02_03_19 ALPHA adds PVoutput


#22

No, it should have uploaded daily outputs for Nov 25 to Nov 30 given that the data exists in your history file. That’s a pretty simple part of the code, so not sure what could be wrong there, but the direction I’m going will change that anyway.

I agree with that. I’m thinking of aggregating the import/export at 1 minute intervals as a trade-off between resolution and overhead.


#23

My data seems to be correct now as well. I forced a reload of history last night, looked today and looks ok. Previously 8th December was showing no export, but is now showing 6.947kwh exported which matches grafana.

So all good I think.


#24


Just had a look. It doesn’t make sense to me that the Export and Net number are exactly the same for the ones highlighted.

Also for these records Generation + Import does not equal Consumption + Export but it does for the days not highlighted.


#25

Had noticed that but was kind of ignoring it as I’m only looking at the Export and $ figures, will have to look at the net value eventually and see what’s going on.


#26

But is your export figure correct. When I downloaded the 5min data and calculated manually I found that the export number reported here was wrong.

There is a law of physics that says energy in must equal energy out.

For example on 24/11:

Energy in is 24.340 (generation) + 7.382(import) = 31.722
Energy out is 18.338 (consumed) + 6.002 (export) = 24.34

Where did the missing energy go ?


#27

So for the 7dec generated 36.422, Export 22.5

24th generated 24.401 exported 13.679

1st Dec generated 42.275, exported 33.158 import 9.008

So still some figures not correct on Export

These figures are from Grafana/InfluxDB

Just forced a history reload from the iotawatt, won’t take long as I’m in donor mode.


#28

Just looking at the 24/11 your 13.679 export from Grafana/Influx is close to what you would expect so still something wrong with PVoutput export numbers


#29

Yep. Just did the upload and now the 8th has gone back to 0 Export which was my initial problem. So must be a bug somewhere.Export on the 8th was 6.947


#30

What’s interesting is that when you look at the live data, it all looks just fine, but the daily outputs seem to be wrong. These are two separate datasets, and I’m not sure exactly what the relationship is. We know that I am using net-export in the output records and that should be calculated as the sum of the net data from the 288 intervals to be consistent with the live data.

But the update of the live statuses is supposed to call addoutput. It looks to me that what gets added gets conflated with the daily net figures somehow. It takes about two days to get a forum answer, and I’m not even sure what the questions should be. So I’ll be doing some trial and error. If I had a solar system, I could be more productive. Maybe I need to build in a PV production simulator for testing.


#31

Happy to help if I can, if you wanted to supply me via PM an api key and Id for a pvoutput system I could have my iotawatt upload my data to it for a dump of data occasionally.


#32

May have just stumbled onto something??
Above the outputs and below the graph, under the More option, there is a recalculate option, you can select 20 outputs at a time and recalculate. My dollars have just gone into positive, and all the Net values have changed and no longer match the Export values that @Giraffe had pointed out.

Is this the simple fix?


#33

The purpose of ‘recalculate’ is if you change your tariffs you can apply the new tariff to some or all of your historical data. Maybe in doing that calculation it also recalculates the daily figures.

That’s a function is not available for free PVoutput access.


#34

@Giraffe, @mobile0408,

This conversation has been very helpful in fleshing out the essense of PVoutput support. I’ve come to the realization that I over engineered this service. I’m now of the belief that the whole notion of writing daily outputs, either to fill holes or as a reload option, is both unnecessary and laden with unintended consequences with respect to the live data.

So, this morning I ripped out all of that code. The result is an astonishingly simple interface that I think will work properly and be easier for everyone to understand. In short it will do what other services like emoncms and influx do in that it will query the last status posted and pick up from there.

On initial launch of the service, it will upload from the “upload history from” date if there are no statuses already there from another source. If the “Reload History” is checked, it will upload all of the statuses from the “upload history from” date regardless of whether there are pre-existing statuses.

I uploaded 35 days worth of status in 4 minutes this morning (donator mode), and I think it will just about do the whole 90 days allowed with donator mode in less than 15 minutes. Possibly you would hit the limit and finish up after the hourly reset. In freeload mode, You would be able to do only 5 or 6 days per hour, but the limit is 14 days so it would take less than three hours to do the allowed 14 days.

I’ll burn these changes in for a day and then cut another ALPHA release.