PVSAT-2: Resultaten/Results

Another conclusion is that apparently there is no loss in power between the electronics in the inverters responsible for the data reading (the inverters are located on top of the building complex, with a 25 meter long AC cable to our appartment), and the PVSAT/WEB'log light data (readings from the end of the AC power cable, before feeding into the net).

Small variations between the two curves in the last graph (rollover image) are most probably the result of:

It is expected that, with fast changing light conditions (strong wind combined with sun and small clouds), the 2 curves will match less well because of the mentioned aberrations. However, if a cloudless sunny day without contrails is logged, it is expected that these 2 curves will match perfectly. A nice compliment for the PVSAT developers, as well as for the OK4 inverter inventor and brilliant software developer Henk Oldenkamp if this prediction will turn out to be true!

---

Quarterly energy yield of several Dutch PV-systems

© March 2005 PVSAT Nieuwsbrief nr. 1/Organisatie voor Duurzame Energie (with kind permission from Corry de Keizer); overlay of Meteocontrol logging data for enhancement data Polder PV (red) by Peter Segaar/Polder PV.

Graph showing quarterly energy yield curves of 18 different PV-systems in the Netherlands obtained from Meteocontrol data extracted from WEB'logs. Since very different sizes of the PV-systems monitored are included in this graph (e.g., 648 Wp for Polder PV, 4400 Wp for the system depicted with a light pink line), energy yields vary considerably; also, systems with different orientations, shading, etc. are included.

---

Visualizing weather and solar flux changes with graphs obtained through PVSAT (November 15 2005)

Collecting the stuff: After the data from the WEB'log light have been automatically uploaded to the German Meteocontrol server, the PV owner is able to download the graphs on a daily basis from that server (visiting the website and logging in with one's private data), which are calculated "on-the-spot". Hence, because it is relatively easy to obtain these average power graphs, one is able to reconstruct the dynamic behaviour of the solar influx through time if one compares the downloaded graphs in a matrix with a sequence of days. Off course, with the caution that one should be able to detect eventual defects in the PV-system. In fact, the latter was one of the goals of the PVSAT-2 project: to be able to decipher PV-system defects or malfunctioning in the mountain of data and within the "normal" fluctuations that occur with the daily and seasonally fluctuating solar energy influx. The last two PVSAT newsletters (available through O.D.E.) provide more details of these interesting fault finding procedures and give nice examples of "typical" aberrations of the normal situation, such as partial shading, snow cover, etc.

Making graphic overviews: To show the normal fluctuations on a daily basis, I used a small trick. I collected all the downloaded graphs on my computer and stored each month in a separate folder. The bitmap (bmp) images generated by the Meteocontrol server are rather large (243 kB) and it is a rather laborious task to make all these graphs smaller to such an extent that a lot of them "fit" on the computer screen. However, when using the browsing option of an image handling program like Paint Shop Pro, one can easily get the same result, with a highly acceptable resolution if one does not put to many day graphs onto the screen. By browsing a "month folder", as I did for October 2005, all the day graphs appear in a single field, or one can make a selection in a separate folder and show the results. By applying the "Alt" and "Print screen" keys at the same time, one is able to make a screendump of the day graph configuration one is interested in, and "dump" the resulting image directly in the same program. After that, it takes one's own phantasy to edit the resulting image. I did this by cutting out "strips" with 4 day graphs each, and importing them into a table in the web page you are reading at this very moment:

© November 2005 Peter J. Segaar/Polder PV Leiden (NL) images generated by the Meteocontrol PVSAT-2 server in Germany
October 5	October 6	October 7	October 8
October 9	October 10	October 11	October 12
October 13	October 14	October 15	October 16
October 17	October 18	October 19	October 20
October 21	October 22	October 23	October 24

^^^
Average power (logging interval fixed at 5 minutes) as measured by the WEB'log light
from the PVSAT-2 project field testing phase, in the day sequence October 5-24, 2005.
Measurements are done on a part of the renovated PV-system of Polder PV
(May 30-31 2005), consisting of 2 108 Wp and 4 93 Wp Shell Solar ACN modules
(total nominal power monitored by the WEB'log: 588 Wp), each fitted with a single
100 Watt OK4E inverter. The inverters have been mounted inside the house during the renovation
commanded by the supplier of the PV-system.

Y-axis max. is 500 Watt (average AC-power per 5 minutes).
Power output on cloudless sunny October days (e.g., Oct. 16 in present example) typically
starts around 8h30 and cumulates at app. 350 Watts average AC power at 13h00 MEZT.
For the 588 Wp monitored that results in an average max. nominal power of 0,6 W/Wp,
which is higher for the 108 Wp panels and lower for the 93 Wp panels.
If there are reflecting clouds around the sun, average AC power at midday can rise to 380 Watts (Oct. 17).
AC power output on cloudless October days grinds to a halt around 18h30 daylight saving time, MEZT.

The most notable weather conditions visualized in this graph: