Oil & Gas

Pipeline leaks cause environmental damage, safety hazards, and massive financial loss. Multi-method approach: 1) Computational Pipeline Monitoring (CPM) — Mass Balance: Install flow meters at pipeline input and output. If input ≠ output (beyond tolerance), leak detected. Sensitivity: can detect 1-2% of flow rate. Limitation: slow for small leaks, affected by measurement accuracy. Implementation: SCADA collects flow data every second, computes real-time mass balance, alarm if imbalance exceeds threshold for sustained period. 2) Pressure Point Analysis: Install pressure sensors every 10-20 km. Leak causes: pressure drop at leak point, pressure wave (rarefaction) travels both directions. Negative Pressure Wave (NPW): detected by pressure transducers. Leak location = calculated from wave arrival time at two sensors. Speed: detects within seconds. Accuracy: ±100m location. 3) Real-Time Transient Model (RTTM): Physics-based model running in real-time — models fluid dynamics (pressure, flow, temperature) along entire pipeline. Compares model predictions with actual measurements. Deviation beyond threshold = leak alarm. Most accurate method, but computationally intensive. 4) Fiber Optic Distributed Sensing: Fiber optic cable along pipeline. DTS (Distributed Temperature Sensing): leak causes temperature change — detected as anomaly along fiber. DAS (Distributed Acoustic Sensing): leak creates acoustic signal — fiber detects vibrations. Can detect: leaks, third-party interference (digging near pipeline), pig passage. 5) Architecture: All methods feed into a central leak detection system. Multi-method consensus reduces false alarms (single method: 10+ false alarms/month, combined: 1-2). Alarm management: prioritize by severity, guide operator response. Regulatory: PNGRB (India) mandates leak detection on all major pipelines.

Blend optimization is one of the highest-value applications in refinery operations — optimizing fuel blending can save ₹100+ crore annually for a large refinery. The Problem: Refinery produces multiple component streams (straight-run naphtha, reformate, FCC gasoline, alkylate, etc.). These must be blended to make final products (petrol, diesel) that meet quality specifications (BS-VI: octane > 91, sulfur < 10 ppm, benzene < 1%, vapor pressure within range). The Optimization: Linear Programming (LP) model: Objective: minimize cost (or maximize margin) of blending. Variables: volume of each component in each product blend. Constraints: a) Quality: each blend must meet all specs. b) Volume: must produce target quantity of each product. c) Availability: limited by refinery production of each component. d) Tankage: limited blending tank capacity. Example: Reformate has high octane (100+) but is expensive. FCC gasoline has octane 90 but is cheaper. Alkylate has octane 95 and low sulfur. Optimizer: find cheapest mix that achieves octane > 91 AND sulfur < 10 ppm AND meets all other specs. Blending Properties: Some properties blend linearly by volume (sulfur content), others blend non-linearly (octane — blending octane ≠ pure component octane). Non-linear models use: blending indices, interaction terms. Online Optimization: Blend optimizer runs in real-time. Near-infrared (NIR) analyzers measure quality inline. If quality drifts: optimizer adjusts component ratios in real-time. Give-away reduction: without optimizer, operators add 'safety margin' (blend to octane 92 when spec is 91). That 1 octane over-blend = wasted expensive reformate. Optimizer runs at exactly spec → saves margin. Implementation: LP solver (CPLEX, Gurobi), integrated with DCS for automatic ratio control, connected to lab/NIR for quality feedback.

The digital oilfield concept transforms traditional oil production through IoT sensors, real-time data, and analytics. Traditional vs Digital: Traditional: operators visit each well daily, manually read gauges, adjust chokes. Data collected weekly/monthly in spreadsheets. Problems detected reactively (after failure). Digital: continuous automated monitoring, remote control, predictive analytics. Architecture: 1) Edge Layer: Wellhead sensors: pressure (wellhead, casing, tubing), temperature, flow rate (multiphase flow meter). Artificial lift sensors: ESP (downhole pressure, temperature, vibration, current). Surface equipment: separator levels, compressor status, storage tank levels. Data: 100+ parameters per well, sampled every 1-10 seconds. Edge computing: local processing (ESP optimization, safety shutdown) even without connectivity. 2) Communication: Onshore: cellular 4G/LTE, VSAT satellite for remote areas. Offshore: VSAT, microwave radio, subsea fiber. Data volume: 1 TB/day for a field with 500 wells. Latency: < 5 seconds for operational data, < 100ms for safety-critical. 3) Central Platform: Data lake: stores all sensor data + production data + maintenance records. Real-time dashboards: production engineers monitor 100+ wells per person (vs 10-20 in traditional). Alerts: automated anomaly detection — well not producing, ESP vibration increasing. 4) Analytics Use Cases: a) Production optimization: well models (IPR/VLP) calibrated with real-time data, recommend optimal choke size and artificial lift settings. 10-15% production uplift typical. b) ESP predictive maintenance: ML model trained on vibration, current, temperature patterns before past failures. Predicts ESP failure 2-4 weeks in advance. Schedule workover vs. wait for catastrophic failure ($500K repair vs $200K preventive). c) Water cut prediction: detect increasing water production early — optimize waterflood, plan intervention. d) Reservoir surveillance: pressure data from all wells → update reservoir model in real-time → better development decisions. ONGC has deployed digital oilfield at Mumbai High (offshore) and Ahmedabad field — monitoring 5,000+ wells.

India moved to daily dynamic fuel pricing in June 2017. How it works: 1) Pricing Formula: Retail price = (International product price + freight) × exchange rate + excise duty + dealer commission + VAT (state). International price: Indian Basket crude price is reference, but actual pricing based on Singapore/Arab Gulf benchmark for refined products (petrol = Singapore gasoline benchmark, diesel = Singapore gasoil). Exchange rate: INR/USD on the day. Taxes: Excise duty (central, ₹20-25/liter) + VAT (state, 15-30% — varies by state, hence different prices in different states). 2) Daily Revision Technology: Every day at 6 AM, new prices effective. System flow: a) International price feed: automated from Platts/Reuters price reporting services — Singapore benchmark prices captured at market close. b) Exchange rate: RBI reference rate. c) Formula engine: computes new retail price per product per state (each state has different VAT). d) Distribution: new prices transmitted to 85,000+ fuel stations by 5:30 AM. e) Dispenser update: modern automated dispensers receive price update electronically. Legacy stations: manual update by station manager. 3) Consumer Communication: Prices published on OMC websites and apps at 6 AM. SMS service for daily price alerts. Media publishes daily. Price difference between cities: ₹5-15/liter due to VAT differences and transportation cost. 4) Technology Challenges: Price consistency: ensure all 85,000+ stations updated simultaneously. Audit: every transaction stored with applicable price — must match the official price for that timestamp and location. Subsidy (LPG): DBT system transfers ₹200-300 subsidy per cylinder to 30 crore+ bank accounts within 48 hours of delivery. 5) Impact: Daily pricing reduces OMC working capital requirement (no accumulation of under/over-recovery). But politically sensitive — government sometimes holds prices before elections.

Hydrocarbon accounting tracks every molecule from wellhead to consumer — it's the 'financial accounting' of oil & gas operations. Why it matters: 1) Revenue: crude oil worth $80/barrel, a 1% measurement error on 100,000 bpd production = $29M/year. Custody transfer (changing ownership) must be accurate. 2) Losses: transit loss (evaporation, leaks), processing loss (refinery yield), distribution loss. Industry norm: 0.2-0.5% acceptable. Beyond that = theft or measurement error. 3) Joint Ventures: Many fields are JV (ONGC 60% + private 40%). Each partner's share calculated from production allocation. Must be auditable and fair. Architecture: a) Measurement: Fiscal metering: high-accuracy flow meters (Coriolis ±0.1%, ultrasonic) at custody transfer points. Crude: measured in barrels (volume) but transacted in metric tons (mass) — temperature correction critical (crude expands with heat). Gas: measured in standard cubic meters, corrected for pressure and temperature. Quality: API gravity, sulfur content, water content (BS&W) — affects price. b) Allocation: Production allocation: field produces commingled (all wells into one pipeline) — must allocate back to individual wells/leases. Well test data (periodic flow test of each well) used for allocation ratios. Process allocation: refinery converts 1 barrel crude into multiple products — yield accounting tracks conversion. c) Reconciliation: Material balance: input = output + inventory change + losses. Close: within 0.5% acceptable. Any larger discrepancy triggers investigation. Monthly close: all measurements verified, tank gauging (dip measurements + temperature + density), pipeline inventory (line fill). d) Systems: SAP IS-Oil Hydrocarbon Management, custom systems with LIMS integration. Connected to: DCS (process data), SCADA (pipeline data), tank gauging (inventory), fiscal meters (custody transfer), lab (quality). Regulatory: DGH (Directorate General of Hydrocarbons) requires production reporting. PSC (Production Sharing Contract) terms define cost recovery and profit petroleum split with government.