11 - Chapter 6

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 21

71

CHAPTER 6

COMPARISON OF RESULTS, ANALYSIS AND


CONCLUSION

6.1 INTRODUCTION

The aim of this thesis is to obtain optimum/minimum value for the


objectives namely cost, weight, losses and volume while delivering its rated
kVA. In order to explain the same both single and three phase transformers
are considered and solved both by conventional method and graphical
method. It is also solved by popular optimizing techniques namely
Lagrangian multiplier method and Random jumping methods. The results are
collected and tabulated.

6.2 COMPARISON OF RESULTS OF SINGLE PHASE


TRANSFORMER

The results obtained for single phase transformer are presented in


Table 6.1 for comparison.
72

Table 6.1 Design results for single phase transformer

DESIGN RESULTS
Method Optimum design
S. Conven-
No Lagrange Random
tional Graphical
multiplier jumping
method method
Particulars method method
Cost ( Cost with
1 Cost of iron optimum
6058.80 6130.07 6131.20 6129.98
Ci, Rs weight )
3707
2 Cost of
6198.00 6130.35 10110 6131.20 6129.83
copper Cc, Rs
3 Total cost,
12256.80 12260.42 13817 12262.40 12259.81
Rs
4 Weight
Weight of 55.08 33.7 33.74 33.88
iron, G i, kg
5 Weight of
20.66 33.7 33.74 33.61
copper, G c, kg
Total weight,
75.74 67.4 67.48 67.49
6 kg

6.3 COMPARISON OF RESULTS OF THREE PHASE


TRANSFORMER

The results of three phase transformer are collected and presented


in Table 6.2. Since 500kVA transformer is more commonly used one, test
certificates for similar rating transformer are obtained from two companies
and their results are also presented for overall comparison. Test certificates
are available in Appendix 6. Out of the available two test certificates, the
higher result test certificate (efficiency) (Kamath Transformers (Pvt) Ltd.,
Bangalore) is presented in the tabulation for comparison.
73

Table 6.2 Design results for three phase transformer

DESIGN RESULTS
Method Optimum design
S. Conventional
Lagrange Random Company
No method Graphical
multiplier jumping Result
method
Particulars method method
Full load
1 loss
2698 2848 2845 2853.23 1048
Full load iron
loss, P i, W
Full load
2 copper loss, 3190 2848 2845 2842.29 5422
P c, W
3 Total full
5888 5696 5690 5695.52 6470
load loss, W
Volume
4 Total volume 0.182 0.097 0.0969 0.0974 ---
of iron,U i, m3
5 Total volume
of copper, U c, 0.07735 0.097 0.0969 0.0964 ---
m3
6 Total
0.25935 0.194 0.1938 0.1938 ---
volume, m3

6.4 COMPARISON OF Bm AND

The flux density, Bm and current density, plays an important

role in the estimation of losses/efficiency. The results of them for a three


phase transformer are presented in Table 6.3 for comparison. Also,
resistance, Per unit reactance and no – load pf are indicated.
74

Table 6.3 Results of Bm and

S. Conventional Graphical
Particulars Symbol Company
No. Design method
1 Maximum flux
density in the Bm 1.2 (assumed) 1.227
core, Wb/m2
2 Current density,
1.4 (assumed) 1.37
A/mm2
3 Resistance/ph,
HV, ohms rp 2.696 2.068 4.42
LV, ohms rs 0.001 0.001068 0.00212
4 Per unit
3.82 4.61
reactance
5 No load power
0.28
factor

The values for Bm and are within the permissible ranges (Refer

Appendix 4)

6.5 ANALYSIS

6.5.1 Practical Optimum Value for Ai

From the conditions for optimum design, we can also write

For minimum cost : Gi Gc cc ci (6.1)

For minimum volume : Gi Gc gi g c (6.2)

For minimum weight : Gi Gc 1 (6.3)

p
For minimum loss : Gi Gc x2 c (6.4)
pi
75

Also the transformer output equation can be written as (say for


example, single phase transformer)

li g i Gi 2
Q 2.22 f Bm A 10 3 (6.5)
Lmt g c Gc i

and Ai is transformed as

Q Lmt g c 10 3 Gi
Ai (6.6)
2.22 f Bm li gi Gc

Depending upon the objective, the corresponding value for the ratio
( Gi Gc ) may be substituted in eqn 6.6 and the net core area may be

determined, which is optimum.

Let us calculate Ai for minimum cost i) from the solutions by

conventional method and ii) from graphical method results.

The values are

i) Conventional Ai 0.00788 m2

ii) graphical Ai 0.007 m2

There is a reduction in area in the proposed graphical method,


which indirectly indicates that the cost is also minimum/optimum. The area
obtained by conventional design is not minimum/ optimum.
76

6.5.2 Improvement in Efficiency

Consider the three phase distribution transformer.

At full load ( x 1) , the full load optimum losses are obtained as

Pi 2848W and Pc 2848W .

Taking 10% of copper loss as stray loss,

Total full load loss = 2848 + (2848 +284.8)

= 2848 + 3132.8 = 5980.8 W

The full load (x=1) efficiency of a 500kVA transformer at UPF is,

500
100 98.82 %
fl 505.9808

Similarly the efficiencies at x (fractions of full load) = 1.25, 0.75,


0.5 and 0.25 are calculated and the values are

x 1.25; 98.47 %
x 0.75; 99.08%
x 0.50; 99.28%
x 0.25; 99.35%

The corresponding results seen in the Test certificate are as follows:

x 1.25; 98.48%
x 1.00; 98.53%
x 0.75; 98.79%
x 0.50; 98.93%
x 0.25; 98.82%
77

Also the full load efficiency at 0.8 pf is calculated and shown in


Table 6.4. The corresponding values of the company is also indicated in the
table for comparison.

Table 6.4 Full load efficiency at 0.8 pf

Fraction of Efficiency %
S.No full load Input, kW Graphical
Company
x method
1. 1.25 407.743 98.10 98.12
2. 1.00 405.981 98.53 98.16
3. 0.75 404.610 98.86 98.42
4. 0.50 403.630 99.10 98.64
5. 0.25 403.040 99.24 98.53

There is an improvement in efficiency at all loads. The


improvement in efficiency in the form of graph is as shown in Figure 6.1. It is
worth referring here that Provincial Electricity Authority (PEA) of Thailand
has the policy for the total loss of 500kVA transformer from 6500W (old
regulation) to 5810W (new regulation) because they would like the customers
to select one of 500kVA transformer rather than two of 250kVA transformer.
(B Suechoey et al 2002).

Normally the distribution transformers have varying load. It is


usual to design such that the maximum efficiency occurs around 50% of the
load. It is seen from a practical transformer test certificate that it is having an
efficiency of load 98.93% in comparison with 99.28% from graphical
optimization. The efficiency calculated from graphical method is higher than
the company’s efficiency at all loads.
78

To show the adoptability of graphical method, the efficiency at


half-full load for UPF is also calculated as under. The optimum losses are
taken from Figure 4.6.

Iron losses, Pi = 2850 W


Copper losses = 700 W
Stray copper losses = 70 W
Total copper loss , Pc = 770 W
Total losses, PT = 2850 + 770 = 3620 W
Efficiency at UPF, = 500/503.620 = 99.28%

which tallies with the calculated value of 99.28 % for x = 0.5

Figure 6.1 Improvement in efficiency

6.6 NEW ALGORITHM FOR OVERALL OPTIMUM DESIGN

This section is not connected with graphical method of optimum


design. However it is presented for academic interest since end results are
improved with less assumption. Equations are arranged only in two variables.
79

Normally optimization will be done for one objective, considering


its constraints. Here a methodology is presented for overall optimization of
design of a transformer. It is evident that since there is no objective, there is
no necessity for objective function. Further, by overall optimization, it means
that the values for the variables which appear in design equations are
dependent upon the optimum value of the variable obtained in the preceding
step. The constraint equations are formed with the help of two independent
variables. By this method, assumption of values for the variables is mainly
avoided. Random jumping method is used at the appropriate places as a
comparator to obtain optimum values. In order to facilitate, the equations are
modified in such a way that there appear only two variables. Already this
method is used and optimum values are arrived which agreed with other
methods. Now a three phase transformer given in section 4.3 of chapter 4 is
considered and the methodology is explained.

6.6.1 Design Methodology

The primary induced emf is approximately equal to the impressed


voltage and hence the per phase induced voltage with Tp turns is written as

E p Vp 4.44 f m T p (6.7)

where Vp is the impressed primary voltage per phase, volts.

E p is the primary induced emf per phase, volts

m is the mutual flux in core, Wb


f is the supply frequency in Hz and
Tp is the turns per phase for primary winding.

Here in the above equation, the values for E p and f are known

and fixed. Hence the above equation may be considered as a constraint


equation of known value with two independent variables m and Tp . This
80

constrained equation may be treated as an unconstrained equation by


specifying lower and upper limits for the variables namely m and Tp . The

limits can be specified out of experience. Hence the problem may now be
specified as an unconstrained minimization problem as

m
Find X which equals f X (6.8)
Tp

with limits for the variables as

l1 m u1 (6.9)

and l1 Tp u1 (6.10)

In the above equation, the value of f X is actually known.


However, random jumping method is used to obtain optimum values for m

and Tp which after evaluation is compared with the known values of f X .

Thus the optimum values for m and Tp are obtained.

Now knowing the optimum value of m and treating the below


mentioned equation as a next constrained equation with variables Bm and A
i

m Bm Ai (6.11)

where Bm is the maximum flux density in the core, Wb/m2 and A is the net
i
core area, m2, the values for Bm and A are optimized as done earlier. The
i
process of optimization is successive in nature.

Now using the normal value for iron stacking factor,


Ki (i.e. Ki 0.9 ), the gross core area, Agi can be obtained. The sizes of

cruciform core are thus obtained and the core design is completed. The
81

cruciform core is as shown in Figure 6.2, having two steps with different sizes
of lamination.

Figure 6.2 Cruciform core

The various symbols for the core are,

d , diameter of circumscribing circle, m


a , largest width of stamping, m
b , smaller width of stamping, m

Now the window dimensions are estimated as described. The


output equation of a three phase transformer is

Q 3.33 f m Ac Ai 10 3 (6.12)

and window area, Aw H w Ww (6.13)

The window space factor K w can be calculated from the empirical


formula. For 500kVA transformer,

10
Kw , (6.14)
30 kv

where kv is the voltage rating of high voltage winding in kV.


82

It is taken as 0.25.

Ac
Also, K w (6.15)
Aw

From Equation (6.12), optimum values for Ac and can be


obtained by adopting earlier procedure. Further from Equation (6.15), Aw
can be found. Using this value for Aw and using equation (6.13), the
optimum values for H w and Ww can be obtained. Thus optimum window
dimensions are obtained as usual. Using the above data, yoke design is also
completed.

The overall frame of a three phase core type transformer is as


shown in Figure 6.3.

Figure 6.3 Overall frame of a three phase transformer

6.6.2 Complete Flow Chart

The flow chart is an indication of methodology which contains


specifications and order of sequence of evaluation. A complete flow chart is
given in Figure 6.4 for the overall optimum design of a transformer. The
standard factors, constants, etc. are given along with specifications.
83

START

READ Q=500kVA,
Ep = 11000, Es =250,
f = 50, m=3, Ki =0.9,
Kw=0.25
-6
s=0.6 x 10 m
-6
c=0.021 x 10 m
t =0.35 x 10-3 m
Kh =191 J/m3
ci= Rs.110/kg
cc= Rs. 300/kg

m= 0.04 to 0.05 Wb
Tp = 1000 to 1200
Bm =1.1 to 1.35 Wb/m2
Ai =0.03 to 0.04 m2
= 1.1x106 to 2.3 x106 A/m2
Ac = 0.03 to 0.07 m2
Hw = 0.4 to 1.2 m
Ww= 0.25 to 0.3m

X= Ep/4.44 f

X= m Tp

m=Bm Ai

Figure 6.4 (Continued) Complete Flow chart


84

Ts = Tp Es/ Ep

Agi = Ai / Ki

d = Ai / 0.56

a = 0.85 d

b = 0.53 d

Qx103/ 3.33 f m = Ac

Aw = Ac / Kw

Aw = Hw Ww

D= Ww + d

Figure 6.4 (Continued) Complete Flow chart


85

Wcw= Ac/2Hw

Lmt=(d+Wcw

Dy = a

Hy = Ai / Dy

li= 4D+3(Hw+Hy)

Ui= Ai li

Gi= 7800 x Ui

Uc=1.5 Ac Lmt

Gc=8900Uc

Ci =ci Gi

Cc =cc Gc

H=Hw +2Hy

Figure 6.4 (Continued) Complete Flow chart


86

W=2D+a

Pi =Bm2 [Kh s f Ai li +( 2 2
t /6 s) f2 Ai li]

2
Pc= 1.5 c Ac Lmt

PT= Pi + Pc

cos =1

fl = Q cos / (Q cos + PT)

PRINT Q, Ep, Es, f, m, m, Tp, Bm, Ts, Ai, , Ac, Hw,


Ww, d, a, b, Aw, Dy, Hy, D, H, W, li, Lmt, Pi, Pc, PT ,
fl, Ui, Uc, Gi, Gc, Ci, Cc

STOP

Figure 6.4 Complete Flow chart


87

6.6.3 Program

A program is written for the algorithm developed and the results are
as shown.

Results

Q=500000.000000 Ep=11000.000000

Es=250.000000 f=50.000000

m=3.000000 piem=0.046212

Tp=1073.061279 Bm=1.191830

Ai=0.039514 delta=1541196.875000

Ac=0.042164 Hw=0.776604

Ww=0.217570 d=0.265632

a=0.225787 b=0.140785

Aw=0.168656 Dy=0.225787

Hy=0.175005 D=0.483202

H=1.126614 Li=4.787634

Lmf=0.919325 Pi=2791.657959

Pc=2900.268066 Pt=5691.925781

ETA=0.988744 ui=0.189178

uc=0.058144 gi=1475.586670

gc=517.478882 Ci=162314.531250

Cc=155243.671875 Ts=24.387756

w=1.192191
88

6.6.4 Comparison of Variables and Performance Quantity

Some of the values for the variables and performance data obtained
both by conventional design and overall optimum design are compared as
shown in Table 6.5. for three phase transformer. The available values of
company test certificate and the values obtained by graphical method are also
indicated for overall comparison.

Table 6.5 Comparison of variables and performance quantity

Overall
S. Conventional Graphical
Particulars Symbols Optimum Company
No design method
design
Maximum flux
1. density in the Bm 1.2 (assumed) 1.19183 1.227 -
core, Wb/m2
Current
2. 1.4 (assumed) 1.541197 1.37 -
density, A/mm2
Full load iron
3. Pi 2698 2792 2848 1048
loss ,W
Full load
4. Pc 3190 2900 2848 5422
copper loss ,W
Total full load
5. PT 5888 5692 5696 6470
loss ,W
Full load
6. efficiency at fl 98.77 98.87 98.87 98.53
UPF, %
% Regulation
at full load and
7. - -
at i) UPF 0.5696 1.42
ii) 0.8 pf 2.75 3.81
89

6.7 CONCLUSION

Rabih (2005) formulated transformer problem in Geometric


Programming (GP) format in his paper. He took two design examples, one is
high frequency transformer (100 kHz) and another is low frequency power
transformer (60Hz) of single phase type. He presented the results of both. He
obtained better values for the constraints, namely for minimum efficiency as
99.56% against the constraint of 97% for high frequency transformer,
similarly for the low frequency transformer as 98.02% against 97%.

In this thesis a Transformer Design Problem one in single phase


and another one in three phase is taken. It is solved by conventional method.
It is seen that none of the objective, namely, cost, volume, weight or loss is
optimized. Whereas the same problem is modified with only two variables
depending upon the objective and solved by the proposed graphical method. It
is observed that the optimum values for the variables are achieved easily,
quickly simultaneously besides satisfying the conditions. Further from Table
6.1 for single phase transformer, it may be noticed that optimization is
achieved for only one objective at a time namely either cost or weight. Also it
may be seen that the use of optimum weight for the calculation of cost does
not yield optimum. The optimum values for the objectives are compared with
the results obtained by other optimizing methods, namely Lagrange multiplier
method and Random jumping method and found agreeing.

In order to emphasis the suitability of the proposed graphical


method, performance quantities are calculated using these optimum values
obtained from graphical method. It is seen that the efficiency, temperature rise
and regulation are improved than that obtained in conventional design.
Further a graph is drawn for efficiency for the results obtained by graphical
method and the result of a transformer company. The efficiency is obtained as
99.28% in comparison with 98.93% for three phase transformer at 50% load.
90

Further the efficiency obtained from the graphical method is higher at all
loads. It is worth specifying that the allowed total loss of 5810 W of
Provincial Electricity Authority (PEA) of Thailand for the same 500 kVA,
three phase transformer, the optimum losses obtained by Graphical method is
only 5696 W, which confirms an improvement in design. Similarly, the kVA
output is verified with the optimum values obtained by graphical method and
found satisfactory. The full load regulation is 0.5696% and 2.75% against
1.42% and 3.81% of company test certificate for UPF and 0.8 p. f.
respectively which shows improvement and falls below 6% stipulated by the
Indian Electricity Rules, 1956.

The optimum values for electric and magnetic parameters are


estimated from the optimum values obtained by graphical method and the
optimality is also verified. Further the values fall within the range. This is an
additional feature of this graphical method.

It is stated that excepting cost optimization graph, other


optimization graphs are direct, i.e. if volume, loss or weight is to be estimated,
the co-ordinates of the graph are directly in terms of their respective
quantities, whereas in cost optimization graph, the co-ordinates are in terms of
area. It is to be noted specially and may be used conveniently.

Above all, a new algorithm for overall optimum design using two
variables at each step of a distribution transformer is presented. The same
three phase transformer design problem as given in section 4.3 is solved. The
efficiency arrived by both graphical method and overall optimum design
approach seems to be same, which strongly indicates that optimum values
estimated by graphical method is correct and accurate. The results obtained
are better than both by the conventional design and results of a transformer
company.
91

Even though, the proposed graphical method is an analytical one,


the results agree with the well known optimizing methods, namely, Lagrange
multiplier method and Random jumping method.

Hence it is concluded that the graphical method using only two


variables may be adopted for optimizing an objective of a distribution
transformer.

You might also like