METHODS AND PROCEDURES

Households that participated in Tucson Water’s low-consumption toilet rebate program in 1991- 1992 were selected to study the functioning of older low-consumption toilets. The functioning of rebate toilets approximately 7 years after installation, or about one-third of the way into their expected life, could then be studied. Data from Tucson Water records showing the installation address, the number of bathrooms, and the type and number of toilets installed were received for 477 households. This number was reduced to the sample size of 200 as follows: one-third of the 477 households were randomly removed from consideration. The 318 households remaining were plotted geographically. Using the geographic center of Tucson, the data set was divided into four quadrants. Approximately 64 homes were randomly selected from each quadrant to assure 50 homes per quadrant, plus several reserve sites in the event of problems obtaining data from selected sites. These sites were well spread-out across an area that is highly diverse in terms of housing age, cost and style.

Data loggers were attached to the water meter on the main water line going to each house for four days(4). Using fifty loggers at a time, four rounds were necessary to obtain data from 200 homes. The logging was completed over 19 days in December 1998. Data from 30 of the households were not usable due to various problems with the connection to the meter or the functioning of the logger, leaving 170 usable traces to be analyzed. Occupants were not notified that their meter was being logged. As a result, the data are not subject to bias due to the “Hawthorne Effect,” when participants change their behavior due to knowledge that their actions are being monitored.

The 170 traces contained data on 20 different models of low-consumption toilet rebated during the program. Table 1 shows the models of toilets and their flush mechanism type.

The data were analyzed using the Trace Wizard™ software for end-use analysis developed by Aquacraft Engineering, Inc.(5) Toilet flushes produce consistent patterns, which can be identified according to their peak flow, duration, and volume of flush (DeOreo et al. 1996). All toilet flushes occurring in the four-day trace were identified and assigned to a particular toilet in the household. The volume of each flush and the number of flushes for each toilet was recorded. The average volume per flush for each toilet and the number of flushes in the trace were then recorded in a database.

A follow-up survey was conducted, initially by phone. The survey asked occupants to confirm the number and type of toilets in their household. Occupants were also asked to state whether any of four common problems were present with any of their toilets: frequent double flushing, toilet runs after flush, flapper leak leading to periodic flushing or refilling, or frequent clogging. Respondents were then asked to rate the functioning of each of their toilets on a 5-point scale with 1 corresponding to very dissatisfied, 2 dissatisfied, 3 neutral, 4 satisfied, and 5 corresponding to very satisfied. The rest of the survey helped assess factors such as whether in-tank toilet bowl cleaners were being used, whether respondents had replaced flush-valve flappers, and the number of adults and children in the household. A copy of the survey is found in Appendix I.

4. Meter-Master model 100EL data loggers were used with a 10 second data storage interval. Meter-Master data loggers are manufactured by F.S. Brainard & Co., P.O. Box 366, Burlington N.J. 08016.
5. Aquacraft Engineering, Inc., 2709 Pine St., Boulder, CO 80304.

Table 1. Low-Consumption Toilet Models and Their Flush Mechanisms

A large percentage of the original participants in the toilet rebate program had moved, and often the phone number of the new occupants was not listed. Also, some households could not be contacted by phone after multiple tries at different times of day or night. Surveys identical to the phone survey were mailed to all households that could not be contacted by phone.

Records from Tucson Water of rebated low-consumption toilets sold to each address, along with the total number of bathrooms and toilets in each household, were combined with surveys of the current residents to inventory toilet types for each household. This expected inventory was compared with the average volume for each toilet identified from the data taken from each household to determine whether low-consumption rebate toilets were flushing at unusually high volumes. When two or more of the same model of low-consumption toilet were being used in a residence, individual low-consumption toilets were often indistinguishable in the data-logging trace. In this case, the average of the same-model low-consumption toilets was reported.

In addition, several common problems with toilet function could be detected from the traces, including high flush volume, double flushing, and flapper leaks. If the average volume of a low- consumption toilet was greater than 2.2 gallons per flush, it was recorded in the database as having a high flush volume.

Only low-consumption rebate toilets with average flush volumes greater than 2.2 gpf were counted as having high flush volumes. Because flush volumes greater than 2.2 gpf were not allowed in testing for meeting ASME/ANSI standards, this seemed an appropriately conservative cutoff volume to account for normal variations in toilet flush volume. Toilet flush volumes can vary slightly depending on human factors such as how long the handle of a gravity flush model is held down and the water pressure at each residence. Officials with Tucson Water state that the average water pressure varies by 5 to 10 psi across pressure zones in the central city, and by 10 to 20 psi in the higher elevation portions of the service area. Differences in pressure of 20 psi would be large enough to cause small systematic variations in flush volume in some toilets, but not enough to cause flush volumes of 1.6 gallon toilets to consistently be measured above 2.2 gpf.

The number of times that each toilet was flushed in succession was recorded as an indication of a possible multiple flushing event to clear one load of waste. Toilet flushes from the same device within four minutes of each other were counted as multiple flush events. Figure 1 shows a sample trace of a multiple flush event.

The proportion of possible multiple flushing events to total flushing events for each toilet in each home was recorded. Toilets with double-flush percentages greater than or equal to 15 percent were counted as having a problem with double flushing. This is a conservative standard. With an average of 28 flushes per toilet over a four day period during the study, this means at least four multiple flushing events, or one per day, were needed for multiple flushing to be considered a problem.

Figure 1. Example of Double Toilet Flushing

The traces also revealed some toilets flushing or refilling on their own. Flapper leaks result in a slow draining of the water in the tank of toilets. As the water drains, the refill valve eventually turns on. As a flapper leak continues, it is detectable on a trace as a recurring pattern of water use spikes at regular intervals. Figure 2 shows a flapper leak recurring at a particular interval. This pattern is easiest to detect at night, when there is little other water use. Only recurring, regular interval leak patterns that were detectable at night were counted as flapper leaks.

Figure 2. Example of Recurring Toilet Flapper Leak