pynite-reporting 0.3.0

pynite_reporting: A 3rd party package to aid in extracting

Easily extract results from Pynite... with pynite_reporting!

PyniteFEA is excellent and it is generally design-ready...if it weren't for all the trouble we have to go through to get results out. This is not unique to Pynite, most FEA programs require a significant amount of post-processing to prepare the actual analysis results for design.

Enter pynite_reporting

This package provides a series of functions that consume a Pynite.FEModel3D object and returns consistently-structured dictionaries of analysis results.

Note: As of 2025-06-30, this package has only been "casually tested" (meaning simple visual checking of outputs). No test suite has been written (but is coming).

Installation

pip install pynite_reporting

Dependencies

• Numpy (>= 2.0.0)

(PyNiteFEA is not a dependency but it is assumed to be in your working environment)

Pynite Compatibility

PyniteFEA >= 1.0.0

(Not compatible with pre-v1.0 versions!)

Examples (typical use)

from Pynite import FEModel3D
import pynite_reporting as pr

model = FEModel3D(...) # Build your model here

# Selected load combinations in your model
lcs = [
    # 'LC1', 
    'LC2',
    'LC3',
    # 'LC4', 
    # 'LC5',
]

# All the below functions optionally take a list of load combos
# so you can select which combos to extract

## Additionally, each function accepts a results_key parameter.
## This optional parameter is set to a default str value, unique for each function.
## When you set the results_key=None, then your results tree will be one level shallower.

# Return reactions for all supports, all load combos
reactions = pr.extract_node_reactions(
    model,
    # load_combinations=lcs,
    # results_key=None
)

# Returns all node deflections for all load combos
node_deflections = pr.extract_node_deflections(
    model,
    # load_combinations=lcs,
    # results_key=None
)

# Return force arrays for all members, all load combos
force_arrays = pr.extract_member_arrays(
    model,
    # n_points=1000,
    # as_lists=False,
    # load_combinations=lcs,
    # results_key=None
)

# Return force min/max/absmax envelope for all members, all load combos
# Values will not necessarily be at concurrent locations
forces_minmax = pr.extract_member_envelopes(
    model,
    # load_combinations=lcs,
    # results_key=None
)

# Return force min/max envelope for each span in all members, all load combos
forces_minmax_spans = pr.extract_span_envelopes(
    model,
    # load_combinations=lcs,
    # results_key=None
)

# Return forces for all load combos at specific locations along the global member length
forces_at_locations = pr.extract_member_actions_by_location(
    model, 
    force_extraction_locations={"Member01": [0, 2000, 3600]},
    # load_combinations=lcs,
    # results_key=None
)

# Return forces for all load combos at 1/4 points for *each span* of the given members
forces_at_location_ratios = pr.extract_member_actions_by_location(
    model, 
    force_extraction_ratios={"Member05": [0.25, 0.5, 0.75]}, 
    by_span=True,
    # load_combinations=lcs,
    # results_key=None
    )

And there you have it! Does that not make your life a little bit easier?

Merge result trees

NEW in v0.2.0

If you are planning on serializing your results to a JSON file, then you can merge all of your result tree (dictionaries) into a single dictionary by using merge_trees:

e.g.

merged_tree = pr.merge_trees([force_arrays, forces_minmax, forces_at_locations])

Convenience functions

There are one-line function for writing/reading dictionaries to JSON files:

pr.to_json("results.json", merged_tree)

round_tripped_tree = pr.from_json("results.json")

FYI (Opinions at work!)

I have made the decision to remove unnecessary results from being returned by some of these functions.

"???WHAA??? I want to see ALL of my results!!!", you say?

I don't think you actually do. Consider the following small amount of results:

{
    'M_col': {
        'shear': {
            'Fy': {
                'LC1': {'max': 0, 'min': 0}, # No loading for this load case on this member
                'LC2': {'max': 4000, 'min': 4000}
            },
            'Fz': {
                'LC1': {'max': 10000, 'min': 10000}, 
                'LC2': {'max': 0, 'min': 0} # No loading for this either...
            },
            
        },
        'moment': {
            'Mz': {
                'LC1': {'max': 0, 'min': 0}, # Or this...
                'LC2': {'max': 20000, 'min': 0}
            }, 
            'My': {
                'LC1': {'max': 50000, 'min': 50000},
                'LC2': {'max': 0, 'min': 0} # Or this...
            }
        },
        ...
    }
}

The above results contain unnecessary data. This structure has loading in the gravity direction and the transverse direction, each on a different load case/combo.

The load cases that show as 0, 0 indicate that the force diagrams are completely flat and without activity.

To avoid confusion in reading and to prevent unnecessary iterations (if you are putting these results through an automated process), I have filtered out the keys that result in null values.

Here is how the above results are returned:

    'M_col': {
        'shear': {
            'Fy': {
                'LC2': {'max': 4000, 'min': 4000}
            },
            'Fz': {
                'LC1': {'max': 10000, 'min': 10000}, 
            },
        },
        'moment': {
            'Mz': {
                'LC2': {'max': 20000, 'min': 0}
            }, 
            'My': {
                'LC1': {'max': 50000, 'min': 50000}
            }
        },
        ...
    },

So, you know that all other load combos result in null values without having to physically read a bunch of zeros or confusing "near zero" values.

The tolerance for this is an absolute tolerance of 1e-7. Currently, this is not parameterized and is hard-coded into the package (because it was easier and made the function signatures cleaner). So, even if you have REALLY small result values (on the order of 0.000001 units), those values will still be returned to you (and not excluded).

This is allows you to see all concurrent forces for a load combination at a given location.