Abaqus version: 2018.hf4

We have created a ODB and writing a Tensor data at the nodes.
There are 4 steps each includes 24 frames for 27k nodes.
So, the data size is very huge --> (272107, 96, 6).

When we are writing this data, we checked the memory usage using following
option:
session.kernelMemoryFootprint

For every frame it is writing the data, the memory usage is
increasing by ~300MB. At the end, the total memory usage is ~25-30GB.
Could anybody please explain why this much memory is required here?

Secondly, I thought closing the odb will release the memory,
however, then also the consumed memory usage (~25-30GB) is NOT released.
Due to this high memory usage, Abaqus/CAE is crashing.
How can we release this memory?

Following function, we have defined to write the data to the odb:

def write_to_tempODB(temp_subodb, isn, steps_info, data, node_labels):
    subodb_inst = temp_subodb.rootAssembly.instances[isn]
    data_count = 0
    for i, (stpname, vals) in enumerate(steps_info.items()):
        # Accessing step related information
        stp_descr, stp_dom, stp_tp, stp_tt, stp_pstp, frame_data = vals

        print('Writing step data', stpname)
        if temp_subodb.steps.has_key(stpname):
            istep = temp_subodb.steps[stpname]
        else:
            istep = temp_subodb.Step(name=stpname, description=stp_descr,
                                     domain=stp_dom, timePeriod=stp_tp,
                                     totalTime=stp_tt,
                                     previousStepName=stp_pstp)

        for (frm_id, frm_incr, frm_val, frm_descr) in frame_data:
            # Creating a frame
            iframe = istep.Frame(incrementNumber=frm_incr, frameValue=frm_val,
                                 description=frm_descr)

            # Creating Total Strain field output
            ifield = iframe.FieldOutput(
                    name='E', type=TENSOR_3D_FULL,
                    description='Total Strain components',
                    componentLabels=('E11', 'E22', 'E33', 'E12', 'E13', 'E23'),
                    validInvariants=(MAX_PRINCIPAL, MIN_PRINCIPAL),
                    )

            ifield.addData(position=NODAL, instance=subodb_inst,
                           labels=tuple(node_labels),
                           data=tuple(data[:, data_count, :]))
            data_count += 1