Short verdictUse of net stress (Qnet = Qgross − overburden removed) is the correct starting point for elastic, increment‑based settlement calculations in frictional soils, but it must be applied with care: Qnet alone does not guarantee zero settlement because of stress‑path effects, hysteresis, redistribution and site specifics.
Why Qnet is the correct starting point- Elastic incremental formulations depend on stress increments (Δp), not absolute gross loads; standard methods (Schmertmann, CPT‑E, influence‑factor summation) explicitly use Δp (net pressure) for settlement prediction.
- If excavation fully removes the pre‑existing overburden and the structural contact pressure exactly restores the original pre‑excavation effective stress at every point beneath the footing, elastic theory predicts no further elastic compression at those points (no positive Δp).
Why Qnet alone can be misleading in practice- Unloading is not perfectly reversible: large decompression can cause irreversible fabric change and loss of contact stiffness; reloading may produce plastic strains even if incremental Δp = 0.
- Redistribution and load spreading: a raft’s gross load spreads with depth; parts of the soil volume will experience non‑zero Δp even if a single shallow point shows Qnet ≈ 0. Settlement is an integral volume response, not a pointwise equality.
- Construction and transient loads: temporary cranes, stockpiles, traffic and vibrations introduce additional positive Δp during works even when design Qnet = 0.
- Cohesive or partially saturated layers: undrained response, excess pore pressures and consolidation mean net effective stress accounting alone does not capture time‑dependent settlements.
- Minimum confinement and arching: shallow post‑cut zones retain some confinement from arching or face support; assuming zero residual confinement is unrealistic and p′min or Emin must be chosen.
- Empirical/parameter issues: Δp‑based methods require representative moduli Es or G that depend on strain level and stress history; using small‑strain E without adjustments can mislead.
How to apply Qnet correctly and safely (practical recipe)- Use Qnet as the incremental input in elastic/CPT methods (Schmertmann, CPT‑E, influence‑factor summation), but distribute load with depth using influence factors Iz and layer discretisation rather than applying Qnet at a single shallow depth.
- Account for unloading/recompression: reduce reload modulus or adopt a recompression law when excavation produces significant decompression; treat reload stiffness separately from virgin stiffness and calibrate to U–R oedometer or plate tests.
- Include a disturbance factor φ and/or minimum confinement p′min (e.g., 5–20 kPa) to prevent unrealistically low stiffness near faces; run sensitivity cases (φ = 1.0, 0.7, 0.5).
- Check elastic settlement sums from Δp together with potential plastic/rearrangement components where the stress path crosses yield boundaries (use OCR, CPT or lab data to estimate plastic parts).
- For profiles with cohesive horizons or low permeability, complement immediate elastic checks with consolidation analysis and staged pore‑pressure dissipation modelling.
- Validate predictions with at least one field test (plate load or instrumented trial) after staged excavation and adjust design moduli accordingly.
Final synthesisQnet is the rightful elastic driver in standard settlement procedures and is widely used. However, in an excavation → construction sequence you must not treat Qnet = 0 as proof of zero settlement. Excavation alters the soil state and stiffness, introduces redistribution and path‑dependence, and can produce irreversible effects not captured by net‑stress algebra. Use Qnet together with influence‑factor depth distribution, recompression adjustments, disturbance factors and validation testing to obtain safe, realistic settlement predictions.
Reference hintSchmertmann method and influence‑factor settlement formulations, CPT‑E procedures, and standard elastic influence‑factor summations (consult original Schmertmann, Mayne / Poulos literature).
