IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 27, NO.

2, APRIL 2012

475

A Real-Time Conductor Sag Measurement

System Using a Differential GPS
Satish M. Mahajan, Senior Member, IEEE, and Uma Mahesh Singareddy

AbstractA real-time and direct method of measuring the absolute value of sag on a transmission-line conductor using a differential global positioning system (DGPS) is presented. The method
includes a real-time data-processing module integrated into the
DGPS system. The GPS signals corresponding to physical movement of the midspan of a transmission-line conductor were obtained so that conductor-to-ground clearance could be recorded
in real time. Several field tests were designed and implemented to
test the DGPS method under varying conditions of the transmission-line current. A comparison of results obtained from the DGPS
method with those from a laser-range finder indicates an accuracy
1 in.
of
Index TermsDynamic thermal circuit rating, global positioning system (GPS), sag measurement.

I. INTRODUCTION

EREGULATION as well as increasing demand for

electrical power has led to additional stress on the existing transmission infrastructure. A short-term solution to
meet higher load demands requires operation of transmission
lines at higher ratings, thereby increasing the power transfer
capability. However, when these lines operate at higher ratings,
the conductor temperature increases and it tends to increase
the sag significantly. In fact, overhead conductor clearance is a
key limiting factor for many transmission lines. This, in turn,
necessitates the need to monitor sag at critical spans. Moreover,
real-time monitoring of conductor sag can be used for warning
purposes to ensure that mandatory clearance limits are not
violated [1], [2]. Several techniques have been developed and
used in the past to measure the conductor sag. Placement of
a resistive wire under a transmission line coupled with an
ammeter is one of these methods [1].
The measurement of the magnetic field under midspan, coupled with knowledge of line current, can also theoretically yield
values of conductor sag [3].

Manuscript received August 17, 2009; revised March 15, 2010; accepted
September 11, 2011. Date of publication February 14, 2012; date of current
version March 28, 2012. This work was supported by the U.S. Department
of Energy under Contracts DE-AC05-00OR22725 and ORNL subcontract
4000051155. Paper no. TPWRD-00623-2009.
S. M. Mahajan is with the Department of Electrical and Computer Engineering, Tennessee Technological University, Cookeville, TN 38505 USA
(e-mail: [email protected]).
U. M. Singareddy was with Tennessee Technological University, Cookeville,
TN 38505 USA (e-mail: [email protected]). He is now with CSG International, Springfield, IL USA.
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TPWRD.2011.2181963

A dynamic thermal circuit rating (DTCR) model involves the

evaluation of conductor temperature based on an energy balance differential equation [4], [5]. The conductor temperature
is then interpreted in terms of sag after taking in to account
wind velocity and ambient temperature. These and other similar methods really are the indirect ways of sag measurement.
In these indirect methods, a parameter related to the conductor
sag is measured and then used to calculate the sag. The indirect methods work well if the system behaves exactly according
to the model based on which they were developed. However,
real-world systems are not as ideal as the models used to describe them. Therefore, direct measurement systems have the
advantage.
Several commercially available systems can also be characterized as direct sag measurement systems. A system based on
line tension measurement (CAT-1) is capable of directly measuring the sag; however, the system must be cut-in to the
transmission system [6]. Another system is based on a sophisticated video camera that records and analyzes a series of picture
frames. This system involves advanced image processing [7].
Another device called Powerdonut2 uses conductor angle measurements to evaluate the sag [8], [9]. The electronics in these
and many modern measurement devices routinely incorporate
GPS for stamping time. Since GPS is inherently a distance measurement device, it has the ability to measure sag. There is no
doubt that accuracy of a GPS device is far better when used for
the measurement of horizontal distances rather than the vertical
distances. However, newer versions of GPS and advancements
in electronics as well as data-processing software are expected
to offer GPS systems with highly improved vertical measurement accuracy. In consideration of these advantages, the initial attempt to develop a differential global positioning system
(DGPS) was made a few years ago [10]. The DGPS method is
also a direct method that measures the conductor sag differentially by measuring the conductor-ground clearance based on
altitude information obtained from the GPS device. The objective of the earlier work was to test the feasibility of using DGPS
for sag measurement through the laboratory-based bench tests
which indicated that errors could be less than 20 cm[10]. Actual field tests in the presence of high voltage were not conducted to verify the practicality of this approach. Moreover, the
communication between the rover, the base, and the computer
system used to monitor sag was established using hard-wired serial connections. A more robust communication system that can
cater to significantly large distances would be needed for practical scenarios. Another issue is that the GPS must be mounted
on the phase conductor and, hence, a rugged shielding package
that can be used to mount the device was needed. This paper at-

0885-8977/$31.00 2012 IEEE

476

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 27, NO. 2, APRIL 2012

Fig. 1. Conceptual representation of the experimental system.

Fig. 2. Schematic details of the GPS enclosure (Rover).

tempts to address these issues so that a real-time DGPS-based

sag measurement system can become a reality. The behavior
of the system was studied on the basis of extensive field tests
conducted in a specially designed test site at the National Transmission Technology Research Center of the Oak Ridge National
Laboratories. The behavior of the GPS antenna in the presence
of high voltage was studied through the high-voltage tests conducted on another site. In addition, this paper attempts to address many of the problems encountered in implementing the
DGPS-based system in real time.
The basic configuration of the experimental system involves
placing the rover GPS integrated with the communication
module in a specially designed GPS enclosure on the transmission line. A geodetic GPS antenna, Leica GMX 902 receiver,
and a wireless modem were put inside the rover. The rover GPS
unit was used to evaluate the conductor to ground clearance
continuously, and this information was transferred in real time
to a distant sag monitoring center.
The base receiver used in this DGPS system was a reference
miles from the test site. A conceptual repstation located
resentation of the DGPS system is shown in Fig. 1 and details
of the rover unit are illustrated in Fig. 2.
II. REAL-TIME DATA-PROCESSING MODULE
The rover unit was made out of stainless-steel supports to a
6-in. diameter HVAC aluminum pipe on the outside and a hightemperature polymer tube on the inside. A geodetic antenna, a

receiver, and a wireless modem were fitted inside the pipe. Although the provision was made to include two 8-h batteries, field
tests were run with 120-V ac and an adapter so that 24-h tests
could be performed without any interruption. The GPS relevant
information was transferred in standard NMEA-0183 sentences
from the GPS receiver to the laptop over the wireless link established by the radio modem. These data were received and processed by the proprietary GPS software and the DGPS data were
subsequently logged in the system. After obtaining the DGPS
data, the altitude in formation from GPS messages was extracted
so that the conductor-to-ground clearance could be evaluated in
real time.
A module was developed in the VB.NET programming environment to process the raw GPS information and obtain the
best estimate of the current clearance value for a particular time
instance. The time instance at which processing took place was
defined by the reporting time value entered in the input data section.
The real-time processing module accessed the GPS data
from the proprietary GPS software over a TCP/IP connection.
After obtaining the raw GPS NMEA-0183 sentences, the
various input parameters pertaining to the processing window
size, reporting time and other parameters were entered in the
Input data section of the window, and the desired processing
technique was selected from the data filtering and smoothing
sections of the window. When the processing functionality was
activated, the raw NMEA sentences from the GPS log file were
first passed to the Parser module where the desired information
was identified and extracted. While extracting the data, the GPS
data corresponding to bad fix were rejected. The timestamp
and altitude values of each time instance corresponding to the
good fix were then added to the Process buffer list. The data
values were added to the list continuously until the current
time instance of the GPS data was equal to the time instance
corresponding to each reporting time. At each reporting time,
all of the data values in the process buffer were then accessed
by the Process Data function where the altitude values in the
Process buffer list were subjected to one of the data-smoothing
techniques selected before initiating the processing functionality. Several data-smoothing techniques were used to evaluate
the best approximate altitude value corresponding to the current
report time instance. The processed clearance value was then
displayed on an output console and subsequently transferred to
a real-time data plotter which plotted the latest clearance value.
III. RESULTS
The performance of the system was tested and analyzed by
conducting field tests on a specially designed test facility at
the National Transmission Technology Research Center of the
Oak Ridge National Laboratory (Fig. 4). The test line operated
at high current and low voltage. The current that was flowing
through the conductor could be set in the range of 1001600 A
dc. A laser-range finder was available to measure conductor-toground clearance.
Measurements from the highly accurate (submillimeter)
laser-range finder were used as the reference for the conductor-to-ground clearance at all times. Measurements obtained

MAHAJAN AND SINGAREDDY: REAL-TIME CONDUCTOR SAG MEASUREMENT SYSTEM

477

respect to the corresponding laser data on a second-by-second

basis (since the GPS records at a rate of one per second) and
study the possible curve-fitting and smoothing techniques
that could be used to improve the accuracy of the GPS data.
This module was developed in the MATLAB programming
environment.
B. Discrete Window-Processing Method

Fig. 3. NMEA-0183 data-processing module data flow.

The second methodology used was a discrete window-processing method. The processing window included the current
value of the GPS data and a set of previous values defined by
the size of the processing window. Data points were averaged
within discrete windows so that smoothing would occur. A relatively large discrete processing window implied the introduction
of a delay in processing.
1) No-Load Test: The first test involved a study of the extent
of sag experienced by the conductor without passing any current through the conductor. This test involved placing the GPS
enclosure system on the test line and evaluating the sag experienced by the line over a period of 24 h.
It was observed that a sag of approximately 30 in could be
experienced by the test line during 24 h. This sag can be attributed to the combined effect of atmospheric conditions, such
as ambient temperature, wind speed, and the solar radiation.
Fig. 5 shows the plot for GPS versus laser measurement data.
A few sharp notches were observed in the raw GPS data. These
notches were removed in the first stage of processing. While
processing the data, actual time measurements were not used.
Instead, time indices (in hours) from the start to the end of each
test were considered for plotting the corresponding clearance
values. Although some notches pertaining to the bad fix were
eliminated, the GPS data curve still has some noise superimposed on it.
After optimization of the window size for both sliding and
discrete windows, processing via sliding window was found to
be slightly more accurate. Processing of data similar to that in
Fig. 5 led to a relatively smooth curve for GPS data as illustrated in Fig. 6. The standard deviation in the absolute error
(GPS minus laser) was evaluated to obtain a measure of the accuracy of the GPS data with respect to the laser data for this test.

Fig. 4. GPS enclosure and the laser-range finder.

from the GPS were compared with these reference values to

calculate any error in the clearance obtained from the GPS
readings. The focus was on the altitude information contained
within the GPS signals from different satellites. The DGPS data
curves were subjected to two different smoothing techniques:
sliding window processing and the discrete window processing.
Detailed information on these two processing methods can be
found in earlier publications [3], [11].
A. Sliding Window-Processing Method
The first postprocessing method consisted of a sliding
window frame having a fixed number of data points prior to the
measurement instant in progress. The window formed by the
fixed number of points slid as the time progressed. The purpose
of this method was to evaluate the accuracy of GPS data with

Variations in solar radiation and the wind speed, which were

responsible for sag (without any current in the conductor) during
Test # 1, are illustrated in Figs. 7 and 8, respectively.
The mean and standard deviations in the error obtained by
using the sliding window processing method were 0.6688 and
0.8311, respectively. Using discrete window processing, the
values obtained for mean and standard deviation in the error
were 0.95002 and 0.837114, respectively. It can also be observed from Fig. 6 that the GPS data curve does not exactly coincide with the laser data curve at certain periods of time during
the test. This could be due to the error in the value reported by
GPS at that time. This variation in the GPS measurements with
respect to the laser may be due to orientation of satellites at that
time of the test, the ambient temperature conditions, the wind
speed, etc.
2) High Current Tests: The subsequent tests involved
subjecting the test line to various specific current profiles as

478

Fig. 5. GPS versus laser clearance plot corresponding to filtered GPS data for
Test # 1 (no current).

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 27, NO. 2, APRIL 2012

Fig. 8. The 3-D wind speed profile for Test #1.

Fig. 9. Test #2.

Fig. 6. GPS versus laser clearance plot corresponding to the smoothed GPS
data for Test # 1.

Fig. 10. Test #3.

Fig. 7. Solar radiation profile for Test #1.

shown in Figs. 915. These tests were designed to measure

conductor sag under fast and slow changes in current as well as
under constant values of current for short and long durations.
Smoothed values of conductor-to-ground clearance obtained
from the DGPS data and from the reference values of the

laser-range finder are embedded within these figures which also

display profiles of the electrical current.
During all of these tests, a reference station located at 10
mi from the rover was used as the base receiver to obtain DGPS
measurements. The data set was filtered and smoothed. It was
observed that the best approximate of the GPS data with respect to the laser data can be obtained by subjecting the data

MAHAJAN AND SINGAREDDY: REAL-TIME CONDUCTOR SAG MEASUREMENT SYSTEM

Fig. 11. Test #4.

Fig. 14. Test #7.

Fig. 12. Test #5.

Fig. 15. Test #8.

479

wire. Measurements of altitude were taken with and without a

voltage of 30 kV and at distances varying from 1 to 5 in. between
the GPS receiver and the wire. No additional shielding was put
between the receiver and wire. There was no effect on the standard deviation in the altitude reading of the GPS. Similar tests
under a 61-kV line (at a separation of 30 ft between the GPS receiver and the 61 kV HV line) indicated no influence on the GPS
readings. A significant separation between GPS frequencies (in
gigahertz) and communication frequencies (in megahertz) with
respect to the power-line frequency (60 Hz) as well as the corona
frequency (in kilohertz) can thus be helpful. Similar observations were reported earlier [12].
IV. CONCLUSION
Fig. 13. Test #6.

set to a 10-s sliding window. In all of the aforementioned tests,

it was observed that the GPS measurement recorded sag with an
error of approximately 1 in when compared with the laser-range
finder.
3) HV Exposure Tests: As illustrated in Fig. 16, a conducting
wire was placed over the GPS receiver and HV applied to the

A working model of a DGPS system was fabricated and

demonstrated to operate at 1.5-GHz carrier signals and with
a 900-MHz communication link. Immunity to HV in close
proximity makes it applicable in real power systems where
online sag information can be very effectively used to monitor
critical spans. Therefore, the DGPS system can be used to
stretch the transmission-line capacity (short term) beyond the
normal rating so that excess thermal capacity of the line could
be exploited. Results from field tests show that the differential

480

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 27, NO. 2, APRIL 2012

Fig. 16. GPS device in the proximity of an HV conductor.

GPS with a remote communication system could measure

the conductor sag to an accuracy of approximately 1 in. Data
acquisition and processing led to a maximum delay of 1 min,
thereby making this a truly online and a direct method of sag
measurement. The inclusion of a power scavenging circuit
for batteries will make this system more practical. In fact, a
commercially available product (powerdonut) uses this type
of circuit. Having a time stamp on the data obtained from
DGPS is an additional advantage that could be very valuable
from a power system fault analysis point of view. A local base
station (instead of a remote one) would guarantee accuracy
of the system. A series of YAGI antenna can certainly extend
the range of communication system where data can reach a
control room operator located at distances of 50 mi or greater.
The relatively expensive system used in this GPS system was
based on carrier frequency. However, a cluster of code-based
GPS devices with the DGPS method could bring the cost down
significantly. The reduction of relatively large error in these
code-based GPS devices through the multisensor data fusion
technique is currently being pursued. It should be noted that the
GPS system is not a controlled environment and is likely to be
affected by various external factors, such as satellite geometry
during the time of the test, ambient weather conditions, etc.

ACKNOWLEDGMENT
The authors would like to thank R. Craven and C. Schuff from
TTU, and Dr. J. Stovall (ORNL) for their assistance.

REFERENCES
[1] R. G. Olsen and K. S. Edwards, A new method for real time monitoring of high-voltage transmission-line conductor sag, IEEE Trans.
Power Del., vol. 17, no. 4, pp. 11421151, Oct. 2002.
[2] A. V. Mamishev, R. D. Nevels, and B. D. Russell, Effects of conductor
sag on spatial distribution of power line magnetic field, IEEE Trans.
Power Del., vol. 11, no. 3, pp. 15711576, Jul. 1996.
[3] U. M. Singareddy, Development of real time sag monitoring system
using differential global positioning system, M.Sc. dissertation, Elect.
Comput. Eng. Dept., Tennessee Tech. Univ., Cookeville, TN, Dec.
2007.
[4] IEEE Standard for Calculating the Current-Temperature Relationship
of Bare Overhead Conductors, IEEE Std. 7381993, 1993.
[5] W. Z. Black and W. R. Byrd, Real-time ampacity model for overhead lines, IEEE Trans. Power App. Syst., vol. PAS-102, no. 7, pp.
22892293, Jul. 1983.
[6] The Valley Group Inc, CAT -1 transmission line monitoring system,
Brochure. [Online]. Available: http://www.nexans.us/US/2008/CAT1_Brochure_1.pdf
[7] Sagometer, Product Overview. EDM Int. [Online]. Available: http://
edmlink.com/EDM_Sagometer.pdf
[8] USI-Power Inc., Power Donut2 system for overhead transmission line
monitoring, product overview. Brochure, 2006.
[9] USI-Power Inc , Power Donut2 system for overhead transmission
line monitoring, communications. Brochure. [Online]. Available:
http://www.usi-power.com/Products%20&%20Services/Donut/PowerDonut2%20Cell.pdf
[10] C. MensahBonsu, U. F. Krekeler, G. T. Heydt, Y. Hoverson, J. Schilleci, and B. A. Agarwal, Application of the global positioning system
to the measurement of overhead power transmission conductor sag,
IEEE Trans. Power Del., vol. 17, no. 1, pp. 273278, Jan. 2002.
[11] S. M. Mahajan and U. M. Singareddy, Real time GPS data processing
for Sag measurement on a transmission line, presented at the IEEE
POWERCON IEEE Power India Conf., paper 01, New Delhi, India,
Oct. 1215, 2008.
[12] J. M. Silva and R. Olsen, Use of global positioning receivers under
power-line conductors, IEEE Trans. Power Del., vol. 17, no. 4, pp.
938944, Oct. 2002.
Satish M. Mahajan (SM09) was born in India and
received the B.E. degree in electrical engineering
from the University of Poona in 1978, the M.S.E.E.
degree from the State University of New York at
Buffalo in 1983, and the Ph.D. degree in electrical
engineering from the University of South Carolina,
Columbia, in 1987.
Since 1987, he has been on the faculty of Electrical
Engineering Department, Tennessee Technological
University, Cookeville. His research interests are
related to physical phenomena of optoelectronic and
high-voltage devices, and the modeling of conventional and renewable power
systems.

Uma Mahesh Singareddy received the M.Sc.

degree in electrical engineering from the Tennessee
Technological University, Cookeville, in 2007, and
the B.Tech. degree in electronics and communication
engineering from Jawaharlal Nehru Technological
University, Hyderabad, India, in 2004.
Currently, he is a Senior Software Development
Engineer with CSG International, Springfield, IL,
with expertise in architecting and developing software for the maintenance of wireless voice and data
networks.