Beruflich Dokumente
Kultur Dokumente
INTRODUCTION
The electric power business is rapidly becoming market driven. However, because of
the increasingly intimate role that electric plays in the national economy, security
remains to the most important aspect of power system operation which cannot be
compromised in a market-driven approach. Recent developments based on the
standard market design (SMD) in restructured electric power system provide an
opportunity for electricity market participants, such as generation companies,
transmission companies, and distribution companies to exercise least-cost or profit-
based operations. In a competitive electricity market, the ISO coordinates the SMD
attributes with market participants for satisfying hourly load demand, limited fuel and
other resources, environmental constraints, and transmission security requirements.
In competitive electricity market, customers expect a least-cost and high-quality
supply of electric energy, which may require additional investments and more
sophisticated operation techniques for enhancing power system security. The
sophistication could in part mitigate severe consequences in the event of cascaded
power system contingencies, which might otherwise result in dramatic property and
human losses, and severely impede the growth in the national economy.
The reliability of a power system is generally judged in terms of the systems
adequacy and security. The term adequacy refers to the systems capacity to meet the
load demand within the components ratings and voltage limits at any time. The term
security refers to the systems ability to withstand the impact of sudden changes due to
equipment outage, such as the loss of a generator, transmission line, etc. The security
of the system is defined in terms of a list of contingencies (i.e., transmission-line and
generator outages) which may cause insecure operation. Clearly, as the system
conditions changes, this list also changes. In order to determine the list of
contingencies, exhaustive load flow analyses should be performed to determine the
impact of each contingency on the set of contingencies which may create problems.
Power system security involves practices designed to keep the system operating when
the components fail.
Examples
1. A generating unit may have to be taken off-line because of auxiliary equipment
failure.
2. By maintaining the proper amount of spinning reserve, the remaining units on the
system can take the deficit without too low a frequency drop or need to shed load.
3. A transmission line may be damaged by a storm and taken out by automatic
relaying.
When one failure results in another failure in the system, it is called cascading failure
which leads to system blackout.
For Good quality of power transfer from source to load operator must have to follow
these constraints.
1. Reliability:-
2. Security:-
3. Stability:-
4. Economic:-
5. etc
The third method security function is security-constrained optimal power flow. In this
function, a contingency analysis is combined with an optimal power flow which seeks
to make changes to the optimal dispatch of generation, as well as other adjustments, so
that when a security analysis runs, no contingencies result in violation.
To understand the above functions, we shall divide the operating states of the power
system into four types:
1. Optimal dispatch
2. Post-contingency
3. Secure dispatch
4. Secure post-contingency
1. Optimal dispatch: It is the state that the power system is in prior to any contingency.
It is optimal with respect to economic operation but it may not be secure.
2. Post-contingency: It is the state of the power sytem after contingency has occurred.
We shall assume here that this state has a security violation (transmission line or
transformer outside its flow limit or bus voltage outside limit).
3. Secure dispatch: It is the state of the power system with no contingency outages but
with corrections to the operating parameters to account for security violation.
4. Secure post-contingency: It is the state of the power system when contingency
analysis is applied to base operating condition with corrections.
It is illustrated with an example. Suppose a trivial power system consisting of two
generators, a load and double circuit line with both generators supplying load. Figure
shows that the system is in economic dispatch that is 500 MW from unit 1 and 700
MW from unit 2 is in optimal dispatch. Further, the two double circuit lines can carry
a maximum of 400 MW so that there is no loading problem with respect to the base
operating condition.
1. Zhihong Jia et.al.[] Here in this paper author discussed about the two available
methods of contingency ranking and screening as continuation power flow
(CPF) and Generalized curve fit (GCF). The new method proposed in this
paper was derived from GCF and the result of the proposed approach was
compared with CPF method. The method calculate the voltage stability margin
considered branch outage . The result based on the IEEE 300 bus and IEEE-
600 bus power system as well as utility system 197 bus and 1254 bus were
presented. Here computation and simulation was performed using IPFLOW
and VSTAB.
3. Robert Fischl et.al.[34] presented the theory and method for systematically
finding the performance index(PI) which was used in automatic contingency
selection (ACS) algorithm. The author solved how to select the set of
weighting coefficients in the currently used PIs for analyzing either the real
power flow or node voltage magnitude problems in order to circumvent some
of the contingency ranking problem .Even more information it was shown how
to select the threshold values of the PI which guarantee proper classification of
the contingencies .Author discussed about the method for finding PI value
using standard SUMI method. One such algorithm was given together with an
illustrative example.
4. K.W.Chan et. al. [ ] described the security analysis methods adopted by
OASIS with an explanation and justification of why they can selected or
developed. An on-line dynamic security contingency screening and ranking
system for large complex power system, online algorithms for system
instability studied (OASIS) was developed. Result and discussion of the trails
carried at the national grid control center NGCC) were also presented.
6. G.B.Jasman et. al. [ ] presented a new technique for ranking outage in power
system. The method of outage simulation by power system injection was used
in this paper and the result of the outage was then further analyzed to
determine the level of voltage instability of the network as a result of the
outage. Result was tested on a 9-bus system.
9. L.D.Arya et.al.[] described a technique for ranking line outage causing voltage
limit violation taking into account the post outage correctability of the network
during contingency selection process. A computationally efficient algorithm
for ranking line outage has been developed taking into account the post outage
corrective capability of the power network and avoiding masking in
optimization and ranking solution to such a problem is computationally
extremely demanding. Two set of compensated shift factor were derived on
IEEE-25,30,91 bus system.
10. A.Mohamed et.al.[ ] presented a fast and accurate technique for ranking
transmission line outage according to the severity of their effects on busbar
voltages. To simulate each contingency case the proposed technique used the
DC formulation with compensation method to compute the post outage angle
and super decoupled load flow formulation and compensation method to
derive the post outage voltage. The aim was to reducing the number of line
contingency cases necessary for the assessment of the voltage security level of
a power system. The result were tested on 14-bus ,24-bus 59bus system.
11. F.Albuyeh et.al.[36] showed that there was little correction between the
contingencies that produce line overload and those that result in unacceptable
voltage profiles. The author provides different approaches for ranking of
contingency analysis and among all the methods reported in the paper, the
distribution factor method by El-Abiab and Stagg was the most widely used.
Fast decoupled power flow used to calculate the performance indices for every
outage. Iterative linear power flow (ILPF) was substituted for the fast
decoupled load flow, but the result of case study was not accurate as that
obtain by ne iteration of the fast decoupled method. The technique presented
here by author was especially flexible for the purpose of accuracy and speed
and that can be trend very easily by just varying the number of iteration. The
main reason for having contingency selection in analyzing online security was
to minimize the computational requirement.
12. S.N.Singh et.al. [ ] suggested simple and efficient method for optimal selection
of weights along with higher order performance indices for voltage
contingency selection. Here author also reviewed existing performance indices
for voltage contingency selection. The proposed performance indices were
able to eliminate misranking and masking effects. The proposed algorithm has
been tested on IEEE-14 bus system and a practical 75-bus Indian system.
13. P.R.Bijwe et. al. [ ] presented a new point for line capability of the power
system for determination of optimum post outage generation corrections a
noval efficient method which was the corner stone of the ranking procedure
presented in this paper. The sat of GSDFs and loss formulas were therefore
applicable for all optimization overload ranking results for the 25 and 95 bus
system with and without post outage corrective rescheduling had obtained.
14. Chong Suk Song et. al.[ ] proposed a algorithm by utilizing the formulation of
PTDFs and LODFs tp calculate the generation shift so that power flow on
transmission line that violate security limits due to single and multiple outage
event were adjusted to its security limits. This paper used the MATLAB
software to build the application where an algorithm was used to suppress the
overloading on a line following a single line and multiline fault or during a
heavy load condition. This program was applied on the new England 10
generator 39-Bus system.
15. Amit Kumar Chowdhury et. al. [ ] discussed about security aspects of power
system by evaluating the severity o transmission line outage. The severity of
contingency was measured using a scalar index called voltage performance
index 1P-1Q method and fast decouple load flow were used as approximation
and exact load flow method for voltage security assessment respectively. The
result of severity of line was evaluated and tested on IEEE-5 bus ,14bus , 30
bus and compare using these load flow method.
16. L.D.Arya et.al.[ ] described a method for security control i.e. line overload ,
alleviation using the switching under a line outage condition. This assessment
was made in pre-contingencies case using the line outage distribution factors.
The author represented a newly developed method which quickly calculate line
flow after a contingency and switched line. The algorithm had implemented on
a 6 bus 11 line and 25 bus test system with help of PC-AT using turbo
PASCAL.
17. Yung-Chung Chang et. al. [ ] presented two sensitivity factors in terms of the
generation shift distribution factor (GSDF) to improve some defects of the
conventional formula for the line outage distribution factor (LODF). The
sensitivity factor were established by using the concepts of generation change
and power injection to simulate the outaged line flow and one of them was
then applied to the case of line addition. A transfer factor was derived to
calculate the power flow of the added line very quickly from the relationship
between the OPF and GSDF. The proposed method had tested by means of a
standard system compare with the conventional OPF in the line outage case
and with DC load flow in line addition case.
Chapter-3
Contingency analysis
Inverse of impedance
In this method only the real part of the power flow equations are considered, that is
the effect of reactive power Q is neglected and all the bus voltages are assumed to be 1
p.u. the matrix B' is computed on the basis that all the resistances are zero from
equation
In this work, the effort has been given on contingency ranking. At starting the
contingency list is formed. Which contains those cases whose probability of getting
outage is found to be high? The list is generally large, is translated in to electrical
network changes, mostly generator and/or line outages. Contingency evaluation using
load flow is then carrying out on the following individual outage in decreasing order
of severity. The activity will be remained up to the point where no post contingency
violations are undergone.
Concept of contingency was introduced by Ejebe and Wollenberg. Contingency is
termed as a unpredictable event in the power system where outage is referred as
temporary suspension of power. Contingency therefore defined as possible
circumstance or as failure outage which is possible but cannot be predicted with
certainty. Contingency analysis need the simulation of each contingency for the given
power system model. In order to analyze the contingency analysis easier it comprises
of three important steps.
Contingency analysis procedure
Contingency creation/definition it is the initial step of contingency analysis. It is
made up of all set of viable contingencies that may happen in a power system. This
process consists of making contingency lists.
Contingency selection it is the second step in contingency analysis. It is the process
which includes finding of severe contingencies from all that may cause to violate bus
voltage and power through lines. Here in these procedure contingencies that list is
reduced by rejection of least severe contingencies and taking into consideration of
most severe outages. In this process the performance index has been used to find the
most severe ones.
Contingency evaluation- is is the third step and the most significant step as it includes
necessary control and security actions which are required in order to reduce the efforts
of most severe contingencies in a power system.
Power engineers need to concentrate in this analysis for power security purpose to
fulfill the motive of supplying power without any interruption i.e. continuity of
supply.
A power system is monitored through the supervisory control and data acquisition
(SCADA) system installed at control center. SCADA collects real-time data from
remote terminal units (RTUs) or now days by using Phase Measurement Unit
(PMUs) installed in substations and power plant and distributed throughout the power
system. The data acquired typically include watt, VARs, Volts, amps, kilowatt hours,
frequency, circuit breaker status, and tap changing and phase shifting transformer
settings. These data are transmitted to the system control center and stored in the
SCADA/EMS real-time database. The system operator the monitors and controls the
system in real time with the help of a state estimator (SE) program. The SE
periodically computes an estimate of operating state of the sub network of interest,
which is almost always a part of a large network. For the purpose of contingency
analysis, the complete model computed by SE consists of the monitored sub network
and an external equivalent that approximately the effect of the surrounding global
network.
Let us consider that n power system components are there in a power system, and if
one component, i.e. one generator or a one line in a transmission system fails or
outage (single failure), then this event is called n 1 contingency analysis. Whereas if
two components, i.e. two lines in a transmission system or, a generator and a
transmission line in the system fail or, outage (two failures), then this event is called n
2 contingency analysis.
The problem of studying thousands of possible outages becomes very difficult to solve
if it is desired to present the results quickly. One of the easiest ways to provide a quick
calculation of possible overloads is to use linear sensitivity factors. These factors
show the approximate change in line flows for changes in generation on the network
configuration and are derived from the dc load flow.
Chapter-4
Contingency selection
PI= wi/2n[fi(z)]2n
Where fi(z) is a liner function of zi, where zi denotes the changes in bus voltage
magnitudes or generator bus injections with respect to their rating etc. The order of the
above performance index in 2n.
S.N.Singh and S.C.Srinivasan proposed voltage performance index (PIv) ,chosen to
quantify deficiency to out-of-limit-bus voltage, is defined as
PIv=
Where Vi and Visp are the post outage voltage mangnitude and specified
(rated)voltage magnitude, respectively, at bus-I ,n is the exponent of the function, and
N is the total number of buses in the system.Wvi is the weighting factor and
Any contingency case with voltage levels outside the limit yields a high value of PIv.
On the other hand, when all the voltages are within the limit, the voltage performance
index PIv is small. Thus, this index measures the severity of the out-of-limit bus
voltage, and for a set of contingences, this index provides a direct means of comparing
the relative severity if the different outages.
The on-line steady-state analysis of the power system requires the evaluation of the
effects of all possible contingencies on the system, for a power system of average size.
It is generally agreed that the analysis of several hundred contingencies is usually
adequate. Full ac analysis using a power flow for several hundred cases presents a
major computational burden and even with the use of fast efficient power flow
algorithms and techniques such as the matrix inversion lemma to simulate
contingencies, it requires substantial amounts of computer time. Since only a few of
the contingencies are security risks at any given time, an automatic contingency
selection method that ranks the contingencies and selects the most severe ones has
been sought.
Contingency ranking and selection methods much have two main properties to be
useful. The computational burden for the selection process and the subsequent AC
analysis of the selected contingencies must be less then, that for the AC analysis of all
the contingencies. This is measured by the ratio of the execution times for the
automatic contingency selection and that for the full AC analysis. The other desired
property is the accuracy of the ranking and hence the selection such that no
contingencies that present contingencies risk are overlooked. This is measured by the
capture ratio which compares the ranking obtained by the selection method to the
actual ranking as obtained by full AC analysis.
The main reason for having contingency selection in analyzing on-line security is to
minimize the computational requirement. However, this has to be balanced against the
accuracy of the total security.
The performance index(PI), based on normalized values of voltage deviation or line
overloads. The most accurate method available to estimate such performance indices,
and to rank them according to their severity, is the AC load flow. However, for several
hundred buses, full AC load flow requires prohibitively high CPU time. Hence, for
such real time application linear, noniterative approximate techniques have to be
employed.
P.R.Bijwe et.al. proposed approach the post-outage line flow are obtained in term of
generation correction (around base case conditions) using GSDFs and line outage
distribution factors. Using these post-outage line flow, the ranking performance index
for line overload is obtained in terms of generation correction. The performance index
is minimized subject to incremental power balance equation and inequality constraints
on generation correction. The efficiency of the proposed approach, however, stems
from the fact that linear analysis (normally used in contingency ranking) permits us to
optimize the generation corrections for every line outage contingency in the same base
case pre-outage condition itself. This allow us to use for all line outage contingency
optimizations and same set of GSDFs and loss formula coefficients. Although it must
be remembered that, the generation corrections for every contingency are found out in
the pre-contingency condition. These are actually applied in respective post-
contingency conditions. To solve the optimization problem a procedure similar to that
involved in classical ED (dispatch) is used. In this work overload ranking of only line
outages contingencies has been considered. The procedure can, however, be easily
extended to include generator outage contingencies. Corrective capability due to
generation rescheduling is considered.
contingencies selection is an important function in a modern energy management
system aimed at alleviating the computational burden involved in online security
analysis. To be specific, the purpose of a contingency selection package is to rank all
possible outages in the system according to their severities. Then, in on-line security
analysis, only the top N outage cases in the ranking list should be studied for more
details using a full AC power flow (FACPF) method.
It has been pointed out that two separate ranking lists are required for real power
problems and voltage problems, respectively, since the contingencies causing line
overloads do not necessarily cause bus voltage violations and vice versa. Thus, two
performance indices, which give measures for line MW overloads and for bus voltage
violations, respectively, are needed for real power and voltage contingency selection.
The performance indices in common use are of the following form
Pip=
PIv=
Where
Pi= MW flow in line i
Pimax= MW capacity limit of line i
Vj=voltage magnitude at bus j
Vjspec=specified nominal voltage at bus j (1)
Vjmax=maximum voltage magnitude at bus j (1.05 pu)
Vjmin=minimum voltage magnitude at bus j (0.95 pu)
p= set of lines used in the computation of PIp index
v= set of buses used in the computation of PIv index
Contingency analysis using sensitivity factors the problem of studying thousands of
possible outages becomes very difficult to solve if it is desired to present the result
quickly. One of the easiest ways to provide a quick calculation of possible overloads
is to use sensitivity factors. These factors show the approximate change in line flow
for changes in generation on the network configutation and are derived from the DC
load flow. These factor can be derived in a variety of ways and basically come down
to two types.
Generation shift factors
Line outage distribution factor
The generation shift factors are designed ali and have following definition
ali=
where l=line index, i=bus index,
power flow on each line in the network could be calculated using a pre calculated set
of a factors as follows:
where , fi= flow on line l after the generator on bus I fails, f = flow before the failure.
The outage flow fi on each line can be compared to its limit and those exceeding their
limit are flagged for alarming would tell the operations personal that the loss of the
generator on bus I would result in an overload on line l. the generation shift sensitivity
factor are linear estimates of the changes in flow with a change in power at a bus.
Therefore, the effects of simultaneous changes on several generatin buses can be
calculated using superposition. The line outage distribution factors are used in a
similar manner, only they apply to the testing for overloads when transmission circuits
are lost. By definition, the line outage distribution factor has the following meaning:
Where ,
Dl,k= line outage distribution factor when monitoring line l after an outage on line k,
Fi= change in MW flow on line l,
Fk= original flow on line k before it was outaged i.e., opened , if one knows the power
on line l and line k , the flow on line l with line k out can be determined using d
factors.
Where fl and fk= pre outage flow on lines l and k , respectively , ft= flow on lone l
with line k out
By pre calculating the line outage distribution factors, a very fast procedure can be set
up to test all lines in the network for overload for the outage of particular line.
Furthermore, this procedure can be repeated for the outage of each line in turn, with
overload reported to the operations personnel in the form of alarm massages. The
generator and line outage be noted that a line flow can be used to program a digital
computer to execute a contingency analysis study of the power system.
Line Outage Distribution Factors (LODFs)
When there is a single-fault or a multi-fault that occurs in the system, the power is
being shifted to the adjacent transmission lines. This normally results in unfavorable
operation conditions where the transmission lines are being overloaded ie.
Transferring power above its capacity limit thereby resulting cascading faults. In order
to avoid such a situation, the overloaded lines need to be relieved from the extra load.
A security analysis study must be executed very quickly in order for it to be of any use
to the operators. This is where the computation of the distribution factors such as the
power transfer distribution factors and line outage distribution factors are required.
These factors, which are based on the DC power flow method, provide approximate
but quick solutions for the change in the power injections in the system.
In order to observe the effect of the line outage, the line outage distribution factors are
being assessed. This factor is being computed using the injection shift factors (ISFs),
which are the building blocks of the other distribution factors [3]. To shift the extra
power being transferred on the factors are being computed. The advantage of these
distribution factors is that since it is linear, quick computation is enabled and is
therefore widely used in security studies [2]. Also, since it is linear, superposition is
possible where different transactions can be superimposed onto each other. These
factors however vary with the change in the topology, when an outage occurs and
when there are change in the loadings on the lines in the system [4]. This would
require a formulation of the factors so that we would be able to compute these factors
at any loading conditions. The formulation of the PTDFs and LODFs is being
described in the next sub-section.
PTDF and LODF matrix formulation
A DC power model for the power system is being formulated where the conditions are
being assumed [5]:
1) All the voltages are considered to have a 1 p.u. agnitude.
2) Line resistance is neglected since it is small compared to
line reactance. The shunt reactance is being omitted.
3) Reactive power flow is disregarded.
For calculating the power flow on a transmission line ij l ,
(1) is being employed.
ij
ij
ij x
P
( )
= (1)
This would require the angle value of the from and to
buses which is obtained from (2).
= B1P (2)
However, this would require the calculation of the
inverse of B for every variation in loading. To facilitate this
process, (3) which is being derived in [6] is being utilized.
1()
ik jk
ij
ij X X
x
= (3)
In order to use this formulation, the B matrix must be
composed. Since the line resistance is being neglected the Y
matrix is being transformed into the B matrix where the offdiagonal
terms Bij and diagonal terms Bii are being
formulated by (4) and (5) respectively.
B = 1 (4)
=
=
N
j ij
ii x
B
1
1 (5)
To the inverse of the B matrix, at the nth bus, which is the
slack bus, a zero entry is being inserted for the nth row and
column (X matrix). The PTDF matrix is being formulated
by (6).
PTDF B A X br = (6)
where Bbr is a nline by nline dimensional diagonal matrix
where the line reactance values are the diagonal entries and
A, is the nline by nbus dimensional the branch incidence
matrix where 1 and -1 stand for the for bus and to bus
respectively. (6) is the matrix formulation of (3) which is
being utilized in the application developed.
In order to compute the line flow on each transmission
line, the PTDF matrix is being multiplied by the power
injection matrix as seen in (7).
l inj f = PTDF P (7)
where Pinj is the vector of the power injected minus the
power consumed at the corresponding bus.
The convenience of this formulation is that we just need
to change the power injection and withdrawal values in the
power injection matrix which would enable the user to
compute the PTDFs for any variation in the loading of the
system. For a column j in the PTDF matrix, which
corresponds to bus j, displays the distribution of the power
on each line for a 1p.u, increase in a power injection on bus
j and a corresponding withdrawal from the slack bus.
i j i slack j slack PDTF PTDF PTDF = (8)
As seen in (8), to compute the PTDFs for an injection in
bus i and a corresponding withdrawal in bus j (a bus other
than the slack bus), we just need to subtract the jth column
from the ith column.
The computation of the LODFs is simply done by
deleting the line from the B matrix, inverting the matrix and
calculating the PTDF matrix. The procedure for the
computation of the distribution factors and adjustment of
line flows, ie. the generation shifts, are being shown in Fig. 1.
PTDF computation and generation shift algorithm.
2.2 Adjustment of Line Flow
For the case when the transmission lines surpass the
security limits, the power flow on these lines needs to be
curtailed. In order to do this, for normal conditions when
there is an increase in the loading of the system, the
formulated PTDF matrix is being used. For the single or
multi-line case, the LODF matrix is being utilized. When
there is a violation in the security limits of the transmission
line, generation needs to be shifted in order to reduce the
line flow on these lines. This is resolved by manipulating
the PTDF and LODF matrix entries. The set of generators,
for a shift in power from generator i to j, the maximum
PTDF value is being selected which means minimum
generation shift is required to bring back the line flow
under secure values. If the flow on the overloaded line is
negative, then the a positive maximum PTDF value is being
selected and vice versa.
Chapter
Contingency Evaluation
contingency state.
Single Line Diagram of IEEE-14 bus syste
IEEE-118 bus system-
The bus data and line data of the IEEE-118 bus test system are detailed in Appendix-B. The
system as show in fig. consists of 1-slack bus, 91-load bus, 35-synchronous condenser, 117-
line (Branches), 9 -Transformers and 19-generator buses. The active power flow in each
transmission line that has been obtained using FDFL (corresponding to the base case loading
condition is also show in fig).The state of the system corresponds to the pre-contingency
state.
Contingency listing-
IEEE-14 bus system-
For this system, all the single line outages and single generator (except slack) outages have
been considered. Thus a total of 23 contingency cases (19 branch outages and 4 generator
outages) were analyzed. The outage of line 6-5 has not been considered as it is same as the
line outage of generator -5, already considered in the contingency list. The performance
indices have been computed using the net work solution from an exact load flow (FDFL)
method.
The voltage performance index (PIV) and power performance index PIp have been computed
for each contingency. For exponential value n=1 using distribution factor and the exact load
flow method respectively PIv of some of the contingencies are presented in table for
calculation of PIp, PIv given in above tables all the weighs have been considered as unity.
Contingency ranking
IEEE-14 bus system-
Contingency
name From To PIP Ranking PIV Ranking
1 1 2 71.82 1 2.752 7
2 2 3 1.294 3 2.087 10
3 2 4 0.4901 22 1.9646 11
4 1 5 0.6026 17 1.0923 19
5 2 5 0.5339 21 1.1699 18
6 3 4 0.8024 6 1.2368 17
7 4 5 0.5496 20 6.1611 2
8 4 7 0.4043 23 5.1387 4
9 7 8 35.371 2 3.3923 5
10 4 9 0.5664 19 2.6689 8
11 7 9 0.857 5 5.7732 3
12 9 10 0.6849 9 0.5243 21
13 6 11 0.6158 16 3.1795 6
14 6 12 0.7676 7 0.9152 20
15 6 13 0.8649 4 0.028 22
16 9 14 0.6712 12 10.344 1
17 10 11 0.6739 11 2.223 9
18 12 13 0.6769 10 1.2923 16
19 13 14 0.6941 8 0.0074 23
20 G2 G2 0.5701 18 1.4467 12
21 G3 G3 0.6712 13 1.3972 13
22 G6 G6 0.6712 14 1.3972 14
23 G8 G8 0.6712 15 1.3972 15