By Mary Sue Lobenstein

Stop going around in circles about what to recommend for multifamily buildings with attached recirculation loops. Here are two options that offer proven energy savings and a two-year payback!

Most recommendations for reducing energy costs in recirculation systems are fraught with disadvantages. For instance, insulating the loop to reduce heat loss, while a good idea for new construction, is impractical as a retrofit since recirculating lines in existing buildings are usually inaccessible. Reducing supply temperatures is a sensible strategy, but depending on demand during peak periods, reductions are not always possible. Because domestic hot water heater use is subject to strong hourly fluctuations, turning off the recirculation pump during periods of low demand, such as at night, is another common approach. The disadvantage is that tenants do not always have the same schedule. Those who need hot water when the pump is off may face a long wait, which can lead to complaints. Turning pumps on and off can also cause premature failure of the coupling. An alternative solution is to reduce loop temperature during light demand. This reduces loop losses, and keeps hot water available continuously.

Recirculation At Lower Temperatures

The standard control on a commercial tank-type water heater used in a multifamily building is a fixed temperature aquastat, which keeps temperatures inside the heater and recirculation loop constant, generally at the setpoint needed to meet maximum demand. High demand periods though usually only occur once or twice a day, and then for only a few hours at a time. A better idea is a control that adjusts temperatures up to a high setpoint when the demand increases, and down to a low setpoint when demand decreases. Controls that provide such an automatic adjustment range from simple mechanical timers, to complicated electronic controls with internal memory that can "learn" patterns of use, anticipate demand, and adjust the temperature setting accordingly.

This retrofit saves energy three ways:

• It reduces direct heat loss from recirculation piping.

• Less energy is required to satisfy fixed volume uses such as clothes and dish washers which use the same volume of hot water, no matter what the temperature.

• The seasonal efficiency of the water heater (the ratio of useful heat delivered during the entire year, over the heat available in the fuel) improves because of reduced off-cycle jacket and flue losses. (If savings from those sources are high enough, this strategy may even be applicable to systems without return loops.)

The first time I heard about this kind of retrofit, it came from the lips of a salesman. I found his suggestion "interesting," but was skeptical of his estimates that his control would save 20% of domestic water heater energy use. After all, he was talking to someone who had heard claims of "20% savings" for just about every new energy conservation gizmo to hit the market in the last fifteen years. But it did seem like a good idea. So good, that we (in conjunction with the St. Paul Energy Resource Center) decided to test it in three buildings.²

The three test sites were apartment buildings built after 1970 with central gas-fired, zoned hydronic heat, and central water heaters. Two sites had 39 units and 78 tenants, while the third had 47 units and 75 residents. The sites were small for multifamily buildings with recirculation loops, but fairly typical for the area. They were large enough that potential savings warranted the expense of the controls we tested. All three buildings had on-site laundries, and two had dishwashers. Each site had two conventional, commercial, gas-fired, tank-type water heaters plumbed in parallel with connected recirculating loops. The recirculation loops in these buildings were uninsulated. All water heaters had standing pilots, no vent dampers, and had rated inputs of roughly 200,000-250,000 Btu/hr.

We retrofitted the domestic hot water systems in each building with two types of controls that provided automatic temperature adjustment. We used an electronic time-based control made by Goldline, with fixed setup and setback temperatures and times, as well as a seven-day program. We also used the Pro-Temp PT-5000, an electronic demand-based control which actually monitors firing time on the water heater (interpreting increased firing as higher demand and decreased firing as lower) and adjusts system temperatures up or down accordingly.³

Measuring Up The Controls

We rotated regulation of the domestic hot water system among three control strategies for consecutive one-week periods throughout the year-long study.

• During the fixed-setpoint mode, the existing aquastats in each building were set at the temperature at which they were initially found (140deg.F, 134deg.F, and 140deg.F for sites 1, 2 and 3, respectively), and the systems were allowed to operate as before the retrofit.

• During the weeks when the system was regulated by the time control, temperatures on the system were switched between a high setpoint (pre-retrofit temperature) and a low setpoint (115deg.F), depending on a preset program. The setback program was the same for all sites and initially included setbacks: at night (11:00 p.m.-5:00 a.m.), in the late morning (9:30-11:00 a.m.), and during the early afternoon (1:00-4:00 p.m.). These were times when we assumed demand would be lower.

• When operated by the demand control, the temperature of the system was maintained between a user-selected minimum (115deg.F) and maximum (145deg.F), but the exact temperature at any given point depended on actual system demand.

Throughout the year, operation of the hot water system in each building was monitored with a computerized data acquisition system collecting hourly average data including: heater run-time and events, hot water supply, return loop, cold water inlet, boiler room and outside temperatures, and water volume.

Crunching The Data

To develop a model of energy use for each of the three control strategies, we used the data to calculate daily inputs and outputs for each control mode. We then performed a linear regression analysis of daily heat input versus output for each control mode (see Figure 2).

After computing a normalized annual water heater output (in Btu/hr) for each control mode at each building, we plugged it into the regression equation to estimate an annual input (Btu/hr) corresponding to each control strategy. We could estimate savings by comparing estimated inputs among control modes. We used PRInceton Scorekeeping Method (PRISM) to analyze whole building gas use, and compared it among modes to evaluate the impact of the controls on overall energy use. (See "PRISM: A Tool for Tracking Retrofit Savings," HE, Nov/Dec '87, p. 27, and "Now that I've RUN PRISM, What Do I Do with the Results?" HE, Sept/Oct '90, p. 27.)

Seeing Is Believing

We found that the aquastat control required the greatest energy input of the three control options. In addition, in most instances, the time-based control required more input than the demand-based control to meet a given demand. Were these savings significant?

Well, the salesman was in the ballpark! The calculated savings for the time control over the aquastat averaged 10%. And the demand controls saved even more (see Table 1). We assumed that the difference in normalized annual output among control modes was due to reduced demand for fixed volumes of hot water for dishwashers and washing machines in the time- and demand-based modes. Installed costs and paybacks for the three strategies are shown in Table 2.

The savings had a seasonal bias, with the highest savings in the summer months (June, July, and August) and the lowest savings in the winter (November, December, January, February, and March). For the time control, summer savings averaged 14%, compared to 11% in shoulder months and 8% in the winter. Similarly the demand control saved an average of 28% in the summer, 18% in the shoulder, and 10% in the winter months. The reason for the seasonal difference was unclear, but may have been related to increased winter demand as a result of lower inlet water temperatures and higher heat loss from the recirculation loop. In addition, consumption levels and demand peaks are greater in winter, therefore cycling and system losses do not play as important a role in energy consumption, possibly resulting in the lower savings we observed. In any case, it is important to note that short-term monitoring of these types of controls may not yield an accurate picture of yearly savings.

PRISM analysis of whole building gas use showed a trend toward lower consumption in the time-based mode, compared to the aquastat mode, and even lower consumption in the demand-based mode. Since the change in gas use for the retrofits was small compared to overall gas use in the buildings, the differences among modes were not statistically significant. Still, the results corroborated the intensive data we collected. Further, there was no trend toward increased energy use for space heating under either test control. Such an increase might have been expected if recirculation loop losses contributed significantly to useful heat gain in the buildings.

This Is Great, But Does It Do the Dishes?

Since tenant satisfaction was as important to evaluate as how much energy the controls saved, we asked the caretakers to carefully log any problems during the first two months of operation. They recorded no complaints although at Site 3 the caretaker felt that during the time control mode, the hot water temperature was too low for various cleaning duties. As a result, the time control at Site 3 was eventually reprogrammed to set back only at night. Tenant surveys conducted in all three buildings and in a control building revealed no significant pattern of complaints between control modes or buildings. Apparently, the controls did not cause difficulties for the tenants (see box "Permanent Reductions an Option?").

It Doesn't Take A Brain Surgeon To Operate

One doesn't need an advanced degree to operate either of the controls, but the time control was harder to use. The time-based equipment was also complicated to install and required a second visit by the contractor to correct faulty wiring which initially caused the timers to operate in reverse.⁴ Although the manufacturer provided good instructions, we found programming the timers difficult compared to programming standard setback thermostats. Since programming was not "intuitive," keeping instructions available in the boiler room was imperative because the time controls required periodic adjustments for daylight savings time changes, clock inaccuracies, and program losses. In contrast, the demand controls were installed by factory-trained contractors without any glitches. The program itself was preset at the factory, and the settings proved satisfactory. Operators can make adjustments to the demand control on-site, through a series of covered, recessed screws. This requires a separate display unit--an additional $500--to accurately complete the adjustments. Consequently, adjustments to the factory program are typically only made by the contractor at the time of installation. This may seem like a disadvantage, but it is actually an advantage. Our experience is that control settings which are not obvious are less likely to be overridden or tampered with. We think the demand-based control is more foolproof than the time-based control.

Defective controls had to be replaced within the first month of operation. One of the time controls was defective, as was one of the demand controls. These problems were revealed by intensive monitoring. We received no tenant complaints during this period.⁵ In a typical installation of the demand control some monitoring is performed by factory authorized installers. We strongly recommend such a monitoring service when either type of control is installed.

Conclusions

The strategy of lowering water heater tank and recirculation loop temperatures during periods of low demand is worthwhile, even in relatively small multifamily buildings of 35-50 units. The time-based control we tested in three buildings had a mean savings of about 10% of annual water heater energy costs, while a demand-based control we tested in the same three buildings had a mean savings of 16%. Corresponding paybacks were about two years for either control, but given the higher savings potential and ease of operation for the demand control, it is probably the better strategy. Also, the building operator is less likely to override the demand control. Initial malfunctions with each of these controls leads us to recommend that contractors include a monitoring service for a brief period after installation to ensure proper operation.

Endnotes 1. Perlman, M. and N.H. Milligan, 1988. "Hot Water and Energy Use in Apartment Buildings," ASHRAE Transactions 94 (1); DeCicco, J.,1988. "Modeling, Diagnosis and Implications for Improving the Energy-Efficiency of Centrally Heated Apartment Buildings," PU/CEES #225, Center for Energy and Environmental Studies, Princeton University, Princeton, NJ; Sachi, M.W., D.L. Bohac, M.J. Hewett, T.S. Dunsworth, M.W. Hancock, R.W. Landry, 1989. "Comparative Energy Performance of Domestic Hot Water Systems in Multifamily Buildings," TR89-5-MF, Center for Energy and The Urban Environment, Minneapolis, MN.

2. The major source of funding for this research was oil overcharge money provided by the Energy Resource Center of St. Paul, Minn., through a grant from the Legislative Committee on Minnesota Resources (Contract #021140-03535). Additional financial support was supplied by Minnegasco Inc., a Minnesota-based gas utility. A copy of the final report can be obtained by contacting CEUE, 510 1st Ave. N., Minneapolis, MN 55417.

3. The control was selected from products made by one manufacturer that holds the patents on demand-based domestic hot water temperature controls. Because of the small building sizes (and hence a smaller hot water load), the least expensive demand control option, which also had the simplest control strategy, was selected for these tests. It should also be noted that this particular control is most appropriate for tank-type water heaters. The manufacturer recommends more complex demand controllers for set-ups in which water is heated by a boiler or by a separate Burkay-type (A.O. Smith) heater with a storage tank.

4. The timers initially lowered temperatures at times of highest demand and vice versa. It should be noted that this was the first time the particular contractor had installed a control like this. Presumably future installations by the same contractor would not have this specific problem.

5. It's not too surprising that no tenants complained about the defective demand-based control. Although not operating properly, it did provide a constant 130deg.F water temperature, which may have been adequate to prevent complaints. However, it is surprising that no tenants complained in the case of the time control malfunctions, since these controls were providing water at the lowest setpoint temperatures during periods of highest demand.

Figure 1. Schematic of a recirculation loop for a multifamily domestic hot water system.

Figure 2: Model of Gas Use at Site 1

                                                    Time         Demand
                                Aquastat          Control        Control
________________________________________________________________________
SITE 1
Annual output (Btu/hr)				41,100			40,100			38,200
Annual input (Btu/hr)				103,500			93,400			86,700
Annual efficiency %					40%			43%			44%
Savings over aquastat 							  	 10%1			16%1
Savings over timer												7%1 
________________________________________________________________________
SITE 2
Annual output (Btu/hr)				32,200			31,900			27,800
Annual input (Btu/hr) 				 84,800			 78,000			 70,300
Annual efficiency %					38%			41%			40% 
Savings over aquastat								  8%1			17%1
Savings over timer												10%1

________________________________________________________________________
SITE 3
Avg annual output (Btu/hr)			35,000			34,300			32,700
Avg annual input (Btu/hr)			128,6600		112,000			109,000
Annual efficiency 					27%			31%			30%
Savings over aquatstat								13%1			15%1
Savings over timer												3% 
________________________________________________________________________
MEAN
Annual efficiency					35%			38%			38% 
Savings over aquastat								10.3%			16%
Savings over timer												7% 
________________________________________________________________________
1 Highly Statistically Significant (P>0.001)
2 Marginally Statistically Significant (0.05 <P< 0.01)

                               Aquastat          Time-         Demand-
                                                 based          based
                                                control        control
________________________________________________________________________
SITE 1
Cost ($/year)                   $4,535           $4,091        $3,797
Control cost                                     $1,010        $1,400
Savings ($/year)                                   $444          $739
Payback (years)                                     2.3           1.9

________________________________________________________________________
SITE 2                                
Cost ($/year)                   $3,716           $3,415        $3,079
Control cost                                       $900        $1,400
Savings ($/year)                                   $301          $637
Payback (years)                                     3.0           2.2

________________________________________________________________________
SITE 3
Cost ($/year)                   $5,631           $4,904        $4,773
Control cost                                       $910         $1400
Savings ($/year)                                   $727          $858
Payback (years)                                     1.3           1.6

________________________________________________________________________
MEAN
Savings ($/year)                                   $491          $744
Payback (years)                                     2.2           1.9
________________________________________________________________________

Note: Fuel cost 50cents per 100,000 Btu (one therm).

Domestic hot water use profiles generated during operation of the demand-based control indicate that in at least one case (Site 1) the temperature of the water heaters can be permanently reduced (in this case, from 140deg.F to 130deg.F). As a result, some of the savings seen in this study may be achieved if building operators reduce system temperatures. It is reasonable to expect that operators will follow through on such a simple recommendation, but our experience is that in practice, building operators are extremely reluctant to make any changes that may result in tenant complaints. A well-intentioned owner or caretaker may think turning down the domestic hot water tank temperature is a good idea and may even plan to do it, but often he or she never actually gets into the boiler room to follow through. Even if the recommendation is carried out, it is often reversed at the first complaint. Prior to the study, the three building operators in the test sites were all asked if they ever tried turning down the hot water temperatures in their buildings. All said they had, but that the current setpoint was the lowest temperature which satisfied demand and prevented complaints.

Related Articles