Although many claim that hydropower units are some of the most robust, reliable machines in the industry, which require little to no machine condition monitoring, current power generation requirements are in fact changing that perception. The automation and digital transformation that is taking place in the hydropower industry is also forcing energy companies to pay more attention to their asset healthcare, not only to the operation, maintenance and reliability of the units but also to the monitoring systems themselves that can support this healthcare.
Traditional vs. new hydropower methods
Traditional hydropower methods feature overdesigned units, whereas newer methods feature value-engineered units that are streamlined to the application. The newer units are more cost-effective to manufacture, but often they can be less robust. Traditional methods feature continuous baseload operation, whereas newer methods are based on load following, peaking and pumped storage, often with many stops and starts. This, plus the effects of partial loading, risk over-stressing the machinery and wearing out components prematurely. Life extension and performance upgrade is available for some older units, but the added performance raises the stress levels applied to those original components that have not been changed.
Traditional maintenance strategies utilised fixed time intervals and were well staffed and well equipped for maintenance and overhauls. This enabled the hydropower units to operate for many years without problems, and therefore condition monitoring requirements were minimal, if needed at all. With newer operation and maintenance strategies, fewer maintenance and diagnostic staff are utilised, even though there is less tolerance for downtime. This means there is more need for predictive maintenance, and condition monitoring is an important part of that.
Finally, whereas traditional methods are typically state-owned and not focused so much on profitability, newer methods are more geared to reducing life cycle costs and improving operation and maintenance practices. Consequently, this means newer hydropower methods are more open to implementing automation and digital transformation technologies.
Reliability and profitability in modern hydro units
One of the biggest differences from the past is that hydropower utilities now place a great deal of importance on reliability and profitability. They must economically fulfill consumer demand (peaking and variable load power), as well as minimse disruption and bolster profit for the utilities and stakeholders. Reliability centered maintenance; ISO 55000 must be in place.
Today, there are more potential failure modes to detect as greater stress is placed on the modern streamlined units and on those older units that have been upgraded. In addition to this, monitoring must be performed at different machine states to handle the variable generation loads now required. Moreover, more lead time is needed for fault detection and diagnosis. Because of all of this, condition monitoring systems must meet these requirements and comply with new data management needs thanks to industry-wide digital transformation.
Condition monitoring’s role in a plant’s digital transformation
Industry 4.0 and IoT are transforming the industrial landscape by driving the need for digital transformation, and the hydropower industry is no exception. These technologies offer unparalleled connectivity, data availability, and automation capabilities, leading to improved efficiency, productivity, and innovation. Embracing digital transformation is essential, with the objective to provide real-time access to enterprise data and predictive performance using management tools for improved operations, maintenance, reliability and logistics. The benefits include optimal resource management, fast, accurate data-driven insights, improved efficiency, and agile production. Condition monitoring will have an important role to play in digitalisation for increasing asset utilisation, driving down asset maintenance costs, reducing downtime, and improving operational efficiency.
Traditional condition monitoring systems have been typically stand-alone systems with their own proprietary servers and data storage. This is changing, however, due to the hydropower digital transformation needs.
Condition monitoring solutions
First and foremost, there is no unique ‘out-of-the-box’ condition monitoring solution that fits all hydropower applications and all current customer needs. This, however, is not a problem by current standards, since there are monitoring solutions today that integrate seamlessly with other data systems in an enterprise-wide platform, and thus provide more value in machine healthcare because of the digital transformation. What is important nowadays is how data is analyzed and managed in a monitoring solution.
Although measurement techniques used in monitoring machine components have not changed much over the years, the way the measurement data is analysed has changed dramatically. Instead of detecting a machine fault based on single vibration symptom that is monitored to fixed alarm limits, several vibration measurements including process parameters can now be automatically analysed together using data-driven AI algorithms, pattern recognition, digital twins, and other means, which ultimately result in more reliable and earlier fault detection. In addition to earlier anomaly detection, AI and ML algorithms can also be used for improving diagnostic insight of a fault, predicting remaining useful life and providing automatic decision support for planning maintenance and optimizing operation. It can be provided by the monitoring system itself or by a remote service.
For data-driven AI tools to be successful, however, a lot of data is needed and it must be managed effectively. Therefore, in addition to new analysis techniques, one of the other big changes from traditional monitoring solutions is how the data is managed, i.e. how and where it is stored, processed, analysed and shared. There are several solutions for this, of which all can be combined.
Off-premise data management and service
Data is stored in off-premise cybersecure cloud services, and there is no need for a data acquisition unit when wireless sensors are used. This setup allows for fast and easy installation, configuration, and commissioning, and no monitoring system servers must be managed. It is also possible to install wired condition monitoring edge devices, if needed, for a more advanced monitoring solution. No matter which sensors or monitoring devices are installed, the ‘condition monitoring as a service’ solution provides actionable insights – supported by AI and approved by experts – through a remote monitoring and diagnostic service. This is the perfect solution for those end-users who do not have in-house diagnostic expertise, or there is a need to reduce the workload on the specialists who must watch over a lot of machines.
In its purest form, an off-premise monitoring solution is typically a stand-alone proprietary solution that is not integrated with an enterprise-wide data management system, but this is sufficient for many small-time operators who are running small hydropower units. Even for larger operators with enterprise data management and large generating units, the off-premise solution can be used for monitoring balance-of-plant pumps and fans, which may not necessarily be integrated in the historian. Of course, it is still possible to save data to the historian while the monitoring and diagnostics are done remotely, off-premise, but this raises a cyber security risk if not properly managed.
On-premises with enterprise-wide data management and service
In this case, the data acquisition unit and sensors are installed on-premise, but the data is stored and monitored in an enterprise-wide historian (or locally, if needed). Event notification, visualisation, and analytics are all done in the historian, using data from the monitoring system. One of the important requirements for a monitoring system to be used in this solution is the ability to store raw dynamic data in the historian in an efficient, effective manner. Not all monitoring systems can do that. A historian database is a practical solution since there is more data available than is provided by a single stand-alone system. This is ideal for correlating measurements with process data correlation, which makes diagnostics faster, easier and more reliable. Various monitoring systems and service providers can access this data for further processing, including for AI and machine learning. Analytics done in the historian are generally more transparent and easier to fine-tune than what can be done in the stand-alone monitoring system.
Some condition monitoring systems include hardware that also provides machine protection in one system. In such a system the monitoring system has to be on-premise, since machine protection cannot be provided by remote services.
B&K Vibro monitoring solutions
B&K Vibro offers both on-premise and off-premise monitoring solutions to the hydropower industry, but in this case study, an on-premise rack-based monitoring system, SETPOINT, was installed. It has a native connection to the AVEVA™ PI System™ for condition monitoring. This same system also includes protection and advanced hydropower unit monitoring capability such as multiple machine states, specific plots and monitoring techniques for monitoring hydropower units (e.g. air gap, magnetic flux), and non-changing data reduction. Another special feature, a data ‘flight recorder,’ is used when the monitoring network is down.
Case study in Brazil
A hydropower unit in Brazil features a storage-based reservoir of 151m2 and is equipped with two 113MW Francis turbines. Originally constructed in the 1970s, the unit underwent significant upgrades between 2013 and 2015. It operates primarily under a fixed load and synchronous condenser mode, with oversight provided by a Generation Operations Center.
The primary challenges faced by the operator at this facility included increasing the availability of the turbines, reducing the frequency of inspections, and transitioning to a predictive maintenance strategy. To address these challenges, the project involved the integration of condition monitoring and diagnostics into the existing AVEVA™ PI System™, allowing for more efficient and effective operation. This integration enabled real-time monitoring and data analysis, significantly enhancing the plant’s operational reliability and maintenance processes.
The operator can monitor the units (as with the units at their other hydropower stations) at partial load to load-specific alarm limits. This monitoring strategy enabled them to avoid subjecting the units to hydraulic disturbances such as vortex rope turbulence and cavitation, which occur at partial load, and yet still generate power for a wider range. They also perform analytics on the data – some of it driven by AI – to more closely keep an eye on the hydropower unit healthcare. One example of this includes taking SETPOINT vibration data (including harmonics and phase) in correlation with process data (temperatures, flow, pressure), and then using machine learning to detect developing faults earlier. The results of this analysis, of which several faults were accurately detected, are then put back into the AVEVA PI System. The algorithm for this diagnostic tool was trained over a period of eight months using data from nine hydropower stations and will be extended as experience is gained.
Summary
Enterprise data management enhances efficiency and reliability by eliminating the limitations and costs associated with proprietary and standalone systems. The hydropower industry has already embarked on digital transformation, leveraging off-premises and enterprise on-premises machine condition monitoring systems to deliver basic and advanced monitoring solutions for small, medium, and large hydropower plants. These systems enable early fault detection and automatic decision support with AI and machine learning, to improve machine healthcare and optimise operation and maintenance. Moreover, third-party service providers can easily access cyber-secure data to offer transparent analytics, such as thermodynamic performance calculations, which can be fine-tuned as experience grows. In contrast, proprietary systems often restrict access and rely on ‘black box’ calculations that are difficult to modify, highlighting the superior flexibility and adaptability of enterprise data management solutions.
The case study in Brazil is an example of a well-developed digital transformation project that combines data provided by various systems, including the SETPOINT system in the AVEVA™ PI System™ historian. Process data is correlated with the monitored data for more accurate, reliable diagnostics.
