This article was originally published in the January/February 1994 issue of Home Energy Magazine. Some formatting inconsistencies may be evident in older archive content.



| Back to Contents Page | Home Energy Index | About Home Energy |
| Home Energy Home Page | Back Issues of Home Energy |



Home Energy Magazine Online January/February 1994



Everything I Know about Energy-Efficient Showerheads I Learned in the Field


by Mike Warwick and Curtis Hickman

Mike Warwick is a program manager at Pacific Northwest Laboratories in Portland, Oregon. Curtis Hickman is in the program evaluation section at Bonneville Power Administration.

In an end-use metering study conducted by Pacific Northwest Laboratories, low-flow showerheads saved less energy and water than expected. The results underscore the importance of using metered data to design conservation programs.

The Bonneville Power Administration (Bonneville) provides wholesale electric power to more than 100 retail distribution utilities in the Pacific Northwest. Faced with growing power demand, Bonneville embraced energy conservation as a resource alternative in 1980, and includes energy-efficient showerheads in residential conservation programs implemented by its retail utility customers. Bonneville gave these utilities the freedom to design their own distribution programs, resulting in distribution of more than 30 models of energy-efficient showerheads, using both professional and owner-installation methods.

Showerheads appear simple on the surface, but many variables affect retrofit savings. Over half of all single-family homes have more than one shower, and retrofits are measured in terms of both households and showers per household. Savings also vary depending on the pre-existing and retrofit showerheads' water flow rates.

Pre-existing flow rates and post-retrofit showerhead performance were both unknowns, so Bonneville's Evaluation Section asked the Pacific Northwest Laboratories (PNL) to conduct a field study to assess the energy-savings potential of energy-efficient showerheads, barriers to their installation, and limits to their effectiveness.1 We conducted the study in two phases: installation and field measurement, and energy savings analysis.

The first phase, installation and field measurement, involved collection of site and occupant data, and development of an engineering equation, or savings algorithm, to estimate savings. The energy savings analysis provided actual savings results that were used to benchmark the savings algorithm so that savings could be estimated under a variety of conditions, such as when not all the showerheads in a home were replaced, or when replacement showerhead flow rates varied by brand.

A Few Good Homes

We designed the field study around 154 single-family homes that Bonneville had already been end-use metering for several years through its Regional End-use Metering Project (REMP).2 We attempted to recruit all homes, with a goal of 120 participants. We offered professional showerhead installation, repair of minor plumbing problems, return of original showerheads upon request, and $40. While REMP study participants are normally willing volunteers for these additional studies, the $40 incentive made a critical difference in recruiting participants.

Once sites were recruited, field staff visited each home to check water flow levels before and after retrofitting each showerhead, install the showerheads, measure water pressure, collect site data, and conduct an occupant survey. We measured water flow rates with a Micro-Weir (see Low-Flow Showers Save Water--Who Cares? HE July/Aug '91 p.27). To avoid introducing errors from different showerhead designs, we selected a single showerhead model for the retrofits, an Ondine brand showerhead rated at 2.5 gallons per minute (gpm), identified from a consumer satisfaction survey (see Does `Low-Flow' Still Mean Low Satisfaction? HE Jul/Aug '91 p.22). We selected a 2.5 gpm device because it represented the types of showerheads distributed by Bonneville in its program. At the time, it was generally assumed that showerhead retrofits would reduce water flow by about 50%, enough to yield statistically valid results in a pre- versus post-retrofit comparison of annual hot water electricity use.

We started the retrofits in the fall of 1991. The initial 12 site visits went smoothly, but the pre-retrofit water flow measurements averaged 3.9 gpm, less than the 5 gpm we expected and generally used in savings estimates. To increase the flow-rate difference between the two showerheads and get statistically significant data, we selected a more efficient showerhead model, retrofitting the balance of sites with an ETL Spa-2000 brand showerhead rated at 2 gpm, which actually performed at 1.7 gpm. The critical site characteristics that emerged in the first phase of the study were initial water flow and flow rate of retrofit showerheads, along with the fraction of showerheads replaced in each home.

`Homes Don't Save Energy--People Do'

Since many homes have more than one shower, participation alone is no guarantee of universal showerhead replacement. Participation results must be couched in terms of both the number of sites participating and the number of showerheads actually installed (see Table 1).

Site results

The success of a conservation program is usually based on the number of sites that participate, but participation can be measured several ways. Our results were as follows:

Program participation is the number of households who initially agreed to have showerheads installed. Of 150 sites contacted, 111, or 74%, agreed to a site visit and potential showerhead replacement.

Program penetration is the fraction of homes that have showerheads installed compared to total homes contacted. Only 98 of 150 study homes had at least one showerhead installed, for a 65% program penetration rate among all households. Alternatively, if only the 111 recruited participants are used to gauge participation, those 98 sites represent an 88% participation rate among interested households.

Program persistence is the number of initial participants who accept showerhead installation and still have at least one in place after the program ends. Without a follow-up site visit, it is difficult to verify persistence. Our study measured persistence in two ways. First, we surveyed participants by telephone to document their levels of satisfaction and any changes that might affect savings. The survey was distributed about eight months after showerhead replacement. The other source of information came from participant requests for return of their original showerheads.

Only 36 sites responded to survey questions about post-retrofit showerhead replacement. Six indicated they removed at least one of the energy-efficient showerheads, indicating an average persistence of 83% at 8 months.

Measure Results

Measure participation is based on the number of showerheads distributed compared to the total number of showers in participating homes. The 150 sites included in the study had an estimated 240 showerheads. A total of 158 showerheads were actually replaced, for a measure participation rate of 66%.

Measure penetration gauges program effectiveness in terms of showerhead replacement. The 98 homes participating in the study had 161 showers, so 158 showerhead retrofits results in a 98% penetration rate among participating showers. This is important, because the key to achieving savings is getting measures installed. This high penetration rate was a result of two related factors. By using professional installers, we ensured the showerheads were actually installed. (Programs that rely on residents rather than professional installation have a low measure penetration rate, and thus fewer savings.) Secondly, the study was designed to allow installers to correct minor plumbing problems that would have otherwise prevented showerhead retrofits or reduced the savings--for instance replacing non-standard showerarms and leaky tub diverter spouts. (Most showers are found in tub/shower combinations where the shower is activated by a diverter valve. If the valves don't seal, some of the water leaks through the bathtub spout instead of going through the showerhead.) These repairs increased the installation rate by 10%, from 88% to 98%.

Measure persistence, like program persistence, is difficult to gauge. Unfortunately, our satisfaction survey did not include data on the number of showerheads replaced in the post-retrofit period. As a result, our only source of measure persistence data came from participant's requests to get their old showerheads back. After 15 months, we had returned 15 showerheads, implying a measure persistence rate of 91%. Only one of these homes indicated on the satisfaction survey that they had replaced their showerheads. If the return requests and survey respnses are added together, they suggest a persistence rate of 86%.

Water Savings

Savings can be expressed in absolute and relative terms. We had access to comparative data on total hot water energy use, along with billing data. However, the hot water energy use data included the combined total for water heating and stand-by energy use--energy consumed by a water heater when the water is not being used.

A previous PNL study of these sites indicated that stand-by energy use was about one-quarter of the total (1,200 kWh out of 4,600 kWh annually), an important rule of thumb.

Although we took one-time water flow measurements, we collected no data on cumulative water use, so it was not possible to estimate actual water savings. But hot water savings are presumably similar to relative electricity savings. Flow rate changes are specific to the showerhead brand and model. Changes for both brands of showerhead used in the study, and study averages, are shown in Table 2, but these averages mask a broad range of observed preexisting flow rates.

Surprising Results

At the outset of the study, we expected existing showerhead flow rates to average around 5 gpm. In fact, flow rates that high were found in less than 10% of the showerheads, and flow rates at about one-half of the sites were between 2.4 and 4 gpm! About a quarter of the sites had flows over 4 gpm and another quarter had flows that were already less than 2.5 gpm. The single factor most strongly associated with low pre-retrofit showerhead flow rates was low water pressure resulting from on-site domestic wells (see Table 3), which typically operate in a pressure range of 25-45 psi. The showerheads used in this study used less water than their design rating at that pressure, though this may not be true of other showerhead models. Low water pressure may affect showerhead performance. Optimizing showerhead design to perform well under low pressure is a key area for improvement in savings. This may require the design of showerheads specifically for low-pressure applications.

Energy Savings

We estimated savings from the pre- and post-retrofit periods in three different ways. The first approach was based on direct measurement of water heater energy use, which we expected to be the most reliable estimate. The second approach used available end-use data to construct simulated billing data (whole house electricity use, net of electric space heating). The third approach attempted to isolate measure-specific savings by sorting hourly data into periods when showers were reportedly taken. We nick-named this approach the likely shower method.

We conducted the study during a region-wide drought, which spawned water conservation programs during the study period. To account for the possibility of lower water use during the drought, we used a control group made up of REMP sites that did not participate in the study. By using a control group, we were able to estimate net and gross savings.

End-use savings estimates

We estimated savings from the end-use data by subtracting hourly consumption in the 12 post-retrofit months from the 12 months preceding the showerhead retrofits. We adjusted the gross savings estimates, 515 kWh annually, by subtracting the change in use by the control group (drought-induced drop in water consumption) to derive an estimate of net savings. The results varied by showerhead brand because of the different flow rates of the two brands and because of the pre-retrofit flow rates at the respective sites (see Table 4). Gross savings of 515 kWh translated into relative savings of 12.6% of total water heating energy use. The net savings of 363 kWh equalled water heating savings of 8.8%.

Billing data savings estimates

This study benefited from the availability of end-use metered data. End-use data is rarely available, so billing data is often used instead. Bonneville was also interested in the precision of savings estimates derived from billing data, as compared with end-use savings estimates. We reconstructed end-use metering data from each REMP site to remove electric space heating loads. The resulting loads mimic normal household billing records.

A collection of sites for which end-use data and billing data were available was selected, to make a comparison between the two methods for estimating savings. Savings estimates were obtained for each site by measuring the change in energy use after low flow showerheads were installed. Both the billing method and end-use methods were used to determine the savings for each site.

Following this, a comparison was made between the estimated savings associated with both methods. The results of these comparisons indicated that there was considerable difference in the precision between the two types of savings estimates. The difference is directly linked to how hot water energy use is determined before savings are estimated.

End-use data contains the actual amounts of energy used only for hot water. Consequently, one can get an accurate measure of the change in energy use for hot water over time. In contrast, billing data contains the total amount of energy consumed, and the energy used only for hot water must be estimated. Consequently, the billing method cannot assure an accurate measure of the change in energy use for hot water over time.

Since the change in energy use serves as a proxy for energy savings, one would expect billing estimates to vary much more than the end-use estimates. In our comparisons, the billing estimates of savings varied twice as much as the end-use estimates. This has an impact on those who intend to use the billing method for estimating savings.

The billing method creates a greater sense of uncertainty about the actual savings one can anticipate. This is due to the underlying uncertainty associated with the fact that billing data does not provide an accurate measure of hot water energy use. Further, across the board adjustments to billing estimates of savings should not be recommended until an analysis of the billing method and its potential bias is done.

The likely shower method is used to determine when participants are likely to take showers. It reduces the time and expense of end-use data analysis. We used a survey of shower times for each occupant to screen the data for the times of day that showers were likely. We used hot water energy for only those hours in the savings analysis. Savings on a per shower basis averaged 15 watts. The relative savings were 12.6%, the same as the end-use savings estimate of gross savings.

Missing Savings

Prior to this study, Bonneville estimated savings from showerhead retrofits at about 400 kWh, using engineering models and assumptions from other studies (see Table 5). Bonneville's engineering estimates were derived from assumptions about the number of showers per household per day, length of showers, flow-rate reductions, water temperature (or ratio of hot to cold water) and the number of showerheads replaced. Many customer utilities considered Bonneville's initial savings estimates too low. The assumptions used by other organizations often produced estimated savings much higher than Bonneville's. Estimates of 50% reductions in water flows from showerhead retrofits are fairly common.

Similarly, assumptions about shower length often fall between seven and ten minutes. Changing these assumptions in engineering models dramatically affects savings estimates. In fact, the data collected in this study boosted savings estimates from Bonneville's engineering model to more than 1,200 kWh. The end-use metered savings of 515 kWh provided a solid basis for calculating program savings. Nevertheless, these results were far lower than many conservation advocates claim. Net savings for the first year were even lower at 363 kWh. These results raise questions about why savings were so low and how programs should be designed for optimal results.

The missing savings must be traceable to errors in the factors used in the engineering estimates. The factors most suspect are reported shower length, number of showers taken per household, and shower temperature--the ones that are most difficult to verify. It is also probable that post-retrofit showering behavior is different from pre-retrofit behavior in ways that are not obvious to participants. For example, the spray pattern from an energy-efficient showerhead may feel cooler at the same temperature setting, and bathers may turn the temperature up, using more hot water. Similarly, some bathing activities, notably hair washing, require a minimum volume of water. Because of reduced water flow from a showerhead retrofit, a longer shower may be required to complete these activities.

Finally, responses to questions about the number of showers taken are open to interpretation. The likely shower technique, which was based on analyses of individual showers, failed to verify almost two-thirds of the showers claimed by participants in detailed surveys. Some of these showers may have been taken during hours that were outside their schedule or in locations outside the home. Showers taken at school or at the gym cannot show up as savings from showerhead retrofits in the home. This underscores the critical role metered data can play in conservation assessment and program evaluation.

Advice for Successful Showerhead Retrofits

We identified several factors that are critical to a successful showerhead retrofit program:

  • Customer-friendly program design. Participation in this study was easy, profitable, and low risk. Nevertheless, and despite personal, telephone recruitment, participation was only 66%.

  • Professional installation. Professional installers replaced 98% of the showerheads, a rate that was possible because the installers were able to resolve minor plumbing problems in 10% of the cases. Programs that do not rely on well-equipped professional installers can expect maximum replacement rates in the range of 90%, while those that rely on self-installation are likely to achieve much lower replacement rates. Professional installation also helps prevent distribution of showerheads to ineligible participants (alternative water-heating fuels, non-target residences, low-flow sites, and so on).

  • Pre-retrofit flow rates. Showerheads are not uniformly wasteful. Showerheads have become more conserving since the 1970s, resulting in lower flow rates in newer homes. Moreover, the actual flow rate of a showerhead is governed by many factors other than its design including water pressure, scaling, and age of supply piping. Average flows are more likely to be near 4 gpm than 6 gpm.

  • Retrofit showerhead performance. There are scores of energy-efficient showerheads available. Each has different performance characteristics and flows at different rates, depending on water pressure. Savings estimates must be calibrated to the performance of the showerheads used in the program.

  • Low water pressure. We found a correlation between low flows and low water pressure. In most cases, low water pressure was associated with domestic wells. However, chronically low water pressure can occur in older municipal water systems and in systems that are stretching to meet rapid growth. Energy-efficient showerheads may improve shower quality at low pressure sites, but this is not true of all showerhead brands.

  • Persistence and choice. Information on the life of showerheads, particularly new models, is scarce, though estimates of measure life range from 12 to 30 years. Factors thought to affect persistence are satisfaction with the performance of the showerhead, and replacement as a function of bathroom remodeling.


To sum it up, low-flow showerheads save considerable amounts of water and energy, but less than has generally been expected. Lower than expected savings may put a showerhead program close to the margin of cost-effectiveness. Care must be taken to design programs that strike a balance between getting the maximum amount of savings per-site and the per-unit cost of savings.

Programs that use professional installers are more likely to capture the maximum savings per site, but at a higher cost per participant. While showerheads are relatively simple to install and self-installation can reduce per participant costs, barriers to self-installation are likely to limit replacement rates to the 50%-60% range (even lower in multi-family and older mobile homes). Further, it is difficult to eliminate ineligible sites in a self-installation program, and far more showerheads may be distributed than are necessary. A self-installation program that is linked to utility records and that requires monitored return of old showerheads in exchange for an incentive is probably a good compromise. n


1. For more information contact Curtis Hickman, at Bonneville Power Administration, PO Box 3621, Portland, OR 97208-3621. Tel: (503)230-5853.

2. In 1983, Bonneville conducted a residential survey of more than 4,000 residences, called the Pacific Northwest Regional Energy Consumption Survey. A sample of these residences were recruited for inclusion in the End-Use Load and Conservation Assessment Project (ELCAP), a multi-year study Bonneville initiated in 1985. Under ELCAP, end-use monitoring equipment was installed to individually meter a variety of electrical circuits in more than 400 selected buildings. ELCAP's successor is the Regional Energy Metering Program (REMP), which continues data collection from 158 of these homes.




Near-Term Shower Power

As of January 1, 1994, all showerheads and faucets made and sold in the U.S. must be low-flow (2.5 gallons per minute at 80 pounds per square inch of water). The regulation is mandated as part of the National Energy Policy Act of 1992. As a result of the law, it is likely that low-flow shower product choices will greatly expand. Showerheads that wear out or get replaced during bathroom remodels after January 1 will be conserving by default. This shifts the importance of showerhead retrofit programs to near- rather than long-term benefits.


Table 1. Installations (105 sites) Number of sites with this number Total Number Number of showerheads number of of of replaced showerheads showers sites 1 2 3 0 replaced _______________________________________________________________ 1 47 44 N/A N/A 3 44 2 48 3 43 N/A 2 89 3 10 1 0 8 1 25 _______________________________________________________________ Total 105 Sites 158
Table 2. Average Water Flow Results (in gpm) Number of Pre- Post- showerheads* Rating flow flow Change % _____________________________________________________________________ ETL sites 136 2.0 3.09 1.67 1.42 -46 Ondine Sites 22 2.5 3.89 2.55 1.34 -34 Over all 158 N/A 3.21 N/A 1.41 -44 _____________________________________________________________________ * for 98 sites where showerheads were retrofit
Table 3. Effect of Pressure on Flow Rates and Savings (80 ETL Sites Only) Low-pressure Normal Sites sites All ETL sites _________________________________________________________________________ Number of sites 18 62 80 Mean pressure (psi) 37 68 61 Mean post-flow (gpm) 2.3 3.3 3.1 Mean post-flow (2.0 gpm 1.4 1.8 1.7 ETL showerhead) (gpm) Difference (gpm savings) .90 1.6 1.4
Table 4. Comparison of Savings by Showerhead Brand Number Pre- Post- Flow Number Gross Net of shower flow flow change of kWh kWh Brand heads (gpm) (gpm) (gpm) sites savings savings _____________________________________________________________________________ ETL 136 3.09 1.67 1.42 73 511 358 Ondine 22 3.89 2.55 1.34 12 542 389 Overall 158 3.21 85 515 363 _____________________________________________________________________________ Estimated control group savings = 153 kWh
Table 5. Comparison of Engineering Estimates with Field Data Assumption Field data _______________________________________________________________ Showerheads retrofit 100% 90% Shower duration 6.5 min. 7.4 min. Water flow reduction 1 gpm 1.4 gpm Hot/cold water ratio 50% hot 70% hot Showers/person/day .77 .95 Persons/home 2.3 2.8 _______________________________________________________________ Estimated savings 400 kWh 1,225 kWh


Related Articles

Big Flush, The: Saving Water in the Big Apple (Anderson) Graywater: An Option For Household Water Reuse (Bennett) Low-Flow Showerheads, Family Strife, and Cold Feet (Meier) Pulling Utilities Together: Water-Energy Partnerships (Jones, Dyer, and Obst) Remodeling Bathrooms: Let the Energy Savings Flow (Johnston) The Rise of Water Service Companies (Berlin) Savings and Showers: It's All in the Head (Proctor, Gavelis, and Miller) Xeriscape: Winning the Turf War Over Water (Iwata)

| Back to Contents Page | Home Energy Index | About Home Energy | | Home Energy Home Page | Back Issues of Home Energy |


Home Energy can be reached at: Home Energy magazine -- Please read our Copyright Notice


  • 1
  • NEXT
  • LAST
SPONSORED CONTENT What is Home Performance? Learn about the largest association dedicated to home performance and weatherization contractors. Learn more! Watch Video