Professional Documents
Culture Documents
Vivek K. Singh1 , Member, IEEE, Timothy R. McJunkin1 , Senior Member, IEEE, and Craig G. Rieger1 , Senior
Member, IEEE
Abstract—Distribution systems are large and complex systems these metrics provide insightful information to system health,
that grow over time. When contingencies occur dispatchers, en- the authors of [18] determined that the system voltage profile,
gineers, and executives need to know the answer to the question, load, loss, and source power depend heavily on the load
“How bad is it?”, in order to make timely and meaningful
decisions. Therefore, a scalable method to add context to system model that is used when considering the distribution system
conditions is needed. In this paper we present novel deviation- and consequently, show that the incorporation of a voltage
based techniques which enable scaled bus voltage and line power dependent load model is essential. In addition, the 33-bus
measurement deviations to quickly assess local and global health, system we will be using for testing features a single generator,
and lend context to the severity of contingencies. The proposed invalidating the use of frequency stability analysis for our
techniques are demonstrated using simulations of line-to-line,
line-to-ground, and three-phases-to-ground fault conditions on purposes.
an IEEE 33-bus distribution model. In this paper, we present a novel approach to health as-
Index Terms—distribution system, heath metrics, line faults, sessment of the distribution system. Local health assessment
voltage stability. indicates the health of an individual aggregated system re-
source (ASR). In this work we define an ASR as a set of
assets that have common objectives. Note that the ASR can
I. I NTRODUCTION
be used for many scales. For example, the components of
Authorized licensed use limited to: National Renewable Energy Laboratory. Downloaded on February 23,2022 at 01:50:43 UTC from IEEE Xplore. Restrictions apply.
2
Authorized licensed use limited to: National Renewable Energy Laboratory. Downloaded on February 23,2022 at 01:50:43 UTC from IEEE Xplore. Restrictions apply.
3
by assessment functions, α() and β(), into deviations which where θ is the threshold for the applicable line power upper
are then scaled and summed into local and global health limit, and m is the slope which controls the rate of conversion
assessments. from deviation to output beyond the threshold. Domain values
lesser than the threshold represents the region where normal
A. System operating points operation is expected to take place. As the power limits placed
System operating points are measurements of system pa- on lines are not directional, meaning one excess watt flowing
rameters taken at particular moments in time that represent out of a line has the same effect on the line as one excess
the status of the system. In this work, the operating points are watt flowing into a line, it is necessary to correct for direction
measures of the bus voltages and line powers of the system and of power prior to using the β function. Other functions, non-
are sampled at regular intervals. Each operating point in the linear functions for example, may be used depending on the
collection is unfiltered data, which when passed through as- desired performance of the assessment, provided that they
sessment functions, create deviation-based information about adhere to the requirement of producing a maximum value of
the state of the system. n sequential samples are used to assess 100.
local and global health.
D. Local health assessment
B. The α() function Local heath assessment provides a quick indication of
α() is an assessment function that takes a root-mean-square voltage or power deviation for a particular ASR. This is
(RMS) voltage as an input and returns a unit-less value accomplished by scaling and summing the deviations in the
between -100 and 100. The input value, ∆V , is the difference bus voltages and line power flows indicated by the assessment
between an actual voltage, from one of the phases of a bus, and functions for a given ASR. For example, λ1 , which represents
the optimal voltage of that bus. The purpose of the α function the local health of ASR1, is given by
is to quantify the extent of the deviation of a bus voltage. It Pn PASR1
produces both negative and positive values so that the output i=1 j=Bus pαj |αj (∆Vji )|
PASR1
may be used to distinguish between under- and over-voltage + j=Line pβj βj (∆Pji )
conditions in other state assessment computations. Each bus in λ1 = 1 − P (3)
ASR1 PASR1
the IEEE 33-bus model has its own α function characterized by 100n j=Bus pαj + j=Line pβj
upper and lower bus voltage limits and slopes. The stereotype where n is the number of sampled operating points taken into
for the α functions implemented in this work is described by consideration, pαj and pβj are importance factors for the jth
−100 : ∆V ≤ −100
m1 + θ1 bus and line, respectively, in ASR1, ∆Vji is the difference
−100
1
m (∆V − θ 1 ) : m1 + θ 1 < ∆V ≤ θ1 between the bus voltage and the optimal bus voltage for the
α(∆V ) = 0 : θ1 < ∆V ≤ θ2 ith sample of the jth bus, and ∆Pji is the difference between
m2 (∆V − θ2 ) : θ2 < ∆V ≤ 100 m2 + θ2
the line power and the optimal line power for the ith sample
PASR1
100 100
: ∆V > m2 + θ2 , of the jth bus. The notation j=Bus means to sum over all
PASR1
(1) of the buses in ASR1; similarly, the notation j=Line means
where θ1 and θ2 are thresholds relating to the lower and upper to sum over all of the lines connected to ASR1. λ can vary
voltage limits, respectively, of the applicable bus, and m1 and between 0.0 and 1.0, where 1.0 indicates that all voltages and
m2 are the lower and upper slopes, respectively, which control line power flows are in their deadbands. A λ not equal to
the rate of conversion from deviation to output beyond the 1.0 indicates that a deviation exists in an ASR and should be
thresholds. Domain values between the thresholds represents investigated. Local health assessments for each of the other
a deadband where normal operation is expected to take place. ASRs are conducted a manner similar to (3).
Other functions, non-linear functions for example, may be
used depending on the desired performance of the assessment, E. Global health assessment
provided that they adhere to the requirement of producing The purpose of the global health assessment is to provide
maximum and minimum values of 100 and -100, respectively. a single point of reference with which to determine if all bus
voltages and line power flows are in their deadbands. This is
C. The β() function indicated by a global health assessment value (γ) of 1.0. The
β() is an assessment function that takes a power as an input global health assessment is given by
and returns a unit-less value between 0 and 100. The input Pλ
pλi λi
value, ∆P , is the difference between an actual line power and γ = Pi=1λ
, (4)
the usual power of that line. The purpose of the β function is i=1 pλi
to quantify the extent of the deviation of a line power. Each where pλi is the importance factor for the ith local health
phase of each line in the IEEE 33-bus model has its own β assessmentPλand λi is the ith local health assessment. The
function characterized by an upper power limit. The stereotype notation i=1 means to sum over all of the local health
for the β functions implemented in this work is described by assessments. The scaling performed by the importance factors
help to convey the severity or impact of deviations in different
0 : ∆P ≤ θ
m(∆P − θ) : θ < ∆P ≤ 100 ASRs. A γ not equal to 1.0 indicates that a deviation exists
β(∆P ) = m +θ (2)
100 : ∆P > 100 in one of the ASRs which should be investigated.
m + θ,
Authorized licensed use limited to: National Renewable Energy Laboratory. Downloaded on February 23,2022 at 01:50:43 UTC from IEEE Xplore. Restrictions apply.
4
1.0
10 1.0
10
ASR1 ASR1
ASR2 ASR2
ASR3 ASR3
ASR4 ASR4
ASR5 ASR5
0.4 ASR6 0.4 ASR6
0 0
1 1
0.2 0.2
Global Global
0.00 0.00
0.00.950 0.975
0.2 1.000 0.41.025 1.050
0.6 1.0750.8 1.100 1.0 0.00.950 0.975
0.2 1.000 0.41.025 1.050
0.6 1.0750.8 1.100 1.0
Time (s) Time (s)
Fig. 2. The voltage of Bus 6, local healths for ASRs 1 through 6, and the Fig. 4. The voltage of Bus 33, local health assessments for ASRs 1 through
global health for a short between phase A and ground in Bus 6. 6, and the global health assessment for a 3-phase short to ground in Bus 33.
1.0
10 TABLE II
Global Health Local Health Bus 18 (kV)
Fig. 3. The voltage of Bus 18, local health assessments for ASRs 1 through depicted in Fig. 3.
3 Fault 3 is the three phase short to ground in Bus 33 depicted
6, and the global health assessment for a short between phases A and B in
Bus 18. in Fig. 4.
V. S IMULATION AND R ESULTS These are indications that those ASRs should be investigated
Simulations demonstrating the health assessment techniques first. As scaling is not performed in these simulations (All of
described in Section III were conducted in Python version the importance factors are 1.0.), the impact of the deviations
3.8.2 using fault data from the model described in Section on local and global healths are considered to be uniform. If
II. To achieve this, all local and global importance factors the bus voltage deviations were scaled by importance factors,
were set to 1.0, and then operating points produced by the then the impact of the deviations would also be scaled and
IEEE 33-bus distribution model described in Section II were reflect the relative detriment of each deviation. Next, the
evaluated under various fault conditions including phase-to- global healths lower due to the collective changes in the local
phase, phase-to-ground, and three-phase-to-ground faults. healths. After a short time, the transient effects of the faults
finish propagating through the system, which cause the local
A. Fault condition response and global healths to stabilize at non-optimal values. Finally,
after five cycles automatic corrective action is taken and the
For the simulations in this section, all importance factors
system normalizes, which is reflected in rising local and global
have been set equal to 1.0. Fig. 2, Fig. 3, and Fig. 4 depict
healths. Table II summarizes the minimum values of each of
the local and global health assessments for a L-G, L-L, and
the local healths and the minimum value of the global health
LLL-G faults, respectively. Fig. 2 depicts a phase-A-to-ground
for the faults described in this section.
fault in Bus 6, Fig. 3 depicts a phase-A-to-B short in Bus 18,
and Fig. 4 depicts a 3-phase-short-to-ground in Bus 33. In all
three figures, the fault occurs at 1 s, and causes deviations B. Importance factors
throughout the system which are reflected in the local and Importance factors scale the deviations of the local health
global health assessments. At first, when the fault occurs, the assessments, and the impact of each local health assessment
local healths are reduced due to the induced deviations in on the the global health assessment. The scaling performed
the system. Notice how in each figure the ASR where the by the local health importance factors, pα and pβ , emphasizes
faulty bus is located is affected most severely; In Fig. 2 Bus the deviations that occur in critical busses or lines, while the
6 is is affected most severely, in Fig. 3 Bus 18 is affected scaling performed by the global health importance factors,
most severely, and in Fig. 4 Bus 33 is affected most severely. pλ , emphasizes deviations in more important ASRs over less
Authorized licensed use limited to: National Renewable Energy Laboratory. Downloaded on February 23,2022 at 01:50:43 UTC from IEEE Xplore. Restrictions apply.
5
Authorized licensed use limited to: National Renewable Energy Laboratory. Downloaded on February 23,2022 at 01:50:43 UTC from IEEE Xplore. Restrictions apply.
6
Authorized licensed use limited to: National Renewable Energy Laboratory. Downloaded on February 23,2022 at 01:50:43 UTC from IEEE Xplore. Restrictions apply.