Operations and Maintenance Special
A key distinction needs to be made between performance monitoring and condition monitoring.
Condition monitoring involves the installation of special instrumentation to measure vibrations, pressure, temperature and other phenomena. This will incur downtime and may have warranty implications. The assessments associated with condition monitoring can be characterised by the time scales typical of the measurements the instruments make. These conditions are typically trended over time.
Performance monitoring requires no additional instrumentation and utilises for its analyses only those data that are already being routinely acquired by the SCADA system.
Condition monitoring is focused on the condition of individual components, subsystems and their planned maintenance whereas performance monitoring is based on assessments of the performance of the system as a whole with a more operational focus. As such performance monitoring and assessment entails a greater emphasis on statistical analysis of power curves for the characterization of anomalous performance and the integration of event data with time series data for the attribution of variance. The time scales performance monitoring represents are therefore those associated with the accumulation of performance statistics.
“Condition monitoring and performance monitoring should be viewed as complementary approaches,” Peter Clive, technical development officer, SGURR Energy Ltd said, in an interview with windenergyupdate.com.
According to Clive, initially an arrangement needs to be made to allow those conducting the performance assessment to access and retrieve the SCADA data.
This can generally be achieved in a manner most convenient to the client. Once it is in place routine analysis of the SCADA data can commence and weekly bulletins and monthly reports are issued detailing performance, accurately attributing revenue variance, identifying key issues and making recommendations for remedial action, pre-emptive interventions and the most effective configuration of O&M infrastructure.
Approach towards analysing the data
The data routinely acquired and stored by SCADA systems represents a vast amount of valuable information that is potentially of great benefit in terms of enabling accurate revenue variance attribution, well informed O&M strategies and infrastructure configuration, pro-active and pre-emptive intervention to alleviate anomalous performance, and the tuning of operational parameters to maximise revenues.
This information has historically been widely neglected because of the resource necessary to extract it from the data, according to a specialist like SGURR Energy.
Clive points out that it has often been considered worthwhile to analyse the data only once something has already gone wrong with a turbine. However, analyses that have hitherto been conducted in a reactive, post hoc manner to diagnose faults or to progress post investment appraisals are now being automated, radically reducing the resource necessary to conduct them, and new tools have been developed which further enable rapid performance assessment and enhance its value. Rapid performance assessment enables routine performance assessment from which all the benefits of a more pro-active approach can be derived.
One current trend is the move towards data aggregation among owners, operators and network management of multiple wind farms: the data from multiple disparate SCADA systems installed at multiple wind farms is put into a common format for centralised asset control in a “SCADA on top of SCADA” setup. This enables more efficient review of the performance of portfolios of wind farms and is also used in some instances to assist forecasting. As yet the full benefits of this trend have not been realized as in general the full potential for performance assessment is not being exploited.
Another interesting trend is the ongoing investigation of exactly how much information can be extracted from routine SCADA data. For example, an active topic of research that is informing our approach to performance assessment is the extent to which the stresses the machine is subject to, arising from, for example, turbulence, wind shear and veer, and flow inclination, can be discerned using standard performance assessment tools. This field is one of the most rapidly progressing fields in a rapidly developing industry and SgurrEnergy engages in continuous innovation to remain at the cutting edge delivering the maximum achievable benefit to its clients.
One possible benefit of third party performance assessment in the future is the possibility it raises for the continuation of the trend towards data aggregation such that performance assessment in terms of turbine inter-comparison is not conducted relative only to other turbines within the same individual wind farm or even within a single portfolio but rather extended to encompass entire fleets. In this way the performance of an individual turbine can be routinely compared to the performance of every other turbine of the same make and model to derive the maximum benefit.
The development of offshore wind farms poses its own unique set of O&M problems, and the ability to conduct routine performance assessment is of particular benefit in this context, where it may ultimately be viewed as a sine qua non.
“In general the most valuable analyses of turbine performance are not implemented by the SCADA system per se but by the tools with which the system may be augmented for the purposes of performance monitoring and assessment and which utilise for that purpose the data routinely acquired by the SCADA system. For example, some limited turbine inter-comparison and performance trending is sometimes performed by reporting modules of SCADA systems but the most sophisticated analyses delivering the greatest benefit are beyond the scope of essential supervisory control and data acquisition implemented by SCADA,” said Clive.
“In order to facilitate performance assessment, a SCADA system should naturally be highly reliable. Event data such as alarms, warnings, and log entries, should adequately record the events they describe. The time series data should include status fields recording the duration of specific conditions during each averaging interval. Good retention of and easy access to historical data is a requisite. Simplicity is a key feature when it comes to integrating the system into a data aggregation strategy.”
There are systems that take the raw data generated by site SCADA, and, using a combination of automated and human processing, to produce the most accurate picture possible of site performance from a variety of perspectives. Every period of turbine downtime is evaluated and categorized, and a power performance model derived for the site in order to accurately measure the cost of losses.
According to Clive, a key consideration has to be the speed: the value added by conducting performance monitoring and assessment comes from the speed with which power performance benchmarks are obtained, performance assessments are made relative to these benchmarks and convergence on any real problems of which the data is evidence is achieved by the automated components of the performance monitoring system, for further detailed and focused investigation.
The key equation is: Rapid performance assessment = Routine performance assessment.
Routine performance assessment is becoming increasingly important as operational assets increase in size and are deployed in ever more demanding and inaccessible environments on an unmanned basis. In such environments a reactive approach simply is not adequate and so routine performance assessment is required to facilitate proactive and pre-emptive action, allowing the operators to take as much control as possible in difficult environments.
The benefits can be summarised by observing: Routine performance assessment = Reduced revenue variance.
Irreducible human contribution to performance assessment
It is said that the reporting modules of SCADA systems rarely have the ability to resolve alarm cascades for a root cause; accurately categorise activity external to the turbine; or have the ability to combine intelligently with a work control system.
In this context, Clive points out that the core functionality of a SCADA system does not extend to analysis.
“The limitations of SCADA systems and the need to augment them with sophisticated performance assessment tools alluded to above notwithstanding: there will always be an irreducible human contribution to performance assessment and the interpretation of the results produced by automated systems. What automated systems provide is a massively accelerated convergence on the real points of interest where the attention of the trained operator can most productively be directed. The identification of anomalous turbine performance and the associated events and alarms previously might have taken days, but now can take minutes or even seconds. This allows investigation of the anomaly to begin immediately where previously it would have taken a significant amount of time and resource just to get to the starting line.”
Many approaches have been adopted to automate the treatment of event cascades, including multivariate auto-regressive statistics and AI techniques from neural networks to simple list processing.
“Often problems arise because adjustments are made to the way in which particular SCADA systems report events, for example in response to the reallocation of turbine faults and penalizing event codes. These adjustments impact differently depending on the approach taken in automating the treatment of event cascade and so a combination of techniques may prove to be the most robust solution. At present, though, trained personnel still retain a central role in scrutinizing and understanding event cascades,” said Clive.
Wind Energy Operations and Maintenance Summit USA
A session, titled `Improving yield by interpreting SCADA data’ will be conducted as part of Wind Energy Operations and Maintenance Summit USA, to be held on April 1-2 in Dallas, Texas this year.
For more information, click here:
Contact: Tom Evans by email firstname.lastname@example.org