VC-16 — EN 1998-3 / NTC Confidence-Factor Chain vs Published Practice

Modules: vernastruct.codes.knowledge + KL/CF plumbing in modules.walls (both NDS and EC6 paths), with the new irregular_stone library entry. Code basis: EN 1998-3 §3.3 (Table 3.1) / Italian NTC 2018 + Circolare 21/01/2019 n.7 §C8.5: knowledge levels KL1/KL2/KL3 (Italian LC1/LC2/LC3) → confidence factors CF = 1.35 / 1.20 / 1.00 dividing material strengths; elastic moduli are adopted at their mean values. Cross-check source (task 2.2): published peer-reviewed case study — "Comparative Seismic Assessment Methods for Masonry Building Aggregates", Frontiers in Built Environment 5:123 (2019), open access: https://www.frontiersin.org/journals/built-environment/articles/10.3389/fbuil.2019.00123/full Status: ✅ PASS — library values, CF values, and the divide-strengths- only convention all match the published application exactly.

1. Why this case matters

The KL/CF system is one of the product's two orthogonal assessment axes (damage factor = physical condition; CF = confidence in material knowledge). VC-10/11 verified that CF divides strength in each engine; this case verifies the whole convention against how practitioners actually publish it in Italian-school seismic assessment — the exact market (ISCARSAH / SAHC-trained conservation engineers) the roadmap targets.

2. Published application

The case study assesses an existing stone-masonry building aggregate to NTC18 at the lowest knowledge level:

Published item	Value
Knowledge level	LC1 (= EN 1998-3 KL1)
Confidence factor	1.35
Masonry typology	irregular stone masonry (NTC18)
Mean compressive strength f_m	1.0 MPa
Shear strength τ₀	0.018 MPa
Elastic modulus E	870 MPa (mean) — not divided by CF
Convention quoted	"the mean values of the resistances … divided by FC = 1.35; the average of the elastic modules have been adopted"

These property values are the Circolare 2019 Table C8.5.I row for muratura in pietrame disordinata (f_m 1.0–2.0, τ₀ 0.018–0.032, E 690–1050 MPa, w 19 kN/m³) applied with the standard LC1 convention: minimum strengths of the range + mean modulus.

3. What VernaStruct reproduces (pinned tests)

New library entry irregular_stone carries exactly those values (f_m 1.0, τ₀ 0.018, E 870, ρ 1900 — ranges + convention documented in the YAML). Through the walls service:

Check	Result
CF values	KL1/KL2/KL3 → 1.35/1.20/1.00 (EN 1998-3 Table 3.1) — exact
NDS path, KL1 vs KL3	axial, shear and bending capacities all scale by exactly 1.35; effective strengths = f/1.35 (= 0.7407 / 0.01333 MPa), matching the published chain
EC6 path, KL1 vs KL3	Φ (Annex G) identical between knowledge levels — λ_t keeps the unreduced stiffness ratio (mean E adopted, as published); f_d and N_Rd scale by exactly 1.35 (f_d = 1.0/2.2 at KL3, /1.35 at KL1)
Report	KL1 warning names the CF and cites EN 1998-3 §3.3 / Circolare 2019

Pinned as tests/test_codes.py::test_vc16_* (4 tests).

4. Notes

This case intentionally verifies a convention, not a single number: the failure mode it guards against is someone "improving" the code by dividing E by CF too (plausible-looking, wrong per both EN 1998-3 practice and the published study) or baking CF into the YAML values (double-counting).
The damage factor remains orthogonal: test_damage_and_kl_stack_orthogonally (VC-11 companion) covers the combined case.
The irregular_stone entry is also a genuine product addition — rubble stone is a core heritage typology (Arabian Peninsula, Mediterranean) that the library previously lacked.