Managing risk with model-based simulations

Historical data analysis and system simulation can help manage the risk associated with meeting solar farm production goals

A reduction in actual solar-farm production compared to expected production will negatively affect the profitability of a solar farm. In the planning stages, detailed simulation models can provide solar-farm developers with the means to determine the spread of expected production. From this, developers can be more confident of the range of the expected return on investment.

Once the solar farm is operational, monitoring and analysis of historical data can help quantify any discrepancies between predicted and actual power outputs, and give the engineering team the means to make well-informed recommendations on improving the efficiency of the overall systems. Using simulation models as an in-service support tool, which can be continually refined and verified through comparison with historical data, provides the means to troubleshoot operational discrepancies in more detail and manage the risk associated with maintenance and plant expansion plans.

“Without simulation, solar farm developers may not be exposed to the detail of system response until the integration and commissioning stage,” says Graham Dudgeon, energy-production industry marketing manager at software developer MathWorks. “By then, if there are errors or the farm is not producing as expected, then you need to find out why and fix it — a potentially time-consuming and expensive task.”

Simulation software, however, lets users design sites to get a feel for how the farm and power-conversion systems will operate and how all components work together. This lets engineers find errors and fix them before construction. Simulation is just a small part of how software, such as MathWorks MATLAB, can help manage risks associated with power generation.

“The ability to access, analyze, and report system data is key to optimizing performance and managing risk,” Dudgeon says. “Solar companies can use historical data to optimize power generation and operational efficiency.”

Accessing data
Today, modern data loggers make it possible to gather enormous amounts of data. But to reduce risks, it must be organized so analysts can find problems and understand system performance before they become bigger issues.

“A large solar farm can generate multiple measurements during operation and have, potentially, gigabytes worth of data,” Dudgeon says. “Measuring and storing is done in a variety of ways, such as with SCADA (supervisory control and data acquisition) systems and data historians. However, once data is available, the challenge lies in how to sort through and use it to find and diagnose problems or improve the system’s efficiency.”

Programs such as MATLAB can access a variety of data sources, Dudgeon explains, such as OPC (open process control) servers and SQL (Structured Query Language) databases, to bring in measured data from an instrument or database — no matter what form the data is in.

Analyzing data
Historical data may be reviewed over a period of hours, days, or years, Dudgeon explains. When looking at large volumes of information, you want to understand the time-line you are interested in and the events on the system, such as component failures. You need a means to do this. MATLAB lets users view and understand historical data to find ways to improve solar-farm operation. The goal is to make sure the solar farm is operating at maximum potential for the maximum amount of time. Software can make this easier.

“Say you’re looking at insolation, energy the sun is actually providing over the solar farm,” Dudgeon says. “The sun is bright and the solar farm is operating. You know the panels should be producing, say, 100 mW. But suppose the equipment is only producing 80 mW. When you see you aren’t getting the expected efficiency, the challenge lies in finding out why.”

The discrepancy could be due to many reasons such as component failures or control systems not doing what you’d like them to do. Analyzing the data can help determine the cause. The main problem will be the large amount of information. Dudgeon says MATLAB can perform computationally intensive tasks faster than traditional programming languages.

Report the data
Once technicians visualize the data and perform numeric computations, they can publish the results. They explore the data, record the information and share it in a Word document, PowerPoint, PDF, etc. Dudgeon says once an engineer identifies the reasons for the mismatch, he can provide recommendations to remedy the problem and report those recommendations in convenient ways.

Historical data can help future operation
Data analysis and simulation software is available to develop predictive models based on historical data to see a system’s output before it’s built. For example, monitoring of a potential solar site will provide a model developer with insolation data that can be used as an input to a simulation model. Simulink, also from MathWorks, works with MATLAB to help engineers develop physical models and the associates control systems.

“One can develop a physical-system simulation of a solar farm to provide an engineering platform that emulates the real system,” Dudgeon says. “Such a physical system would consist of a model of the panels, converter connection to the grid and management system. This way, the model shows the physics of what’s going on and will help perform process improvement.”

One method of improving a system is to use historical data to verify the physical model. Dudgeon says it’s important to gain confidence that the model is accurate and representative of the real world. The model becomes a tool for in-service support. Historical data helps continually refine and improve the original model to get a better simulation with more accuracy each time.

Simulation allows examining retrofits and upgrades before committing significant capital expenditures and provides confidence that component selection is right-sized and fit-for-purpose.

“Simulation allows exploring different options,” Dudgeon says. “You can simulate the modules’ current and voltage response, and input a measured solar intensity. Suppose you need an inverter that will use maximum power-point tracking and power factor control to provide electricity to the grid. Simulation helps choose the right control strategy and the right size of inverter or other component for the task at hand, without having to commit an expenditure on something larger or more powerful than you need.”

Simulation studies help build confidence when actually buying and integrating equipment because you’ll know what you expect it to do, Dudgeon says. Model-based design help close the gap between what’s expected and reality. SPW