Example 2: Manually tagged elements#

The drawing from papermodels’ Example 2, where FB2.5, FB1.14 and CB0.10 were tagged by hand in the PDF and the other elements were tagged around them. loadtracer uses the tags that papermodels assigns, so the manually tagged elements appear in the results under their own names.

Note: papermodels 0.21 keeps the manual tags FB1.14 and CB0.10, but names the beam that the PDF tags FB2.5 (page 2) FB2.0. That is a papermodels issue; loadtracer reports whatever tags the graph has.

The notebook and its PDF are in Examples/Example 2 in the repository.

%matplotlib inline
import sys
import warnings
from decimal import Decimal

import pandas as pd
from papermodels.datatypes.geometry_graph import GeometryGraph
from papermodels.datatypes.joist_models import JoistArrayModel
from loadtracer.load_tracer import LoadTracer

sys.path.insert(0, "..")
from helpers import (
    bottom_members,
    compare_modes,
    load_check,
    load_table,
    occupancies_in,
    quietly,
    reaction_table,
)

pd.options.display.float_format = "{:,.2f}".format
pd.options.display.max_rows = 40

1. Build the graph (papermodels)#

pdf_path = "sketch_to_scale-w_manual_tag.pdf"

graph = GeometryGraph.from_pdf_file(pdf_path, scale=Decimal(1 / 72 * 4))
graph.assign_collector_behaviour(JoistArrayModel, spacing=1.0)

/home/engineering/papermodels/src/papermodels/datatypes/joist_models.py:237: UserWarning: Joist array J2.0: support WT2.0 is a polygon with a point reaction type; the joists bear on it along its long axis. If it is a wall, add 'Reaction Type: Linear' to its legend entry.
  warnings.warn(msg)

graph.plot_annotations(1)
../../_images/e420d54f5bf38f7525c4891ffd330de755ecbf9c5229972f52bc4e1acf9f470f.png

2. Occupancies#

loadtracer needs the loads of each occupancy in the drawing, as pressures by load case. These are the occupancies that the loading areas use:

occupancies_in(graph)

['big stuff', 'residential', 'roof']

The drawing is in feet, so these pressures (psf) give loads in lb/ft and lb.

occupancies = {
    "big stuff": {"D": 50.0, "L": 100.0},
    "residential": {"D": 15.0, "L": 40.0},
    "roof": {"D": 20.0, "S": 35.0},
}

3. Run loadtracer#

Every element is analysed from the top of the graph down. Only the graph and the occupancies are needed; element heights and floor elevations are not inputs. artifacts/ receives the analysis models (models.json) and the reactions (reaction_forces.json).

lt = LoadTracer(geometry_graph=graph, occupancies=occupancies)
with warnings.catch_warnings(record=True) as caught:
    warnings.simplefilter("always")
    quietly(lt.execute_analyses, artifacts_dir="artifacts")

print(f"{len(lt.reactions)} members analysed, {len(lt.errors)} errors")
for w in caught:
    print(f"{type(w.message).__name__}: {w.message}")

55 members analysed, 0 errors
JoistLoadVariabilityWarning: Member FB1.14: the reactions of joist array J1.0 (D) vary by 3.0:1 along 0.116 to 18.274; they were converted to a uniform load. Review the joist loads.
JoistLoadVariabilityWarning: Member FB1.14: the reactions of joist array J1.0 (L) vary by 2.5:1 along 0.116 to 18.274; they were converted to a uniform load. Review the joist loads.
JoistTribWidthWarning: Member FB1.14: joist array J1.0 carries big stuff, residential; its trib widths are expressed with 'big stuff' and are approximate.
JoistLoadVariabilityWarning: Member FB1.0: the reactions of joist array J1.0 (D) vary by 2.7:1 along 0.0 to 18.159; they were converted to a uniform load. Review the joist loads.
JoistLoadVariabilityWarning: Member FB1.0: the reactions of joist array J1.0 (L) vary by 2.3:1 along 0.0 to 18.159; they were converted to a uniform load. Review the joist loads.
JoistTribWidthWarning: Member FB1.0: joist array J1.0 carries big stuff, residential; its trib widths are expressed with 'big stuff' and are approximate.

All of the load applied to the building reaches the bottom of the graph:

load_check(lt)
applied at the bottom
D 8,666.48 8,666.48
S 6,563.13 6,563.13
L 11,832.06 11,832.06

lt.issues lists input that loadtracer corrected so that the rundown could continue (for example, load drawn past the end of a member, which is moved to the member end):

{tag: issues for tag, issues in lt.issues.items()}

{}

4. One member, end to end: FB1.14#

FB1.14 is one of the manually tagged beams. It is continuous over three supports and carries the J1.0 joist array.

pd.DataFrame(lt.converted_joist_loads("FB1.14"))
member array joists case dir profile start_loc end_loc w_start w_end total max_min_ratio occupancy pressure trib_width_start trib_width_end approximate
0 FB1.14 J1.0 20 D FY variable 0.12 18.27 141.11 141.11 2,562.36 3.04 big stuff 50.00 2.82 2.82 True
1 FB1.14 J1.0 20 L FY variable 0.12 18.27 337.30 337.30 6,124.63 2.53 big stuff 100.00 3.37 3.37 True

The loads on FB1.14 after expansion, by load case:

load_table(lt, "FB1.14")
kind source case start end w_start / P w_end total profile
0 point CB1.0 D 4.98 4.98 1,196.47 NaN 1,196.47
1 point CB1.0 S 4.98 4.98 2,093.82 NaN 2,093.82
2 distributed J1.0 D 0.12 18.27 141.11 141.11 2,562.36 variable
3 distributed J1.0 L 0.12 18.27 337.30 337.30 6,124.63 variable

The loads in the Pynite model of FB1.14:

member = lt.models["FB1.14"]["members"]["FB1.14"]
pd.DataFrame(member["DistLoads"] + member["PtLoads"])
0 1 2 3 4 5 6
0 FY 141.11 141.11 0.12 18.27 D False
1 FY 337.30 337.30 0.12 18.27 L False
2 FY 1,196.47 4.98 D NaN NaN None
3 FY 2,093.82 4.98 S NaN NaN None

The reactions of FB1.14 (positive downwards on the supports):

reaction_table(lt, ["FB1.14"])
D L S
member support
FB1.14 0 1,361.71 1,791.11 1,071.64
1 2,952.78 4,728.17 1,705.66
2 -555.67 -394.64 -683.48

The manually tagged elements that papermodels kept, under their own tags:

reaction_table(lt, ["FB1.14", "CB0.10"])
D L S
member support
FB1.14 0 1,361.71 1,791.11 1,071.64
1 2,952.78 4,728.17 1,705.66
2 -555.67 -394.64 -683.48
CB0.10 0 2,952.78 4,728.17 1,705.66

The PDF’s FB2.5 is not in the graph; the level 2 beams are:

sorted(tag for tag in lt.reactions if tag.startswith("FB2"))

['FB2.0']

5. Joist loads as point loads or distributed loads#

With joist_loads="point" each joist reaction is applied as a point load, which is the most detailed representation. The total load is the same either way. Individual reactions are the same for uniform joist reactions and differ more the more the joist reactions vary; arrays that vary by 2:1 or more are listed in the warnings in section 3. For those, check the point load results.

lt_point = LoadTracer(geometry_graph=graph, occupancies=occupancies, joist_loads="point")
with warnings.catch_warnings():
    warnings.simplefilter("ignore")
    quietly(lt_point.execute_analyses)

pd.concat({"point": load_check(lt_point)["at the bottom"], "distributed": load_check(lt)["at the bottom"]}, axis=1)
point distributed
D 8,666.48 8,666.48
S 6,563.13 6,563.13
L 11,832.06 11,832.06 
compare_modes(lt_point, lt, ["FB1.14"])
point distributed difference
D L S D L S D L S
member support
FB1.14 0 1,404.01 1,876.25 1,071.59 1,361.71 1,791.11 1,071.64 -42.29 -85.15 0.05
1 3,140.38 5,101.49 1,705.58 2,952.78 4,728.17 1,705.66 -187.59 -373.32 0.08
2 -785.61 -853.11 -683.45 -555.67 -394.64 -683.48 229.95 458.47 -0.03

The J1.0 joists carry the heavier “big stuff” occupancy over part of FB1.14, so their reactions vary by up to 3:1 (see the JoistLoadVariabilityWarning in section 3). As a uniform load, that concentration is spread along the beam, which moves load from the interior support to the ends. Here, the point load results are the ones to use.

6. Reactions at the bottom#

The loads delivered to the elements at the bottom of the graph (the foundation):

reaction_table(lt, bottom_members(lt))
D S L
member support
WB0.0 0 1,357.43 2,375.49 0.00
CB0.2 0 -555.67 -683.48 -394.64
CB0.4 0 1,361.71 1,071.64 1,791.11
CB0.10 0 2,952.78 1,705.66 4,728.17
CB0.0 0 2,425.06 1,615.36 3,642.06
CB0.1 0 469.33 -123.20 1,308.75
CB0.3 0 655.83 601.65 756.62