def get_CEA_generation(row, col, year, source_url):
"""Extract a generation data point from CEA data."""
try:
if row[col] == u'-':
generation = pw.PlantGenerationObject()
else:
gen_gwh = float(row[col])
generation = pw.PlantGenerationObject.create(gen_gwh,
year,
source=source_url)
except:
generation = pw.PlantGenerationObject()
return generation
print("Loading Form 860-2")
wb860_2 = xlrd.open_workbook(RAW_FILE_NAME_860_2)
ws860_2 = wb860_2.sheet_by_name(TAB_NAME_860_2)
print("Loading Form 860-3")
wb860_3 = xlrd.open_workbook(RAW_FILE_NAME_860_3)
ws860_3 = wb860_3.sheet_by_name(TAB_NAME_860_3)
# read in plants from File 2 of EIA-860
print("Reading in plants...")
plants_dictionary = {}
for row_id in xrange(2, ws860_2.nrows):
rv = ws860_2.row_values(row_id) # row value
name = pw.format_string(rv[COLS_860_2['name']])
idnr = pw.make_id(SAVE_CODE, int(rv[COLS_860_2['idnr']]))
capacity = 0.0
generation = pw.PlantGenerationObject()
owner = pw.format_string(str(rv[COLS_860_2['owner']]))
try:
latitude = float(rv[COLS_860_2['lat']])
except:
latitude = pw.NO_DATA_NUMERIC
try:
longitude = float(rv[COLS_860_2['lng']])
except:
longitude = pw.NO_DATA_NUMERIC
location = pw.LocationObject(u"", latitude, longitude)
new_plant = pw.PowerPlant(idnr, name, plant_country=COUNTRY_NAME,
plant_location=location, plant_coord_source=SOURCE_NAME,
plant_owner=owner, plant_capacity=capacity,
plant_generation=generation,
plant_cap_year=YEAR, plant_source=SOURCE_NAME, plant_source_url=SOURCE_URL)
# first check this isn't an ID overlap
country2 = plants_dictionary[idnr_full].country
if country != country2:
if idnr_full not in overlapping_ids.keys():
overlapping_ids[idnr_full] = {'country1': country, 'country2': country2}
# don't treat this as a unit
continue
# update plant
existing_plant = plants_dictionary[idnr_full]
existing_plant.capacity += capacity
# append generation object - may want to sum generation instead?
if generation:
if not isinstance(existing_plant.generation, list):
existing_plant.generation = [pw.PlantGenerationObject(),]
existing_plant.generation.append(generation)
# if lat/long for this unit, overwrite previous data - may want to change this
if latitude and longitude:
new_location = pw.LocationObject(location, latitude, longitude)
existing_plant.location = new_location
# unclear how to handle owner, source, url, commissioning year
else:
new_location = pw.LocationObject(location, latitude, longitude)
new_plant = pw.PowerPlant(plant_idnr=idnr_full, plant_name=name, plant_country=country,
plant_location=new_location, plant_coord_source=geolocation_source_string,
plant_capacity=capacity,
plant_owner=owner, plant_generation=generation,
plant_source=source, plant_source_url=url,
except:
print(u"-Error: Can't read fuel type for plant {0}.".format(name))
fuel = pw.NO_DATA_SET
try:
latitude = float(row[latitude_col])
longitude = float(row[longitude_col])
except:
latitude, longitude = pw.NO_DATA_NUMERIC, pw.NO_DATA_NUMERIC
try:
location = pw.format_string(row[location_col])
except:
location = pw.NO_DATA_UNICODE
try:
gen_gwh = float(
pw.format_string(row[generation_col].replace(",", "")))
generation = pw.PlantGenerationObject(gen_gwh)
except:
generation = pw.NO_DATA_OTHER
try:
owner = pw.format_string(row[owner_col])
except:
owner = pw.NO_DATA_UNICODE
try:
source = pw.format_string(row[source_col])
if source == u"Open Government Portal": # avoid duplication (can remove after updating FT)
continue
except:
print(u"-Error: Can't read source for plant {0}.".format(name))
source = pw.NO_DATA_UNICODE
try:
url = pw.format_string(row[url_col])
row_grid = pw.format_string(rv[COLS['grid']], None)
if row_grid == u"AISLADO":
continue # don't add islanded generators (not grid-connected)
if row_fuel:
if row_name:
if current_plant_name:
# assign ID number, make PowerPlant object, add to dictionary
idnr = pw.make_id(SAVE_CODE, count)
annual_generation = pw.PlantGenerationObject(
gwh=current_generation_sum,
start_date=gen_start,
end_date=gen_stop,
source=SOURCE_NAME)
new_plant = pw.PowerPlant(plant_idnr=idnr,
plant_name=current_plant_name,
plant_owner=current_owner,
plant_fuel=current_fuel_types,
plant_country=COUNTRY_NAME,
plant_capacity=current_capacity_sum,
plant_cap_year=YEAR_OF_DATA,
plant_source=SOURCE_NAME,
plant_source_url=SOURCE_URL,
plant_generation=annual_generation)
plants_dictionary[idnr] = new_plant
count += 1