SIMULATION OF PROJECTED WATER DEMAND
Water demand in the Union County WSA was simulated to year 2050 for two scenarios. The normal-growth scenario reflects normal historical growth for population, employment, and median household income for Union County. The high-growth scenario considers the recent period (1995 to 1998) of rapid growth for population, employment, and median household income for the WSA.
The Forecast Manager module of the IWR-MAIN software provides accounting and analysis tools for estimating future municipal and nonmunicipal water demand. The user's manual and suite description provide additional details for much of the discussion presented in this section of the report (Planning and Management Consultants, Ltd., 1999). The water-use forecasting algorithm of Forecast Manager is built to operate on data corresponding to the study area, water-use sectors and subsectors, months, and forecast years. The study area for the model is the Union County WSA. Forecasts were devised for the residential (single-family subsector) and nonresidential sectors (commercial and industrial subsectors) of municipal use and for the nonresidential sectors (commercial and industrial subsectors) of nonmunicipal use for years 2010, 2020, 2030, 2040, and 2050. The projection of residential water use to 2050 was simulated using a linear-predictive model with IWR-MAIN to incorporate demographic, socioeconomic, and climatic variables. The projection of non-residential water use to 2050 was simulated using a constant-rate model incorporating water use per employee and employment projections.
Residential water use for the Union County WSA was calculated using the number of households on public supply and the average per-household water use per day. Per-household water use is a function of demographic, socioeconomic, or climatic variables (Boland and others, 1984; Linaweaver, 1965; Maidment and others, 1985).
Residential water use for 2010, 2020, 2030, 2040, and 2050 was projected with a linear-predictive model using an estimated model intercept (inelastic demand) and linear coefficients for demographic, socioeconometric, and climatic factors. The factors commonly assumed to control daily household water use are monthly temperature and precipitation, marginal price, and median income. Daily per-household water use can therefore be expressed in the following form:
q = a + bt + cp + dm + ei, (1)
where, for each month and year,
q is the estimated rate of daily use per single-family household,
a is the inelastic demand, and
b, c, d, and e are linear coefficients of elasticity of per-household use for the factors t (temperature), p (precipitation), m (marginal price), and i (median income).
Values for the inelastic demand and the coefficients of elasticity in equation (1) are developed or estimated prior to calculating projected water use. Based on regression analysis of monthly data, changes in temperature and precipitation may account for only 11 percent of the observed temporal variation in residential water use. Though climatic variables (temperature and precipitation) may change over the period of scenarios run, defining these changes was beyond the scope of this effort, and temperature and precipitation were held constant. Based on regression analysis of monthly variations in water use, the results validate residential water-use concepts and indicate that water use would increase with increasing temperature and decrease with increasing price.
Median household income is constant for each system for each month for 1998, but is likely to change from year to year in the future. Therefore, the coefficient of elasticity for median household income was determined from literature values (William Y. Davis, Planning and Management Consultants, Ltd., oral commun., 1999). The coefficient for median household income was calculated based on 1998 daily per-household water use and 1998 median household income. The coefficient is interpreted as an elasticity term and is determined as follows:
e = ( q1998 * βs)/X1998, (2)
e is the coefficient of the explanatory variable, median household income;
q is the estimated daily water use per household, 193 gallons per day (gal/d) in 1998;
β is 0.4, the literature value for elasticity of per-household water use for median household income; and
X is the median household income, in thousands of dollars. The value is 32.458 for 1998.
The elasticity (β) is a dimensionless measure of the relation between a percentage change in water use and a percentage change in median household income when other factors affecting demand remain unchanged (Boland and others, 1984). The literature value for β is 0.4 (Planning and Management Consultants, Ltd., 1995). The analysis assumes that for a 1-percent increase in median household income, water demand increases 0.4 percent.
For equation (1) inelastic water use is 58.1, water use associated with climatic factors is constant at 89, elasticity coefficient for marginal price is -15.7, and elasticity coefficient for median household income is 2.378. Per-household water use (q) is calculated as:
q = 58.1 + (89) - (15.7 m) + (2.378 i), (3)
where, m and i represent estimates of marginal price and median income.
Under the scenarios of normal- and high-economic growth in this study, climatic conditions are held constant and small adjustments are made to the marginal price. Extreme drought or larger changes to the marginal price can be used as alternative assumptions to develop alternate modeling scenarios. Marginal price is expected to change only slightly over the forecast years. The range of values for water use associated with marginal price is from -31 gal/d in 1998 to -42 gal/d in 2050. These negative numbers indicate a tendency for increases in marginal price to decrease water use. Median income is projected to change substantially over the forecast years and has the greatest influence on increasing water use per household for each forecast year. In the high-growth scenario, the component of water use associated with median income ranges from 77 gal/d in 1998 to 167 gal/d in 2050.
Total residential water use for years 2010, 2020, 2030, 2040, and 2050 is determined as follows:
Qy = (qy * hy)/106, (4)
Qy is the total residential water use in year (y), in million gallons per day;
qy is the per-household water use in year (y), as determined by equation 3; and
hy is the number of households served by public supply in year (y).
Commercial and industrial water use projections to 2050 were determined using a constant-rate model based on the amount of water use per employee in each subsector and the projected number of employees to 2050. The base-year (1998) per-employee water-use rate was calculated from the base-year water use and the number of employees for the commercial and industrial subsectors (Planning and Management Consultants, Ltd., 1999). Future changes in water use per employee in the commercial and industrial subsectors are unknown. The water use per employee for 1998 remains constant for all of the forecast years for each subsector. The change in the number of employees (N) determines the change in the water-use forecast from year to year. Thus, the quantity of water use in a given subsector, month, and forecast year is calculated as:
Q s,m,y = N s,m,y * q s,m,y, (5)
Q is gallons per day used in subsector (s) in month (m) in year (y),
N is the number of employees in subsector (s) in month (m) in year (y), and
q is the average daily water-use rate per employee in subsector (s) in month (m) in base year for 1998.
With the constant-rate model, the change in employees (N) explains the change in the water-use forecast from year to year. For municipal use, the per-unit use rates in gallons per employee per day for the commercial and industrial subsectors are 58 and 19 gallons per employee per day (ged), respectively; for nonmunicipal use, 12 and 143 ged, respectively.
The results of the simulation for a high-growth scenario show that average demand could increase 131 percent from 2.9 Mgal/d in 1998 to 6.7 Mgal/d in 2050. Peak daily demand for municipal water was estimated as 1.65 times the average daily rate of use for both scenarios. The ratio 1.65 is the average peak demand ratio for public-supply systems using ground water in Mississippi (American Water Works Association, 1992).
Housing, employment, climatic, and economic data were prepared as input to the water-use models in IWR-MAIN. Several assumptions (about the character of the data for the base year and about the structure of socioeconomic conditions in future years) were necessary to model the Union County WSA. These assumptions are detailed within the respective data sections and in Appendix A. Data were prepared for the municipal (residential and nonresidential sectors) and nonmunicipal (nonresidential sectors) use for the base year 1998 and for future years 2010, 2020, 2030, 2040, and 2050.
Occupied housing units for the residential sector are counted as single-family households. Union County, in 1998, averaged 2.51 persons per household (U.S. Department of Commerce, 1992a). The number of households served by public supply for the forecast years (table 2) was estimated using projected population growth and assuming 2.51 persons per household. Population projections were derived by Dr. James H. Eblen (TVA, Economic Development, Technical Services, written commun., 1999). See Appendix A in this report for an explanation of the methodology and the projections for population for Union County, Mississippi. The number of occupied-housing units served by public supply was incrementally increased from 80 percent in 1998 to 95 percent of the total occupied-housing units in 2050 for the normal-growth scenario and to 97 percent for the high-growth scenario.
Other residential data are marginal price, median household income, temperature, precipitation, and water-conservation savings. The water and wastewater price-rate structures for each system were used to specify marginal price for the base and future years. An average marginal price of $1.97 for the systems was used for the base year. For those areas in the WSA most likely to acquire sewer lines by year 2050 (Glenn Duckworth, Executive Director, Union County Development Association, oral commun., 1999), marginal price was adjusted for future years, and a revised marginal price was input to the model. The model assumes that customers connected to public-supply systems with sewer capacity will use public wastewater treatment. For the purposes of this model, the dollars are expressed as 1998 constant dollars.
The only complete assessment of median household income in Mississippi occurs in each decennial census. For the base and forecast years, median household income was estimated using the methodology described in Appendix A. The dollars are expressed as 1998 constant dollars.
Public/unaccounted water use was estimated as a percentage of the total municipal use. For the base year of 1998, the public/unaccounted water use was about 23 percent, which reflects the average rate observed for the public-supply systems. For future years, the percentage remains constant through time at 15 percent, which is the water-industry average for Mississippi (American Water Works Association, 1992).
The average daily maximum temperature for each month and the total monthly precipitation for the base year 1998 and for the period 1956 to 1998 (table 3) were used as input for the forecast years. Monthly data for years 1956 to 1998 at each of these stations were evaluated to define average climatological conditions for future years for residential water demand.
Water-conservation savings that would result from installing low-flow plumbing fixtures as required by the Federal Energy Conservation Act of 1992 were factored into the model. For this study, a conservative estimate of water savings was used to account for the uncertainties of the performance of the low-flow technology. Water use per household was reduced for all estimated new housing units that would be built from 1994 to 2050. The estimated savings are for one person per household instead of 2.51 persons per household. Estimated water savings of 14 gallons per unit per day were entered for 50 percent of the occupied households on public supply for the year 2010, for 75 percent of the occupied households on public supply in 2020, and for 100 percent of the occupied households on public supply for 2030 through 2050 (Planning and Management Consultants, Ltd., 1995; American Water Works Association, written commun., 1997). In the residential model, the estimated per-unit use is reduced by the given amount before the unit use is multiplied by the number of housing units (Planning and Management Consultants, Ltd., 1999).
The constant-rate model projected municipal water use for the commercial (nonmanufacturing) and industrial (manufacturing) subsectors based on the estimated increases in employees. The employee counts (Appendix A) were multiplied by the corresponding unit-use coefficients of gallons per employee per day. The coefficients were derived from water-production records maintained by the public-supply systems and from employee counts reported to the Union County Development Association (Glenn Duckworth, Executive Director, written commun., 1999).
Nonresidential water usage for nonmunicipal water supplies, such as self-supplied industry, is held constant for the base and forecast years for the normal- and high-growth scenarios. This decision assumes that additional nonresidential water for new or expanding facilities would be provided by public-supply systems.
The IWR-MAIN linear-predictive and constant-rate models applied to the Union County WSA were used to estimate water demand for years 2010, 2020, 2030, 2040, and 2050. The estimates for the municipal residential and nonresidential sectors and the nonmunicipal nonresidential sectors were aggregated to yield totals for the WSA for each year. The values are reported as two significant figures. Estimates for a normal-growth scenario from 1998 to year 2050 (table 4) show that:
Estimates for a high-growth scenario show that average demand could increase 131 percent from 2.9 Mgal/d in 1998 to 6.7 Mgal/d in 2050. Peak daily demand for municipal water was estimated as 1.65 times the average daily rate of use for both scenarios. The ratio 1.65 is the average peak demand ratio for public-supply systems using ground water in Mississippi (American Water Works Association, 1992). Historical (1990, 1995, and 1998) and projected (2000, 2010, 2020, 2030, 2040, and 2050) water withdrawals for a normal- and a high-growth scenario for the municipal sector are shown in figure 6. An average annual increase of 1.03 percent was used to estimate baseline water use for areas outside of Union County.
The water-demand models were used primarily to test assumptions and the effects that various assumptions or changes would have on water use in the county rather than as a predictive tool to generate absolute values showing future water use. As with any model, the degree of uncertainty increases as the length of time of the projections increase. Projecting 50 years involves assuming many political, environmental, economic, and technical factors will not shift radically. If the assumptions are changed (for example, population decreases in the area) the water-demand results will change. The results depend on the validity of the assumptions.
The uncertainty in the forecast of water demand is embedded in the projections of values for population, employment, and median household income for normal- and high-growth scenarios (Appendix A) and assumptions about water conservation. Together, the normal- and high-growth projections provide a range within which growth can reasonably occur as summarized in table 5. Further, within the residential water-demand model, uncertainty is introduced in calculating the coefficient of elasticity for median household income. The elasticity used to calculate the coefficient is a literature value (0.4) rather than one determined by site-specific data.