def test_stack_on_colname():
""" Test that we can stack column with an implicit double level
in the column names indicated by a separator string"""
dframe = pd.DataFrame(
columns=["EQLNUM", "STATIC", "OWC@2000-01-01", "OWC@2020-01-01"],
data=[[1, 1.2, 2000, 1900], [2, 1.3, 2100, 2050]],
)
stacked = common.stack_on_colnames(dframe)
assert "DATE" in stacked
assert "OWC" in stacked
assert len(stacked.columns) == 4
assert len(stacked["DATE"].unique()) == 2
assert len(stacked) == 4
assert not stacked.isnull().sum().sum()
dframe = pd.DataFrame(
columns=[
"EQLNUM",
"STATIC",
"OWC@2000-01-01",
"OWC@2020-01-01",
"GOC@2000-01-01",
"GOC@2020-01-01",
],
data=[[1, 1.2, 2000, 1900, 1800, 1700],
[2, 1.3, 2100, 2050, 2000, 1950]],
)
stacked = common.stack_on_colnames(dframe)
assert "DATE" in stacked
assert "OWC" in stacked
assert "GOC" in stacked
assert len(stacked.columns) == 5
assert len(stacked["DATE"].unique()) == 2
assert len(stacked) == 4
assert not stacked.isnull().sum().sum()
dframe = pd.DataFrame(columns=["OWC@2000-01-01", "OWC@2020-01-01"],
data=[[2000, 1900], [2100, 2050]])
stacked = common.stack_on_colnames(dframe)
assert "DATE" in stacked
assert "OWC" in stacked
assert len(stacked.columns) == 2
assert len(stacked["DATE"].unique()) == 2
assert len(stacked) == 4
assert not stacked.isnull().sum().sum()
dframe = pd.DataFrame(columns=["EQLNUM", "STATIC"],
data=[[1, 1.2], [2, 1.3]])
stacked = common.stack_on_colnames(dframe)
assert "DATE" not in stacked
assert "OWC" not in stacked
assert "EQLNUM" in stacked
assert "STATIC" in stacked
assert len(stacked.columns) == 2
assert len(stacked) == 2
assert not stacked.isnull().sum().sum()
def df(
eclfiles: EclFiles,
region: str = None,
rstdates: Optional[Union[str, datetime.date, List[datetime.date]]] = None,
soilcutoff: float = 0.2,
sgascutoff: float = 0.7,
swatcutoff: float = 0.7,
stackdates: bool = False,
) -> pd.DataFrame:
"""Produce a dataframe with pillar information
This is the "main" function for Python API users
Produces a dataframe with data for each I-J combination
(in the column PILLAR), and if a region parameter is
supplied, also pr. region.
PORV is the summed porevolume of the pillar (in the region),
VOLUME is bulk volume, and PORO is porevolume weighted porosity
PERM columns contain unweighted value averages, use with caution.
If a restart date is picked, then SWAT and SGAS will
be used to compute volumes pr. phase, WATVOL, OILVOL and GASVOL. The
columns with dynamic data will include the date in the column headers
like SWAT@2009-01-01
Args:
region: A parameter the pillars will be split
on. Typically EQLNUM or FIPNUM. Set to empty string
or None to avoid any region grouping.
rstdates: Dates for which restart data
is to be extracted. The string can
be in ISO-format, or one of the mnenomics
'first', 'last' or 'all'. It can also be a list
of datetime.date.
soilcutoff: If not None, an oil-water contact will
be estimated pr. pillar, based on the deepest cell with
SOIL above the given cutoff. Value is put in column OWC.
sgascuttof: If not None, a gas contact will be
estimated pr pillar, based on the deepest cell with
SGAS above the given cutoff. Value is put in column GOC.
swatcutoff: OWC or GWC is only computed for pillars
where at least one cell is above this value.
stackdates: If true, a column
called DATE will be added and data for all restart
dates will be added in a stacked manner.
"""
# List of vectors we want, conservative in order to save memory and cputime:
vectors = []
if region:
vectors.append(region)
vectors.extend(["POR*", "PERM*", "SWAT", "SGAS", "1OVERBO", "1OVERBG"])
grid_df = grid.df(eclfiles,
rstdates=rstdates,
vectors=vectors,
dateinheaders=True)
rstdates_iso = grid.dates2rstindices(eclfiles, rstdates)[2]
grid_df["PILLAR"] = grid_df["I"].astype(str) + "-" + grid_df["J"].astype(
str)
logger.info("Computing pillar statistics")
groupbies = ["PILLAR"]
if region:
if region not in grid_df:
logger.warning("Region parameter %s not found, ignored", region)
else:
groupbies.append(region)
grid_df[region] = grid_df[region].astype(int)
for datestr in rstdates_iso:
logger.info("Dynamic volumes for %s", datestr)
volumes = compute_volumes(grid_df, datestr=datestr)
grid_df = pd.concat([grid_df, volumes], axis="columns", sort=False)
aggregators = {
key: AGGREGATORS[key.split("@")[0]]
for key in grid_df if key.split("@")[0] in AGGREGATORS
}
# Group over PILLAR and possibly regions:
grouped = (grid_df.groupby(groupbies).agg(aggregators)).reset_index()
# Compute correct pillar averaged porosity (from bulk)
if "PORV" in grouped and "VOLUME" in grouped:
grouped["PORO"] = grouped["PORV"] / grouped["VOLUME"]
# Compute contacts:
for datestr in rstdates_iso:
if "SWAT@" + datestr in grid_df and ("SOIL@" + datestr in grid_df
or "SGAS@" + datestr in grid_df):
contacts = compute_pillar_contacts(
grid_df,
region=region,
soilcutoff=soilcutoff,
sgascutoff=sgascutoff,
swatcutoff=swatcutoff,
datestr=datestr,
)
if not contacts.empty:
grouped = pd.merge(grouped, contacts, how="left")
if stackdates:
return common.stack_on_colnames(grouped,
sep="@",
stackcolname="DATE",
inplace=True)
return grouped
def test_stack_on_colname():
"""Test that we can stack column with an implicit double level
in the column names indicated by a separator string"""
dframe = pd.DataFrame(
columns=["EQLNUM", "STATIC", "OWC@2000-01-01", "OWC@2020-01-01"],
data=[[1, 1.2, 2000, 1900], [2, 1.3, 2100, 2050]],
)
orig_dframe = dframe.copy()
common.stack_on_colnames(dframe, inplace=False)
pd.testing.assert_frame_equal(dframe, orig_dframe)
stacked = common.stack_on_colnames(dframe)
assert not dframe.equals(orig_dframe) # It was modifid in the process
assert "DATE" in stacked
assert "OWC" in stacked
assert len(stacked.columns) == 4
assert len(stacked["DATE"].unique()) == 2
assert len(stacked) == 4
assert not stacked.isnull().sum().sum()
dframe = pd.DataFrame(
columns=[
"EQLNUM",
"STATIC",
"OWC@2000-01-01",
"OWC@2020-01-01",
"GOC@2000-01-01",
"GOC@2020-01-01",
],
data=[[1, 1.2, 2000, 1900, 1800, 1700], [2, 1.3, 2100, 2050, 2000, 1950]],
)
stacked = common.stack_on_colnames(dframe)
assert "DATE" in stacked
assert "OWC" in stacked
assert "GOC" in stacked
assert len(stacked.columns) == 5
assert len(stacked["DATE"].unique()) == 2
assert len(stacked) == 4
assert not stacked.isnull().sum().sum()
dframe = pd.DataFrame(
columns=["OWC@2000-01-01", "OWC@2020-01-01"], data=[[2000, 1900], [2100, 2050]]
)
stacked = common.stack_on_colnames(dframe)
assert "DATE" in stacked
assert "OWC" in stacked
assert len(stacked.columns) == 2
assert len(stacked["DATE"].unique()) == 2
assert len(stacked) == 4
assert not stacked.isnull().sum().sum()
dframe = pd.DataFrame(columns=["EQLNUM", "STATIC"], data=[[1, 1.2], [2, 1.3]])
stacked = common.stack_on_colnames(dframe)
assert "DATE" not in stacked
assert "OWC" not in stacked
assert "EQLNUM" in stacked
assert "STATIC" in stacked
assert len(stacked.columns) == 2
assert len(stacked) == 2
assert not stacked.isnull().sum().sum()
