Power measurement

[BobM]

Hi,

I've just installed a new GEM for residential energy monitoring and I've already learned a lot. But a couple questions come to mind as I begin to explore further.

I am monitoring the main feeds from the utility and all circuits (22) but one individually including the GEM. I am currently acquiring at 10s intervals and sending data to emoncms (OpenEnergyMonitor) at 30s intervals. I grabbed the monitored values for a single interval from emoncms and dumped them into a spreadsheet and the sum of individual circuits was greater than the main feed value by about 100W. If it was less I'd simply attribute it to the unmonitored circuit. Main feed CTs are installed in opposing directions. All other CTs alternate orientation vertically in the panel on each side. I am wondering if this observation is simply due to sample timing. Presumably there is an AtoD in the GEM somewhere that samples channels serially? Any thoughts?

Second, I was thinking about trying to compare cumulative energy usage between the monitor and what is reported by my utility per month. I see two potential issues. 1) Knowing the exact times when the utility read the meter; and 2) I presume my digital meter does something that could be approximated with integration. I might be able to deal with 2) using numerical methods and 1) I could just pick a time like midnight on the dates the meter was read and assume a reasonable amount of error in the result. Maybe I'm in the weeds here.

Anyway, it's cool and now that I have a data source I think I have a chance to have some fun, learn a lot, and maybe save some $$.

Appreciate any advice/insights

[ben]

I am monitoring the main feeds from the utility and all circuits (22) but one individually including the GEM. I am currently acquiring at 10s intervals and sending data to emoncms (OpenEnergyMonitor) at 30s intervals. I grabbed the monitored values for a single interval from emoncms and dumped them into a spreadsheet and the sum of individual circuits was greater than the main feed value by about 100W. If it was less I'd simply attribute it to the unmonitored circuit. Main feed CTs are installed in opposing directions. All other CTs alternate orientation vertically in the panel on each side. I am wondering if this observation is simply due to sample timing. Presumably there is an AtoD in the GEM somewhere that samples channels serially? Any thoughts?

How many unmonitored circuits and what's the mains vs the sum of all loads? The split CTs have about a 1% error, whereas the micros can be anywhere from 1% to 3%, 100W does seem a bit large though.

Second, I was thinking about trying to compare cumulative energy usage between the monitor and what is reported by my utility per month. I see two potential issues. 1) Knowing the exact times when the utility read the meter; and 2) I presume my digital meter does something that could be approximated with integration. I might be able to deal with 2) using numerical methods and 1) I could just pick a time like midnight on the dates the meter was read and assume a reasonable amount of error in the result. Maybe I'm in the weeds here.

Does it say on your bill when they read?

Ben

[BobM]

I am monitoring the main feeds from the utility and all circuits (22) but one individually including the GEM. I am currently acquiring at 10s intervals and sending data to emoncms (OpenEnergyMonitor) at 30s intervals. I grabbed the monitored values for a single interval from emoncms and dumped them into a spreadsheet and the sum of individual circuits was greater than the main feed value by about 100W. If it was less I'd simply attribute it to the unmonitored circuit. Main feed CTs are installed in opposing directions. All other CTs alternate orientation vertically in the panel on each side. I am wondering if this observation is simply due to sample timing. Presumably there is an AtoD in the GEM somewhere that samples channels serially? Any thoughts?

How many unmonitored circuits and what's the mains vs the sum of all loads? The split CTs have about a 1% error, whereas the micros can be anywhere from 1% to 3%, 100W does seem a bit large though.

One unmonitored circuit - our electric range, 22 of 32 channels used, no pulse, no temp. I just checked again and the mains value was 5914 and the sum of all monitored loads was 6462. I have one 240v circuit w/o a neutral so I subtracted 1 from the range value in setup. It's the A/C circuit which we are not using this time of year

Second, I was thinking about trying to compare cumulative energy usage between the monitor and what is reported by my utility per month. I see two potential issues. 1) Knowing the exact times when the utility read the meter; and 2) I presume my digital meter does something that could be approximated with integration. I might be able to deal with 2) using numerical methods and 1) I could just pick a time like midnight on the dates the meter was read and assume a reasonable amount of error in the result. Maybe I'm in the weeds here.

Does it say on your bill when they read?

Ben

Yup. I have a month and day but not a time. I would use that day and then just be consistent with the time on each end. BTW - it also occurred to me that the values are watt-seconds so I guess integration is not required - just sum up the mains values from btmon assuming it stored differences between consecutive values in the db?

[mwall]

BTW - it also occurred to me that the values are watt-seconds so I guess integration is not required - just sum up the mains values from btmon assuming it stored differences between consecutive values in the db?

btmon stores the watt-second values directly. so just query the db for the record at start of the period and the record at the end of the period, and the total will be the difference.

[BobM]

BTW - it also occurred to me that the values are watt-seconds so I guess integration is not required - just sum up the mains values from btmon assuming it stored differences between consecutive values in the db?

btmon stores the watt-second values directly. so just query the db for the record at start of the period and the record at the end of the period, and the total will be the difference.

Thanks!
I just got the db working with MySQL. A little trouble because I'm currently using a Windows 7 laptop with a 64bit architecture and it looks like the Python MySQL module is 32 bit so I had to go back and re-install 32 bit Python to avoid btmon import problem with MySQLdb. I also seem to have to use 127.0.0.1 for mysql_host - I could not connect using localhost even though I used localhost as the hostname when I configured MySQL. I'm using MySQL Community Server 5.5, Python 2.7.3 and MySQL-python-1.2.4b4.win32-py2.7.exe. Anyway, all is good now.

[ben]

One unmonitored circuit - our electric range, 22 of 32 channels used, no pulse, no temp. I just checked again and the mains value was 5914 and the sum of all monitored loads was 6462. I have one 240v circuit w/o a neutral so I subtracted 1 from the range value in setup. It's the A/C circuit which we are not using this time of year

Can you take a look at every load in real-time, and see if anything looks too high? You could also try flipping on a few light switches, your mains and the associated circuit should increase by the rating of the light bulb.

Ben

[BobM]

One unmonitored circuit - our electric range, 22 of 32 channels used, no pulse, no temp. I just checked again and the mains value was 5914 and the sum of all monitored loads was 6462. I have one 240v circuit w/o a neutral so I subtracted 1 from the range value in setup. It's the A/C circuit which we are not using this time of year

Can you take a look at every load in real-time, and see if anything looks too high? You could also try flipping on a few light switches, your mains and the associated circuit should increase by the rating of the light bulb.

Ben

I am uploading data to OpenEnergyMonitor and am monitoring individual loads as well as the mains. I don't see anything obviously wrong yet although I've discovered some potential areas for savings. I can experiment with a 60W light on different circuits to see if I get the expected change. I have also created a feed from one of my unused GEM channels that subtracts all monitored inputs from the mains input to give me an estimate of my one unmonitored circuit which is either on or off except for some power for always-on displays. I'll send you a plot when I've got a little more data. I just had to go "off-line" for a bit to change my monitoring setup a bit. I'll be posting a new topic on that.