Geotechnical Engineering — Good-to-Know Rules of Thumb
Practitioner judgment for geotechnical engineers, engineering geologists, and exam candidates. Heuristics, not code — verify against the governing standard and the site-specific data.
Site investigation
- Boring spacing: 100–200 ft for buildings, closer for erratic geology; depth ≥ 1.5× the foundation width below the bearing level, or to competent strata. One boring is an anecdote; two are a disagreement; three start to be data.
- Never design off a single data point: the most expensive geotechnical failures come from interpolating between too few borings. If the geology is variable, the investigation budget is the cheapest insurance on the project.
- SPT N ≈ relative density: N < 4 very loose, 4–10 loose, 10–30 medium dense, 30–50 dense, > 50 very dense. In clays, N ≈ 2–4 soft, 4–8 medium, 8–15 stiff, 15–30 very stiff, > 30 hard — then confirm with lab tests.
- Water table discipline: measure it in every boring, note the date and season, and assume it rises in the wet season. Half of all "unexpected" geotechnical problems are water the investigation found and the design ignored.
- CPT over SPT for stratigraphy: when the budget allows one premium method, CPT's continuous profile beats SPT's 5-ft anecdotes — especially for liquefaction analysis and soft-soil delineation.
Soil mechanics sanity checks
- Effective stress first: σ' = σ − u on every calculation, every time. If your answer doesn't change when the water table moves, you forgot the water.
- Friction angles: clean sand 28–34°, gravel 34–40°, silt 26–30°, normally consolidated clay φ' ≈ 25–30° with c' ≈ 0. If your lab says φ' = 45° for a sand, question the lab.
- Undrained strength: su ≈ 0.2–0.3 σ'v for normally consolidated clays (the SHANSEP ratio). A quick check on every UU result — lab su values far from this ratio deserve scrutiny.
- Consolidation: 90% consolidation at Tv = 0.848; doubling the drainage path quadruples the time (t ∝ Hdr²). Wick drains work by cutting Hdr, not by changing the soil.
- Permeability ranges: gravel 10⁻¹–10⁻³ cm/s, sand 10⁻³–10⁻⁵, silt 10⁻⁵–10⁻⁷, clay < 10⁻⁷. Seven orders of magnitude — get the order right before the digit.
Foundations
- Bearing capacity FS = 3 on qult is the traditional allowable-stress comfort zone; LRFD resistance factors (0.45–0.55 for shallow foundations) encode the same caution differently.
- Settlement controls: bearing capacity rarely governs building foundations — settlement does. Run the settlement check before celebrating the bearing number.
- Differential settlement budget: ½ in. for ordinary structures, ¾ in. tolerable for flexible frames, near-zero for precision facilities. Angular distortion > 1/300 starts cracking finishes; > 1/150 starts structural distress.
- Pile tip: end bearing in dense stratum is worth multiples of shaft friction in soft soil — get to bearing if it's reachable. A pile that "floats" in soft clay is a settlement machine.
- Downdrag: any fill placed after pile installation, or any consolidating layer the pile passes through, adds drag load — size the pile structurally for it, don't just reduce capacity.
- Expansive soil: if PI > 25 and clay fraction is high, assume swell pressure matters — design the foundation for it (post-tensioned slab, drilled piers with void boxes) rather than hoping it won't swell.
Slopes and retaining
- FS targets: 1.5 static, 1.1–1.2 pseudo-static/seismic, 1.25–1.3 for rapid drawdown on dams. FS = 1.0 is failure — 1.2 is not "safe," it's "conditionally stable with good data."
- Water is the enemy: most slope failures involve water the design didn't account for — raise the water table in your analysis and watch the FS collapse. Drainage is the cheapest stabilization there is.
- Infinite slope for shallow slides: translational failures on long uniform slopes — the quick check before the method-of-slices production run.
- Retaining walls: Ka ≈ 0.3 for φ = 30° (active), Kp ≈ 3.0 (passive) — the 10:1 ratio is why passive resistance is precious and active pressure is relentless. Never count full passive in front of a wall that might get excavated later.
- MSE walls: the reinforced mass IS the retaining structure — check internal (pullout, rupture), external (sliding, overturning, bearing), and global stability separately. Global failure through the reinforced zone is the one most often missed.
Earthquake geotechnics
- Liquefaction screening: saturated + loose (N1,60 < ~20–30) + sand/silt + within 50 ft of the surface + strong shaking = suspect until proven otherwise. Run the triggering analysis; don't hand-wave it.
- amax rules everything: liquefaction, slope displacement, and lateral earth pressure increments all scale with peak ground acceleration — get the site-specific amax right before refining anything else.
- Lateral spreading: needs a free face or gentle slope toward water — flat inland sites settle; waterfront sites spread. The consequence analysis differs completely.
- Post-liquefaction bearing: assume the liquefied layer contributes zero bearing and zero lateral support — design foundations to punch through it or improve it.
- Ground improvement selection: densification (vibro, stone columns) for clean sands; deep soil mixing or drains for silty soils; deep foundations when improvement is uneconomical. Match the method to the fines content.
Construction and field judgment
- The observational method: the design is a hypothesis; the excavation is the experiment. Have contingency plans (Peck's observational method) for the ground being worse than modeled — because sometimes it is.
- Compaction QC: 90% relative compaction for fills under structures is the floor, not the target; test every lift, and don't let the contractor "make it up on the next lift."
- Changed conditions: differing site conditions clauses exist because the ground always surprises — document everything, photograph everything, and notify early. The claim you win is the one with the daily logs.
- Dewatering: drawdown causes settlement — a dewatering system that saves the excavation can settle the neighbor's building. Monitor adjacent structures, and recharge if needed.
Professional judgment
- The geotechnical report is a risk document: state what you know, what you inferred, and what could still surprise — in writing. The report that hides uncertainty is the one that gets read in court.
- Talk to the driller: the person running the rig knows things the logs don't capture — drilling behavior, water strikes, heave. Buy the coffee and ask.
- Coordinate with the structural engineer early: foundation type, settlement tolerance, and lateral system are joint decisions — the cheapest foundation is the one designed with the structure, not after it.
- Know your lane: geologists characterize hazards, engineers design mitigations — in California the CEG/GE boundary is statutory. Respect it; the collaboration is the point.
- When the ground scares you, say so: recommending against a site, or demanding more investigation, is the hardest and most valuable thing a geo-professional does.
File: geotechnical/good-to-know-rules-of-thumb.md — heuristics for study and practice; always verify against the governing standard and site data.
Original ETS reference writing, verified against public sources on 2026-09-30. Thin sourcing is flagged inline. Not legal advice — confirm current editions with the issuing body.