Professional Documents
Culture Documents
To cite this article: Omer Sarfraz & Christian K. Bach (2018) Update to office equipment diversity
and load factors (ASHRAE 1742-RP), Science and Technology for the Built Environment, 24:3,
259-269, DOI: 10.1080/23744731.2017.1365765
The current article gives an overview of methods used for the determination of updated recommended diversity factors for various
office equipment. The update to the diversity factors was required because of the introduction of new types of office equipment, such
as laptop docking stations, and both increased equipment performance as well as improved power management. The present study,
in particular, investigates the effects of the number of tested pieces of equipment as well as the interval duration for data reduction.
Analysis and results on the effect of the different number of tested pieces of equipment and averaging intervals on the diversity
factor are presented. Load factors for various office types are calculated using heat gain and diversity factor values of various office
equipment. These updated office load factor values are included in the 2017 edition of the ASHRAE Fundamentals Handbook are
found to be 10% to 33% lower than those published in the previous ASHRAE Fundamentals Handbook (2013). The article also
presents recommendations for minimum equipment count required to determine diversity factors for various office equipment in a
typical office space. For most tested equipment, 8 to 9 pieces of equipment tested in parallel appears to give acceptable results, for
example, ±6% of the predicted value for diversity factor within the predicted 95% confidence interval.
Introduction by the plug load can be up to 1.5 times the power used by the
load itself. Because plug loads significantly affect the build-
The HVAC community focuses on improving The energy effi- ing’s overall energy consumption, it is important to accurately
ciency of heating, ventilation, and air-conditioning (HVAC) estimate the energy consumed by the plug loads. Over-
systems and on developing better building envelope systems. estimation of the energy consumed by the plug loads can
In the past years, this has substantially decreased heating lead to an oversized HVAC system, which results in increased
and cooling loads, as well as the associated electricity and capital and operational costs of HVAC systems. According to
heating fuel consumption. Plug loads are now one of the Sheppy et al. (2014), oversized HVAC systems can result from
main contributors to overall modern building power con- the over-estimation of the plug loads at the design phase.
sumption (Sheppy et al. 2011). Plug loads are the electrical In the United States, plug loads account for almost 5% of
loads that are plugged directly into an electrical outlet. Plug the total primary energy consumption, or up to 50% of the
loads do not include large appliances, general building light- total electricity consumption of buildings with high-efficiency
ing, heating and ventilation, cooling, or water heating systems HVAC systems (Sheppy et al. 2011). Thomas and Moller
(Moorefield et al. 2008). (2006) found that the energy consumption of two exemplary
Plug loads affect the overall building’s energy consump- office buildings could be reduced by 3% to 4% by properly
tion in two ways: first, by direct electricity consumption; and designing the HVAC system using accurate plug load data.
second, by the increase in HVAC load due to heat addition. The most useful form of data for practitioners and building
According to CIBSE Guide F (CIBSE 1998), the power energy modelers are the load factor, such as watts per square
required by the HVAC system to overcome the heat generated meter (W/m2 or W/ft2 ) for various office spaces (Wilkins
and Hosni 2011). Accurate determination of load factors for
various office spaces requires an accurate estimation of equip-
Received April 22, 2017; accepted July 28, 2017 ment power consumption and diversity factor data. Equip-
Omer Sarfraz, Student Member ASHRAE, is a PhD Student. ment power consumption data based on nameplate values
Christian K. Bach, PhD, Associate Member ASHRAE, is an result in an oversized HVAC system because the nameplate
Assistant Professor. value is based on the maximum working capacity of the
∗ equipment. Hosni et al. (1999) found that for office equip-
Corresponding author e-mail: sarfraz@okstate.edu
Color versions of one or more of the figures in the article can be ment with a nameplate power consumption below 1000 W,
found online at www.tandfonline.com/uhvc. the total heat gain is 25% to 50% of the nameplate value.
260 Science and Technology for the Built Environment
Day Time Equip. 1 Equip. 2 Equip. N Sum averaged using a 15-minute moving average, shown in
Power(t) Power(t) Power(t) Power(t)
Figure 4.
Sun Sum_Sun(t) It can be seen that some of the desktops, LDS, and screens
never switched to idle or off mode during the test duration.
Mon Sum_Mon(t) As a result, the average power consumption of these equip-
ment during nonbusiness hours was 33%–50% of the power
consumption during business hours. It can also be seen that
Sat Sum_Sat(t) the average idle mode power consumption of printers is
Peak _
Peak Power(t) Equip. 1(t)
Peak _ Peak _ negligible (<10 W) if compared to their active mode power
Equip. 2(t) Equip. N(t)
consumption.
Fig. 3. Diversity factor calculation for 15-minute averaging intervals: An example case for equipment with 4-hour hypothetical power
consumption data.
Fig. 4. Power consumed over 1-week period (15-minute averages) for different types of standard office equipment. All the tested
equipment pieces are included in the average. For screens, the sudden decrease in average power between Thursday and Friday is due
to missing power consumption data.
Volume 24, Number 3, March 2018 263
Fig. 5. Diversity factor as a function of averaging interval for different types of standard office equipment (min→ minute).
Fig. 6. Equipment diversity factor over a period of 1 week for different types of standard office equipment. For screens, the sudden
decrease in diversity between Thursday and Friday is due to the missing data.
264 Science and Technology for the Built Environment
Table 1. Diversity factor for common office equipment. For example, if selecting 2 desktops from 12 available
12
Diversity Diversity factor (ASHRAE desktop computers, there are ( ) = 66 possible individual
2
Equipment factor [%] Handbook 2013; %)
combinations. Each of these combinations leads to a different
Desktop PC 75 75 power consumption value, as can be seen in Figure 7. The
Laptop docking station 70 NA average power consumption and diversity factor of combi-
∗
Notebook computer 75 75 nations of two selected desktop computers with maximum
∗∗
Screen 91 60 and minimum peak diversity factors for a 1-week duration
Printer 51 NA are shown in Figures 8a and 8b. Both these combinations
have an average and peak power consumption within ±1%,
NA→ Not available. but have different peak diversity factors. The difference in
∗ Insufficient data from RP-1742, values based on previous data from 2013
the peak diversity factor of these two combinations is due to
ASHRAE Handbook and principal investigator/project monitoring sub-
committee judgment. the different offset of peak power consumption for the two
∗∗ Screen diversity factor is found to be 91%; this is different from the sub- desktop computers in these combinations.
mitted value (70%) for next Handbook (2017) update since screen testing for Increasing the number of selected computers for diversity
diversity factor was performed after the tables’ submission. Since screen aver- factor calculation causes the variation in average diversity fac-
age power consumption is a small value, for example, 21W, using higher value
of diversity factor (i.e., 91%) for load factor determination will not cause any
tor to decrease.
significant change in load factor values for most offices. Table 2 shows the percentage difference of the 95% confi-
dence interval of different number of arbitrarily chosen com-
binations for a selected number of tested desktops to the final
of the equipment is in off mode during the test diversity factor value (i.e., j = 12). It can be seen from Table 2
duration. that the selection of 7 desktops results in a reasonable estima-
tion of the diversity factor value as the maximum percentage
difference of the confidence interval from the final value (with
Desktop computers all the desktops selected) is <6%.
In order to study the effect of the number of selected desk-
top computers on the diversity factor, power consumption LDS
data from 12 desktop computers is used. The number of
possible combinations of tested desktops from the test data To analyze the effect of the number of selected LDS on the
can be determined using the combination formula, such diversity factor, the power consumption data for 11 simulta-
as: neously tested LDS is used.
Figure 9 shows the change in diversity factor for a different
number of tested LDS. It can be seen that the change in the
n n!
= , (2) diversity factor decreases with the increase in the number of
r r! (n − r)!
selected LDS.
Table 3 shows the percentage difference of a 95% confi-
where n = number of tested desktop computers, n = 12, dence interval of different number of arbitrarily chosen com-
and r = number of selected desktops from test data, r = binations for a selected number of tested LDS from the
1, 2, .., 12. final diversity factor value (i.e. j = 11). It can be concluded
from Table 3 that the selection of 8 LDS results in a rea-
sonable estimation of the diversity factor value as the maxi-
mum difference of the confidence interval from the final value
is <6%.
Printers
For printers, the power consumption data was recorded for
1 week for 10 different printers tested in the library (printing
room) of an academic institution. The data were used to ana-
lyze the effect of a different number of selected printers from
test data (j = 10) on the diversity factor.
Figure 10 shows the change in diversity factor for a differ-
ent number of tested printers. It can be seen that the change
in diversity factor decreases with the increase in the number
of printers.
Table 4 shows the percentage difference of a 95% confi-
dence interval of different number of arbitrarily chosen com-
Fig. 7. Effect of number of selected desktop computers on the binations for a selected number of tested printers from the
diversity factor. final diversity factor value (i.e., j = 10). It can be concluded
Volume 24, Number 3, March 2018 265
Fig. 8. Average power consumption and diversity factor for the 2 desktops selected from the 12 available desktop computers.
Fig. 11. The effect of the number of selected screens on the diver-
Fig. 9. The effect of the number of selected laptop docking sta- sity factor.
tions on the diversity factor.
Screens
To analyze the effect of the number of selected screens on
the diversity factor, the power consumption data for 15
screens tested simultaneously is used. Figure 11 shows the
change in diversity factor for a different number of tested
screens.
Table 5 shows the percentage difference of a 95% confi-
dence interval of different number of arbitrarily chosen com-
binations for a selected number of tested screens from the final
diversity factor value (i.e., j = 15). It can be observed that
the selection of only 5 screens results in a reasonable estima-
Fig. 10. The effect of the number of selected printers on the diver- tion of the diversity factor value as the maximum difference
sity factor. of the confidence interval from the final value is <6%. Only
266 Science and Technology for the Built Environment
Table 2. Percentage difference of 95% confidence interval from desktop final diversity factor value.
95% confidence interval (%) Interval % difference from final value
Number of Number of
selected desktops combinations L A U L A U
Table 3. Percentage difference of 95% confidence interval from docking station final diversity factor value.
95% confidence interval (%) Interval % difference from final value
Number of
selected laptop Number of
docking stations combinations L A U L A U
Table 4. Percentage difference of 95% confidence interval from printer final diversity factor value.
95% confidence interval (%) Interval % difference from final value
Number of Number of
selected printers combinations L A U L A U
Table 5. Percentage difference of 95% confidence interval from screen final diversity factor value.
95% confidence interval (%) Interval % difference from final value
Number of Number of
selected screens combinations L A U L A U
100% Laptop docking Light 3.67 NA 15.5 m2 /workstation, all laptop docking station
station use, 1 printer per 10
Medium 4.91 NA 11.6 m2 /workstation, all laptop docking station
use, 1 printer per 10
50% Laptop docking Light 4.75 NA 15.5 m2 /workstation, 50% laptop docking
station station/50% desktop, 1 printer per 10
Medium 6.35 NA 11.6 m2 /workstation, 50% laptop docking
station/50% desktop, 1 printer per 10
100% Desktop Light 5.83 6.46 15.5 m2 /workstation, all desktop use, 1 screen, 1
printer per 10
Medium 7.79 8.61 11.6 m2 /workstation, all desktop use, 1 screen, 1
printer per 10
100% Laptop docking 2 screens 7.44 NA 11.6 m2 /workstation, all laptop docking station
station use, 2 screens, 1 printer per 10
100% Desktop 2 screens 9.06 10.76 11.6 m2 /workstation, all laptop use, 2 screens, 1
printer per 10
3 screens 10.33 NA 11.6 m2 /workstation, all desktop use, 3 screens, 1
printer per 10
100% 2 screens 11.00 16.15 7.9 m2 /workstation, all desktop use, 2 screens, 1
Desktop—heavy printer per 8
3 screens 12.49 NA 7.9 m2 /workstation, all desktop use, 3 screens, 1
printer per 8
100% Laptop docking 2 screens 12.23 NA 7.9 m2 /workstation, all laptop docking station
station—full on use, 2 screens, 1 printer per 8, no diversity
100% Desktop—Full 2 screens 14.35 21.52 7.9 m2 /workstation, all desktop use, 2 screens, 1
on printer per 8, no diversity
3 screens 16.48 NA 7.9 m2 /workstation, all desktop use, 3 screens, 1
printer per 8, no diversity