2020 IEEE Canadian Conference on Electrical and Computer Engineering (CCECE)

M-Model: A New Precise Medium-Length

Transmission Line Model
Ali R. Al-Roomi Mohamed E. El-Hawary
Department of Electrical and Computer Engineering Department of Electrical and Computer Engineering
Dalhousie University Dalhousie University
1459 Oxford Street, Halifax, NS B3H 4R2 Canada 1459 Oxford Street, Halifax, NS B3H 4R2 Canada
Email: [email protected] | Tel.: +1(902) 989-4589 Email: [email protected] | Tel.: +1(902) 473-6198

Abstract—Real transmission lines are translated into mathe-

matical models using either the lumped parameter approach or
the distributed parameter approach. The first one is used for
short- and medium-length transmission lines, while the other
is used for long-length transmission lines where the accuracy
and precision are required. For medium transmission lines, the
lumped parameter approach can be applied using one of four
popular circuit representations known as gamma (Γ), opposite-
gamma (Γ), tee (T), and pi (Π). This study presents a new circuit
representation called em (M). This model is inspired by the sag- Fig. 1. Short-length transmission line model.
ging phenomenon where, at the sag point, the distributed series
impedance of the Π-model is divided into two equal/unequal parts
and the distributed shunt admittance at the center is bigger than
that at both ends. For some numerical experiments, the M-model
shows a stunning performance in estimating transmission line
readings. It wins in most cases and, for the few remaining cases,
the M-model shows very competitive results.
Index Terms—telegrapher’s equations, distributed parameters,
lumped parameters, transmission lines, line modeling.

I. I NTRODUCTION
F someone opens any popular electric power engineering Fig. 2. Two-port network representation of a transmission line.
I textbook, he/she will realize that the most important stage
is the power flow (PF) analysis [1]–[4]. The reason is that all
the other stages (such as economic load dispatch, unit commit- Any transmission line with a length of 80 km (∼50 miles)
ment, optimal PF, fault analysis, contingency analysis, stability or less can be simplified as a short line model where the effects
and control, state estimation, protective relays coordination, of the per-phase capacitance (C) and conductance (G) are
etc) mainly depend on PF. Therefore, if the data received from neglected. The simplified circuit of this model is shown in
PF is incorrect or inaccurate, then the whole process will be Fig. 1. The goal here is to represent that circuit as a two-port
affected. However, the PF analysis itself depends on the quality network similar to the one shown in Fig. 2. Thus, the ABCD
and accuracy of the model used to represent real transmission parameters of Fig. 1 are [6]:
[ ] [ ][ ] [ ][ ]
lines. Thus, even going with highly precise PF solvers, weak VS 1 Z VR A B VR
representation of transmission lines leads to significant errors = = (1)
IS 0 1 IR C D IR
in all power system analysis.
where, by analogy, A = 1, B = Z, C = 0, and D = A.
In the circuit theory, some approximations are applied to
For medium-length transmission lines, where the total
simplify Maxwell’s equations of electromagnetism. For exam-
length exceeds 80 km (∼50 miles) but not longer than 250
ple, in the period between 1884-1885, Heaviside successfully
km (∼155 miles) [4], [6], four popular circuits could be used
re-wrote Maxwell’s 20 fundamental equations to obtain a
where the shunt capacitance is considered. These circuits are
new set of four compact equations, which were standardized
shown in Fig. 3 [7], [8]. The two-port network of these four
by the late 1890s [5]. To model real transmission lines as
lumped models are:
mathematical equations, there are two possible representations:
1) The lumped parameter model; also called the lumped • Nominal Γ-circuit:
[ ] [ ][ ]
component model and the lumped element model. VS 1 Z VR
2) The distributed parameter model. = (2)
IS Y 1 + ZY IR

978-1-7281-5442-8/20/$31.00
Authorized ©2020 of
licensed use limited to: Auckland University IEEE
Technology. Downloaded on December 22,2020 at 23:04:14 UTC from IEEE Xplore. Restrictions apply.
 2020 IEEE Canadian Conference on Electrical and Computer Engineering (CCECE)

II. M ATHEMATICAL M ODELLING OF THE M-M ODEL

In the preceding section, the known transmission line mod-

els have been introduced. This section presents the proposed
new transmission line model. Based on our knowledge with
(a) Nominal Γ-circuit (b) Nominal Γ-circuit
extensive search, there is no such model presented in the
literature. The idea behind developing this model is to pre-
cisely account for the variation in the distributed parameters
of a line under sag. Two possible scenarios could be faced
with real overhead transmission lines. For W towers installed
between the sending-end and receiving-end terminals, these
(c) Nominal Π-circuit (d) Nominal T-circuit two scenarios can be illustrated in Fig. 4. The single-line
Fig. 3. Popular medium-length transmission line models. diagram of the proposed lumped parameter model is shown in
Fig. 5. It is called the M-model because the lumped parameters
are arranged as the letter em. There are two approaches to
• Nominal Γ-circuit: model the left impedance (ZL ) and the right impedance (ZR ):
[ ] [ ][ ]
VS 1 + ZY Z VR
= (3)
IS Y 1 IR • The two quantities are unequal:

• Nominal Π-circuit: ∑
W −1

[ ] [ ][ ] ZL = zL,i (7)
VS 1( + ZY
2 ) Z VR i=1
= (4) −1
IS Y 1 + ZY 4 1 + ZY
2
IR ∑
W
ZR = zR,i (8)
• Nominal T-circuit: i=1
[ ] [ ( )] [ ]
VS 1 + ZY
2 Z 1 + ZY4 VR where zL,i is the impedance between the ith tower and
= (5)
IS Y 1 + ZY
2
IR the sag point, and zR,i is the impedance between the sag
The distributed parameter model is used when the length is point and the (i + 1)th tower.
longer than 250 km (∼155 miles). In this model, all the per- The two-port network of this approach can be established
unit-length passive components are considered to have very with the following matrix elements:
accurate calculations. The two-port network of this highly
precise model is: A = 1 + ZL YC + YS (ZL + ZR + ZL ZR YC ) (9)
[ ] [ ][ ] B = ZL + ZR + ZL ZR YC (10)
VS cosh (γL) Zc sinh (γL) VR
= 1 (6) C = YC + YS [2 + YC (ZL + ZR )]
IS Zc sinh (γL) cosh (γL) IR
+ YS2 (ZL + ZR + ZL ZR YC ) (11)
where Zc and γ are called the characteristic impedance and
the propagation constant, respectively [4], [6]. Also, L denotes D = A (12)
the total length.
• The two quantities are equal:
A. Main Contribution
W −1
This study tries to enhance the existing lumped model of 1 ∑
ZL = ZR = zi (13)
real transmission lines. A new lumped circuit called the M- 2 i=1
model is proposed, which is much superior to the other known
lumped parameter models (i.e., short-, Γ-, Γ-, T- and Π- where zi is the total impedance of the cable connected
models). To validate its accuracy, the performance is evaluated between the ith and the (i + 1)th towers.
against all the preceding lumped models for different currents The two-port network of this approach can be easily
and cable lengths. obtained by equating ZL and ZR . For not messing up,
let’s call them ZL = ZR = ZV . Thus, replacing ZL and
B. Paper Organization
ZR with ZV in (9)-(12) yields:
The remaining part of the study is arranged as follows:
Section II introduces the M-model and the fundamental equa- A = 1 + ZV (YC + 2YS + ZV YC YS ) (14)
tions derived from its circuit. The numerical experiments are
B = ZV (2 + ZV YC ) (15)
conducted in Section III with some discussion in Section IV.
Finally, the conclusion and scope of future work are presented C = (1 + ZV YS ) [YC + YS (2 + ZV YC )] (16)
in Section V. D = A (17)

Authorized licensed use limited to: Auckland University of Technology. Downloaded on December 22,2020 at 23:04:14 UTC from IEEE Xplore. Restrictions apply.
 2020 IEEE Canadian Conference on Electrical and Computer Engineering (CCECE)

(a) Inclined-spans (b) Leveled-spans

Fig. 4. Two different sag scenarios could be faced in real overhead transmission lines.

( )
γL
Ŷ coth 2
= Y γL
3
[ 2
√ ]
2 − 5 + 4 cosh (γL) + cosh (γL)
× √ (23)
3 − 5 + 4 cosh (γL) + 2 cosh (γL)
Fig. 5. Medium-length transmission line M-circuit. where the full derivations can be found in [9].

III. N UMERICAL E XPERIMENTS

To evaluate the performance of the M-model and having
a fair comparison with other nominal models, the following
three experiments are taken with no-sag effect:

A. Testing Transmission Line Models for Different Scenarios

Fig. 6. Equivalent M-circuit for the long line model. For this experiment, a special graphical user interface (GUI)
is designed using MATLAB. Figure 7 shows the program used
in this experiment. The Γ- and Γ-models will be entered into
the competition later for more comprehensive performance
If there is no sag, then the ABCD parameters of the nominal
comparison. As can be seen from GUI, the transmission line
M-model can be easily found by taking ZV = Z0 /2 and YC =
parameters and sending-end variables are given and then, from
YS = Y0 /3 as follows:
the models, the receiving-end variables are calculated. The
( ) performance of each calculated variable can be measured by
ZY ZY considering the long line model as a reference.
A = 1+ 3+ (18)
6 6 For this experiment, four different scenarios are covered.
( )
ZY Table I lists the line parameters and the sending-end variables
B = Z 1+ (19) used in each scenario. By taking the absolute difference
12
2
( ) between the lumped parameter models and the distributed
ZY ZY
C = Y + 2+ (20) parameter model, the performance of these lumped parameter
9 12
models can be evaluated. Table II shows the performance of
D = A (21) four different lumped parameter models. From this table, it is
obvious that the M-model is the closest one to the distributed
where Z0 and Y0 are the total series impedance and shunt parameter model. This is correct for all or most measurements
admittance under no-sag. The subscript “0” is dropped in (18)- recorded in each scenario. Even for a few records, where the T-
(21) to match with other nominal lumped models. model wins, the M-model can also provide highly competitive
Also, the ABCD parameters of the long-length transmission solutions.
line model given in (6) can be transformed into an equivalent
B. Testing Transmission Line Models by Varying IS
M-model similar to that shown in Fig. 6. This can be done by
using the following two equations: Except for the total length, the power factor and the sending-
end current, the line settings given in [10] are considered:
( ) L = 165 km, f = 50 Hz, R = 96.10 mΩ/km, C =
γL
Ẑ −Z tanh 2 0.010215 µF/km, L = 1.1051 mH/km, G = 44.5 nf/km,
= · γL VS = 275 kV and pf S = 0.9 lags.
2 4
[ √
2 ] The total length is calculated at the center between the mini-
× 1 − 5 + 4 cosh (γL) (22) mum and maximum allowable lengths of medium transmission

Authorized licensed use limited to: Auckland University of Technology. Downloaded on December 22,2020 at 23:04:14 UTC from IEEE Xplore. Restrictions apply.
 2020 IEEE Canadian Conference on Electrical and Computer Engineering (CCECE)

Fig. 7. Transmission line receiving-end analyzing software.

Fig. 8. Absolute errors between the lumped parameter models and the distributed parameter model—calculated by varying the sending-end current.

lines (i.e., L = 80 + 250−80

2 = 165 km), so it is not far Also, it can be seen that the Γ- and Γ-models could provide
away from the ranges of short- and long-length transmission good results. However, they last just for small periods before
lines. To generate 2D plots, the sending-end current (IS ) is the M-model wins again.
varied from 200 A to 1 kA. For each receiving-end variable,
the solutions obtained by the lumped parameter models are C. Testing Transmission Line Models by Varying Total Length
subtracted from that of the distributed parameter model (i.e., The same preceding line settings are considered in this
the long-length transmission line model) and then the absolute experiment. However, here, IS is set equal to 300 A with
values are taken. Figure 8 shows the absolute error plots of f = 60 Hz. To cover the three categories of transmission
these lumped parameter models. Except for VR and REG lines (i.e., the short-, medium-, and long-length transmission
where the T-model is the winner, it is clear that the M-model lines), L is varied from 10 to 1000 km. The absolute errors
performs better than all the other models. For PR and η, the of the receiving-end variables are plotted in Fig. 9 for all the
competition is almost equal between the M- and T-models. lumped parameter models. Again, the M-model proves itself

Authorized licensed use limited to: Auckland University of Technology. Downloaded on December 22,2020 at 23:04:14 UTC from IEEE Xplore. Restrictions apply.
 2020 IEEE Canadian Conference on Electrical and Computer Engineering (CCECE)

Fig. 9. Absolute errors between the lumped parameter models and the distributed parameter model—calculated by varying the total length.

TABLE I
PARAMETERS AND VARIABLES U SED IN E ACH S CENARIO where Y0 and Y are respectively the total line admittance
before and during sag.
Variables/ Scenario No. The question now is about how to find Y . We have devel-
Parameters 1 2 3 4
oped two innovative techniques to calculate Y directly with
VS (kV) 345 345 220 500
easy steps. Because of the limited space, these two techniques
IS (A) 400 570 1000 200
will be discussed in the future.
R (Ω/km) 0.0173 0.036 0.15 0.045
L (mH/km) 1.0105 0.9549 1.3263 0.97
C (µF/km) 0.0117 0.125 0.0538 0.0115 V. C ONCLUSION AND S COPE OF F UTURE W ORK
G (pf/km) 82.407 0 0 105.83 This study proposes a new lumped circuit to model real
f (Hz) 60 50 60 50
transmission lines. In addition to its capability to account for
L (km) 478 130 40 350
the sag effects on the distributed parameters, the steady-state
pf S 0.89 0.95 0.8 0.86
M-model can also beat the other known lumped parameter
Angle Mode lagging lagging leading lagging
models. From the numerical experiments, the M-model proves
that it can even be used for short- and long-length transmission
line models. Its absolute error is very low and thus it is very
as a highly precise approximator of the distributed parameter
close to the long line model.
model. Except for a few spots, the M-model wins in all the
Based on the encouraging results, it would be good to test
competitions.
the performance of this precise lumped parameter model for
different power system problems. The other useful applications
IV. F URTHER D ISCUSSION of this model could be utilized in future work where the
impacts of loading and surrounding weather conditions are
The preceding experiments are done by assuming that there
considered. To calculate the shunt admittance located at the
is no sag. For sag transmission lines, the M-model can also
center of the M-model, which is called the slack admittance,
be very useful. First of all, (18)-(21) are not valid for leveled-
two innovative techniques will be discussed in the future.
and inclined-spans sag scenarios shown in Fig. 4. The series
parameters, i.e. ZL and ZR , can be calculated either by (7) R EFERENCES
and (8) or by (13). For the shunt parameters, the two-end shunt
[1] M. E. El-Hawary, Introduction to Electrical Power Systems, Lajos
admittances (i.e., the left and right YS ) are not changed, and Hanzo, Ed. Hoboken, New Jersey: John Wiley & Sons Inc., 2008.
they are equal to the initial value (Y0 /3). Thus, the difference [2] J. D. Glover, M. S.Sarma, and T. J. Overbye, Power System Analysis
in the shunt admittance due to sag can be compensated by the and Design, 5th ed. Stamford, CT: Cengage Learning, 2012.
[3] J. J. Grainger and W. D. Stevenson, Power System Analysis. New York:
slack admittance (YC ) located at the center of the M-model. McGraw-Hill Education, 1994.
Thus, YC can be calculated as follows: [4] H. Saadat, Power System Analysis, ser. McGraw-Hill series in electrical
and computer engineering. Singapore: WCB/McGraw-Hill, 1999.
2 [5] B. J. Hunt, “Oliver Heaviside: A First-Rate Oddity,” Physics Today,
YC = Y − 2YS = Y − Y0 (24) vol. 65, no. 11, p. 48, Nov. 2012, [Accessed March 09, 2016].
3

Authorized licensed use limited to: Auckland University of Technology. Downloaded on December 22,2020 at 23:04:14 UTC from IEEE Xplore. Restrictions apply.
 2020 IEEE Canadian Conference on Electrical and Computer Engineering (CCECE)

[6] A. R. Al-Roomi and M. E. El-Hawary, “Effective Weather/Frequency-

TABLE II
Based Transmission Line Models—Part I: Fundamental Equations,” in
A BSOLUTE E RRORS B ETWEEN S OME L UMPED PARAMETER M ODELS AND
2017 IEEE Electrical Power and Energy Conference (EPEC), Oct 2017,
THE D ISTRIBUTED PARAMETER M ODEL - F IRST E XPERIMENT
pp. 1–6.
[7] G. Bousaleh, F. Hassoun, and R. Hage Chehade, “Study of Current
Receiving-End Performance of Scenario No.1
Distribution Over a Power Cable Presenting Non-Uniform Geometry
Variables Short T-Model Pi-Model M-Model Using the Partial Differential Equations Approach,” in Melecon 2010
VR (kV) 58.951 1.3581 2.9399 1.14950 - 2010 15th IEEE Mediterranean Electrotechnical Conference, April
2010, pp. 933–938.
θVR (◦ ) 3.008 0.8195 2.0311 0.96205 [8] Li Juan, Gao Hou-lei, Xue Yong-duan, Zou Gui-bin, and Xu Bing-
IR (A) 213.380 20.6670 14.0320 8.29750 yin, “Calculation Method for One-Section Lumped-Circuit Model of
Uniform Line,” in 2013 IEEE Power Energy Society General Meeting,
θIR (◦ ) 34.395 1.3360 0.5779 0.39890 July 2013, pp. 1–5.
pf R 0.199 0.0233 0.0268 0.01417 [9] A. R. Al-Roomi, “Improved Power System Realization and
Integration,” Ph.D. dissertation, Dalhousie University, Halifax, NS,
|SR | (MVA) 49.328 7.2420 8.9815 4.71110
Canada, Apr. 2020, [Accessed May 7, 2020]. [Online]. Available:
PR (MW) 2.543 0.4595 0.4515 0.14337 http://hdl.handle.net/10222/78503
QR (MVAr) 135.450 12.2460 14.6600 7.70650 [10] R. Benato, S. D. Sessa, and F. Guglielmi, “Determination of Steady-State
and Faulty Regimes of Overhead Lines by Means of Multiconductor
REG (%) 59.802 2.2789 3.4337 1.24420 Cell Analysis (MCA),” Energies, vol. 5, no. 8, pp. 2771–2793, Jul.
η (%) 1.195 0.2160 0.2122 0.06739 2012. [Online]. Available: https://doi.org/10.3390/en5082771
Receiving-End Performance of Scenario No.2
Variables Short T-Model Pi-Model M-Model
VR (kV) 33.457 0.2762 1.1294 0.46440
θVR (◦ ) 0.295 0.1296 0.2082 0.10545
IR (A) 670.560 28.4010 15.6880 9.98960 Ali R. Al-Roomi (S’13-M’14) received the B.Sc.
θIR (◦ ) 48.314 0.7901 0.3858 0.25697 degree in process instrumentation and control en-
gineering and the M.Sc. degree (Hons.) in electrical
pf R 0.488 0.0141 0.0090 0.00551
engineering from the University of Bahrain, Sakheer,
|SR | (MVA) 311.920 14.0030 10.4670 6.12870 Bahrain, in 2006 and 2014, respectively. He is cur-
PR (MW) 6.706 2.3172 0.5357 0.48106 rently pursuing the Ph.D. degree in electric power
systems engineering, Dalhousie University, Halifax,
QR (MVAr) 488.190 17.2610 12.3410 7.31230 NS, Canada.
REG (%) 24.360 0.3506 0.7229 0.27752 In 2007, he was with Moore Control and Engi-
neering (MCE) Middle East, Manama, Bahrain, as
η (%) 2.072 0.7161 0.1656 0.14867 a Project Engineer. In 2007, he joined Yokogawa
Receiving-End Performance of Scenario No.3 Middle East, Nuwaidrat, Bahrain, as a DCS Subsystem and Graphics Engineer.
From 2008 to 2012, he was with Aluminum Bahrain B.S.C., Askar, Bahrain,
Variables Short T-Model Pi-Model M-Model as an Instrumentation and Control Maintenance Engineer in its power stations,
VR (kV) 1.8617 0.012710 0.018336 0.009564 in both generation and auxiliary sides. From 2013 to 2015, he was an
Instrumentation, Control, and Protection Engineer with Riffa Power Station,
θVR (◦ ) 0.0911 0.012587 0.024283 0.012177
Riffa, Bahrain. His current research interests include power operation, power
IR (A) 66.2780 0.228710 0.098951 0.069297 protection, state estimation, smart grid, power system design and analysis,
θIR (◦ ) 4.9050 0.011438 0.006251 0.004046 system realization, load forecasting, machine learning, and meta-heuristic
algorithms.
pf R 0.0591 0.000270 0.000344 0.000183
|SR | (MVA) 29.9810 0.112880 0.069604 0.043444
PR (MW) 1.1416 0.015686 0.076499 0.035693
QR (MVAr) 45.4210 0.193580 0.198380 0.109570
REG (%) 1.5188 0.006145 0.006530 0.003586
η (%) 0.3745 0.005146 0.025095 0.011709 Mohamed E. El-Hawary (S’68-M’72-SM’76-F’90-
LF’09) received the B.Sc. degree (Hons.) in electri-
Receiving-End Performance of Scenario No.4 cal engineering from the University of Alexandria,
Variables Short T-Model Pi-Model M-Model Alexandria, Egypt, in 1965, and the Ph.D. degree
from the University of Alberta, Edmonton, AB,
VR (kV) 32.983 0.07269 0.96389 0.41627 Canada, in 1972.
θVR (◦ ) 0.444 0.05119 0.06614 0.03489 He was a Killam Memorial Fellow with the
University of Alberta. He has been a Professor of
IR (A) 278.940 7.13750 3.83260 2.47160 electrical and computer engineering with Dalhousie
θIR (◦ ) 39.456 0.46485 0.23479 0.15569 University, Halifax, NS, Canada, since 1981. He
pf R 0.500 0.00831 0.00483 0.00306 served on the Faculty and was the Chair of the
Electrical Engineering Department, Memorial University of Newfoundland,
|SR | (MVA) 203.260 5.43210 3.74290 2.24570 St. John’s, NL, Canada, for eight years. He was an Associate Professor of
PR (MW) 3.641 0.99558 0.30707 0.24800 electrical engineering with the Federal University of Rio de Janeiro, Rio
de Janeiro, Brazil, for two years, and an Instructor with the University of
QR (MVAr) 263.980 6.30090 4.19820 2.54480 Alexandria. He has authored many textbooks and monographs and over 160
REG (%) 15.810 0.11489 0.35744 0.14453 refereed journal articles. He pioneered many computational and artificial
intelligence solutions to problems in economic/environmental operation of
η (%) 2.445 0.66837 0.20615 0.16649
power systems.

Authorized licensed use limited to: Auckland University of Technology. Downloaded on December 22,2020 at 23:04:14 UTC from IEEE Xplore. Restrictions apply.