NATURE OF RESIDENTIAL WATER USE
AND EFFECTIVENESS OF CONSERVATION PROGRAMS

In the August 1998 issue of Colorado Water, Michelsen, McGuckin, and Stumpf summarized the results of their effort to evaluate the effectiveness of residential price and nonprice programs. They used a "macro" approach and developed three estimating models. The Regional model compares water use patterns across cities. The Season Specific model is a variation of the Regional model that looks at water use behavior during specific seasons of the year. Finally, the City Specific model evaluates water use patterns in individual cities. In all cases, historical monthly water use data were utilized to do this analysis. Their results indicate that water price has a significant and negative impact on water use but that water demand is very price-inelastic. Thus, increasing water rates as a conservation measure will not cause a major decline in water use. Their results indicate that nonprice conservation programs can be effective but the results were mixed. Based on their monthly database, they concluded that outdoor water use does vary with monthly temperature, but not with monthly precipitation. A general conclusion of these authors is that:

S S Overview of Our Urban Water Demand Studies -- Beginning in 1993, Professor Heaney and a graduate student, Lynn Buhlig, began exploring the nature of urban water use and the possible effectiveness of water conservation. Using the City of Boulder as the case study and relying on aggregate monthly water use data for the entire city, we attempted to estimate the effectiveness of a variety of conservation practices that had been installed beginning in 1988. Data from 1971 through 1987 were used to describe the pre-conservation water use patterns. These data were compared to the post-conservation period of 1988 through 1994. A wide variety of conservation practices were installed by the City of Boulder including public education, a demonstration xeriscaping garden, increasing water rates, automation of the City's irrigation systems, and rebate programs for buffalo grass and soil moisture sensors. The results of this aggregate analysis using monthly data were disappointing. No statistically significant difference in the pre and post-conservation water use patterns could be discerned. This does not mean that the conservation practices were ineffective; rather, it means that the use of monthly data for the entire city probably disguises the impact of a small change in one component of water use. Based on this finding, we decided to move from doing statistical analysis of city-wide monthly water use data to evaluating individual houses. The problem was how to do detailed, non-intrusive measurements of household water use. The largest micro study that had been done was work by Brown and Caldwell (1984). This study sampled only a small number of houses and some of the metering was intrusive which may have affected the usage patterns.

Bill DeOreo, a consulting engineer in Boulder, solved the measurement problem by developing a computerized sensing device that is attached to the water meter. It measures flow into the house at ten-second intervals. Signal processing software was developed to convert the ten-second flow signals to individual water using events. The initial evaluation of this technique was done in cooperation with the Water Conservation Office of the City of Boulder as part of Peter Mayer's 1995 MS thesis. The Heatherwood neighborhood near Boulder was the selected study area. The results were very encouraging.

After graduation, Peter Mayer joined Bill DeOreo at Aquacraft and they promoted the idea to cities across North America and to the American Water Works Association Research Foundation. As a result, a $900,000 monitoring study was initiated. For each of 12 cities across North America, a sample of 1,000 houses was selected based on evaluation of local demographics and historical water use. A questionnaire was sent to each of these 1,000 houses. The average response rate was 46%. Based on the returned questionnaires, a sample of 100 houses was selected. Then, detailed monitoring was done on each of these houses during two 14-day periods, one warmer and one cooler. Data was successfully obtained from all but 12 of the 1,200 homes. About 28,000 complete days of water use data were collected including more than 1.9 million water-use events (toilet flushes, showers, clothes washer cycles, faucet usage, irrigation, etc.). Graduate students from the University of Colorado were employed to work on this project as part of their MS thesis research. This research project ended earlier this year and the results are now becoming available. A brief summary of findings to date is presented below. More detailed information about this entire effort can be found in a series of reports, papers, and theses, i.e., Buhlig (1995), Mayer (1995), DeOreo et al. (1996), DeOreo and Mayer (1996), Mayer et al. (1997), Courtney (1997), Harpring (1997), Stadjuhar (1997), or by contacting http://www.aquacraft.com.

S S Demographics of Study Participants -- The study group consists of a wide variety of single family homes. Study homes included mansions in gated communities and dilapidated one bedroom cabins. The landscapes ranged from lush turf grass and elegant xeriscape to horse pastures, hardscape to untamed weeds. The average household size in the study was 2.8 people and the median annual household income was between $50,000 and $60,000. Seventy-seven percent of survey respondents had completed at least some college and nearly 20 percent reported having either a Master's or higher degree. Nearly 92 percent of the surveyed homes were owner occupied and 8 percent were rental units. Of the study homes, 67.8 percent were built before 1980, 23.5 percent were built between 1980 and 1992, and 4.2 percent were built since 1993 when new plumbing codes went into effect.

S S General Results -- The 12 study sites represent a diverse collection of single-family water use patterns. In each of the 12 cities, a sample of 1,000 houses was selected. One year of historical metered water use was obtained from billing records for each of the 12,000 houses. Annual water use and estimated indoor and outdoor water use for each city is shown in Table 1. Indoor water use is estimated by averaging water use during the non-irrigation season. The majority of residential water use in Boulder (57%) and Denver (60%) is for outdoor purposes, primarily lawn watering. While the variability in indoor water use for cities across North America is low, it is much higher for outdoor water use. The results of the detailed measurements of water use in 100 houses in each of the 12 cities are presented below.

S S Indoor Water Use -- Indoor water use patterns for Boulder and Denver are compared to indoor use in the other 10 cities in Table 2. These results are based on the four weeks of continuous measurements of household water use for 1,200 houses across North America. Toilets are the major use of water indoors comprising 26.7% of the total. Clothes washers (21.6%), showers (16.7%), faucets (15.7%), and leaks (13.7%) are the other major components of indoor water use. The distribution of indoor water use is quite stable across the major water use components. The main sources of variability are in minor uses and leaks. The average indoor water use rates per capita for Boulder and Denver are 64.9 and 69.2 gallons per capita per day, respectively. The 12 city average indoor water use is 69.7 gallons per capita per day. These results for indoor water use are somewhat higher than previous studies that estimated indoor water use at about 60 gpcd (Maddaus 1987). The major source of the difference is probably in how leaks are evaluated. It is difficult to separate leaks into indoor or outdoor. The value for leaks shown in Table 2 assumes that leaks are chargeable to indoor water use. If they were assigned to outdoor water use, then the average per capita indoor water use rate would decrease to about 60 gpcd. Indoor residential water use per capita is quite stable in the United States reflecting the fact that indoor water use is for relatively essential purposes.

Indoor water use does not vary significantly over the year. Some daily variability occurs between weekdays and weekends. Peak usage occurs during the early morning hours of 7 to 10 am. Most of this peak is due to toilet and shower use. Toilet flushing continues at a similar rate for the rest of the day and into the evening. On the other hand, showers are taken primarily in the morning. Peak clothes washing activity occurs from 9 am to 1 pm. In general, water use in houses declines during the middle of the day since fewer people are at home. Use increases in the evening as people return home and prepare dinner, and then reaches its lowest level between midnight and 6 am when people are asleep. A general discussion of individual indoor water use components is presented below.

Toilet Flushing: Toilet flushing is the most regular and predictable of all of the indoor water uses with an average of 18.5 gpcd. Conservation options for toilets have focused on reducing the gallonage per flush from 4-5 gallons to 1.6 gallons which is mandated nationally in the plumbing codes beginning in 1993. An important concern with regard to lower volume per flush is that people would double or triple flush. Mayer et al. (1998) divided the NAREUS database into those houses that had only ultra-low flush (ULF) toilets and those that didn't. The results, shown in Table 3, indicate the same number of flushes per day with the ULF houses using only 9.5 gpcd as compared to 19.5 gpcd for non-ULF houses, a major savings of 10 gpcd. The Boulder sample only contained 1.0% of houses that fell into the ULF category while Denver had 6.9% (Mayer et al. 1998). As people replace toilets around the country, the impact of using ULF toilets will become apparent. It is evident from Table 3 that double flushing is not a problem with ULF toilets.

The volume per flush can be reduced to 0.5 gallons using pressurized systems. This technology may gain more widespread use in the future. Dual flush toilets are employed in Australia wherein the user selects whether to use more or less flushing water depending upon the need.

Clothes Washing: Clothes washers use an average of 15.0 gpcd. The traditional Monday wash day has been replaced by a more uniform pattern of clothes washing which is done throughout the day with peaks in the morning and early afternoon. More efficient clothes washers are expected to reduce water use per load by about 25 percent. The timing on clothes washing could be affected by electric or water utility rates that provide time of day incentives and disincentives. For example, water users in Great Britain tend to wash clothes late at night to take advantage of lower electricity rates.

Showers and Baths: Showers (11.6 gpcd) are much more popular than baths (1.2 gpcd) for all 12 cities in the NAREUS study. For Boulder, Colorado, the morning shower is the predominant time for this activity. The other peak in showering occurs during the evening. Showers are taken on a daily basis in Boulder. Thus, no significant variability occurs from day to day. The main conservation option for showers is to use low-flow showerheads.

Results to date indicate only limited reduction in water use since users did not set the older showerheads to the higher flow rates. Federal law mandates a maximum flow rate for showers of 2.5 gallons per minute (gpm). Results of the NAREUS study indicate that most people set their shower flow rate below this level. Thus, conservation savings may not be that significant (Mayer et al. 1998)

The results of the NAREUS study indicate that the average shower used 17.2 gallons and lasted for 8.2 minutes and the average flow rate was 2.1 gallons per minute (gpm). Most showers use between 5 and 20 gallons of water. This indicates that on average people shower at a flow rate below the 1992 plumbing code standard of 2.5 gpm. The LF shower homes used an average of 29.9 gpd and 11.3 gpcd for showering, while the non-LF shower homes used an average of 34.4 gpd and 13.4 gpcd. The net savings for the LF shower homes is therefore 2.1 gpcd. A more significant difference was observed in the mean daily per capita shower duration of the LF and non-LF shower homes. While the occupants of non-LF shower homes averaged 4.6 minutes per person per day of showering, occupants of the LF homes averaged 5.7 minutes per person per day. Nevertheless, the net difference in water use between the two groups is 2.1 gpcd.

Faucet Use: Faucet use includes drinking water, water for washing and rinsing dishes, flushing solids down the garbage disposal, shaving, and numerous other personal needs. Faucet use averages 10.9 gpcd. No breakdown among these uses is available although one can make educated guesses as to the amounts of water used for these purposes. Best estimates of actual drinking water use are in the range of 0.25 to 0.5 gallons per capita per day with a mean of 0.35 gallons per day (Cantor et al. 1987). Garbage disposals add about 1 gpcd to total indoor consumption (Karpiscak et al. 1990). Faucet use requires the highest water quality because it is the potable water source.

Dishwashers: Dishwashers are a relatively minor water use and newer dishwashers are being designed to conserve energy and water. Present per capita water use averages only 1.0 gpcd.

Water Use for Cooling: For some houses, and for many commercial and industrial establishments, water use for cooling is a significant part of the water budget. Swamp coolers are used in the more arid areas of the United States. Karpiscak et al. (1994) estimate that residential evaporative coolers use about 6 gpcd in Tuscon, Arizona. Because of the relatively small number of houses using coolers, the average usage is quite low, only 0.4 gpcd.

S S Outdoor Water Use -- Whereas indoor residential water use is very constant across the United States and does not vary seasonally, irrigation water use varies widely from little use to being the dominant water use. Also, it varies seasonally. The 12 cities in the NAREUS are not a representative sample of the United States with regard to climate types. Also, the amount of natural precipitation that occurred during the study periods can have a significant impact on the results. Nevertheless, the results certainly suggest the potential major impact of irrigation on average and peak water use.

Irrigation water use follows a definite pattern of high use rates in the morning and evening with low use rates during the day and late at night. Thus, these customers are following the common recommendations to not water during the middle of the day. Watering late at night is discouraged because of the noise from some types of sprinklers.For the entire NAREUS study, outdoor water use averaged 85,800 gallons per house per year as was shown in Table 1, significantly more than the 61,800 gallons per house per year for indoor water use. Of course, these 12 cities do not constitute a representative sample of all cities in North America. Nevertheless, the dominance of outdoor water use in the more arid western United States is apparent. In Boulder and Denver, outdoor water use averaged over the entire year exceeds indoor water use for the residential users. Thus, for residential areas in the more arid and warmer parts of the country, lawn watering is the largest single use on an annual average basis and is the dominant component of peak daily and hourly use during the summer months. In more arid areas, evapo-transpiration (E-T) requirements are much greater than natural rainfall. In warmer parts of the country, even those with abundant rainfall, e.g., Florida, irrigation water use rates are high because of the long growing season which includes some dry periods. Irrigation water use is a major input to the urban water budget during the growing season. A growing number of people are installing automatic sprinkling systems. These systems tend to use more water than manual systems (Mayer 1995). Also, the timers on these systems are seldom adjusted. Thus, lawn watering occurs even during rainy periods. Experience with soil moisture sensors to control sprinkling use has been mixed. Automatic sprinkling systems do offer the potential for more efficient use of water if they are properly calibrated and operated (Courtney 1997).

Peak hourly use in Boulder, Colorado occurs between 6 and 8 am and is caused predominantly by irrigation (Harpring 1997). Indoor water use at 6 am is about 7.5 gallons per house while the total water use at the same time is about 41 gallons per house. Thus, irrigation constitutes over 80% of the peak hourly use. Options for reducing outdoor water use include using less water loving plants, applying water more efficiently, reducing the irrigated area, and using nonpotable water including stormwater runoff and treated wastewater (Courtney 1997). Sakrison (1996) projects a potential decrease of 35 % in the demand for irrigation water in King County, Washington if higher density urbanization occurs. For King County, the main way that water use is managed is by restrictions on outdoor water use for landscaping. A maximum permissible E-T is allotted that forces the property owner to reduce the amount of pervious area devoted to turf grass. Stormwater runon to the pervious area can be used for an extra credit.

Lawn watering has increased in the United States as population migration occurs to warmer, more arid areas. Also, urban sprawl means much larger irrigable area per dwelling unit. Lawn watering needs are a dominant component of peak water use in urban areas. Reuse of treated wastewater and stormwater for lawn watering appears to be very attractive possibilities for more sustainable communities.

S S Summary and Conclusions -- The results of these process-oriented monitoring studies during the past four years provide a major improvement in our understanding of the nature of residential water use. For the 12 cities studied, indoor per capita water use averaged 69.3 gpcd with toilets, clothes washers, showers, faucets, and leaks being the largest indoor end use components. Cost-effective reduction in indoor use can be achieved by using low-flush toilets. This change is occurring nationwide due to the requirements of the national plumbing codes. Retrofitting showerheads is less effective since people do not operate showers at the higher flow rates anyway. Continuing improvements in household appliances are expected to significantly reduce indoor water use. Leaks are primarily the result of faulty toilet flapper valves and miscellaneous faucet and irrigation system leaks and can be repaired. Overall, for indoor water use, the picture is relatively optimistic in terms of reducing per capita water use. The current per capita use of about 65-70 gpcd should be reduced to 40-45 gpcd when existing conservation measures are used for all residential areas. This reduction saves not only on water supply costs but also on wastewater treatment costs since virtually all of the indoor water use must be collected and treated at the wastewater treatment plant.

While indoor water use is expected to decline as described above, the gains in reducing indoor water use may be offset by increases in outdoor water use. Outdoor water use exceeds indoor water use in more arid parts of the country. Also, outdoor water use constitutes the majority of the peak summer demand that taxes the capacity of urban water systems. The trend towards lower density housing increases the irrigable area per capita. Also, more people are installing automatic sprinkling systems. People vary widely in how they use water outdoors. This causes much uncertainty in estimating peak flow rates. A concerted effort is needed to devise more effective ways to reduce outdoor water use in urban areas. Intensive monitoring is needed to evaluate how irrigation water is actually used in urban areas. The possibility of reusing treated wastewater and stormwater for lawn watering should be given serious consideration, especially as the requirements for more stringent water treatment are imposed on cities. It is increasingly difficult to justify providing a very high level of treatment to all of the water brought into a city only to have the majority of it used to irrigate landscapes and flush toilets.