Alstom Grid Develops High-Voltage Direct Current Transmission Control System Using Model-Based Design

“Using Model-Based Design we developed a complex control system in significantly less time than our traditional process would have required. We eliminated months of hand-coding by generating code from our models, and we used simulations to enable early design verification.”

Challenge

Accelerate control system development for high-voltage direct current voltage source converters

Solution

Use Model-Based Design to model, simulate, verify, and generate code and documentation for the control and protection systems

Results

  • Quantifiable process improvements
  • Rapid integration with power system simulation software
  • Protection systems implemented in one week
Alstom Grid’s HVDC demonstrator system with power converter modules. The improved controllability of the VSC in this system makes it well-suited for smart grid applications.

High-voltage, direct current (HVDC) electric power transmission systems offer several advantages over high-voltage, alternating current (HVAC) systems, including financial advantages when used across long distances or with underground or underwater cables. HVDC enables greater power density, improved power flow control, and more efficient use of energy sources than HVAC. Changing the voltage from AC to DC in an HVDC transmission, however, requires complex converter stations.

The line-commutated converters commonly used today require multiple filter banks, which can be expensive and quite large. For HVDC applications that need compact site layouts, such as offshore and onshore wind farms where space is limited, voltage source converters (VSCs) provide a better solution.

Engineers at Alstom Grid built a 24 megawatt demonstrator system to support development and testing of VSC technology, prepare for large-scale production, and enable potential customers to visit a fully functioning VSC facility. The team used Model-Based Design to accelerate development. “Model-Based Design enabled us to manage the complexity of the VSC system, verify our control design early in development, and meet our reliability, quality, and time-to-market targets,” says Anthony Totterdell, Deputy Control Systems Manager – VSC Expert at Alstom Grid.

Challenge

The demonstrator control system included a high-level sequencing layer running on a microprocessor that managed interaction with the power system, intermediate current and phase control layers running on DSPs, and low-level power electronic switch controls that monitor capacitor voltages every 100 microseconds.

“To develop such a complex system, our systems engineers needed to verify the design as early as possible,” says Totterdell. “In the past, we sometimes found design and implementation issues late in the project, when we were ready to test on a real-time simulator.”

With wind farm and smart grid demand increasing worldwide, Alstom Grid saw a market opportunity that would demand an aggressive production schedule. “Our goal was to get from concept to a complete demonstrator in 24 months,” explains Totterdell. “To meet this timeline, we needed to accelerate the software development process while minimizing the number of coding errors found late in the process.”

Solution

Alstom Grid used Model-Based Design with MATLAB® and Simulink® to model, simulate, document, and generate code for the HVDC VSC control system.

An Alstom Grid Senior Fellow developed the conceptual design with Simulink and Stateflow®. Using Simulink and Simscape Electrical™, he also built a plant model that included an AC grid connection, transformers, and loads, as well as insulated gate bipolar transistors (IGBTs) and capacitors for the lower-level power electronics submodules.

To verify the control design and plant model functionality, Alstom Grid engineers ran closed-loop simulations in Simulink. Next, they refined the control algorithms in preparation for deployment to a real-time target.

To help manage the system’s complexity, Alstom Grid engineers used Simulink model referencing to partition the model into library blocks that corresponded to the DSP hardware on which the components would be deployed. They added interfaces between blocks, converted the model to discrete mathematics, and switched from a variable-step to a fixed-step discrete solver.

Using Embedded Coder® the engineers generated C code functions to define sequencing, system-level control, and individual phase controls. They deployed the C code to the DSPs in a 1.2 kilowatt–scale hardware simulator to verify the system’s real-time operation.

Finally, the engineers ported the control algorithms to the production hardware for the 24 megawatt demonstrator and verified that it operated as designed.

The final system model comprised approximately 2,000 subsystems containing 33,000 blocks, 564 discrete states, and 250 Stateflow charts. From the controller part of this model, which included 2,000 blocks, Alstom Grid generated about 10,000 lines of production code. Using Simulink Report Generator™, they also generated more than 300 pages of system description and functional specification documentation.

The demonstrator system is currently in operation, and is receiving positive evaluations from prospective customers.

Results

  • Quantifiable process improvements. “Model-Based Design enabled us to make measurable improvements to our development process,” says Totterdell. “Time from design to code (including testing) dropped from approximately a year to 3 months, design iterations dropped from 2 months to less than 2 weeks, and documentation updates that used to take 2 weeks were done in minutes.”

  • Rapid integration with power system simulation software. “Our customers asked us to use the PSCAD/EMTDC environment for dynamic performance and transient analysis studies, which previously required rewriting our models in PSCAD and months of integration time,” says Totterdell. “Our experts worked with MathWorks consultants to reuse our existing MATLAB and Simulink models with Embedded Coder, enabling us to implement a change in functionality in about five minutes.”

  • Protection systems implemented in one week. “The protection algorithms for our conventional HVDC system took about six months to develop and test in C,” says Totterdell. “I re-implemented the same algorithms in Simulink and Stateflow and had them working in a single week.”