Using LabVIEW and CompactRIO to Develop Monitoring Technologies Based on Synchrophasors Measurement for ‘Smart Metering’

Gallery

Figure 1: Synchrophasor Estimation: Reconstructed Signal s1(t) and Phase Redefinition

Author(s):
M. Paolone - UNIVERSITÀ DI BOLOGNA
A. Borghetti - UNIVERSITÀ DI BOLOGNA
C.A. Nucci - UNIVERSITÀ DI BOLOGNA

The evolution of power distribution networks from passive to active creates major changes in their operational procedures, especially monitoring networks in real time. We need to use advanced, smart monitoring tools to quickly and reliably estimate the real-time state of these networks. One of the most promising technologies in this field is distributed monitoring based on PMUs.

Synchrophasor estimation algorithms are based on the Discrete Fourier Transform (DFT) applied to quasi-steady state signals representing network node voltages and/or branch current waveforms. We can group these DFT-based algorithms into one-cycle DFT estimators and fractional-cycle DFT estimators performing recursive and nonrecursive updates. We created a DFT algorithm so we could use PMUs in active distribution networks and to keep the synchrophasor measurement accuracy within specific limits, even in the presence of distorted signal waveforms and electromechanical transients (namely, frequency-varying signals).

Compared to transmission networks, active distribution networks are characterized by reduced line lengths and limited power flows. When using bus voltage synchrophasors to estimate the state of the network, these characteristics result in very small phase differences between bus voltage phasors (generally in the order of tens of mrad or less). These characteristics call for PMU devices characterized by synchrophasor phase uncertainty well below the limits provided by IEEE C37.118. Distribution networks are characterized by distortion levels much higher than those of transmission networks. Additionally, active distribution networks are expected to operate even when islolated from the main transmission networks so PMUs are useful in supporting distribution system operators during the islanding and reconnecting maneuvers. In this respect, using a PMU to monitor electromechanical transients, generally characterized by non-negligible deviations from the rated network frequency, could lead to incorrect estimations of the synchrophasors phases and frequencies.

Synchrophasor Estimation Algorithm

Traditional synchrophasor estimation algorithms based on DFT generally perform the synchrophasor measurement directly from the output of the DFT applied to a signal typically sampled with a rate in the order of few kHz. Our algorithm on the other hand, though still based on DFT, makes use of a two-step approach in which the first step is a DFT analysis of the input signal and the second step is a time-domain analysis of the reconstructed time-domain signal corresponding to the fundamental frequency tone. The first step is unique because it uses the proposed algorithm to identify the fundamental frequency tone. Such an algorithm provides accurate results in the case of high sampling frequencies (for example, to 100 kHz). The following presents a brief description of the synchrophasor estimation algorithm.

The synchrophasor estimation algorithm involves the following three steps:
1. Sampling the three-phase voltages within a time window (T) of 80 ms (that is four cycles at 50 Hz), starting in correspondence to the UTC-GPS pulse-per-second (PPS) wavefront (typically 1 or 10 PPS).

2. Reconstructing the fundamental frequency tone as a sinusoidal signal characterized by a single frequency within a specific frequency window Δf (that is f₀ ± Δf, where f₀ indicates the rated value of the network frequency). We use the LabVIEW Real-Time Module and the CompactRIO real-time microcontroller to implement this step.

3. Estimating the synchrophasor amplitude, phase, and frequency with reference to the reconstructed fundamental frequency tone waveform. We implemented this step using the LabVIEW Real-Time Module and the CompactRIO real-time microcontroller.

Figure 1 summarizes the signal analysis obtained using the procedure described above. The dash-point line represents the generic distorted signal to estimate the synchrophasor, the continuous line represents the time-domain reconstructed fundamental frequency tone, and the dashed line represents the PPS signal.

PMU Prototype

We implemented the synchrophasor estimation algorithm on an NI CompactRIO embedded real-time microcontroller equipped with a 3M-gate FPGA. We sampled the voltage waveforms using an NI 9215 C Series module with a ±10 V dynamic signal input operating at a 100 kHz sampling frequency. The UTC-GPS time frame is provided by the S.E.A GPSIB Mobile Module characterized by a 100 ns time synchronization uncertainty. We used an NI 9401 Digital I/O module as a counter to perform a measurement between the PPS front rise (provided by the GPS device) and the first sample of the digitized waveforms. Figure 2 shows the structure of the PMU implementation.

As shown in Figure 2, the real-time microcontroller sends the FPGA the set point of the following quantities: sampling frequency fs; observation time window T; and number of GPS-PPS.

The FPGA forwards the number of PPS to the GPS device generating a PPS signal sent to both the NI 9215 and the NI 9401. These connections trigger the start of the waveform sampling in correspondence to the PPS front (for a duration corresponding to the observation time window T). At the same time, the PPS front sent to the NI 9401 module triggers the start of the FPGA counter running at the FPGA clock frequency, which for the adopted hardware, is equal to 40 MHz. This counter stops in correspondence with the first sample of sampled waveforms for the calculation (see Figure 1). The sampled data, as well as the GPS time tag, are inserted into a DMA-FIFO memory and retrieved by the real-time microcontroller to perform the synchrophasor estimation algorithm. The number of PPS corresponds to the number of synchrophasor estimations per second.

PMU Experimental Characterization and Conclusions

Experimental characterization refers to periodic signals characterized by spectrum components with constant frequency. We generated reference signals with an NI PXI chassis connected to an NI PXI arbitrary waveform generator, an NI PXI timing and synchronization module, an NI PXI high-accuracy data acquisition module, and an NI PXI high-performance  embedded controller. We analyzed two cases: single tone (50 Hz) and distorted signals. For distorted signals, we generated the reference signal with spectrum components equal to the limit values provided by the standard EN 50160. Table 1 summarizes the PMU uncertainties proving the characteristics of the developed device are compatible with the requirements of active power distribution networks applications.

A more extensive description of the experimental characterization is presented below and shows that the performances of the PMU prototype are not influenced by frequency-varying signals representing slow electromechanical transients.

Table 1: Mean Values and Standard Deviations of the Error Distributions of the PMU Prototype With Reference to Steady State Conditions: Single Tone and Distorted Signals

Author Information:
M. Paolone
UNIVERSITÀ DI BOLOGNA