PVBC Export-Import calculations results inconsistent with meter board

[GoogleCodeExporter]

What steps will reproduce the problem?
1.
2.
3.

What is the expected output? What do you see instead?

Expected that PVPC calculations of export and import would be consistent with 
meter-board measurements.  They are not.  In trying to analyse why I realised 
that my PVBC records do not include any input from me on export and import.  I 
wonder how the net graphs in PCOutput.org show my daily export and import.

What version of the product are you using (e.g. 1.4.1.9 32bit)?

What Operating system (XP, Vista, Windows 7 - 32bit or 64bit)?

Win 7
Which Database Type (MySQL, MS Access, SQLite - If Access, 32bit or 64bit)? 
SQLite

Are there any unusual messages in PVService.log? (upload a copy to this
issue if there are)

Please upload your settings.xml file with the issue. This answers many
questions I would otherwise need to ask.

Please provide any additional information below.

Original issue reported on code.google.com by jakelly...@bigpond.com on 6 Sep 2011 at 12:47

[GoogleCodeExporter]

Accompanying graph shows good linear relationship between PVBC estimates and 
meter board measurements of exports.

Original comment by jakelly...@bigpond.com on 6 Sep 2011 at 1:14

[GoogleCodeExporter]

Current Cost consumption is generally accurate to within about 10%. My 
experience not any official evaluation. Many factors affect CC readings. 
Official electricity meters are designed to measure actual power consumed or 
generated. They are sensitive to voltage and current direction as well as 
current amplitude. They work effectively with reactive devices such as electric 
motors (especially during startup, stop and speed change). The CC Meters assume 
240 volts (not measured) and they detect current via the electromagnetic field 
the AC current produces. They are not sensitive to current direction. They 
should not be used on cables where the energy flow direction can change.

Reactive loads (generally anything that can use electricity to store energy and 
then release it later) are poorly handled by CC meters. Large electromagnets 
(motors and transformers) use current to create magnetic fields. When the 
voltage drops they create current again as the magnetic field collapses often 
with energy flowing in the opposite direction. If the discharge current returns 
to the grid, power is not consumed, it is simply stored and then handed back. 
This is a reactive load. Resistive loads like incandesant lights and heaters 
are more effectively measured by CC meters.

Anything with a motor that starts and stops can reduce the accuracy of the CC 
readings.

Other issues relate to the way readings are delivered from the CC device. The 
CC device provides power (not energy) at about 6 second intervals. Energy is 
calculated in PVBC by extrapolation across the time period (usually 6 seconds). 

The meter history is delivered as 2 hour energy readings. The accuracy of these 
2 hour readings is unknown. They are not directly comparable with the 6 second 
power readings used for the immediate readings.

CC Meters are a useful guide but no replacement for a proper meter. All of the 
"Cable Clamp" style energy meters suffer from the same issues.

Dennis M-F

Original comment by DennisMackayFisher on 6 Sep 2011 at 12:44

[GoogleCodeExporter]

Dennis,  Thank you for your advice about the accuracy of the Current Cost (CC) 
estimates of power. All good stuff, but I think that is not our problem.

Assuming that the inverted power data are accurate, using the commercial meter 
input and output data I calculate that my  power production (16 Aug - 6 Sep) 
was 242.3 kWh.  The CC data for that period was nearly the same, 240.0 kWh.  
These estimates are surprisingly similar given the uncertainty about the 
accuracy of the CC data.

My concern is about the underlying logic of power measurement and estimates in 
PVBC.  The relationship between power components is Export - Import  = 
Generation - Consumption.  I have Generation (Inverter) and Consumption (CC 
Meter) but no input to PVBC for Export or Import.  Export - Import is known, 
but without knowing export, then import is unknown, and conversely without 
knowing import, then export is unknown.  However PVBC produces estimates for me 
of both export and import.  How can it do that when there are two unknowns in 
the basic equation relating Export, Import, Generation and Consumption?

Original comment by jakelly...@bigpond.com on 6 Sep 2011 at 11:09

[GoogleCodeExporter]

Dennis

I added Comment#3 on-line before reading your email.  I wrote:

"Dennis,  Thank you for your advice about the accuracy of the Current Cost
(CC) estimates of power. All good stuff, but I think that is not our
problem.

Assuming that the inverted power data are accurate, using the commercial
meter input and output data I calculate that my power production (16 Aug - 6
Sep) was 242.3 kWh.  The CC data for that period was nearly the same, 240.0
kWh.  These estimates are surprisingly similar given the uncertainty about
the accuracy of the CC data.

My concern is about the underlying logic of power measurement and estimates
in PVBC.  The relationship between power components is Export - Import  =
Generation - Consumption.  I have Generation (Inverter) and Consumption (CC
Meter) but no input to PVBC for Export or Import.  Export - Import is known,
but without knowing export, then import is unknown, and conversely without
knowing import, then export is unknown.  However PVBC produces estimates for
me of both export and import.  How can it do that when there are two
unknowns in the basic equation relating Export, Import, Generation and
Consumption?"

I think that I could contrive good correspondence between my PVBC and Meter
board data if I added a calculated 'Cook's constant', say k, equally to
PVBC's estimates of Export and Import.  This would presumably have
consequences in PVOutput, including the estimates of FIT and system
pay-back.

With ongoing compliments for your fascinating program,

James Kelly

Original comment by jakelly...@bigpond.com on 6 Sep 2011 at 11:31

[GoogleCodeExporter]

Revision of part of my comment 4...

The suggested improvement to PVBC's estimates of my export would be to multiply 
them by a factor, not add a constant.

I should have written "I think that I could contrive good correspondence 
between my PVBC and Meter board data if I multiplied a calculated 'Cook's 
constant', say k, to
PVBC's estimates of Export.  This would presumably have consequences in 
PVOutput, including the estimates of FIT and system pay-back.

Original comment by jakelly...@bigpond.com on 7 Sep 2011 at 1:55

[GoogleCodeExporter]

The "Calibrate" column on the Appliance list is a multiplier. Try using it. 

To my knowledge you would be the first user of this feature. 

Dennis M-F

Original comment by DennisMackayFisher on 7 Sep 2011 at 10:34

• Changed state: Accepted
• Added labels: Type-Setup
• Removed labels: Type-Defect

[GoogleCodeExporter]

There are two issues, firstly how PVBC gets separate estimates for Export and 
Import, when neither is measured in my system, and secondly the correction of 
CC data.

How does PVBC produce separate Export and Import estimates when the only input 
data  appear to be power Generation and power Consumption?  I see that the 
difference (Generation - Consumption) must be the combined value of (Export - 
Import) but could you explain for me the procedure that PVBC uses to identify 
(separate) each of the two components in (Export - Import)?

The second matter is the adjustment of the CC data that PVBC receives 
automatically.
My CC data relates  well to my meter board data (see the attached file) and it 
looks as if I could use a multiplication factor as a constant to get improved 
estimates of Consumption as measured by the CC device.  

I did as you suggested, and will report back after I get a couple of days data 
with the calibration set to 1.1386, k in the equation Consumption = k * CC 
reading.

Regards

James Kelly

Original comment by jakelly...@bigpond.com on 7 Sep 2011 at 11:43

[GoogleCodeExporter]

Apologies.  I am too careless!

As mentioned in paragralh 3 of comment 4 my CC meter produces Consumption data 
that are closely consistent with my meterboard data.  I reset the calibration 
to 1.000 when I realised that I wrongly used my Export data 'k' to adjust CC 
Consimption data.

I await your help in explaining how PVBC produces separate estimates of Import 
and Export.

Regards

James Kelly

Original comment by jakelly...@bigpond.com on 7 Sep 2011 at 12:16

[GoogleCodeExporter]

PVBC does not do the calculations you mentioned. PVBC uploads energy and power 
values for yield and consumption at 5 or 10 minute intervals. All other 
calculations are performed by pvoutput.

I suggest you browse:

http://forums.whirlpool.net.au/forum-replies.cfm?t=1558447&p=-1&#bottom

I believe you will get an explanation there

Dennis M-F

Original comment by DennisMackayFisher on 7 Sep 2011 at 1:27

[GoogleCodeExporter]

More thanks!

Your Comment 9 link did not lead me direct to an explanation of the balance of 
power estimates, but it did point me to a forum where I was able to read some 
of the PVOutput.org posts on net power and import and export, enough to allow 
me to have a better understanding.

Dennis, I unwittingly hounded you with queries that I did not know should have 
been addressed to PVOutput.org.  Sorry.  This was only obvious after reading 
Comment 9.

Regards

James Kelly 

Original comment by jakelly...@bigpond.com on 8 Sep 2011 at 2:01

[GoogleCodeExporter]

Thanks James

I hope you enjoy PVBC anmd pvoutput. It is addictive!!!

Closing this now

Original comment by DennisMackayFisher on 8 Sep 2011 at 11:53

• Changed state: Done

[GoogleCodeExporter]

Actually I've been using that calibrate multiplier on my current cost data for 
a while! 
It's set to 0.87 and that seems to get close to the meter figures although I 
haven't checked recently!
Turns out my Panasonic microwave is the biggest cause of my 'errors'. The CC 
sees it as a constant 100-150w load when the plug-in meter thingy rates it at 
3w!! Useful thing is, as it's a constant load it works well with the multiplier 
for canceling it out over the whole day!!
Anyhow, no point to this other than to say 'i use it...so please don't drop it 
from the app!'
I've seen useful stuff disappear from apps before due to perceived 'non-use'

Cheers
Pete

Original comment by PeteRown...@gmail.com on 30 Sep 2011 at 12:33

[GoogleCodeExporter]

Peter

Thanks for your input. I am fortunate in that my average daily current cost 
data is nearly identical to my neter data.  The difference id less than one per 
cent. Except that I adopted a multiplier like 0.995 I would get no improvement 
in using it.

I think that my problem is more subtle.  PVOutput.org's calculation of export 
and import result in both being under-stated.  In my case meter measurements of 
export are PVOutput org produce export estimates that are 1.13 times 
PVOutput.orgs's estimates, and import estimates are 1.20 times PVOutput.org's 
estimates.  The relationship is surprisingly constant day to day.

I now think that this difference is generated whenever there is both export and 
import within any five-minute period.  It seems that if there are very 
short-term power compenents totalling A for Export and B for Import, where one 
is postive, the other negative,  then PVOutput.org sees only their sum.  So it 
sees (A-B), and depending on the sign of this expression takes (A-B) as a net 
import or extport during the five minute period when there is both import and 
export.  (A-B) is declared as new A or new B thus understating real values for 
A and B during that five-minute period when there is both import and export, 
and thereby understating both in the daily totals.

I welcome comments on my conjecture from readers who disagree with it or agree 
with it.

Original comment by jakelly...@bigpond.com on 30 Sep 2011 at 9:50