# Refiner Energy Optimizer

*/Opportunities/Refiner_Energy_Optimizer*

## Opportunity Overview

**Wedge**: The beachhead is the Crude Distillation Unit heater network, which is the single largest energy consumer in any refinery. Winning the Crude Distillation Unit provides immediate, highly measurable fuel savings to prove the system's mathematical validity. Expansion proceeds unit-by-unit to Fluid Catalytic Crackers and Reformers, ultimately encompassing the entire site-wide steam and utility grid.
**Timing**: Physics-informed neural networks now process thousands of noisy time-series tags from plant historians without hallucinating unphysical states. Additionally, secure edge-compute hardware allows heavy model inference to run safely inside the plant's operational technology network boundaries without cloud latency.
**Why This I C P**: Refineries operate on extremely tight margins where energy consumption constitutes up to sixty percent of non-feedstock operating expenses. This makes plant managers highly sensitive to minor efficiency gains, driving faster procurement cycles than in discrete manufacturing sectors.
**Size Of Prize**: There are roughly 2,700 large-scale oil refineries and petrochemical complexes globally that spend tens of millions annually on fuel gas and power. Capturing a software value of $250,000 per site annually for energy optimization yields a total addressable prize of approximately $675M.
**Gap Narrative**: Refinery process engineers manage energy-intensive furnaces and boilers using static, physics-based simulations that lag behind real-time feed variations. Operators manually adjust setpoints, leaving a persistent efficiency gap where excess fuel is burned to maintain safe operational margins. The ICP requires dynamic, multi-variable optimization that adjusts control limits continuously without compromising unit yields.
**Defensibility**: The system compounds value through localized data accumulation and model tuning. As the software ingests months of specific sensor data, it learns the unique thermodynamic quirks and fouling degradation curves of individual furnaces. Ripping out the software forces the refinery to revert to generic models, imposing an immediate financial penalty in lost efficiency that acts as a severe switching cost.
**Why This Thesis**: An advisory software layer directly addresses the multi-variable complexity that human operators cannot calculate in real time. Deploying this as a software overlay on existing Distributed Control Systems extracts immediate value from sunk sensor and hardware investments without requiring physical retrofits.

## Opportunity Linked Thesis

**Thesis**: [Software](/Theses/Software)

## Opportunity Linked I C P

**Icp**: [Petrochemical Refinery](/CompanyTypes/Petrochemical_Refinery)

## Opportunity Market Sizing

_Illustrative — target and order-of-magnitude estimate figures, not an achieved track record (this Thing is concept-stage)._

**S A M**: ~$200-300M North American and European modernized refineries
**S O M**: ~$15-30M
**T A M**: ~1,000 global petrochemical refineries × ~$600k/yr ≈ ~$600M
**Growth Rate**: ~12-18%/yr, driven by tightening emissions regulations and volatile utility feedstock costs forcing active margin management
**Paid Comparable Spend**: ~$300k-600k/yr per facility spent on outsourced process engineering consultants and static thermodynamic modeling software

## Opportunity Incumbents

- [KBC Visual MESA](/Products/KBC_Visual_MESA) — Tool
- [Aspen Energy Analyzer](/Products/Aspen_Energy_Analyzer) — Tool
- [Honeywell Forge](/Products/Honeywell_Forge) — Tool
- [AVEVA Process Optimization](/Products/AVEVA_Process_Optimization) — Tool
- [Custom Excel Models](/Products/Custom_Excel_Models) — Spreadsheet
- [Engineering Consulting Audits](/Products/Engineering_Consulting_Audits) — Service

## Opportunity Win Conditions

**Kill Thresholds**:
- Implementation time exceeds 60 days for the initial plant unit
- Recommendation acceptance rate falls below 20 percent after 30 days of live data
- Demonstrated energy cost savings remain under $10k per week during the pilot
- Customer refuses to transition from paid pilot to annual contract after 90 days
**Leading Metrics**:
- Time-to-first-baseline-model in days
- Recommendation acceptance rate percentage
- Historian connection uptime percentage
- Weekly active process engineers per facility
- Identified utility savings in dollars per week
**What Proves Right**: Process engineers connect live plant historian data and accept the system's dynamic energy routing recommendations over 40 percent of the time during the pilot. Plant managers upgrade from single-unit trials to full-facility deployments within 90 days based on proven utility cost reductions. The product successfully displaces static spreadsheet models and commands a base pricing tier of $150k per year.
**What Proves Wrong**: Operators refuse to trust the optimization recommendations, manually overriding or ignoring more than 80 percent of the system alerts. Integration with legacy on-premise historian databases takes longer than 60 days to stabilize. The actual utility savings generated fall below the 2 percent efficiency threshold required to justify the software cost.

## Opportunity Build Profile

**Hardest Part**: Building a physics-informed model that strictly obeys thermodynamic constraints and safety envelopes, ensuring the system never recommends a setpoint that risks an emergency shutdown.
**Min Viable Scope**: Deliver open-loop setpoint recommendations to human operators for a single high-impact unit like a crude distillation heater. Deliberately exclude closed-loop automated control, predictive maintenance alerts, and plant-wide optimization.
**Cold Start Problem**: The system lacks equipment-specific degradation models and sensor baselines until deployed at a specific plant. Break this by requiring a design partner to provide 36 months of historical OSIsoft PI data to train the initial local models before pilot launch.
**Time To First Value**: 3-4 weeks of historical data ingestion and baseline model training
**Data Moat Available**: true
**Technical Difficulty**: High

## Neighborhood

### Where the gap lives

- [Predict Freeness Variability](/Tasks/Predict_Freeness_Variability) — latent gap · Tasks

### Incumbent in

- [KBC Visual MESA](/Products/KBC_Visual_MESA) — incumbent in · Products
- [Engineering Consulting Audits](/Products/Engineering_Consulting_Audits) — incumbent in · Products
- [Honeywell Forge](/Products/Honeywell_Forge) — incumbent in · Products
- [AVEVA Process Optimization](/Products/AVEVA_Process_Optimization) — incumbent in · Products
- [Aspen Energy Analyzer](/Products/Aspen_Energy_Analyzer) — incumbent in · Products
- [Custom Excel Models](/Products/Custom_Excel_Models) — incumbent in · Products

### Applies thesis

- [Petrochemical Refinery](/CompanyTypes/Petrochemical_Refinery) — applies thesis · CompanyTypes

### Embodies

- [Software](/Theses/Software) — embodies · Theses

### Similar Opportunities

- [Algorithmic Refiner Energy Optimization](/Opportunities/Algorithmic_Refiner_Energy_Optimization) — similar · Opportunities
- [Feedstock Blending Optimizer](/Opportunities/Feedstock_Blending_Optimizer) — similar · Opportunities
- [Distillation Yield Engine](/Opportunities/Distillation_Yield_Engine) — similar · Opportunities
- [Refinery Dynamic Deadband Adjustment](/Opportunities/Refinery_Dynamic_Deadband_Adjustment) — similar · Opportunities
- [Thermal Energy Telemetry for Mills](/Opportunities/Thermal_Energy_Telemetry_for_Mills) — similar · Opportunities
- [Combustion Optimization API](/Opportunities/Combustion_Optimization_API) — similar · Opportunities
- [Cracking Unit Diagnostics](/Opportunities/Cracking_Unit_Diagnostics) — similar · Opportunities
- [Refiner Tuning Agent](/Opportunities/Refiner_Tuning_Agent) — similar · Opportunities
- [Furnace Stoichiometry Automation](/Opportunities/Furnace_Stoichiometry_Automation) — similar · Opportunities
- [Autonomous Loop Tuning](/Opportunities/Autonomous_Loop_Tuning) — similar · Opportunities
- [Acoustic Burner Diagnostics](/Opportunities/Acoustic_Burner_Diagnostics) — similar · Opportunities
- [Steam Load Balancing](/Opportunities/Steam_Load_Balancing) — similar · Opportunities
- [Real-Time Kiln Optimization](/Opportunities/Real-Time_Kiln_Optimization) — similar · Opportunities
- [Predictive Kiln Thermal Optimization](/Opportunities/Predictive_Kiln_Thermal_Optimization) — similar · Opportunities
- [AI Yield Optimization for Oleochemical Plants](/Opportunities/AI_Yield_Optimization_for_Oleochemical_Plants) — similar · Opportunities
- [Combustion Optimization for Glass Producers](/Opportunities/Combustion_Optimization_for_Glass_Producers) — similar · Opportunities
- [Refiner Tuning Agent](/CompanyTypes/BCTMP_Mills/Opportunities/Refiner_Tuning_Agent) — similar · Opportunities
- [Kiln Energy Controller](/CompanyTypes/Specialty_and_Oil_Well_Cement_Producers/Opportunities/Kiln_Energy_Controller) — similar · Opportunities
- [Kiln Energy Controller](/CompanyTypes/Vertically_Integrated_Wallboard_Giants/Opportunities/Kiln_Energy_Controller) — similar · Opportunities
- [Real-Time Freeness Prediction](/Opportunities/Real-Time_Freeness_Prediction) — similar · Opportunities