Solpy is a python library to model solar system power performance similar to PVWatts or NREL's System Advisor Model(SAM). I initially started writing this while working in Bangladesh as fairly crude calculator to go from a fisheye panorama to a csv of vectors for shading calculations, however there were several pieces that were added to make it a bit more useful for both analyis and design.
Daniel Thomas did work adding the Tang evacuated glass tube model. Pyephem was added for solar positioning. There is also a simple module for reading TMY3 data. This tool is rudimentary, but functional.

This is primarily a research and analysis tool and there is no guarantee on the calculations.

-Liu & Jordan diffuse irradiance model
-Perez et al. diffuse irradiance model
-Sandia Inverter model
-NEC voltage drop caculations

-enphase.py - Enphase API wrapper
-epw.py - EPW weather data
-expedite.py - calculate information needed for the expedited permit process
-fisheye.py - fisheye image to shading vectors
-forecast.py - forecast.io API wrapper
-noaa.py - NOAA weather data API wrapper
-pv.py - system performance prediction
-pvcli - cli interface to pv modeling-- -tmy3.py - read tmy3 data
-vd.py - voltage drop

PV systems are descibed with json. For example:

{"system_name":"System Name",
    "zipcode":"17601",
    "tilt":34,
    "azimuth":180,
    "phase":1,
    "voltage":240,
    "array":[
        {"inverter":"SMA America: SB6000US 240V",
        "panel":"Mage Solar : Powertec Plus 250-6 MNCS",
        "series":14,
        "parallel":2}
        ]
    }

If that json was in a file called template.json, the command to model it would be;

pvcli -f template.json

This is the sort of project that lends itself nicely to ipython. Since discovering that project I've tried to make things flow naturally in that enviroment. I really like the inline graphics of the qtconsole.

$ipython qtconsole --colors=Linux --pylab=inline