Adapted perforation strategy to allow for layering + bug fixes (#296)

* src/flownet/data/perforation_strategy.py * LAYER_ID now recognized in bottom_point, top_point, multiple and multiple_based_on_workovers perforation strategies * Only wells in layer loop * Changed multiple function in perforation_strategy.py to work correctly and added test data to a csv * Bugfix in boundingbox of foreign connections * Updated unit test to also test synthetic well K, which penetrates both layers * Added changelog item Co-authored-by: Wouter J. de Bruin <[email protected]>
equinor · Jan 15, 2021 · 2e44905 · 2e44905
1 parent 5546204
commit 2e44905
Show file tree

Hide file tree

Showing 4 changed files with 117 additions and 293 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -5,6 +5,7 @@ This project adheres to [Semantic Versioning](https://semver.org/).
 ## Unreleased
 
 ### Added
+- [#296](https://github.com/equinor/flownet/pull/296) Adapted perforation strategy to allow for layering + bug fixes in the 'multiple' and 'multiple_based_on_workovers' perforation strategies.
 - [#284](https://github.com/equinor/flownet/pull/284) Added the option to specify cumulative phase rates as observations (WOPT, WWPT, WGPT, WGIT, WWIT)
 
 ### Fixes

diff --git a/src/flownet/data/perforation_strategy.py b/src/flownet/data/perforation_strategy.py
@@ -18,7 +18,9 @@ def bottom_point(df: pd.DataFrame) -> pd.DataFrame:
     """
     df_multiple = multiple(df)
     df_multiple_ever_true = (
-        df_multiple.groupby(["X", "Y", "Z", "WELL_NAME"]).sum().reset_index()
+        df_multiple.groupby(["X", "Y", "Z", "WELL_NAME", "LAYER_ID"])
+        .sum()
+        .reset_index()
     )
     df_first_dates = (
         df[["WELL_NAME", "DATE"]]
@@ -50,7 +52,9 @@ def top_point(df: pd.DataFrame) -> pd.DataFrame:
     """
     df_multiple = multiple(df)
     df_multiple_ever_true = (
-        df_multiple.groupby(["X", "Y", "Z", "WELL_NAME"]).sum().reset_index()
+        df_multiple.groupby(["X", "Y", "Z", "WELL_NAME", "LAYER_ID"])
+        .sum()
+        .reset_index()
     )
     df_first_dates = (
         df[["WELL_NAME", "DATE"]]
@@ -84,19 +88,12 @@ def multiple(df: pd.DataFrame) -> pd.DataFrame:
         DataFrame with all connections
 
     """
-    df = df[["WELL_NAME", "X", "Y", "Z", "DATE", "OPEN"]].sort_values(
-        ["WELL_NAME", "X", "Y", "Z", "DATE"]
+    df = df[["WELL_NAME", "X", "Y", "Z", "DATE", "OPEN", "LAYER_ID"]].sort_values(
+        ["WELL_NAME", "X", "Y", "Z", "DATE", "LAYER_ID"]
     )
+    df["SHIFT"] = df.groupby(["WELL_NAME", "X", "Y", "Z", "LAYER_ID"])["OPEN"].shift(1)
 
-    return df[
-        (df["OPEN"] != df["OPEN"].shift(1))
-        | (
-            (df["X"] != df["X"].shift(1))
-            & (df["Y"] != df["Y"].shift(1))
-            & (df["Z"] != df["Z"].shift(1))
-        )
-        | (df["WELL_NAME"] != df["WELL_NAME"].shift(1))
-    ]
+    return df[(df["OPEN"] != df["SHIFT"])].drop("SHIFT", axis=1)
 
 
 # pylint: disable=too-many-locals
@@ -105,7 +102,7 @@ def multiple_based_on_workovers(df: pd.DataFrame) -> pd.DataFrame:
     This strategy creates multiple connections per well when the well during the historic production period has been
     straddled or plugged (i.e., individual connections have been shut).
 
-    The following steps are performed:
+    The following steps are performed per layer:
 
         1. Split connections into groups of connections per well, based on their open/closing history. That is,
            connections that have seen opening or closure at the same moment in time are considered a group. This is
@@ -128,34 +125,37 @@ def multiple_based_on_workovers(df: pd.DataFrame) -> pd.DataFrame:
     df = multiple(df)
 
     df_w_groups = pd.DataFrame(
-        [], columns=["WELL_NAME", "X", "Y", "Z", "DATE", "OPEN", "GROUPID"]
+        [], columns=["WELL_NAME", "X", "Y", "Z", "DATE", "OPEN", "GROUPID", "LAYER_ID"]
     )
     df_groups = pd.DataFrame([], columns=["X", "Y", "Z", "GROUPID"])
     groupid = 0
 
     # Step 1
-    for well_name in df["WELL_NAME"].unique():
-        df_well = df.loc[df["WELL_NAME"] == well_name][
-            ["X", "Y", "Z", "WELL_NAME", "OPEN", "DATE"]
-        ]
-        df_well_piv = df_well.pivot_table("OPEN", ["X", "Y", "Z", "WELL_NAME"], "DATE")
-        df_well_piv.fillna(method="ffill", axis=1, inplace=True)
-        df_well_piv.fillna(False, inplace=True)
-        df_well_piv = df_well_piv.apply(lambda x: hash(tuple(x)), axis=1)
-
-        for group in df_well_piv.unique():
-            df_group = (
-                df_well_piv.loc[df_well_piv == group]
-                .index.to_frame()
-                .reset_index(drop=True)[["X", "Y", "Z"]]
+    for layer in df["LAYER_ID"].unique():
+        for well_name in df[df["LAYER_ID"] == layer]["WELL_NAME"].unique():
+            df_well = df.loc[
+                (df["WELL_NAME"] == well_name) & (df["LAYER_ID"] == layer)
+            ][["X", "Y", "Z", "WELL_NAME", "LAYER_ID", "OPEN", "DATE"]]
+            df_well_piv = df_well.pivot_table(
+                "OPEN", ["X", "Y", "Z", "WELL_NAME", "LAYER_ID"], "DATE"
             )
-            df_group["GROUPID"] = groupid
-            groupid += 1
-            df_groups = df_groups.append(df_group)
+            df_well_piv.fillna(method="ffill", axis=1, inplace=True)
+            df_well_piv.fillna(False, inplace=True)
+            df_well_piv = df_well_piv.apply(lambda x: hash(tuple(x)), axis=1)
+
+            for group in df_well_piv.unique():
+                df_group = (
+                    df_well_piv.loc[df_well_piv == group]
+                    .index.to_frame()
+                    .reset_index(drop=True)[["X", "Y", "Z"]]
+                )
+                df_group["GROUPID"] = groupid
+                groupid += 1
+                df_groups = df_groups.append(df_group)
 
-        df_w_groups = df_w_groups.append(
-            df_well.merge(df_groups, how="left", on=["X", "Y", "Z"])
-        )
+            df_w_groups = df_w_groups.append(
+                df_well.merge(df_groups, how="left", on=["X", "Y", "Z"])
+            )
 
     # Step 2
     for groupid in df_w_groups["GROUPID"].unique():
@@ -167,8 +167,8 @@ def multiple_based_on_workovers(df: pd.DataFrame) -> pd.DataFrame:
         df_foreign = df_w_groups.loc[
             (
                 ((df_w_groups["X"] > xmin) & (df_w_groups["X"] < xmax))
-                | ((df_w_groups["Y"] > ymin) & (df_w_groups["Y"] < ymax))
-                | ((df_w_groups["Z"] > zmin) & (df_w_groups["Z"] < zmax))
+                & ((df_w_groups["Y"] > ymin) & (df_w_groups["Y"] < ymax))
+                & ((df_w_groups["Z"] > zmin) & (df_w_groups["Z"] < zmax))
             )
             & (df_w_groups["GROUPID"] != groupid)
         ]
@@ -203,7 +203,11 @@ def multiple_based_on_workovers(df: pd.DataFrame) -> pd.DataFrame:
             )
 
     df_w_groups["OPEN"] = df_w_groups["OPEN"].astype(int)
-    result = df_w_groups.groupby(["WELL_NAME", "DATE", "GROUPID"]).mean().reset_index()
+    result = (
+        df_w_groups.groupby(["WELL_NAME", "DATE", "GROUPID", "LAYER_ID"])
+        .mean()
+        .reset_index()
+    )
     result["OPEN"] = result["OPEN"].astype(bool)
     result.drop("GROUPID", axis=1, inplace=True)
 

diff --git a/tests/data/well_perforations_2layers.csv b/tests/data/well_perforations_2layers.csv
@@ -0,0 +1,42 @@
+WELL_NAME,X,Y,Z,DATE,OPEN,LAYER_ID
+A,1.0,1.0,1.0,2021-01-12 09:53:52.832254,True,0
+A,2.0,2.0,2.0,2021-01-12 09:53:52.832254,False,0
+A,1.0,1.0,1.0,2021-01-13 09:53:52.832254,False,0
+A,2.0,2.0,2.0,2021-01-13 09:53:52.832254,True,0
+B,3.0,3.0,3.0,2021-01-12 09:53:52.832254,True,0
+B,4.0,4.0,4.0,2021-01-12 09:53:52.832254,True,0
+B,3.0,3.0,3.0,2021-01-13 09:53:52.832254,False,0
+B,4.0,4.0,4.0,2021-01-13 09:53:52.832254,False,0
+C,5.0,5.0,5.0,2021-01-13 09:53:52.832254,False,0
+C,6.0,6.0,6.0,2021-01-13 09:53:52.832254,True,0
+D,7.0,7.0,7.0,2021-01-11 09:53:52.832254,False,0
+D,7.0,7.0,7.0,2021-01-12 09:53:52.832254,False,0
+D,7.0,7.0,7.0,2021-01-13 09:53:52.832254,True,0
+E,8.0,8.0,8.0,2021-01-11 09:53:52.832254,True,0
+E,8.0,8.0,8.0,2021-01-13 09:53:52.832254,False,0
+F,9.0,9.0,9.0,2021-01-13 09:53:52.832254,False,1
+G,10.0,10.0,10.0,2021-01-11 09:53:52.832254,True,1
+G,11.0,11.0,11.0,2021-01-11 09:53:52.832254,True,1
+G,10.0,10.0,10.0,2021-01-12 09:53:52.832254,True,1
+G,11.0,11.0,11.0,2021-01-12 09:53:52.832254,False,1
+H,12.0,12.0,12.0,2021-01-11 09:53:52.832254,True,1
+H,13.0,13.0,13.0,2021-01-11 09:53:52.832254,False,1
+H,12.0,12.0,12.0,2021-01-12 09:53:52.832254,True,1
+H,13.0,13.0,13.0,2021-01-12 09:53:52.832254,True,1
+I,14.0,14.0,14.0,2021-01-11 09:53:52.832254,True,1
+I,14.0,14.0,14.0,2021-01-12 09:53:52.832254,True,1
+I,14.0,14.0,14.0,2021-01-13 09:53:52.832254,True,1
+J,15.0,15.0,15.0,2021-01-11 09:53:52.832254,True,1
+J,16.0,16.0,16.0,2021-01-11 09:53:52.832254,True,1
+J,17.0,17.0,17.0,2021-01-11 09:53:52.832254,True,1
+J,15.0,15.0,15.0,2021-01-13 09:53:52.832254,True,1
+J,16.0,16.0,16.0,2021-01-13 09:53:52.832254,False,1
+J,17.0,17.0,17.0,2021-01-13 09:53:52.832254,True,1
+K,18.0,18.0,4.0,2021-01-12 09:53:52.832254,True,0
+K,18.0,18.0,4.0,2021-01-13 09:53:52.832254,True,0
+K,18.0,18.0,7.0,2021-01-12 09:53:52.832254,True,0 
+K,18.0,18.0,7.0,2021-01-13 09:53:52.832254,True,0
+K,18.0,18.0,10.0,2021-01-12 09:53:52.832254,True,1
+K,18.0,18.0,10.0,2021-01-13 09:53:52.832254,True,1
+K,18.0,18.0,12.0,2021-01-12 09:53:52.832254,True,1
+K,18.0,18.0,12.0,2021-01-13 09:53:52.832254,True,1