This script is designed to be run periodically and will gather current data on your laptop battery and append it to a file. If scheduled to run every few minutes for an extended period of time, it will build up data about your laptop battery charge and process usage for later analysis.

python battery.py [logfile]

If no logfile is supplied, current battery info is printed to stdout.

This script uses the Windows Management Instrumentation Command-line tool (WMIC) to request system data from Windows. Commands are issued as sub-processes and the results are parsed and organized.

The following fields are requested using the Win32_Battery class:

• BatteryStatus: Information about whether the battery is charged, charging, discharging, etc.
• EstimatedChargeRemaining: Estimated battery charge [%]
• EstimatedRunTime: Estimated battery time remaining based on current usage [min]
• Status: Health of the battery

More information on the WMI Win32_Battery class can be found at:
http://msdn.microsoft.com/en-us/library/aa394074(v=vs.85).aspx

The following fields are requested using the Win32_PerfRawData_PerfProc_Process class:

• Name: Name of each process
• PercentProcessorTime: Despite its name, this field is not the CPU usage of this process. In this class, this field contains the total time that the process has run instructions on the CPU. This time is represented in 100-nanosecond intervals. Unfortunately this is not a great indicator of current CPU usage, just an indicator of total CPU usage since the process was created. This seems to be the best I can get from WMI.

Note: There also exists a Win32_PerfFormattedData_PerfProc_Process class, as well as several others. This class also has a "PercentProcessorTime" field, which actually corresponds to CPU usage, but when tinkering with this the results seemed inconsistent. Processes would often report varied CPU usage when they were in the background the entire time. The data seemed to be dependent on whatever instant you decided to request the data from WMI. I decided the above class was a slightly better option.

More information on the Win32_PerfRawData_PerfProc_Process class can be found at: http://msdn.microsoft.com/en-us/library/aa394323(v=vs.85).aspx

Task scheduling on windows can be accomplished by using Schedule Tasks in the Control Panel, or by using schtasks.exe on the command line.

python analyze.py [logfile] [outfile]

In this graph, the color of the markers represents the age of the data. The oldest data is in green, and the newest data is in red. We can see that, for the same charge level, the average remaining runtime decreases as the age of the battery increases. This data was taken over 1.5 years.

• How does Windows' estimated remaining runtime compare to the actual time it takes to drain the battery?
• Does the estimate's accuracy change as the battery ages ?

For working with more data from Windows, Tim Golden's WMI Python package may be more useful

Inspired by http://www.ifweassume.com/2013/08/the-de-evolution-of-my-laptop-battery.html