By Lauren DeLorenzo, Assistant Editor, Pump Industry Magazine
Pump monitoring technology has been around since the 1970s, but as the industry launches into the digital revolution, pumps equipped with artificial intelligence (AI) systems are expected to be the rule, not the exception.
As pump manufacturers and operators embrace these new technologies, the industry is expected to transition from a product-based industry to a service-based one.
Now, the focus is not just on ensuring a high-quality product, but partnering with original equipment manufacturers (OEMs) to envision and build more efficient, more connected pump ecosystems.
This swift shift has changed the focus for pump manufacturers. Between 50 and 60 per cent of OEMs’ revenue will come from service-related activities (real-time monitoring and reliability services) by 2025, according to a Frost & Sullivan analysis, 2025 Vision: Future of Pumps in a Connected World.
Pump OEMs are looking to diversify portfolios with new services that can provide meaningful data insights. These service areas might include cloud computing, edge analytics, smart sensor mesh networks, blockchain, predictive maintenance and other IIoT-based technologies, while leveraging AI to reduce energy consumption and control operational costs for pumps. With a new age of digital pumps on the horizon, how are AI-enabled systems expected to change the future of the industry?
Predictive tools
With AI-enabled predictive analytics, there is an overall reduction in both operating and lifecycle costs. A Deloitte analysis found that predictive maintenance can reduce overall maintenance costs by up to 30 per cent, and reduce breakdowns by 70 per cent.
Wireless technology can connect advanced pump sensors to the cloud for analysis by AI software, offering a more comprehensive and long-term vision of equipment performance, as opposed to manual checks. Typically, manual pump equipment assessments will be carried out periodically.
This can give an indication of whether there are any potential operating issues which could cause a shutdown or disruptions in the short term.
However, during the gaps between these manual equipment checks, issues can arise without an operator’s knowledge. With innovative pump monitoring devices, operators can check on pumps daily in real time, leaving no time for potential issues to arise.
A proactive approach
This shift from a reactive to a proactive approach to maintenance has been happening for a number of years, and is key to extending equipment life. AI, the industrial internet of things (IIoT) and other smart pumping methods are transforming not just practical operations, but also our understanding of holistic equipment management.
Data from sensors can offer early detection and diagnostic processes, troubleshooting problems before they impact plant efficiency and incur extra costs. This means that plant managers can be on top of signs of equipment failure or maintenance concerns before they result in problems that can affect the operation.
Smart sensors can operate, control and protect pumps and their associated systems, helping operators to more easily avoid common causes for equipment failure. With predictive AI systems, it is possible to anticipate when a pump is continuously running below minimum flow or dry running due to closed suction valves.
These sensors can also indicate when insufficient net positive suction head available causes cavitation, or when closed discharge valves result in heat buildup and subsequent liquid vaporisation.
For example, excessive vibration is a common cause of wear and tear for pumps, and can quickly and drastically cut down on expected equipment life.
With AI-enabled pumps, operators can more accurately sense too-high vibration levels. Even if this issue is overlooked, and a pump breaks down because of improper operation, managers will be able to look back at a continuous record of high vibration levels, examine why the breakdown occurred, and make an informed plan for how to prevent future issues. This type of data analysis is also available for other common causes of poor equipment health, such as temperature, run-time hours and battery life.
Visualising trends
AI-based predictive analytics can use historical data to identify performance trends and estimate the remaining lifetime of equipment.
This ability gives operators a much clearer picture of when repair and replacement actions may need to be scheduled. This is extremely useful, as it can help managers determine how to go about maintenance issues in a way which will have the least impact on operations, rather than forcing a reactive shutdown or outage.
Advanced product data management combines AI-based analytics with neural networks to compress data into visual, usable information. This can involve dashboards full of graphics and information, which users can look at to quickly survey equipment conditions and trends, leading to more informed, data-driven decisions.
Advanced condition monitoring
Condition monitoring can involve the observation of a number of different elements, depending on the type of pump being used. In condition monitoring-based AI, data is enhanced through pre-processing methods before it is sent through the AI system. AI has been used for automatic fault detection of centrifugal pumps to allow for remote monitoring of multiple pumps at the same time.
With cloud-based computing, operators can manage facilities halfway around the world, with pumps in different locations, but connected to the same system, learning from each other and adjusting their performance accordingly through a connected AI system.
Thanks to cloud-based computing, pumps in vastly different locations can be controlled through smart sensors, which detect changes in vibration, temperature, pressure and electrical signals. These sensors can track data in real time, and can be connected to AI systems through a complex diagnostic system.
Pumps that ‘learn’ from data
AI can also teach the equipment how it is supposed to operate, allowing it to detect anomalies and adjust its performance accordingly.
Employing self-diagnostic features can also give users time to schedule and prioritise the maintenance needs of equipment, giving operators more control over the health of the system.
AI-driven machine learning can not only help operators monitor the performance of pump system components – it can also be used to optimise the output of the machinery. Compressors, turbines and drives can all be tracked with visualisation tools, which in turn can reveal and minimise any inefficiencies.
When coupled with digital twins, AI-enabled machine learning can simplify maintenance with predictive tools to boost the reliability of pumping equipment and get the most out of the assets. These tools can also identify and amplify strengths in the operating system.
More recent developments are exploring how AI can be further integrated with IIoT in other areas of machine condition monitoring.
Collated data
A common challenge is that companies can monitor hundreds, sometimes thousands of machines at one time, making collecting, organising and analysing data extremely difficult. This can result in oversights, and limit the operator’s ability to monitor equipment in real time.
A proactive, AI-enabled system, however, can tackle this problem by converting massive volumes of data into collated patterns and trends for a more comprehensive analysis. Innovative AI algorithms, which are built from software and neural networks, can analyse advanced sensor data in multiple formats, distilling them into trends that can identify normal and abnormal operation patterns.
AI algorithms can adapt to a specific facility or system, learning from previous data to provide more valuable insights and highlight correlations between specific factors and equipment performance. These analytics can get to the true underlying cause of the issue, rather than addressing it solely at the surface level. This approach reduces downtime and extraneous costs by giving operators more time to deploy solutions to foreseen issues.
Centrifugal pumps and AI
Centrifugal pumps can be prone to hydraulic faults, including cavitation, waterhammer and turbulence. They can also fall victim to mechanical defects, such as misalignment, imbalance, bearing damages and impeller damage.
Recent research has investigated how these faults can be mitigated with the use of AI. One study examined a centrifugal pump under seven different testing conditions – normal, imbalance, misalignment, impeller damage, bearing damage, mechanical looseness and cavitation.
An accelerometer was used to collect vibration data, which was then put through pre-processing algorithms (AI system). By integrating the AI with an advanced diagnostic system, faults could be processed and classified for further machine performance analysis.
The future of AI-enabled pumps
AI system improvements aren’t just limited to motors and pumps – the entire operating system can benefit from how AI interacts with sensors, relays and optical sensing equipment.
As IIoT continues to set higher service expectations for OEMs and operators alike, the pump industry must shift towards a new business model that encompasses the support, management and maintenance of equipment, while making operations more efficient and sustainable.
It’s clear that the industry is going through a transformation that is here to stay, and the growing realisation of how AI can unlock cost and efficiency benefits will prove to be one of the greatest tools for operators who are looking towards a more user-friendly and efficient future.
Comments are closed.
Truely said.. there is a lot of data collected from IoT devices, the industry can effeciently analyze the trend and make AI based predictive model.