... __ _ Attachment 5

AMBIENT TEMPERATURE AND ELECTRICAL LOADS IN COMMERCIAL BUILDINGS

ABSTRACT

This paper discusses the impact of ambient temperature on electrical end-use loads in commercial buildings. The results are based on an analysis of two buildings in Seattle for which a full year of data is available. One building is a supermarket, the other an office building. The basic characteristics of the buildings and their loads are summarized in Table 1. Since the sample is too small to be representative, the discussion focuses on basic properties of the temperature dependence which may arise in any building. The major points can be summarized as follows:

1) The relationship of loads to ambient temperature is well determined, typically showing less scatter than in all-electric residences. However, the dependence tends to be highly nonlinear in a variety of ways. Operating schedules, manual control, undersizing of equipment, and multiple modes of operation (heating, cooling, economizer, etc.) all contribute to the nonlinearity. To allow for the nonlinear nature of the dependence, we used a robust method called lowess for fitting a nonlinear smooth curve (Chambers et al., Graphical Methods for Data Analysis, Wadsworth 1983). The figures show daily average data for the year from August 1984 through July 1985, with the smooth fit superimposed. Typically, these curves account for 80 to 95 percent of the variance. These smooth curves can then be used to weather-adjust the end-use loads.

2) The loads which relate to temperature are of a thermal nature, including heating, air conditioning, and refrigeration. Other loads, such as interior lighting and general use, show no relationship with temperature. Some loads, like outdoor lighting controlled by photocell and domestic hot water, show a strong correlation with ambient temperature. However, these loads are really determined by otfier factors: the length of daylight in one case and the mean water temperature in the other case. Both of these variables show an annual sinusoidal variation like that of ambient temperature, but reach their minima at different times of the year.

3) The effects of temperature on loads are best studied on a daily average basis to eliminate the effect of hour dynamics which are controlled largely by unknown schedules, building response time, HVAC system dynamics, etc. The temperature­dependent loads do, in some cases, show a very strong time-of-day variation. This can be studied by making a separate analysis for each hour of the day. There are also strong weekday, weekend, and holiday effects. The shape of the temperature response may be quite different on weekends than weekdays. It should be noted that "weekends" can be arbitrary days of the week; for example, the supermarket is stocked at night every day except Sundays and Thursdays (or sometimes Wednesdays). Weekends, in this case, would be Sundays and Thursdays. Days when the building is cleaned may also show unusual use.

4) Other variables were examined along with temperature. Solar radiation was statistically significant in both cases, but explained very little of the variation. Use of multiple explanatory variables is complicated by the non­linear nature of the temperature effect. Indoor temperatures show a large variation, ranging from less than 60 to more than 80°F during working hours. However, use of indoor temperature as an additional explanatory variable pro­duced only a small improvement in the goodness of fit. Further complications are introduced by multiple types of occupancy (for example, an office building containing a restaurant and small retail store), multiple thermal zones with different control settings, and multiple types of thermal equipment (for example, space heating provided partially by resistance and partially by heat pumps or partially electricity and partially by natural gas).

Table 1 Characteristics of Buildings and Loads

Building ID Use Sguare Footage Annua 1 ConsumQtion % HVAC % Thermal

2403 Grocery 24,800 65.4 kWH/sqft 20.5 65.8* 2404 Office 14,920 20.9 kWH/sqft 50.4

*Includes food refrigeration.

Bldg 2404 TOT Daily Average kW

1 o' . I I ' . 7 .82 j

7 281 l

6 .74 j ~

6 20 j

s.ss 1 5.11 l

1 4.57 i

4 .03 1 l

3.431

2.95 1

..

2 .41 +-1 .....,.-....-.....-~--,,-..-~....,.-r----r---r-2.30 3.15 4.00 4.84 5.69 6.54 7.38 10

1

Daily Average Ambient Temperature !F)

Bldg 2403 TOT Daily Average kW

102

1

j • .

2 .48 .a •• :P. ~ : ~~o---':" _• .;:• \ 11o 0

2 .37 j a B g 0° ' 2 .26 l • ••

::::j ••

1.93 1

1 .81 1 1 .70 . 1 .59 j

1 .48 i i

1 .371-' --.-..--....,-......--,--...,....--r---,....---.---2 .30 3.15 4 .00 4.84 5.69 6 .54 7 .38 10

1

Daily Average Ambient Temperature (F)

Figure 1 TEMPERATURE DEPENDENCE OF TOTAL

ELECTRICAL USE

This figure shows the nonlinear nature of the relationship between weekday use and ambient temperature. The top panel shows an office building w>th package HVAC. The steep downslope on the left is due to space heat while the shallow upslope on the right is due to space coo 1 i ng. The middle portion reflects days where both heating and coo 1 i ng are used. The 1 ower pane 1 shows a supermarket with heat pump and resistance heat. The downslope is due to space heat, while the shallow upslope on the right is due to food refrigeration. The flat top occurs where space heat reaches maximum capacity.

10 1

8 .12

8 .02

7 .91

7 .81

7.71

7.60

7 .50

7 .40

7 .29

Bldg 2404 IT3 Daily Average Temperature (F)

7.19

7.09~....,.......,.-,....-~~--r-----r---r--..---r--r---···- ..... 2 .30 3.15 4 .00 4.84 5 .69 6 .54 7.38 101

Daily Average Ambient Temperature (F)

:::1 7.31

7.12

8 .93

6 .74

8.55

8.36

6 .17

5.79

Bldg 2403 IT1 Daily Average Temperature (F)

5 .98 1

1

.. . 5.60~....,.......,.--,....-~......---,....-~-r--r----r--rl---

2.30 3 .15 4.00 4.84 5 .69 6 .54 7.38101

Daily Average Ambient Temperature (F)

Figure 2 RELATIONSHIP OF INTERIOR TEMPERATURES TO AMBIENT

The top panel shows an interior temperature on weekdays in the office building. The thermostat setting at about 71 °F shows clearly. In the cooling mode the temperature shows a 1 i near trend with ambient due to a deadband and low capacity of the cooling equipment. The lower panel shows an interior temperature on weekdays for the supermarket. The temperature increases on the right because the cooling mode of the heat pumps is rarely used. The bump is due to manual control of the HVAC. The heat is shut off until temperatures are about 60°F, producing the dip. The heat is then turned on resu 1 ti ng in a s 1 i ght increase in temperature. The drop at the far left is due to undersizing cr underuse of the heating equipment.

Bldg 2404 MIX Daily Average kW 10

1

I <

5.67 ~ • <

5.1 3 <

4 .50 1 , 4 061 3.53 '

1 2.99•

1 2.46 <

1 .92 j

1.391

o.8s J

. . •

0 .31 +-1 .....,.-..---.---,--r--"1'1 -.--...--,---,.-.--..,---2.30 3 .15 4.00 4.84 5.69 6.54 7 .J8 101

Deily Aver8ge Ambier>t Temper8ture (f)

Bldg 2403 MIX Daily Average kW

101

4.51

4.04

3.57

3.10

2 .63

2.15

1 .68

1.21

0.74

0 .27 1

-0.20

• ..,i._ ..... .. • q,

2.30 3 .15 4.00 4 .84 5 .69 6 .54

Deily Average Ambient Temperature iF)

Figure 3 RELATIONSHIP OF HVAC LOADS TO

AMBIENT TEMPERATURE

7.38 101

The top panel shows the package HVAC units in the office building. The left slope is due to the electric resistance duct heat, while the right slope is due to the air-conditioned compressors. The increased COP of the cooling mode is reflected in a lower slope. The middle portion with intermediate slope reflects days in which both heating and coo 1 i ng are used. The 1 ower pane 1 shows the heat pumps in the supermarket. The flat top occurs at maximum capacity. The use then falls linearly with temperature, in several discrete steps, until the units are shut off at about 60°F. There is a slight indication of cooling or economizer operation between 60°F and 65°F.

101

9.53

9 .34

9 .15

8.96

8.n

8.58

8.39

8.20

8 .01

7.82

Bldg 2403 REF Daily Average kW

• •

7 .63+-..,.--r---r_,.,.-.,.-...,.....-r---r--,r---r--...,.....~-2.30 3 .15 4 .00 4 .84 5 .69 6 .54 7.38 10 1

Deily Average Ambient Temper8ture (f)

Bldg 2403 HOT Daily Average kW

41:~J ~· -:.· 4.14 •

3.65 I .g ~ ..

3.16 ..

2.67

2 .18

1 .70

1.21

0.72

0.23

I

I

'II.

• og

··"-i _-r~--~~--~~~·~·~---~~~~~:1:1:·:·~ -0 .264-2.30 3 .15 4.00 4 .84 5.69 6.54 7.38 10

1

Daily Average Ambient Temper8ture (f)

Figure 4 RELATIONSHIP OF OTHER THERMAL LOADS

TO AMBIENT TEMPERATURE

The top panel shows the food refrigeration loads in the supermarket. The loads tend to be fairly constant from 31°F to 60°F. Above 60°F the refrigeration use increases linearly with ambient. The refrigeration load actually varies linearly with an inside temperature which is not shown here. The refrigeration load consists of two components, the frozen food and medium temperature cases, which show different temperature dependencies. The lower panel shows the electric resistance heat in the supermarket. The relationship with temperature is 1 i near, and is bounded by maximum capacity on the left and shut off on the right. Compare with the lower panel of Figure 3.

Bldg 2404 MIX Weekdays and Weekends Daily Average kW

101

I

4.96 j

4461

3.97 1

3 .47 1 2.97

2 .48 ;

1 .98

1 .49

0.99 1

0 .50 '

0 .00 +-..,....-,--.,.-..,........,.-.,..-..,....-,--,.--..,....-,--,.---2.30 3.10 3.90 4 .70 5 .50 6.30 7.1010

1

Daily Average Ambient Temperature iF)

Bldg 2404 TOT Weekdays and Weekends Daily Average kW 10

1 . ,I 5 .87 1 5 .22

4 .57

2.61 ::::l 1 96 i

1 .30 l 0 .65

0. 00 +---r---..---,--,.---,--.,.-.,..-..,....-.,.---2 .30 3 .10 3.90 4 .70 5 .50 6.30 7 .10 10

1

Daily Average Ambient Temperature !Fl

Figure 5 TEMPERATURE DEPENDENCE OF USE ON

WEEKDAYS VERSUS WEEKENDS

This figure shows the lowess smooths fit separately for weekdays and weekends. The top panel shows HVAC use in the office building . The lower curve is the weekend curve. In co 1 d weather the space heat use is about the same on weekends as weekdays while in warmer weather the office is coo 1 ed on weekdays but not on weekends. Thus, the temperature dependence may have a different shape on different days of the week. The lower panel shows the total use for the office building on weekdays and weekends. The much lower use of lights and receptables on weekends produces a constant offset during the heating mode. The larger difference on the right is due to the absence of weekend cooling.

2404 MIX Average May Profile

50~--------~------------------~

40

30

20

10

6 9 12 15 18 21 24

Hour of Day

2403 MIX Average May Profile

50r-------------------------------~

40

30

20

10 e. e e ...... ~ ..... a e .......... a a II. IJ ......

0~--~--~--~--~--~--~--~~ 0 3 6 9 12 15 18

Hour of Day

Figure 6 TIME OF DAY DEPENDENCE OF

HVAC LOADS

21 24

This figure shows the average hourly weekday HVAC loads for the month of May. The top panel shows the package HVAC in the office building. The sharp spike in the morning is space heat while the afternoon peak is space coo 1 i ng. The intermediate hours include days with both heating and cooling as well as economizer operation. The night setback or shutoff is evident. The lower panel shows the heat pumps in the supermarket. In contrast to the office, there i s almost no time of day variation in HVAC use. The main reason for the lack of variation is manual control of the units.