diff --git a/src/power_grid_model/_utils.py b/src/power_grid_model/_utils.py
index 23e230d4b..7f79eebe2 100644
--- a/src/power_grid_model/_utils.py
+++ b/src/power_grid_model/_utils.py
@@ -539,9 +539,28 @@ def is_columnar(component_data: ComponentData) -> bool:
     return not isinstance(component_data, np.ndarray)
 
 
-def is_nan_or_equivalent(array):
+def is_nan_or_default(x: np.ndarray) -> np.ndarray:
+    """
+    Check if elements in the array are NaN or equal to the min of its dtype.
+
+    Args:
+        x: A NumPy array to check.
+
+    Returns:
+        A boolean NumPy array where each element is True if the corresponding element in x is NaN
+        or min of its dtype, and False otherwise.
+    """
+    if x.dtype == np.float64:
+        return np.isnan(x)
+    if x.dtype in (np.int32, np.int8):
+        return x == np.iinfo(x.dtype).min
+    raise TypeError(f"Unsupported data type: {x.dtype}")
+
+
+def is_nan_or_equivalent(array) -> bool:
     """
     Check if the array contains only nan values or equivalent nan values for specific data types.
+    This is the aggregrated version of `is_nan_or_default` for the whole array.
 
     Args:
         array: The array to check.
@@ -549,7 +568,7 @@ def is_nan_or_equivalent(array):
     Returns:
         bool: True if the array contains only nan or equivalent nan values, False otherwise.
     """
-    return isinstance(array, np.ndarray) and (
+    return isinstance(array, np.ndarray) and bool(
         (array.dtype == np.float64 and np.isnan(array).all())
         or (array.dtype in (np.int32, np.int8) and np.all(array == np.iinfo(array.dtype).min))
     )
@@ -749,3 +768,19 @@ def get_dataset_type(data: Dataset) -> DatasetType:
         raise ValueError("The dataset type could not be deduced because multiple dataset types match the data.")
 
     return next(iter(candidates))
+
+
+def get_comp_size(comp_data: SingleColumnarData | SingleArray) -> int:
+    """
+    Get the number of elements in the comp_data of a single dataset.
+
+    Args:
+        comp_data: Columnar or row based data of a single batch
+
+    Returns:
+        Number of elements in the component
+    """
+    if not is_columnar(comp_data):
+        return len(comp_data)
+    comp_data = cast(SingleColumnarData, comp_data)
+    return len(next(iter(comp_data.values())))
diff --git a/src/power_grid_model/validation/errors.py b/src/power_grid_model/validation/errors.py
index 58bc84649..48e1fb6d9 100644
--- a/src/power_grid_model/validation/errors.py
+++ b/src/power_grid_model/validation/errors.py
@@ -325,12 +325,14 @@ def __init__(  # pylint: disable=too-many-arguments
         self,
         component: ComponentType,
         field: str,
-        ids: list[int],
-        ref_components: ComponentType | list[ComponentType],
+        ids: Optional[list[int]] = None,
+        ref_components: Optional[ComponentType | list[ComponentType]] = None,
         filters: Optional[dict[str, Any]] = None,
     ):
         # pylint: disable=too-many-positional-arguments
-        super().__init__(component=component, field=field, ids=ids)
+        self.ids = ids if ids is not None else []
+        ref_components = ref_components if ref_components is not None else []
+        super().__init__(component=component, field=field, ids=self.ids)
         self.ref_components = [ref_components] if isinstance(ref_components, (str, ComponentType)) else ref_components
         self.filters = filters if filters else None
 
diff --git a/src/power_grid_model/validation/rules.py b/src/power_grid_model/validation/rules.py
index fdb4df991..7046fb9f6 100644
--- a/src/power_grid_model/validation/rules.py
+++ b/src/power_grid_model/validation/rules.py
@@ -40,6 +40,7 @@
 import numpy as np
 
 from power_grid_model import ComponentType
+from power_grid_model._utils import get_comp_size, is_nan_or_default
 from power_grid_model.data_types import SingleDataset
 from power_grid_model.enum import FaultPhase, FaultType, WindingType
 from power_grid_model.validation.errors import (
@@ -678,26 +679,39 @@ def all_not_two_values_equal(
     return []
 
 
-def all_ids_exist_in_data_set(
-    data: SingleDataset, ref_data: SingleDataset, component: ComponentType, ref_name: str
-) -> list[IdNotInDatasetError]:
+def ids_valid_in_update_data_set(
+    update_data: SingleDataset, ref_data: SingleDataset, component: ComponentType, ref_name: str
+) -> list[IdNotInDatasetError | InvalidIdError]:
     """
-    Check that for all records of a particular type of component, the ids exist in the reference data set.
+    Check that for all records of a particular type of component, whether the ids:
+    - exist and match those in the reference data set
+    - are not present but qualifies for optional id
 
     Args:
-        data: The (update) data set for all components
+        update_data: The update data set for all components
         ref_data: The reference (input) data set for all components
         component: The component of interest
-        ref_name: The name of the reference data set, e.g. 'input_data'
+        ref_name: The name of the reference data set, e.g. 'update_data'
     Returns:
         A list containing zero or one IdNotInDatasetError, listing all ids of the objects in the data set which do not
         exist in the reference data set.
     """
-    component_data = data[component]
+    component_data = update_data[component]
     component_ref_data = ref_data[component]
-    if not isinstance(component_data, np.ndarray) or not isinstance(component_ref_data, np.ndarray):
-        raise NotImplementedError()  # TODO(mgovers): add support for columnar data
-
+    if component_ref_data["id"].size == 0:
+        return [InvalidIdError(component=component, field="id", ids=None)]
+    id_field_is_nan = np.array(is_nan_or_default(component_data["id"]))
+    # check whether id qualify for optional
+    if component_data["id"].size == 0 or np.all(id_field_is_nan):
+        # check if the dimension of the component_data is the same as the component_ref_data
+        if get_comp_size(component_data) != get_comp_size(component_ref_data):
+            return [InvalidIdError(component=component, field="id", ids=None)]
+        return []  # supported optional id
+
+    if np.all(id_field_is_nan) and not np.all(~id_field_is_nan):
+        return [InvalidIdError(component=component, field="id", ids=None)]
+
+    # normal check: exist and match with input
     invalid = np.isin(component_data["id"], component_ref_data["id"], invert=True)
     if invalid.any():
         ids = component_data["id"][invalid].flatten().tolist()
diff --git a/src/power_grid_model/validation/validation.py b/src/power_grid_model/validation/validation.py
index dbbbd7ad6..24ba03fd9 100644
--- a/src/power_grid_model/validation/validation.py
+++ b/src/power_grid_model/validation/validation.py
@@ -33,6 +33,7 @@
 )
 from power_grid_model.validation.errors import (
     IdNotInDatasetError,
+    InvalidIdError,
     MissingValueError,
     MultiComponentNotUniqueError,
     ValidationError,
@@ -47,7 +48,6 @@
     all_greater_or_equal,
     all_greater_than_or_equal_to_zero,
     all_greater_than_zero,
-    all_ids_exist_in_data_set,
     all_less_than,
     all_not_two_values_equal,
     all_not_two_values_zero,
@@ -58,6 +58,7 @@
     all_valid_enum_values,
     all_valid_fault_phases,
     all_valid_ids,
+    ids_valid_in_update_data_set,
     none_missing,
     valid_p_q_sigma,
 )
@@ -149,13 +150,11 @@ def validate_batch_data(
     for batch, batch_update_data in enumerate(batch_data):
         row_update_data = compatibility_convert_row_columnar_dataset(batch_update_data, None, DatasetType.update)
         assert_valid_data_structure(row_update_data, DatasetType.update)
-        id_errors: list[ValidationError] = list(validate_ids_exist(row_update_data, input_data_copy))
 
-        batch_errors = input_errors + id_errors
-        if not id_errors:
-            merged_data = update_input_data(input_data_copy, row_update_data)
-            batch_errors += validate_required_values(merged_data, calculation_type, symmetric)
-            batch_errors += validate_values(merged_data, calculation_type)
+        batch_errors = input_errors
+        merged_data = update_input_data(input_data_copy, row_update_data)
+        batch_errors += validate_required_values(merged_data, calculation_type, symmetric)
+        batch_errors += validate_values(merged_data, calculation_type)
 
         if batch_errors:
             errors[batch] = batch_errors
@@ -216,23 +215,26 @@ def validate_unique_ids_across_components(data: SingleDataset) -> list[MultiComp
     return all_cross_unique(data, [(component, "id") for component in data])
 
 
-def validate_ids_exist(update_data: SingleDataset, input_data: SingleDataset) -> list[IdNotInDatasetError]:
+def validate_ids(update_data: SingleDataset, input_data: SingleDataset) -> list[IdNotInDatasetError | InvalidIdError]:
     """
-    Checks if all ids of the components in the update data exist in the input data. This needs to be true, because you
-    can only update existing components.
+    Checks if all ids of the components in the update data:
+     - exist and match those in the input data
+     - are not present but qualifies for optional id
 
     This function should be called for every update dataset in a batch set
 
     Args:
         update_data: A single update dataset
-        input_data: A power-grid-model input dataset
+        input_data: Input dataset
 
     Returns:
-        An empty list if all update data ids exist in the input dataset, or a list of IdNotInDatasetErrors for
-        all update components of which the id does not exist in the input dataset
+        An empty list if all update data ids are valid, or a list of IdNotInDatasetErrors or InvalidIdError for
+        all update components that have invalid ids
 
     """
-    errors = (all_ids_exist_in_data_set(update_data, input_data, component, "input_data") for component in update_data)
+    errors = (
+        ids_valid_in_update_data_set(update_data, input_data, component, "update_data") for component in update_data
+    )
     return list(chain(*errors))
 
 
diff --git a/tests/unit/validation/test_batch_validation.py b/tests/unit/validation/test_batch_validation.py
index 26520dae9..064f9853d 100644
--- a/tests/unit/validation/test_batch_validation.py
+++ b/tests/unit/validation/test_batch_validation.py
@@ -119,6 +119,7 @@ def test_validate_batch_data_input_error(input_data, batch_data):
 def test_validate_batch_data_update_error(input_data, batch_data):
     batch_data["line"]["from_status"] = np.array([[12, 34], [0, -128], [56, 78]])
     errors = validate_batch_data(input_data, batch_data)
-    assert len(errors) == 2
-    assert [NotBooleanError("line", "from_status", [5, 6])] == errors[0]
-    assert [NotBooleanError("line", "from_status", [5, 7])] == errors[2]
+    assert len(errors) == 3
+    assert NotBooleanError("line", "from_status", [5, 6]) == errors[0][0]
+    assert NotBooleanError("line", "from_status", [5, 7]) == errors[1][1]
+    assert NotBooleanError("line", "from_status", [5, 6]) == errors[2][0]
diff --git a/tests/unit/validation/test_validation_functions.py b/tests/unit/validation/test_validation_functions.py
index 646ecf1f8..73c1102d3 100644
--- a/tests/unit/validation/test_validation_functions.py
+++ b/tests/unit/validation/test_validation_functions.py
@@ -9,8 +9,16 @@
 import pytest
 
 from power_grid_model import CalculationType, LoadGenType, MeasuredTerminalType, initialize_array, power_grid_meta_data
+from power_grid_model._utils import compatibility_convert_row_columnar_dataset
 from power_grid_model.core.dataset_definitions import ComponentType, DatasetType
-from power_grid_model.enum import Branch3Side, BranchSide, CalculationType, FaultType, TapChangingStrategy
+from power_grid_model.enum import (
+    Branch3Side,
+    BranchSide,
+    CalculationType,
+    ComponentAttributeFilterOptions,
+    FaultType,
+    TapChangingStrategy,
+)
 from power_grid_model.validation import assert_valid_input_data
 from power_grid_model.validation.errors import (
     IdNotInDatasetError,
@@ -27,7 +35,7 @@
 from power_grid_model.validation.validation import (
     assert_valid_data_structure,
     validate_generic_power_sensor,
-    validate_ids_exist,
+    validate_ids,
     validate_required_values,
     validate_unique_ids_across_components,
     validate_values,
@@ -113,7 +121,7 @@ def test_validate_unique_ids_across_components():
     assert len(unique_id_errors[0].ids) == 4
 
 
-def test_validate_ids_exist():
+def test_validate_ids():
     source = initialize_array("input", "source", 3)
     source["id"] = [1, 2, 3]
 
@@ -135,10 +143,47 @@ def test_validate_ids_exist():
 
     update_data = {"source": source_update, "sym_load": sym_load_update}
 
-    invalid_ids = validate_ids_exist(update_data, input_data)
+    invalid_ids = validate_ids(update_data, input_data)
+
+    assert IdNotInDatasetError("source", [4], "update_data") in invalid_ids
+    assert IdNotInDatasetError("sym_load", [7], "update_data") in invalid_ids
+
+    source_update_no_id = initialize_array("update", "source", 3)
+    source_update_no_id["u_ref"] = [1.0, 2.0, 3.0]
+
+    update_data_col = compatibility_convert_row_columnar_dataset(
+        data={"source": source_update_no_id, "sym_load": sym_load_update},
+        data_filter=ComponentAttributeFilterOptions.relevant,
+        dataset_type=DatasetType.update,
+    )
+    invalid_ids = validate_ids(update_data_col, input_data)
+    assert len(invalid_ids) == 1
+    assert IdNotInDatasetError("sym_load", [7], "update_data") in invalid_ids
+
+    source_update_less_no_id = initialize_array("update", "source", 2)
+    source_update_less_no_id["u_ref"] = [1.0, 2.0]
 
-    assert IdNotInDatasetError("source", [4], "input_data") in invalid_ids
-    assert IdNotInDatasetError("sym_load", [7], "input_data") in invalid_ids
+    update_data_col_less_no_id = compatibility_convert_row_columnar_dataset(
+        data={"source": source_update_less_no_id, "sym_load": sym_load_update},
+        data_filter=ComponentAttributeFilterOptions.relevant,
+        dataset_type=DatasetType.update,
+    )
+    invalid_ids = validate_ids(update_data_col_less_no_id, input_data)
+    assert len(invalid_ids) == 2
+    assert IdNotInDatasetError("sym_load", [7], "update_data") in invalid_ids
+
+    source_update_part_nan_id = initialize_array("update", "source", 3)
+    source_update_part_nan_id["id"] = [1, np.iinfo(np.int32).min, 4]
+    source_update_part_nan_id["u_ref"] = [1.0, 2.0, 3.0]
+
+    update_data_col_part_nan_id = compatibility_convert_row_columnar_dataset(
+        data={"source": source_update_part_nan_id, "sym_load": sym_load_update},
+        data_filter=ComponentAttributeFilterOptions.relevant,
+        dataset_type=DatasetType.update,
+    )
+    invalid_ids = validate_ids(update_data_col_part_nan_id, input_data)
+    assert len(invalid_ids) == 2
+    assert IdNotInDatasetError("sym_load", [7], "update_data") in invalid_ids
 
 
 @pytest.mark.parametrize(