Opening result files with DPF / Defining datasources

Pernelle Marone-Hitz

There are different ways of referencing the datasources in DPF, which are they?

Pernelle Marone-Hitz

The first method is to use streams. A stream opens the data files and these files will stay opened with some data cached to make the next evaluation faster. Remember to close the streams after you're finished with them!

A typical example using streams would look like this:

import mech_dpf
import Ans.DataProcessing as dpf
mech_dpf.setExtAPI(ExtAPI)
#Data sources
dataSources = mech_dpf.GetStreams()
#Scoping
scoping = dpf.Scoping()
scoping.Ids = [1]
scoping.Location = 'Nodal'
#Stress X direction
stressXOp = dpf.operators.result.stress_X()
stressXOp.inputs.streams_container.Connect(dataSources)
stressXOp.inputs.mesh_scoping.Connect(scoping)
sX = stressXOp.outputs.fields_container.GetData()
print('sX is: ')
print(sX)
print('sX[0] is:')
print(sX[0])
print('sX on node #1 is: ')
print(sX[0].Data)
# Release stream
dataSources = dataSources.ReleaseHandles()

Another method is to directly set the path to the result file. The same example would then look like this:

import mech_dpf
import Ans.DataProcessing as dpf
mech_dpf.setExtAPI(ExtAPI)
#Get path to result file
import os
folder = r"D:\DummyModel_files\dp0\SYS\MECH"
filename = "file.rst"
filepath = os.path.join(folder,filename)
#Data sources
dataSources = dpf.DataSources()
dataSources.SetResultFilePath(filepath)
#Scoping
scoping = dpf.Scoping()
scoping.Ids = [1]
scoping.Location = 'Nodal'
#Stress X direction
stressXOp = dpf.operators.result.stress_X()
stressXOp.inputs.data_sources.Connect(dataSources)
stressXOp.inputs.mesh_scoping.Connect(scoping)
sX = stressXOp.outputs.fields_container.GetData()
print('sX on node #1 is: ')
print(sX[0].Data)

Note: A very convenient way to get the path to the result file from the Mechanical interface is to use the Mechanical automation API:

analysis = ExtAPI.DataModel.Project.Model.Analyses[0]
filepath = analysis.ResultFileName

Finally, another very convenient way of opening a file through DPF is to use the Model class. This class facilitates the access to the various information about the result file. The same example as before, using the Model class:

import mech_dpf
import Ans.DataProcessing as dpf
mech_dpf.setExtAPI(ExtAPI)
#Get path to result file
analysis = ExtAPI.DataModel.Project.Model.Analyses[0]
filepath = analysis.ResultFileName
#Data sources
dataSources = dpf.DataSources()
dataSources.SetResultFilePath(filepath)
# Model
model=dpf.Model(dataSources)
print(model)
print(model.AvailableNamedSelections)
print(model.Mesh)
#Scoping
scoping = dpf.Scoping()
scoping.Ids = [1]
scoping.Location = 'Nodal'
#Stress X direction
stressXOp = dpf.operators.result.stress_X()
stressXOp.inputs.data_sources.Connect(dataSources)
stressXOp.inputs.mesh_scoping.Connect(scoping)
sX = stressXOp.outputs.fields_container.GetData()
print('sX on node #1 is: ')
print(sX[0].Data)

Many information can be fetched from the Model class:

model=dpf.Model(dataSources)
print(model)
print(model.ResultInfo)
print(model.AvailableNamedSelections)
print(model.TimeFreqSupport)
print(model.ResultInfo.UnitSystem)
print(model.ResultInfo.AnalysisType)
print(model.Mesh)
# Access mesh
my_nodes = model.Mesh.Nodes
my_nodes [0].Id
my_nodes [0].X
model.Mesh.ElementScoping.Ids[10]

total_obliteration

Is there any calculation-time-wise benefit of using streams? Currently, my script to extract contacts reaction force (using nmisc) for every of 77 load steps from only 3 named selection runs more than 3 hours.

Could you provide insights how to optimize working with big .rst files (150GB+)?

Pernelle Marone-Hitz

Evaluation will be faster with steams as the data is cached. For big .rst files, you could consider using incremental operator.