate Project: PRJ-06

Value Analysis¶

Value Types: Efficiency gain + Throughput gain (de-bottleneck) Value Formula:

(transactions_per_day × time_saved_per_transaction_minutes / 60 × labor_rate × 250)
+ (approval_cycle_reduction_days × orders_per_month × downtime_cost_per_delayed_order × 12)
+ (e_market_adoption_increase_% × transactions_per_year × automation_savings_per_transaction)

Variable	Value	Source	Status
current_automated_transaction_rate	60-65%	John Sabo: "low-to-mid 60s" (was higher pre-Cliffs)	workshop-confirmed
target_automated_rate	70%+	John Sabo's target	workshop-confirmed
transactions_per_day (high-volume MRO)	"hundreds"	John Sabo: "high-volume MRO buyers: hundreds of transactions/day"	workshop-confirmed
time_saved_per_transaction_minutes	[TBD]	Reduced system switching (2→1 interface)	needs-corporate
labor_rate (buyer)	[TBD]	Loaded purchasing agent rate	needs-corporate
approval_cycle_reduction_days	[TBD]	Tabware flow vs Oracle flow vs unified	needs-corporate
part_creation_time_current	36+ hours	Warehouse admin: "36+ hours for part creation during breakdowns"	workshop-confirmed
systems_during_EAM_transition	3	John Sabo: "buyers will work in 3 systems" during transition	workshop-confirmed

Workshop-Sourced Range: $1-3M/yr Confidence: High — John Sabo (corporate cataloging) validated at corporate level, MDT readout pitched procurement as self-funding starter Key Quote: "[Tabware flow] buyer may work for nothing if financial approval fails." — John Sabo

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, PM	[TBD]	Map Tabware + Oracle + Ellipse approval flows
Data engineering	Data engineer	[TBD]	Multi-CMMS API integration
ML/AI development	ML engineer	[TBD]	Conversational AI front-end, intent classification
Application/UX	Frontend dev	[TBD]	Unified buyer interface
Infrastructure	Moderate	[TBD]	AI inference, multi-system connectors
Change management	—	[TBD]	Medium — buyer workflow change during EAM transition. 20%.

BH-04: Inventory Intelligence & Master Data Cleanup¶

Card Type: A — Anchored Corporate Project: PRJ-06

Value Analysis¶

Value Types: Inventory optimization + Cost avoidance Value Formula:

(inventory_value × carrying_cost_% × reduction_%)
+ (obsolete_parts_annual_reorder_cost)
+ (cross_site_sharing_savings_from_visibility)
+ (space_recovery_value_from_20yr_old_parts)

Variable	Value	Source	Status
inventory_value	$63M	Warehouse admin: "$63M inventory"	workshop-confirmed
unique_parts	19,000	Warehouse admin: "19,000 parts"	workshop-confirmed
warehouses	6	Warehouse admin: "6 warehouses"	workshop-confirmed
carrying_cost_%	25%	Industry standard	estimated
reduction_%	10-15%	Conservative (MDT confirmed 10% duplicates)	estimated
obsolete_parts_on_auto_reorder	[TBD]	Parts sitting 20+ years still auto-ordering	needs-corporate
annual_obsolete_reorder_cost	[TBD]	Count × avg cost	needs-corporate
cross_site_visibility_status	"Tabware siloed per plant — IH can't see BH inventory"	John Sabo	workshop-confirmed
mining_benchmark	Mary reviews recommended orders daily, 15 yrs	John Sabo	workshop-confirmed

Workshop-Sourced Range: $2-5M/yr Confidence: High — $63M inventory confirmed, 19K parts confirmed, John Sabo validated mining vs steel maturity gap Key Quote: "I'm more concerned about the actual on-hand counts than the min/maxes." — John Sabo. "Mining does this right. Steel doesn't have an equivalent."

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, PM	[TBD]	Inventory audit, duplicate taxonomy
Data engineering	Data engineer	[TBD]	Tabware + Oracle cross-reference, Pi-Log integration
ML/AI development	Data scientist	[TBD]	Duplicate detection (beyond Pi-Log), obsolescence scoring
Application/UX	Frontend dev	[TBD]	Inventory dashboard, alert system
Infrastructure	Minimal	[TBD]	—
Change management	—	[TBD]	High — inventory policy change, cross-plant visibility. 25%.

BH-02: Maintenance Copilot (Voice Capture + Technician Assist)¶

Card Type: B — Structured Corporate Project: PRJ-06

Value Analysis¶

Value Types: Efficiency gain + Data quality uplift Value Formula:

(diagnosis_time_saved_per_repair × repairs_per_month × labor_cost_per_hour)
+ (documentation_compliance_improvement × work_order_quality_uplift_value)

Variable	Value	Source	Status
diagnosis_time_saved_per_repair	[TBD]	Estimate 30-60 min/repair (per CLV evidence)	needs-corporate
repairs_per_month	[TBD]	Tabware work order volume	needs-corporate
labor_cost_per_hour	[TBD]	Blended maintenance tech rate	needs-corporate
documentation_compliance_improvement	[TBD]	Current vs target work order completion rate	needs-corporate
wifi_coverage	[TBD]	Gaps in coke plant, plate mill, BF area	needs-corporate

Workshop-Sourced Range: $0.5-2M/yr + data quality uplift (enabler for PdM and analytics) Confidence: Medium — cross-site validated (CLV, MDT, TLD) but not deeply discussed at BH. Wi-Fi coverage is the constraint.

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, UX researcher, PM	[TBD]	Field shadowing, connectivity assessment
Data engineering	Data engineer	[TBD]	Tabware API, knowledge base ingestion
ML/AI development	ML engineer	[TBD]	Voice-to-structured (LLM/STT), RAG over repair history
Application/UX	Frontend dev, mobile dev	[TBD]	Voice-first mobile app, offline-capable
Infrastructure	Moderate	[TBD]	STT/LLM inference, offline sync, Wi-Fi dependency
Change management	—	[TBD]	High — USW receptivity unknown, trust critical. 25%.

BH-27: Part Visual Identification¶

Card Type: C — Absorbed Corporate Project: PRJ-06 Reason: Quick win within BH-P06 scope. Image catalog for 19K parts accelerates receiving and cycle counts. Value Contribution: $0.5-1M/yr — absorbed into BH-P06. Enabler for cycle count digitization and inventory accuracy. Cost Contribution: Image capture workflow + AI classifier — bounded within BH-P06 scope.

BH-28: Cycle Count Digitization (Paper → Tablet)¶

Card Type: C — Absorbed Corporate Project: PRJ-06 Reason: Already in testing. Warehouse Admin leading. Low complexity, high morale impact. Value Contribution: $0.3-0.5M/yr — labor savings + accuracy. Absorbed into BH-P06. Cost Contribution: Tablet app development — minimal ML, bounded scope.

BH-29: Min/Max Intelligent Management & Reorder Optimization¶

Card Type: B — Structured Corporate Project: PRJ-06

Value Analysis¶

Value Types: Cost avoidance + Efficiency gain Value Formula:

(obsolete_parts_auto_reorder_cost_per_year)
+ (stockout_events_from_bad_minmax × downtime_cost_per_event)
+ (minmax_management_labor_hours × labor_rate × 12)

Variable	Value	Source	Status
parts_on_auto_reorder	[TBD]	Tabware replenish module settings	needs-corporate
obsolete_parts_auto_reorder_cost	[TBD]	"Huge waste of money" — Warehouse Admin	needs-corporate
stockout_events_per_year	[TBD]	From bad min/max settings	needs-corporate
downtime_cost_per_event	[TBD]	Production loss per stockout	needs-corporate
minmax_change_volume_per_month	[TBD]	Currently tracked via email only	needs-corporate
labor_hours_on_minmax_management	[TBD]	Manual email search + coordination	needs-corporate

Workshop-Sourced Range: $1-3M/yr Confidence: Medium — Warehouse Admin building v1, clear pain articulated Key Quote: "Min/max numbers are very arbitrary — someone out in the mill tells us and we just do it." "It has never been built." (change log)

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Data scientist, PM	[TBD]	Order history analysis, criticality classification
Data engineering	Data engineer	[TBD]	SQL data warehouse extraction
ML/AI development	Data scientist	[TBD]	Probability-based reorder thresholds
Application/UX	Frontend dev	[TBD]	Change log + notification system
Infrastructure	Minimal	[TBD]	—
Change management	—	[TBD]	Low — Warehouse Admin already building foundation. 15%.

BH-32: Vendor Follow-Up & Procurement Tracking Automation¶

Card Type: C — Absorbed Corporate Project: PRJ-06 Reason: Quick win — automated PO follow-up from existing data. Absorbed into BH-P06 procurement automation scope. Value Contribution: $0.5-1M/yr — reduced stockouts from forgotten orders. Absorbed into BH-P06. Cost Contribution: Rules engine + notification system within BH-P06 scope.

BH-33: Requisition Real-Time Alerting & Pick List Automation¶

Card Type: C — Absorbed Corporate Project: PRJ-06 Reason: Quick win addressing 24-hour Tabware data refresh lag. Dependent on BH-P17 (read-replica). Value Contribution: $0.3-1M/yr — reduced service delays. Absorbed into BH-P06. Cost Contribution: Event trigger + notification system. Blocked by Tabware refresh lag (BH-P17).

BH-40: Buyer Intelligence & Cross-Plant Analytics¶

Card Type: B — Structured Corporate Project: PRJ-06

Value Analysis¶

Value Types: Cost avoidance + Efficiency gain Value Formula:

(pricing_improvement_% × annual_procurement_spend)
+ (buyer_time_saved_hours × labor_rate × buyers × 12)

Variable	Value	Source	Status
annual_procurement_spend (steel plants)	[TBD]	Aggregated across all steel plants	needs-corporate
pricing_improvement_%	1-3%	From cross-plant visibility on commodity pricing	estimated
buyer_time_saved_hours	[TBD]	Manual query building → AI query	needs-corporate
labor_rate (buyer)	[TBD]	Loaded purchasing agent rate	needs-corporate
buyers_per_commodity	1 across all steel plants	John Sabo: "one buyer per commodity"	workshop-confirmed

Workshop-Sourced Range: $1-3M/yr Confidence: Medium — John Sabo validated the need, one-buyer-per-commodity structure is right for cross-plant AI Key Quote: "Over the past 10 years, what has been the average price on this? — I think it would be very beneficial." — John Sabo

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, PM	[TBD]	Commodity group prioritization
Data engineering	Data engineer	[TBD]	Cross-plant procurement data aggregation
ML/AI development	Data scientist	[TBD]	Pricing analytics, agreement tracking
Application/UX	Frontend dev	[TBD]	Buyer intelligence dashboard
Infrastructure	Moderate	[TBD]	Cross-plant data access
Change management	—	[TBD]	Low-Medium — buyers already savvy for blankets. 15%.

BH-P07: Through-Process Quality & Yield¶

BH-09: Through-Process Quality Traceability¶

Card Type: B — Structured Corporate Project: PRJ-04

Value Analysis¶

Value Types: Yield improvement + Cost avoidance Value Formula:

quality_loss_per_year × traceability_improvement_% × yield_recovery_rate
+ cross_process_defect_attribution_value (root cause across BF→BOF→caster→HSM→plate)

Variable	Value	Source	Status
quality_loss_per_year (BH)	[TBD]	Scrap + downgrade + rework across all process stages	needs-corporate
traceability_improvement_%	15-25%	Conservative — longest process chain in CLF amplifies this	estimated
yield_recovery_rate	[TBD]	What % of identified losses are actionable	needs-corporate
PA_domains	4 (Patrick → Doug → Eric → Matt)	PA group structure maps to process chain	workshop-confirmed
quality_review_delay	~24 hrs for coils	QMS flags but manual review next day	workshop-confirmed

Workshop-Sourced Range: $5-15M/yr Confidence: Medium — data exists across every process stage but 4 PA domains = 4 organizational silos to bridge

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, domain specialist, PM	[TBD]	Cross-PA-domain data mapping
Data engineering	Data engineer (senior)	[TBD]	4 PA SQL domains + QMS + IBA integration
ML/AI development	ML engineer, data scientist	[TBD]	Cross-process defect attribution model
Application/UX	Frontend dev	[TBD]	Through-process quality dashboard
Infrastructure	Significant	[TBD]	4 read-replicas needed (BH-P17)
Change management	—	[TBD]	High — cross-PA-domain coordination. 25%.

BH-10: Surface Defect Detection / SIS Enhancement¶

Card Type: C — Absorbed Corporate Project: PRJ-04 Reason: Seed — Palmer priority but SIS status at BH is unknown. Strongest evidence at MDT (Ametek 60% accuracy). Deferred until BH SIS baseline established. Value Contribution: $2-8M/yr industry estimate. Absorbed into BH-P07. Cost Contribution: Classifier retraining — bounded ML. Entry at MDT, scale to BH.

BH-11: Cobble Prediction & Prevention (HSM)¶

Card Type: B — Structured Corporate Project: PRJ-05

Value Analysis¶

Value Types: Cost avoidance + Throughput gain Value Formula:

cobbles_per_year × (equipment_damage_cost + downtime_hours × production_value_per_hour + scrap_tons × margin_per_ton) × prevention_rate

Variable	Value	Source	Status
cobble_rate	0.4% last year (higher recently)	Senior ops	workshop-confirmed
cobbles_per_year	[TBD]	Cobble rate × bars rolled per year	needs-corporate
equipment_damage_cost_per_cobble	[TBD]	Drive spindle, work roll damage	needs-corporate
downtime_hours_per_cobble	[TBD]	HSM delay data	needs-corporate
production_value_per_hour (HSM)	[TBD]	HSM throughput × margin	needs-corporate
scrap_tons_per_cobble	[TBD]	Quality records	needs-corporate
margin_per_ton	[TBD]	Product mix average	needs-corporate
prevention_rate	15-20%	LOWER than CLV estimate — prior AI failure at BH (2017-2018) demands humility	estimated
prior_AI_attempt	Failed — California startup, 6 months, "tried and tried and faded away"	Senior ops	workshop-confirmed
GE_rolling_model_access	Source code available (unlike MDT Siemens black box)	Process control	workshop-confirmed

Workshop-Sourced Range: $2-8M/yr Confidence: Low-Medium — prior AI failure is institutional memory. GE source code access is a genuine advantage. Modern LLM/transformer approaches differ from 2017-era ML. Key Quote: "100% we are interested." — Senior ops (despite prior failure). "What they see, smell, hear — isn't captured." (missing piece identified)

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, data scientist, PM	[TBD]	Forensic review of 2017-18 failure FIRST
Data engineering	Data engineer	[TBD]	IBA server + GE model data extraction
ML/AI development	ML engineer, data scientist	[TBD]	Transformer-based risk model (not 2017-era ML)
Application/UX	Frontend dev	[TBD]	Operator risk score integrated with HMI
Infrastructure	Moderate	[TBD]	Real-time inference at HSM
Change management	—	[TBD]	High — prior failure, operator trust critical. 25%.

BH-37: Strip Steering / Bruise Prediction¶

Card Type: B — Structured Corporate Project: PRJ-05

Value Analysis¶

Value Types: Cost avoidance + Yield improvement Value Formula:

bruise_rejection_rate × annual_HSM_tons × margin_per_ton × reduction_%
+ cobble_prevention_from_steering × cobbles_prevented × cost_per_cobble

Variable	Value	Source	Status
bruise_rejection_rate	0.23% (Feb), 0.4% per group	Senior ops: "millions in value"	workshop-confirmed
annual_HSM_tons	[TBD]	HSM production volume	needs-corporate
margin_per_ton	[TBD]	Product mix average	needs-corporate
reduction_%	20-30%	With TDF optimization + camera-based steering	estimated
TDF_program_exists	Yes — predicts differential force in F2	Process control	workshop-confirmed
GE_source_code_access	Yes — unlike MDT Siemens black box	Process control	workshop-confirmed
capital_constraint	Clutch removal = unfunded capital requirement	Senior ops	workshop-confirmed

Workshop-Sourced Range: $3-8M/yr Confidence: Medium — software-only path may exist (TDF optimization + GE model), but capital constraint (clutch removal) limits full potential Key Quote: "If we had the technology 50 years ago, we would have done it. If we had the money 20 years ago, we would have done it."

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, domain specialist	[TBD]	TDF utilization analysis, camera feasibility
Data engineering	Data engineer	[TBD]	GE rolling model data + TDF + camera integration
ML/AI development	ML engineer	[TBD]	Steering prediction, TDF optimization
Application/UX	Frontend dev	[TBD]	Operator steering guidance display
Infrastructure	Moderate	[TBD]	Real-time inference, possible camera hardware
Change management	—	[TBD]	Medium — operator-dependent (TDF utilization varies). 20%.

BH-P08: PdM Platform — Belt System & Multi-Asset¶

BH-53: Belt System Instrumentation & PdM¶

Card Type: B — Structured Corporate Project: PRJ-03

Value Analysis¶

Value Types: Cost avoidance + Throughput gain Value Formula:

belt_failure_events_per_year × (BF_downtime_hours × production_value_per_hour + repair_cost)
× prevention_rate

Variable	Value	Source	Status
belt_system_length	7.5 miles	BF Process Engineer: "seven and a half miles"	workshop-confirmed
belt_failure_events_per_year	[TBD]	BF delay records — belt-attributed	needs-corporate
BF_downtime_hours_per_failure	[TBD]	Historical repair times	needs-corporate
production_value_per_hour (BF)	[TBD]	BF throughput × margin	needs-corporate
repair_cost_per_failure	[TBD]	Belt replacement + labor	needs-corporate
prevention_rate	20-30%	Motor amp Phase 1; higher with hardwired sensors Phase 2	estimated
motor_amp_data_availability	Exists across most equipment	BF Process Engineer	workshop-confirmed
prior_sensor_failure	Battery-powered vibration sensors failed	"We no longer buy anything. Much rather have been hardwired"	workshop-confirmed

Workshop-Sourced Range: $2-5M/yr Confidence: Medium — motor amp data as Phase 1 is low-cost, but full instrumentation requires capital approval. Belt failures directly shut BFs. Key Quote: "It's seven and a half miles of relying on people to walk that and listen and see and hear. And sometimes you can't hear temperature."

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, data scientist, PM	[TBD]	Motor amp baseline, critical belt segment identification
Data engineering	Data engineer	[TBD]	Motor amp data extraction from HMI
ML/AI development	ML engineer	[TBD]	Anomaly detection on motor amp patterns
Application/UX	Frontend dev	[TBD]	Belt health dashboard
Infrastructure	Phase 1: Minimal. Phase 3: Significant	[TBD]	Phase 3 = hardwired temp/vibration sensors ($$$)
Change management	—	[TBD]	Low — BF engineer is champion. 10%.

BH-05: PdM Platform (Multi-Asset)¶

Card Type: B — Structured Corporate Project: PRJ-03

Value Analysis¶

Value Types: Cost avoidance + Throughput gain Value Formula:

Σ (asset_class_failures_per_year × (downtime_cost + repair_cost) × prevention_rate)
across: BFs, BOFs, coke ovens, cranes, critical rotating equipment

Variable	Value	Source	Status
asset_classes	2 BFs, 3 BOFs, 164 coke ovens, cranes, plate mill	Largest equipment base in CLF	workshop-confirmed
failures_per_year_per_class	[TBD]	Tabware + delay records per asset	needs-corporate
downtime_cost_per_failure_per_class	[TBD]	By asset class	needs-corporate
repair_cost_per_failure_per_class	[TBD]	By asset class	needs-corporate
prevention_rate	30-50%	H2 target across multiple assets	estimated
drew_taylor_endorsement	PA-vouched "forward thinker"	PA group	workshop-confirmed

Workshop-Sourced Range: $3-12M/yr Confidence: Medium — largest equipment base in CLF = highest volume, but broad scope requires prioritization

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, data scientist, PM	[TBD]	Asset prioritization, data audit per class
Data engineering	Data engineer	[TBD]	Multi-source: Tabware + HMI + third-party PdM reports
ML/AI development	ML engineer, data scientist	[TBD]	Per-asset anomaly models
Application/UX	Frontend dev	[TBD]	Asset health dashboard, alert management
Infrastructure	Moderate	[TBD]	ML inference, multi-system data pipeline
Change management	—	[TBD]	Medium — cross-department rollout. 20%.

BH-55: BF Alert Triage & Intelligent Alarm Management¶

Card Type: C — Absorbed Corporate Project: PRJ-03 / PRJ-01 Reason: Nuisance alarm suppression and failure trending across 100+ HMI screens. Enabler for BH-P08 PdM platform. Value Contribution: $1-3M/yr — reduced unplanned downtime from missed trends. Absorbed into BH-P08. Cost Contribution: Alert analytics layer within BH-P08 scope. Key Quote: "I think the thing that the AI could help us with is trending failures that we don't see." — BF Process Engineer

BH-45: PdM Alert Triage & Automated Escalation (IH-Sourced)¶

Card Type: A — Anchored Corporate Project: PRJ-03

Value Analysis¶

Value Types: Cost avoidance (prevent already-predicted failures) Value Formula:

unread_PdM_reports_per_month × critical_alert_% × avg_failure_cost × prevention_rate

Variable	Value	Source	Status
unread_PdM_reports_per_month	282	Al: "282 since the beginning of the month that I haven't even looked at"	workshop-confirmed
critical_alert_%	[TBD]	What % of reports contain actionable severity levels	needs-corporate
avg_failure_cost_per_missed_alert	[TBD]	Historical failures with unread prior warnings	needs-corporate
prevention_rate	50-70%	High — these are ALREADY predicted failures, just not actioned	estimated
third_party_providers	ITR (vibration + thermography), Shell (oil sampling)	Al	workshop-confirmed

Workshop-Sourced Range: $1-3M/yr Confidence: High — zero infrastructure required, failures are already predicted, just unread. Immediate impact. Key Quote: "We've had plenty of failures where we look back — it's sitting in my inbox from two weeks ago. Nobody looked at it." — Al

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, PM	[TBD]	Map third-party report formats
Data engineering	Data engineer	[TBD]	Email/PDF ingestion pipeline
ML/AI development	NLP engineer	[TBD]	Report parsing, severity extraction, auto-escalation
Application/UX	Frontend dev	[TBD]	Critical alert dashboard, Tabware WO auto-creation
Infrastructure	Minimal	[TBD]	Email integration, no new hardware
Change management	—	[TBD]	Low — Al is champion, new supervisors need this. 10%.

BH-P09: BF Process Intelligence & Raw Materials¶

BH-13: BF Stove Optimization & Raw Material¶

Card Type: B — Structured (WITH CRITICAL CAVEAT) Corporate Project: PRJ-05

Value Analysis¶

Value Types: Energy efficiency + Throughput gain Value Formula:

(stove_energy_optimization_% × annual_stove_energy_cost)
+ (edge_case_prevention_events × production_loss_per_event)
+ (multi_furnace_coordination_improvement × throughput_gain)

Variable	Value	Source	Status
annual_stove_energy_cost	[TBD]	Gas consumption for stove heating	needs-corporate
stove_energy_optimization_%	[TBD]	Incremental over existing thermal model	needs-corporate
edge_case_events_per_year	[TBD]	Events beyond physics model prediction	needs-corporate
production_loss_per_edge_case	[TBD]	BF downtime during unmodeled scenarios	needs-corporate
existing_thermal_model_quality	"I go weeks on end without adjustments"	BF Process Engineer	workshop-confirmed
prior_AI_vendor_trial	4 months, NO incremental value found	BF Process Engineer	workshop-confirmed

Workshop-Sourced Range: $3-10M/yr (HIGHLY UNCERTAIN — existing model is very good) Confidence: Low — BF Process Engineer's 4-month AI vendor trial found zero value. Palmer flagged BF stove optimization but BH counter-evidence must be communicated. Value must come from areas BEYOND the existing thermal model.

Framing caution: The BF Process Engineer is the most technically sophisticated individual across 4 sites. Any AI pitch must demonstrate value beyond what he already has. Frame as augmenting his system, not replacing it.

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, domain specialist, PM	[TBD]	Existing model audit — what CAN'T it do?
Data engineering	Data engineer	[TBD]	Thermal model output + edge case data
ML/AI development	ML engineer	[TBD]	Complement to physics model (hybrid AI)
Application/UX	Frontend dev	[TBD]	Edge case alerting
Infrastructure	Minimal	[TBD]	On-prem
Change management	—	[TBD]	High — BF engineer must believe in it. 25%.

BH-14: BF Burden Mix / Raw Material Optimization¶

Card Type: B — Structured Corporate Project: PRJ-05

Value Analysis¶

Value Types: Cost avoidance + Throughput gain Value Formula:

(burden_cost_optimization_% × annual_burden_cost)
+ (BF_productivity_gain_% × annual_BF_throughput × margin_per_ton)

Variable	Value	Source	Status
annual_burden_cost	[TBD]	Ore + sinter + coke volumes × prices	needs-corporate
burden_cost_optimization_%	[TBD]	Mix optimization potential	needs-corporate
BF_productivity_gain_%	[TBD]	From burden chemistry optimization	needs-corporate
annual_BF_throughput	[TBD]	2 BFs × throughput	needs-corporate
margin_per_ton	[TBD]	Hot metal margin	needs-corporate
BH_closed_loop_advantage	On-site coke + sinter = unique control	BF session	workshop-confirmed
stock_house_ML	Already handles feed rates (NOT burden chemistry)	BF Process Engineer	workshop-confirmed

Workshop-Sourced Range: $5-15M/yr Confidence: Low-Medium — BH's unique coke + sinter closed loop is a genuine advantage, but stock house ML already handles feed rates. Gap is burden chemistry optimization.

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, domain specialist	[TBD]	Burden chemistry parameter mapping
Data engineering	Data engineer	[TBD]	Coke quality + sinter quality + BF process data
ML/AI development	ML engineer, optimization specialist	[TBD]	Multi-objective burden optimization
Application/UX	Frontend dev	[TBD]	Burden recommendation dashboard
Infrastructure	Moderate	[TBD]	Cross-system (coke plant + sinter + BF)
Change management	—	[TBD]	Medium — BF engineer buy-in critical. 20%.

BH-19: Sinter Plant Optimization¶

Card Type: C — Absorbed Corporate Project: new (BH-unique) Reason: Seed — no existing model to compete with (unlike BF). Known ML application in steel. BH-unique asset (2,800 kt/yr). Clearest open gap in BH-P09. Value Contribution: $2-5M/yr — absorbed into BH-P09. Phase 1 priority within BH-P09 because no existing model competition. Cost Contribution: ML model within BH-P09 scope. Requires sinter plant DCS data.

BH-23: Operator Decision Support (BF/BOF/HSM)¶

Card Type: C — Absorbed Corporate Project: PRJ-05 Reason: Broad scope — value captured across BH-P02 (BOF), BH-P07 (HSM/quality), BH-P09 (BF). Not standalone. Value Contribution: $1-5M/yr — distributed across parent projects. Senior ops explicitly asked for recipe recommendations by grade. Cost Contribution: Decision models per process area within respective parent project scopes.

BH-P10: Knowledge Capture / Virtual SME¶

BH-08: Knowledge Capture / Virtual SME¶

Card Type: A — Anchored Corporate Project: Virtual SME (cross-site)

Value Analysis¶

Value Types: Risk mitigation + Efficiency gain Value Formula:

(knowledge_flight_risk_cost × probability_of_departure × capture_rate)
+ (new_employee_ramp_time_reduction × new_hires_per_year × labor_rate)
+ (rediscovery_avoidance_events × cost_per_rediscovery)

Variable	Value	Source	Status
coke_plant_retirement_risk	3 of 5 section managers at retirement age, age-74 electrical manager (54 yrs tenure)	Coke Plant Div Mgr	workshop-confirmed
BF_two_person_dependency	BF Process Engineer + Bill hold entire process control capability	BF session	workshop-confirmed
IH_turnover	~100 people since 2019, 3 of 4 supervisors new within last year	John (IH)	workshop-confirmed
PhD_model_loss	Coal blend models lost with PhDs let go ~1.5 yrs ago	Coke Plant Div Mgr: "Nobody seems to know where they're at"	workshop-confirmed
Bill_HMI_value	100+ screens built over 15 years	BF session: "He's a brilliant man"	workshop-confirmed
John_already_building	Uploading to SharePoint, training personal AI model	IH session	workshop-confirmed
caster_alignment_study_found	"From years ago, describes EXACT current problem with fix plan — filed away and forgotten"	John (IH)	workshop-confirmed
knowledge_flight_risk_cost	[TBD]	Replacement + capability loss + operational degradation	needs-corporate
new_hires_per_year	[TBD]	Hiring/turnover rate	needs-corporate

Workshop-Sourced Range: $0.5-2M/yr + incalculable risk mitigation Confidence: High on need (5th consecutive site validation, most acute knowledge flight risk in CLF), Medium on execution (expert receptivity is the bottleneck — "One of them, maybe?") Key Quotes: "A caster alignment study from years ago describes the EXACT current problem with a fix plan — filed away and forgotten." — John (IH). "People too embarrassed to ask for help." — John. "He's a brilliant man. He's not a people person." — about Bill.

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, UX researcher, PM	[TBD]	Expert interview strategy (non-intrusive)
Data engineering	Data engineer	[TBD]	SharePoint + iFix + Vault + document ingestion
ML/AI development	ML engineer	[TBD]	RAG system, knowledge graph, per-department agents
Application/UX	Frontend dev	[TBD]	Chat interface per department, search over knowledge base
Infrastructure	Moderate	[TBD]	LLM inference, document processing pipeline
Change management	—	[TBD]	Critical — expert resistance, IT policy blockers (John needs Vault access). 25%. Frame as "succession planning for the automation you built."

BH-P11: Cross-System Data Unification & AI Query Layer¶

BH-39: Cross-System Data Unification / AI Query Layer¶

Card Type: A — Anchored Corporate Project: PRJ-01

Value Analysis¶

Value Types: Efficiency gain + Foundation (enabler) Value Formula:

(analyst_hours_saved_per_week × labor_rate × analysts × 52)
+ (decision_quality_improvement × decisions_per_year × avg_decision_value)
+ (EAM_transition_bridge_value)

Variable	Value	Source	Status
Eric_manual_hours	~80% of day	Eric: "80% of his day pulling and manipulating reports manually across 3 databases"	workshop-confirmed
databases_with_different_schemas	3+ (Tabware, Oracle, Ellipse)	Eric + John Sabo	workshop-confirmed
prior_AI_attempt	Failed — "actually took longer to answer the AI questions"	Eric	workshop-confirmed
Lisa_architecture_docs	Exists in SharePoint — agreed to share	Lisa	workshop-confirmed
EAM_migration_timeline	Sep 2026 Cleveland → mid-2027 all plants	John Sabo	workshop-confirmed
cross_plant_instance	"EAM still won't fix this — still siloed, no cross-plant instance"	John Sabo	workshop-confirmed
analyst_hours_saved_per_week	[TBD]	Quantify Eric's manual work	needs-corporate
labor_rate (analyst)	[TBD]	Loaded rate	needs-corporate
analysts_affected	[TBD]	How many Erics across CLF	needs-corporate

Workshop-Sourced Range: $1-3M/yr + foundation value for every data-dependent initiative Confidence: High — Lisa has the architecture documentation and agreed to share. Problem is precisely defined. EAM migration creates both urgency and opportunity. Key Quote: "He takes English, Spanish, and German, and makes it all speak one language." — about Eric's manual work

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, PM	[TBD]	Lisa's SharePoint architecture ingestion
Data engineering	Data engineer (senior)	[TBD]	Cross-system data dictionary, schema mapping
ML/AI development	ML engineer	[TBD]	Natural language query layer over unified schema
Application/UX	Frontend dev	[TBD]	Query interface for buyers/analysts
Infrastructure	Moderate	[TBD]	LLM inference, multi-system connectors
Change management	—	[TBD]	Low — Eric IS the demand signal. 10%.

Note: If we get one deliverable from Burns Harbor, it should be Lisa's architecture documentation. It describes the corporate-level data flow that creates every site's information flow problem.

BH-52: Integration Handoff Monitoring & Auto-Remediation¶

Card Type: C — Absorbed Corporate Project: PRJ-01 Reason: Monitoring layer for SAP ↔ legacy system handoffs. 13+ integration projects, failed handoffs cascade. Absorbed into BH-P11. Value Contribution: $0.5-1M/yr — data quality preservation. Absorbed into BH-P11. Cost Contribution: Event monitoring + alerting within BH-P11 scope.

BH-P12: Enterprise Scheduling & S&IOP¶

BH-25: Cross-Stage Scheduling / S&IOP¶

Card Type: B — Structured Corporate Project: PRJ-02

Value Analysis¶

Value Types: Throughput gain + Efficiency gain Value Formula:

(additional_heats_per_day_from_optimization × margin_per_heat × 365)
+ (changeover_reduction_% × changeover_hours_per_year × production_value_per_hour)
+ (missed_shipdate_reduction_% × penalty_per_missed × shipments_per_year)

Variable	Value	Source	Status
BH_scheduling_complexity	Dual product (flat-rolled + plate) + coke/sinter + 2BFs + 3BOFs	Most complex in CLF	workshop-confirmed
additional_heats_per_day	[TBD]	Scheduling optimization potential	needs-corporate
margin_per_heat	[TBD]	~300 tons × margin/ton	needs-corporate
changeover_hours_per_year	[TBD]	HSM + plate mill changeovers	needs-corporate
production_value_per_hour	[TBD]	Combined throughput	needs-corporate
missed_shipdate_events	[TBD]	Commercial data	needs-corporate

Workshop-Sourced Range: $10-30M/yr Confidence: Low-Medium — massive scope, BH is most complex scheduling environment. H3 for a reason.

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, domain specialist, PM	[TBD]	Cross-functional scheduling mapping
Data engineering	Data engineer (senior)	[TBD]	SAP + MES + L-scheduler + legacy integration
ML/AI development	ML engineer, optimization specialist	[TBD]	Constraint optimization, demand forecasting
Application/UX	Frontend dev	[TBD]	Planning dashboard
Infrastructure	Significant	[TBD]	Real-time multi-system feeds
Change management	—	[TBD]	Very high — cross-functional. 30%.

BH-49: Demand Forecasting & Market Intelligence¶

Card Type: C — Absorbed Corporate Project: PRJ-02 Reason: Strategic H2 — AI-enhanced demand forecasting from SAP IBP. All data centralized. Absorbed into BH-P12. Value Contribution: $3-10M/yr estimated. Absorbed into BH-P12 roll-up. Cost Contribution: Forecasting model within BH-P12 scope.

BH-50: Cross-Plant Order Reallocation Automation ("Fast Path")¶

Card Type: A — Anchored Corporate Project: PRJ-02

Value Analysis¶

Value Types: Efficiency gain Value Formula:

reallocation_events_per_month × manual_reentry_hours_per_event × labor_rate × 12
+ reallocation_cycle_time_reduction_days × events_per_month × margin_per_order × 12

Variable	Value	Source	Status
reallocation_events_per_month	[TBD]	Cross-plant order transfers	needs-corporate
manual_reentry_hours_per_event	[TBD]	Lisa: "all customer data exists in SAP master data but must be recreated in destination legacy system"	needs-corporate
labor_rate	[TBD]	Customer service / planning rate	needs-corporate
reallocation_cycle_time_current	[TBD]	Current turnaround time	needs-corporate
SAP_master_data_completeness	"That's fairly easy to manage"	Lisa	workshop-confirmed

Workshop-Sourced Range: $0.5-2M/yr Confidence: Medium-High — Lisa confirmed "fairly easy," SAP master data exists. Bounded, testable quick win. Key Quote: "We need to leverage that information and take this record and go from Burns Harbor to Indiana Harbor for production." — Lisa

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, PM	[TBD]	SAP → legacy format mapping per plant
Data engineering	Data engineer	[TBD]	SAP API + legacy system connectors
ML/AI development	Minimal	[TBD]	Template generation, format translation
Application/UX	Frontend dev	[TBD]	"Fast Path" button interface
Infrastructure	Minimal	[TBD]	SAP integration
Change management	—	[TBD]	Low — Lisa is champion. 10%.

BH-P13: Intra-Plant Logistics & Warehouse Digitization¶

BH-16: Intra-Plant Slab & Coil Logistics Optimization¶

Card Type: B — Structured Corporate Project: PRJ-07

Value Analysis¶

Value Types: Efficiency gain + Throughput gain Value Formula:

(current_moves_per_unit - target_moves) × units_per_month × cost_per_move
+ (truck_routing_time_saved_per_delivery × deliveries_per_day × labor_rate × 365)

Variable	Value	Source	Status
external_trucks_per_day	50-100	Warehouse: "50-100 external trucks/day"	workshop-confirmed
doors	~200 but 95% go to ~24 locations	Warehouse	workshop-confirmed
deliveries_unloaded_in_mill	90%	Warehouse: "90% of deliveries unloaded in-mill, not central spares"	workshop-confirmed
current_moves_per_unit	[TBD]	Movement data	needs-corporate
target_moves_per_unit	[TBD]	Optimized routing	needs-corporate
cost_per_move	[TBD]	Crane/vehicle + labor	needs-corporate
IE_prior_slab_study	Exists	IE previously studied slab movement at BH	workshop-confirmed

Workshop-Sourced Range: $2-5M/yr Confidence: Medium — IE prior slab study gives head start. Dual product streams (flat-rolled + plate) = most complex routing in CLF.

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, PM	[TBD]	Build on IE prior slab study
Data engineering	Data engineer	[TBD]	Movement tracking + GPS integration
ML/AI development	Optimization specialist	[TBD]	Routing optimization
Application/UX	Frontend dev	[TBD]	Real-time material flow dashboard
Infrastructure	Moderate	[TBD]	GPS, possible RFID expansion
Change management	—	[TBD]	Medium — crane/truck operator workflow. 20%.

Card Type: B — Structured Corporate Project: PRJ-07

Value Analysis¶

Value Types: Efficiency gain + Safety Value Formula:

(lost_driver_time_per_delivery × deliveries_per_day × labor_rate × 365)
+ (misdelivery_events_per_month × redelivery_cost × 12)
+ (fire_extinguisher_check_time_saved × locations × checks_per_year)

Variable	Value	Source	Status
lost_driver_time_per_delivery	[TBD]	GPS coordinates texted as workaround	needs-corporate
deliveries_per_day	50-100 external trucks	Warehouse	workshop-confirmed
labor_rate (driver)	[TBD]	External truck driver rate	needs-corporate
misdelivery_events_per_month	[TBD]	Wrong door deliveries from copy-pasted PO numbers	needs-corporate
redelivery_cost	[TBD]	Rerouting + delay	needs-corporate
Ford_model_reference	"GPS-activated within 2-mile radius, routes by PO"	Warehouse admin	workshop-confirmed
language_barriers	"Truck drivers with limited English"	Warehouse	workshop-confirmed

Workshop-Sourced Range: $1-3M/yr Confidence: Medium — Ford Vehicle Plant Locator is a proven model, but requires facility 3D scanning and cell coverage validation

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, PM	[TBD]	Plant mapping, GPS coverage assessment
Data engineering	Data engineer	[TBD]	PO → location mapping, GPS integration
ML/AI development	Minimal	[TBD]	Routing algorithm
Application/UX	Mobile dev	[TBD]	Driver-facing GPS app (multilingual)
Infrastructure	Moderate	[TBD]	GPS/cell coverage, possible 3D scanning
Change management	—	[TBD]	Low — drivers already receiving GPS coordinates. 10%.

BH-44: Hot Metal Logistics Optimization (IH-Specific)¶

Card Type: A — Anchored Corporate Project: PRJ-07

Value Analysis¶

Value Types: Throughput gain + Cost avoidance Value Formula:

logistics_delay_days_per_10 × avg_delay_cost_per_event × (365/10)
+ frozen_ladle_events_per_year × ladle_recovery_cost
+ empty_ladle_return_delay_hours × production_value_per_hour

Variable	Value	Source	Status
logistics_delay_frequency	6 out of 10 days	Al: "6 out of 10 days we have a delay"	workshop-confirmed
ladles_per_day	70	Al: "70 ladles/day"	workshop-confirmed
ladle_capacity	150-220 tons each	Al	workshop-confirmed
transit_time_across_canal	20-30 min	Al: rail bridge over Indiana Harbor Canal	workshop-confirmed
turnaround_cost	$200K+ per event	Cross-site evidence	workshop-confirmed
frozen_ladle_recovery	"weeks to recover"	Al: "metal sitting too long = freezes"	workshop-confirmed
coordination_contact_points	7:45am + 3pm calls only	"a lot happens between those two calls"	workshop-confirmed
avg_delay_cost_per_event	[TBD]	Production loss + ladle recovery	needs-corporate
ladle_recovery_cost	[TBD]	Thawing + refurbishment	needs-corporate

Workshop-Sourced Range: $3-8M/yr Confidence: Medium — 6/10 days have logistics delays is extraordinarily high, but IH is culturally challenging. Unique to IH (no other site has dual steel shop + canal). Key Quote: "A lot of this is coordination problems." — Don Zuki

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, PM	[TBD]	Ladle dispatch process mapping
Data engineering	Data engineer	[TBD]	GPS/scanner + scheduling + shop status integration
ML/AI development	Optimization specialist	[TBD]	Dispatch optimization, predictive routing
Application/UX	Mobile dev, frontend dev	[TBD]	Mobile dispatch for hot metal coordinator
Infrastructure	Moderate	[TBD]	Mobile access, real-time shop status feeds
Change management	—	[TBD]	Very high — 3 separate groups, cultural resistance. 30%.

BH-54: Hot Metal Temperature & Heat Loss Optimization¶

Card Type: C — Absorbed Corporate Project: PRJ-07 / PRJ-04 Reason: RFID + pyrometer infrastructure already built at BH BF area. Quick win on existing data — sub temperature at fill vs dump. Value captured in BH-P13 roll-up. Value Contribution: $1-3M/yr — temperature loss minimization improves downstream steel quality. Cost Contribution: Analytics on existing RFID/pyrometer data — minimal within BH-P13 scope.

BH-P14: Environmental Compliance & Carbon Capture¶

BH-07: Environmental Compliance Automation¶

Card Type: B — Structured Corporate Project: new (BH-unique)

Value Analysis¶

Value Types: Efficiency gain + Risk mitigation Value Formula:

manual_compliance_hours_per_year × labor_rate
+ compliance_gap_fine_risk × probability_reduction
+ push_opacity_correlation_value (Method 9 → charging event optimization)

Variable	Value	Source	Status
manual_compliance_hours_per_year	[TBD]	COMS reporting, Method 9 records, EPA filings	needs-corporate
labor_rate (environmental staff)	[TBD]	Loaded rate	needs-corporate
compliance_gap_fine_risk	[TBD]	100+ CWA violations (2016-2020), lead pollution record	needs-corporate
probability_reduction	50-80%	Automated tracking eliminates human gaps	estimated
push_opacity_data	Method 9 (manual visual) + COMS (continuous)	Coke Plant Div Mgr	workshop-confirmed
desulfurization_facility	None — low-sulfur coal is mandatory	Coke Plant Div Mgr	workshop-confirmed

Workshop-Sourced Range: $1-3M/yr + regulatory risk mitigation (fines, consent decrees) Confidence: Medium — politically sensitive ("everything here consider business confidential"), but highest EPA exposure in CLF

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Solution architect, PM	[TBD]	Compliance requirement mapping
Data engineering	Data engineer	[TBD]	COMS + Method 9 + charging event correlation
ML/AI development	Minimal	[TBD]	Rules-based automation, opacity correlation
Application/UX	Frontend dev	[TBD]	Compliance dashboard + automated reporting
Infrastructure	Minimal	[TBD]	—
Change management	—	[TBD]	Medium — data sensitivity concerns. 20%.

BH-24: Carbon Capture Monitoring & Optimization¶

Card Type: C — Absorbed Corporate Project: new (BH-unique) Reason: Seed — $50M system coming online but timeline and data streams unknown. Greenfield data opportunity when ready. Value Contribution: $1-3M/yr + regulatory/ESG value. Absorbed into BH-P14. Cost Contribution: Performance monitoring model — deferred until system commissioning.

BH-P15: Safety Analytics¶

BH-06: Safety Analytics & Incident Trend Prediction¶

Card Type: C — Absorbed Corporate Project: new Reason: Seed — no BH-specific evidence beyond safety incident history (BF explosion 2020, slag pit explosion 2021). Needs champion. Low direct $ but high political value — Palmer cares about this. Value Contribution: Low direct $, high political value. Not sized independently. Cost Contribution: Low — analytics on existing safety reporting data.

BH-P16: Warehouse Operations & Admin Automation¶

BH-30: Warehouse Scheduling & Admin Automation¶

Card Type: C — Absorbed Corporate Project: new Reason: Bundle of small repetitive tasks (shift scheduling, KPI auto-population, turn report). Warehouse Admin already building. Quick wins that compound. Value Contribution: $0.2-0.5M/yr — absorbed into BH-P16. Cost Contribution: Web app development, Tabware API integration — minimal. Key Quote: "52 hours a year, compound that over all career." — Warehouse Admin

BH-31: Inventory Forecasting for Budget Planning¶

Card Type: C — Absorbed Corporate Project: new Reason: Warehouse Admin already built v1 in Power BI + SQL. Wants AI enhancement (standard deviation, probability). Scalable to all sites. Value Contribution: $0.3-1M/yr — better budget accuracy. Absorbed into BH-P16. Cost Contribution: Statistical model on existing Power BI pipeline — minimal.

BH-P17: Infrastructure Enablers¶

BH-56: OT Network / Cloud Bandwidth Upgrade Assessment¶

Card Type: C — Absorbed (Enabler — PREREQUISITE) Corporate Project: PRJ-01 (enabler) Reason: No direct $ value. Blocks all cloud-based AI at BH. IT problem, not PA problem. Requires corporate IT. Value Contribution: Enabler — blocks $10M+ in AI project value. Not sized independently. Cost Contribution: Network assessment + corporate IT engagement. Not a Vooban/IE deliverable. Key Quote: "We have a very delicate and small pipe between Burns Harbor and the cloud... that's been the case for years and years and years." — Patrick (PA Mgr)

BH-57: Production Database Read-Replica Provisioning¶

Card Type: A — Anchored (PREREQUISITE) Corporate Project: PRJ-01 (enabler)

Value Analysis¶

Value Types: Enabler (risk mitigation) Value Formula:

No direct value formula — this is a prerequisite.
Enables: BH-P02 ($11-33M), BH-P07 ($15-43M), BH-P01 ($22-60M), BH-P05 ($8-22M)
Total blocked value: $56-158M/yr across dependent projects

Variable	Value	Source	Status
PA_areas_needing_replicas	3 of 4 (Doug Fortner steel making FIRST, Eric Carter finishing, Patrick PA)	PA group	workshop-confirmed
existing_read_replica	1 of 4 (Matt Barney's hot mill only)	PA group	workshop-confirmed
query_risk_without_replica	"you may not take anything down, but you might kill a bunch of jobs"	PA group	workshop-confirmed
implementation_cost	$0.1-0.3M	Standard SQL Server replication	estimated

Workshop-Sourced Range: $0.1-0.3M implementation cost (enabler, not revenue) Confidence: High — standard SQL Server work, PA group understands it, quick win

Cost Analysis¶

Component	Vooban Team	IE	Notes
Discovery & design	Minimal	[TBD]	PA group already scoped it
Data engineering	Data engineer	[TBD]	SQL Server replication setup per area
ML/AI development	None	[TBD]	—
Application/UX	None	[TBD]	—
Infrastructure	Low	[TBD]	Additional SQL Server instances
Change management	—	[TBD]	Minimal — PA group offered. 5%.

Project Roll-Ups¶

Project	Initiatives	Anchored (A)	Structured (B)	Absorbed (C)	Workshop Range	Confidence
BH-P01 Coil Velocity & Shipping	BH-34, BH-35, BH-38, BH-17	3	0	1	$22-60M/yr	High
BH-P02 BOF/Caster Chemistry	BH-41, BH-15, BH-42, BH-12	3	0	1	$11-33M/yr	High
BH-P03 Coke Plant Ops	BH-46, BH-18, BH-48, BH-47	1	3	0	$7-17M/yr	Med-High
BH-P04 Plate Mill Shipping	BH-43, BH-20	1	1	0	$4-13M/yr	High
BH-P05 Ops-Maint Integration	BH-01, BH-36, BH-21, BH-22	1	1	2	$8-22M/yr	High
BH-P06 Maint Workflow & Inventory	BH-03, BH-04, BH-02, BH-27, BH-28, BH-29, BH-32, BH-33, BH-40	2	2	5	$7-19M/yr	High
BH-P07 Quality & Yield	BH-09, BH-35*, BH-10, BH-11, BH-37	0	3	1	$15-43M/yr	Medium
BH-P08 PdM Belt & Multi-Asset	BH-53, BH-05, BH-55, BH-45	1	2	1	$7-23M/yr	Medium
BH-P09 BF Process Intelligence	BH-13, BH-14, BH-19, BH-23	0	2	2	$11-35M/yr	Low-Med
BH-P10 Knowledge Capture	BH-08	1	0	0	$0.5-2M + risk	High
BH-P11 Data Unification	BH-39, BH-52	1	0	1	$1.5-4M/yr	High
BH-P12 Scheduling & S&IOP	BH-25, BH-49, BH-50	1	1	1	$14-42M/yr	Low-Med
BH-P13 Logistics & Warehouse	BH-16, BH-26, BH-44, BH-54	1	2	1	$7-19M/yr	Medium
BH-P14 Environmental & Carbon	BH-07, BH-24	0	1	1	$2-6M/yr	Medium
BH-P15 Safety Analytics	BH-06	0	0	1	Low	Low
BH-P16 Warehouse Admin	BH-30, BH-31	0	0	2	$0.5-1.5M/yr	High
BH-P17 Infrastructure Enablers	BH-56, BH-57	1	0	1	Enabler ($0.1-0.3M cost)	High
TOTAL	57	17	18	22	$110-340M/yr

BH-35 is shared between BH-P01 and BH-P07. Value counted once in BH-P01.

Card type distribution: 17 Anchored (30%), 18 Structured (32%), 22 Absorbed (39%). The 35 cards with formulas (A+B) cover the bulk of the value — the 22 Absorbed initiatives contribute within parent projects.

Corporate Inquiry Table — Burns Harbor¶

Purpose: All variables tagged needs-corporate in one table. Send to IE for Cleveland-Cliffs data request.

Production & Throughput¶

#	Variable	Needed For	Question to Ask CLF	Priority
1	production_value_per_hour (BH overall)	BH-01, BH-36, BH-09, BH-53	What is Burns Harbor's production value per hour? (tons/hour × margin/ton across all production units)	Critical — used across 4+ cards
2	margin_per_ton	BH-34, BH-17, BH-09, BH-11, BH-37, BH-25	What is the average product margin per ton at Burns Harbor? (by product mix: flat-rolled vs plate if possible)	Critical — used across 6+ cards
3	margin_per_heat	BH-25, BH-42	What is the average margin per heat at Burns Harbor? (~300 tons × $/ton)	Critical
4	heats_per_year (3 BOFs)	BH-41, BH-42, BH-25	How many heats per year across Burns Harbor's 3 BOFs?	High
5	annual_HSM_tons	BH-37, BH-11, BH-17	What is annual HSM production volume in tons?	High
6	plate_throughput_tons_per_year	BH-20	What is annual plate mill production volume?	Medium
7	coils_per_day	BH-35	How many coils are produced/shipped per day? (derived from 220K+ tons/month ÷ avg coil weight)	High
8	monthly_shipping_tons_breakdown	BH-34	Breakdown of 220K+ tons/month: truck vs rail vs barge, flat-rolled vs plate	Medium

Chemistry & Quality¶

#	Variable	Needed For	Question to Ask CLF	Priority
9	off_chemistry_heats_per_month	BH-41	How many heats per month are off-chemistry? (Dave said 5% overall, 3% carbon+sulfur)	Critical — Dave's #1
10	chemistry_transitions_per_day	BH-15	How many chemistry transitions does each caster make per day?	High
11	off_spec_tons_per_transition	BH-15	How many tons of scrap/downgrade per caster transition on average?	High
12	quality_holds_per_day	BH-35	How many coils does QMS flag per day? What % pass manual review?	High
13	cobbles_per_year (HSM)	BH-11	How many cobble events at the HSM in last 12 months? Average downtime per cobble?	High
14	equipment_damage_cost_per_cobble	BH-11	Average repair cost per cobble? (drive spindle, work rolls, etc.)	High
15	bruise_rejection_tons_per_year	BH-37	Tonnage rejected/downgraded for bruise defects per year?	Medium
16	quality_loss_per_year (total)	BH-09	Total annual quality losses across all process stages? (scrap + downgrade + rework)	Medium

Maintenance & Reliability¶

#	Variable	Needed For	Question to Ask CLF	Priority
17	misattributed_delay_hours_per_month	BH-01	How many delay hours per month have no matching work order? (or are disputed between ops/maint)	High
18	repeat_failure_annual_cost	BH-01	Annual cost of repeat equipment failures? (frequency × avg cost per event)	High
19	Tabware_work_order_volume_per_month	BH-01, BH-02	How many maintenance work orders per month in Tabware at BH?	High
20	belt_failure_events_per_year	BH-53	How many BF conveyor belt failure events per year? Duration and production impact of each?	High
21	unplanned_downtime_per_asset_class	BH-05	Top 5 unplanned downtime events by asset class (BF, BOF, coke, crane, plate mill) — frequency and cost?	High
22	critical_alert_%_in_PdM_reports	BH-45	What % of third-party PdM reports (ITR vibration, Shell oil) contain actionable alerts?	Medium
23	avg_failure_cost_per_missed_alert	BH-45	Historical failures where prior warning existed in unread reports — what did they cost?	Medium

Procurement & Inventory¶

#	Variable	Needed For	Question to Ask CLF	Priority
24	annual_procurement_spend (BH)	BH-40	Annual procurement spend at Burns Harbor?	High
25	transactions_per_day (buyers)	BH-03	Average daily procurement transactions per buyer? (Tabware + Oracle combined)	High
26	obsolete_parts_auto_reorder_cost	BH-04, BH-29	Estimated annual cost of auto-reordering obsolete parts? (parts sitting 20+ years still on auto-order)	Medium
27	parts_on_auto_reorder_count	BH-29	How many of 19K parts are on auto-reorder? What % are obsolete?	Medium
28	stockout_events_per_year	BH-29	Production delays per year attributable to parts unavailability at BH?	Medium

Labor & Cost Rates¶

#	Variable	Needed For	Question to Ask CLF	Priority
29	labor_cost_per_hour (maintenance tech)	BH-01, BH-02, BH-36	Loaded hourly rate for maintenance technician at BH? (wages + benefits + overhead)	Critical — used across 3+ cards
30	labor_cost_per_hour (buyer/admin)	BH-03, BH-40	Loaded hourly rate for purchasing agent?	Medium
31	labor_cost_per_hour (manager)	BH-46, BH-48	Loaded hourly rate for division manager/supervisor?	Medium
32	new_hires_per_year	BH-08	How many new operators/technicians hired per year at BH?	Low

Shipping & Logistics¶

#	Variable	Needed For	Question to Ask CLF	Priority
33	coil_cycle_time_birth_to_ship	BH-34	Average days from coil birth to customer shipment?	Critical — core BH-P01 metric
34	reprocessing_events_per_month	BH-34	How many coils per month are rerouted/reprocessed due to quality or scheduling issues?	High
35	plate_orders_per_month	BH-43	Plate orders per month? Partial rail car frequency?	Medium
36	OTIF_rate_current	BH-43	Current OTIF rate for plate? For hot strip?	Medium
37	IH_hot_metal_delay_cost	BH-44	Average cost per hot metal logistics delay at IH? (production loss + ladle recovery)	High
38	lost_driver_time_per_delivery	BH-26	Estimated time lost per delivery due to navigation confusion at BH plant?	Low

Coke Plant & BF¶

#	Variable	Needed For	Question to Ask CLF	Priority
39	annual_energy_cost (coke plant)	BH-18	Annual energy/gas cost for coke plant operation? (164 ovens × 19hr cycles)	High
40	green_push_events_per_year	BH-18	How many undercooked/green push events per year? Cost per event?	Medium
41	annual_coal_spend	BH-47	Annual coal procurement spend? (8 coal types × volume × price)	Medium
42	coke_quality_variability (CSR/CRI)	BH-47, BH-18	Current coke quality variability? (standard deviation of CSR/CRI)	Medium
43	BF_stove_energy_cost	BH-13	Annual stove energy cost? Incremental opportunity beyond existing thermal model?	Medium
44	annual_burden_cost	BH-14	Annual BF burden cost? (ore + sinter + coke volumes × prices)	Medium
45	sinter_quality_BF_impact	BH-19	What is the relationship between sinter composition and BF productivity?	Low

Compliance & Environmental¶

#	Variable	Needed For	Question to Ask CLF	Priority
46	manual_compliance_hours_per_year	BH-07	Person-hours per year on manual environmental compliance tracking at BH?	Medium
47	compliance_fine_history	BH-07	Historical environmental fines/penalties at BH? Penalty schedule for violations?	Medium
48	carbon_capture_timeline	BH-24	When does the $50M carbon capture system come online? What data streams will it generate?	Low

Infrastructure¶

#	Variable	Needed For	Question to Ask CLF	Priority
49	cloud_bandwidth_current	BH-56	Current bandwidth capacity between BH and cloud? What would adequate bandwidth look like?	Medium
50	wifi_coverage_map	BH-02	Wi-Fi/cell coverage map for BH — specifically coke plant, plate mill, BF area, warehouses?	Medium

Summary: 50 variables needed. 5 Critical (used across many cards), 17 High, 22 Medium, 6 Low priority.