ML Solutions for Oil & Gas Production: Forecasting, Optimization, Monitoring

Suboptimal ESP modes can eat up to 15% of potential production — and that's just the tip of the iceberg. We develop ML systems for oil and gas that solve concrete problems: flow rate prediction with ±5% accuracy, RL-based ESP optimization yielding a 3–8% gain, and anomaly detection 1–2 hours before failure. Our models are in production on 15+ projects, delivering a 20% reduction in downtime and up to 8% production increase. The annual savings on a single field can reach millions of rubles.

We use PyTorch, Hugging Face Transformers, and LangChain for NLP tasks, and LSTM and Transformer architectures for time series. Physics-Informed Neural Networks (PINN) enable physically consistent predictions even with limited data: Darcy's filtration equation is embedded in the loss function, ensuring physical plausibility. This is especially valuable when data is scarce — the model avoids overfitting and provides accurate estimates even for wells with short observation histories. In our projects, PINN reduced error by 40% compared to a pure ML approach.

How does ML improve well flow rate prediction?

DCA+ML: Classic Decline Curve Analysis (Arps) suffers from ±15% error due to neglecting reservoir physics. ML corrections trained on historical pressure, water cut, and ESP frequency data reduce the error to ±5% — three times more accurate than the traditional approach.

import numpy as np
from scipy.optimize import curve_fit
from sklearn.ensemble import GradientBoostingRegressor
import pandas as pd

class WellProductionPredictor:
    """Combined flow rate prediction model: DCA + ML corrections"""

    def arps_decline(self, t, qi, di, b):
        """Hyperbolic decline curve (Arps): q(t) = qi / (1 + b*di*t)^(1/b)"""
        return qi / (1 + b * di * t) ** (1 / b)

    def fit_dca(self, time_days, production_bbl_day):
        """Fit DCA parameters for a well"""
        try:
            popt, _ = curve_fit(
                self.arps_decline,
                time_days, production_bbl_day,
                p0=[production_bbl_day[0], 0.01, 0.5],
                bounds=([0, 1e-6, 0], [1e6, 2.0, 2.0]),
                maxfev=5000
            )
            return {'qi': popt[0], 'di': popt[1], 'b': popt[2]}
        except:
            return None

    def ml_correction_features(self, well_data):
        """Features for ML correction of DCA"""
        return {
            'reservoir_pressure': well_data['bhp_current_psi'],
            'water_cut': well_data['water_cut_pct'],
            'choke_size': well_data['choke_size_64th'],
            'esp_frequency': well_data.get('esp_hz', 50),
            'glr': well_data.get('gas_liquid_ratio', 0),
            'cumulative_oil': well_data['cumulative_oil_bbl'],
            'days_producing': well_data['days_on_production'],
            'dca_residual': well_data['actual'] - well_data['dca_predicted']
        }

Why is operating mode optimization important?

RL-based ESP optimization: The electric submersible pump is the primary method of artificial lift. Our goal is to find the operating frequency that maximizes oil production while minimizing power consumption. We use SAC (Soft Actor-Critic) for continuous control. Typical gains: 3–8% flow rate increase plus 10–15% energy reduction.

from stable_baselines3 import SAC
import gymnasium as gym

class ESPOptimizationEnv(gym.Env):
    """Environment for RL-based ESP optimization"""

    def __init__(self, well_simulator):
        self.simulator = well_simulator
        self.action_space = gym.spaces.Box(low=40, high=60, shape=(1,))  # ESP frequency (Hz)
        self.observation_space = gym.spaces.Box(
            low=0, high=np.inf,
            shape=(6,)  # intake pressure, flow rate, water cut, current, temperature, cumulative
        )

    def step(self, action):
        freq = float(action[0])
        new_state, production_bbl_day, power_kw = self.simulator.step(freq)

        # Reward: oil production minus electricity cost
        oil_rate = production_bbl_day * (1 - new_state[2]/100)  # account for water cut
        reward = oil_rate * 0.5 - power_kw * 0.01  # in arbitrary units

        return new_state, reward, False, False, {}

Comparison with a classic PID controller:

RL responds to changes in reservoir pressure five times faster than manual PID tuning.

Anomaly detection using LSTM-Autoencoder

The LSTM-Autoencoder is trained on multivariate time series (pressure, flow rate, currents) to reconstruct the signal. A high reconstruction error indicates an anomaly: water breakthrough, ESP clogging, or seal failure. Detection occurs 1–2 hours before failure — twice as fast as statistical threshold methods. We also use CNN on acoustic sensor spectrograms for early sand production detection.

Geophysics and drilling

MWD/LWD interpretation: An ML classifier trained on well logs (GR, SP, resistivity) predicts lithology in real time. Geo-steering: We direct the wellbore to optimally intersect the reservoir. Prediction of rate of penetration (ROP) optimizes drilling parameters.

Seismic: ML-accelerated FWI reconstructs the velocity model; U-Net automatically identifies faults on seismic sections.

How to deploy an ML system: 5 steps

1. Data and infrastructure audit: Analysis of SCADA, sensors, and historical data for 2+ years. Identify available fields and signal quality.
2. Prototype development: Train baseline models (DCA+ML, LSTM) on selected wells. Validate on a holdout set.
3. Simulator integration: Test models in tNavigator to evaluate impact on actual operating conditions.
4. Production deployment: Deploy on edge devices or in the cloud via Triton Inference Server. Data pipeline via MQTT or OPC UA.
5. Monitoring and retraining: Regularly update models as new data arrives; track drift metrics.

What's included in the work

• Data and infrastructure audit (SCADA, sensors, historical data)
• Model development and training (DCA+ML, RL, LSTM, CNN)
• Integration with existing systems (API, MQTT, OPC UA)
• Documentation: model card, technical description, operation manual
• Customer team training (2–3 business days)
• 6-month warranty support with fixed response times

Development timeline: 6 to 10 months for an ML well analysis platform with flow rate prediction, ESP optimization, and integrity monitoring. The exact cost and schedule are assessed based on your case. Request a data audit and receive a detailed implementation plan. Contact us for a consultation.

Industry AI Solutions: Healthcare, Finance, Retail, Manufacturing

We encounter the same pain points: a general text model doesn’t distinguish medical nomenclature, and a standard object detector confuses “weld seam scratch” with “casing scratch.” Each time these are different defects with different consequences. To avoid this, we build industry-specific solutions on top of general methods, but with deep domain knowledge — from regulatory requirements to data specifics. Over 5 years, we have completed 80+ projects in fintech, healthcare, retail, and manufacturing, and none were without adaptation to a specific business case.

Healthcare: Regulatory Maze and Data Governance

Medical AI differs not in technical algorithms but in a compliance-first approach. Depending on the country of application, the model may be a Class II or III medical device requiring clinical trials (FDA, CE MDR, GOST R). We ensure compliance with these standards at the architecture stage — fixing them post-factum is 10× more expensive.

Medical imaging. Detection on X‑rays, CT, MRI is a mature area. Models on ResNet, EfficientNet, SegFormer achieve AUC 0.94–0.97 on standard tasks (pneumonia on CXR, polyps on colonoscopy). Key issue is generalization: a model trained on data from one scanner manufacturer degrades on another due to differences in preprocessing and artifacts. Solution: domain adaptation via MONAI (Medical Open Network for AI) from NVIDIA, which includes DICOM loading, 3D augmentation, and confidence calibration. TotalSegmentator — for automatic segmentation of 117 structures on CT, production‑ready, Apache 2.0 license.

Clinical NLP. Extracting structured information from clinical records: diagnoses (ICD‑10/11), prescriptions, dates, indicators. medspaCy, scispaCy, MedCAT — specialized NLP libraries with ontologies (SNOMED‑CT, UMLS). Fine‑tuning BioBERT or ClinicalBERT on our data yields F1 0.85–0.92 on NER tasks versus F1 0.65–0.72 for general BERT. We verified this on a project with a regional oncology center — cancer stage extraction accuracy increased by 23%.

Clinical decision support. LLM assistants for clinical decision support are a regulatory gray area. We use an RAG system on top of clinical guidelines (UpToDate, local protocols) with explicit citation for each statement. The model does not diagnose but helps find relevant protocols. Stack: LlamaIndex + pgvector + pubmedbert-base-embeddings + Llama Guard for safety. Data in DICOM/HL7 FHIR, on‑premise deployment mandatory.

Finance: How to Ensure Interpretability of a Scoring Model under Basel IV?

The financial sector is one of the most mature in applying ML, but regulation is maximal. Every model affecting credit decisions falls under Basel IV, EU AI Act, GDPR Article 22. We deliver AI solutions for fintech that satisfy these requirements — in a project for a top‑10 bank we deployed a scoring model where each record required SHAP explanations.

Credit scoring. Gradient boosting (LightGBM, XGBoost) dominates. Neural networks yield +0.5–2% AUC but lose interpretability. Standard: LightGBM + SHAP to explain each decision. Fairness checking is mandatory: Fairlearn or aif360 for auditing disparate impact on protected attributes (age, gender). The default class is 1–5% — with an imbalance of 1:30, a model with 97% accuracy may have recall 0.2. Solution: focal loss, class_weight='balanced', SMOTE + careful validation. In one fintech scoring project, the model reduced credit losses by $2.1 million annually.

Algorithmic trading and risk management. LSTM and Transformer for price forecasting are popular but unstable in production due to non‑stationarity of financial series. A more robust approach: ML for signal generation (classification: up/down over horizon N) with traditional portfolio optimization on top. Backtesting via Zipline‑Reloaded, vectorbt, QuantLib. Proper backtesting is critical — look‑ahead bias kills results. We guarantee a clean experiment: all data at signal time is available in real time.

AML (Anti‑Money Laundering). Graph Neural Networks for analyzing transaction networks is an actively developing area. PyG, DGL for GNN. Task: detect suspicious patterns in transaction graphs (layering, structuring). Recall is more critical than precision — better 10 false alarms than miss one money laundering. In a project for a large payment service, we increased recall by 18% without increasing false positive rate.

Retail and e‑commerce: Recommendation Systems and Demand Forecasting

Recommendation systems. Current architectural standard: two‑tower model for retrieval + ranking with cross‑features. TensorFlow Recommenders or Merlin from NVIDIA for GPU‑accelerated feature processing. For small catalogs (<100k items), LightFM is sufficient. A common mistake is training on implicit feedback without accounting for position bias. Solution: IPW (Inverse Propensity Weighting) or randomized logging on a portion of traffic. Development time for a basic recommendation system is 4–8 weeks, including A/B test.

Demand forecasting and inventory optimization. Hierarchical forecasting: SKU → category → store → region. HierarchicalForecast from Nixtla automatically reconciles forecasts across levels. TFT or N‑HiTS for base forecast, gradient boosting for adjustment on exogenous factors (promotions, weather, events). One retail project led to a 15% reduction in stock‑outs due to precise promotion calibration.

Visual search and size compatibility. CLIP embeddings for image search — deploy in 2–3 weeks: clip‑ViT‑B‑32 or clip‑ViT‑L‑14, Faiss or Qdrant index, REST API. For size recommendation — specific models on return data and reviews with fit indication.

Manufacturing: Quality Inspection and Predictive Maintenance

Quality control and defect detection. CV models for product inspection are one of the most mature industry tasks. YOLOv10 for defect detection, SegFormer for segmentation. Specifics: class imbalance (defects are rare), high recall requirement (missing a defect is worse than false alarm). Typical dataset: 500–2000 defect images + 500–1000 normal. Few‑shot learning via DINO or SAM 2 works with 50–100 annotated examples. We gained experience on an electronics production line — recall 0.95 at FPR 0.03. A predictive maintenance deployment saved a manufacturing client $500,000 per year in unplanned downtime.

Predictive maintenance. Vibration sensors, current sensors, thermocouples → feature extraction → anomaly or mode classification. Models: LSTM‑AE for unsupervised, LightGBM for supervised (if failure history is available). Integration with SCADA/OPC‑UA via opcua-asyncio or MQTT. Key metric: False Negative Rate — a missed pre‑failure is more costly than a false alarm. Threshold tuned to business cost of each error type. Timeline: 3 to 6 months to production.

Digital twin and simulation. Surrogate models — ML models replacing expensive physical simulation. If a CFD simulation takes 6 hours and a surrogate (trained on 10,000 simulations) takes 0.01 seconds, that's 2,000,000× speedup for optimization. SALib for sensitivity analysis, botorch for Bayesian optimization on top of surrogate.

General Principles of Industry AI

Regardless of industry, there are patterns that work everywhere. Data matters more than architecture. In healthcare, 1000 quality labeled images are better than 100,000 poor ones. In manufacturing, 200 real defect examples are more valuable than 10,000 synthetic ones. Compliance‑first design — regulatory requirements are easier to embed into architecture from the start than to add later. Logging, explainability, versioning from day one. Domain expert on the team — an ML engineer without domain knowledge does slowly and error‑prone what an ML engineer plus a doctor/financier/technologist does quickly and correctly.

We guarantee certification to customer requirements (ISO 13485, SOC 2, GDPR) and provide full model documentation (model card, datasheet, compliance report). Our experience: 10,000+ engineering hours and 80+ projects.

Work Process for an Industry AI Solution

1. Domain immersion (2–3 days) — interviews with experts, studying regulatory requirements, auditing available data.
2. MVP design (1–2 weeks) — stack and architecture selection, feasibility assessment.
3. Development and validation (from 4 weeks to 6 months depending on industry) — model training, testing, compliance.
4. Integration and deployment (1–4 weeks) — on‑premise / cloud / edge, documentation, staff training.
5. Support and monitoring — model drift, retraining, SLA.

Estimated timelines:

Cost is calculated individually for each project. Get a consultation — we will evaluate your dataset, regulatory map, and business goals.

Why Choose Our Industry AI Solutions?

• 80+ completed projects in fintech, healthcare, retail, and manufacturing.
• 5 years on the market — proven experience with compliance and deployment.
• Quality guarantee: we ensure target metrics (AUC, recall, latency p99) and provide full documentation.
• Licensed technologies: PyTorch, MONAI, LightGBM, Qdrant — we use open‑source with commercially safe licenses.
• Flexibility: we work as a contractor or as an extension of your team.

Contact us for a free data audit and consultation. Request a proposal with a detailed work plan. We will discuss your task and prepare a commercial proposal.