Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
A Comparison of Real
Time Thermal Rating
Systems in the U.S. and
the UK
David M. Greenwood
Jake P. Gentle
Grant L. Ingram
Peter J. Davison
Kurt S. Myers
Isaac J. West
Jason W. Bush
Matthias C. M. Troffaes
March 2013
David M. Greenwood
Jake P. Gentle
Grant L. Ingram
Peter J. Davison
Kurt S. Myers
Isaac J. West
Jason W. Bush
Matthias C. M. Troffaes
March 2013
http://www.inl.gov
1
Abstract— Real-Time Thermal Rating is a smart grid
technology that allows the rating of electrical conductors to be
increased based on local weather conditions. Overhead lines are
conventionally given a conservative rating based on worst case
scenarios. We demonstrate that observing the conditions in real
time leads to additional capacity and safer operation. This paper
compares two pioneering schemes—one in the United States of
America and one in the United Kingdom—in which Real-Time
Thermal Ratings were applied. Even though the systems had the
same aims, comparing the approaches taken by two independent
engineering teams has led to insights that will be inform and
improve future Real-Time Thermal Rating projects.
I. INTRODUCTION
A. The Concept of Real-Time Thermal Ratings Figure 1. Diagram of the heat balance within a conductor
Real-Time Thermal Ratings (RTTR) (also referred to as
Dynamic Thermal Ratings, Dynamic Line Rating, Real Time This energy balance is described by Michiorri, Taylor, and
Rating, and Concurrent Cooling) are based on the observation Jupe in their 2009 article [2]:
that the first limit of a current carrying conductor is its
temperature. Power lines, cables, and transformers are ܫଶ ܴሺܶ ሻ ݍ௦ ൌ ݍ ݍ (1)
operated using a static rating based on conservative seasonal
conditions. Consequently, there is unused headroom within the where qs, qc, and qr are the heating through solar radiation,
power system because of the cooling effect of the cooling through convection, and cooling through radiation,
environment. This technology could be applied to defer costly respectively. Tc is core temperature of the conductor. I2R(Tc)
upgrades, increase the yield of Distributed Generation (DG) represents the internal heating due to line current from the
and support the network during outages. Joule Effect. Equation (1) represents a steady state case, where
This paper contributes a unique comparison of two the line has heated up to the point at which the heat losses and
independently developed RTTR methods from the United gains are equal.
States and the United Kingdom. The work documented in this Power lines have a maximum operational temperature that
report focused on overhead lines (OHLs), the component should not be exceeded. This maximum temperature is used to
which stands to benefit the most from the adoption of RTTR calculate the maximum current that can be accommodated by
[1]. Fig. 1 shows the energy balance in an OHL between the line. There are three widely used models for calculating
environmental conditions and heating by the Joule Effect. overhead line capacity, produced by CIGRE [3], International
Electrochemical Commission (IEC) [4], and the Institute of
This work is supported by the U.S. Department of Energy Wind Power Electronics Engineers (IEEE) [5].
Technology Office contract with Idaho National Laboratory, the Durham
Energy Institute and EPSRC Grant EP/J501323/1. The authors would also like
to acknowledge the contributions and support of Idaho Power. B. Applications
David M Greenwood, Peter J. Davidson and Grant L. Ingram are with the
School of Engineering and Computer Science, Durham University, UK
RTTR can be applied during planning, design, and
(d.m.greenwood@dur.ac.uk) operation of a transmission system. The benefits it provides
Matthias C.M. Troffaes is with the Department of Mathematical Sciences, include increasing the capacity for potential wind generation,
Durham University, UK (Matthias.troffaes@dur.ac.uk)
Jake P. Gentle, Kurt S. Myers, Isaac J. West and Jason W. Bush are with
due to the natural synergy between generation and increased
the Idaho National Laboratory, USA (Jake.Gentle@inl.gov) conductor capacity at times of high local wind speed. The
INL/EXT-13-28553 2
need for network reinforcement can be deferred due to the investor-owned utilities, public utilities, and electric
additional capacity. Scheduled outages can be planned cooperatives. Figure 3 shows the major transmission lines and
considering the potential higher line ratings, and the increased substations within the U.S..
network capacity can help reduce the number of disconnected
customers during unscheduled outages.
upon the level of risk allowed. The scaling factor used to do Fig. 7. U.S. trial site in southern Idaho, showing local terrain, conductors,
weather stations, and model points.
this is known as the correlation term (CT). Figures of
exceedence are pre-determined for all distribution networks in Presently, the dynamic ratings of two 138kV lines and two
the UK. For single circuit supply systems the figure is 0% and 230kV lines are being studied with 17 weather stations
for multi circuit 3%. Realistically the figure of 0% is instated at strategic locations on tower structures along the
unobtainable from a log curve therefore 0.001% is used. lines. All of the weather stations are spaced between 1 and 5
Upon selection of the exceedence value required, the miles.
correlation term can be trivially found using the graph shown
in Fig. 6. The product of the design current and correlation B. Description of Local Climate
term gives the thermal rating for the circuit under A historical database of climatology characteristics has been
consideration. collected since 2007. Over the years, however, weather
100 stations have been added or moved. Currently, the largest and
most complete data set from all 17 weather stations ranges
from April 25, 2012 to present. Fig. 8 below shows a sample
10
wind rose plot of data from one weather station (WS03),
collected from August 14, 2011 to August 14, 2012.
1
Te(%)
0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2
0.1
0.01
0.001
CorrelationTermCT
and wind power production. almost all of these systems is that they typically do not provide
The local terrain comprises of a large valley, containing enough measurements to obtain an accurate assessment of the
small towns, villages and forests. The total change in terrain varying climate conditions, line temperatures, and sags along
height across the area is approximately 200 m. A map of the each span. More weather data (wind speed, wind direction,
area is shown in Fig. 9. The total area covered by the trial site ambient temperature, and solar irradiance) are needed along
is 16.7 square miles. the transmission lines to improve the calculations and
accuracy of existing systems.
The system developed by INL uses weather and
environmental measurements to dynamically rate transmission
lines. This system incorporates larger quantities of weather
measurement equipment installed along the transmission lines
to be dynamically rated. The weather measurement equipment
collects data such as wind speed, wind direction, ambient air
temperature, and solar irradiance levels at predetermined
locations, and is used to model and calculate a more complete
and accurate picture of the weather conditions and
temperatures along the full length of transmission lines of
interest. Weather parameters are obtained from local weather
stations, mounted on the power line poles.
A computational fluid dynamics (CFD) program is being
used to estimate wind conditions on a transmission line
Fig. 9. The UK site trial in North Wales, showing the local area and the between weather stations using information from the stations
location of the conductors and weather stations (TBC).
along the line. The technology used for the wind estimation
B. Description of Local Climate was originally developed for wind farm annual energy
The five weather stations at the trial site have been in place production and a challenge in the present work is using the
since 2008. The local climate data presented here is calculated CFD software to simulate wind conditions over a much larger
from data spanning January 1, 2009 to December 31, 2009. geographic area and that the direction vector is critical. The
The average temperatures for each season are shown in CFD model is refined using historical wind information
Table I. obtained from 17 weather stations located within the region of
Fig. 10 shows a wind rose plot for the UK trial site. The interest. Simulations are then performed to estimate the wind
prevailing winds come from the south west, though winds velocity along the length of the line in distinct 500–1000-m
from all directions are observed throughout the year. sections, using the nearest weather station on similar terrain,
by expert judgment. The CFD program includes information
about the variation and surface roughness of the terrain in the
modeling and therefore is better able to accurately simulate the
wind speed and direction, taking into account the topography
of the land.
Additional INL custom built software programs use CFD
simulated wind speeds and a line’s currents to determine
cooling effects and real time conductor temperatures. From
this process the ampacity rating of the line can be adjusted
dynamically.
The convection heat loss has two equations: the value qc1
for low air speed(<3 mph) and qc2 for higher air speed:
ୈ౭ ୮ Ǥହଶ
ୡଵ ൌ ൜ͳǤͲͳ ͲǤͲ͵ʹ ቀ ቁ ൠ ୟ୬୪ୣ ሺୡ െ ୟ ሻ (5)
ஜ
ୈ౭ ୮ Ǥ
ୡଶ ൌ ൜ͲǤͲͳͳͻ ቀ ቁ ൠ ୟ୬୪ୣ ሺୡ െ ୟ ሻ (6)
ஜ
σሾቆଵ൘ మ ቇୱ ሿ
୪ǡౡ
୩ ൌ (13)
σቆଵ൘ మ ቇ
୪ǡౡ
This method is used to estimate Wd, Sr, and Ta. For the
wind speed, Ws, the ground roughness is taken into account
using the log law shown in (15).
C. Included Components
The UK method includes models for overhead lines,
underground cables and power transformers. The potential
benefits of each of these components is discussed in [1].
However, for the trial site considered by this paper, only
overhead lines are considered.
D. Monte Carlo Simulations
In order to account for the uncertainties present in the Fig. 14. An illustration of the Monte Carlo method employed as part of the
system, rather than using fixed values for each variable, a UK rating estimation method.
probability distribution over each variable is assumed. For
simplicity, all variables are assumed to be statistically
independent from one another.
Because the functions involved are quite complicated,
analytic derivation of the probability distribution of the line
rating is not really feasible. Hence, the Monte Carlo method
was used. In this method, samples from the probability
distributions of the input variables are drawn to perform n
deterministic calculations of the output variable, that is, the
line rating. The results of these calculations approximate the
probability distribution of the line rating. The approximation
improves as n is higher. An illustration of this is shown in
Fig. 14 [2].
Fig. 15 shows results from a trial conducted in the winter of
2008/2009 [2]. This is a comparison of measured and
Fig. 15. A comparison of measured and estimated conductor temperatures for
estimated conductor temperature. During the winter period, winter 2008, at the north Wales trial site [2].
the line cooling is dominated by low ambient temperatures.
Fig. 16 shows the same results, but from summer 2009
when wind speed and direction are the dominant factors in line
cooling. The method does not perform as well in the summer,
and as a result, work to improve wind and direction estimation
using CFD calculations is ongoing [15].
VIII. KEY DIFFERENCES IN METHODOLOGY quantify some of the error and improve wind prediction, but
The two methodologies share many common elements, there is certainly potential for further improvement in this
which implies that the general approach is appropriate for the direction.
application. Even given all these factors, it is encouraging how far we
The UK methodology’s inclusion of Monte Carlo can get by with the current simple methods.
calculations to account for some level of uncertainty is a key
difference. If RTTR were to be deployed on operational X. CONCLUSIONS
networks, understanding the uncertainties would be essential Real-Time Thermal Ratings systems have been developed
to its success. This may be driven by the fact that the UK does independently in the U.S. and the UK. Both systems are
not identify a critical span. However, this method is more currently in the prototype phase and active use of Real-Time
computational intensive since many calculations must be run Thermal Ratings is imminent. The methodologies share
in place of one. Furthermore the UK distribution network common elements, but the U.S. system has a more
operators are conservative, so a method for highlighting the sophisticated wind model while the UK system has a better
level of risk was considered essential. Perhaps one of the key uncertainty model. In future, we hope to develop a method
features of the UK method is that it shows operators that they which takes advantage of the best aspects of both approaches.
are already running at a measurable level of risk, because there
are times when the static line ratings are exceeded. The U.S. REFERENCES
system has done some initial quantification of uncertainty [1] Michiorri, A., Taylor, P.C., Jupe, S.C.E., Berry, C.J., Investigation into
using field measurements, with the conventional engineering the influence of environmental conditions on power system ratings. Proc.
IMechE Vol. 223 Part A: J. Power and Energy, 2009: p. 743-757.
factor of safety method. [2] Michiorri, A., Taylor, P.C., Jupe, S.C.E, Overhead line real-time rating
The U.S. method includes advanced CFD wind simulation estimation algorithm: description and validation. IMechE Vol. 224 Part
which is driven by the large domain and complex terrain and A: J. Power and Energy, 2009. 224(A): p. 293-304.
topology. In the UK the small domain size and a closer [3] Cigre Working Group 22.12., The thermal behaviour of overhead line
conductors. Electra, 1992. 114(3): p. 107-25.
average spacing of MET stations allows a simpler method to [4] IEC, Standard TR 1597 Overhead Electrical Conductors - Calculation
give accurate results. Methods for Stranded Bare Conductors. 1985.
In the UK the peak load is experienced in winter when low [5] IEEE, Standard 738 Standard for calculating the current-temperature
relationship of bare overhead line conductors. 1993.
temperatures dominate the line cooling. Temperature is a more [6] Michiorri, A., Taylor, P.C. Forecasting real-time ratings for electricity
diffusive rather than convective property, and as such, varies distribution networks using weather forecast data. in CIRED'09. 2009.
less over the distances considered. As a result a crude method Prague.
[7] Jupe, S.C.E., P.C. Taylor, and A. Michiorri, Coordinated output control
for estimating wind speed and direction can be accepted. The of multiple distributed generation schemes. Renewable Power
limitations of this method become apparent in the summer, Generation, IET. 4(3): p. 283-297.
when the wind flows dominate the cooling [8]. [8] Jake Gentle, K.S.M., Thomas Baldwin, Isaac West, Hart, Bruce Savage,
Mikle Ellis, Phil Anderson, Concurrent Wind Cooling in Power
In the U.S., there is no clear seasonal load peak since it Transmission Lines, in Western Energy Policy Research Conference.
varies from region to region, so the model must be robust at all 2012: Boise, Idaho.
times of year. This has led to the development of a more [9] Office, M. Scotland 1971–2000 averages. [cited 2012 21/08]; Available
from:
sophisticated wind estimation approach. http://www.metoffice.gov.uk/climate/uk/averages/19712000/areal/scotla
nd.html.
IX. FUTURE DEVELOPMENTS [10] Office', M. England 1971–2000 averages. [cited 2012 21/08]; Available
from:
The UK method could be improved by adopting a more http://www.metoffice.gov.uk/climate/uk/averages/19712000/areal/engla
sophisticated wind estimation technique. In fact this is the nd.html.
subject of a current paper by several of the authors. In order to [11] D.E.C.C., Digest of UK energy statistics. 2012.
[12] Price, C.F. and R.R. Gibbon, Statistical approach to thermal rating of
implement RTTR on real networks, proper quantification of overhead lines for power transmission and distribution. Generation,
all uncertainties involved is crucial. Transmission and Distribution, IEE Proceedings C, 1983. 130(5): p.
First, various model assumptions, such as independence and 245-256.
[13] Catherine Meissner, K.V., Agder Energi and U.W.-S. Horn. Park
shapes of distributions, could be better validated, and where optimization using IEC constraints for wind quality in EWEA Anual
necessary, modified to any specific situation. Second, many Event 2011. 2011. Brussels, Belgium.
uncertainties are currently not accounted for at all, including [14] Shepard, D., A two-dimensional interpolation function for
irregularly-spaced data, in Proceedings of the 1968 23rd ACM national
for instance structural uncertainty due to imperfect terrain conference. 1968, ACM.
shape and roughness modeling, limited resolution, boundary [15] Greenwood, D.M., Ingram, G.L., Taylor, P.C. Improving Real-Time
conditions, and steady-state assumptions which underlie all Thermal Ratings Using Computational Wind Simulations. in 10th Wind
Integration Workshop. 2011. Aarhus, Denmark.
CFD simulations, at least in the current approach.
The U.S. method did include historical data in order to