Study Reveals Surprising Results

By Anna Lynn Spitzer

Irvine, CA, October, 29th, 2014 — A pair of studies conducted by the California Plug Load Research Center at Calit2 and sponsored by the California Energy Commission has yielded some unexpected results.  

The studies show that computers are not entering sleep mode or automatically turning off as often as their users think, and therefore are consuming more energy than previously thought. The biggest impact is in the workplace, where computers are often left on overnight or for other periods when they’re not in use.  

The first study surveyed more than 2,000 members of the UC Irvine community. Faculty, staff, students and retirees answered questions about their behaviors with computer power-management features of all the computers they use regularly, including their office desktops, their home desktops and their laptops []. 

"People think when they leave their workstation, the computer will reduce its own energy consumption after a specified amount of time," said Commissioner Andrew McAllister, the agency's lead on energy efficiency issues, in a CEC press release. "These studies show a strong desire and intent by computer users to reduce energy use. Identified user error and knowledge gaps indicate significant room for improvement in the power management options and interfaces available to computer users."

In essence, computers are not delivering their full energy savings potential even though they have built-in automatic power-management settings and manual power-saving options, including sleep, hibernate and shutdown. Previous studies in offices and homes have shown that computers are used inefficiently, but there has been little agreement on the numbers and little attention paid to differences in users' behaviors towards power management. The UCI study looked more closely at user behavior, includng how users combine taking manual action and relying on automatic settings.

The CEC press release detailed these additional findings:

•    Respondents report using automatic power-management features over manual modes. The survey showed that 39 percent of the time, users regularly use manual controls to put office desktop computers into sleep, hibernate or off modes. Of those office desktops not taking advantage of automatic power management, 61 percent are left on all the time. Interestingly, the respondents were more efficient with their home desktops, and even more energy-conscious with their laptops. “Comparing across the three types of computers for the same users was a big value-added component of the survey,” says Joy Pixley, CalPlug’s co-lead researcher for the project.

•    Users changed power-management settings themselves in 50 percent of laptops, 41 percent of home desktops and 20 percent of office desktops. The respondents who felt more control over their office desktops were more likely to have changed the settings.

•    According to the survey, the two main reasons computers were left on, even when they wouldn't be used for hours, are that users felt restarting is too slow and they believed the computer would automatically go into sleep or other lower-power mode. For office desktops, two other main reasons were that the computer needed to stay on for updates or backups, or needed to stay on for remote access. 

The second study used specialized software to monitor workplace desktops used by 119 respondents of the first survey []. The computers were monitored 24 hours a day for several weeks and actual computer usage patterns were tracked. Findings were compared to those of the first study.

Among the results of the second study, the CEC press release detailed:

•    The monitoring study showed a large difference between direct observation of users’ computer settings and their responses in the previous survey. Researchers observed that 20 percent of computers had automatic power-management enabled, whereas the survey responses indicated that 84 percent of these computers had at least one automatic power setting enabled. Closer analysis of the whole data set suggests users incorrectly believe automatic settings are engaged when they are not.

•    Workplace desktop computers are on 76 percent of the day, even though they were only being used 16 percent of the day. Sleep mode was enabled for only about 7 percent of the day.

•    Overall, workplace desktop computers in the study were on and not being used more than 60 percent of the time. 

•    The majority of computers (69 percent) are off for less than 5 percent of the time, and most of those are off for a few minutes a day, likely when rebooting.

"The considerable amount of energy that is being consumed by computers that are on, but not in use, shows that with better power management alternatives, a large amount of energy could be saved with improved power management features," the CEC’s McAllister said.

Pixley says the survey’s unique approach allowed researchers to glean novel data. “It’s unusual to have survey data from so many people about all their different computers,” she said. “Not only can we compare user behavior for laptops versus desktops, but also how people use desktops differently at home than at work. 

She adds: “Comparing what monitoring study participants reported in the survey to what their power settings actually were gave us a unique opportunity to explore the issue of user confusion about power management.”

Pixley's co-author Stuart Ross noted: "The survey study also considered several other variables that we thought might affect power-management behaviors. For example, we found that how a computer's power levels are managed is affected in part by whether a computer is shared, the user's self-rated knowledge of computers and power management, and his/her role in the university."

The California Plug Load Research Center conducted the two studies to better understand computer use patterns and identify potential ways to reduce energy waste in California. The CEC plans to use these studies to supplement other research as it develops a draft staff proposal for computer energy-efficiency standards.