"IRP Approach To Water Supply Alternatives for Duck River Watershed,"
December 27, 2001

For presentation at Tennessee AWRA Conference, April 2002
William W. Wade,
Energy and Water Economics,
Columbia TN, 38401 Tel: 931-490-0060

The south central region of Tennessee was left without a reliable water supply to meet future needs when TVA stopped construction of the Columbia Dam on the Duck River. The Duck River region extends from the City of Columbia and Spring Hill, home to the Saturn automotive plant, upstream to Tullahoma, and encompasses Maury, Marshall, Coffee and Bedford Counties. This region has experienced significant population growth in the last 15 years; forecast growth is equally significant.

As part of the agreement not to construct the Columbia dam, TVA reserved easements on the Fountain Creek watershed for a possible future impoundment. Other options include the raising of the Normandy Dam impoundment level. As with any water supply planning problem, all proposed options had advantages and disadvantages.

The Duck River Agency commissioned a study to evaluate the various options for supplying the region a 50-year source of water. The study used state-of-the-art Integrated Resource Planning (IRP) methods to compare each option's contribution to supply reliability and instream flow requirements required by regulations. The study revealed that DRA Watershed agencies face four targets that must be met from Normandy Reservoir downstream to City of Columbia:

• Normandy Reservoir minimum water level;
• Flow Rate for wastewater assimilation and water supply at Shelbyville -- 165 cfs;
• Flow Rates for wastewater assimilation at Columbia (130 cfs) and for ecosystem protection (100 cfs);
• Future water demands for multiple agencies in the watershed thru 2050.

The December 2001 study revealed how the various alternatives met the four targets and called for added studies to complete missing data. This short article discusses the application of Integrated Resource Planning methods within the project.

1 Downstream Hydrologic Determinants of Supply Requirements

A prior study by TVA assumed that during drought conditions most of the Duck River inflow comes from the minimum release supplied by Normandy Dam to meet the Shelbyville target. During low rainfall periods in the summer, TVA assumed that165 cfs flow at Shelbyville results in ~130 cfs flow at Columbia. Actual River flow data reveal a different statistical pattern. Flows at the USGS Columbia station are mostly higher than the flows at Shelbyville under the most rigorous drought conditions.

• Of 601 daily observations when flows at Shelbyville are below 165 cfs, Columbia flows are higher on 538 days (90% of the days).
• Flows at Columbia average 83 cfs higher than low flows at S’ville.

Chart 1 shows the flow data arrayed to emphasize the few days in the 21 year history when Columbia flows are lower than S’ville flows. This record is used to measure water supplies in the Duck River to meet Columbia and Spring Hill demands and instream flow requirements.

2 Upstream Hydrologic Determinants of Supply Requirements

DRUC, which withdraws and treats water from Normandy Reservoir, reports that providing high quality drinking water under conditions of drought and expected future daily spike demand is difficult and expensive. (20 mgd = daily spike demand.) Chart 2 shows that DRUC’s target operating range is between 865 - 875 elevation. TVA modeling of the 20 mgd demand brought the reservoir down to 858. Whether DRUC can achieve drinking water quality, without excessive treatment costs, and meet the 165 cfs Shelbyville flow target remains to be studied if demand target is 20 mgd.

3 Can Duck River flows at Columbia meet future withdrawal demands and instream flow minimum target?

Statistical analysis allows hydrologic data post Normandy dam closure to be "run past" future demands to determine if minimum instream flow is breached. Table 1 shows the equivalent benchmark minimum flows that will meet future demands without breaching instream flow criteria. If flows drop below these benchmarks, added supplies are needed. The crucial question is how much supply for how many days.

Water supply shortfalls are measured in terms of frequency, duration and magnitude. Table 2 shows the results of the statistical analysis for five demand scenarios measured against the 100 cfs benchmark. (Other benchmarks are examined within the project.)

• Line 1 shows the number of times in 21 years that flows breach 100 cfs.
• Line 2 converts this to an average annual occurrence--1 day increasing to 18 days.
• Line 3 shows the annual probability of shortfall.
• Line 4 represents the average duration if shortfall occurs.
• Line 5 shows the average daily flow shortfall in cfs.
• Line 6 calculates the needed new water supply to avoid the shortfall.

Table 2 results imply:

• A 21 % chance of any shortfall in 2020 increasing to 68 % by 2050, high demand, or 47 % low demand.
• The average daily shortfall would range near 15 - 16 cfs, when it occurs.
• Shortfall would last ~3.4 days in 2020 with or without extraordinary industrial growth upstream of Columbia;
• Duration of the 2050 average shortfall would increase to ~5 days; ~7 days with added industrial growth. More than one summer occurrence could be expected.

Table 3 adds two measures of magnitude.

• Maximum one-day shortfalls in 21 years range from 42 to 83 cfs over the scenarios.
• Maximum shortage event once in 21 years last 8 to 25 days and range from 124 to 228 million gallons without added industrial growth; 432 mil. gal. with new industry.

A storage solution must target the maximum likely shortage event. Table 3 showed this to range from 124 to 228 million gallons. [380 - 700 AF.] Chart 3 interpolates the estimates to show the growth in this maximum requirement from 2020 to 2050.

Table 4 adds another measure of shortage magnitude for the130 cfs benchmark.

• Maximum shortfalls for one day in 21 years range from 72 to 113 cfs over the scenarios w/o industrial growth.
• Maximum shortage event one time in 21 years ranges from 25 - 37 days with 228 to 794 million gallons shortfall; 38 days with 1,087 mil. Gal. with new industry.

4 New Water Supply Requirements

Added supply that will provide ~15-20 cfs during a few days in the summer will meet future average water demands and protect the 100 cfs requirement.

Protecting the 21-year drought of record requires 228 million gallons in storage to meet demand in a 25-day 2050 drought. (Conservation might also offset this demand.)

Protecting the higher Columbia benchmark, 130 cfs, will require substantially more new water supply.

• Protecting the 21-year drought of record requires 794 million gallons to meet a 37-day 2050 drought; 1,087 million gallons to meet assumed new industry demand.

5 Alternatives to Enhance Supply Reliability

Objective of IRP is to find most economic way of adding increments of reliability to meet policy targets.

• Cost of new water supply created by alternative only can be amortized over new water required to meet targets.
• Other purposes than water supply could absorb some costs not born by DRA.
• Costs of effects of not meeting DRA reliability targets are needed to set targets.
• How much to spend depends on cost of missing target.

Supply reliability planning entails a water management strategy that provides the watershed with a reliable and affordable water supply for the next 50 years. Four DRA reliability targets emerged from the study:

• Maintain Normandy above 860 elevation. [or ? (to be determined).]
• Meet 120/155 cfs at Shelbyville. [depends on Shelbyville wastewater treatment.]
• Avoid breaching instream flow benchmarks at Columbia more than 1 day every five years. [or ? (to be determined). (Flow benchmark may need further support.)]
• Meet all consumptive demands in the watershed [except during droughts of some quantifiable measure (to be determined).]

The study determined that DRA must establish its own reliability targets before evaluating adequacy of resource options. Cost effects of less than 100% reliability of meeting any target need to be compared to cost of achieving target. Sensitivity of Duck River habitat to flow targets needs added study.

Reliability targets require a storage solution to supplement summer-fall shortages with wet season surpluses. Available storage options:

• Store more water in Normandy by raising dam.
• Store existing water in Tims Ford:
•     To use for pipeline to Duck River.
•     To meet Tullahoma demand, assuming Tullahoma shifted to Tims Ford.
• Store water behind Fountain Creek dam.

DRA's challenge is to find the most economic mix of resources to achieve an acceptable reliability objective. An affordable alternative with acceptable risk may be preferable to elimination of any chance of shortage at much higher costs. Solution depends on the costs of effects of shortage compared to costs of supply options.

Empirical determinants of the best mix are hydrologic, economic and regulatory driven.

• What added supply do you need to achieve acceptable reliability?
• What can you pay for?
• What can you permit?

Four of the six alternatives are subjected to hydrology analysis to show how they improve the 100 cfs standard against the 2050 high demand requirement.

Reliability tests of this benchmark are presented on Table 5 for TVA alternatives: How do the alternatives meet the 100 cfs criterion?

Table 5 shows information needed to evaluate supply alternatives -- effect of the alternative on likely shortage, magnitude and duration.

• Line 3 shows that alternatives reduce likelihood of a shorfall from 2 years in 3 to 1 year in 3 or 4.
• Line 2 shows that the resulting average occurrence would be reduced from 8.1 days per year to 1.3 days per year.
• Line 4 shows that when a shortfall occurs, average duration is reduced from 5 days to 3.4 days. (Conservation anomalous result is accurate.)
• Line 5 shows that the average daily shortfall remains 15 - 17 cfs, while likelihood and duration decline.

Table 6 shows the simplified but corrected needed supplies and added storage based on the worst case analysis. Worst case requirements are shown in top of Table 6. Approx. new supplies in storage are shown on bottom.

• Fountain Creek planned as 46,000 AF. supply. [>100,000 million gallons]
• Normandy planned as 17,000 AF. added supply. [5,500 million gallons]

6 Reliability Summary

Hydrologic modeling reveals likelihood, duration of shortages, average and worst case amounts of needed new water.

About 15 cfs, 3 days duration, 1 year in 4: 2020.

About 16 cfs, 18 days duration, 2 years in 3: 2050.

Max shortage event:

• 100 cfs benchmark: 1 event in 20.7 years: 228 mil gal, 25 days duration
• 130 cfs benchmark: 1 event in 10.7 years: 794 mil gal, 37 days duration.

Low-flow days (100 cfs standard) could exceed 20 cfs

• about 1 day annually, low 2050 forecast;
• about 2 days annually, high 2050 forecast.

Clearly, the magnitude of the water supply problem in the Duck River Watershed depends on policy targets chosen with a clear view of their economic cost. Instream flow requirements and measures to offset demand growth each greatly influence needed new supply facilities. The problem needs a solution; but creative solution steps will likely reduce size and cost of facilities needed to assure reliable water supplies for the next 50 years and adequate protection for the sensitive ecosystem of the Duck River.