A new Data Architecture for Solar Farm Management

By Paraic O’Toole, CEO Automsoft

New Requirements for a New Environment

Industrial automation software such as SCADA systems, distributed control systems and data historians are generally designed to manage complex environments with many moving parts. Think of a traditional fossil fuel burning power station with the big cooling towers, once a common feature of the landscape. These environments have complex rotating equipment like turbines, pumps and generators. They have disparate systems like fuel management, emission control systems and so on. Now contrast this to a typical solar installation. Some do have moving parts where arrays track the movement of the sun. However in most cases solar arrays are static, relatively straightforward environments in terms of equipment and processes.

I’m not suggesting that there aren’t layers of complexity associated with managing a solar farm, rather that the solar industry requires software designed and configured specifically for the industry, rather than the shoehorning of traditional IA software into this environment.

The Correct Data Set

To start, because there is a distinction between the constructor of solar plants and the contractor to operate and maintain (O&M), systems are frequently minimal in scope. Data collection and analytics are frequently left as a problem for the O&M contractor to resolve. For their part, the O&M contractor is typically contractually held to a reliability, uptime and output series of KPIs.

What is really needed is the efficient collection of the correct data set: We need to know how the assets are performing, both against spec and relative to identical assets in similar environments. In certain environments like the desert, windy conditions may lead to sand obscuring some panels accounting for performance deficiencies for example. Inverters may underperform or overheat and plants may have interference from humans or animals. We want to know when something goes wrong – what were the associated factors when it went wrong. We want to know how long and how much it costs to repair.

Environmental vs Operational Factors

In choosing KPIs to monitor, it is really important to evaluate them under two headings – are they factors or metrics we can influence i.e. their value acts as a trigger for action; or do they indicate a change in performance due to an event in the operating environment. Overall, we need to compare the actual power generated by the plant with what it should be and what are the reasons for any deviation. Like most things, when you distil it down, it’s quite simple.

Meaningful KPIs

KPIs that are meaningful, therefore, compare the maximum theoretical output (taking into account clipping) against actual performance. This tells us how we are performing and indicates if we are underperforming. KPIs from the inverters identify if there is a problem with the inverters and the combiner boxes feeding into the inverters. KPIs from the CMMS (maintenance system) measure how efficient we are at resolving errors and minimising downtime. Integrating real-time weather data from the weather station factors in real world conditions.

A New Data Architecture and its Benefits

I would propose the following features for a new integrated data architecture: A SCADA-lite (since its function is mainly to get data from the panels) and a good but not necessarily high cost CMMS. A data repository to manage the large volumes of time series data generated and to receive data from the CMMS and external data sources such as weather data. An analytics and reporting engine to calculate and present the KPIs and reports to the various audiences within O&M and the solar installation owner.

The consolidated data repository, or data lake, then becomes an enterprise-wide corporate resource that enables advanced analytics for solar farm management. An enterprise-wide solution has the advantage, particularly for O&M companies managing multiple solar installations, of aggregated learnings regarding asset performance, efficiencies and causes of errors. The ability to predict future outcomes become more accurate the larger the dataset and results in less downtime, faster recovery time and lower cost of operation.

Contact [email protected] if you would like to discuss your Solar Farm data management requirements.