LOSS MINIMISATION IN INDUSTRIAL POWER SYSTEM OPERATION D. Lukman K. Walshe T.R.

Blackburn School of Electrical and Telecommunication Engineering University of New South Wales, Kensington Sydney, NSW 2052, Australia
E-mail : d_lukman@hotmail.com

Abstract
This paper presents a new approach to the use of load flow by proposing the incorporation of the Blosses coefficient method to bias the operation of control devices such as OLTC, FACTS, etc inside the traditional Newton-Raphson algorithm. In this way the choice of control settings can be determined whilst minimising losses. It is believed that this technique will be of particular advantage to industrial systems where the far more complex loss minimisation methods used in EHV transmission systems are not implemented at present. The causes, types and effects of losses are first described. Then follows an extract from a real life load flow study to illustrate present usage. A modified technique of loss minimisation is then proposed using modified B-losses coefficients and changes to transformer modelling in power flow software with a view to arriving at a load flow suited to an industrial application.

1. INTRODUCTION
The quantification and minimization of losses is important because it can lead to a more economic operation of a power system. If we know how the losses occur, we can take steps to limit the losses. Hence, if more losses can be minimised, the power can be consumed efficiently. Existing power generation and transmission can be used effectively without having to build new installations and at the same time save the cost of losses. Losses result from current flow in resistive materials and magnetising energy in transformers and motors. Other sources such as dielectric and rotational losses are usually so small as to fall outside the scope of this paper. Considering losses associated with resistive material , three things need to be considered in order to prevent the unnecessary losses; either reducing the resistance/impedance, or decreasing the current, or maximising voltages. Magnetising losses are however minimised by minimising line voltages (in per unit terms). Hence, the ability to reduce the losses is an important consideration. Several papers had been written in the

1950s and 1960s such as [1], which gives the comparison of actual measured transmission loss in an extensive power system with that calculated from a set of constant loss coefficients for wide shifts in load between two plants. Reference [2] discusses an improved digital-computer method of calculating loss-formula coefficients. This paper also discusses B-loss coefficients and how these can be implemented in the digital load flow calculation in power system operation.

2. LOSSES IN POWER SYSTEMS

To determine B-loss coefficients, it is necessary to determine the precise loss mechanisms, which occur in the system. Traditionally, B-loss coefficients have been applied to transmission line analysis where the losses are predominant by only line loss determined by I2R. Transformer losses are not significant in such systems. The only other losses are corona, which will occur only under foul weather conditions. However, in industrial power systems, the losses are more diverse and thus B-coefficients will be more complicated to utilize. 2.1. Transmission Losses Consider a simple three-phase radial transmission line between two points of generating/source and receiving/load as illustrated by Figure I.

Figure II Radial System with One Additional Generation to Load Bus Losses can now be expressed by the equation:
2 2 PL = P 1 B11 + 2P 1P 2 B12 + P 2 B22

Figure I Radial Line with One Generation and One Load We can deduce that the line loss is:

We refer to B as the loss coefficients. These will be discussed in detail in section 6. Transmission losses become a major factor to be considered when it is needed to transmit electric energy over long distances or in the case of relatively low load density over a vast area. The active power losses may amount to 20 to 30 % of total generation in some situations [5]. 2.2. Industrial System Losses Losses in complex industrial systems can arise from the following mechanisms: Line and cable losses Transformer losses (core and copper) Machine losses (core and copper) Eddy current losses in metal housings, etc Corona losses on contaminated insulators These are thus a complex combination of fixed (core and corona) and variable (I2 dependent) losses. Thus PL = B0 + B1PG2 where B0 represents fixed loss B1 represent variable loss PG is the generated power Thus the calculation of B-loss coefficients is more complex in large industrial systems, but the potential improvements or loss reduction is likely to be greater if the B coefficients can be incorporated in load flow software.

P loss = 3 I R
where R is the resistance of the line in ohms per phase. The current I can be obtained:

I =

where PG is the generated power (load power and losses) VG is the magnitude of the generated voltage (line-to-line) cosG is the generator power factor Combining the above two equations, we have:

( 3 )V

PG
G

cos G

PL =

R VG
2

cos 2 G

( PG2 )

Assuming fixed generator voltage and power factor, we can write the losses as:

P L = BP G2
where in this case

B=

R VG cos2 G
2

Losses are thus approximated as a second order function of generation. If a second power generation is present to supply the load as shown in Figure II, we can express the transmission losses as a function of the two plant loadings

3. EFFECTS OF LOSSES
An early method used by engineers to minimize the cost of delivered power was by supplying power from only the most efficient plant at light loads. As load increased, power would be supplied by the most efficient plant until full capacity of this plant was reached. For further increase in load, the next most efficient plant would start to supply the additional power and so on. This method fails to minimize the cost because the losses in transmission from the plant with the lower incremental cost may be so great that economical consideration may require lowering the

load at the plant with lower incremental cost and increasing it at the plant with higher incremental cost. A very efficient generating unit with low incremental operating cost may be located far from load centers. Transmission losses associated with this unit may be so high that the economic dispatch solution, which is how the real power output of each controlled generating unit is selected to meet specified demand, requires the unit to decrease its output, while other transmission units with higher incremental operating costs but lower transmission losses increase their outputs. We need to express total transmission loss as a function of plant loadings. Thus, losses increase the operating cost of running a power system and determine how to operate various generating plants. In addition to that, thermal losses reduce the overall lifetime of the electrical equipments.

Figure III. Electrical One Line Diagram including Train H Electrical System The electrical system has the flexibility to be controlled in either normal ordinary operation with the tie breaker at MV/LV busses turned off or as a load-shifted operation with one of the paired generators from the 13.8 kV bus switched off or shut down putting the demand to the other generator to supply the whole loads connected to it and letting the tie circuit breaker switched on allowing operation as a double-ended substation. The switching sequence is depicted in Figure IV where the closed breakers are shaded and the open breakers are left unfilled.

4. AN INDUSTRIAL POWER SYSTEM

Consider the case of the electrical system in Train H LNG Refinery Plant in Bontang, Borneo, Indonesia which was commissioned in December 1999 after three years of Engineering Procurement Construction (EPC) where one of the authors was involved in the power system study of the overall electrical system. The system can be used as a model for application of B-coefficient losses. The Train H electrical system is added as part of Module II of a two module system (Module I and Module II). The two modules are interconnected by current limiting reactors (CLR) for fault current limiting at a ring bus voltage of 34.5 kV. Each module consists of four LNG trains connected to 13.8 kV substation busses. Train H LNG refinery plant electrical system is paired with Train G and with Train E and F pair form module II. The system has the capability to be operated as either a modular system when the CLR is switched off or operated as an integrated system when the two modules are connected through the reactor as shown in Figure III. Generator 7 in Figure III has flexibility to be switched between buses 30PS-4 and 30PS-5 to overcome deficient of Module II for case study LF-int-5 and LF-int-6, which is given in Table I.

Figure IV. Normal and Abnormal/Load Shifting Operation

A load flow study was performed using the GE power system software PSLF. An interim study was performed to quickly confirm the ratings of major equipment and load flow modelling. The results of a load flow study were useful in determining optimum taps for transformers and to enable prediction of equipment overloading and excessive voltage drops, which may be encountered in the future and which increase equipment losses. Several case studies were created based on both normal and abnormal/load shifting operating conditions under both modular and integrated operation. The listing below provides all the cases having been considered and gives the summary of the significant outcomes/results out of the load flow study.
Case No. LF-int-1 LF-int-2 LF-int-3 LF-int-4 LF-int-5 Description System normal integrated operation System normal modular operation Integrated operation with load shifting from Bus 7 to Bus 8 Integrated operation with load shifting from Bus 8 to Bus 7 Modular operation with load shifting from Bus 7 to 8 Comments Acceptable operation Acceptable operation Acceptable operation. Requires some tap adjustments to achieve acceptable voltages. Acceptable operation. Requires some tap adjustments to achieve acceptable voltages. Acceptable operation. Requires some tap adjustments to achieve acceptable voltages. Generation deficiency in Module II

threshold. In the instance of the bus voltage, which falls below 97.5 % threshold even with the presence of transformer taps, boost taps can be used to make the desired improvements. However, load shifting can result in voltage reduction because load shifting causes the transformer to be more heavily loaded. The problem is more significant if the system is operated in modular operation, which does not have as much power generation. If voltage drops occur in feeders, transformer taps must be used to boost the load voltage if the natural voltage sag results in voltages outside acceptable range. As it can be given by the load flow case example, more concentration was given in terms of voltage acceptance criteria rather than achieving loss minimization. This is a shortcoming in traditional load flow. Given the NR load flow, the ability to find the least cost way of running an industrial power system would result in major improvements to both the design and operation of such systems. We are going to look at a different way of using load flow in industrial power systems. We know that, at minimum load, the OLTC of a transformer would be set at the minimum tap setting and at maximum load, the tap setting is adjusted at the maximum setting. In the loss minimisation technique, correct tap settings are to be controlled by plant loadings and network loss minimization with voltage criteria setting boundaries for operation rather than absolutes as in traditional load flow.

Table I Load Flow Case List and Summary Several analyses were made and are described as follows: The five cases evaluated in the load flow study indicate that the overall system design planned for the Train H expansion is capable of supplying the required loads at acceptable voltage level Voltage acceptance criteria (0.975 > V > 1.05 per unit) can be easily achieved during normal operation, but some adjustment in transformer taps are required to obtain acceptable voltages under load shift conditions However, load-shifting operation would not be recommended for modular operation because it may result in generation loss in Module 2. PT Badaks, Bontang LNG Refinery Operator, system has grown to over 500 busses. Therefore, the ability to extract useful information out of the excessive information is necessary. The total system load is calculated to be 124 MW for 8 trains, or about 15.5 MW per train. The results of load flow are also able to determine if the voltage at particular bus has fallen below 97.5 %

5. LOSS TECHNIQUES

MINIMIZATION

Contemporary loss minimization techniques during the system design phase described in the literature include the use of low resistivity materials, reducing impurities in conductor material, reducing eddy currents by using lamination and transposition, the use of conductor with large cross sectional area, bundling the conductors in case of transmission line in order to reduce the resistance/reactance. Some other techniques relate to the reduction of current since the losses are proportional to I2R. The use of FACTS devices such as the series compensated capacitor, upgrading of transmission line, adjusting the control of transformers and capacitor banks to increase the loadability or maximum power transfer

of a line and improve power factors have been implemented.[9] Several formulae were proposed to calculate the power losses by looking at the patterns of generation and loads, kWh/km, price/kWh and RPI-X formula.[10] However, the method which is of our interest and under current research here is the use of B-losses coefficient which was first proposed by E.E.George [5] in 1943 and later perfected by G.Kron by applying the tensor analysis method to power system. The idea is to include the B-coefficient into the load flow analysis, which will be described in the section 6.

Vm, Vn are voltages at bus m and n Nkm, Nkn are current distribution factors pfm, pfn are power factors A number of assumptions are made for the calculation of loss formula: 1. Assume a constant power factor at each source 2. Assume constant generator angular positions, m 3. Assume constant generator bus voltage magnitudes 4. Assume all load currents maintain constant ratio to total current Although it is valid only for certain loading restrictions, in practice, however, this formula produces close answers with errors only up to a few percent. Sophisticated methods of calculating Bcoefficient do exist and are currently being used by electricity utilities to calculate losses to determine economical dispatch operation. [5 p. 722][7 p. 242] 6.1. Example of Bmn Calculation Consider a three-line two plant system as shown in Figure II before. The following data [4] are given in per unit system: |V1| = 1.05 |V2| = 1.03 |V3| = 1.00 (pf)1 = 0.95 (pf)2 = 0.95 (pf)3 = 0.85 R1D = 0.04 R2D = 0.05 R3D = 0.03 Using the general expression for Bmn above:

6. APPLICATION OF B-LOSS COEFFICIENT INTO POWER SYSTEM ANALYSIS/LOAD FLOW TECHNIQUES

In this method, the power flow is used to account for power transmission losses in the power system. It is common practice to express the system losses in terms of active power generations only. This is commonly referred to as the loss formula or Bcoefficient method. The simplest form of loss equation is called Georges formula, which is given by:

PL = Pm Bmn Pn
m =1 n =1

B11 =

PL is the power losses Pm, Pn is the power generation from all sources The coefficients Bmn are commonly referred as the loss coefficients with the units of reciprocal Watt/MWatt. The B coefficients are not truly constant but vary with unit loadings. A more general formula (Krons loss formula) is

where

0.04 0.03 + 2 = 0.0817 2 2 (1.05) (0.95) (1) (0.85) 2 0.05 0.03 B22 = + 2 = 0.0937 2 2 (1.03) (0.95) (1) (0.85) 2 0.03 B12 = 2 = 0.0415 (1) (0.85) 2
2 2 PL = 0.0817 P 1 + 0.083P 1P 2 + 0.0937 P 2

Thus in per unit, we have the loss formula given by:

PL = K L 0 + Bm 0 Pm + Pm Bmn Pn
m =1 m =1 n =1

A linear term Bm0Pm and a constant KL0 have been added to the original quadratic equation. This shows that losses depend on the active power generations P only. Bmn is known as loss coefficient and is given by a general expression:

6.2. Economic Loading of Power Plants This method using B-coefficient enables us to coordinate transmission loss in scheduling output of each plant for maximum economy for a given load. The total fuel cost is a function of power inputs. Using Lagrangian method, minimum fuel cost FT is obtained when /Pn = 0 for all values of n, where:.

B mn =
where

cos ( m n ) N km N kn Rk V m V n ( pf m )( pf n ) k

m, n are phase angles of currents Im,In

= multiplier Pn = individual plant input to the network

k = FT Pn PL PR n=1

PL = power loss PR = total power received by the loads Minimum fuel cost is obtained when the incremental fuel cost of each plant, multiplied by its penalty factor, is the same for all plants in the system.

conditions, identify transformer tap settings and minimise kW losses. The use of B-losses coefficient in fast digital computer power flow would be required to determine the amount of power loss and minimise the transmission loss, thus increasing net savings. The research project will look at how the loss coefficient approach can be used to manipulate transformer OLTC, shunt capacitors, synchronous condensers, excitation controllers and the likes of FACTS device setting inside a Newton Raphson load flow with the object of loss minimisation. This also requires revised modelling of transformers, motors, capacitor banks FACTS and lumped loads [10] in order to incorporate the shunt and voltage dependency of their losses.

dF n Ln = dP n
where Ln is the penalty factor of plant n and is given by:
Ln = 1 P L 1

Pn For k plants, partial differentiation with respect to Pn yields:

PL = Pn Pn

P B
m m =1 n =1

mn

Pn = 2 Pm Bmn
m =1

8. REFERENCE
1. Kirchmayer, L.K., Stagg, G.W., Analysis of Total and Incremental Losses in Transmission Systems, Trans. AIEE, vol.70, pt. II, pp. 1197-1205, 1951 2. Kirchmayer, L.K, Happ, H.H., Stagg, G.W., Hohenstein, J.F., Direct Calculation of Transmission Loss Formula-I, Trans. AIEE, vol. 79, pt. III, pp.962-969, 1960 3. Eaton, J.R., 1972, Electric Power Transmission Systems, Prentice-Hall Inc., Englewood Cliffs, New Jersey 4. El-Hawary, M.E., 1983, Electric Power Systems: Design and Analysis, Reston Publishing Company, Inc., A Prentice-Hall Company, Reston, Virginia, USA 5. Glover, J.D., Sarma, M., 1994, Power System Analysis and Design, second edition, PWS Publishing Company, Boston, USA 6. Stevenson, W.D., 1982, Elements of Power System Analysis, fourth edition, McGraw-Hill Book Company, New York, USA 7. GE International Inc., 1998, Train H Expansion Study-Interim Study Report for HCJV, LNG Refinery Project, Bontang, Indonesia 8. B.C.G Shin, Development of the Loss Minimization Function for Real Time Power System Operations: A New Tool, IEEE Transaction on Power Systems, Vol. 9, No. 4, pp. 2028-2034, November 1994 9. Cory, B.J., 1998, Electric Power Systems, John Wiley & Sons, Sussex, England 10. Gosbell, V.J., System Load Characteristics, 1981 Residential School in Electrical Power Systems, 1981, Lect. Nt., Vol. 1, pp. B5.1-B5.17

Economical loading of each plant is found by assuming a value for . By solving equations for several values of , data are found by plotting generation at each plant against total generation. Total received load can also be plotted against the plant outputs to determine the net savings. The current research is aiming to implement the Blosses coefficient into the simulation, to extend the traditional implementation to include shunt losses, and to use the information as a basis for controlling tap changers, FACTS settings, switched capacitor banks and the like.

7. CONCLUSION
Most research focuses on the large transmission system. Tools developed for such systems are normally inaccessible to the operator of medium sized industrial networks. This research project addresses this shortcoming. Losses increase the cost of operation of a power system. Hence, loss minimization would be a very advantageous goal to use to optimize the generation and delivery of electrical power. This is particularly the case for industrial systems, where substantial loss reduction can be achieved by using loss minimisation rather than voltage drop criteria in system operation. A load flow study can be used to calculate the kW and the kVar flows in power system, determine operating voltage levels for selected operating