Whitepaper: New ultra-fast, FPGA-based, floating-point tools for real-time power system simulation and advanced control

[BMac]

Problem Statement

As many of you have encountered, a significant obstacle to the successful design and deployment of new digital energy technologies for the smart grid has been a lack of suitable high speed real-time simulation tools that enable the type of comprehensive validation and verification of electric power control systems that is standard practice in other industries. Real-time hardware-in-the-loop (HIL) testing is necessary for the design of high reliability digital control systems because it enables complete test coverage and automation of testing, including conditions that would be difficult, impossible or dangerous to produce in the real world.

HIL testing works by connecting the embedded control system (unit under test) to a real-time simulator that mimics the behavior of the physical world. Since the physical world is simulated, it can represent virtually any test case scenario. This enables the control system to be tested thoroughly, systematically and comprehensively. HIL testing plays a major role in the reliability and robustness we typically associate with automotive and aerospace combustion engine control systems. In those industries, it would be unheard of to release an embedded control system to the market without the level of comprehensive validation and verification testing that is enabled by HIL simulation.

Unfortunately, the designers of switched mode power electronics, micro grid, smart grid and power distribution control systems have struggled to utilize HIL testing methods due to the high speed of the physical phenomenon associated with short distance electrical systems. Conventional processor-based HIL simulators are unable to digitally simulate fast (short distance) electrical phenomenon in real-time, due to ultra-fast (MHz speed) input-to-simulator-to-output timing requirements. Because of this, most smart grid, micro grid and electric motor drives control systems have not been validated with complete test coverage that is inclusive of extreme or rare conditions not easily created or replicated. (In the case of long distance (> 10 km) transmission grid simulations, processor-based HIL systems are capable of achieving the necessary 10-50 microsecond speeds and well known commercial solutions exist.)

At the same time, the fast, high bandwidth nature of electrical phenomenon necessitates advanced model-based control algorithms executing at high speed in cost effective embedded control hardware that’s rugged enough and lifetime supportable for long term deployment to the grid. Conventional processor and DSP based control systems typically must be augmented with custom circuitry and/or field programmable gate arrays (FPGAs) to meet the demanding timing requirements needed to implement advanced model-based, non-linear and highly custom control and pulse width modulation (PWM) algorithms. To achieve the necessary speeds on cost effective hardware often required full custom circuitry design, brittle and labor intensive floating to fixed-point math conversion and difficult to maintain register level text-based programming.

The grid interconnection of distributed energy resources (DERs), including intermittent renewable energy sources, is a key technology barrier holding back new waves of innovation in electric power industry. Despite exponential improvements in the performance-per-dollar (PPD) of solar photovoltaic and other renewable energy sources, power electronics transistors, battery energy storage technologies, real-time networks and embedded computing, a limiting factor in the desirable proliferation of these technologies is our ability to resolve the stability, power quality, and power flow issues associated with high penetration DERs.

The successful integration of DERs into the power grid is nothing less than a grand cyber-physical systems (CPS) design challenge of critical importance to energy security, economic growth and the worldwide response to climate change.

NI Commitment to Complete the Design V

National Instruments (NI) is engaged in sustained R&D outlay to provide a comprehensive Design V Toolchain, present status shown in the figure below, for next generation distributed grid energy control systems. The NI goal is to significantly improve the model based design and simulation, commercial deployment, product validation and verification, production testing, real-time networking, and remote asset management of DER CPS embedded systems. Significant progress towards these goals has been made in recent years. To enable an average cost savings of $950,000 USD in engineering development cost per embedded systems design (114 person-months), NI developed the NI sbRIO General Purpose Inverter Controller (GPIC) in collaboration with 20-30 industry design teams and NREL researchers. To enable over 10 fold productivity improvements in the model based design process, NI developed patented new co-simulation tools that enable power system and power electronics domain experts, rather than low-level hardware design engineers, to design advanced algorithms for heterogeneous FPGA hardware and deploy them to pre-validated embedded targets designed for commercial deployment using high level LabVIEW FPGA graphical programming tools. Nonetheless, significant continued investments and deep collaboration with researchers in industry, government and academia will be required to complete the Design V Toolchain for networked distributed power grid control systems.

Introducing new ultra-fast, heterogeneous FPGA-based, floating-point tools for real-time electric power system simulation and control

To address these challenges, National Instruments is pleased to introduce several new technologies for ultra-fast, heterogeneous FPGA-based, floating point simulation and control. A principal goal in developing these technologies is to reduce the cost, risk and development time necessary to bring grid tied power electronics and power system control products to commercial deployment in the clean energy, smart grid, micro grid and electric vehicle industries.

The new tools for ultra-fast, floating point, FPGA-based control and simulation have been developed for the control and real-time simulation of electric power systems, including switched mode power electronics. This includes classical linear model-based control solvers (transfer function and state-space) and floating point math blocks for building non-linear control & simulation solvers suitable for FPGA-based power system control and HIL simulation.

In addition to linear and non-linear solvers based on the new FPGA Floating Point Toolkit, an exciting new technology is in development that promises to automate the process of converting models from high level nodal analysis simulation environments such as NI Multisim to ultra high speed graphical floating point implementations for LabVIEW FPGA.

Key Enablers

     1. The availability of heterogeneous field programmable gate array (FPGA) hardware containing an array of integrated digital signal processor (DSP) cores capable of efficiently executing floating point math operations. These modern FPGAs are actually hybrid devices containing hard-core DSP processing elements, integrated within the reconfigurable computing fabric, and capable of achieving MHz speeds for simulation and control.

Figure 1. Heterogeneous FPGA system-on-a-chip (SOC) devices contain an array of programmable hardware logic, DSP cores, and communication elements that can now be programmed using graphical floating point math.

     2. A high level graphical programming tool chain that requires no knowledge of register level programming languages such as VHDL and Verilog (LabVIEW FPGA).

Figure 2. Using open, high level graphical programming tools, advanced algorithms can be automatically compiled to heterogeneous FPGA hardware like this floating point IP core for a 9x9 state space solver which executes at 1.9 MHz

     3. A new palette of floating point IP cores for LabVIEW FPGA available with resource optimized (control oriented) and speed optimized (simulation oriented) cores.

Figure 3. LabVIEW FPGA graphical floating point palette cores include arithmetic and transcendental math functions in resource optimized (ending in 00) and speed optimized (ending in SCTL) versions

     4. Technology to automatically convert electric circuit, power electronics and power system models to graphical floating point solvers for ultra-high speed heterogeneous FPGA hardware.

    

Figure 4. Automatic Multisim to LabVIEW FPGA conversion enables this power system model to be simulated at 2 MHz speeds using heterogeneous FPGA hardware

     5. Technology to automatically convert finite element analysis (FEA) results to high fidelity, ultra-high speed real-time electric motor and inverter drive simulations.

Figure 5. National Instruments has partnered with JSOL Corporation to enable the ability to export JMAG and JMAG-RT FEA models for PMSM and SRM machines to special high-fidelity, high speed solvers for heterogeneous FPGA hardware

It is hoped that the release of these new technologies and tools is sufficient to empower lead users in the developer community to begin taking an improved approach to power electronics and power system control design.

To learn more, please download the white paper PDF attachment below. 

Contents

Problem Statement    3

NI Commitment to Complete the Design V    4

Introducing new ultra-fast, heterogeneous FPGA-based, floating-point tools for real-time electric power system simulation and control    5

Key Enablers    5

FPGA-Based Electrical Motor Drive Simulation Toolkit    8

     High fidelity non-linear motor simulation using Finite Element Analysis (FEA) based solver    11

     Switched Reluctance Machines (SRMs)    12

     Signal Level Control System HIL Testing    13

     Full Power Motor Inverter ECU HIL testing (80 to 320 kW)    13

     Full Power Medium Voltage HIL Testing (up to 15 MW continuous)    14

     Subaru Case Study    15

Automatic MultiSim to FPGA Conversion for Ultra-High Speed Simulation    16

     Multisim2FPGA – Transmission Line with Breaker Switchgear    17

     Multisim2FPGA – Boost Converter    20

     Multisim2FPGA – Three Phase Inverter with Line Reactor Filter    21

     Multisim2FPGA – High Order Filter    22

     Multisim2FPGA – Induction Motor with 3-Phase Inverter    23

Floating Point Toolkit and FPGA-Based Solvers    26

     1. Intro to FPGA Floating Point Toolkit IP Library & Examples    26

     2. Multichannel Transfer Function Simulator    27

          Timing and resource benchmarks- resource optimized transfer function    32

          Timing and resource benchmarks- speed optimized transfer function    33

     3. Multichannel State-Space Solver    34

     4. AC Induction Motor Simulator (Linear/Nonlinear)    38

     5. GPIC Sensorless Magnetic Levitation (Nonlinear Control System)    43

          Pre-compile co-simulation with Multisim for FPGA Control Algorithm Development    46

Software Downloads and Installers    60

     Floating Point Toolkit and FPGA-Based Solvers (Transfer Function, State-Space, AC Induction Motor, Sensorless Maglev)    60

     Power Electronics Design V Training Workshop    60

     LabVIEW 2013 SP1 Platform DVDs    61

     NI Electric Motor Simulation Toolkit    62