Rotating Equipment Condition Monitoring Maintenance & Fault Prediction using Wireless Sensors

Problem

An industrial conglomerate faced challenges with their existing maintenance practices for rotating equipment, particularly screw compressors and their motors. Despite regular preventive maintenance processes and periodic vibration analyses, the company experienced a year-over-year increase in equipment failures. The delays in procuring components for rebuilding further complicated the situation. The traditional inspection regime did not yield the intended results, and the company needed a more effective and proactive approach to prevent failures and optimize maintenance.

Approach

To address the challenges and enhance equipment maintenance, the company adopted a Condition Monitoring Maintenance (CMM) application with wireless sensors and AI-based analytics. The following approach was implemented:

• Smart Wireless Sensors: Proprietary wireless sensors were installed on multiple refrigeration chillers, each comprising a 500HP AC motor driving an 800-ton Frick ammonia twin screw compressor. These sensors continuously monitor tri-axis vibration and temperature data from the rotating equipment. 
• AI + Physics-based Analytics: The CMM application utilizes AI and physics-based analytics models developed by Algo8 to analyze the vibration and temperature data in real-time. The AI model learns from the collected data to create an evolving condition map of the equipment across its operating range. 
• Early Fault Detection: By continuously analyzing small snippets of vibration and temperature information, the application detects anomalies and deviations from normal equipment behavior. It can predict potential failure modes and provides actionable insights for timely corrective actions. 
• Real-time Alerts: The CMM application immediately alerts the maintenance team through the app and email when it identifies increasing overall vibration levels and an increasing noise floor, indicating potential issues with the equipment. 

Benefit

The implementation of the CMM application with wireless sensors and AI-based analytics yielded significant benefits: 

• Cost Savings: The early detection of potential failures and condition-based maintenance allowed the company to save 90% of direct costs associated with rebuilding, including rigging and labor expenses. 
• Avoided Downtime and Production Loss: By identifying issues early, the application helped the company avoid unplanned downtime of 3 hours and prevented production losses during the same period. 

Impact

The adoption of the CMM application resulted in notable financial benefits for the industrial conglomerate:

• Direct Cost Savings: The early detection of equipment issues and consequent cost savings amounted to $40,000, providing a greater than 10x return on investment (ROI) for the motor-compressor. 
• Avoided Production Losses: Additionally, the application’s proactive approach prevented 3 hours of line downtime and production loss, resulting in further undisclosed dollar savings. 

By leveraging smart wireless sensors and AI-based analytics for rotating equipment condition monitoring, the company achieved cost savings, reduced downtime, and improved maintenance efficiency, ultimately leading to enhanced operational reliability and profitability. 

Real-Time Optimization of Refinery FCC

Problem

Optimizing plant performance is crucial to maximize output and achieve optimum profit, yield, production, and energy consumption without significant capital investments. However, many plants operate based on outdated design conditions and set points, leading to suboptimal performance. Additionally, the implementation of Advanced Process Control (APC) does not always result in true optimization as its primary function is process control, not maximizing performance. To address these challenges, a more effective and dynamic approach to optimize the Refinery Fluid Catalytic Cracking (FCC) process is required.

Approach

To achieve real-time optimization for the Refinery FCC process, Algo8’s solution employs the following approach:

• Plant Model and AI Optimizer: The Real-Time Optimization (RTO) system consists of a hybrid plant model that incorporates both first principles and data science. This model replicates the plant behavior and response accurately. The system also includes an AI optimizer that works with the plant model to optimize its objective function (e.g., production, yield, or profit). 
• Autotraining Capability: The hybrid plant model has autotraining capabilities, enabling it to adapt and learn from real-time plant behavior and conditions. This ensures accurate predictions and optimal performance even as the plant’s behavior and feedstock conditions change over time. 
• Manipulation of Independent Variables: The AI optimizer manipulates the independent variables within the plant model towards the boundaries to push the plant’s operation closer to its optimal limits. It identifies the best possible operating conditions to achieve the desired objectives while adhering to constraints and limitations. 
• Frequency of Optimization: The RTO system runs on a regular frequency (e.g., every 4 hours) or whenever significant changes in feed or operating conditions occur. It provides optimal target set points for the manipulated variables either as advisory inputs to DCS engineers (open loop) or directly communicates these values to the Advanced Process Control (APC) system for implementation in DCS (close loop). 

Benefit

The implementation of Algo8’s Real-Time Optimization (RTO) solution for Refinery FCC brings substantial benefits:

• Increased Profit: The RTO system helps achieve a 2% increase in profit by optimizing the product mix (e.g., propylene vs gasoline), optimizing catalyst consumption, and gaining higher total liquid production. 
• Energy Consumption Reduction: The RTO system leads to a significant reduction in energy consumption, resulting in 7-10% energy savings.

Impact

By adopting Algo8’s RTO solution, a refinery operating a 105kbpd FCC experienced substantial financial benefits:

The RTO solution added $8-9 million in annual benefits, primarily driven by improved profit and energy savings. The optimized FCC process resulted in enhanced plant performance, increased profitability, and reduced energy costs, making the refinery more competitive and efficient. 

Predictive Maintenance for Rotating Equipment (Compressor – MAB)

Problem

The main air blower (MAB) in the Fluid Catalytic Cracking (FCC) unit of a refinery is a critical centrifugal compressor that provides air necessary for combustion in the regenerator. The smooth functioning of the MAB is essential for the overall performance of the FCC unit. Unplanned shutdowns of the MAB can result in significant losses, including direct production losses, additional maintenance costs, and indirect efficiency losses due to unwanted downtime. The challenges faced by the refinery include:

• High Downtime Impact: Unplanned shutdown of the MAB leads to the shutdown of the entire FCC unit, resulting in substantial losses in terms of production, maintenance, and efficiency. 
• Frequency of Unplanned Shutdowns: The MAB has experienced multiple unplanned shutdowns, leading to financial losses and decreased production efficiency. 
• Monitoring Complexity: The MAB requires monitoring of over 200+ tags, and the lack of a single source of truth for 20 assets causes incomplete visibility into the system. 

Approach

To minimize unplanned shutdowns and gain visibility over MAB operations, the refinery adopted Algo8’s Predictive Maintenance application. The approach includes the following steps:

• Process and Machine Data Integration: The Predictive Maintenance application leverages both process and machine data to detect issues early. It uses historical data from the equipment and the process to train and fine-tune anomaly and predictive models. 
• Anomaly Detection Engine: Algo8’s Anomaly Detection Engine uses supervised and unsupervised machine learning algorithms to identify anomalous behavior in the MAB. It also determines the contributing factors responsible for the anomalous behavior, which could be related to process or operational issues. 
• Root Cause Analysis with Generative AI: The application uses Generative AI enabled Failure Mode and Effects Analysis (FMEA) to identify the failure mode responsible for the primary cause of the problem, leading to a better understanding of the underlying issues. 
• Predictive Model: The Predictive Maintenance application includes a predictive model that forecasts the remaining time to failure of the equipment. This early prediction allows operators to take action and prevent unplanned shutdowns. 
• Recommendations with Generative AI: The application provides recommendations through Generative AI enabled what-if scenarios. These recommendations help manage equipment failures proactively and extend the operating window of the MAB. 

Benefit

The implementation of Algo8’s Predictive Maintenance application resulted in several benefits for the refinery:

• Minimized Production Losses: By avoiding unplanned shutdowns, the refinery experienced minimized production losses, ensuring continuous operation and optimized productivity. 
• Reduced O&M Spend: The reduced frequency of unscheduled maintenance downtime led to cost savings in operations and maintenance expenses. 
• Decreased Safety Incidents: Predictive maintenance reduced the likelihood of safety incidents caused by unexpected shutdowns and equipment failures. 

Impact

The successful implementation of Algo8’s Predictive Maintenance application had a significant impact on the refinery: 

The refinery managed to save $7.5 million per shutdown, ensuring 100% asset availability by proactively addressing issues and preventing unplanned shutdowns. This not only improved the financial performance of the refinery but also increased overall operational efficiency and safety. 

Heat Exchangers Fouling Prediction and Cleaning Schedule Optimization

Problem

The refining and chemical industries face a significant challenge with heat exchanger fouling, where deposits of impurities accumulate on heat exchanger surfaces over time. This fouling reduces heat transfer efficiency, leading to several issues: 

• Increased CO2 Emissions: Reduced heat transfer efficiency requires higher energy consumption, leading to increased greenhouse gas emissions and environmental impact. 
• Decreased Plant Profitability: Heat exchanger fouling leads to reduced production throughput and increased energy costs, affecting the overall profitability of the plant. 
• Higher Maintenance Costs: Frequent fouling requires more frequent cleaning, leading to increased downtime and maintenance expenses. 
• Energy Consumption: Infrequent cleaning results in limited heat transfer capabilities, leading to higher energy consumption to achieve the desired process temperatures. 

Approach

To address the challenges of heat exchanger fouling and optimize the cleaning schedule, the following approach can be implemented:

• Data Collection and Monitoring: Install sensors on heat exchangers to continuously monitor key parameters, such as temperature, pressure, flow rates, and fouling indicators (e.g., pressure drop). Collect historical data on fouling patterns and cleaning schedules. 
• Machine Learning for Fouling Prediction: Utilize machine learning algorithms to analyze the sensor data and predict fouling tendencies. The models can identify early signs of fouling and estimate the rate of fouling accumulation for each heat exchanger. 
• Cleaning Cost Analysis: Analyze the cost associated with heat exchanger cleaning, considering factors like cleaning materials, labor, downtime, and energy consumption during the cleaning process. 
• Optimization Algorithm: Develop an optimization algorithm that balances the trade-off between cleaning frequency and energy consumption. The algorithm considers factors such as fouling prediction, cleaning cost, energy savings due to improved heat transfer, and operational constraints. 
• Real-Time Cleaning Decision Support: Implement a real-time decision support system that takes input from the fouling prediction models and optimization algorithm. The system recommends the most cost-effective cleaning schedule for each heat exchanger based on real-time conditions. 
• Condition-Based Cleaning: Move from a fixed schedule-based cleaning approach to a condition-based cleaning strategy. Clean the heat exchangers only when the fouling accumulation reaches a predetermined threshold, optimizing cleaning efforts and reducing unnecessary downtime. 
• Continuous Learning: Continuously update and improve the machine learning models based on new data to enhance fouling prediction accuracy over time. 

Benefit

• Improved Heat Exchanger Performance: The prediction of fouling tendencies and optimized cleaning schedule leads to improved heat transfer efficiency, resulting in reduced energy consumption and greenhouse gas emissions. 
• Enhanced Plant Profitability: Optimal cleaning schedules lead to increased production throughput, reduced energy costs, and enhanced plant profitability. 
• Reduced Maintenance Costs: By adopting a condition-based cleaning approach, unnecessary cleaning and associated maintenance costs are minimized. 
• Enhanced Equipment Reliability: Regular cleaning at the right intervals helps maintain the integrity and reliability of heat exchangers, extending their operational life. 
• Data-Driven Decision Making: The use of machine learning and optimization algorithms ensures data-driven decisions, maximizing the benefits of heat exchanger cleaning. 
• Environmental Sustainability: Reduced energy consumption and CO2 emissions contribute to the overall environmental sustainability of the refining and chemical industries. 
• Improved Asset Management: By better understanding fouling patterns and performance trends, plant operators can implement more efficient asset management strategies. 

By implementing a data-driven approach for heat exchanger fouling prediction and cleaning schedule optimization, the refining and chemical industries can significantly improve heat exchanger performance, reduce energy consumption and emissions, enhance plant profitability, and optimize maintenance costs. 

EOR Optimization and Polymer Injection Reduction for Upstream Oil & Gas Operator

Problem

The upstream oil and gas operator uses Polymer Enhanced Oil Recovery (EOR) to improve the sweep efficiency of existing water flood systems. Polymer injection involves increasing the viscosity of injected water with polymers to improve sweep efficiency. However, maintaining the right viscosity at the wellhead level presents several challenges:

• Viscosity Loss in Pipelines: The polymer solution experiences viscosity loss as it travels through pipelines from the centralized preparation facility to the wellhead. This leads to variations in the target viscosity at the injection points. 
• Lack of Visibility in Dilution Ratio: Maintaining the right dilution ratio of the polymer solution at the wellhead becomes challenging due to lack of real-time visibility and monitoring. 
• Limited Laboratory Testing: The remote location of multiple wells makes it difficult to perform lab testing of wellhead viscosity on a shift basis, leading to difficulties in adjusting injection rates.

Approach

To address the challenges in polymer injection and improve EOR optimization, the upstream oil and gas operator can implement the following approach:

• Real-Time Monitoring and Data Collection: Install sensors and monitoring devices along the pipelines to track the viscosity and other relevant parameters of the polymer solution in real-time. Gather data on flow rates, pressure, and temperature at various points to understand the viscosity loss during transportation. 
• IoT Connectivity and Data Integration: Utilize IoT technology to connect the monitoring devices and sensors to a centralized data platform. Integrate the real-time data with reservoir and geo data from the wells for a comprehensive understanding of the injection process. 
• Machine Learning Models: Develop machine learning models that analyze the real-time data and predict the viscosity at the wellhead based on historical patterns and pipeline characteristics. These models can adjust the target viscosity based on real-time conditions. 
• Automated Injection Adjustment: Implement an automated control system that adjusts the injection rates at the wellhead to maintain the desired viscosity level. The control system can take inputs from the machine learning models and dynamically adjust the polymer injection rates. 
• Remote Monitoring and Control: Enable remote monitoring and control of the polymer injection process from a central control room. Operators can access real-time data, make adjustments, and optimize injection rates without the need for on-site visits. 
• Data Visualization and Reporting: Develop a data visualization and reporting dashboard that provides a clear overview of the injection process, including real-time viscosity, injection rates, and efficiency. This allows for informed decision-making and performance analysis. 

Benefit

• Optimized Polymer Injection: The real-time monitoring and automated control system ensure that the right viscosity is maintained at the wellhead, leading to improved sweep efficiency and enhanced EOR optimization. 
• Reduced Polymer Consumption: With better control over polymer injection rates, the operator can reduce polymer consumption, resulting in cost savings and improved overall operational efficiency. 
• Enhanced Visibility and Efficiency: The IoT connectivity and data integration provide better visibility into the injection process, allowing for more efficient decision-making and troubleshooting. 
• Minimized Downtime: Proactive adjustments based on real-time data help minimize downtime caused by viscosity variations, leading to continuous and stable injection operations. 
• Remote Operations: Remote monitoring and control capabilities enable efficient management of multiple wells from a centralized location, reducing the need for on-site visits and associated costs. 
• Improved Production: By optimizing sweep efficiency and EOR operations, the operator can achieve improved oil production rates and enhanced recovery from existing wells.

In conclusion, implementing real-time monitoring, data integration, and automated control for polymer injection in EOR can lead to significant benefits, including enhanced sweep efficiency, reduced polymer consumption, and improved overall operational efficiency for the upstream oil and gas operator.

Real-Time Calculation of Asset Health based on Corroded Area Percentage

Problem

Operating plants and facilities generate a massive amount of data from various systems like DCS (Distributed Control Systems), historians, SCADA (Supervisory Control and Data Acquisition), and APM (Asset Performance Management). However, operators often struggle to gain meaningful insights from this data due to its sheer volume and lack of integration. As a result, they are unable to effectively monitor and optimize plant performance. Similarly, maintenance engineers and experts face challenges in planning and executing maintenance tasks and turnarounds because they lack centralized information and visibility of plant and equipment accessibility from remote locations or headquarters. This lack of real-time and comprehensive insights presents significant challenges in improving plant health and asset uptime, optimizing productivity, ensuring safety, and fostering collaboration among different departments and stakeholders.

Approach

The approach to solving the challenges in plant performance and maintenance planning involves the implementation of a Digital Twin for the operating plant or facility. Here’s a breakdown of the key steps:

• Data Integration and Centralization: Collect data from various sources, such as DCS, historians, SCADA, APM systems, IoT sensors, and other relevant databases. Integrate and centralize this data in a unified platform or database, making it easily accessible and enabling real-time analysis. 
• Digital Twin Creation: Develop a digital twin of the operating plant or facility, representing a virtual replica of the physical plant and its assets. The digital twin should accurately simulate the behavior of the real plant, including process dynamics, equipment performance, and response to operational changes. 
• Real-Time Monitoring and Analytics: Implement advanced analytics and machine learning algorithms to continuously monitor and analyze data from the operating plant and the digital twin. This enables real-time performance monitoring, anomaly detection, predictive maintenance, and optimization of plant processes. 
• Remote Accessibility and Visualization: Enable remote access to the digital twin and real-time plant data, providing operators, maintenance engineers, and experts the ability to monitor and control plant operations from remote locations or headquarters. Use immersive visualization techniques like Virtual Reality (VR) or Augmented Reality (AR) for enhanced situational awareness and decision-making. 
• Collaborative Environment: Create a collaborative platform where different departments and stakeholders can access and share relevant information, insights, and recommendations derived from the digital twin and real-time data. Foster cross-functional collaboration for more effective decision-making and problem-solving. 
• Maintenance Planning and Execution: Utilize the digital twin to improve maintenance planning and execution. Predictive maintenance insights can help prioritize maintenance tasks, schedule turnarounds, and optimize the use of resources, ultimately reducing downtime and enhancing asset reliability. 
• Operational Optimization: Leverage the insights from the digital twin to optimize plant processes, improve energy efficiency, and identify opportunities for productivity enhancement and cost reduction.

Benefit

• Improved Plant Performance: The digital twin and real-time analytics enable better monitoring and optimization of plant processes, leading to improved performance and increased operational efficiency. 
• Enhanced Asset Uptime: Predictive maintenance insights help identify potential equipment failures in advance, reducing downtime and increasing asset uptime. 
• Safety and Productivity: Real-time monitoring and analytics enhance safety by detecting anomalies and potential hazards early on. Additionally, optimized processes and maintenance reduce unplanned shutdowns, enhancing productivity. 
• Centralized Data and Insights: By centralizing data and insights, the digital twin facilitates quick access to relevant information, supporting informed decision-making across different departments. 
• Remote Accessibility: The ability to access and control the plant remotely improves operational agility and enables better coordination and collaboration among teams. 
• Cost Savings: Predictive maintenance and operational optimization lead to cost savings by reducing maintenance expenses and optimizing resource utilization. 
• Accelerated Turnarounds: Effective maintenance planning based on the digital twin’s insights streamlines turnaround processes, minimizing downtime during maintenance activities.

In summary, implementing a digital twin for an operating plant or facility offers a holistic solution to the challenges faced in performance monitoring, maintenance planning, and collaboration. It empowers operators, maintenance teams, and stakeholders with real-time insights, fostering a more efficient, safe, and productive plant operation.

Real-Time Calculation of Asset Health based on Corroded Area Percentage

Problem

Asset management in industries like manufacturing, oil and gas, and infrastructure often involves monitoring and inspecting assets for signs of corrosion, which can lead to significant safety risks and operational inefficiencies if left unaddressed. Traditional manual inspection methods are time-consuming, costly, and may not provide real-time insights into the health of assets. Identifying areas with high corrosion rates is crucial to prioritize maintenance activities and prevent potential failures. The challenge is to develop a system that can accurately and efficiently calculate the health of assets in real-time based on the percentage of corroded area, reducing the need for manual inspections and enabling prioritized maintenance planning.

Approach

The approach to implementing real-time asset health calculation based on the percentage of corroded area involves leveraging IoT sensors, computer vision, and data analytics. Here’s a breakdown of the key steps:

• IoT Sensor Deployment: Install IoT sensors on assets, both internally and externally, to collect data on corrosion levels. These sensors can measure environmental factors like temperature, humidity, and corrosive gases, as well as asset-specific parameters like wall thickness or coating condition. 
• Data Acquisition and Preprocessing: The IoT sensors continuously gather data on corrosion-related parameters. This data is preprocessed to remove noise, handle missing values, and ensure its accuracy and consistency. 
• Computer Vision for Corrosion Detection: For assets with visual surfaces, deploy computer vision algorithms to analyze images and detect areas of corrosion. These algorithms can segment corroded regions and calculate the percentage of the corroded area on the asset surface. 
• Data Integration: Combine data from IoT sensors and computer vision outputs to create a comprehensive view of asset health. The integrated data should include the percentage of corroded area, environmental conditions, and other relevant factors. 
• Health Index Calculation: Develop a health index or score based on the percentage of corroded area and other relevant parameters. The health index should represent the asset’s overall condition and potential risk of failure. 
• Real-Time Data Processing: Utilize data analytics and machine learning algorithms to process the integrated data in real-time. This ensures that the health index is continuously updated as new data is collected, providing real-time insights into asset health. 
• Prioritization of Maintenance Activities: Based on the health index, prioritize maintenance activities and intervention strategies. Assets with higher corrosion levels or deteriorating health would receive higher priority for maintenance, allowing for proactive planning and resource allocation. 
• Alerts and Notifications: Implement an alerting system that notifies maintenance teams when an asset’s health index reaches a critical threshold or requires immediate attention. This enables timely responses and reduces the risk of unexpected failures.

Benefit

• Reduced Manual Inspection: By automating the corrosion detection process using computer vision and IoT sensors, manual inspection requirements can be reduced by up to 90%, saving time and labor costs. 
• Real-Time Asset Health Monitoring: The system provides real-time updates on asset health, allowing proactive maintenance planning and minimizing the risk of unplanned downtime or failures. 
• Cost Savings: Prioritizing maintenance activities based on asset health helps allocate resources more efficiently, reducing unnecessary maintenance costs and optimizing asset performance. 
• Enhanced Safety and Reliability: Timely identification of corroded areas allows for prompt maintenance and repair, enhancing safety and prolonging asset life. 
• Improved Asset Management: Accurate and continuous monitoring of asset health enables better asset management decisions, leading to increased operational efficiency and reduced operational risks. 
• Increased Sustainability: Proactive maintenance and optimized resource allocation contribute to a more sustainable approach to asset management by reducing waste and unnecessary interventions. 
• Compliance and Regulatory Requirements: Efficient asset health monitoring aids in meeting regulatory requirements related to asset inspection and maintenance.

By implementing a real-time calculation of asset health based on the percentage of corroded area, organizations can streamline maintenance operations, enhance asset reliability, and reduce inspection costs, leading to improved overall asset management and operational performance.