Archive Walker User Manual V1.0

User Manual:

Open the PDF directly: View PDF PDF.
Page Count: 48

DownloadArchive Walker User Manual V1.0
Open PDF In BrowserView PDF
PNNL-28016

Archive Walker Software
Setting Up and Reviewing Analyses with
the Archive Walker GUI
September 2018
J Follum
H Wang
P Etingov
F Tuffner
U Agrawal

Y Huang
D Kosterev (BPA)
S Yang (BPA)
G Matthews (BPA)
E Heredia (BPA)

Prepared for the U.S. Department of Energy
under Contract DE-AC05-76RL01830

DISCLAIMER
This report was prepared as an account of work sponsored by an agency of the
United States Government. Neither the United States Government nor any agency
thereof, nor Battelle Memorial Institute, nor any of their employees, makes any
warranty, express or implied, or assumes any legal liability or responsibility
for the accuracy, completeness, or usefulness of any information, apparatus,
product, or process disclosed, or represents that its use would not infringe
privately owned rights. Reference herein to any specific commercial product,
process, or service by trade name, trademark, manufacturer, or otherwise does not
necessarily constitute or imply its endorsement, recommendation, or favoring by
the United States Government or any agency thereof, or Battelle Memorial
Institute. The views and opinions of authors expressed herein do not necessarily
state or reflect those of the United States Government or any agency thereof.

PACIFIC NORTHWEST NATIONAL LABORATORY
operated by
BATTELLE
for the
UNITED STATES DEPARTMENT OF ENERGY
under Contract DE-AC05-76RL01830

Printed in the United States of America
Available to DOE and DOE contractors from the
Office of Scientific and Technical Information,
P.O. Box 62, Oak Ridge, TN 37831-0062;
ph: (865) 576-8401
fax: (865) 576-5728
email: reports@adonis.osti.gov
Available to the public from the National Technical Information Service
5301 Shawnee Rd., Alexandria, VA 22312
ph: (800) 553-NTIS (6847)
email: orders@ntis.gov 
Online ordering: http://www.ntis.gov

PNNL-28016

Archive Walker Software
Setting Up and Reviewing Analyses with the Archive Walker
GUI

September 2018

J Follum
H Wang
P Etingov
F Tuffner
U Agrawal

Y Huang
D Kosterev (BPA)
S Yang (BPA)
G Matthews (BPA)
E Heredia (BPA)

Prepared for
the U.S. Department of Energy
under Contract DE-AC05-76RL01830

Pacific Northwest National Laboratory
Richland, Washington 99352

Acknowledgments
The Archive Walker Software and associated functionality were primarily funded by the Bonneville
Power Administration through its Technology Innovations Program, with funding associated with
projects TIP-348, TIP-349, and TIP-350. Further improvements were funded by the U.S. Department of
Energy’s Grid Modernization Laboratory Consortium under project GM0072.

iii

Contents
Acknowledgments........................................................................................................................................ iii
1.0 Introduction .......................................................................................................................................... 1
2.0 Installing and Launching Archive Walker............................................................................................ 1
3.0 Project Panel ......................................................................................................................................... 3
4.0 Signal Selection Panel .......................................................................................................................... 3
5.0 Data Sources Tab .................................................................................................................................. 6
5.1 File Sources Panel ........................................................................................................................ 6
5.1.1 Example File Field ............................................................................................................ 6
5.1.2 File Directory Field ........................................................................................................... 6
5.1.3 File Type Field .................................................................................................................. 6
5.1.4 Mnemonic Field ................................................................................................................ 8
5.2 Mode Panel................................................................................................................................... 8
5.2.1 Mode Selection Dropdown................................................................................................ 8
5.2.2 Start Time Field................................................................................................................. 8
5.2.3 End Time Field .................................................................................................................. 8
5.2.4 Waiting Fields ................................................................................................................... 8
5.2.5 Hybrid Transition Field ..................................................................................................... 8
6.0 Data Quality and Customization Tab ................................................................................................... 9
6.1 Data Quality Filters ...................................................................................................................... 9
6.1.1 Status Flags ....................................................................................................................... 9
6.1.2 Zeros .................................................................................................................................. 9
6.1.3 Missing .............................................................................................................................. 9
6.1.4 Nominal Voltage ............................................................................................................... 9
6.1.5 Nominal Frequency ......................................................................................................... 10
6.1.6 Outlier ............................................................................................................................. 10
6.1.7 Stale Data ........................................................................................................................ 10
6.1.8 Data Frame ...................................................................................................................... 10
6.1.9 Channel ........................................................................................................................... 10
6.1.10 Entire PMU ..................................................................................................................... 11
6.1.11 Angle Wrapping .............................................................................................................. 11
6.2 Signal Customizations ................................................................................................................ 11
6.2.1 Scalar Repetition ............................................................................................................. 11
6.2.2 Addition........................................................................................................................... 11
6.2.3 Subtraction ...................................................................................................................... 11
6.2.4 Multiplication .................................................................................................................. 12
6.2.5 Division ........................................................................................................................... 12

v

6.2.6 Exponential...................................................................................................................... 12
6.2.7 Sign Reversal................................................................................................................... 12
6.2.8 Absolute Value ................................................................................................................ 12
6.2.9 Real Component .............................................................................................................. 12
6.2.10 Imaginary Component ..................................................................................................... 12
6.2.11 Angle Calculation............................................................................................................ 12
6.2.12 Complex Conjugate ......................................................................................................... 12
6.2.13 Phasor Creation ............................................................................................................... 13
6.2.14 Power Calculation ........................................................................................................... 13
6.2.15 Signal Type/Unit ............................................................................................................. 13
6.2.16 Metric Prefix Customization ........................................................................................... 13
6.2.17 Angle Conversion............................................................................................................ 13
7.0 Signal Processing Tab ........................................................................................................................ 13
7.1 Unwrap Angles........................................................................................................................... 13
7.2 Interpolation ............................................................................................................................... 13
7.3 Filtering and Multirate Processing ............................................................................................. 14
7.3.1 Filters............................................................................................................................... 14
7.3.2 Multirate Processing ........................................................................................................ 14
7.4 Wrap Angles............................................................................................................................... 15
7.5 Name, Type, and Unit Selection ................................................................................................ 15
8.0 Post-Processing Customization Tab ................................................................................................... 15
9.0 Detection Tab ..................................................................................................................................... 15
9.1 Detector Configuration............................................................................................................... 15
9.1.1 Periodogram Based Forced Oscillation Detector ............................................................ 15
9.1.2 Spectral Coherence Based Forced Oscillation Detector .................................................. 17
9.1.3 Out-of-Range Detector .................................................................................................... 18
9.1.4 Ringdown Detector ......................................................................................................... 19
9.1.5 Wind Ramp Detector ....................................................................................................... 19
9.1.6 Voltage Stability Detector ............................................................................................... 20
9.1.7 Mode Meter ..................................................................................................................... 21
9.2 Alarm Configuration .................................................................................................................. 23
9.2.1 Periodogram Based Forced Oscillation Alarming........................................................... 23
9.2.2 Spectral Coherence Based Forced Oscillation Alarming ................................................ 23
9.2.3 Ringdown Alarming ........................................................................................................ 24
10.0 Reviewing Results .............................................................................................................................. 24
10.1 Results Hierarchy ....................................................................................................................... 24
10.2 General Approach ...................................................................................................................... 25
10.3 Interacting with the GUI ............................................................................................................ 26
10.4 Examples .................................................................................................................................... 26

vi

10.4.1 Forced Oscillation Detectors ........................................................................................... 27
10.4.2 Out-of-Range Detector .................................................................................................... 28
10.4.3 Ringdown Detector ......................................................................................................... 30
10.4.4 Wind Ramp Detector ....................................................................................................... 32
10.4.5 Voltage Stability Detector ............................................................................................... 33
10.4.6 Mode Meter ..................................................................................................................... 35
11.0 Transferring the Location of Results and Data ................................................................................... 36

vii

1.0

Introduction

Archive Walker (AW) is a tool designed to ingest, analyze, and visualize PMU data and the power system
events that can be found within it. It is open source, with development funded by Bonneville Power
Administration (BPA) and the US Department of Energy (DOE). The goal of AW is to bring value to
archives of PMU data by enabling analysis with a set of built in event detectors and reducing the overhead
required to prototype and deploy new analysis methods.

2.0

Installing and Launching Archive Walker

The AW tool is compatible with the Windows operating system. It is composed of two components: an
underlying MATLAB engine and a user interface written in the .NET framework. These components rely
on the MATLAB runtime and the .NET framework, both of which can be downloaded for free. First,
download .NET framework 4.6.1 at https://www.microsoft.com/en-US/download/details.aspx?id=49982.
Next, download the R2017a (9.2) MATLAB runtime at
https://www.mathworks.com/products/compiler/matlab-runtime.html. Finally, transfer the executables
downloaded from these links to the computer that will host AW and run them to install the .NET
framework and MATLAB runtime.
Next, AW can be downloaded as a zip file at https://github.com/pnnl/archive_walker/releases. Once
downloaded, unzip the file by right clicking on it. The resulting folder contains the following:

•
•
•
•
•

dlls (folder)
ExampleData (folder)
Projects (folder)
BAWGUI.exe
BAWGUI.exe.config

There is no installation necessary. Simply double click the BAWGUI.exe file to launch AW. There will
be a brief delay the first time the tool is launched.
Examples of the GUI’s layout are presented in Figure 1 and Figure 2. The Project panel on the left side of
the screens will be discussed in Section 3. The Signal Selection panel visible on the right side of the
screen when the Settings page is selected will be described in Section 4. Sections 5-9 discuss the Settings
tabs used to configure AW, and Section 10 covers how to review results under the Results page.
The discussion in Section 10 utilizes the preloaded example data and results included in the download of
AW. These examples cover each of the detectors. The data is stored in the ExampleData folder and the
results are stored in the Projects folder. To access these examples, ensure that the specified results storage
location is the Projects folder (see Section 3 for instructions). When the GUI is opened for the first time,
the results path defaults to this folder.

1

Figure 1: Example GUI layout with Settings page selected.

Figure 2: Example GUI layout with Results page selected.
The final step in preparing to use AW is choosing where results should be stored. AW was designed to
allow detailed review of detection results while minimizing required disk space. Still, analyses of large
archives may require significant storage. The amount of storage required is dependent on factors such as
the number of PMUs in the dataset, the sampling rate, and the number of preprocessing steps, but it is on
the order of one GB for each day of data. The Projects folder mentioned earlier in this section serves as

2

the storage location for results the first time AW is opened. If this is undesirable, create a new folder
where you would like results to be stored. The path to this folder will be entered in the GUI before
beginning analysis.

3.0

Project Panel

The panel along the left side of the GUI (see the red box in Figure 3) allows the user to set up projects and
specify where results should be stored. The path at the top of the panel specifies where results will be
stored. Clicking the folder button below the path allows the user to select a different path for results. The
first time AW is opened, the Projects folder included in the zip file is selected as the results storage
location. This folder contains examples that are described in Section 10.4.

Figure 3: Project panel (left) and Settings page (right).
Results are organized into projects and tasks. A task is comprised of a set of results associated with a
specific analysis setup. Tasks are stored within projects for organizational clarity. After a project is added
by clicking the “Add Project” button, tasks can be created by right clicking on the project name (see the
green box in Figure 3). Right clicking also allows the entire project to be deleted. New tasks are also
created when the save button is clicked after the setup of an existing task is altered.
Once the setup for a task is complete, it can be applied to data by clicking the play button (green arrow).
While a task is running, a green arrow appears next to the task name. The remaining buttons allow a task
to be paused/resumed, stopped, or deleted.

4.0

Signal Selection Panel

The panel on the right side of the GUI, indicated by the red box in Figure 1, is used to select signals when
setting up analyses. The dropdown at the top of the panel provides options for how to select signals. The
options fall in three categories: 1) Initial set of signals contained in the data, 2) output signals from

3

previous tabs, and 3) inputs and outputs from steps on the current tab. For the first two categories, the
signals can be organized by PMU or by signal type. Signals are also organized by their sampling rate
because all signals at the input to any step in AW must have the same sampling rate. In the following, the
ringdown detector example, which is discussed in more detail in Section 10.4.3, will be used to illustrate
the functionality of the Signal Selection panel.
Examples of the three categories of signal selection are illustrated in Figures 4, 5, and 6. In Figure 4, the
first step of the analysis setup is to unwrap voltage angles included in the input data. The data is being
organized by signal type, rather than PMU, because only voltage angles are of interest. After unwrapping,
a filter is applied to the signals before passing them into a multirate step. When the same set of signals is
desired for multiple steps, it is convenient to use the Output Channels by Step option in the dropdown, as
in Figure 5. In the figure, the inputs to multirate processing can be selected easily by clicking either the
Step 1 or Step 2 checkboxes (note that the signals are identical). After signal processing, the signals are
ready to be passed to the ringdown detector. In Figure 6, these signals are selected by displaying outputs
from the Signal Processing Tab. At each step, the functionality of the Signal Selection Tab makes it easier
to locate the signals of interest, making it significantly easier to configure AW. Details on the setup
process are provided in the following sections.

Figure 4: Selecting voltage angle signals to be unwrapped from the input data using the Signal Selection
panel.

4

Figure 5: Selecting signals from previous steps for input to the multirate processor using the Signal
Selection panel.

Figure 6: Selecting signals from the Signal Processing Tab for analysis with the ringdown detector using
the Signal Selection panel.

5

5.0

Data Sources Tab

This tab allows the user to specify which files of PMU data should be examined.

5.1 File Sources Panel
The user specifies example files, from which the GUI automatically determines the locations, formatting,
and naming conventions of the data. Multiple file sources can be specified, allowing the user to, for
example, analyze their own data and data from partner utilities at the same time. All files from different
sources must be of the same type and have corresponding timestamps. However, different sources can
have different sampling rates.

5.1.1

Example File Field

AW must know which signals will be available for analysis, so an example file is required. The path to
the example file can be entered into the field manually, or the user can navigate to it using the folder
button next to the field (see Figure 3). Once the field is filled, clicking the Read File button will load the
PMU and signal names into the GUI for display while setting up analyses. It also automatically fills the
File Directory, File Type, and Mnemonic fields for the user’s review. These fields cannot be edited
directly, but their entries are described in Sections 5.1.2, 5.1.3, and 5.1.4 to provide additional information
about how data should be stored.
After a task is executed, it cannot be modified without saving it as a new task to prevent results from
being destroyed inadvertently. The exception is the example file. As detailed in Section 11, the example
file can be updated so that data and results can be shared by different users.

5.1.2

File Directory Field

Specifies the main folder containing a source of data files. It can have any name, but subfolders must
adhere to the following convention. Inside the main folder, PMU data is sorted by timestamp into folders
with names corresponding to the four-digit year of the timestamp. Within each year folder, PMU data is
sorted by timestamp into folders with names corresponding to the convention YYMMDD, where YY
corresponds to the two-digit year, MM corresponds to the two-digit month, and DD corresponds to the
two-digit day of the month. Individual data files are stored within the folders specifying the date. For
example, with a main folder path of C:\PMUdata, the path to the folder containing data files timestamped
at any time on November 2, 2014 would be
C:\PMUdata\2014\141102.

5.1.3

File Type Field

Specifies the format of the data. All file sources must use the same file type.
5.1.3.1
PDAT
A custom file type used by the Bonneville Power Administration (BPA).

6

5.1.3.2
JSIS CSV
A generic format for CSV files developed by the Western Electricity Coordinating Council (WECC) Joint
Synchronized Information Subcommittee (JSIS). Figure 7 contains the relevant section from the JSIS
document “WECC Guideline for Data Format Used in Engineering Analysis Applications of Disturbance
and Simulated Data.” The full list of signal types and units compatible with Archive Walker are listed in
Table 1.

Figure 7: Information on the JSIS-CSV format from the JSIS document “WECC Guideline for Data
Format Used in Engineering Analysis Applications of Disturbance and Simulated Data.”
Table 1. List of Signal Types and Units Compatible with Archive Walker.
Type
Description
Units
F
Bus frequency
Hz
VPM
Positive Sequence RMS voltage magnitude
kV
VPA
Positive Sequence voltage angle
DEG
IPM
Positive Sequence RMS current magnitude
Amps or kAmps
IPA
Positive Sequence current angle
DEG
P
Active power
MW
Q
Reactive power
MVAR

7

5.1.4

Mnemonic Field

Files are expected to adhere to the following convention:
MNEMONIC_YYYYMMDD_HHMMSS.csv(pdat)
where MNEMONIC is a set of characters excluding underscore, YYYYMMDD specifies the file’s start
date, and HHMMSS specifies the file’s start time. The MNEMONIC portion of the name should appear in
the Mnemonic field.

5.2 Mode Panel
The user specifies which data should be analyzed and how the tool should move through the files.

5.2.1

Mode Selection Dropdown

The user specifies the mode that the tool should operate in.
5.2.1.1
Archive
Runs from a user-specified range of times, immediately skipping over any missed files.
5.2.1.2
Real Time
Begins processing from the most recently available file (selected based on computer’s system clock in
UTC) and continues to process files as they become available. If a file is missing, the tool will wait a
user-specified amount of time for it to be written.
5.2.1.3
Hybrid
The tool begins in Archive mode before transitioning to Real Time mode.

5.2.2

Start Time Field

In Archive and Hybrid modes, specifies the timestamp of the first file to be processed.

5.2.3

End Time Field

In Archive mode, specifies the timestamp of the last file to be processed.

5.2.4

Waiting Fields

When Real Time or Hybrid mode is selected, a set of fields will appear allowing the user to specify how
the tool should handle data no longer being available (first sentence) and gaps in the data (second
sentence).

5.2.5

Hybrid Transition Field

When Hybrid mode is selected, a field will appear as part of a sentence to allow the user to specify when
the tool should transition from Archive mode to Real Time mode.

8

6.0

Data Quality and Customization Tab

The user specifies which data quality filters should be implemented and how custom signals should be
created from the base signals in the data files. Filters and customizations are added by clicking the ⊕
dropdown in the upper right corner of the panel.

6.1 Data Quality Filters
These filters allow the user to examine signals for data that is missing, corrupt, or otherwise
compromised. Later, in the Data Processing Tab, the user can specify how to interpolate through periods
of bad/missing data.

6.1.1

Status Flags

Filters the data based on status flags stored alongside the data in the processed files. Note that this is the
only data quality filter that operates on all signals from a particular PMU, rather than on specific signals.
Also note that JSIS-CSV files do not contain status flags.

6.1.2

Zeros

This filter flags missing data where zero was used as a placeholder.

6.1.3

Missing

This filter flags missing data that was not replaced by a placeholder.

6.1.4

Nominal Voltage

This filter flags voltage phasors based on the nominal value of the voltage magnitude. The user specifies
the nominal voltage along with upper and lower thresholds as multipliers. For example, if the nominal
voltage is set at 100 kV, the lower threshold is 0.7, and the upper threshold is 1.2, any voltage magnitude
measurements outside the range [70 kV, 120 kV] will be flagged. Things to note:

•
•

•
•

The filter accepts signals that are either voltage phasors (complex representation
combining magnitude and angle) or voltage magnitudes.
If a voltage magnitude is selected, the tool will attempt to flag the corresponding voltage
angle as well. This functionality depends on the naming convention used in PDAT files.
Thus, if JSIS-CSV files are being used, voltage magnitudes and angles should be
combined into phasors using customizations before applying this filter.
The nominal voltage should always be specified in kV. The tool automatically accounts
for signals with units of volts.
The tool does not account for line-neutral versus line-line orientations. Thus, the nominal
voltage must be specified to match the way measurements are reported.

9

6.1.5

Nominal Frequency

This filter flags frequency data based on the nominal system frequency. To allow short excursions that are
unrealistic for long time ranges, the filter employs two stages.
In the first stage, the entire file is considered. If frequency measurements fall outside the upper and lower
thresholds for more than the portion of the file specified by the Portion of File field, all frequency
measurements from the file are discarded.
In the second stage, individual samples that fall outside the specified upper and lower threshold are
discarded.

6.1.6

Outlier

This filter is intended to flag measurements that fall unrealistically far from surrounding measurements.
The standard deviation of the signal is calculated, and then measurements falling more than the standard
deviation times the multiplier specified in the Number of Standard Deviations field away from the
signal’s median are flagged. Things to note:

•
•

6.1.7

It is not appropriate for angle signals.
If a voltage or current magnitude is selected, the tool will attempt to flag the
corresponding angle as well. This functionality depends on the naming convention used
in PDAT files. Thus, if JSIS-CSV files are being used, the corresponding angle signals
may not be flagged.

Stale Data

This filter flags measurements that take on the same value for multiple samples, i.e., are stale. If
measurements remain unchanged for the number of samples specified by the Threshold field, they are
flagged.
Because stale frequency measurements often indicate that all measurements from the PMU are unreliable,
the user also has the option to flag all measurements from the PMU if the considered signal is a frequency
signal. This option should generally not be used with JSIS-CSV files containing measurements from
multiple PMUs.

6.1.8

Data Frame

If several signals from a PMU are found to be unreliable for a set of timestamps, all signals from the
PMU may be unreliable for those timestamps. This filter will flag all signals from the PMU at timestamps
where the percentage of signals in the PMU with flagged data exceeds the user-specified limit. This filter
should generally not be used with JSIS-CSV files containing measurements from multiple PMUs.

6.1.9

Channel

If many of a signal’s measurements are found to be unreliable, all of the signal’s measurements in the file
may be unreliable. This filter will flag all measurements in signals where the percentage of flagged
measurements in the signal exceeds the user-specified limit.

10

6.1.10

Entire PMU

If many of a PMU’s measurements are found to be unreliable, all of the PMU’s measurements in the file
may be unreliable. This filter will flag all measurements in PMUs where the percentage of flagged
measurements in the PMU exceeds the user-specified limit. This filter should generally not be used with
JSIS-CSV files containing measurements from multiple PMUs.

6.1.11

Angle Wrapping

Angle measurements from PMUs are typically wrapped to [-π, π] or [-180º, 180º]. In some JSIS-CSV
files, the wrapping may be performed improperly, leaving a value near zero in the middle of the wrap.
This filter identifies such occurrences by flagging successive angle jumps larger than the user-specified
threshold. Typically, 90º (π/2 radians) is a good choice for the threshold.

6.2 Signal Customizations
Signal customizations allow the user to manipulate the signals stored in the data files to create new
signals for analysis. These new signals are stored in “custom PMUs” which are structured just as the
original PMUs are. When parameterizing each of the customizations, the user is required to specify the
name of the custom PMU and signal. Custom signals of identical length created in separate steps can be
stored in the same custom PMU.
When multiple signals are involved in a customization, they must have the same sampling rate.
Customizations can again be performed after multirate processing (implemented on the Signal Processing
tab) is used to align sampling rates. This second round of customizations is implemented on the PostProcessing Customization tab.

6.2.1

Scalar Repetition

This customization allows creating a signal composed of a constant value. Such signals can be useful
within other customizations. The signal type and units should be chosen carefully to ensure compatibility
in later customization steps. For example, signals must have identical types and units to be added
together. The PMU for Time Source field specifies which PMU’s timestamps should be used for the
creation of the new signal.

6.2.2

Addition

This customization allows any number of signals with identical signal types and units to be added
together.

6.2.3

Subtraction

This customization allows one signal to be subtracted from another with identical signal type and unit.

11

6.2.4

Multiplication

This customization allows any number of signals to be multiplied together. If more than one non-scalar
signal is included, the custom signal type and unit are set to Other. Otherwise, the non-scalar signal’s type
and unit are retained.

6.2.5

Division

This customization allows one signal to be divided by another. If signal units agree, the result is a scalar.
If divisor is a scalar, the custom signal retains the type and unit of the dividend. Otherwise, the custom
signal’s type and unit are set to Other.

6.2.6

Exponential

This customization allows each measurement in a signal to be raised to an exponent.

6.2.7

Sign Reversal

This customization changes the sign of each measurement, e.g., positive to negative and negative to
positive.

6.2.8

Absolute Value

This customization takes the absolute value of each input signal and returns corresponding custom
signals.

6.2.9

Real Component

This customization takes the real component of each input signal and returns corresponding custom
signals.

6.2.10

Imaginary Component

This customization takes the imaginary component of each input signal and returns corresponding custom
signals.

6.2.11

Angle Calculation

This customization takes the angle of each input signal (assumed to be complex) and returns
corresponding custom signals with units of radians. Input signals with a phasor signal type produce
custom signals with corresponding angles, e.g., inputting a phase B voltage phasor results in a phase B
voltage angle type for the custom signal.

6.2.12

Complex Conjugate

This customization takes the complex conjugate of each input signal and returns corresponding custom
signals.

12

6.2.13

Phasor Creation

This customization combines input magnitude and angle signals and produces custom signals in phasor
representation. Signal types for magnitudes and angles must be compatible, i.e., current magnitudes with
current angles and voltage magnitudes with voltage angles. The resulting custom signals are assigned
types and units based on the inputs.

6.2.14

Power Calculation

This customization allows active, reactive, complex, and apparent power to be calculated from input
voltages and currents. The inputs can be specified as either phasors or magnitude/angle pairs. The custom
signal type and unit are assigned based on the units of the inputs.

6.2.15

Signal Type/Unit

This customization allows the type and unit of a signal to be specified. It makes no change to the input
signal’s measurements. This customization is useful for ensuring a signal’s compatibility with other
signals in customizations or for certain detectors.

6.2.16

Metric Prefix Customization

This customization changes the metric prefix for signals and adjusts the signal’s scaling appropriately.

6.2.17

Angle Conversion

This customization changes the units of angle signals either from radians to degrees or from degrees to
radians, depending on the input signal. The units of the custom signals are selected appropriately.

7.0

Signal Processing Tab

On this tab, the user can implement a variety of signal processing functions to prepare the data for
analysis by detectors.

7.1 Unwrap Angles

Angle measurements from PMUs are typically wrapped to [-π, π] or [-180º, 180º]. The jumps introduced
by wrapping prevent proper application of filters and bad/missing data interpolation. Angle signals listed
in this section are unwrapped to allow filters and interpolation to be performed.

7.2 Interpolation
To replace measurements that are flagged as bad/missing, interpolation can be applied to each signal. If
Constant is selected in the Type field, corrupt measurements are replaced by the last good value. If Linear
is selected, corrupt measurements are replaced by selecting values along the line connecting the good
value preceding the corrupt measurements to the good value following the corrupt measurements. The
Limit field specifies the maximum number of samples that can be interpolated. Sections of flagged data
longer than this limit are not replaced and are excluded from further analysis.

13

7.3 Filtering and Multirate Processing
In this section, filters are applied to signals to remove unwanted frequency content and multirate
processing is applied to change the sampling rates of signals.

7.3.1

Filters

Filters can be specified based on their coefficients, or the user can specify parameters and the tool will
automatically design the filter. The Filter Type dropdown offers the following options.
7.3.1.1
Rational
The user specifies the numerator and denominator coefficients of an IIR filter (FIR if the denominator
coefficient is one). Each coefficient should be separated by a comma. Note that Rational refers to how the
filter is specified; the high-pass and low-pass filters described in the following sections are also rational
filters.
7.3.1.2
High-Pass
The user specifies the order and cutoff frequency for a high-pass Butterworth filter designed by the tool.
7.3.1.3
Low-Pass
The user specifies the passband ripple as a positive dB value, stopband ripple as a positive dB value,
passband cutoff frequency in Hz, and stopband cutoff frequency in Hz of a low-pass Parks-McClellan
equiripple FIR filter designed by the tool.

7.3.2

Multirate Processing

Multirate processing is used to adjust the sampling rate of signals to improve or speed up analysis and to
make signals at different sampling rates compatible.
Similar to the custom PMUs created during customization steps, multirate PMUs are created to store the
outputs of multirate processing steps. A unique name must be specified for each different multirate
processing step.
The user can choose to specify the new sampling rate in the following ways.
7.3.2.1
Specify New Sampling Rate
The desired sampling rate, which must be an integer multiple or factor of the current sampling rate, is
specified in units of samples per second. An appropriate low-pass filter is applied as part of the
processing.
7.3.2.2
Specify Upsampling/Downsampling Factors
If only the upsampling factor is specified, then the signal is expanded (zeros placed between each
sample). If only the downsampling factor is specified, then samples are removed. In these cases, no
filtering is applied. If both upsampling and downsampling factors are specified, then the signal is
expanded, appropriately low-pass filtered, and then downsampled.

14

7.4 Wrap Angles
If desired, angle measurements unwrapped in the Unwrap Angles section can be wrapped again in this
section.

7.5 Name, Type, and Unit Selection
Processing may effectively change the type and units of a signal, e.g., high-pass filtered voltage angle
signals are closely related to frequency. Further, some of the tool’s detectors allow only specific signal
types to be analyzed. Thus, this section allows the user to easily change the type and unit of several
signals, rather than one at a time.
Alternatively, the name of a single signal can be changed. This can improve clarity. For example, suppose
a voltage angle signal named Bus1.ANG is high-pass filtered and becomes closely related to frequency. It
may improve clarity for the user if the signal is renamed Bus1.DerivedFrequency. When a single signal is
considered, the type and units can also be changed.

8.0

Post-Processing Customization Tab

After signal processing has been completed, further signal customizations can be completed. This is
particularly useful for combining signals that were initially at different sampling rates. All of the
customizations are implemented as described in the Data Quality and Customization Tab section. Note
that only customizations are available; all data quality checks are completed before the signal processing
steps.

9.0

Detection Tab

On this tab, detectors are configured to examine data for events. For some detectors, alarms to highlight
the most important events are also parameterized.

9.1 Detector Configuration
In this section, individual detectors are configured. Fields that can be left empty have an initial value of
Optional. If these fields are not populated, default values will be selected. The following sections describe
each of the detectors and their parameters. Along with the detectors included in the open-source release of
AW, a modified version of the tool also includes the voltage stability detector and mode meter described
in Sections 9.1.6 and 9.1.7.

9.1.1

Periodogram Based Forced Oscillation Detector

The periodogram based forced oscillation detector operates on relatively long record lengths (greater than
10 minutes), is very sensitive, and produces highly accurate estimates of forced oscillation frequencies.
For details see:
J. Follum, F. Tuffner and U. Agrawal, "Applications of a new nonparametric estimator of ambient
power system spectra for measurements containing forced oscillations," 2017 IEEE Power &
Energy Society General Meeting, Chicago, IL, 2017, pp. 1-5.

15

This detector can only be applied to signals of the following types: real power, reactive power, apparent
power, frequency, and OTHER.
Parameters are as follows.
9.1.1.1
Mode
Signals listed for the detector can be processed individually or simultaneously. When operated in
multichannel mode, the detector is more sensitive to system-wide events and less sensitive to local events.
9.1.1.2
Analysis Length
The total amount of data that is to be examined. Typical values range between 600 and 1800 seconds.
This is the parameter K in the supplied reference.
9.1.1.3
Window Type
Different windows applied to the data will provide slightly different detection performance.
9.1.1.4
Frequency Interval
Specifies the distance between evaluated frequencies. This interval should be no larger than the inverse of
the analysis length.
9.1.1.5
Window Length
The amount of data in overlapping windows used to estimate the ambient spectrum. This is the parameter
M in the supplied reference.
9.1.1.6
Window Overlap
The amount of overlap between windows when the ambient spectrum is estimated. A typical selection is
one half the window length.
9.1.1.7
Median Filter Width
The frequency range corresponding to the order of the frequency-domain median filter used while
estimating the ambient spectrum. Using the median filter order Nm from the supplied reference, this is Nm1 multiplied by the frequency interval.
9.1.1.8
Probability of False Detection
The probability that a detection will occur even though a forced oscillation is not present. The probability
can only be reduced at cost to the related probability that an oscillation will be detected when it is present.
This probability is expected as a value between zero and one, with typical values close to zero.
9.1.1.9
Minimum Frequency
The lower end of the frequency range to be examined for forced oscillations.
9.1.1.10 Maximum Frequency
The upper end of the frequency range to be examined for forced oscillations. The selected value cannot
exceed one-half of the sampling rate.

16

9.1.1.11 Frequency Tolerance
Forced oscillations must be separated in frequency by at least this amount to be considered separate
events.

9.1.2

Spectral Coherence Based Forced Oscillation Detector

The spectral coherence based forced oscillation detector operates on relatively short record lengths
(approximately one to two minutes), allowing it to detect large oscillations quickly. For details see:
J. Follum, “Detection of Forced Oscillations in Power Systems with Multichannel Methods,”
PNNL-24681, Pacific Northwest National Laboratory, Richland, WA, 2015,
which can be found using the search tool at https://www.pnnl.gov/publications/. This detector can only be
applied to signals of the following types: real power, reactive power, apparent power, frequency, and
OTHER.
Parameters are as follows.
9.1.2.1
Mode
Signals listed for the detector can be processed individually or simultaneously. When operated in
multichannel mode, the detector is more sensitive to system-wide events and less sensitive to local events.
9.1.2.2
Analysis Length
The length of overlapping segments used in the spectral coherence calculation. Typical values are in the
neighborhood of 60 seconds.
9.1.2.3
Delay
The amount of delay between the overlapping segments used in the spectral coherence calculation.
Typical values are in the neighborhood of 10 seconds. This is the parameter Δ𝑛𝑛 in the supplied reference.

9.1.2.4
Number of Delays
The number of overlapping segments included in the spectral coherence calculation. The minimum value
is two. This is parameter D+1 in the supplied reference.
9.1.2.5
Threshold Scale
This parameter controls detection performance. Lower values will increase sensitivity but will increase
false detections. Larger values require stronger evidence to trigger a detection but will decrease false
detections. The lowest acceptable value is one, but many false detections will occur for this setting. A
typical selection is three. This is parameter c in the supplied reference.
9.1.2.6
Window Type
Different windows applied to the data will provide slightly different detection performance.

9.1.2.7
Frequency Interval
Specifies the distance between evaluated frequencies. This interval should be no larger than the inverse of
the window length.

17

9.1.2.8
Window Length
The amount of data in overlapping windows used to calculate the spectral coherence. Typically, a value
several times smaller than the analysis length should be selected.
9.1.2.9
Window Overlap
The amount of overlap between windows when the spectral coherence is estimated. A typical selection is
one half the window length.
9.1.2.10 Minimum Frequency
The lower end of the frequency range to be examined for forced oscillations.
9.1.2.11 Maximum Frequency
The upper end of the frequency range to be examined for forced oscillations. The selected value cannot
exceed one-half of the sampling rate.
9.1.2.12 Frequency Tolerance
Forced oscillations must be separated in frequency by at least this amount to be considered separate
events.

9.1.3

Out-of-Range Detector

This detector operates by comparing incoming data to a set of thresholds. The thresholds are referenced
from a baseline specified as a nominal value by the user or by averaging over a window. The detector has
two stages. In the first stage, detection occurs if the signal remains outside of the thresholds for a certain
duration. In the second stage, detection occurs if the signal exceeds its thresholds while simultaneously
exceeding a rate-of-change threshold.
Parameters are as follows.
9.1.3.1
Nominal Value
The nominal value of all included signals. Serves as the reference for detection thresholds.
9.1.3.2
History for Baseline
The amount of data used to establish a baseline. The resulting average serves as the reference for
detection thresholds.
9.1.3.3
Window Length
The amount of data considered in the duration based detector.
9.1.3.4
Duration Threshold
The amount of data that must exceed thresholds to trigger the duration based detector.
9.1.3.5
Rate of Change Threshold
The amount of change per second that must be achieved for detection to occur with the rate-of-change
based detector.

18

9.1.3.6
Positive Deviation
How far above the baseline value the upper threshold resides. Specified separately for each stage of the
detector as a positive value.
9.1.3.7
Negative Deviation
How far below the baseline value the lower threshold resides. Specified separately for each stage of the
detector as a negative value.

9.1.4

Ringdown Detector

This detector is based on the root mean square (RMS) energy in a signal. It operates by comparing the
RMS energy over recent data to a baseline established from a longer history of data. This detector can
only be applied to signals of the following types: real power, frequency, and OTHER.
Parameters are as follows.
9.1.4.1
Analysis Length
The amount of data to use in the calculation of the RMS energy. This value should approximate the length
of ringdowns.
9.1.4.2
History for Baseline
The amount of data used to establish a baseline. Typical values are in the neighborhood of 120 seconds.
9.1.4.3
Threshold Scale
This parameter controls detection performance. Lower values will increase sensitivity but will increase
false detections. Larger values require stronger evidence to trigger a detection but will decrease false
detections. The lowest acceptable value is one, but many false detections will occur for this setting. A
typical selection is three.

9.1.5

Wind Ramp Detector

This detector examines data (typically conceived as measurements of power production from wind farms)
for evidence of ramping. To do this, the tool applies low pass filters to the measurements to capture
trends. The user can select from two filters included with the tool, one for short trends on the order of
minutes and one for long trends on the order of hours. This detector can only be applied to signals of the
following types: real power, reactive power, and OTHER. The default values for the wind ramp detector
parameters are guidelines for application to real power measurements.
Parameters are as follows.
9.1.5.1
Minimum Ramp Size
The smallest an upward or downward trend in the signal may be in order to trigger a detection.
9.1.5.2
Maximum Ramp Size
The largest an upward or downward trend in the signal may be in order to trigger a detection.

19

9.1.5.3
Minimum Ramp Duration
The shortest a trend in the signal may be in order to trigger a detection.
9.1.5.4
Maximum Ramp Duration
The longest a trend in the signal may be in order to trigger a detection.

9.1.6

Voltage Stability Detector

This detector seeks to identify oncoming local voltage instability using approaches based on representing
the surrounding power system with a Thevenin equivalent. Five algorithms are available for calculating
the Thevenin equivalent. Based on the resulting Thevenin impedance and the bus’s load impedance a
stability metric is calculated. The AW tool detects when the metric, which typically varies between 0 and
1, falls below a user-defined threshold.
When setting up a voltage stability detector, checkboxes are used to indicate which of the Thevenin
estimation algorithms are to be implemented. Some of the algorithms operate on a window of data. The
length of these windows are specified in the Analysis Length fields. After methods are selected, sites can
be configured. A site typically corresponds to a single substation. However, if multiple substations are
closely linked, they can be considered together as a single site. In the parameter names that follow,
branch refers to a line sourcing or sinking power for the site, while shunt refers to a line acting
exclusively as a sink, such as a reactor bank. The full list of parameters is as follows.
9.1.6.1
Name
A name used to identify the site when results are displayed.
9.1.6.2
Stability Threshold
A threshold for the stability metric between zero and one. If the metric falls below this value, an event
will be triggered.
9.1.6.3
Voltage Bus Magnitude
The signal supplying measurements of the site’s bus voltage magnitude. Measurements must be for the
positive sequence component with units of Volts measured line-neutral. If multiple entries are specified,
the voltage bus will be represented as their average.
9.1.6.4
Voltage Bus Angle
The signal supplying measurements of the site’s bus voltage angle. Measurements must be for the
positive sequence component with units of degrees measured line-neutral. If multiple entries are
specified, the voltage bus will be represented as their average.
9.1.6.5
Branch (Shunt) Active Power
The signal supplying measurements of active power flowing on a branch (shunt) connected to the site.
Units must be three-phase MW. If omitted, the active power is calculated based on the supplied voltage
bus and branch (shunt) current measurements.

20

9.1.6.6
Branch (Shunt) Reactive Power
The signal supplying measurements of reactive power flowing on a branch (shunt) connected to the site.
Units must be three-phase MVAR. If omitted, the reactive power is calculated based on the supplied
voltage bus and branch (shunt) current measurements.
9.1.6.7
Branch (Shunt) Current Magnitude
The signal supplying measurements of the magnitude of current flowing on a branch (shunt) connected to
the site. Measurements must be for the positive sequence component with units of Amps.
9.1.6.8
Branch (Shunt) Current Angle
The signal supplying measurements of the angle of current flowing on a branch (shunt) connected to the
site. Measurements must be for the positive sequence component with units of degrees.

9.1.7

Mode Meter

The mode meter tool in AW is intended to be used as a data generator for baselining studies, rather than
as an actual detector. Along with estimates of electromechanical modes, it captures system conditions as a
set of measurements. Together, this data can be used to better understand how system conditions impact
the modes. A mode meter can associate estimates of several different modes with a set of system
condition signals. A separate mode meter implementation is needed for each set of system condition
signals to be tracked. Descriptions of the parameters for the mode meter tool follow.
9.1.7.1
Mode Meter Name
The name of the mode meter to be configured. This entry is used to distinguish the mode meter in the
results.
9.1.7.2
Oscillation Baselining Signals
The list of measurements that comprise the system conditions. The value of these measurements are
captured each time the modes are estimated so that a baseline can be established.
9.1.7.3
Mode of Interest
Name of the mode that is being estimated. This entry is used to distinguish the mode in the results.
9.1.7.4
Minimum Frequency
The lowest acceptable frequency estimate for the mode of interest.
9.1.7.5
Nominal Frequency
An estimate of the mode’s frequency used in selecting a mode estimate from the set of pole estimates
produced by the mode meter algorithms.
9.1.7.6
Maximum Frequency
The maximum acceptable frequency estimate for the mode of interest.
9.1.7.7
Maximum Damping Ratio
The maximum acceptable damping ratio estimate for the mode of interest.

21

9.1.7.8
Signal Selection
The signals that are selected specific to each mode are the signals used to estimate the mode.
9.1.7.9
Analysis Length
The amount of data used to generate mode estimates.
9.1.7.10 Retroactive Continuity Status
To obtain a mode estimate, it must be selected from a set of pole estimates produced by the algorithms.
When retroactive continuity is on, AW will return to time periods where more than one pole estimate fell
in the specified frequency range. It then makes adjustments where it appears that the wrong pole estimate
was selected.
9.1.7.11 Retroactive Continuity Maximum Length
To ensure that the retroactive continuity functionality does not alter estimates from long periods where
mode estimates remained stable, a maximum length for the time periods that qualify to be adjusted is
specified.
9.1.7.12 Method
This dropdown allows the user to select a mode meter algorithm. Here YW stands for Yule Walker, LS
stands for Least Squares, ARMA stands for the AutoRegressive Moving Average model, and +S indicates
that detected forced oscillations are incorporated into the model as sinusoids.
9.1.7.13 AR Model Order
Model order for the autoregressive (AR) portion of the ARMA model used to estimate the modes.
9.1.7.14 MA Model Order
Model order for the moving average (MA) portion of the ARMA model used to estimate the modes.
9.1.7.15 Exaggerated AR Model Order
The model order of the high-order autoregressive (AR) model used in the first stage of least squares
algorithms.
9.1.7.16 Number of Equations
The number of equations to use in the Yule-Walker algorithms.
9.1.7.17 Number of Equations with FO Present
The number of equations to use in the YW-ARMA+S algorithm when a forced oscillation (FO) is
detected.
9.1.7.18 Forced Oscillation Detection Parameters
For the LS-ARMA+S and YW-ARMA+S algorithms, the periodogram-based forced oscillation detector
is implemented. See Section 9.1.1 for descriptions of the parameters.

22

9.2 Alarm Configuration
To help ensure that the most significant forced oscillation results are displayed prominently, parameters
can be selected to set alarm flags associated with each detected event. The ringdown detector also has one
parameter intended to limit displayed results to true ringdowns.

9.2.1

Periodogram Based Forced Oscillation Alarming

Alarms for forced oscillations detected with the periodogram algorithm are based on the duration and
signal-to-noise ratio (SNR) of the oscillation, as depicted in Figure 8. Alarm flags are set for forced
oscillations with SNR/duration pairs exceeding the threshold in red. The user selects the five parameters
in the figure.

Figure 8: Alarm configuration diagram for periodogram based forced oscillation detection.

9.2.2

Spectral Coherence Based Forced Oscillation Alarming

Alarms for forced oscillations detected with the spectral coherence algorithm are based on the duration
and coherence of the oscillation, as depicted in the figure below. Alarm flags are set for forced
oscillations with coherence/duration pairs exceeding the threshold in red. The user selects the five
parameters in the figure. All Coherence values should range from zero to one.

23

Figure 9: Alarm configuration diagram for spectral coherence based forced oscillation detection.

9.2.3

Ringdown Alarming

The ringdown detector is based on signal energy and will detect any event of sufficient energy, even if it
is too long to realistically be a ringdown. Thus, the user may set a maximum duration for detected
ringdowns. High-energy events of longer duration will be discarded.

10.0 Reviewing Results
This section covers all of the tabs under the Results page. Before the appearance of results in the GUI is
described, it will be useful to provide background on how results are stored in Section 10.1 and a general
approach to reviewing results that follows naturally in Section 10.2. An overview of methods for
interacting with the GUI is provided in Section 10.3. With this foundation, specific examples are provided
in Section 10.4 for each detector using the examples included with the download of AW. Note that the
mode meter is designed as a data generator, rather than an actual detector, so it is an exception to much of
what follows. Details are provided in Section 10.4.6).

10.1 Results Hierarchy
The goal of the storage approach implemented within AW is to allow detailed review of a detector’s
behavior while minimizing the amount of storage required. To accomplish objective, results are stored in
a hierarchy, which is represented in Figure 10. At the top level, a set of XML files store high-level

24

information about individual events, such as start time, extreme value, and number of channels that the
event was detected in. These results require very little disk space, but they contain a very limited amount
of information. At the next level of the hierarchy, extrema (maximum and minimum values) of analyzed
signals are stored at regular intervals. These intervals are much longer than PMU reporting rates to keep
required disk space to a minimum and to allow fast loading into the GUI. Sudden deviations and long
trends in signals can be identified and compared with the high-level event information stored in XMLs.
At the final level of the hierarchy, information needed to perfectly recreate an analysis is stored at regular
intervals. Using this information, detector performance and analyzed signals can be reviewed for specific
intervals of time. This hierarchy lends itself to the general approach to reviewing results that was
implemented across all detectors. It is described in the following section.

Figure 10: Diagram of the results storage hierarchy.

10.2 General Approach
Though the results tabs for each detector are unique, they utilize aspects of the results hierarchy in similar
ways. For the time period of interest, high-level event information is presented in tables. These tables are
automatically updated based on the start- and end-time fields at the top of the tabs. They list when the
event occurred, which signals it was detected in, and provide an indication of severity specific to each
detector. These tables can be used to identify which events are of highest interest.
Once the tables have been used to identify a general time period of interest, the Overview button can be
used to retrieve signal extrema. The resulting plots also indicate where events were detected, providing

25

the user with an overview of what transpired. From these overview plots, specific time periods for more
detailed analysis can be selected.
After zooming in on the overview plots, clicking the Retrieve Detail button will rerun the detectors. While
the analysis is rerunning, a green arrow will appear next to the task name. The time required to rerun is
approximately the same as the time the analysis took when it was initially conducted, so it is important to
choose periods for detailed analysis wisely. In many cases, a meaningful rerun can be conducted in
several seconds. If the rerun requires too much time, clicking the Cancel button will end the rerun and
display the detailed results that were retrieved. Once it completes, plots are generated to display the
analyzed signals and details about each detector’s operation. These can be used to gain insight into the
event and check detector settings.
The general approach holds for each of the detectors, though not all of them use each level of the results
hierarchy. Details are provided in the following section.

10.3 Interacting with the GUI
The tables and plots used in the AW GUI are interactive and can be manipulated in several ways. The best
way to learn these capabilities is through experience with the tool, but a list of common features and
helpful hints is provided here to help the user get started.

•
•
•
•
•
•
•
•

Plots can be zoomed by scrolling the mouse wheel while hovering over the plot or its
vertical or horizontal axes
Holding the left mouse button while dragging the cursor will zoom to the highlighted
region of a plot
To pan within a plot, hold down the right mouse button while moving the cursor
To reset the range of a plot, press a on the keyboard
After clicking on a plot, it can be copied to the clipboard by pressing Ctrl+c on the
keyboard
Additional information about a trace in a plot can be obtained by hovering over the trace
or holding down the left mouse button while pointing at it
Entries in tables can be selected to display additional information in associated plots or
tables
Sections in the GUI can be collapsed by clicking the
symbols

10.4 Examples
This section discusses navigation of results for each detector using the preloaded example data and results
included in the AW download (see Section 2). The data is stored in the ExampleData folder and the
results are stored in the Projects folder. To access these examples, ensure that the specified results storage
location is the Projects folder (see Section 3 for instructions). When the GUI is opened for the first time,
the results path defaults to this folder. The following subsections assume that this step has already been
completed.
Because AW reruns analyses to provide detailed results, the user must input the path to the proper
example file for each task. Instructions will be provided in each of the following subsections to navigate
this procedure. Section 11 provides further explanation.

26

10.4.1

Forced Oscillation Detectors

In the Project panel, select the ForcedOscillation task. If this step has not been completed previously, the
GUI will produce a dialogue box indicating that the example file cannot be found. Navigate to the Data
Source tab on the Settings page. In the Example File field, enter the full path to
*\ExampleData\ForcedOscillation\2017\170203\ ExData_20170203_000000.csv
The ExampleData folder was included in the download of AW (see Section 2). Note that if you navigate
to the example file using the GUI, you will need to select JSIS_CSV files in the lower right hand corner of
the File Explorer window to see the files. Next, click the Read File button to load the settings and results.
Finally, navigate to the Forced Oscillations tab on the Results page.
The GUI automatically selects the starting time range as the most recent day with an event, in this case
February 3, 2017, and will appear as in Figure 11. The plot in the upper left displays the frequencies (yaxis) and time ranges (x-axis) of detected oscillations. The start- and end-points of oscillations that
triggered an alarm are highlighted with red dots. Results are also presented in tables. The table below the
plot lists oscillations by frequency. Clicking on an entry in the table updates the table in the upper right,
which displays individual occurrences. Clicking on an occurrence entry highlights the occurrence in the
plot (note the thicker blue line in the plot) and lists the signals that were involved in the occurrence in the
table in the lower right of the screen.

27

Figure 11: GUI display after loading results for the forced oscillation detector.

10.4.2

Out-of-Range Detector

In the Project panel, select the OutOfRange task. If this step has not been completed previously, the GUI
will produce a dialogue box indicating that the example file cannot be found. Navigate to the Data Source
tab on the Settings page. In the Example File field, enter the full path to
*\ExampleData\OutOfRange\2017\170425\ ExData_20170425_000000.csv
The ExampleData folder was included in the download of AW (see Section 2). Note that if you navigate
to the example file using the GUI, you will need to select JSIS_CSV files in the lower right hand corner of
the File Explorer window to see the files. Next, click the Read File button to load the settings and results.
Finally, navigate to the Out of Range Events tab on the Results page.
The GUI automatically selects the starting time range as the most recent day with an event, in this case
April 25, 2017. The upper table lists events, while the lower table lists each of the signals that the event
was detected in. In this case, two events were detected. Click the Overview button to load the extrema for
the analyzed signals. Once done, the GUI will appear as in Figure 12, with the overview plot to the left.
Note that even with this course data, distinguishing characteristics of the events can be observed. The first
event involves a sudden decrease in frequency, while the second was captured because of a sudden

28

increase. In the overview plot, the event selected in the upper table is highlighted. To select a different
event, simply click it in the table.
To obtain detailed information about the second event, begin by zooming in on it, as in Figure 13. Note
that the table of events automatically updates to contain only events within the selected time period.
Clicking the Retrieve Detail button reruns the analysis and creates additional plots, as displayed in Figure
13. The top plot in the Detection Details section displays all of the analyzed signals. Below it, thumbnails
are provided for each of these signals. Clicking on the thumbnail will display three plots: the signal with
detection limits, performance of the “duration” stage of the detector, and performance of the “out-ofrange” stage of the detector (see Section 9.1.3).

Figure 12: GUI display after loading overview results for the out-of-range detector.

29

Figure 13: GUI display after retrieving detailed information about an out-of-range event.

10.4.3

Ringdown Detector

In the Project panel, select the Ringdown task. If this step has not been completed previously, the GUI
will produce a dialogue box indicating that the example file cannot be found. Navigate to the Data Source
tab on the Settings page. In the Example File field, enter the full path to

30

*\ExampleData\Ringdown\2017\170523\ExData_20170523_000000.csv
The ExampleData folder was included in the download of AW (see Section 2). Note that if you navigate
to the example file using the GUI, you will need to select JSIS_CSV files in the lower right hand corner of
the File Explorer window to see the files. Next, click the Read File button to load the settings and results.
Finally, navigate to the Ringdowns tab on the Results page.
The GUI automatically selects the starting time range as the most recent day with an event, in this case
May 23, 2017. The upper table lists events, while the lower table lists each of the signals that the event
was detected in. In this case, two events were detected. Click the Overview button to load the extrema for
the analyzed signals. Once done, the GUI will appear as in Figure 14, with the overview plot to the left.
Note that even with this course data, it is clear that the first event is less severe than the second. In the
overview plot, the event selected in the upper table is highlighted. To select a different event, simply click
it in the table.
To obtain detailed information about the second event, begin by zooming in on it, as in Figure 15. Note
that the table of events automatically updates to contain only events within the selected time period.
Clicking the Retrieve Detail button reruns the analysis and creates additional plots, as displayed in Figure
15. The top plot in the Detection Details section displays all of the analyzed signals. Below it, thumbnails
are provided for each of these signals. Clicking on the thumbnail will display a plot of the signal and a
plot of the detector’s operation (see Section 9.1.4).

Figure 14: GUI display after loading overview results for the ringdown detector.

31

Figure 15: GUI display after retrieving detailed information about a ringdown.

10.4.4

Wind Ramp Detector

In the Project panel, select the WindRamp task. If this step has not been completed previously, the GUI
will produce a dialogue box indicating that the example file cannot be found. Navigate to the Data Source
tab on the Settings page. In the Example File field, enter the full path to
*\ExampleData\WindRamp\2017\170203\ExData_20170203_000000.csv
The ExampleData folder was included in the download of AW (see Section 2). Note that if you navigate
to the example file using the GUI, you will need to select JSIS_CSV files in the lower right hand corner of
the File Explorer window to see the files. Next, click the Read File button to load the settings and results.
Finally, navigate to the Wind Ramps tab on the Results page.

32

The GUI automatically selects the starting time range as the most recent day with an event, in this case
February 3, 2017. The table lists events active during the selected time range. In this case, two events
were detected. Click the Overview button to load the extrema for the analyzed signals. Once done, the
GUI will appear as in Figure 16, with the overview plot below the table. In the overview plot, the event
selected in the table is highlighted. To select a different event, simply click it in the table.

Figure 16: GUI display after loading overview results for the wind ramp detector

10.4.5

Voltage Stability Detector

In the Project panel, select the VoltageStability task. If this step has not been completed previously, the
GUI will produce a dialogue box indicating that the example file cannot be found. Navigate to the Data
Source tab on the Settings page. In the Example File field, enter the full path to
*\ExampleData\VoltageStability\2017\170425\ExData_20170425_000000.csv
The ExampleData folder was included in the download of AW (see Section 2). Note that if you navigate
to the example file using the GUI, you will need to select JSIS_CSV files in the lower right hand corner of
the File Explorer window to see the files. Next, click the Read File button to load the settings and results.
Finally, navigate to the Voltage Stability tab on the Results page.
The GUI automatically selects the starting time range as the most recent day with an event, in this case
April 25, 2017. The upper table lists events, while the lower table lists each of the sites that the event was
detected in. In this case, nine events were detected. Click the Overview button to load the extrema for the
analyzed signals. Once done, the GUI will appear as in Figure 17. Under the Stability Indices banner, the
extrema of the stability metrics derived for each of the included Thevenin estimation algorithms are
plotted. In the example all five methods are implemented, but in the figure the results from three of them

33

have been collapsed. Under the Out-of-Range Data Trends banner, overview plots from all out-of-range
detectors are plotted. The red rectangles that are overlaid indicate detected events. These plots are
provided because out-of-range voltage events are useful when considering voltage stability.
To obtain detailed information about the decaying stability, zoom from approximately 00:04:00 to
00:15:59 and set the Prediction Delay field to 30 seconds. Next, click the Retrieve Detail button to rerun
the analysis and generate additional plots, as displayed in Figure 18. The plots under the Validation
banner include the measured bus voltage and an estimate of the voltage based on the Thevenin equivalent.
Good agreement between these values is one indicator that the Thevenin equivalent is a good
approximation. Clicking on the thumbnails allows the results from different Thevenin equivalent
estimation methods to be reviewed.

Figure 17: GUI display after loading overview results for the voltage stability detector.

34

Figure 18: GUI display after retrieving detailed information about the decaying voltage stability.

10.4.6

Mode Meter

In the Project panel, select the ModeMeter task. If this step has not been completed previously, the GUI
will produce a dialogue box indicating that the example file cannot be found. Navigate to the Data Source
tab on the Settings page. In the Example File field, enter the full path to
*\ExampleData\ModeMeter\2017\170523\ExData_20170523_000000.csv
The ExampleData folder was included in the download of AW (see Section 2). Note that if you navigate
to the example file using the GUI, you will need to select JSIS_CSV files in the lower right hand corner of
the File Explorer window to see the files. Next, click the Read File button to load the settings.
As discussed in Section 9.1.7, the mode meter tool is used to generate data, rather than to detect events.
Thus, the output of the mode meter tool is not displayed on the Results page. To see the output of the
mode meter example, navigate to the folder
*\Projects\Project_Examples\Task_ModeMeter\Event\MM\Meter1
The Projects folder was included in the download of AW (See Section 2). The folder name Meter1
corresponds to the Mode Meter Name parameter in the setup of the tool. The file 170523.csv contains the
results for May 23, 2017. These results are displayed in Figure 19. Column A contains timestamps,
columns B through H contain measurements to capture system conditions, and columns I and J contain
damping ratio and frequency estimates, respectively. The timestamps specify the time that the mode
estimates were generated and the system conditions were collected. For the system condition
measurements, row 2 lists the signal names, row 3 contains the signal type, and row 4 specifies the units.
For the mode estimates, row 2 contains the name of the mode specified in the Mode of Interest field in the
tool’s setup, row 3 specifies the signal used to generate the estimate, and row 4 lists the mode meter
algorithm. In this example, system frequency measurements are collected alongside estimates of a system
mode near 0.22 Hz.

35

Figure 19: Example output of the mode meter tool contained in 170523.csv.

11.0 Transferring the Location of Results and Data
Results and data can be moved for convenience or collaboration between AW users. The processes for
handling changes to the location of results and data are distinct and will be described separately.
Moving results is straightforward, but requires a high-level understanding of AW’s storage structure. For
the results storage path, projects, and tasks in the example Project panel of Figure 20, AW automatically
generates the subfolders in Figure 21. The results of each task are stored in a folder together. These task
folders are stored in the folder for their associated project. To transfer results, simply move the folders at
any level in Figure 21. Note that if individual tasks are moved, they must be placed under new folders that
match the naming convention Project_ProjectName, where ProjectName can be selected by the user.
When moving whole project folders, it is best practice to place them in a location occupied only by other
AW project folders. After relocating the results, simply select the new results storage location in the
GUI’s Project panel (see Section 3). The results can then be viewed from their new location.

Figure 20: Example Project panel.

36

Figure 21: Structure of results storage corresponding to the projects and tasks in Figure 20.
After AW analyzes a set of data, it still relies on the data to produce a detailed review of results (see
Section 10.1). If the data is moved after a task is saved, the user must specify the new location in the GUI.
Each time a task is selected, the GUI determines whether the example file entered on the Data Source tab
is available. If it is, it reads the file and populates the Settings page and the Signal Selection panel. If it is
not, the GUI requests that the user enter a new example file. When selecting a task after moving the
analyzed data, the GUI will make this request. In response, navigate to the Data Source tab under the
Settings page. Enter the new path to the example file and click the Read File button. The GUI will update
the configuration file for the task and then populate the Settings page and the Signal Selection panel. Once
this is done, results can be reviewed as usual.

37



Source Exif Data:
File Type                       : PDF
File Type Extension             : pdf
MIME Type                       : application/pdf
PDF Version                     : 1.5
Linearized                      : Yes
Author                          : Tuffner, Francis K
Company                         : Pacific Northwest National Laboratory
Content Type Id                 : 0x010100905F6E5C82F48540BC753A02C30AD17A
Create Date                     : 2018:09:26 08:34:57-07:00
Modify Date                     : 2018:09:26 08:35:10-07:00
PNNL Number                     : PNNL-XXXXX
Source Modified                 : D:20180926153443
Language                        : EN-US
Tagged PDF                      : Yes
XMP Toolkit                     : Adobe XMP Core 5.4-c006 80.159825, 2016/09/16-03:31:08
Metadata Date                   : 2018:09:26 08:35:10-07:00
Creator Tool                    : Acrobat PDFMaker 11 for Word
Document ID                     : uuid:0d5e70d6-ace1-466c-b585-2c92c8dd7ae9
Instance ID                     : uuid:6c631421-f584-49ee-abe8-f4284f3d44ec
Subject                         : 7
Format                          : application/pdf
Title                           : Title
Creator                         : Tuffner, Francis K
Producer                        : Adobe PDF Library 11.0
Page Layout                     : OneColumn
Page Count                      : 48
EXIF Metadata provided by EXIF.tools

Navigation menu