fix(gap): mark aquifer boundary cells as FIPLEG 5

docs/gap_roadmap.md

@@ -86,7 +86,7 @@ less often. The technical detail follows for whoever picks the item up.
 | 8 | Give user-facing names a clear meaning | Open |
 | 9 | Represent a localized casing hole | Experimental; static opening/isolation and one-year pressure-driven transfer verified |
 | 10 | Generate hole-transfer reporting from the workbook | Done; verified on CIRRUS (Smeaheia, one year) |
-| 11 | Align FIPLEG aquifer reporting with the current coarse-grid boundaries | Open; region 5 definition needs review |
+| 11 | Align FIPLEG aquifer reporting with the current coarse-grid boundaries | Implemented; CIRRUS INIT check pending |
 | 12 | Run only new or unfinished workbook cases | Open; batch reruns every case and rewrites the manifest |
 | 13 | Clean up the CIRRUS template deck | Partial; generated `BRINE` now matches salt tables; CIRRUS comparison and cleanup pending |
 
@@ -560,10 +560,23 @@ invalid locations, and unchanged cases without reporting.
 
 ### 11. Align FIPLEG aquifer reporting with the current coarse-grid boundaries
 
-**Status: open; outdated definition reported by the user (2026-10-12).**
-FIPLEG 5 was intended to identify aquifer cells with multiplied pore volume.
-The aquifer representation has since moved to the outer edges of the coarse
-grid; its reporting zonation needs to be checked against that current placement.
+**Status: implemented; CIRRUS INIT check pending.** Outdated definition
+reported by the user (2026-10-12): FIPLEG 5 was intended to identify aquifer
+cells with multiplied pore volume. The aquifer representation moved to the
+outer edges of the coarse grid, but region 5 stayed on the bottom
+`aquifer_layers` of the reservoir.
+
+**Decision (2026-10-17):** region 5 is the four outer coarse-grid columns in the
+reservoir layers. Water and overburden edge cells keep regions 1 and 2, even
+though their pore volume is also multiplied, so ocean and overburden regions
+keep their identity. Region 3 is now the whole reservoir minus those edge
+columns, including the bottom `aquifer_layers`, which keep `aquifer_permx` but
+are no longer labelled 5. The edge boxes for `FIPLEG 5` and the `PORV`
+multiplier come from one shared list in `prepare_init_case.py`, so they cannot
+drift apart. The wellbore cells (region 6: pipe, open hole and casing-cement boxes) sit in
+a central parent cell (10, 10 in the default 20 x 20 grid), away from the edge
+columns; other grid sizes were not checked. Cases generated earlier are not
+comparable region-for-region.
 
 **In practice:** regional inventories and transfers labelled as aquifer
 quantities should refer to the actual boundary-aquifer cells, not their former

runscripts/prepare_init_case.py

@@ -142,6 +142,12 @@ def region(keyword: str, value: object, start: int, end: int) -> str:
         formatted_value = f"{value:g}" if isinstance(value, (int, float)) else str(value)
         return f"{keyword} {formatted_value} 1 {nx} 1 {ny} {start} {end} /"
 
+    # Outer coarse-grid columns (i0 i1 j0 j1): the aquifer boundary used for both PORV and FIPLEG 5.
+    edge_columns = ((1, 1, 1, ny), (nx, nx, 1, ny), (2, nx - 1, 1, 1), (2, nx - 1, ny, ny))
+
+    def edge_region(keyword: str, value: object, start: int, end: int) -> list[str]:
+        return [f"{keyword} {value:g} {i0} {i1} {j0} {j1} {start} {end} /" for i0, i1, j0, j1 in edge_columns]
+
     lines = [
         "EQUALS",
         "FIPLEG 4  /",
@@ -163,14 +169,15 @@ def region(keyword: str, value: object, start: int, end: int) -> str:
         "",
         "-- Reservoir and aquifer properties",
     ]
+    lines.append(region("FIPLEG", 3, reservoir_start, reservoir_end))
     if aquifer_start is None:
-        lines.extend([region("PERMX", reservoir_permx, reservoir_start, reservoir_end), region("FIPLEG", 3, reservoir_start, reservoir_end)])
+        lines.append(region("PERMX", reservoir_permx, reservoir_start, reservoir_end))
     else:
         if aquifer_start > reservoir_start:
-            lines.extend(
-                [region("PERMX", reservoir_permx, reservoir_start, aquifer_start - 1), region("FIPLEG", 3, reservoir_start, aquifer_start - 1)]
-            )
-        lines.extend([region("PERMX", aquifer_permx, aquifer_start, reservoir_end), region("FIPLEG", 5, aquifer_start, reservoir_end)])
+            lines.append(region("PERMX", reservoir_permx, reservoir_start, aquifer_start - 1))
+        lines.append(region("PERMX", aquifer_permx, aquifer_start, reservoir_end))
+    lines.append("-- FIPLEG 5: aquifer boundary, the outer coarse-grid columns in reservoir layers")
+    lines.extend(edge_region("FIPLEG", 5, reservoir_start, reservoir_end))
     lines.extend(
         [
             "",
@@ -187,10 +194,7 @@ def region(keyword: str, value: object, start: int, end: int) -> str:
             "",
             "MULTIPLY",
             "-- Multiply pore volume on the four outer coarse-grid sides",
-            f"PORV {porv_multiplier:g} 1 1 1 {ny} 1 {nz} /",
-            f"PORV {porv_multiplier:g} {nx} {nx} 1 {ny} 1 {nz} /",
-            f"PORV {porv_multiplier:g} 2 {nx - 1} 1 1 1 {nz} /",
-            f"PORV {porv_multiplier:g} 2 {nx - 1} {ny} {ny} 1 {nz} /",
+            *edge_region("PORV", porv_multiplier, 1, nz),
             "",
             "-- Do not reduce permeability in the water layer",
             f"PERMZ {permz_multiplier:g} 1 {nx} 1 {ny} {water_end + 1} {nz} /",

src/GaP/README.md

@@ -264,6 +264,14 @@ uv run --extra visualization python runscripts/plot_region_fluxes.py \
 	--output-root cases --zonation FIPLEG --list-vectors
 ```
 
+Generated coarse grids label `FIPLEG` as: 1 water, 2 overburden, 3 reservoir,
+5 aquifer boundary, and 6 the wellbore cells inside the LGR (pipe, open hole
+and casing-cement boxes; the rest of the LGR is not labelled 6). Region 5 is the four outer coarse-grid
+columns in the reservoir layers, the cells whose pore volume is multiplied
+(`porv_multiplier`); the water and overburden edge cells keep 1 and 2. Cases
+generated before this definition (region 5 on the bottom `aquifer_layers`) are
+not comparable region-for-region.
+
 The list reports the exact keys and the cases exporting each one. If `1-6`
 exists but `6-1` does not, only `1-6` is selectable; reversing a pair is not
 inferred. Select a pair with `--vector`, repeat it for multiple panels, or

tests/gap/test_prepare_init_case_from_xlsx.py

@@ -1,3 +1,4 @@
+import importlib.util
 from pathlib import Path
 import subprocess
 import sys
@@ -11,6 +12,12 @@
 from src.WellClass.libs.well_pressure.pressure_table import PressureTable
 
 
+PREPARE_PATH = Path(__file__).parents[2] / "runscripts/prepare_init_case.py"
+PREPARE_SPEC = importlib.util.spec_from_file_location("prepare_init_case", PREPARE_PATH)
+prepare_init_case = importlib.util.module_from_spec(PREPARE_SPEC)
+PREPARE_SPEC.loader.exec_module(prepare_init_case)
+
+
 def _write_minimal_workbook(path: Path, *, salinity: float = 0.032) -> None:
     metadata = pd.DataFrame({"key": ["namespace", "name", "author"], "value": ["screen", "xlsx-test", "pytest"]})
     header = pd.DataFrame(
@@ -127,13 +134,43 @@ def test_prepare_init_case_from_xlsx_stages_files(tmp_path):
     assert "PERMX 10000 1 20 1 20 1 3 /" in grdecl_text
     assert "PERMX 0.001 1 20 1 20 4 18 /" in grdecl_text
     assert "PORO 1 1 20 1 20 1 3 /" in grdecl_text
-    assert "FIPLEG 3 1 20 1 20 19 55 /" in grdecl_text
-    assert "FIPLEG 5 1 20 1 20 56 58 /" in grdecl_text
+    assert "FIPLEG 3 1 20 1 20 19 58 /" in grdecl_text
+    assert "FIPLEG 5 1 1 1 20 19 58 /" in grdecl_text
     assert "PERMX 1000 1 20 1 20 56 58 /" in grdecl_text
     assert "TRANZ 0 1 20 1 20 19 19 /" in grdecl_text
     assert "PERMZ 0.1 1 20 1 20 4 58 /" in grdecl_text
 
 
+def _boxes(text: str, keyword: str) -> list[tuple[str, ...]]:
+    return [tuple(line.split()[2:]) for line in text.splitlines() if line.startswith(f"{keyword} ")]
+
+
+@pytest.mark.parametrize("aquifer_layers", [0, 3])
+def test_fipleg_aquifer_marks_reservoir_edge_columns_with_multiplied_pore_volume(tmp_path, aquifer_layers):
+    root = Path(__file__).parents[2]
+    grdecl = tmp_path / "grid.grdecl"
+    grdecl.write_text((root / "test_data/examples/wildcat-pflotran/include/TEMP_GRD.grdecl").read_text(encoding="utf-8"))
+    prepare_init_case.parameterize_grdecl(
+        grdecl, water_layers=3, overburden_layers=15, reservoir_layers=40,
+        aquifer_layers=aquifer_layers, aquifer_permx=250.0,
+    )
+    text = grdecl.read_text(encoding="utf-8")
+    edges = [("1", "1", "1", "20"), ("20", "20", "1", "20"), ("2", "19", "1", "1"), ("2", "19", "20", "20")]
+    porv = _boxes(text, "PORV")
+    assert [box[:4] for box in porv] == edges
+    assert {box[4:] for box in porv} == {("1", "58", "/")}
+    fipleg5 = [tuple(line.split()[2:]) for line in text.splitlines() if line.startswith("FIPLEG 5 ")]
+    assert [box[:4] for box in fipleg5] == edges
+    assert {box[4:] for box in fipleg5} == {("19", "58", "/")}
+    # Water, overburden and the interior of the reservoir keep their own regions.
+    assert "FIPLEG 1 1 20 1 20 1 3 /" in text
+    assert "FIPLEG 2 1 20 1 20 4 18 /" in text
+    assert "FIPLEG 3 1 20 1 20 19 58 /" in text
+    assert text.index("FIPLEG 3 1 20 1 20 19 58 /") < text.index("FIPLEG 5 1 1 1 20 19 58 /")
+    # Physical aquifer properties are unchanged: the bottom layers keep aquifer_permx.
+    assert ("PERMX 250 1 20 1 20 56 58 /" in text) == bool(aquifer_layers)
+
+
 def test_workbook_keeps_simulation_assumptions_outside_well_model(tmp_path):
     workbook = tmp_path / "well_input.xlsx"
     _write_minimal_workbook(workbook)