Wind Farm Performance Management
What the wind farm Operator sees
Working from the ‘outside-in’, let’s look firstly at a few sample dashboards used to monitor local wind and power generation at four wind farms. These dashboards are built in RAPID’s HTML 5 visualization - RAPIDVision. (Click on the images to enlarge.)
Rotating dashboards are used in the main control room to provide Operators with an overview of the wind farms. Authorised Operators can view dashboards through a PC or a hand held device and have the ability to create, save and share their own private web pages within the system.
What goes on behind the dashboards
Diving deeper behind these dashboards, a number of things are happening which are necessary to avoid fines from the grid operators:
- Upper and lower power generation tolerances are calculated from the required dispatch load, the current generation load, frequency and grid limits
- Tolerance limits are compared with the actual generation and alarm conditions raised using RAPID Alarm and Events in real time if there are deviations
- A history of tolerances are stored for longer term reporting
All data calculations are carried out using the RAPID Calculation Engine and an inbuilt library of calculations for wind farm power generation. Operators can use this library and simply wire their data into a calculation block and out for storage and visualization.
What needs to be reported and how
Management requires formalized power generation reports on a daily basis, whereas Operators need to perform their own drilled down analysis and report on a less formal basis.
The RAPIDReports system is used to format formal management reports and schedule them for creation as PDFs. These are secure and cannot be edited, ensuring complete confidence in the reported data.
- Daily Reports – per wind farm, combined into a single daily report, automatically emailed to recipients at the start of each day - summary data for the previous 24 hours and rolled up data for the month and year to date.
- Meteorological Data - wind speed, wind direction, temperature, atmospheric pressure and relative humidity - is summarized into 5 minute periods and sent via a dedicated email account to a modelling system (M2M). The data is used for machine learning to increase its ability to predict local weather at the wind farm sites.
Operators produce informal and less structured reports using:
- Private pages in the dashboarding system
- Data integration with Microsoft Excel
- Quick Queries in RAPIDVision - a scratch pad to visualize any combination of data by means of either tables or trends, which can be stored and shared
How data is collected from the wind farms
By their nature, wind farms are widely dispersed with potentially high latency network connections between the wind farms and the central database at HQ. So to cope with this, RAPID’s Data Broker system collects data locally and feeds it to the central database using IoT protocols. Its inbuilt Store and Forward capability prevents data loss during network interruptions.
The external interfaces from the control systems use standard OPC-DA, OPC DA-XML and Modbus protocols. Custom development interfaces the Electronic Dispatch Logging system (EDIL) to RAPID. The EDIL system is used to send power production settings from the grid to the generator, and is implemented in a relational database table.
Manual data is added either by means of a simple User Interface to enter values, or by means of ingestion of larger volumes of data from structured files (CSV, JSON), for example, the yearly generation targets, monthly scheduled maintenance down time and monthly, weekly and daily weather forecast data.
How data is stored
Data is stored in a central RAPID database, hosted in a corporate cloud.
Data is replicated in real time to a separate test database, which is used for report development. This test database also acts as a disaster recover backup.
The data is stored using RAPID’S zero loss compression system, minimizing the database footprint without compromising on data integrity or performance.
Data is stored currently for 5 years, but is likely to increase as more users of the data extend their analytics to include longer time periods.
Wind Farm Application Architecture
Here is a diagram of how the application is structured: