Professional Documents
Culture Documents
Unlocking Distribution Network Capacity Through Real-Time Thermalrating For High Penetration of DGs
Unlocking Distribution Network Capacity Through Real-Time Thermalrating For High Penetration of DGs
a r t i c l e
i n f o
Article history:
Received 5 December 2013
Received in revised form 12 May 2014
Accepted 31 July 2014
Keywords:
Active network management
Distributed generation control
Load forecasting
Power ow
Real-time thermal rating
State estimation
a b s t r a c t
Highly stochastic loading in the emerging active distribution networks means that electric utilities need
to use their assets to the fullest by deploying intelligent network management tools. Real-time thermal
rating (RTTR) provides possibility for short term and even real-time active distribution network management enabling the network to run closer to an overload state without damage. In this study, a RTTR
based active distribution network management framework is formulated giving hour-by-hour network
capacity limits. Relationships of stochasticities in customer loads and DG output with thermal responses
of underground cables, overhead lines and distribution transformers are explained. RTTR is applied on all
distribution network components with simulated scenarios involving various levels of DG penetration.
This study quanties the potential for an increased DG utilization and an increased potential for new DG
installations when RTTR is integrated with distribution management systems.
2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND
license (http://creativecommons.org/licenses/by-nc-nd/3.0/).
1. Introduction
An increasing number of renewable energy resources are
being installed in distribution systems. Larger units, usually above
10 MW, are installed by commercial power producers and are generally connected to transmission facilities. Smaller units which are
no larger than 1 or 2 MW are, however, installed in distribution
systems and referred to as distributed generation (DG) [1]. It is
anticipated that with the proliferation of DG some distribution
networks will face power ow congestion due to the thermally vulnerable components restricting the connection capacity and active
energy yield of DG [2]. In addition, switching operations that recongure the feeder will be common in distribution systems with the
advent of DGs [3].
Nevertheless, traditionally, the main task of distribution system
analysis tools has been to solve the power ow for one specic
point in time, which is the predicted peak demand. To increase DG
connectivity and also to harness the generated renewable energy
efciently there is a need for active distribution management
systems (DMS). It is also in the interest of distribution network
operators (DNOs) to increase asset utilization in a safe manner,
potentially allowing latent capacity to be used under strictly controlled conditions [2]. Currently, the thermal limits applied by DNOs
tend to be based on xed or assumed meteorological conditions
http://dx.doi.org/10.1016/j.epsr.2014.07.032
0378-7796/ 2014 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).
M.Z. Degefa et al. / Electric Power Systems Research 117 (2014) 3646
37
Table 1
Radial distribution network data.
1 Primary substation (40 MVA, 110/20 kV)
16 secondary substations (1.6 MVA, 20/0.4 kV)
Heating type
No.
1800 households
Direct electric (DE)
District heating (DIST)
Ground source HP (GSHP)
Electric storage heating (STORE)
673
960
109
58
38
M.Z. Degefa et al. / Electric Power Systems Research 117 (2014) 3646
kWh/ hour
0
2
1
0
Ground Source Heatpump Household
5
3
0
10
5
0
01-Jul
20-Aug
09-Oct
28-Nov
17-Jan
Time
08-Mar
27-Apr
16-Jun
Fig. 2. Household types and their typical hourly load prole for one year (July, 2008 to June, 2009).
Table 2
Distribution of LV customers into the four heating types under the sixteen secondary
substations.
No.
DE
DIST
GSHP
STORE
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
151
116
85
101
20
101
99
0
0
0
0
0
0
0
0
0
6
0
0
0
75
8
0
213
126
150
84
70
105
119
0
4
0
0
0
0
0
0
0
0
0
0
0
0
3
4
56
46
0
0
0
0
0
0
0
5
26
0
0
27
0
0
0
0
however this would incur 1% loss of life per emergency if the emergency loading is maintained for 1 h. Hence, in this study dynamic
thermal rating utilization is considered in real-time normal network operation where the compromise of loss of life is avoided.
Component thermal models are pre-requested for a power
system real-time thermal rating. In this study the three major
distribution network components are used to compute the system rating. These components are overhead lines, electric cables
and power transformers, and their respective thermal models are
used. Except for cables buried inside unlled conduit and secondary
transformers installed in cabins, IEEE standards are used to develop
the thermal models. For cables inside unlled conduit a suitable
air-gap thermal model has been proposed by the authors in [14]. A
dynamic thermal model for prefabricated MV/LV substations, presented in [15], is used for secondary substation transformer thermal
ratings.
3.1. Component thermal modeling
The component models can either be dynamic or static. The
dynamic model involves time constants, which are functions
of thermal resistances and capacitances emulating the transient
responses. The time constant for overhead bare lines is very small,
as the maximum possible hotspot steady state temperature for
a given loading scenario shows up in the order of minutes. IEEE
standard [16] is used for the thermal modeling of overhead lines
and [13] is used for oil-immersed distribution transformer indoor
installations. However, for underground cables installed inside
M.Z. Degefa et al. / Electric Power Systems Research 117 (2014) 3646
39
Underground cable
conductor temperature (degC)
Fig. 3. Seven loop thermal model for underground cable inside unlled conduit.
55
47.63
45
40
35
30
25
20
15
0
a)
5 7.9 10
20
25
Hours
30
35
40
45
90
80
70
62.42
50
40
30
20
10
0 0.250.5
Underground cable
100
1.5
2.5
3
Hours
3.5
4.5
Hotspot
Top Oil
90
Temperature (degC)
15
100
Component
71.26
60
53.87
50
0.25
0.358
3.733
80
40
30
20
0.3583
3.733
10
Hours
1400
DTR
Static Seasonal
1000
600
250
01-Jul
20-Aug
09-Oct
28-Nov
17-Jan
08-Mar
27-Apr
16-Jun
20-Aug
09-Oct
28-Nov
17-Jan
Time
08-Mar
27-Apr
16-Jun
50
25
0
-25
01-Jul
Fig. 5. RTTR vs Static rating for overhead bare line (July, 2008 to June, 2009).
40
M.Z. Degefa et al. / Electric Power Systems Research 117 (2014) 3646
60
50
50
40
40
kWh/hour
Underground cable
conductor temperature (degC)
(a)
30
20
10
20
10
0
01-Jul
12-Sep
24-Nov
05-Feb
19-Apr
01-Jul
0
01-Jul
Time
(b)
30
20-Aug
09-Oct
28-Nov
17-Jan
Time
08-Mar
27-Apr
16-Jun
400
Fig. 7. One year hourly generated energy from a 50 kW wind turbine (kWh/h) (July,
2008 to June, 2009).
300
200
100
0
01-Jul
12-Sep
24-Nov
05-Feb
19-Apr
01-Jul
Time
Fig. 6. One year conductor temperature (a) and the corresponding hourly loading
prole (b) of an underground cable (July, 2008 to June, 2009).
3
i=1
i Yi,t +
4
j=1
j xj,t +
2
k Dk,t + t
(1)
k=1
t N(0, 2 )
where yt is the output (target variable); C is constant or the intercept of the model; Yi,t : lagging output values.
They include:
Previous Day Maximum Load
Previous Week Same Hour Load, Yi,t168
Previous Day Same Hour Load, Yi,t24
Dk,t denotes external inputs (explanatory variables).
They include:
S wmeas + K,
wcut
in
Pwind =
Pnomin,
Pnomin
S=
wnomin wcut
K = S wcut
wmeas wnomin
(2)
in
in
Pnomin is the rated power; wnomin is the wind speed at rated power;
wcut in is the cut-in wind speed (m/s); wmeas is the wind speed (m/s);
S is the slope; K is the constant.
4.2. Solar panel model
A practical model developed by Jones and Underwood for the
production of optimal output power from a photovoltaic module
is used in this study [19]. The approach uses the simple relationships from the diode model of irradiance and temperature with
short circuit current and open circuit voltage to calculate maximum
power output. In Eq. (3), except for ambient temperature and solar
irradiance values, all other values can easily be taken from the manufacturers datasheet of the considered module. The one year hourly
weather data for ambient temperature is from the Finnish Meteorological Institute (FMI), and the solar irradiation data is from the
Solar Energy Services for Professionals (SoDa) website [18,20]. The
M.Z. Degefa et al. / Electric Power Systems Research 117 (2014) 3646
0.15
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
01-Jul
20-Aug
09-Oct
28-Nov
17-Jan
Time
08-Mar
27-Apr
16-Jun
Fig. 8. One year hourly generated energy from a single 215 Wp solar panel (kWh/h)
(July, 2008 to June, 2009).
Ppv
max
FF =
= FF
G
Isc
Gref
Voc
Tjref
ln(P1 G)
Tj
ln(P1 Gref )
PVmax
Voc Isc
(3)
Isc
P1 =
G
Tj = Tair +
NOCT 20
S
80
Tj is the PV cell temperature (K); Tjref is the reference cell temperature (K); FF is the lling factor; S is the insolation (mW/cm2 ); Isc
is the short circuit current (A); Voc is the open circuit voltage (V);
PVmax is the maximum power under Standard Test Condition (STC)
(for our PV: irradiance of 1000 W/m2 and cell temperature of 25 C);
Tair is the air temperature (degC); G is the irradiance (W/m2 ); Gref
is the standard irradiance (W/m2 ); PVmax is the maximum power
under STC conditions (W); Ppv max is the maximum power output
(W).
5. Algorithm for RTTR and optimal power ow
The OPF is a technique deployed in power systems to address
problems ranging from economic dispatch to loss minimization
[21]. In this study the OPF function utilizes static rating or realtime thermal rating of the system to decide the permissible loads
of network components. Whenever static rating or thermal violation is experienced, only curtailment of DGs or loads is considered
as correction measures, as shown in the ow chart in Fig. 9. The
objective for the OPF is to minimize the total cost of load and DG
curtailment whenever ratings are exceeded. Within the scope of
this paper, only the potential that a dynamic thermal rating function brings to a distribution network has been shown, especially
with the increasing connection of distributed generation. Real-time
network reconguration can also be an alternative remedy for network capacity violations, but it is not considered in the presented
framework.
As shown in Fig. 9 there is a decision making step depending either on whether thermal violation or static rating violation
is detected. The logic behind the limit violation decision is given
in Table 3. Since the loading scenario in this study is hourly, the
emergency limits of components with less than an hour loading
limit are not an option. For oil immersed transformers, normal
41
i,t i,t
i,t i,t
i,t i,t
(lc
Plc + gc
Plvgc + gc
Pmvgc )
(4)
i
i,t
where t and i are indices for time span and buses. lc
is com-
i,t
i,t
and gc
is compensation cost
pensation cost for curtailed load Plc
i,t
for curtailed low voltage side generation (Plvgc
) and medium volt-
i,t
) at time span t and bus i. The objective
age side generation (Pmvgc
function (4) is subjected to the following constraints:
Table 3
Static loading thermal limits for decision making.
ij
ij
ij
ij
ij
ij
ij,t
ij,t
1 h maximum
OH (conductor)
UG (conductor)
TRAFO (hotspot)
Pf
Static rating
Temperature
430 A
110 C
330 A
90 C
1600 kVA
110 C/120 C
Qf
ij
(5)
ij
(6)
cos(f )
sin(f )
42
M.Z. Degefa et al. / Electric Power Systems Research 117 (2014) 3646
Table 4
DGs prevalence scenarios for MV distribution network operating radially.
Scenario#1
Scenario#2
Scenario#3
Scenario#4
Scenario#5
Scenario#6
Scenario#7
Scenario#8
ij,t
Sf
=
ij,t 2
Pf
DG level (out
of 40 MVA)
DG cumulative
maximum (MVA)
0
50%
100%
110%
120%
150%
200%
315%
0
20
40
44
48
60
80
126
ij,t 2
+ Qf
WIND
SOLAR
Node 12 (number
of 50 kW turbines)
Node 14 (number
of 50 kW turbines)
Total
(kVA)
Node 10 (number of
215 Wp solar panels)
Node 12 (number of
215 Wp solar panels)
Total
(kVA)
0
107
214
235
257
321
428
675
0
285
570
627
684
855
1140
1799
0
19,600
39,200
43,120
47,040
58,800
78,400
123,676
0
1017
2034
2237
2441
3051
4068
6418
0
1763
3525
3878
4231
5288
7051
11,125
0
400
800
880
960
1200
1600
2524
(7)
i,t
i,t
Pmvg
Pmvgc
Psi,t
ij,t
Pf
=0
(11)
j
i,t
i,t
i,t
Psi,t = Pli,t Plc
Plvg
+ Plvgc
(8)
Qsi,t
(9)
Qli,t
Qlci,t
i,t
Qlvg
i,t
+ Qlvgc
i,t
i,t
Qmvg
Qmvgc
Qsi,t
ij,t
Qf
=0
(12)
Ssi,t =
Psi,t + Qsi,t
(10)
Vi Vit Vi
(13)
M.Z. Degefa et al. / Electric Power Systems Research 117 (2014) 3646
43
Table 5
Comparison of real-time thermal rating and static rating for different penetration levels of DGs.
Number of hours in year
Violations of RTTR
Load curtailed
Load curtailed
0
54
430
433
603
618
893
1295
0
0
191
258
294
440
624
1087
0
0
0
0
0
0
0
0
DG curtailed
0
77.65
2243.73
2993.36
3858.65
6881.58
13,111.73
31,224.89
1000
500
-400
01-Jul
20-Aug
09-Oct
28-Nov
17-Jan
Time
08-Mar
27-Apr
16-Jun
i,t
0 Plc
Pli,t
(14)
i,t
i,t
0 Plvgc
Plvg
(15)
i,t
i,t
0 Pmvgc
Pmvg
(16)
i,t i,t
Plc
Ql Qlci,t Pli,t = 0
(17)
i,t
i,t
Plvgc
Qlvg
(18)
i,t
i,t
i,t
i,t
Pmvgc
Qmvg
Qmvgc
Pmvg
=0
Current Load
(19)
i,t
i,t
i,t
i,t
(Qlvg
) and Plvgc
(Qlvgc
)
active (reactive) powers at time t and bus i. Plvg
are LV side generation and curtailment of active(reactive) power.
i,t
i,t
i,t
i,t
Pmvg
(Qmvg
) and Pmvgc
(Qmvgc
) are MV side generation and curtailij
300
250
37
200
35
33
150
31
100
29
50
27
25
Fig. 10. Secondary substation 12 transformer load for scenario#6 (July, 2008 to June,
2009) (negative power is associated with LV connected solar panels power supply
to MV network).
=0
0
0
969.97
1434.80
1972.13
4046.98
8822.90
23,594.88
39
0
0
i,t
i,t
Qlvgc
Plvg
DG curtailed
0
0
0
0
0
0
0
0
Temperature
41
Underground cable conductor
temperature (degC)
S1
S2
S3
S4
S5
S6
S7
S8
Violations of static
rating
DG scenario
10
15
20
Hours
Fig. 11. Line section 12 underground cable forecasted load and the corresponding
conductor temperature for scenario#3.
ij,t
Table 6
Percentage of load increment upto the static and RTTR rating limits for the 16 secondary substations.
Substation number
SUB#1
SUB#2
SUB#3
SUB#4
SUB#5
SUB#6
SUB#7
SUB#8
SUB#9
SUB#10
SUB#11
SUB#12
SUB#13
SUB#14
SUB#15
SUB#16
1.2
0.970
0.635
0.839
0.317
0.760
0.810
0.578
0.533
0.415
0.322
0.503
0.283
0.382
0.480
0.371
RTTR (120 )
Difference
Overloaded
20.64%
84.33%
39.50%
269.46%
53.89%
44.42%
102.41%
119.69%
181.84%
262.86%
132.75%
312.70%
206.57%
143.93%
215.53%
10.00%
40.00%
100.00%
60.00%
310.00%
70.00%
60.00%
170.00%
150.00%
230.00%
530.00%
170.00%
400.00%
260.00%
170.00%
240.00%
10.00%
19.36%
15.67%
20.50%
40.54%
16.11%
15.58%
67.59%
30.31%
48.16%
267.14%
37.25%
87.30%
53.43%
26.07%
24.47%
44
M.Z. Degefa et al. / Electric Power Systems Research 117 (2014) 3646
that the power factor remains the same after and before either
load or generation curtailment.
Although the formulation presented in (4) is general, a DC power
ow is used in both pre-optimization load ow and the optimal
power ow of the test case scenarios (see Fig. 9). DC power ow
is used for its computational simplicity. It should be noted that
the accuracy of the DC power ow model is not of concern, since
the focus of this paper is on capacity limit violations (not under
voltage). Besides, for the test network utilized in this study, the
difference between AC and DC power ow has been checked to be
trivial. The real-time thermal rating is solved in MATLAB whereas
the optimal power ow is solved by General Algebraic Modeling
System (GAMS).
Fig. 12. Unserved DG generation due to capacity limit with RTTR and Static rating.
M.Z. Degefa et al. / Electric Power Systems Research 117 (2014) 3646
45
Table 7
Percentage of DG installment capacity from annual peak up to the static and RTTR rating limits for the 16 secondary substations.
Substation number
Percentage of Solar panel installment compared to the peak load connected to each secondary substation
SUB#1
SUB#2
SUB#3
SUB#4
SUB#5
SUB#6
SUB#7
SUB#8
SUB#9
SUB#10
SUB#11
SUB#12
SUB#13
SUB#14
SUB#15
SUB#16
RTTR (120 )
Difference
1.2
0.970
0.635
0.839
0.317
0.760
0.810
0.578
0.533
0.415
0.322
0.503
0.283
0.382
0.480
0.371
90.00%
130.00%
200.00%
150.00%
390.00%
170.00%
160.00%
220.00%
240.00%
300.00%
370.00%
250.00%
430.00%
320.00%
260.00%
330.00%
170.00%
150.00%
230.00%
170.00%
440.00%
240.00%
180.00%
300.00%
270.00%
340.00%
420.00%
280.00%
490.00%
370.00%
340.00%
380.00%
80.00%
20.00%
30.00%
20.00%
50.00%
70.00%
20.00%
80.00%
30.00%
40.00%
50.00%
30.00%
60.00%
50.00%
80.00%
50.00%
S8
S7
Scenarios
S6
S5
S4
S3
S2
S1
0
100000
200000
300000
400000
emphasized the need for a deeper investigation of the stochasticity of loading due to intermittent DG output and its impact on the
mechanical strength of components.
The implementation of price and network based demand
response programs, the pervasiveness of electric vehicles, the
increasing installation of DGs, the rising customer load base of
urban areas and other emerging factors are creating a distribution
network loading curve which is stochastic in nature and previously
unseen in shape. Hence, static distribution network rating is no
longer a good enough network management tool. Rather, utilities
need to adapt to the dynamism of load and generation behavior by
applying programs such as RTTR for efcient and secure network
operation.
Acknowledgements
The authors of this paper would like to acknowledge that this
work is jointly funded by the Aalto Energy Efciency Research Program (AEF) through the SAGA and STEEM projects and, the SGEM
project.
Appendix A.
Line and cable parameters
Underground cable (240Wiski)
Overhead line (Al132)
Res (/m)
React (/m)
0.000138
0.000279
0.00011
0.000344
330
430
References
[1] R.C. Dugan, T.E. McDermott, Operating conicts for distributed generation on
distribution systems, in: IEEE Rural Electric Power Conference, 2001, A31.
[2] S.C.E. Jupe, P.C. Taylor, Distributed generation output control for network power
ow management, IET Renew. Power Gener. 3 (2008) 371386.
[3] R.F. Arritt, R.C. Dugan, Distribution system analysis and the future smart grid,
IEEE Trans. Ind. Appl. 47 (2011) 23432350.
[4] A. Neumann, P. Brinckerhoff, Unlock distribution capacity using dynamic thermal ratings, Energize (2008) 1819.
[5] D.A. Douglass, D.C. Lawry, A.A. Edris, E.C. Bascom, Dynamic thermal ratings
realize circuit load limits, IEEE Comput. Appl. Power 13 (2000) 3844.
[6] A. Michiorri, P.C. Taylor, Forecasting real-time rating for electricity distribution
networks using weather forecast data, in: 20th International Conference on
Electricity Distribution, CIRED, 2009, pp. 14.
[7] R.C. Dugan, T.E. McDermott, G.J. Ball, Planning for distributed generation, IEEE
Ind. Appl. Mag. 2 (2001) 8088.
[8] R.A. Walling, R. Saint, R.C. Dugan, J. Burke, L.A. Kojovic, IEEE summary of distributed resources impact on power delivery systems, IEEE Trans. Power Deliv.
23 (2008) 16361644.
[9] R.A.F. Currie, G.W. Ault, J.R. McDonald, Methodology for determination of economic connection capacity for renewable generator connections to distribution
46
[10]
[11]
[12]
[13]
[14]
[15]
M.Z. Degefa et al. / Electric Power Systems Research 117 (2014) 3646
networks optimized by active power ow management, IEE Proc. Gener.
Transm. Distrib. 153 (2006) 456462.
A. Borghetti, M. Bosetti, S. Grillo, S. Massucco, C.A. Nucci, M. Paolone, F. Silvestro,
Short-term scheduling and control of active distribution systems with high
penetration of renewable resources, IEEE Syst. J. 4 (2010) 313322.
H.T. Yip, C. An, G.J. Lloyd, P. Taylor, Dynamic thermal rating and active control
for improved distribution network utilization, Dev. Power Syst. Prot. (2010)
15.
R.J. Millar, E. Saarijrvi, M. Lehtonen, M. Hyvrinen, J. Niskanen, P. Hmlinen,
Electricity distribution network planning algorithm based on efcient initial and radial-to-full network conversion, Int. Rev. Electr. Eng. 8 (2013)
10761090.
IEEE Guide for Loading Mineral-Oil-Immersed Transformers, IEEE Std C57. 912011.
M.Z. Degefa, M. Lehtonen, R.J. Millar, Comparison of air-gap thermal models for
MV power cables inside unlled conduit, IEEE Trans. Power Deliv. 27 (2012)
16621669.
M.Z. Degefa, R.J. Millar, M. Lehtonen, P. Hyvnen, Dynamic thermal modeling of MV/LV prefabricated substations, IEEE Trans. Power Deliv. 29 (2014)
786793.
[16] IEEE Standard for Calculating the Current Temperature Relationship of Bare
Overhead Conductors, IEEE Std. 738-2006.
[17] R.J. Millar, M. Lehtonen, A robust framework for cable rating and temperature
monitoring, IEEE Trans. Power Deliv. 21 (2006) 313321.
[18] FMI Finnish Meteorological Institute: http://en.ilmatieteenlaitos./
[19] A.D. Jones, C.P. Underwood, A modeling method for building-integrated photovoltaic power supply, Build. Serv. Eng. Res. Technol. 23 (2002) 167177.
[20] Solar
Energy
Services
for
Professionals,
http://www.soda-is.com/eng/index.html
[21] M.Z. Degefa, R.J. Millar, M. Koivisto, M. Humayun, M. Lehtonen, Load ow analysis framework for active distribution networks based on smart meter reading
system, Engineering 5 (2013) 18.
[22] S. Kazemi, Reliability Evaluation of Smart Distribution Grids (Ph.D. Dissertation), Depts. Elec. Eng., Sharif University of Technology and Aalto University,
2011 http://lib.tkk./Diss/2011/isbn9789526042411/
[23] J.Z. Hansen, Failure in MV joints (XLPE cable) in heavy loaded cable systems
connecting large windmills to the distribution system, in: NORDAC, Danish
Energy Association, 2012.
[24] G. Celli, E. Ghiani, F. Pilo, G.G. Soma, Reliability assessment in smart distribution
networks, Electr. Power Syst. Res. 104 (2013) 164175.