Back to Insights

Production Operations Software: Surveillance, Optimization, and the AI Opportunity

Dr. Mehrdad Shirangi | | Published by Groundwork Analytics LLC
Editorial disclosure

This article reflects the independent analysis and professional opinion of the author, informed by published research, vendor documentation, and practitioner experience. No vendor reviewed or influenced this content prior to publication.

A production engineer managing 200 wells across a basin spends the first two hours of every morning doing the same thing: pulling up SCADA data, checking which wells went down overnight, reviewing production rates against expectations, identifying anomalies, and prioritizing the day's field activities. This ritual has not fundamentally changed in 20 years, despite billions of dollars invested in production operations software.

The tools have gotten better. SCADA systems are more reliable. Dashboards are prettier. Some operators have deployed machine learning models for anomaly detection. But the core workflow -- a human engineer sifting through data, making judgments, and dispatching field crews -- remains largely intact. In an industry that has adopted horizontal drilling, hydraulic fracturing, and pad drilling as transformative technologies, the software that manages producing wells is remarkably conservative.

This article surveys the production operations software landscape: what is available, what it actually does, and where AI and machine learning can deliver genuine improvements over the status quo.

The Production Software Stack

Production operations software is not a single category. It is a stack of interconnected systems, each handling a different layer of the production management workflow:

  1. SCADA / Field Data Acquisition -- Collecting real-time data from wellheads, separators, tank batteries, and artificial lift equipment
  2. Production Surveillance -- Visualizing production data, detecting anomalies, comparing actual versus expected performance
  3. Artificial Lift Optimization -- Optimizing the performance of rod pumps, ESPs, gas lift, plunger lift, and other lift methods
  4. Production Accounting / Allocation -- Allocating commingled production to individual wells, reconciling meter readings, generating regulatory reports
  5. Production Forecasting -- Decline curve analysis, rate-transient analysis, reserves estimation
  6. Field Optimization -- Gas lift gas allocation, water handling, compression optimization, facility debottlenecking

Most operators use software from multiple vendors across this stack, and the integration between layers ranges from tight (within a single vendor's platform) to nonexistent (between competing vendors).

SCADA and Field Data Acquisition

SCADA (Supervisory Control and Data Acquisition) systems form the data foundation for production operations. In upstream oil and gas, SCADA systems collect data from remote terminal units (RTUs) installed at wellheads and facilities, transmit it via radio, cellular, or satellite to a central host, and display it for operators and engineers.

The major SCADA platforms used in upstream production include:

Emerson (OpenEnterprise / ROC)

Emerson is one of the largest SCADA providers in the upstream space, with a strong installed base of ROC (Remote Operations Controller) RTUs and the OpenEnterprise SCADA host software. Emerson's strength lies in the breadth of field instrumentation -- pressure transmitters, flow meters, level sensors, and valve controllers -- all of which integrate natively with their SCADA infrastructure.

Emerson has been investing in its Plantweb digital ecosystem, which extends beyond traditional SCADA into predictive analytics and remote monitoring. Their acquisition of companies like Zedi strengthened their cloud-based production monitoring capabilities.

ABB (ABB Ability)

ABB provides SCADA and automation systems for upstream production, particularly in larger, more complex production facilities and offshore environments. ABB's Ability platform extends traditional SCADA with cloud-based analytics, predictive maintenance, and digital twin capabilities.

ABB tends to be more prevalent in international operations and offshore environments than in U.S. land production, where Emerson has a stronger presence.

WellAware

WellAware takes a different approach from traditional SCADA vendors. Rather than providing a full SCADA system, WellAware focuses on wellsite data acquisition and cloud-based monitoring, targeting the gap between traditional SCADA (which requires significant infrastructure) and manual data collection (which is slow and incomplete).

WellAware's hardware (wellsite sensors and communication devices) is designed for rapid deployment, and their cloud platform provides monitoring and alerting without the complexity of a full SCADA implementation. This approach appeals to mid-size operators who want real-time field data but find traditional SCADA systems too expensive or complex for their scale.

Strengths: Low-cost, rapid-deployment field monitoring, cloud-native platform, simpler than traditional SCADA for basic well surveillance.

Limitations: Less functionality than full SCADA systems, limited control capabilities (monitoring focused), smaller installed base compared to Emerson or ABB.

Production Surveillance Platforms

Production surveillance sits between SCADA (which provides raw data) and optimization (which provides recommendations). Surveillance platforms aggregate data from SCADA and other sources, apply analytics to detect anomalies and performance changes, and present the results to production engineers in actionable dashboards.

OspreyData

OspreyData provides cloud-based production surveillance with a focus on AI-driven anomaly detection. Their platform ingests SCADA data and applies machine learning models to identify production issues -- wells that have gone down, artificial lift systems that are underperforming, and unexpected changes in fluid rates or pressures.

OspreyData's differentiator is the combination of automated anomaly detection with operational workflows. The platform does not just flag anomalies; it creates work orders, routes them to the appropriate field personnel, and tracks resolution. This workflow integration is often more valuable than the anomaly detection itself, because the bottleneck in production management is frequently not detecting problems but ensuring they are addressed in priority order.

Strengths: AI-driven anomaly detection, integrated operational workflows, cloud-native architecture, works across artificial lift types.

Limitations: Dependent on the quality and completeness of incoming SCADA data, ML models need training and tuning for each operator's wells and operating conditions.

SLB Production Operations (ForeSite)

SLB's ForeSite platform covers production surveillance, artificial lift optimization, and well performance analysis in an integrated environment. ForeSite is part of SLB's broader digital portfolio and benefits from integration with SLB's artificial lift equipment (particularly ESPs), downhole sensors, and reservoir modeling tools.

ForeSite includes automated well test validation, virtual metering capabilities, and artificial lift performance diagnostics. For operators using SLB artificial lift systems, the integration between ForeSite and SLB's downhole equipment provides detailed diagnostic capabilities that third-party platforms cannot match.

Strengths: Deep integration with SLB artificial lift equipment, comprehensive surveillance and optimization features, global support infrastructure.

Limitations: Most valuable when paired with SLB equipment and services, significant implementation effort for large deployments, premium pricing.

Artificial Lift Optimization

Artificial lift is the single largest category of production operations software, because the majority of producing wells in North America require some form of artificial lift -- rod pumps, ESPs (electric submersible pumps), gas lift, plunger lift, or progressive cavity pumps.

Ambyint

Ambyint focuses on AI-driven artificial lift optimization, particularly for rod pump and ESP systems. Their platform uses machine learning to continuously optimize artificial lift settings -- pump speed, stroke length, cycle timers for rod pumps; frequency, intake pressure targets for ESPs.

Ambyint's approach is notable for its emphasis on autonomous optimization. Rather than providing recommendations that an engineer must review and implement, Ambyint's system can directly adjust artificial lift parameters through integration with the well's controller, subject to configurable constraints and guard rails.

The company claims measurable production uplift (typically 3-8%) and reduced failure rates through optimized operating conditions. These claims are supported by several published case studies, though results vary significantly depending on the existing optimization baseline and the specific well conditions.

Strengths: Autonomous optimization with direct controller integration, focus on measurable production uplift, works across major rod pump and ESP manufacturers, continuous learning from well data.

Limitations: Requires reliable SCADA connectivity and controller integration, value proposition depends on how well-optimized the wells are before deployment, autonomous control raises concerns for some operators regarding well integrity and safety.

SLB Lift IQ

Lift IQ is SLB's artificial lift optimization platform, primarily focused on ESP monitoring and optimization. Given SLB's position as the world's largest ESP manufacturer, Lift IQ has deep integration with SLB ESP systems, providing detailed diagnostic capabilities based on motor current, intake pressure, discharge pressure, vibration, and temperature data from SLB's downhole sensors.

Lift IQ's ESP analytics include health monitoring (detecting bearing wear, gas interference, scale buildup), performance optimization (adjusting frequency to optimize production rate and efficiency), and remaining useful life estimation. For gas lift optimization, SLB offers integrated solutions that combine surface injection control with downhole diagnostics.

Strengths: Unmatched ESP diagnostic depth for SLB equipment, integration with SLB's global artificial lift engineering team, comprehensive health monitoring and predictive maintenance.

Limitations: Most valuable for SLB ESPs specifically -- operators using ESPs from other manufacturers (Baker Hughes, Borets, Novomet) will not get the same level of diagnostic depth. The platform is part of SLB's broader digital ecosystem, which can be complex to adopt for operators seeking a point solution.

Weatherford (Production Optimization)

Weatherford provides artificial lift optimization software across multiple lift types, with particular strength in rod pump systems (reflecting Weatherford's large rod pump manufacturing and service business) and managed pressure operations.

Weatherford's ForeSite (not to be confused with SLB's ForeSite -- the naming overlap is an ongoing source of confusion) covers rod pump optimization, ESP monitoring, plunger lift optimization, and gas lift management. Weatherford has invested in predictive analytics for artificial lift failure prediction and performance optimization.

Strengths: Broad artificial lift coverage across multiple lift types, strong rod pump optimization capabilities, integrated with Weatherford's artificial lift services.

Limitations: Weatherford's financial restructuring in recent years created uncertainty about the company's long-term software investment. The platform works best with Weatherford equipment.

Where AI Can Help: The Production Operations Opportunity

Production operations is one of the most promising domains for AI in upstream oil and gas, for a simple reason: the data is relatively available (through SCADA), the decisions are frequent and repetitive, and the value of incremental optimization is large when multiplied across hundreds or thousands of wells.

Opportunity 1: Automated Anomaly Detection and Triage

Production engineers currently spend a disproportionate amount of time on surveillance -- scanning dashboards for problems. AI-based anomaly detection can identify production issues faster and more consistently than manual review, but the real value is in triage: not just detecting that something is wrong, but assessing the likely cause, estimating the production impact, and recommending the appropriate response.

Current anomaly detection systems (OspreyData, Ambyint, and others) are making progress here, but most still generate more false positives than engineers would like. The challenge is not the anomaly detection algorithm itself but the contextual reasoning required to distinguish between a genuine problem and an expected operational change (shut-in for offset frac, planned maintenance, weather-related downtime).

Opportunity 2: Continuous Artificial Lift Optimization

Artificial lift optimization is not a one-time exercise. Well conditions change continuously -- fluid levels decline, water cuts increase, gas-liquid ratios shift, equipment degrades. The optimal artificial lift settings today may not be optimal next month. Manual re-optimization happens infrequently (quarterly or annually for most operators), leaving significant production on the table between adjustments.

AI-driven continuous optimization -- where the system adjusts artificial lift parameters in real time based on current well conditions -- is technically feasible and commercially available (Ambyint and others offer this today). The barriers to broader adoption are cultural (trusting an algorithm to control a well) and practical (ensuring fail-safe behavior when the algorithm encounters conditions outside its training data).

Opportunity 3: Production Forecasting With Operational Context

Traditional production forecasting (decline curve analysis) treats each well as a mathematical curve. In reality, production performance is driven by operational decisions -- artificial lift changes, workovers, restimulations, offset well completions -- that create discontinuities in the production curve. An AI-based forecasting system that incorporates operational context (what interventions are planned, how the artificial lift is being managed, what offset activity is expected) could provide significantly better forward-looking production estimates than traditional DCA.

Opportunity 4: Field-Level Optimization

Most production optimization happens at the individual well level. But the highest-value optimization problems are at the field level: How should gas lift gas be allocated across 50 wells to maximize total field production? How should water handling capacity be distributed across facilities? Which wells should be shut in during a compression outage to minimize total production loss?

These are combinatorial optimization problems that are mathematically well-suited to AI techniques. The challenge is that they require integrated data from across the field -- well-level production data, facility constraints, pipeline capacities, compression availability -- that is often siloed in different systems.

At Groundwork Analytics, we approach production optimization as a systems problem, not a well-by-well problem. The greatest untapped value in most producing fields lies not in optimizing individual wells in isolation but in optimizing the allocation of shared resources -- gas lift gas, compression, water handling capacity, personnel -- across the portfolio. These are optimization problems that require both domain expertise and computational methods, and they are the problems we focus on.

Opportunity 5: Predictive Maintenance for Artificial Lift

Equipment failure is the largest single source of production downtime in artificial lift operations. ESP failures, rod pump failures, and compressor outages cost the industry billions of dollars annually in lost production and workover expenses. Predictive maintenance -- detecting the early signatures of impending failure and intervening before catastrophic failure occurs -- has been a target for AI application for years.

The challenge is that predictive maintenance requires high-quality, high-frequency data from the equipment being monitored. For ESPs with downhole gauges, the data is often available. For rod pumps in remote locations with intermittent SCADA connectivity, the data may be insufficient for reliable prediction. The AI models are ready. The sensor infrastructure and data pipelines, in many cases, are not.

What Is Still Manual (And Why)

Despite the available software, a surprising amount of production engineering work remains manual:

Well test analysis -- Interpreting the results of production tests (flow rates, pressures, fluid samples) is still largely a manual, expert-driven process. Software provides tools for plotting and analysis, but the interpretation requires judgment that current AI cannot reliably replicate.

Workover prioritization -- Deciding which wells to work over, in what order, with what intervention, is a complex decision that involves production impact, cost, rig availability, permitting, and strategic priorities. Software can inform this decision but rarely drives it.

Production allocation -- In fields with commingled production (multiple wells flowing into shared facilities), allocating measured production at the facility level back to individual wells is a persistent challenge. Allocation methods range from well test prorations to virtual metering, but the process often involves manual adjustments and engineering judgment.

Regulatory reporting -- State production reporting requirements involve data collection, validation, and submission workflows that are only partially automated for most operators.

These manually intensive workflows are not manual because the technology does not exist to automate them. They are manual because the data integration, contextual reasoning, and exception handling required exceed what current production software provides. This is precisely the gap that well-designed AI systems can fill.

Practical Recommendations

For production engineering teams evaluating software and AI investments:

Fix your data pipeline first. No AI model will compensate for unreliable SCADA data. Invest in sensor maintenance, communication reliability, and data quality monitoring before deploying analytics platforms.

Start with surveillance, not optimization. Automated anomaly detection with proper contextual reasoning will free up engineering time that can then be redirected to optimization activities. This is a lower-risk, higher-confidence starting point than autonomous control.

Measure the baseline. Before deploying any optimization software, measure your current performance: average uptime, mean time to detect failures, mean time to respond, production efficiency. Without a baseline, you cannot evaluate whether the new system is delivering value.

Plan for the integration challenge. The most valuable production optimization opportunities require data from multiple systems (SCADA, artificial lift controllers, facility management, production accounting). Plan for integration costs and complexity upfront.

Demand vendor-neutral solutions. If your artificial lift fleet includes ESPs from SLB, rod pumps from Weatherford, and gas lift controllers from a third vendor, you need an optimization platform that works across all of them, not one that is optimized for a single manufacturer's equipment.

The production operations software market is mature in its data acquisition layer (SCADA) and maturing in its analytics layer (surveillance, AI-based anomaly detection). Where the industry has the most room to grow is in the optimization layer -- using the data and analytics to actually change how wells are operated, in real time, at scale. That is where AI will deliver its most significant impact.

Dr. Mehrdad Shirangi is the founder of Groundwork Analytics and holds a PhD from Stanford University in Energy Systems Optimization. He has been building AI solutions for the energy industry since 2018. Connect on X/Twitter and LinkedIn, or reach out at info@petropt.com.

Related Articles

Have questions about this topic? Get in touch.

Back to Insights