Inverter strings and array orientation

I checked in today on my dad's setup

He has 13 panels (@415W each = 5.4kW) across 4 roofs
a/ 4x east facing
b/ 4x south facing
c/ 2x east facing
d/ 3x south facing (1 of which gets some shade from a chimney)

What's the optimum way to string these together? I was thinking it would be:
string1: b/ and d/ (all south) [7 panels]
string2: a/ and c/ (all east) [6 panels]

Yet this is the string output chart from the bottom of the "My Inverter" > "PV data" page:

Note how there aren't two distinct distributions... I was expecting if the allocation was above, an earlier 'east' bell curve followed by a later 'south' bell curve. This output graph suggests to me that the installer has mixed and matched east and south roofs i.e.
string 1: a/ and b/ (mix east and south) [8 panels]
string 2: c/ and d/ (mix east and south) [5 panels]

Which would be the optimal configuration please?

A single string with optimisers would be my guess, but startup voltages are going to fun.

I actually have 7 panels on my south facing roof, and 6 on my east roof - 5.46kw array. Obviously, the south roof produces most as it has 7 panels. On your solar pic you can see what each string is doing by clicking on the words string 1 or string 2.
The east roof peaks earlier than the south roof but drops away by 4.30 to virtually nothing , but the south keeps producing till about 7pm now. A couple of months ago I was still getting some sun off the south roof till after 8pm.
I attach a pic of my output on 4th where the curves were pretty smooth with clear unbroken blue skys, although I'm beginning now to get a bit of shading from a neighbours tree on the east roof which will get worse over the next few months, but it only lasts for just over an hour or so. The one point worth mentioning is that this hot weather has really reduced the output.

TimHutchings Thank you - your chart look much more like what I was expecting. And confirms what I thought that the plots are on top of each other, not stacked. Your string1:string2 peaks look more like a 7:6 ratio, rather than an 8:5 too.

26Left I would suggest that is looks about right when there is/was a lot of cloud cover (and Sahara dust). To see the bell curve comparison you really need to pick a day that was relatively cloud free as reflected solar can make quite a difference. e.g. I have E/W orientation and on a recent 'clear' afternoon the east string gave 266w on the East String at 16.30 but yesterday (with a not dissimilar solar pattern to your dad's) at 16:30 it was 642w

SilverArt I'm not convinced as the two strings are exactly the same shape throughout the day - just different amplitude

Definitely looks like string 1 is a+b, string 2 = c+d.

For fun, download the PV data into a spreadsheet, multiply the string 2 values by 8/5 (=1.6) and plot. I expect they'll line up very closely.

26Left I would agree that they follow a similar trend, however I would strongly suggest that they are not "exactly the the same shape", there are a few very noticeable differences and some not quite as obvious.

rjp - thanks - I'll get the data from him and do that

SilverArt - it definitely looks that way to me but I'll do the analysis rjp suggests to check

one question outstanding though - what would be optimal? Would it be better to have all the same orientation on the same string? Or is mixing east and south a good idea?

26Left Mixing would be a bad idea as I understand it.

26Left As I understand it from Gary does Solar’s videos, as the panels are plugged together in series on the string so you want to have all the panels on the same string ‘doing the same thing’ in terms of receiving similar solar. You can mix and match different ‘things’ (orientation, shading, etc) on different strings on the same inverter but keep all the panels on a string the same

You want to have all the east on one string and all south on a separate string, both with their own MPPT.

I clearly misread earlier when I thought you had e/w and s rather than e/s only.