[andreachiocca]

Preview (once the preview check succeeds):

• FIG. R04.1, chapter 03: https://pr-59.academic-website-preview.pages.dev/research/#rosenthal-map
• FIG. R05.1, chapter 04: https://pr-59.academic-website-preview.pages.dev/research/#spr-fatigue
• Italian: https://pr-59.academic-website-preview.pages.dev/it/research/

FIG. R04.1: Rosenthal map of metal AM (chapter 03)

The figure plots 254 literature operating conditions for SLM and LMD in the dimensionless $\tilde U$–$\tilde P$ plane of Moda, Chiocca et al., Materials & Design 223 (2022) 110991. It is the analogue of the article's Fig. 5, and the data come from the updated workbook (revision 5).

What it shows:

• point colour gives the alloy family, point shape gives the process;
• dashed lines are aspect-ratio isolines (Eq. 12);
• the white curve is the boundary $R=\bar R$ (Eq. 22), with the feasible region shaded above it;
• chips filter by family and process.

Interaction. Selecting a point draws its melt pool seen from above. The outline is the exact isotherm of Eq. (2), written in closed polar form as $r=W_0(Ry(1+\cos\theta))/(1+\cos\theta)$. The panel also lists $U$, $P$, $\bar R$, $Ry$, $A_r$ and the source, with its DOI where the workbook has one.

Generator. scripts/generate_rosenthal_map.py reads sheet Parametri:

• it recomputes $\tilde U$ and $\tilde P$ from the raw columns (Eqs. 20–21) and fails if they disagree with the workbook;
• it keeps the workbook's own selection, i.e. the rows that have U* and P*;
• it leaves out EBM, as the article does.

Dropped rows are counted in the JSON: 3 without a process, 24 EBM, 6 incomplete, 55 outside the workbook's selection.

Tests. tests/test_rosenthal_map.py checks, in pure Python:

• $Ry=\tilde P\tilde U$;
• Eqs. (6) and (12) against the exact isotherm;
• the tips $X^\pm$ (Eqs. 9, 10);
• that every outline point satisfies Eq. (2);
• the boundary and isolines;
• the chapter binding and the i18n keys.

Note on the cited accuracies. The maximum deviation of Eq. (12) from the exact isotherm is 2.009 % (at $Ry\approx0.32$), and that of Eq. (6) is 0.7265 % (at $Ry\approx1.5$). The values quoted from Wang et al. are 1.994 % and 0.7236 %. Both were cross-checked with a 30-digit mpmath solution. The test bounds are set accordingly.

FIG. R05.1: hybrid SPR joints in fatigue (chapter 04)

The figure shows 45 fatigue tests at $R=0.1$ on Al 5005 / Fe P05 self-piercing-riveted joints, with and without Terokal 4497 adhesive. The data come from the outputs of the manuscript in preparation; the tests were run within the PRIN project.

Three views:

• force amplitude: lap-shear and coach-peel joints form two separate populations, about 10× apart, so no curve is drawn;
• Rupp structural stress, rivet only: median S–N curve with its ±2s band, band ratio ×2.33;
• Rupp structural stress, rivet + adhesive: band ratio ×2.04.

Points animate between the views. Colour gives the geometry, fill marks the adhesive and shape gives the thickness stack.

Generator and tests. scripts/generate_spr_fatigue.py only collects fatigue_assessment_output, and checks that the stored S–N intercept, slope and SD are the least-squares fit of the same points. tests/test_spr_fatigue.py repeats those checks.

One raw-data inconsistency, reported and not corrected. In Data/FATICA/LAP/2_15_NO_GLUE/2_15_NO.txt, row 5 reads $F_{max}=2860$, $F_{min}=286$, $F_a=1400$, $F_m=1573$ N. $F_a$ and the stress column are consistent with $F_{max}\approx3111$ N, while $F_{max}$, $F_{min}$ and $F_m$ give $F_a=1287$ N. The article's structural stress uses 1400 N, and the figure keeps it until the author resolves the row. The test lists it as the only known exception.

Other changes

• Both figures sit in their chapter through figures in _data/research_areas.json.
• CLAUDE.md and the roadmap are updated. Neither figure has a figure_sources.yml pin: their inputs are files from the author, not a repository.

Verification (local)

• unittest, the Ruby checks, the production build and all page verifiers pass.
• Browser checks match the master baseline: the only errors come from the CDNs blocked in this environment.
• Screenshots checked at 1280 and 375 px, EN and IT, with no horizontal overflow.

---

Generated by Claude Code