diff --git a/src/Model/Results/Results.py b/src/Model/Results/Results.py
index f44c9c20..dd04dfb6 100644
--- a/src/Model/Results/Results.py
+++ b/src/Model/Results/Results.py
@@ -14,7 +14,9 @@
 # You should have received a copy of the GNU General Public License
 # along with this program.  If not, see <https://www.gnu.org/licenses/>.
 
+import struct
 import logging
+import itertools
 import numpy as np
 
 from copy import deepcopy
@@ -28,12 +30,10 @@ logger = logging.getLogger()
 
 
 class Results(SQLSubModel):
-    _SQL_TABLE = "solver_results"
-
-    def __init__(self, study=None, solver=None,
+    def __init__(self, id=-1, study=None, solver=None,
                  repertory="", name="0"):
         super(Results, self).__init__(
-            status=study.status,
+            id=id, status=study.status,
             owner_scenario=study.status.scenario.id
         )
 
@@ -78,3 +78,122 @@ class Results(SQLSubModel):
             self._repertory,
             qlog=None,
         )
+
+    def timestamps_to_struct(self):
+        ts = self._meta_data["timestamps"]
+        sf = ">" + ''.join(itertools.repeat("d", len(ts)))
+
+        return struct.pack(sf, ts)
+
+    @classmethod
+    def _db_create(cls, execute, ext=""):
+        execute(f"""
+          CREATE TABLE results{ext} (
+            {cls.create_db_add_pamhyr_id()},
+            solver TEXT NOT NULL,
+            study_revision INTEGER NOT NULL,
+            creation_data DATE NOT NULL,
+            nb_timestamps INTEGER NOT NULL,
+            timestamps BLOB NOT NULL,
+            {Scenario.create_db_add_scenario()},
+            {Scenario.create_db_add_scenario_fk()},
+            PRIMARY KEY(pamhyr_id, scenario)
+          )
+        """)
+
+        return cls._create_submodel(execute)
+
+    @classmethod
+    def _db_update(cls, execute, version, data=None):
+        major, minor, release = version.strip().split(".")
+
+        if major == "0" and int(minor) <= 2:
+            cls._db_create(execute)
+
+        return cls._update_submodel(execute, version, data)
+
+    @classmethod
+    def _db_load(cls, execute, data=None):
+        new = []
+
+        study = data['study']
+        status = data['status']
+        scenario = data["scenario"]
+        loaded = data['loaded_pid']
+
+        values = execute(
+            "SELECT pamhyr_id, solver_name, solver_type, " +
+            "study_revision, creation_data, nb_timestamps, timestamps, " +
+            "scenario " +
+            "FROM results " +
+            f"WHERE scenario = {scenario.id} " +
+            f"AND pamhyr_id NOT IN ({', '.join(map(str, loaded))}) " +
+            "ORDER BY ind ASC"
+        )
+
+        for v in values:
+            it = iter(v)
+
+            pid = next(it)
+            solver = next(it)
+            revision = next(it)
+            creation_date = next(it)
+            nb_timestamps = next(it)
+            timestamps_bytes = next(it)
+            owner_scenario = next(it)
+
+            new_results = cls(
+                id=pid, status=status,
+                owner_scenario=owner_scenario
+            )
+            new_results.set("solver_name", solver_name)
+            new_results.set("solver_type", solver_type)
+            new_results.set("study_revision", revision)
+            new_results.set("creation_date", creation_date)
+
+            sf = ">" + ''.join(itertools.repeat("d", len(nb_timestamps)))
+            ts = struct.unpack(sf, timestamp_bytes)
+            new_results.set("timestamps", ts)
+
+            new_results._river = River._db_load(execute, data)
+
+            loaded.add(pid)
+            new.append(new_results)
+
+        return new
+
+    def _db_save(self, execute, data=None):
+        execute(
+            "DELETED FROM results " +
+            f"WHERE scenario = {self._owner_scenario}"
+        )
+        execute(
+            "DELETED FROM results_data " +
+            f"WHERE scenario = {self._owner_scenario}"
+        )
+
+        pid = self._pamhyr_id
+        if self._solver is None:
+            solver_name = self.get("solver_name")
+            solver_type = self.get("solver_type")
+        else:
+            solver_name = self._solver._name
+            solver_type = self._solver._type
+
+        ts = self.get("timestamps")
+        sf = ">" + ''.join(itertools.repeat("d", len(ts)))
+
+        execute(
+            "INSERT INTO " +
+            "results (pamhyr_id, solver_name, solver_type, " +
+            "study_revision, creation_data, nb_timestamps, timestamps, " +
+            "scenario) VALUES (?, ?, ?, ?, ?, ?, ?)",
+            self._pamhyr_id, solver_name, solver_type,
+            self._owner_scenario.revision, self.get("creation_data"),
+            len(ts), struct.pack(sf, ts), self._owner_scenario
+        )
+
+        data["result"] = self._pamhyr_id
+        self._river._db_save(execute, data)
+
+        return True
diff --git a/src/Model/Results/River/River.py b/src/Model/Results/River/River.py
index d068ec1b..7f89a7af 100644
--- a/src/Model/Results/River/River.py
+++ b/src/Model/Results/River/River.py
@@ -14,15 +14,25 @@
 # You should have received a copy of the GNU General Public License
 # along with this program.  If not, see <https://www.gnu.org/licenses/>.
 
+import struct
 import logging
+import itertools
 
+from functools import reduce
 from datetime import datetime
 
+from Model.Tools.PamhyrDB import SQLSubModel
+
 logger = logging.getLogger()
 
 
-class Profile(object):
+class Profile(SQLSubModel):
     def __init__(self, profile, study):
+        super(Profile, self).__init__(
+            id=-1, status=study.status,
+            owner_scenario=study.status.scenario.id
+        )
+
         self._study = study
         self._profile = profile  # Source profile in the study
         self._data = {}  # Dict of dict {<ts>: {<key>: <value>, ...}, ...}
@@ -65,9 +75,122 @@ class Profile(object):
     def has_sediment(self):
         return any(map(lambda ts: "sl" in self._data[ts], self._data))
 
+    @classmethod
+    def _db_create(cls, execute, ext=""):
+        execute(f"""
+          CREATE TABLE results_data{ext}(
+            {cls.create_db_add_pamhyr_id()},
+            result INTEGER NOT NULL,
+            key TEXT NOT NULL,
+            reach INTEGER NOT NULL,
+            section INTEGER NOT NULL,
+            len_data INTEGER NOT NULL,
+            data BLOB NOT NULL,
+            {Scenario.create_db_add_scenario()},
+            {Scenario.create_db_add_scenario_fk()},
+            FOREIGN KEY(result) REFERENCES results(pamhyr_id),
+            FOREIGN KEY(reach) REFERENCES river_reach(pamhyr_id),
+            FOREIGN KEY(section)
+            REFERENCES geometry_profileXYZ(pamhyr_id),
+            PRIMARY KEY(pamhyr_id, result, key, scenario)
+          )
+        """)
 
-class Reach(object):
+        if ext == "_tmp":
+            return True
+
+        return cls._create_submodel(execute)
+
+    @classmethod
+    def _db_update(cls, execute, version, data=None):
+        major, minor, release = version.strip().split(".")
+
+        if major == "0" and int(minor) <= 2:
+            cls._db_create(execute)
+
+        return cls._update_submodel(execute, version, data)
+
+    @classmethod
+    def _db_load(cls, execute, data=None):
+        new = []
+
+        study = data['study']
+        status = data['status']
+        scenario = data["scenario"]
+        loaded = data['loaded_pid']
+        timestamps = data['timestamps']
+
+        values = execute(
+            "SELECT pamhyr_id, result, key, " +
+            "reach, section, len_data, data, scenario " +
+            "FROM results_data " +
+            f"WHERE scenario = {scenario.id} " +
+            f"AND pamhyr_id NOT IN ({', '.join(map(str, loaded))})"
+        )
+
+        for v in values:
+            it = iter(v)
+
+            pid = next(it)
+            result = next(it)
+            key = next(it)
+            reach = next(it)
+            section = next(it)
+            len_data = next(it)
+            data = next(it)
+            owner_scenario = next(it)
+
+            new_data = cls(
+                id=pid, status=status,
+                owner_scenario=owner_scenario
+            )
+
+            sf = ">" + ''.join(itertools.repeat("f", len_data))
+            values = struct.unpack(sf, data)
+
+            for timestamp, value in zip(timestamps, values):
+                new_data.set(timestamp, key, value)
+
+            loaded.add(pid)
+            new.append(new_data)
+
+        return new
+
+    def get_keys(self):
+        return reduce(
+            lambda acc, ts: acc.union(d[ts].keys())
+        )
+
+    def _db_save(self, execute, data=None):
+        pid = self._pamhyr_id
+        result = data["result"]
+
+        for key in self.get_keys():
+            data = self.get_key(key)
+
+            sf = ">" + ''.join(itertools.repeat("f", len(data)))
+            data_bytes = struct.pack(sf, data)
+
+            execute(
+                "INSERT INTO " +
+                "results_data (pamhyr_id, result, , " +
+                "study_revision, key, len_data, data, " +
+                "scenario) VALUES (?, ?, ?, ?, ?, ?, ?)",
+                pid, result, self._owner_scenario.revision,
+                key, len(data), data_bytes,
+                self._owner_scenario
+            )
+
+        return True
+
+
+class Reach(SQLSubModel):
     def __init__(self, reach, study):
+        super(Reach, self).__init__(
+            id=-1, status=study.status,
+            owner_scenario=study.status.scenario.id
+        )
+
         self._study = study
         self._reach = reach  # Source reach in the study
         self._profiles = list(
@@ -101,9 +224,31 @@ class Reach(object):
     def has_sediment(self):
         return any(map(lambda profile: profile.has_sediment(), self._profiles))
 
+    @classmethod
+    def _db_create(cls, execute, ext=""):
+        return cls._create_submodel(execute)
 
-class River(object):
+    @classmethod
+    def _db_update(cls, execute, version, data=None):
+        return cls._update_submodel(execute, version, data)
+
+    @classmethod
+    def _db_load(cls, execute, data=None):
+        return cls._db_load(execute, data)
+
+    def _db_save(self, execute, data=None):
+        for profile in self._profiles:
+            data["profile"] = profile.geometry.pamhyr_id
+            profile._db_save(execute, data)
+
+
+class River(SQLSubModel):
     def __init__(self, study):
+        super(River, self).__init__(
+            id=-1, status=study.status,
+            owner_scenario=study.status.scenario.id
+        )
+
         self._study = study
 
         # Dict with timestamps as key
@@ -137,3 +282,20 @@ class River(object):
                 self._reachs
             )
         )
+
+    @classmethod
+    def _db_create(cls, execute, ext=""):
+        return cls._create_submodel(execute)
+
+    @classmethod
+    def _db_update(cls, execute, version, data=None):
+        return cls._update_submodel(execute, version, data)
+
+    @classmethod
+    def _db_load(cls, execute, data=None):
+        return cls._db_load(execute, data)
+
+    def _db_save(self, execute, data=None):
+        for reach in self._reachs:
+            data["reach"] = reach.geometry.pamhyr_id
+            reach._db_save(execute, data)