Archived Forum

I have a set of networked smart meters from EKM Metering in Berkely, CA, and I have my own version of an EMONCMS type software that reads them, logs them and displays the data (you can see it live at http://ecovillageithaca.org/evi_metering).  These meters are capable of reporting both instantaneous power as well as running KWH accumulation.

I am interested in also experimenting with your interesting software (you have much richer functionality for web-based configuration and customization), but I am a bit puzzled by the way you track energy usage and wondered if someone could clarify.

As I understand it, EMONCMS generally uses regular writes of instantaneous power values, and then extrapolates energy usage (KWH) by looking at the difference in time between readings and assuming that power was constant during that time period. True?

It seems to me that this method is a decent approximation for simple meters which can only report instantaneous power, but I would think it would introduce a significant amount of inaccuracy over time when calculating KWH unless you log power readings with a very high frequency.

My question is, does EMONCMS have any way work directly with KWH data from KWH-capable meters instead of extrapolating power readings?

My system works this way; It reads the meters every minute or so and logs the current running accumulated KWH value.  It then stores the difference between KWH readings to indicate energy used in that timespan.  I store and graph this data on an hourly basis (KWH consumed per HOUR), which I think gives the time-based data most useful to my users, and I can easily aggregate it for larger time periods (days, months, years).  

I also have a way for users to pop up a temporary self-updating graph (similar to your "smoothie" visualization) of running instantaneous power.  This can be used to experiment by showing the relative power used by various appliances as things are turned on an off, and looking at vampire loads, but is generally not useful for tracking ongoing usage patterns or trends.

Anyway, I am both curious how much of this functionality I could replicate in EMONCMS as well as what folks think about the relative merits of the two approaches (power-centric vs energy-centric).  I find that the power-centric approach you seem to use is the most common, and several people have recommended it to me, but I have always personally been more interested in tracking energy usage over time, so I tend to optimize for that.

Thoughts?

P.S. You might also find it interesting to look at the special display I have in my software for solar net-metered situations (bring up my website and click on "SW" in the left-hand tree control, for example.  This display combines the data from 2 meters, one tracking solar output and one tracking the connection to the utility, and combines them into one graph and table of historical data.  It extrapolates actual usage by summing the flows of the two meters.  The cool thing is that it can show in real time how much you are using, generating and buying/selling from the utility, and it tracks the percentages over various historical time periods.  For example, it shows that in this "SW" (i.e southwest cluster of 8 homes in my neighborhood) we have generated 59% of our power from our shared solar PV system so far this year. 

This approach can be applied just as easily to an individual home with grid-tied PV, as long as the user has meters that can report independently on PV and utility flows, and might be something worth implementing in EMONCMS.