FAQ — frequently asked questions¶

General¶

Is RSL III free?¶

RSL III is part of the GeoStru NX suite with a freemium plan. For up-to-date details visit geostru.ai or write to info@geostru.ai.

Do I have to install anything?¶

No. RSL III is a web app that opens from nx.geostru.ai/rsl/ in any modern browser.

What does "equivalent linear" mean?¶

It is a site response analysis method that approximates the non-linear soil behaviour with successive iterations of linear analyses. See metodo.md for details.

Stratigraphy¶

Where do I get Vs values?¶

Direct measurement with seismic tests:

MASW (Multichannel Analysis of Surface Waves) — the most common, from the surface with aligned geophones
ReMi (Refraction Microtremor) — from ambient seismic noise
Cross-hole / Down-hole — paired or single boreholes with seismograph (more accurate)
HVSR (passive microtremor seismic, Nakamura 1989) — only to estimate the site fundamental frequency, not point-wise Vs(z)
Seismic Cone Penetration Test (SCPT) — continuous Vs along a CPTU

If you don't have direct measurements, do not estimate Vs from literature or N-SPT — it is too imprecise for RSL.

How deep must the stratigraphy be?¶

Down to the seismic reference bedrock (Vs ≥ 800 m/s).

In Italy typically:

Po Plain / coastal sedimentary: 50-300 m
Intermontane basins (Umbria, Marche, Abruzzo): 30-100 m
Mountain / cliff areas: 5-30 m

If you don't reach Vs > 800 m/s with the boreholes, extrapolate with constant or linearly increasing Vs to an assumed depth (document this in the technical report).

Can I use SPT instead of Vs?¶

No, not directly. SPT measures penetration resistance, not shear-wave velocity. There are empirical correlations (Ohta-Goto 1978, Ohsaki- Iwasaki 1973, Pitilakis 1999) but they are very uncertain (typical error 30-50%).

For liquefaction SPT is fine (see LiquiTer). For RSL you need direct seismics.

How do I choose the degradation curve?¶

Depends on the soil:

Sands → Seed-Idriss (1970), Darendeli (2001)
Clays → Vucetic-Dobry (1991), as a function of PI
Gravels → specific curves (limited library, rarely available)
Anthropic fills → sand curves with increased ξ_min
Soft marine soils → Vucetic-Dobry for medium PI + smaller γ_min

When in doubt, try 2 different curves (e.g. Seed-Idriss + Darendeli) and compare the results. If FA/FH factors change little (< 10%), the choice is not critical.

Input accelerogram¶

How many accelerograms do I need?¶

For NTC 2018 / ICMS 2008-2018 compliant studies: at least 7 spectrum-compatible accelerograms (NTC par. 7.3.5). RSL III in multi-input mode processes them all and produces the mean spectrum.

For sensitivity or educational studies: 1-3 accelerograms are enough.

Where do I find real accelerograms?¶

PEER NGA-West (USA, but also European events): ngawest2.berkeley.edu
ITACA (Italian, INGV): itaca.mi.ingv.it
ESM (European Strong-Motion Database): esm.mi.ingv.it
REXEL (Iervolino-Galasso, University Federico II Napoli): public platform for spectrum-compatible accelerogram selection — NTC 2018 reference standard
GeoStru Spectra (geostru.ai): GeoStru suite for spectrum-compatible accelerogram selection

How do I do "spectral matching" / scaling?¶

RSL III does not do automatic scaling. You must pre-process the accelerograms with dedicated software. We recommend:

GeoStru Spectra (geostru.ai) — produces NTC-spectrum-compatible series ready to import into RSL III
REXEL (Iervolino-Galasso, University Federico II Napoli) — public free academic platform, NTC 2018 reference

Then import the already-matched accelerograms into RSL III.

Outcrop or within?¶

See dati-input.md#outcrop-vs-within. For archive recordings (PEER, ITACA): outcrop is the right choice in 99% of cases.

Calculation¶

How many iterations are needed?¶

Typically 4-8 for standard accelerograms. If the calculation requires >15 iterations and does not converge, it usually means:

Soil with γ_max > 1% (strongly non-linear regime → EQL method not reliable, use non-linear SRA)
Anomalous degradation curve
Too-fine discretisation (reduce f_max)

What to do if the calculation does not converge?¶

Raise the tolerance to 1% (default 0.5%)
Raise max iterations to 50
Check that the computed γ_max is < 1% in all layers
If γ_max > 1% in any layer → the site is in a significantly non-linear regime, RSL III is not the right tool

γ_eff = 0.65 · γ_max — why 65%?¶

It is an empirical value calibrated by Idriss (1968) on experimental tests. It represents the "average equivalent strain" during the cyclic history. For seismic histories with a single dominant pulsation it can be 0.5-0.55, for long-tail histories 0.7-0.85. RSL III uses 0.65 by default; you can change it under Advanced.

Output¶

FA, FH, FT — which one for my project?¶

See icms.md for the full table.

In short:

Low-rise buildings (1-3 storeys): FA
Medium-rise buildings (4-7 storeys): FH
Tall buildings (10+ storeys): the RSL spectrum directly (FT is only a summary)

The RSL spectrum is below the NTC one, is that OK?¶

Yes, it can happen. It means your site amplifies less than the standard NTC spectrum predicts for the assumed ground category. Consequence: you can design less conservatively but only if your RSL analysis is well supported by:

Measured Vs (not estimated)
7+ spectrum-compatible accelerograms
ICMS 2008/2018 compliance

Otherwise use the standard NTC spectrum.

Post-seismic settlements?¶

RSL III does not compute post-seismic settlements from liquefaction. If your site has liquefiable layers, after the RSL analysis use LiquiTer NX for liquefaction + settlements.

My question isn't here¶

Contact us and we'll add it.