PERINI PRAVEENA SRI
Department of Social Science, Faculty of Economics
Ethiopia, Aksum University
ABSTRACT
The purpose of this paper is to examine water use estimation in hydel and thermal electric power plants in selected regions i.e. Coastal, Rayalaseema and Telangana regions of Andhra Pradesh. The study primarily focuses on the realistic fundamental premise that thermal electric and hydro electric energy generation is responsible for the largest monthly volume of water withdrawals in four seasons (i.e. summer, rainy, winter and post monsoon season) of a year. These enormous water withdrawals by these hydel and thermal power plants can have significant influence on local surface water resources. However there are very few studies of determinants of water use in hydel and thermal electric generation. Analysis of hydel and thermal electric water use data in the existing power plants clearly indicates that there is wide variability in unitary hydel and thermal electric water use within the system. The multivariate regression procedures were used to identify the significant determinants of thermal and hydel water withdrawals in various power plants i.e. five hydel and four thermal power plants. The estimated regression coefficients indicate that the best explanatory variables for the total quantity of hydel water withdrawals are storage capacity, tail water level and actual generation and thermal water withdrawals are condenser cooling and ash disposal. The unit variability of unit water usage indicates that there is significant potential for water conservation in existing power plants.
Keywords:
Thermal water withdrawals, hydel water withdrawals, storage capacity, tail water level, actual generation, condenser cooling and ash disposal.
 INTRODUCTION
Water has become a growing source of tension especially in power sector in many parts of the World. For India hydro and thermal power projects are vital to fill in the serious electric energy shortfalls that crimp its economy. About 40 percent of India’s population is off the power grid and due to this the welfare of the economy was badly affected. The main stumbling block for this kind of worse situation are a genuine water shortage problem in India and the country’s inability to properly manage large quantities of water during rainy season has made matters worse, exposing it to any small variation in rainfall or river flow. Though the country has invested heavily on nuclear power to generate 30,000 MW and $ 19 billion to produce factories of major thermal, hydro and nuclear power stations, the electric energy shortages were very much prevalent in most parts of the country. For this the first and foremost thing is to judiciously manage the vital resource “water”. The country also planned for setting up of 20,000 MW solar power by 2020. The Government of India has an ambitious mission of Power for All By 2012. This would require an installed generation capacity of atleast 20,000 MW by 2012 from the present level of 144,564.97 MU. However the power requirement will double by 2020 to 400,000 MW. How India is able to meet this target with the ongoing water shortage plight in Electricity Generation Industry is a matter of great concern. However the Electricity Generation Industry strategy should primarily focus on this invisible culprit “Water” causing huge generation losses through better water efficiency techniques and lay emphasis on technology up gradation and massive utilization of renewable sources of energy.
The purpose of this paper was to examine water use estimation at hydel and thermal electric power plants in selected regions i.e. Coastal, Rayalaseema and Telangana regions of Andhra Pradesh. The study primarily focuses on the realistic fundamental premise that thermal electric and hydro electric energy generation is responsible for the largest monthly volume of water withdrawals in four seasons (i.e. summer, rainy, winter and post monsoon season) of a year. These enormous water withdrawals by these hydel and thermal power plants can have significant influence on local surface water resources. Water use at the power station level (by fuel type) can be estimated indirectly by using multiple regression analysis. In regression models, water use relationships are expressed in the form of mathematical equations, showing water use as a mathematical function of one or more independent (explanatory) variables. The mathematical form (eg. Linear, multiplicative and exponential) and the selection of the Right hand side (RHS) or independent variables depend on the category and on aggregation of water demand represented by Left Hand side (LHS) or dependent variable.
2.0 THEORETICAL AND CONCEPTUAL REVIEW OF LITERATURE: DIFFERENT APPROACHES OF WATER USE ESTIMATION
The various studies relating to water demand for thermal power plants and its significant determinants are reviewed for explicit understanding of thermal electric energy water use. Cootner, Paul and George O Golf (1965) have build upon a systematic model for estimating water demand in conventional steam electric utility industry. They have regarded water as a common factor input along with fuel. Here
TWD= f (Q_{f}, Cw, E_{He}, C_{WH })
Where in TWD = Thermal water withdrawal demand, Q_{f }= Quantity and cost of fuel, C_{w} = Cost of water, E_{He} = Economics of heat exchange and recycle and C_{WH}= other costs of thermal power plant associated with the disposal of waste heat.
In other words the quantity of the fresh water withdrawals depends upon the above mentioned factors. In another study Wollman and Bonem (1971) found that the quantity of fresh water withdrawals for steam electric power generation depends upon (1) Thermal efficiency (with higher thermal efficiency less heat will be dissipated. Due to this smaller amount of cooling water are needed) (2) The extent to which sea or brackish water can substitute for fresh water (3) The rate of recirculation. Recirculation is a function of price of water availability. Young and Thompson (1973) in their study identified three factors that affect water use in thermal electric energy generation. They can be listed as water pricing, change in generation, technology, price of electricity, price of substitutes used in electricity i.e. oil and gas, population and level of general economic activity. The other factors include waste and heat discharge to water and the changes in cooling technologies.
Gleick (1993) in his study reviewed the water requirement of electric energy. Taking as base of earlier studies, he estimated the consumptive water use in Electricity Generation Industry using different technologies. The system efficiency for conventional coal combustion (Once through Cooling Towers), natural gas combustion (Once Through Cooling Towers) and nuclear generation (CTs) stood at 35 percent, 36 percent and 40 percent. The estimates specifies that with the help of Once Through Cooling Technologies, the average consumptive use ranges from 1.2 m^{3}/MWH in case of conventional coal, for oil and natural gas consumption the average consumption use is less by 1.1 m^{3}/MWH , where as with cooling towers it was 2.6 m^{3}/MWH. For nuclear power generation the average consumptive use of water with the aid of CTs was more that stood at 3.2 m3/MWH. There is a need for use of high efficient technology in cooling towers for water conservation. Electric Power Research Institute 2002, estimated the evaporation water loss from recirculating towers i.e., roughly 480 gal/MWH for a coal fired power plant. Mortenson, 2006 in his study have provided a technological breakthrough i.e. small scale tests of one technology (that uses crosscurrents of ambient air for condensation) as a counteracting measure for these evaporation losses. By this technology the evaporation losses can be reduced to about 60140 gallons/MWH (that can be applied even to hotter climates). In value terms, EPRI 2004 notified that the savings from reduction of evaporation losses will be $870,000.
There are very few studies of determinants of water use in hydel and thermal electric generation. The literature available relating to water use estimations is very few. Water use experts have to opt for estimation methods for many of the water withdrawals classes i.e. domestic, agriculture and industry because of the true fact that many legal, financial and political constraints limit for getting the hard data. For instance water withdrawals in domestic and live stock water use are usually estimated by multiplying population figures by coefficient. In case of agricultural sector, the irrigation water withdrawals are often estimated by multiplying the acreage by assumed water requirements of the crop rather than by measuring actual water pumped and applied.
Snavely (1986), explicitly details the water use data collection programs and maintaining regional data base of the Great Lakes St. Lawrence River Basin States. The results are very much appealing indicating as how broad the range of estimation coefficient for water use can be within a geographic area with similar water availability. Mostly the estimated coefficients used for agriculture and domestic use vary by a factor of 10. The econometric studies relating to water use estimation in public supply use and thermo electric power use have the potential to explain temporal and geographic variability across USA. The aggregated water use estimates were provided by the National water Use Information Programme. These estimates primarily focus on measuring total water withdrawals (that includes annual extraction of fresh surface water and ground water) for the period 19801985 to 19901995 in each of 48 states of USA for public supply water withdrawals , domestic, commercial, irrigation and live stock. The saline water withdrawals were estimated for industrial, mining and thermal electric categories. The public supply water withdrawals are estimated within geographical area i during year t using a set of explanatory variables that includes air temperature, precipitation, price of water, median household income and others.
Cohn et.al (1989) and Christensen et.al (2000) have used examples of such kind by using statistical techniques. The shorter time period used has the advantage of highlighting the recent trend of declining water use since the 1980 compilation. The mean withdrawal for the period (19801995) clearly indicates that it was 183.7 gallons per capita per day. This average water withdrawals would decrease by 7.8 gpcd, if the state GDP per capita increased by $1000. The inclusion of this state GDP captures the effects of relative volume of non residential uses (along with their ability to pay for water). The model also indicates that US was able to withdraw 17.2 gpcd, because of its surface water rights in comparison with riparian law states. The inclusion of temperature and precipitation variables also clearly shows the effect of weather on water withdrawals and can be used in normalizing water use for weather. The model indicates that average per capita demand for water in the state decreases by 2.1gallons per day per one inch increase in precipitation and vice versa i.e. water demand increases during summer months. i.e. average temperature.
Billings and Jones, 1996 employed indirect estimation of water use in urban and municipal planning using coefficient based methods. It calculates water use for commercial, residential and industrial categories. They assume constant water use rates and ignores trends i.e. changes in water use due conservation, technological change or economic forces. Mullusky et.al (1995), Wood Well and Desjardin (1995) for Washington D.C. metropolitan area have employed this water use coefficients for three categories of water users i.e. simple family homes, multiple family homes and employment water use. Another approach of estimating National Water Use in USA includes Stratified random sampling followed by Census of Agriculture. They employed various methods of collecting data such as telephone, mail survey instruments to develop detailed country level estimates of national agricultural activities. According to Hutson et.al 2004 the thermo electric power water use refers to water that is removed from the ground or diverted from surface water sources (that includes fresh water and saline water) for use in the process of generating electricity with steam driven turbine generators. In this paper the term water withdrawals is used more often precisely. The term designates the amount of water that is extracted from natural water sources. Again it is essential to demarcate between water withdrawals and discharge as consumptive use. Water consumption is the quantity of water with drawn that is evaporated, transpired, incorporated in to crops, consumed by human or live stock.
At the end it can be said that different authors have notified different methods for estimation of water use for various uses of the economy. This paper employs multivariate models of water use for estimation of significant determinants of thermal and hydel water withdrawals.
Objectives of the paper
The objective is to determine if multiple regression models of unit hydel and
thermo electric water use have the potential
To identify significant determinants of total hydel and thermo electric water withdrawals across selected region wise power plants in AP using aggregated category wise water use estimates.
To estimate the future water withdrawals for hydel and thermal electricity generation plants expressed as cubic meters per second. (Cumecs) and cubic meters using the growth rate phenomenon.
The types of data used for estimation are monthly water withdrawals data (For surface fresh water resources)
Region level models for hydro and thermo electric water withdrawals
The potential dependent and independent variables for water withdrawals are identified for estimation purpose. Regional level data for thermal and hydel water withdrawals are more accurate data. The underlying reason being water withdrawals are usually metered.
Dependent Variable: Total Hydel Water Withdrawals
Total Thermal Water Withdrawals
Independent Variables of Hydel Power Plant:
(a) Reservoir levels, (b) Inflows, (c) Storage capacity, (d) Evaporation losses, (e) Tail water level and (f) Gross Head
Independent Variables of Thermal Power Plant:
(a) Demineralised water, (b) Boiler Feedback, (c) Condenser Cooling (d) Ash disposal, (e) Others include colony domestic, drinking, sanitation, fire fighting, back wash filter, (f) Installed generation capacity, (g) Actual electric energy production (h) Total no. of cooling towers, (i) Water temperatures in summer, rainy and winter.
Multiple Regression analyses were performed using the data related to category wise water use in power plant, generating facility and weather conditions from month wise 199596 to 200809 data in respective thermal and hydel power plants. The effect of variables such as quantities of water used exclusively for the production of electricity i.e. Boiler feed, Demineralised water, Condenser cooling, Ash Disposal, colony domestic (Drinking, Sanitation, Fire Fighting, Back wash filter ), installed capacity generation, number of cooling towers, cooling temperature and electric energy generation on total water withdrawals of thermal power plants are explicitly analyzed. In addition to this, the effect of variables such as reservoir elevation, storage capacity, tail water level, evaporation losses, inflows, gross head, actual generation on total hydel withdrawals have also been looked in to. This paper explores the structure of power plant level aggregated water use data based on corresponding and routinely collected economic and climatic data. The purpose of this enquiry is to determine if multiple regression models have the potential to explain the temporal and climatic variability across various thermal and hydel power plants in Andhra Pradesh using aggregated water use estimates and most importantly to identify significant determinants of total water withdrawals of thermal and hydel power plants. The statistical models examined here are derived using data estimates of total water withdrawals for hydel and thermo electric power use.
Specification of Mathematical Model
WHE_{im} = a +∑ bj X_{j }
j
Where WHE_{im } = Fresh water withdrawals for Hydel Electric Energy within region wise i during particular months m in a year.
X_{j }is a set of explanatory variables. (Mentioned above)
WTE_{im} = a +∑ bj X_{j }
j
WTE_{im }= Fresh water withdrawals for Thermal Electric Energy within region wise i during particular months m in a year.
X_{j }is a set of explanatory variables. (Mentioned above Coefficients a and bj can be estimated using multiple regression model.
Specification of the Econometric Model:
In Linear forms, these equations can be estimated as follows
Y_{t} = B_{1}+B_{2}X_{2}+B_{3}X_{3}+B_{4}X_{4}+B_{5}X_{5}+B_{6}X_{6}+B_{7}X_{7}+ µ
Model: 1 WTE_{im} = B_{1}+B_{2} CT+B_{3}DB+B_{4}CD+B_{5}AS+B_{6}WT+B_{7}AG+µ ……… (1)
Where, WTE_{im} = Water withdrawals for thermal electric energy in region i for particular months m.
CT = Condenser cooling (with Cooling Towers), DB = Demineralized water and Boiler Feed
CD = Colony Domestic, AS = Ash Slurries, WT= Water Temperature, AG= Actual generation
µ= random error term
Condenser Cooling: Water required for cooling hot turbines and condensers
Demineralized Water: Water that is, free of minerals and salts. Water runs through active resin beds to remove metallic ions and filtered through sub micron filter to remove suspended impurities.
Colony Domestic: Water that is used for the purpose of colony maintenance, drinking purpose and plantation.
Ash Slurries: As coal burns, it produces carbon –dioxide, sulphur –dioxide and nitrogen oxides. These gases together with lighter ash are called fly ash. The electro static precipitators remove all the fly ash and are mixed with water to make in to ash slurries.
Water temperature: Recording the temperature of water during summer, rainy and winter seasons.
Actual Generation: The generation of electricity that is actually generated apart from installed generation.
Model 2: WHE_{im} = B_{1}+B_{2} RE+B_{3}SC+B_{4 }TW+B_{5}GH+B_{6}WT+B_{7}AG+µ ……. (2)
Where WHE_{im}= Water withdrawals for hydel electric energy in region i for particular months m.
RE = Reservoir Elevation, SC= Storage Capacity ,TW= Tail water level, El= Evaporation losses, GH= Gross Head, WT= Water Temperature, AG= Actual Generation,µ= random error term
Reservoir Elevation: This is defined as the foot of the dam. i.e. the level from which the reservoir storage level and the height of the dam are measured.
Storage Capacity: This corresponds to the flood level usually designated as the upper limit of the normal operational range, above which the spill gates come in to operation
Tail water Level: Water immediately below the power plant. Tail water elevation refers to the level that water which can rise as discharges increase. It is measured in the feet above sea level. 1 foot = 0.305 meters.
Inflows: The inflow may be monsoonal rains or lakes, rivers. The average volume of incoming water, in unit period of time.
Evaporation Losses: Conversion of liquid to vapor state by latent heat. Water gets saturated in the form of vapor due to rise in water temperature.
Discharge: Volume of water released from power dam at a given time measured as cubic feet per second.
Gross Head: A dam’s maximum allowed vertical distance between upper stream’s surface water fore bay elevation and the down stream’s surface water (tail water) elevation at the tail race for reaction wheel dams.
Actual Generation: The amount of electricity actually generated apart from installed generation.
Selected power plants in three regions of Andhra Pradesh
Power Plant by
Fuel Type 
Rayalaseema Region  Telangana Region  Coastal Region 
Thermal  Rayalaseema Thermal Power Plant  .Kothagudaem Thermal Power Station O & M
.Kothagudaem Thermal Power Station Stage V 
Narla Tata Rao Thermal Power Plant 
Hydel  Nagarjuna Sagar Main Power House
Nagarjuna Sagar Left Canal Power House
Nagarjuna Sagar Right Canal Power House 
Srisailam Left canal power house
Srisailam right Canal Power House

The collection of data includes a monthly time series data analysis during the period (199596 to 200809). Analysis of hydel and thermal electric water use data in the existing power plants clearly indicates that there is wide variability in unitary thermal and hydel electric water use within the system. The multi variate regression procedures were used to identify the significant determinants of thermal and hydel water withdrawals in various power plants i.e. five hydel and four thermal power plants. The unit variability of unit water usage indicates that there is significant potential for water conservation in existing hydel and thermal electric power plants.
3.0 Approach and Methodology
The study includes three main components. (a) A series of site visits and interviews with power plant personnel. (b) Field surveys of selected hydel and thermal power plants of Andhra Pradesh (c) The multiple regression analysis of power generation data and other associated information.
Summary of site visits: Site visits for selected five hydel namely Nagarjuna Sagar Main Power House, Nagarjuna Sagar Left Canal Power House, Nagarjuna Sagar Right Canal Power House, Srisailam Left canal power house and Srisailam right Canal Power House and four thermal namely Rayalaseema Thermal Power Plant, Kothagudaem Thermal Power Station O & M, Kothagudaem Thermal Power Station Stage V and Narla Tata Rao Thermal Power Plant have been made to assess the overall performance scenario of power plants and also to examine the extent of water irregularities .Appraisal of Power Plant Survey: The research estimates of hydel and thermal Electric Energy water withdrawals are based upon the authenticated sources of data provided by respective hydel and thermal power plants of Andhra Pradesh Generation Corporation of India Limited. In order to transparently clarify the way that power generation officials responded to this kind of field survey in practice and to solicit information from them on factors responsible for water use at power generation facilities, site visits have been taken up. At various Power plants several personal interviews with power plant officers helped to identify the types of onsite water uses, the measurement of these uses and provision of information on various types of cooling systems and water use procedures employed by hydel and thermal electric energy generation facilities.
The purpose of conducting a series of personal interviews with power plant officials can be listed as follows:
(a) Scrutinize and examine the power generation water use and water withdrawals from intake (surface water) to discharge in various types of facilities.
(b) Observing the fact that all the water with drawals (hydel and thermal) are metered.
(c) Detailed analysis about important onsite uses of water and its significant determinants
(d) To obtain feedback on the cooling system level of water use in power stations.
Multiple Regression Models of Water Use
The principal sources of data used in the multi variate analyses of thermal and hydel power plants are most accurate and provides a fairly comprehensive review of plant characteristics, power generation and water withdrawal details. The data in electronic format and in official records was available for the years 199697 to 200809. The weather data i.e. especially related to water temperatures during summer, rainy and winter was collected in order to examine the influence of it on total thermal and hydel water withdrawals.
At the end it can be concluded that the site visits and field surveys helped to identify important concerns about water measurement and use at thermal and hydel electric power plants. Added to this, these factors have received attention in the development of models to describe hydro and thermal electric water use. All the above mentioned information proved very much useful in the design of data analysis that was used to develop water use bench marks.
4.0 RESULTS AND DISCUSSION: ESTIMATION AND INTERPRETATION OF MODEL SPECIFICATIONS
Hydel based Electric Energy Power Plants
Model Specification I Nagarjuna Sagar Main Power House
(Appendix table: A1)
In model 1 the estimated regression equation for total hydel water withdrawals is in the linear form as follows:
* * * *
WHE = 146.2380.080RE0.258SC+0.350TW+0.133GH+50.67AG
(3.96) (3.144) (119.87)
N= 154, R^{2} =0.99, f= 5543.05
 The estimated equation indicates that the total hydel water withdrawals are inelastic with respect to storage capacity. This kind of negative relationship indicates that the hydel water withdrawals are somewhat in responsive to changes in the storage capacity. The coefficients are statistically significant at 1 % level.
 The total hydel water withdrawals are elastic with tail water level and actual generation that hold a positive relationship. The coefficients are statistically significant at 5 % and 1 % level.
 The tratio of regression coefficients is highly significant for three independent variables namely SC, TW and AG. As the t ratio value is greater than 2.58 indicates that the relation between dependent variable and independent variables observed in the sample holds good.
 The t ratio of regression coefficient is not at all significant for other independent variables such as reservoir elevation and gross feet, as the t value is very small.
 The R^{2 }(coefficient of determination) is 0.99. It means that the independent variables tail water level, actual generation and storage capacity can explain 99 percent of variation in the dependent variable (WD) and remaining 1 percent variation is unexplained by the model. As R^{2 }is very high, the estimated equation is considered as an equation of very good fit.
 The overall model is statistically significant as f value is higher and more significant at 1% level. This clearly indicates that the regressors are significantly associated with dependent variable.
Model SpecificationII Nagarjuna Sagar Left Canal Power House
(Appendix Table: A2)
* * * *
WHE = 1660.7703.516RE21.705SC+9.653TW+491.286AG+0.130EL
(3.314) (4.16) (3.84) (15.67)
N= 166, R^{2}= 0.78, f = 116.22
 The estimated regression coefficients indicate that the best independent that have significant effect are storage capacity and actual generation with significant levels at 1 % for each of independent variables.
 The tratio of regression coefficients is highly significant with two independent variables namely storage capacity and actual generation. As t ratio value is greater than 2.58, it indicates that the relation between Hydel Water withdrawal and independent (SC) and (AG) observed in the sample holds good.
 The R^{2} is 0.78. It means that the independent variables SC and AG can explain 78 percent variation in the dependent variable and the remaining 22 % variation is unexplained by the model. The estimated equation is considered as an equation of very good fit.
 The overall model is statistically significant as f value is higher (116.22) and more significant at 1 % level. This indicates that the regressors SC and AG are significantly associated with dependent variable.
Model Specification III Nagarjuna Sagar Right Canal Power House
(Appendix Table: A3)
* * *
WHE = 6133.252+0.628 RL58.029 SC+0.414EL+37.493TW+486.057 AG
(7.314) (6.063) (16.232)
N= 166, R^{2}= 0.78, f value = 116.22
 The estimated regression coefficients indicate that the best independent variables that have significant effect are storage capacity and actual generation with significant levels at 1 % for each of independent variables.
 The tratio of regression coefficients is highly significant with two independent variables namely storage capacity and actual generation. The relation between water withdrawals and Storage capacity and actual generation in the sample holds good as the tvalue is greater than 2.58.
 The tratio of regression coefficients is not at all significant for other independent variables such as reservoir level, storage capacity and evaporation losses.
 The R^{2} is 0.78. It means that the independent variables SC and AG can explain 78 % variation in the dependent variable and remaining 22 % variation is unexplained by the model. The estimated equation is considered as the equation of very good fit.
 The overall model is statistically significant as f value is higher (116.22) and more significant at 1 % level. This indicates that the regressors are significantly associated with dependent variable (WD)
Model Specification IV Srisailam Left Bank Power House
(Appendix Table: A4)
* *
WHE = 2243.5010.766RE+1.195SC+57.47AG+0.592EL+4.24TW+0.000IF
(2.27) (18.81) (2.69)
N= 58 , R^{2}= 0.96, f value = 221.872
 The estimated regression coefficients indicate that the best independent variables that have significant effect are actual generation and tail water level with significant levels at 1 % and 10 % for independent variables.
 The tratio of regression coefficients is highly significant with three independent variables namely reservoir elevation, actual generation and tail water level. The tratio value is greater than 1.96 value for reservoir level and greater than 2.58 value for actual generation and tail water level. This indicates that the relation between WD and independent variables AG and reservoir elevation observed in the sample holds good.
 The t ratio of regression coefficients is not at all significant for other independent variables such as evaporation losses and inflows.
 The R^{2} is 0.96. It means that the independent variables reservoir level, actual generation and tail water level can explain 96 % of variation in the dependent variable and remaining 4% is unexplained by the model. Thus the estimated regression coefficient is considered as an equation of very good fit.
 The overall model is statistically significant as f value is higher (221.872) and more significant at 1 % level. This indicates that the regressors AG and TW are significantly associated with dependent variable. (WD)
Model Specification V Srisailam Right Bank Power House
(Appendix Table: A5)
* * *
Y = 7630.3801.78RE+0SC+56AG+0.051EL+0.627TW+0.289GH
(4.199) (4.3) (122.65)
N= 138 , R^{2 } = 0.99 and f value = 4.59
 The estimated regression coefficients indicate that the best independent variables that have a significant effect are storage capacity and actual generation with significant levels at 1 % level each of independent variable.
 The tratio of regression coefficients is highly significant with two independent variables namely storage capacity and actual generation. The t ratio value is greater than 2.58 for SC and AG that indicates that the relation between WD and independent variables SC and AG holds good.
 The t ratios of regression coefficients is not at all significant for other independent variables such as evaporation losses, tail water level and gross head.
 The R^{2} is 0.99. It means that the independent variables such as storage capacity and actual generation can explain 99 % variation in the dependent variable and remaining 1 % is unexplained by the model. Thus the estimated regression coefficient is considered as an equation of very good fit.
 The overall relationship was statistically significant as f value is 4.59 and more significant at 1 % level. This indicates that the regressors SC and AG are significantly associated with WD.
Thermal based Electric Energy Power Plants
Model Specification VI Kothagudaem Thermal Power Plant O &M
(Appendix Table: A6)
* *
Y= 787978.047 + 1.021CC2.130DB12.190CD+146.699 OT +1.152 AD+4616.497 CT817.112AG
(3.259) (3.841)
N= 84, R^{2} = 0.55, f value = 13.710
 The estimated regression coefficients indicate that the best explanatory (independent) variables with significant effect on quantity of water with drawals per Kilowatt hour are condenser cooling with cooling towers (Natural Draft cooling system) and ash disposal with significant levels of 5 % and 1 % level.
 The estimated equation indicates that the total thermal water withdrawals are elastic with respect to condenser cooling and ash disposal. This kind of positive relationship indicates that the thermal water withdrawals are responsive to changes in condenser cooling and ash disposal.
 The tratio of regression coefficients have expected signs and is highly statistically significant for two independent variables namely condenser cooling with Natural Draft CTs and Ash Disposal. The t ratio value is greater than 2.58.
 This indicates that the importance of technological alternatives (i.e. Condenser Cooling with natural draft CTs) is the significant determinant of water withdrawals. Next ash disposal takes second place as significant determinant of total thermal water withdrawals.
 The tratio of regression coefficient is not at all significant for other independent variables such as DM and Boiler feedback, colony domestic, others (Drinking, Sanitation, Fire fighting, Back Wash Filter), cooling temperature and actual electric energy generation.
 The R^{2} is 0.55. It means that the independent variables such as condenser cooling and ash disposal can explain 55 % of variation in the dependent variable and remaining 45 % variation is unexplained by the model. The estimated equation is considered as good fit.
 The overall model is statistically significant as f value is higher (13.710) and highly significant at 1 % level. This indicates that the regressor’s condenser cooling with Natural Draft CT’s and Ash Disposal are significantly associated with dependent variable WDs.
Model Specification VII Kothagudaem Thermal Power Station Stage V
(Appendix Table: A7)
* *
Y= 98233.879+0.873 CC+1.186AD+0.111 DB1688.373CT+32.019 AG
(20.91) (15.247)
N= 83, R^{2}= 0.97, f value = 706.164
 The estimated regression coefficients indicate that the best independent variables with significant effect on quantity of WD per million tonnes are Condenser cooling and ash disposal with significant levels at 1% level each.
 The tratio of regression coefficients have expected signs and is highly statistically significant for two independent variables namely Condenser cooling with natural draft CT’s and Actual Generation. The t ratio value is greater than 2.58. Here the significant determinant of WD’s are CC with natural draft CT’s. Next comes ash disposal as second good determinant.
 The t ratio of regression coefficient is not at all significant for other independent variables such as BF & DM, cooling temperature and Energy Generation.
 The R^{2} is 0.97. It means that independent variables such as CC and AD can explain 97 % of variation in the dependent variable (Water withdrawal) and remaining 3 % variation are unexplained by the model. Thus the estimated equation is considered as an equation of very good fit.
 The overall model is statistically significant as f value is higher (706.164) and highly significant at 1 % level. This indicates that the regressors condenser cooling with NDCT’s and Ash Disposal are significantly associated with Water withdrawal’s (Dependent Variable)
Model Specification VIII Rayalaseema Thermal Power Plant
(Appendix Table: A8)
*
Y = 10334.674+0.745 CC+8.725 BF+0.847 AS4.143 AG145.408 CT
(2.677) (3.007)
N= 35, R^{2} = 0.87 and f value = 33.145
 The estimated regression coefficients indicate that the best independent variables with significant effect on quantity of Water Withdrawal Condenser cooling with significant levels at 5%.
 The tratio of regression coefficients have expected signs and is highly statistically significant for one independent variables namely Condenser cooling with natural draft CT’s .The t ratio value is greater than 2.58. Here the significant determinant of WD’s are CC with natural draft CT’s.
 The t ratio of regression coefficient is not at all significant for other independent variables such as BF & DM, Ash Disposal cooling temperature and Energy Generation.
 The R^{2} is 0.87. It means that independent variables such as CC can explain 87 % of variation in the dependent variable (WD) and remaining 13 % variation are unexplained by the model. Thus the estimated equation is considered as an equation of very good fit.
 The over all model is statistically significant as f value is higher (33.145) and highly significant at 1 % level. This indicates that the regressors condenser cooling with NDCT’s are significantly associated with WD’s (Dependent Variable)
Model Specification IX Narla Tata Rao Thermal Power Plant
(Appendix Table: A9)
* *
Y = 139993.709 + 1.002CC 0.863CD + 1.031 AS 373.483 CT 56.843 AG
(1277.966) (19.88)
N= R^{2 }= 1.00, f value = 907849.564
 The estimated regression coefficients indicate that the best explanatory (independent) variables with significant effect on quantity of water with drawals per Kilowatt hour are condenser cooling with cooling towers ( Induced l Draft cooling system) and ash disposal with significant levels of 1 % and 1 % level.
 The estimated equation indicates that the total thermal water withdrawals are elastic with respect to condenser cooling and ash disposal. This kind of positive relationship indicates that the thermal water withdrawals are responsive to changes in condenser cooling and ash disposal.
 The tratio of regression coefficients have expected signs and is highly statistically significant for two independent variables namely condenser cooling with Induced Draft CTs and Ash Disposal. The t ratio value is greater than 2.58.
 This indicates that the importance of technological alternatives (i.e. Condenser Cooling with Induced draft CTs) is the significant determinant of water withdrawals. Next ash disposal takes second place as significant determinant of total thermal water withdrawals.
 The tratio of regression coefficient is not at all significant for other independent variables such as, colony domestic, cooling temperature and actual electric energy generation.
 The R^{2} is 1.00. It means that the independent variables such as condenser cooling and ash disposal can explain 100 % of variation in the dependent variable. This shows that we have accounted for almost all the variability with the variables specified in the model. The estimated equation is considered as very good fit.
 The overall model is statistically significant as f value is higher (907849.564) and highly significant at 1 % level. This indicates that the regressor’s condenser cooling with Induced Draft CT’s and Ash Disposal are significantly associated with dependent variable WDs.
The pertinent conclusion of this study is there may be significant potential for water conservation after having identified the significant determinants of total thermal water withdrawals i.e. condenser cooling and ash disposal. The choice of explanatory variable for eg: Induced draft and natural draft technological innovation was able to address the significant changes of water withdrawals.
5.0 CONCLUSION AND RECOMMENDATION
The thermal and hydel power plants sustenance is very much under stake due to major reason of fresh water shortages in power generation. The most sophisticated technology followed in advanced countries namely Concentrated solar thermal power integrated with combined system of conventional steam plant, Fresnel Solar Collector and Solar Flower Tower can be used as a replica even in developing countries like India though not cost effective in order to counteract the water shortage problem
REFERENCES
Benedy Kt Dziegielewski, Thomas Bik (August 2006), “ Water Use Bench Marks for Thermo Electric Power Generation” Project report, Southern Illinois University, United States
Geological Survey, 2004, USGS National Competitive Grants Program.
Gbadebo Oladosu, Stan Hadley, Vogt D.P. and Wilbanks J.J. (September, 2006), “Electricity
Generation and Water Related Constraints: An Empirical Analysis of Four South Eastern
States”, Oak Ridge National Laboratory, Oak Ridge Tennessee.
Sitanon Jesdapipat and Siriporon Kiratikarnkul, “ Surrogate pricing of water: The Case of micro Hydro –Electricity Cooperatives in Northern Thailand”.
Xiaoying Yang & Benedy Kt Dziegielewski (February,2007), “ Water Use by Thermo Electric power plants in the United states” Journal of the American Water Resources Association, Vol 43, No.1.
“Estimating Water Use in United States: A new Paradigm for National Use Water Use Information Programme”(2002),
http://books.nap.edu/openbook.php?record_id=10484&page=95
Data Sources
Annual Report on the Working of SEBs and Electricity Departments, Planning Commission, Various Issues
Administrative Reports of Andhra Pradesh Generation Corporation of India Limited (APGENCO),Various Issues. Field Level data of selected thermal and hydel power stations authenticated by APGENCO.
APPENDIX TABLES
Table: A1: Nagarjuna Sagar Main Power House
Variables Entered/Removed  
Model  Variables Entered  Variables Removed  Method  
1  acutal_generation, tail_water_level, Reser_elevation, Gross_feet, Storage_capacity^{a}  .  Enter  
a. All requested variables entered.  
b. Dependent Variable: water_discharge_cums  
Model Summary  
Model  R  R Square  Adjusted R Square  Std. Error of the Estimate  
1  .997^{a}  .995  .995  512.92868  
a. Predictors: (Constant), acutal_generation, tail_water_level, Reser_elevation, Gross_feet, Storage capacity  
ANOVA^{b}  
Model  Sum of Squares  df  Mean Square  F  Sig.  
1  Regression  7291771208.745  5  1458354241.749  5543.053  .000^{a}  
Residual  38675087.446  147  263095.833  
Total  7330446296.191  152  
a. Predictors: (Constant), acutal_generation, tail_water_level, Reser_elevation, Gross_feet, Storage capacity  
b. Dependent Variable: water_discharge_cums  
Coefficients^{a} 

Model  Unstandardized Coefficients  Standardized Coefficients  t  Sig.  
B  Std. Error  Beta  
1  (Constant)  146.238  1555.816  .094  .925  
Reser_elevation  .080  .093  .012  .865  .389  
Storage capacity  .258  .065  .091  3.966  .000  
tail_water_level  .350  .111  .031  3.144  .002  
Gross_feet  .133  .094  .026  1.419  .158  
acutal_generation  50.669  .423  1.041  119.869  .000  
a. Dependent Variable: water_discharge_cums  
Table: A 2 Nagarjuna Sagar Left Canal Power House
Variables Entered/Removed^{b}  
Model  Variables Entered  Variables Removed  Method  
1  evaporation, energe_bus, twl_ft, storage capacity, reservior_level^{a}  .  Enter  
a. All requested variables entered.  
b. Dependent Variable: water_drawals  
Model Summary  
Model  R  R Square  Adjusted R Square  Std. Error of the Estimate  
1  .864^{a}  .747  .739  2350.84646  
a. Predictors: (Constant), evaporation, energe_bus, twl_ft, storage capacity, reservior_level  
ANOVA^{b} 

Model  Sum of Squares  df  Mean Square  F  Sig.  
1  Regression  2626964399.664  5  525392879.933  95.068  .000^{a}  
Residual  889763133.646  161  5526479.091  
Total  3516727533.310  166  
a. Predictors: (Constant), evaporation, energy bus, twl_ft, storage capacity, reservior_level  
b. Dependent Variable: water_drawals  
Coefficients^{a}  
Model  Unstandardized Coefficients  Standardized Coefficients  t  Sig.  
B  Std. Error  Beta  
1  (Constant)  1660.770  501.102  3.314  .001  
reservior_level  3.516  3.411  .157  1.031  .304  
storage capacity  21.705  5.219  .538  4.159  .000  
twl_ft  9.653  2.510  .394  3.846  .000  
energy bus  491.286  30.765  .987  15.969  .000  
evaporation  .130  .508  .015  .255  .799  
a. Dependent Variable: water_drawals  
Table: A3 Nagarjuna Sagar Right Canal Power House
Model  Variables Entered  
1  generation bus, reservior_level, evaporation, storage capacity, tailwaterlevel^{a}  
b. Dependent Variable: water_drawals
Model Summary 

Model  R  R Square  Adjusted R Square  Std. Error of the Estimate  
1  .885^{a}  .784  .777  3767.05581  
a. Predictors: (Constant), generation bus, reservior_level, evaporation, storage capacity, tailwaterlevel  
ANOVA^{b}  
Model  Sum of Squares  df  Mean Square  F  Sig.  
1  Regression  8246365913.182  5  1649273182.636  116.222  .000^{a}  
Residual  2270513515.133  160  14190709.470  
Total  10516879428.315  165  
a. Predictors: (Constant), generation bus, reservior_level, evaporation, storage capacity, tailwaterlevel  
b. Dependent Variable: water_drawals  
Coefficients^{a}  
Model  Unstandardized Coefficients  Standardized Coefficients  t  Sig.  
B  Std. Error  Beta  
1  (Constant)  6133.252  838.604  7.314  .000  
reservior_level  .628  7.571  .016  .083  .934  
storage capacity  58.029  9.570  .832  6.063  .000  
Evaporation  .414  .810  .027  .511  .610  
Tailwaterlevel  37.493  21.598  .263  1.736  .084  
generation bus  486.057  29.945  1.045  16.232  .000  
a. Dependent Variable: water_drawals  
Table: A4 Srisailam Left Canal Power House
Variables Entered/Removed^{b}  
Model  Variables Entered  Variables Removed  Method  
1  inflow, Reservoir, evaporat, Actual generation, Tail water, storage_capacity^{a}  .  Enter  
a. All requested variables entered.  
b. Dependent Variable: water_withdra  
Model Summary  
Model  R  R Square  Adjusted R Square  Std. Error of the Estimate  
1  .981^{a}  .963  .959  1454.18057  
a. Predictors: (Constant), inflow, Reservoir, evaporat, Actual generation, Tail water, storage capacity  
ANOVA^{b}  
Model  Sum of Squares  df  Mean Square  F  Sig.  
1  Regression  2815082375.894  6  4.692E8  221.872  .000^{a}  
Residual  107846697.597  51  2114641.129  
Total  2922929073.491  57  
a. Predictors: (Constant), inflow, Reservoir, evaporat, Actual generation, Tail water, storage capacity  
b. Dependent Variable: water_withdra  
Coefficients^{a}  
Model  Unstandardized Coefficients  Standardized Coefficients  t  Sig.  
B  Std. Error  Beta  
1  (Constant)  2243.501  2527.275  .888  .379  
Reservoir  .766  .337  .239  2.272  .027  
storage capacity  1.195E6  .000  .000  .004  .997  
Actual generation  57.476  3.055  .953  18.814  .000  
evaporat  .592  .939  .081  .631  .531  
Tail water  4.237  1.572  .248  2.695  .010  
inflow  .000  .002  .017  .339  .736  
a. Dependent Variable: water_withdra  
Table: A5 Srisailam Right Canal Power House
Model  Variables Entered  Variables Removed  Method  
1  Gross head, Tailwaterlevel, actual generation, Evaporation, storage, Reservoir  .  Enter  
a. All requested variables entered.  
b. Dependent Variable: water withdrawals  
Model Summary 

Model  R  R Square  Adjusted R Square  Std. Error of the Estimate  
1  .998^{a}  .995  .995  631.39218  
a. Predictors: (Constant), Gross head, Tailwaterlevel, actual generation, Evaporation, storage, Reservoir  
ANOVA^{b}  
Model  Sum of Squares  df  Mean Square  F  Sig.  
1  Regression  1.099E10  6  1.832E9  4.596E3  .000^{a}  
Residual  5.222E7  131  398656.090  
Total  1.105E10  137  
a. Predictors: (Constant), Gross head, Tailwaterlevel, actual generation, Evaporation, storage, Reservoir  
b. Dependent Variable: water withdrawals  
Coefficients^{a}  
Model  Unstandardized Coefficients  Standardized Coefficients  t  Sig.  
B  Std. Error  Beta  
1  (Constant)  7630.380  1817.341  4.199  .000  
Reservoir  .178  .322  .027  .553  .581  
storage  .000  .000  .068  4.288  .000  
actual generation  56.314  .459  1.022  122.651  .000  
Evaporation  .051  .139  .005  .365  .716  
Tailwaterlevel  .627  .334  .059  1.874  .063  
Gross head  .289  .320  .036  .904  .368  
a. Dependent Variable: water withdrawals  
Table: A6 Kothagudaem Thermal Power Plant O &M
Variables Entered/Removed^{b}  
Model  Variables Entered  Variables Removed  Method  
1  energy generation , cooling temp, DM Water & Boiler Feed back , Ash Disposal , Condenser Cooling , Colony domestic , (Drin, Sani, Firefigh, Backwarhfiler) ^{a}  .  Enter  
a. All requested variables entered.  
b. Dependent Variable: Total water consumption  
Model Summary 

Model  R  R Square  Adjusted R Square  Std. Error of the Estimate  
1  .747^{a}  .558  .517  289298.132  
a. Predictors: (Constant), energy generation , cooling temp, DM Water & Boiler Feed back , Ash Disposal , Condenser Cooling , Colony domestic , (Drin, Sani, Firefigh, Backwarhfiler)  
ANOVA^{b}  
Model  Sum of Squares  df  Mean Square  F  Sig.  
1  Regression  8.032E12  7  1.147E12  13.710  .000^{a}  
Residual  6.361E12  76  8.369E10  
Total  1.439E13  83  
a. Predictors: (Constant), energy generation , cooling temp, DM Water & Boiler Feed back , Ash Disposal , Condenser Cooling , Colony domestic , (Drin, Sani, Firefigh, Backwarhfiler)  
Coefficients^{a}  
Model  Unstandardized Coefficients  Standardized Coefficients  t  Sig.  
B  Std. Error  Beta  
1  (Constant)  787978.047  1.334E6  .591  .557  
Condenser Cooling  1.021  .313  .551  3.259  .002  
DM Water & Boiler Feed back  2.130  5.717  .038  .373  .710  
Colony domestic  12.190  15.642  .250  .779  .438  
(Drin, Sani, Firefigh, Backwarhfiler)  146.699  201.477  .467  .728  .469  
Ash Disposal  1.152  .300  .409  3.841  .000  
cooling temp  4616.497  10000.955  .039  .462  .646  
energy generation  817.112  1096.318  .295  .745  .458  
a. Dependent Variable: Total water consumption  
Table: A7 Kothagudaem Thermal Power Plant Stage V
Variables Entered/Removed^{b}  
Model  Variables Entered  Variables Removed  Method  
1  Energy Generation, ASH DISPOSAL (MT), Cooling Temperature , Boiled Feed and DM plant Regeneration, COOLING TOWER MAKEUP (MT)^{a}  .  Enter  
a. All requested variables entered.  
b. Dependent Variable: TOTAL CONS. (MT)  
Model Summary  
Model  R  R Square  Adjusted R Square  Std. Error of the Estimate  
1  .989^{a}  .979  .977  64726.513  
a. Predictors: (Constant), Energy Generation, ASH DISPOSAL (MT), Cooling Temperature , Boiled Feed and DM plant Regeneration, COOLING TOWER MAKEUP (MT)  
ANOVA^{b} 

Model  Sum of Squares  df  Mean Square  F  Sig.  
1  Regression  14792454121098.932  5  2958490824219.786  706.164  .000^{a}  
Residual  322593153570.889  77  4189521474.947  
Total  15115047274669.820  82  
a. Predictors: (Constant), Energy Generation, ASH DISPOSAL (MT), Cooling Temperature , Boiled Feed and DM plant Regeneration, COOLING TOWER MAKEUP (MT)  
b. Dependent Variable: TOTAL CONS. (MT)  
Coefficients^{a}  
Model  Unstandardized Coefficients  Standardized Coefficients  t  Sig.  
B  Std. Error  Beta  
1  (Constant)  98233.879  76676.230  1.281  .204  
COOLING TOWER MAKEUP (MT)  .873  .042  .577  20.912  .000  
ASH DISPOSAL (MT)  1.186  .078  .484  15.247  .000  
Boiled Feed and DM plant Regeneration  .111  .978  .003  .114  .910  
Cooling Temperature  1688.373  2158.260  .014  .782  .436  
Energy Generation  32.019  115.619  .005  .277  .783  
a. Dependent Variable: TOTAL CONS. (MT)  
Table: A 8 Rayalaseema Thermal Power Plant
Variables Entered/Removed^{b}  
Model  Variables Entered  Variables Removed  Method  
1  Cooling Temp, Ash slurry, Actual Generation, Power Generation, Boiler feed, Condenser cooling, BCW^{a}  .  Enter  
a. All requested variables entered.  
b. Dependent Variable: Water consumption  
Model Summary  
Model  R  R Square  Adjusted R Square  Std. Error of the Estimate  
1  .934^{a}  .873  .846  1324.085  
a. Predictors: (Constant), Cooling Temp, Ash slurry, Actual Generation, Power Generation, Boiler feed, Condenser cooling, BCW  
ANOVA^{b}  
Model  Sum of Squares  df  Mean Square  F  Sig.  
1  Regression  3.487E8  6  5.811E7  33.145  .000^{a}  
Residual  5.084E7  29  1753200.788  
Total  3.995E8  35  
a. Predictors: (Constant), Cooling Temp, Ash slurry, Actual Generation, Power Generation, Boiler feed, Condenser cooling, BCW  
Coefficients^{a}  
Model  Unstandardized Coefficients  Standardized Coefficients  t  Sig.  
B  Std. Error  Beta  
1  (Constant)  10334.674  3861.078  2.677  .012  
Condenser cooling, BCW  .745  .248  .432  3.007  .005  
Boiler feed  8.725  4.628  .244  1.885  .069  
Ash slurry  .847  .501  .230  1.692  .101  
Power Generation  .595  .388  .138  1.532  .136  
Actual Generation  4.143  5.478  .077  .756  .456  
Cooling Temp  145.408  94.141  .109  1.545  .133  
a. Dependent Variable: Water consumption  
Table : A 9 Narla Tata Rao Thermal Power Plant
Variables Entered/Removed^{b}  
Model  Variables Entered  Variables Removed  Method  
1  Energy Generation, Condenser cooling & BCW (KL), Cooling Temperature , Ash slurry water (KL), Colony Domestic (KL)^{a}  .  Enter  
a. All requested variables entered.  
b. Dependent Variable: Totalwaterconsumption  
Model Summary  
Model  R  R Square  Adjusted R Square  Std. Error of the Estimate  
1  1.000^{a}  1.000  1.000  50290.302  
a. Predictors: (Constant), Energy Generation, Condenser cooling & BCW (KL), Cooling Temperature , Ash slurry water (KL), Colony Domestic (KL)  
ANOVA^{b}  
Model  Sum of Squares  df  Mean Square  F  Sig.  
1  Regression  11480277367590772.000  5  2296055473518154.000  907849.564  .000^{a}  
Residual  42994946072.977  17  2529114474.881  
Total  11480320362536844.000  22  
a. Predictors: (Constant), Energy Generation, Condenser cooling & BCW (KL), Cooling Temperature , Ash slurry water (KL), Colony Domestic (KL)  
b. Dependent Variable: Totalwaterconsumption  
Coefficients^{a}  
Model  Unstandardized Coefficients  Standardized Coefficients  t  Sig.  
B  Std. Error  Beta  
1  (Constant)  139993.709  137540.088  1.018  .323  
Condenser cooling & BCW (KL)  1.002  .001  .987  1277.966  .000  
Colony Domestic (KL)  .863  .584  .001  1.476  .158  
Ash slurry water (KL)  1.031  .052  .018  19.879  .000  
Cooling Temperature  373.483  3763.081  .000  .099  .922  
Energy Generation  56.843  138.469  .000  .411  .687  
a. Dependent Variable: Totalwaterconsumption  
3 thoughts on “Econometric Models to Water Use Estimation in Power Plants: An Experiential Analysis”