simcenter-dlml 3.0__tar.gz → 3.2__tar.gz

{simcenter_dlml-3.0 → simcenter_dlml-3.2}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.4
+Metadata-Version: 2.5
 Name: simcenter-dlml
-Version: 3.0
+Version: 3.2
 Summary: Damage and Loss Model Library
 Project-URL: Homepage, https://github.com/NHERI-SimCenter/DamageAndLossModelLibrary
 Project-URL: Documentation, https://nheri-simcenter.github.io/DamageAndLossModelLibrary/
@@ -28,7 +28,7 @@ Requires-Dist: codespell; extra == 'dev'
 Requires-Dist: colorlover; extra == 'dev'
 Requires-Dist: mypy; extra == 'dev'
 Requires-Dist: numpydoc; extra == 'dev'
-Requires-Dist: pelicun<3.10,>=3.9; extra == 'dev'
+Requires-Dist: pelicun<4,>=3.10; extra == 'dev'
 Requires-Dist: plotly; extra == 'dev'
 Requires-Dist: pytest-cov; extra == 'dev'
 Requires-Dist: pytest>=7; extra == 'dev'
@@ -39,7 +39,8 @@ Requires-Dist: sphinx; extra == 'dev'
 Requires-Dist: sphinx-design; extra == 'dev'
 Requires-Dist: sphinx-rtd-theme; extra == 'dev'
 Requires-Dist: sphinxcontrib-bibtex; extra == 'dev'
-Requires-Dist: streamlit>=1.48.0; extra == 'dev'
+Requires-Dist: streamlit>=1.48.0; (python_version < '3.10') and extra == 'dev'
+Requires-Dist: streamlit>=1.58.0; (python_version >= '3.10') and extra == 'dev'
 Requires-Dist: tqdm; extra == 'dev'
 Provides-Extra: doc
 Requires-Dist: colorlover; extra == 'doc'
@@ -56,13 +57,14 @@ Requires-Dist: colorlover; extra == 'explorer'
 Requires-Dist: plotly; extra == 'explorer'
 Requires-Dist: qdrant-client[fastembed]>=1.14.0; extra == 'explorer'
 Requires-Dist: scipy; extra == 'explorer'
-Requires-Dist: streamlit>=1.48.0; extra == 'explorer'
+Requires-Dist: streamlit>=1.48.0; (python_version < '3.10') and extra == 'explorer'
+Requires-Dist: streamlit>=1.58.0; (python_version >= '3.10') and extra == 'explorer'
 Provides-Extra: lint
 Requires-Dist: codespell; extra == 'lint'
 Requires-Dist: mypy; extra == 'lint'
 Requires-Dist: ruff==0.7.0; extra == 'lint'
 Provides-Extra: test
-Requires-Dist: pelicun<3.10,>=3.9; extra == 'test'
+Requires-Dist: pelicun<4,>=3.10; extra == 'test'
 Requires-Dist: pytest-cov; extra == 'test'
 Requires-Dist: pytest>=7; extra == 'test'
 Description-Content-Type: text/markdown
@@ -180,6 +182,10 @@ The library is the data backbone for
 damage-and-loss engine, and is available automatically through the **PBE** and
 **R2D** desktop applications that run Pelicun under the hood.
 
+Pelicun 3.10 and later install this library from PyPI and need no further setup.
+Pelicun 3.9 and earlier read the model data from this repository instead; see
+[Release policy](#release-policy) if you are still on one of those versions.
+
 ### Work with the raw data
 
 You can also read the CSV/JSON files directly from a clone of this repository:
@@ -192,6 +198,36 @@ git clone https://github.com/NHERI-SimCenter/DamageAndLossModelLibrary.git
 > with auto-generated model pages also remains available; it will be revised over
 > time to complement the DLML Explorer.
 
+## Release policy
+
+Releases are published to PyPI as [`simcenter-dlml`](https://pypi.org/project/simcenter-dlml/)
+and tagged here, with matching entries in [`CHANGELOG.md`](CHANGELOG.md). Version
+numbers use two segments (`3.0`, `3.1`, ...), and every model in a release is a
+fixed snapshot, so an analysis can cite the exact data it used.
+
+### Migrating from Pelicun 3.9 and earlier
+
+Pelicun 3.9 and earlier downloaded the model data from this repository's latest
+GitHub release when you imported them. Since v3.0 the data lives under
+`src/dlml/data/`, a layout those versions cannot read, so they now fail with a
+download error.
+
+We recommend upgrading. Pelicun 3.10 and later depend on `simcenter-dlml` directly
+and download nothing:
+
+```bash
+pip install --upgrade pelicun
+```
+
+If you need to stay on Pelicun 3.9 or earlier, pin the model data to the last
+release those versions can read:
+
+```bash
+pelicun dlml update v2.1.0
+```
+
+The `v2.1.0` release stays available for this purpose.
+
 ## Contributing
 
 We welcome new and improved models from researchers and practitioners. A formal

{simcenter_dlml-3.0 → simcenter_dlml-3.2}/README.md

@@ -111,6 +111,10 @@ The library is the data backbone for
 damage-and-loss engine, and is available automatically through the **PBE** and
 **R2D** desktop applications that run Pelicun under the hood.
 
+Pelicun 3.10 and later install this library from PyPI and need no further setup.
+Pelicun 3.9 and earlier read the model data from this repository instead; see
+[Release policy](#release-policy) if you are still on one of those versions.
+
 ### Work with the raw data
 
 You can also read the CSV/JSON files directly from a clone of this repository:
@@ -123,6 +127,36 @@ git clone https://github.com/NHERI-SimCenter/DamageAndLossModelLibrary.git
 > with auto-generated model pages also remains available; it will be revised over
 > time to complement the DLML Explorer.
 
+## Release policy
+
+Releases are published to PyPI as [`simcenter-dlml`](https://pypi.org/project/simcenter-dlml/)
+and tagged here, with matching entries in [`CHANGELOG.md`](CHANGELOG.md). Version
+numbers use two segments (`3.0`, `3.1`, ...), and every model in a release is a
+fixed snapshot, so an analysis can cite the exact data it used.
+
+### Migrating from Pelicun 3.9 and earlier
+
+Pelicun 3.9 and earlier downloaded the model data from this repository's latest
+GitHub release when you imported them. Since v3.0 the data lives under
+`src/dlml/data/`, a layout those versions cannot read, so they now fail with a
+download error.
+
+We recommend upgrading. Pelicun 3.10 and later depend on `simcenter-dlml` directly
+and download nothing:
+
+```bash
+pip install --upgrade pelicun
+```
+
+If you need to stay on Pelicun 3.9 or earlier, pin the model data to the last
+release those versions can read:
+
+```bash
+pelicun dlml update v2.1.0
+```
+
+The `v2.1.0` release stays available for this purpose.
+
 ## Contributing
 
 We welcome new and improved models from researchers and practitioners. A formal

simcenter_dlml-3.2/data_sources/seismic/transportation_network/component/California RC Bridges 2025/source/README.md

@@ -0,0 +1,394 @@
+# Source data
+
+`Sa_1_Fragility Database.csv` is a copy of the file of the same name in
+DesignSafe project PRJ-5910, version 1 (published 2025-04-29,
+doi:10.17603/ds2-c73m-nj37, Open Data Commons Attribution License):
+
+Chen, S., Y. Xie, C. Wu, H. V. Burton, J. E. Padgett, and Á. Zsarnóczay. 2025.
+Second-Generation Component and System-Level Seismic Fragility Models for
+Reinforced Concrete Bridges in California. DesignSafe-CI, PRJ-5910.
+
+The data paper describing it is Chen et al. (2025), Earthquake Spectra 41(4):
+3234–3253, doi:10.1177/87552930251343634.
+
+The same project version also publishes `EDP_Fragility Database.csv`, which
+lists the median capacities of each component for its damage states, in terms of
+the engineering demand parameter (EDP) of the component. That file is kept in
+`../../California RC Bridges 2025 EDP/source/`. The generator in
+`../../California RC Bridges 2025 EDP/` builds the EDP-based dataset
+`seismic/transportation_network/component/California RC Bridges 2025 EDP` from
+it. The generator of this folder, `../generate_library_files.py`, reads it only
+to confirm the capacities quoted in the damage-state descriptions (see
+[Damage-state descriptions](#damage-state-descriptions)); it changes no
+fragility parameter here.
+
+This file is the record of every change the generator makes to the published
+labels and values and of every published row it omits. The same changes are
+recorded in the `correction` column of `../id_crosswalk.csv` and in the Comments
+of each affected model.
+
+## File format
+
+`Sa_1_Fragility Database.csv`:
+
+- UTF-8 with a byte-order mark, 1,018 data rows; CRLF line endings in the
+  DesignSafe file, stored here with LF endings.
+- Columns: `Bridge group`, `Fragility notation`, the lognormal median (`_λ`, in
+  g) and dispersion (`_ζ`) of Sa(1.0 s) for the `Slight`, `Moderate`,
+  `Extensive`, and `Complete` damage states, `Notes`, and `Reference`.
+- `N/A` marks a damage state with negligible probability of damage from ground
+  shaking; an empty cell marks a state the source does not define.
+
+Row numbers used in the generator, the crosswalk, the model metadata, and this
+file are 0-based data rows of `Sa_1_Fragility Database.csv`; the spreadsheet row
+is the data row plus 2.
+
+Keep the file unchanged. Corrections are applied by
+`../generate_library_files.py`.
+
+## Number formatting
+
+The source values have at most two decimals and no trailing zeros (3,968 cells
+with two decimals, 489 with one, and 47 with none); the generator stops if a
+cell has more. Every value read from the source is written to `fragility.csv`
+exactly as published, with only the float noise of the conversion removed. The
+recomputed values of data rows 0 to 7 (see [Recomputed single-span
+diaphragm-abutment curves](#recomputed-single-span-diaphragm-abutment-curves))
+are rounded to four decimals, and trailing zeros are dropped, so 32.5740 is
+written 32.574. The Comments of those models quote each published value as
+printed and each recomputed value as stored.
+
+## Changes to the published labels and values
+
+### Column-shape labels
+
+Three rows carry the wrong column shape in their `Notes`. The values are
+unchanged; only the label, and with it the shape token of the model ID, is
+corrected.
+
+| Row | Published shape | Stored shape |
+|---|---|---|
+| 298 | circular | oblong |
+| 299 | oblong | circular |
+| 741 | circular | rectangular |
+
+The corrected models are `CRCB.E2.S3P.C1.D.AbPass.BoxGirder.Obl.M17` (row 298),
+`CRCB.E2.S3P.C1.D.AbPass.BoxGirder.Span5.MF.Circ.MCP18` (row 299), and
+`CRCB.E1.S3P.C1.S.System.BoxGirder.Rect.M17` (row 741).
+
+Evidence:
+
+- Row 298: the published values match the oblong-column class S-E2-S34-O-D of
+  the Mangalathu (2017) thesis.
+- Row 299: Mangalathu et al. (2018) model circular columns only (Fig. 2c, Fig.
+  3, and Table 1 of the study). The source exchanges the shape labels of rows
+  298 and 299; the values are in the correct rows.
+- Row 741: the published values match the rectangular-column class S-E1-S34-R-S
+  of the thesis and the omitted rectangular-column Soleimani (2017) entry of the
+  same group (data row 744).
+
+### Abutment directions of the single-span seat-abutment curves
+
+The four Cavalcante et al. (2022) active and passive abutment curves of group
+E123-S1-C0-S carry the values that Cavalcante et al. (2022, Table 9) give for
+the other direction. The notation is exchanged; the values are unchanged.
+
+| Data row | Deck | Published notation | Stored notation | Model ID |
+|---|---|---|---|---|
+| 26 | slab | AbAct | AbPass | `CRCB.E123.S1.C0.S.AbPass.Slab.CPR22` |
+| 27 | T-beam | AbAct | AbPass | `CRCB.E123.S1.C0.S.AbPass.TBeam.CPR22` |
+| 28 | slab | AbPass | AbAct | `CRCB.E123.S1.C0.S.AbAct.Slab.CPR22` |
+| 29 | T-beam | AbPass | AbAct | `CRCB.E123.S1.C0.S.AbAct.TBeam.CPR22` |
+
+The published values of these rows, as median in g and dispersion for the slight
+and the moderate state:
+
+| Row | Slight λ | Slight ζ | Moderate λ | Moderate ζ |
+|---|---|---|---|---|
+| 26 | 24.8 | 0.95 | 104 | 0.95 |
+| 27 | 27.7 | 0.78 | 117 | 0.78 |
+| 28 | 10.9 | 0.95 | 35.6 | 0.95 |
+| 29 | 12.9 | 0.77 | 43 | 0.77 |
+
+In the tables of this file, λ is the median in g and ζ the dispersion, as in the
+column names of the source file.
+
+### Recomputed single-span diaphragm-abutment curves
+
+In the literature they surveyed, Chen et al. (2025) found no viable set of
+fragility models for single-span bridges with diaphragm abutments. They derived
+the active and passive abutment curves of group E123-S1-C0-D (data rows 0 to 7)
+from the seat-abutment curves of E123-S1-C0-S with modification factors. A
+modification factor is the ratio of a diaphragm-abutment to the corresponding
+seat-abutment fragility parameter in the two-span, single-column groups of one
+era: Era 1, or Eras 2 and 3, as the `Notes` of each row state. The published
+values show the factors applied to the medians and the dispersions; the paper
+describes the adjustment of the medians only. The database does not list the
+factors.
+
+The factors were applied by the labels of the seat rows. The slight-median
+factors of Eras 2 and 3 can be computed from the source file. The Era 2 and Era
+3 two-span, single-column groups publish identical AbAct and AbPass curves, so
+the Era 2 rows serve for both: E2-S2-C1-D (diaphragm abutments) and E2-S2-C1-S
+(seat abutments), each with a circular-column and an oblong-column curve. F is
+the mean of the two diaphragm-to-seat ratios:
+
+AbAct:  F = mean(0.50 / 0.73, 0.42 / 0.66) = 0.6606   (data rows 114, 115 over
+536, 537) AbPass: F = mean(1.05 / 1.31, 0.94 / 1.24) = 0.7798   (data rows 116,
+117 over 538, 539)
+
+Applied to the seat rows as labeled, these factors reproduce the published
+slight medians of rows 1 and 5: 0.6606 × 24.8 = 16.38 g (seat row 26, labeled
+AbAct) and 0.7798 × 10.9 = 8.50 g (seat row 28, labeled AbPass). Applied to the
+seat rows that hold the direction of each row, they give 0.6606 × 10.9 = 7.20 g
+and 0.7798 × 24.8 = 19.34 g. The AbAct diaphragm curves were therefore built on
+the seat rows labeled AbAct (26 and 27), which hold the passive-direction
+values, and the AbPass curves on rows 28 and 29, which hold the active-direction
+values, so the swap corrected above propagates into rows 0 to 7.
+
+Each published value is F × s_labeled, where F is the factor for the direction
+and parameter of the row and s_labeled is the value of the seat row with the
+same deck and the same published label. Keeping F and using the seat row that
+holds the direction of the row, s_correct, gives
+
+recomputed = F × s_correct = published × s_correct / s_labeled
+
+so the factor itself need not be known. Every populated value is recomputed this
+way and rounded to four decimals. The ratios s_correct / s_labeled are listed
+below; the seat rows are given as the labeled row, then the correct row:
+
+| Rows | Curve | Seat rows | Slight λ | Slight ζ | Moderate λ | Moderate ζ |
+|---|---|---|---|---|---|---|
+| 0, 1 | active, slab | 26, 28 | 0.4395 | 1.0000 | 0.3423 | 1.0000 |
+| 2, 3 | active, T-beam | 27, 29 | 0.4657 | 0.9872 | 0.3675 | 0.9872 |
+| 4, 5 | passive, slab | 28, 26 | 2.2752 | 1.0000 | 2.9213 | 1.0000 |
+| 6, 7 | passive, T-beam | 29, 27 | 2.1473 | 1.0130 | 2.7209 | 1.0130 |
+
+The ratios are shown to four decimals; the generator uses them unrounded. The
+eight recomputed models are:
+
+| Row | Model ID |
+|---|---|
+| 0 | `CRCB.E123.S1.C0.D.AbAct.Slab.Mod1.CPR22` |
+| 1 | `CRCB.E123.S1.C0.D.AbAct.Slab.Mod23.CPR22` |
+| 2 | `CRCB.E123.S1.C0.D.AbAct.TBeam.Mod1.CPR22` |
+| 3 | `CRCB.E123.S1.C0.D.AbAct.TBeam.Mod23.CPR22` |
+| 4 | `CRCB.E123.S1.C0.D.AbPass.Slab.Mod1.CPR22` |
+| 5 | `CRCB.E123.S1.C0.D.AbPass.Slab.Mod23.CPR22` |
+| 6 | `CRCB.E123.S1.C0.D.AbPass.TBeam.Mod1.CPR22` |
+| 7 | `CRCB.E123.S1.C0.D.AbPass.TBeam.Mod23.CPR22` |
+
+Their published and stored values (each cell gives the published value, then the
+stored value):
+
+| Row | Slight λ | Slight ζ | Moderate λ | Moderate ζ |
+|---|---|---|---|---|
+| 0 | 17.56 → 7.7179 | 0.95 → 0.95 | 78.73 → 26.9499 | 0.95 → 0.95 |
+| 1 | 16.38 → 7.1993 | 1.17 → 1.17 | 95.16 → 32.574 | 1.17 → 1.17 |
+| 2 | 19.62 → 9.1371 | 0.78 → 0.77 | 88.57 → 32.5514 | 0.78 → 0.77 |
+| 3 | 18.3 → 8.5224 | 0.96 → 0.9477 | 107.06 → 39.3468 | 0.96 → 0.9477 |
+| 4 | 7.75 → 17.633 | 1 → 1 | 26.3 → 76.8315 | 1 → 1 |
+| 5 | 8.5 → 19.3394 | 1.19 → 1.19 | 37.22 → 108.7326 | 1.19 → 1.19 |
+| 6 | 9.17 → 19.6906 | 0.81 → 0.8205 | 31.77 → 86.444 | 0.81 → 0.8205 |
+| 7 | 10.06 → 21.6017 | 0.96 → 0.9725 | 44.96 → 122.333 | 0.96 → 0.9725 |
+
+The extensive and complete states of these rows are not defined. The generator
+repeats the calculation of F above from the source rows and stops unless F ×
+s_labeled equals the published slight median of rows 1 and 5, and F × s_correct
+equals the recomputed one, each within 0.01 g.
+
+### Moderate dispersion of the Shao et al. (2022) 10 ft slope-offset curves
+
+Ten rows repeat one Shao et al. (2022) configuration, an abutment on piles with
+a 10 ft approach slab and damage through slope offset, in different groups. The
+source prints a large dispersion of 1.99, as does Shao et al. (2022, Table 6);
+the dataset stores 0.99. Shao et al. (2022, Sec. 4.4) defines the dispersion of
+the integrated model as the square root of the sum of squares of the dispersions
+of its three methods, and for this configuration their Tables 3 to 5 give 0.41,
+0.62, and 0.65, which combine to 0.987. The moderate median, 0.92 g, and the
+slight state are unchanged.
+
+| Data row | Group | Model ID |
+|---|---|---|
+| 17 | E123-S1-C0-D | `CRCB.E123.S1.C0.D.Approach.Pile.Slope10.SXR22` |
+| 41 | E123-S1-C0-S | `CRCB.E123.S1.C0.S.Approach.Pile.Slope10.SXR22` |
+| 63 | E1-S2-C1-D | `CRCB.E1.S2.C1.D.Approach.Pile.Slope10.SXR22` |
+| 103 | E1-S2-C2P-D | `CRCB.E1.S2.C2P.D.Approach.Pile.Slope10.SXR22` |
+| 247 | E1-S3P-C1-D | `CRCB.E1.S3P.C1.D.Approach.Pile.Slope10.SXR22` |
+| 286 | E1-S3P-C2P-D | `CRCB.E1.S3P.C2P.D.Approach.Pile.Slope10.SXR22` |
+| 466 | E1-S2-C1-S | `CRCB.E1.S2.C1.S.Approach.Pile.Slope10.SXR22` |
+| 522 | E1-S2-C2P-S | `CRCB.E1.S2.C2P.S.Approach.Pile.Slope10.SXR22` |
+| 739 | E1-S3P-C1-S | `CRCB.E1.S3P.C1.S.Approach.Pile.Slope10.SXR22` |
+| 795 | E1-S3P-C2P-S | `CRCB.E1.S3P.C2P.S.Approach.Pile.Slope10.SXR22` |
+
+### Filled lower states of nine Unseating curves
+
+Nine Unseating rows define the extensive and complete states only: the source
+studies define unseating capacity for those two states, and the slight and
+moderate cells are empty. The generator sets the slight and moderate median and
+dispersion equal to the extensive ones. The extensive and complete curves are
+unchanged and keep the numbers DS3 and DS4. Note that with this setup, the
+slight and moderate states cannot occur which is in line with the intent of
+model developers. The LS1 and LS2 descriptions of these models say that the
+state is not defined by the source.
+
+The nine models are:
+
+| Row | Model ID |
+|---|---|
+| 24 | `CRCB.E123.S1.C0.S.Unseating.Slab.CPR22` |
+| 25 | `CRCB.E123.S1.C0.S.Unseating.TBeam.CPR22` |
+| 249 | `CRCB.E1.S3P.C1.D.Unseating.BoxGirder.Span5.MF.Disc.Circ.MCP18` |
+| 316 | `CRCB.E2.S3P.C1.D.Unseating.BoxGirder.Span5.MF.Disc.Circ.MCP18` |
+| 400 | `CRCB.E3.S3P.C1.D.Unseating.BoxGirder.Span5.MF.Disc.Circ.MCP18` |
+| 851 | `CRCB.E2.S3P.C2P.S.Unseating.BoxGirder.TCB.Span3.Obl.JML19` |
+| 852 | `CRCB.E2.S3P.C2P.S.Unseating.BoxGirder.TCB.Span3.Flr1.JML19` |
+| 853 | `CRCB.E2.S3P.C2P.S.Unseating.BoxGirder.TCB.Span3.Flr1T.JML19` |
+| 854 | `CRCB.E2.S3P.C2P.S.Unseating.BoxGirder.TCB.Span3.Flr2.JML19` |
+
+The extensive median and dispersion copied to the slight and moderate states:
+
+| Row | Extensive λ | Extensive ζ |
+|---|---|---|
+| 24 | 0.94 | 0.73 |
+| 25 | 0.8 | 0.64 |
+| 249 | 3.87 | 1.12 |
+| 316 | 9.8 | 0.99 |
+| 400 | 11.72 | 1.04 |
+| 851 | 7.79 | 0.9 |
+| 852 | 7.66 | 0.9 |
+| 853 | 7.18 | 0.87 |
+| 854 | 7.28 | 0.85 |
+
+## Omitted rows
+
+The generator omits 101 of the 1,018 rows; the crosswalk lists them with
+`omitted` in its `dataset` column.
+
+### Soleimani (2017) entries that repeat Mangalathu (2017) curves
+
+In three single-column, multi-span seat-abutment groups the source lists six
+Mangalathu (2017) System curves a second time, attributed to Soleimani (2017),
+with the same column shape. All eight parameters of each repeated row equal
+those of its Mangalathu (2017) partner, which the generator checks. The repeated
+rows are omitted, and the Mangalathu (2017) models are kept.
+
+| Omitted | Partner | Shape | Kept model ID |
+|---|---|---|---|
+| 742 | 740 | circular | `CRCB.E1.S3P.C1.S.System.BoxGirder.Circ.M17` |
+| 744 | 741 | rectangular | `CRCB.E1.S3P.C1.S.System.BoxGirder.Rect.M17` |
+| 835 | 833 | circular | `CRCB.E2.S3P.C1.S.System.BoxGirder.Circ.M17` |
+| 837 | 834 | oblong | `CRCB.E2.S3P.C1.S.System.BoxGirder.Obl.M17` |
+| 957 | 955 | circular | `CRCB.E3.S3P.C1.S.System.BoxGirder.Circ.M17` |
+| 959 | 956 | oblong | `CRCB.E3.S3P.C1.S.System.BoxGirder.Obl.M17` |
+
+The shape of row 741 is its corrected label; the source labels it circular.
+
+### Rows without parameters
+
+Ninety-five component rows are N/A in the slight and moderate median and
+dispersion, the only states the source defines for these components (the
+extensive and complete cells are empty). The database uses N/A to mark a
+negligible probability of damage from ground shaking. They hold no curve and are
+omitted; no System row is among them. They are 71 column foundation rotation
+(FndRot) rows, in every group where the component occurs except for one row of
+E1-S2-C2P-S (data row 504, kept as
+`CRCB.E1.S2.C2P.S.FndRot.BoxGirder.MCB.Rect.M17`), and 24 shear key (Key) rows,
+in the Era 2 and Era 3 multi-span seat-abutment groups.
+
+| Component | Group | Rows | Data rows |
+|---|---|---|---|
+| FndRot | E1-S2-C1-D | 2 | 48, 49 |
+| FndRot | E1-S2-C1-S | 2 | 453, 454 |
+| FndRot | E1-S2-C2P-D | 4 | 90, 91, 92, 93 |
+| FndRot | E1-S2-C2P-S | 3 | 501, 502, 503 |
+| FndRot | E1-S3P-C1-D | 2 | 238, 239 |
+| FndRot | E1-S3P-C1-S | 2 | 726, 727 |
+| FndRot | E1-S3P-C2P-D | 4 | 273, 274, 275, 276 |
+| FndRot | E1-S3P-C2P-S | 4 | 774, 775, 776, 777 |
+| FndRot | E2-S2-C1-D | 2 | 122, 123 |
+| FndRot | E2-S2-C1-S | 2 | 546, 547 |
+| FndRot | E2-S2-C2P-D | 4 | 153, 154, 155, 156 |
+| FndRot | E2-S2-C2P-S | 4 | 593, 594, 595, 596 |
+| FndRot | E2-S3P-C1-D | 2 | 305, 306 |
+| FndRot | E2-S3P-C1-S | 2 | 816, 817 |
+| FndRot | E2-S3P-C2P-D | 4 | 356, 357, 358, 359 |
+| FndRot | E2-S3P-C2P-S | 4 | 891, 892, 893, 894 |
+| FndRot | E3-S2-C1-D | 2 | 181, 182 |
+| FndRot | E3-S2-C1-S | 2 | 636, 637 |
+| FndRot | E3-S2-C2P-D | 4 | 209, 210, 211, 212 |
+| FndRot | E3-S2-C2P-S | 4 | 683, 684, 685, 686 |
+| FndRot | E3-S3P-C1-D | 2 | 392, 393 |
+| FndRot | E3-S3P-C1-S | 2 | 938, 939 |
+| FndRot | E3-S3P-C2P-D | 4 | 424, 425, 426, 427 |
+| FndRot | E3-S3P-C2P-S | 4 | 989, 990, 991, 992 |
+| Key | E2-S2-C1-S | 2 | 550, 551 |
+| Key | E2-S2-C2P-S | 4 | 601, 602, 603, 604 |
+| Key | E2-S3P-C1-S | 2 | 820, 821 |
+| Key | E2-S3P-C2P-S | 4 | 899, 900, 901, 902 |
+| Key | E3-S2-C1-S | 2 | 640, 641 |
+| Key | E3-S2-C2P-S | 4 | 691, 692, 693, 694 |
+| Key | E3-S3P-C1-S | 2 | 942, 943 |
+| Key | E3-S3P-C2P-S | 4 | 997, 998, 999, 1000 |
+
+## Soleimani (2017) models
+
+Six Soleimani (2017) System curves remain after the repeated entries are
+omitted. Each is the regular base model of the unbalanced-frame study of that
+thesis (Tables 7.10 and I.5).
+
+| Data row | Model ID |
+|---|---|
+| 743 | `CRCB.E1.S3P.C1.S.System.BoxGirder.Circ.S17` |
+| 745 | `CRCB.E1.S3P.C1.S.System.BoxGirder.Rect.S17` |
+| 836 | `CRCB.E2.S3P.C1.S.System.BoxGirder.Circ.S17` |
+| 838 | `CRCB.E2.S3P.C1.S.System.BoxGirder.Obl.S17` |
+| 958 | `CRCB.E3.S3P.C1.S.System.BoxGirder.Circ.S17` |
+| 960 | `CRCB.E3.S3P.C1.S.System.BoxGirder.Obl.S17` |
+
+The author states in the Acknowledgements of the thesis that its curves "are
+illustrative and should not be used for deployment in ShakeCast, or other risk
+analysis software". Chen et al. (2025) included the curves in the database. This
+library distributes models and is not a risk-analysis application, but its
+models are meant for damage and loss assessments such as pelicun's, which is the
+kind of use the author cautions against. Users must judge for themselves whether
+these curves suit their use before deploying them. The Comments of each of the
+six models carry the same caution.
+
+## Damage-state descriptions
+
+The LimitStates descriptions of the models are written for this library from the
+wording in `../metadata_text.json` (`limit_states`). Their sources are:
+
+- System: the Hazus verbal definitions of the four bridge damage states.
+- Column, Unseating, and the secondary components (abutments, bearings, deck
+  displacement, column foundations, shear keys, and joint seals): Mangalathu
+  (2017) Sec. 6.1 and Table 6.13, Ramanathan (2012) Chapter 5, and Chen et al.
+  (2025) Sec. 3. The two-state unseating capacities of 6 and 9 in. come from the
+  EDP-file row for groups E1-S2-C2P-S, E1-S3P-C2P-S, E123-S1-C0-D, and
+  E123-S1-C0-S, which cites Mangalathu, Soleimani, and Jeon (2017) and
+  Cavalcante et al. (2022).
+- Approach: Shao et al. (2022) Table 2.
+- Cavalcante et al. (2022) bearings: Cavalcante et al. (2022) Table 8.
+
+The median capacities that the descriptions quote come from `EDP_Fragility
+Database.csv` in `../../California RC Bridges 2025 EDP/source/`, where they are
+expressed in the EDP of each component (curvature ductility, column drift in
+percent, displacement or settlement in inches, or rotation in radians). The
+generator keeps them in its `CAPACITIES` table and stops unless every quoted
+value equals the value of the matching row of the EDP file; its function
+`edp_row_key` documents how rows are matched by component, EDP, and the era or
+group in `Notes`. Some values of the EDP file are deliberately not quoted:
+
+- The moderate joint seal description quotes no capacity. The EDP file gives 5
+  in., which the published seal curves do not reproduce (see the seal sentence
+  in the model Comments).
+- The Era 3 row of unseating capacities is not quoted for the Era 3 Mangalathu
+  (2017) curves. Their descriptions quote the Era 2 set, because those curves
+  equal the Era 2 curves of the thesis (Appendix C).
+- The two-state unseating models (the filled rows above) quote only the
+  extensive and complete capacities of their row.
+- No description quotes the `Dispersion` column.
+
+The capacities are text only; no fragility parameter is derived from or changed
+by the EDP file.