rename series_modules, parallel_modules

[wholmgren]

The PVSystem class and the system_def function accept keyword arguments series_modules and parallel_modules. Until PR #171, pvlib code did not actually use them and they only existed for user extensions. As brought up in #171 by @jforbess, these names are confusing at best.

Now is a good time to change them since they didn't actually do anything before. I would support something like series_modules --> modules_per_string and parallel_modules --> strings.

I'd normally wait for 0.4 to implement a change that could break user code, but in this case I'm leaning towards 0.3.3 release because 1) it's unlikely that anyone is actually using these arguments and 2) #171 is going in 0.3.3.

Thoughts?

[adriesse]

Actually, I don't think the original terms were all that bad. At least there seemed little chance that they would be interpreted incorrectly, which is an important criterion. "modules_per_string" is fine, but I do not think "strings" is quite enough. Maybe "number_of_strings" could be considered, if long names are favoured. Personaly I prefer shorter ones, like "string_length" and "n_strings".

Any of these new alternatives would seem imply one specific wiring scheme for the NxM matrix of modules, whereas the calculations for a homogenous system and conditions apply to all possible wiring schemes.

[jforbess]

I think it's assuming the same number of modules per string behind a single
inverter. You can build up systems from inverter units. It does assume the
same number of modules per string for all strings in an inverter, which is
perhaps a bit less normal these days with multi MPPT handling.

I think parallel_modules is confusing because in my head it should be
parallel_strings, not modules. But I see what you mean.

On Fri, May 20, 2016 at 11:32 PM, Anton Driesse [email protected]
wrote:

Actually, I don't think the original terms were all that bad. At least
there seemed little chance that they would be interpreted incorrectly,
which is an important criterion. "modules_per_string" is fine, but I do not
think "strings" is quite enough. Maybe "number_of_strings" could be
considered, if long names are favoured. Personaly I prefer shorter ones,
like "string_length" and "n_strings".

Any of these new alternatives would seem imply one specific wiring scheme
for the NxM matrix of modules, whereas the calculations for a homogenous
system and conditions apply to all possible wiring schemes.

—
You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHub
#176 (comment)

[cwhanse]

My vote is for

Modules_per_string
Strings_per_inverter

Cliff

[wholmgren]

I like Cliff's suggestions.

[jforbess]

Yes, I think they are good. The only possible future proofing is
strings_per_mppt, (maximum power point tracker). Though the current sandia
inverter model doesn't work that way, I could see it or another model being
extended that way in the future.

Jessica

On Thu, May 26, 2016 at 12:02 PM, Will Holmgren [email protected]
wrote:

I like Cliff's suggestions.

—
You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHub
#176 (comment)

[adriesse]

If we allow the possibiity of not having an inverter or an mppt unit, how about "strings_per_array"?

[jforbess]

But isn't the point to allocate a set amount of kWdc to a given mppt, so
that it can be modeled? I am sure I could be overlooking a system type, so
please feel free to describe what I'm missing.

I was viewing the goal defining these parameters as the finest grain to be
handled uniquely in a model. Multiple systems can be ganged together to
make up one site's worth of PV.

Even if it is not, I think that array is too general, and would mean
different things to different people. An array could contain many inverters
or mppts, right?

On Fri, May 27, 2016 at 2:45 PM, Anton Driesse [email protected]
wrote:

If we allow the possibiity of not having an inverter or an mppt unit, how
about "strings_per_array"?

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#176 (comment),
or mute the thread
https://github.com/notifications/unsubscribe/AH66Ae658e7QTlLbdHF8R94Jqf-HL99vks5qF2YFgaJpZM4IjZx1
.

[adriesse]

Yes, the word "array" is certainly open to interpretation. Feel free to
ignore this suggestion!

[wholmgren]

At the risk of getting off topic... I think that at one point, maybe in #17, there was some consensus on defining a PVSystem as being a set of modules connected to a single inverter. It seemed like a fine place to start. But perhaps part of the reason for the disagreement/discussion here is that the existing PVSystem may not provide the right level of abstraction for many users. We may need a set of PVSystem-like classes to keep things reasonable. Off the top of my head, one related class could be PVPowerPlant, something that collects and aggregates multiple PVSystems. I don't see an obvious good solution for @jforbess's multi-mppt inverters using the existing PVSystem class. Let's make a new PVSystem refactoring issue if anyone wants to discuss this in more detail. Maybe we can also set aside some time at PVSC to discuss these ideas in person.

That being said, I think that our existing PVSystem can model a lot of MWs with a simple modules_per_string and strings_per_inverter (assuming the same number of modules per string for all strings behind a single inverter).

Long story short, I prefer to forge ahead with modules_per_string and strings_per_inverter for now because we may need to make much larger changes in the future. Is that reasonable?

[jforbess]

Yes, I think it's fine for now, considering the current state of the
inverter models. Just pointing out future trends.

On Sun, May 29, 2016 at 12:29 PM, Will Holmgren [email protected]
wrote:

At the risk of getting off topic... I think that at one point, maybe in
#17 #17, there was some
consensus on defining a PVSystem as being a set of modules connected to a
single inverter. It seemed like a fine place to start. But perhaps part of
the reason for the disagreement/discussion here is that the existing
PVSystem may not provide the right level of abstraction for many users. We
may need a set of PVSystem-like classes to keep things reasonable. Off the
top of my head, one related class could be PVPowerPlant, something that
collects and aggregates multiple PVSystems. I don't see an obvious good
solution for @jforbess https://github.com/jforbess's multi-mppt
inverters using the existing PVSystem class. Let's make a new PVSystem
refactoring issue if anyone wants to discuss this in more detail. Maybe we
can also set aside some time at PVSC to discuss these ideas in person.

That being said, I think that our existing PVSystem can model a lot of MWs
with a simple modules_per_string and strings_per_inverter (assuming the
same number of modules per string for all strings behind a single
inverter).

Long story short, I prefer to forge ahead with modules_per_string and
strings_per_inverter for now because we may need to make much larger
changes in the future. Is that reasonable?

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#176 (comment),
or mute the thread
https://github.com/notifications/unsubscribe/AH66AQBIcTP1PvOTDaZI39WiAF9HCpzVks5qGekzgaJpZM4IjZx1
.

[wholmgren]

Closed by #188.