Foundations, Seepage and Hydraulic–Geotechnical Interaction

Introduction and context

Hydraulic structures transfer water, soil stresses and seepage forces through a foundation system. The safety of a dam, weir, barrage, canal head regulator, levee or cofferdam therefore depends not only on structural strength but also on the interaction between hydraulic gradients, pore-water pressures, soil permeability and effective stress. Many failures attributed to “structural” causes are initiated by foundation seepage, concentrated leakage, piping, inadequate drainage, uplift or loss of shear strength.

This chapter develops methods for analysing seepage, estimating uplift and exit gradients, assessing internal erosion and piping, and selecting cut-offs, grout curtains, filters, drains and dewatering systems. The treatment is intended for postgraduate civil-engineering practice: simplified analytical methods are used first, but their assumptions, limitations and links with numerical modelling and field verification are made explicit.

Learning objectives

By the end of this chapter, the reader should be able to:

1. formulate seepage problems using Darcy’s law, flownets and flow nets;

2. calculate pore pressure, seepage force, uplift and exit gradients;

3. distinguish piping, suffusion, backward erosion and concentrated leakage;

4. assess internal stability using critical-gradient and filter criteria;

5. select and proportion cut-offs, grout curtains, drainage layers and relief wells;

6. evaluate bearing, sliding, slope stability, basal heave and dewatering at a preliminary design level;

7. integrate geotechnical, hydraulic, construction and monitoring requirements in a foundation design.

1. Groundwater flow and effective stress

For saturated soil, total stress is divided into effective stress and pore-water pressure:

\[

\sigma' = \sigma-u

\]

where \(\sigma'\) is effective normal stress, \(\sigma\) is total normal stress and \(u\) is pore-water pressure. Soil shear strength is commonly represented by the effective-stress Mohr–Coulomb relation

\[

\tau_f=c'+\sigma'\tan\phi'

\]

where \(c'\) and \(\phi'\) are effective cohesion and friction angle. Seepage changes \(u\), and therefore changes both bearing resistance and shear strength.

Darcy’s law for saturated, laminar flow is

\[

q=k i A

\]

where \(q\) is discharge, \(k\) is hydraulic conductivity, \(i=\Delta h/L\) is hydraulic gradient, and \(A\) is the area normal to flow. The sign convention is often written \(\mathbf q=-k\nabla h\), indicating flow toward decreasing total head \(h=z+p/\gamma_w\). Darcy’s law is appropriate when flow is sufficiently slow for viscous forces to dominate; it is not automatically valid for coarse gravel at very high velocities or for turbulent flow.

Hydraulic conductivity is directional in stratified or fissured ground. For horizontal flow through layers of thickness \(H_j\), the equivalent conductivity is

\[

k_h=\frac{\sum k_jH_j}{\sum H_j}

\]

For vertical flow through the same layers,

\[

k_v=\frac{\sum H_j}{\sum(H_j/k_j)}.

\]

Thus a thin, low-permeability layer may have a modest effect on horizontal flow but dominate vertical seepage resistance. Anisotropy can be transformed approximately by replacing coordinates with \(x'=x\sqrt{k_v/k_h}\), provided the medium is homogeneous, saturated and described by constant principal conductivities.

2. Seepage analysis and flownets

A flownet is composed of flow lines and equipotential lines. In a homogeneous isotropic soil, they intersect approximately at right angles. For a two-dimensional flownet with \(N_f\) flow channels, \(N_d\) equipotential drops and total head loss \(H\), discharge per unit width is

\[

q=kH\frac{N_f}{N_d}.

\]

The equation follows by considering one curvilinear square: each channel carries approximately \(k(H/N_d)\) times its width-to-length ratio, which is near unity for a properly constructed net. The relation is a graphical approximation, not a universal formula for arbitrary distorted meshes, anisotropic ground or three-dimensional flow.

The head loss per equipotential interval is \(\Delta h=H/N_d\). Pore pressure at a point is obtained from its elevation and pressure head:

\[

u=\gamma_w(h-z).

\]

For a point located at elevation \(z=4\,m\), with total head \(h=9\,m\), taking \(\gamma_w=9.81\,kN/m^3\), \(u=9.81(9-4)=49.05\,kPa\).

Worked example 1: flownet discharge and pressure

A two-dimensional flownet beneath a weir has \(k=2\times10^{-5}\,m/s\), upstream-to-downstream head difference \(H=6\,m\), \(N_f=4\), and \(N_d=12\). Then

\[

q=(2\times10^{-5})(6)(4/12)=4.0\times10^{-5}\,m^3/(s\,m).

\]

If a point lies at elevation \(z=2\,m\) and its total head is 8.5 m, its pressure is \(9.81(6.5)=63.8\,kPa\). The result should be checked against the actual boundary heads and the assumed datum.

3. Uplift, seepage force and exit gradient

The uplift force beneath a floor is the integral of pore pressure over the base:

\[

U=\int_0^B u(x)\,dx

\]

per unit width, where \(B\) is the floor length. If the pressure distribution is approximated by a trapezoid with end pressures \(u_1\) and \(u_2\),

\[

U=B\frac{u_1+u_2}{2}.

\]

This approximation must not be used blindly where drains, cracks, layered foundations or three-dimensional effects cause a nonlinear pressure distribution. Drain efficiency is often represented during preliminary design by a specified reduction in downstream uplift, but the assumed reduction requires verification by piezometers.

Seepage force per unit volume is \(i\gamma_w\), acting in the direction of seepage. It is not a separate “water pressure” to be added twice; it is the body-force representation of the hydraulic gradient in an effective-stress analysis.

At a downstream exit, the upward gradient is critical when the submerged unit weight is balanced by seepage force:

\[

i_c=\frac{\gamma'}{\gamma_w}=\frac{G_s-1}{1+e}

\]

for saturated soil, where \(G_s\) is specific gravity of solids and \(e\) is void ratio. For \(G_s=2.65\) and \(e=0.65\),

\[

i_c=\frac{2.65-1}{1+0.65}=1.00.

\]

A factor of safety against heave based on gradient is commonly written \(FS_i=i_c/i_{exit}\). This criterion addresses uplift instability at the exit; it does not by itself prove resistance to internal erosion.

4. Internal erosion, piping and filters

Internal erosion is particle detachment and transport within a soil mass. Important modes are:

• Backward erosion: erosion progresses upstream from an unfiltered downstream exit.
• Concentrated leakage: water flows through a crack, poorly compacted zone, animal burrow or interface.
• Suffusion: finer particles migrate through the constraining framework of coarser particles without gross distortion.
• Contact erosion: flow along the boundary between soils of contrasting grading removes particles from the finer soil.

Piping is often used broadly for progressive internal erosion, but a high exit gradient is only one possible trigger. Assessment should consider particle grading, density, fabric, cracks, interfaces, dispersivity, hydraulic loading history and the presence of a properly designed filter.

A granular filter must satisfy two conflicting functions: retain the protected soil and remain sufficiently permeable to pass water without unacceptable pressure. Widely used empirical criteria include a retention condition such as

\[

D_{15,f}\leq 4\text{ to }5D_{85,b}

\]

and a permeability condition such as

\[

D_{15,f}\geq 4D_{15,b},

\]

where subscripts \(f\) and \(b\) denote filter and base soil. These are screening rules, not universal laws. Gap-graded, broadly graded, dispersive or internally unstable soils require specialist criteria and often laboratory testing. Filter thickness must also permit continuity, construction tolerance and resistance to segregation; a nominal particle-size criterion alone is inadequate.

Geotextile filters require assessment of apparent opening size, permittivity, in-plane transmissivity, clogging, puncture, creep and compatibility with the soil. A geotextile should not be selected solely from a catalogue opening size.

5. Control measures: cut-offs, grout curtains and drainage

Cut-offs

A cut-off lengthens the seepage path and reduces the hydraulic gradient. Options include a concrete diaphragm wall, sheet piles, slurry trench, secant piles, grout-filled trench or compacted low-permeability core. A cut-off should penetrate a competent low-permeability stratum where practicable. Partial penetration may redirect flow beneath the tip and create a concentrated exit gradient.

For a simple one-dimensional vertical cut-off, the seepage resistance is approximately proportional to \(L/k\), where \(L\) is flow length. This is only a conceptual measure: real seepage redistributes around the cut-off, and a two-dimensional flownet or finite-element model is required for final assessment.

Grout curtains

Grout curtains reduce seepage through fractured rock. Design requires investigation of joint orientation, aperture, connectivity, permeability and groutability. Curtain depth and spacing are selected from water-pressure tests, exploratory holes and staged grouting results. Grout pressure must avoid hydraulic fracturing, uplift of overburden and uncontrolled grout migration. Verification includes post-grouting water-pressure tests, grout take interpretation and piezometric response.

Drainage and relief systems

Drainage blankets, toe drains, chimney drains, drainage galleries and relief wells lower pore pressure and provide controlled outlets. A drain must have adequate hydraulic capacity, filter compatibility, access for inspection and redundancy. For a relief well, the preliminary discharge estimate is often based on Darcy flow through the screened interval, but well efficiency, skin effects, partial penetration and drawdown interference must be included in detailed design.

6. Foundation stability: postgraduate design framework

Bearing capacity

For a shallow foundation, a conventional drained bearing-capacity expression is

\[

q_{ult}=c'N_c s_c d_c i_c+qN_qs_qd_qi_q+0.5\gamma'BN_\gamma s_\gamma d_\gamma i_\gamma.

\]

Here \(q\) is effective surcharge at foundation level, \(B\) is width, \(\gamma'\) is effective unit weight, \(N_c,N_q,N_\gamma\) are bearing factors, and \(s,d,i\) account for shape, depth and load inclination. The exact factors depend on the adopted theory and friction angle. For saturated clay under rapid loading, an undrained analysis using \(s_u\) rather than drained parameters may be appropriate. Eccentricity, uplift, foundation inclination and seepage-modified pore pressure must be treated explicitly.

A practical design should compare: (a) ultimate capacity with partial factors, (b) allowable settlement, (c) sliding and overturning, and (d) hydraulic heave. A high bearing-capacity factor of safety does not compensate for excessive differential settlement or piping.

Sliding and overturning

For a gravity hydraulic structure, the sliding resistance may include base friction and allowable cohesion, but cohesion should be used cautiously unless the interface is demonstrably sound. A simplified check is

\[

FS_s=\frac{\mu(V-U)+P_p}{H}

\]

where \(V\\) is downward load, \(U\) uplift, \(\mu\) interface friction coefficient, \(P_p\) reliable passive resistance and \(H\) total horizontal driving force. Passive resistance should generally be reduced or omitted if excavation, scour or future access could remove it. Overturning is checked using moments about the toe, including uplift distribution and eccentricity. Sliding, bearing and global stability should be assessed with consistent load factors.

Slope stability

For a soil slope adjacent to a reservoir or canal, use effective-stress analyses for long-term conditions and total-stress analyses only where the loading rate and drainage conditions justify them. Bishop’s simplified method is suitable for many circular-slip assessments; Spencer or Morgenstern–Price methods are preferable where force and moment equilibrium and noncircular mechanisms matter. Reservoir drawdown can leave high pore pressure while external water support falls, producing a critical transient condition. Seepage forces and anisotropic permeability should be represented in the pore-pressure model.

Basal heave

Excavations in soft clay may fail by basal heave when upward undrained shear resistance is insufficient to balance excavation-induced stress relief and hydraulic pressure. A preliminary stability check compares available undrained resistance around the excavation base with the net driving stress, using a method appropriate to geometry and clay strength profile. For deep excavations, calculate stability using a recognised limit-equilibrium or finite-element method, include construction stages, wall stiffness and strut loads, and consider artesian pressure separately. Lowering groundwater outside the excavation can cause settlement and may destabilise neighbouring foundations.

7. Dewatering design

Dewatering is a temporary hydraulic structure. Its design objectives are safe excavation, acceptable drawdown, control of piping and protection of adjacent assets.

For an unconfined aquifer, a preliminary steady radial-flow estimate to a well is

\[

Q=\frac{\pi k(h_o^2-h_w^2)}{\ln(R/r_w)},

\]

where \(h_o\) and \(h_w\) are saturated thicknesses at radius \(R\) and well radius \(r_w\), respectively. For a confined aquifer,

\[

Q=\frac{2\pi k b(s)}{\ln(R/r_w)},

\]

where \(b\) is aquifer thickness and \(s\) is drawdown at the well. These equations assume homogeneous, isotropic, extensive aquifers, steady flow, fully penetrating wells and no recharge boundaries; they are unsuitable without modification for layered, leaky or transient systems.

A design workflow is: establish groundwater levels and seasonal extremes; define allowable excavation inflow and drawdown; identify aquifer boundaries; select sump pumping, wellpoints, deep wells, ejectors or cut-off-plus-pumping; calculate preliminary flows; check well capacity and redundancy; evaluate settlement and contaminant migration; and specify monitoring and emergency power. Sump pumping is acceptable only where inflow can be collected without removing soil fines. Otherwise, filtered wells or a low-permeability enclosure are safer.

8. Worked integrated design example

An illustrative, not project-specific, weir has a 12 m impervious floor. The estimated uplift pressures at the upstream and downstream ends are 110 kPa and 35 kPa, varying approximately linearly. The vertical load excluding uplift is 1,900 kN per metre width, the horizontal water load is 420 kN/m, and the interface friction coefficient is 0.55. A reliable passive resistance of 60 kN/m is available.

Uplift is

\[

U=12(110+35)/2=870\,kN/m.

\]

Net vertical force is \(V-U=1,030\,kN/m\). The preliminary sliding resistance is

\[

0.55(1,030)+60=626.5\,kN/m,

\]

so

\[

FS_s=626.5/420=1.49.

\]

Whether this is acceptable depends on the governing code and load combination; it should not be labelled safe without partial-factor checks. If drains reduce the average uplift by 20%, uplift becomes 696 kN/m, net vertical force becomes 1,204 kN/m, and resistance becomes \(0.55(1,204)+60=722.2\,kN/m\), giving \(FS_s=1.72\). This demonstrates both the value of drainage and the danger of assuming drain efficiency without monitoring. The pressure distribution must also be checked for local uplift, cracking and exit-gradient conditions.

9. Applied case: illustrative cofferdam excavation

Background

A temporary cofferdam is proposed in a sand layer overlying clay. The excavation is 6 m deep, groundwater is 1 m below ground level, and the downstream side contains an artesian sand seam. Initial calculations based only on a sump predicted manageable inflow.

Problem and analysis

Trial pumping produced cloudy water and increasing inflow. Piezometers showed a steep upward gradient near the excavation base. The issue was not merely discharge volume: removal of fines indicated incipient internal erosion, while artesian pressure reduced effective stress at the base. A basal-heave check in the clay and a piping check in the sand were therefore both necessary.

Approach and solution

The revised design used a sheet-pile cut-off keyed into the clay, staged excavation, filtered relief wells inside the enclosure, and a graded filter layer beneath the working surface. Pumping was increased gradually while piezometers, turbidity and settlement points were monitored. The cut-off was checked for seepage bypass around its tip, and well spacing was adjusted after a pumping test.

Outcome and learning points

The revised system reduced inflow and turbidity, maintained positive effective stress at the base and limited drawdown outside the cofferdam. The case demonstrates that dewatering, seepage control and foundation stability are one coupled problem. It also illustrates the importance of observational methods: measured piezometric and turbidity trends were used to validate or revise the design.

10. Numerical modelling, monitoring and governance

Finite-element seepage models can represent anisotropy, staged construction, unsaturated zones, drains and coupled consolidation. However, model precision cannot exceed the quality of permeability, boundary and stratigraphic data. Conduct sensitivity analyses for hydraulic conductivity, grout effectiveness, drain clogging, reservoir level and transient drawdown. Calibrate against piezometric and flow observations rather than adjusting parameters merely to obtain a preferred answer.

Monitoring should include reservoir and tailwater levels, piezometers, uplift cells, drain discharge, turbidity, settlement, deformation, relief-well flow and visual leakage inspections. Alert and action levels must be defined before construction. Data governance requires traceable calibration, secure records and clear responsibility for decisions. AI-assisted anomaly detection may help identify changes in piezometric trends or drain discharge, but it cannot replace engineering judgement, sensor validation or emergency procedures.

Common professional failures include ignoring construction defects, using average permeability in strongly layered ground, treating a filter rule as a proof of internal stability, counting passive resistance that may be excavated, and failing to model transient drawdown. Designs should follow applicable dam, foundation, groundwater and occupational-safety regulations, obtain independent review for high-consequence structures, and control grout, pumped water and contaminated discharge responsibly.

Chapter Summary

Foundation performance in hydraulic structures is governed by the interaction of hydraulic head, pore pressure, effective stress and soil or rock resistance. Darcy’s law and flownets provide first-order estimates of seepage and uplift, while exit-gradient calculations identify potential heave. Internal erosion is broader than piping at the exit and requires attention to soil grading, defects, interfaces and filters. Cut-offs and grout curtains lengthen or block seepage paths; drains and relief wells control pore pressure but require filters, capacity, access and monitoring. Bearing capacity, sliding, overturning, slope stability, basal heave and dewatering must be assessed with consistent drainage assumptions and construction stages. Analytical equations are valuable for screening, but layered ground, anisotropy, transient conditions and high-consequence projects require numerical analysis, field testing and observational control.

Key Terms

Effective stress: Total stress minus pore-water pressure.

Hydraulic gradient: Head loss per unit length in the direction of flow.

Hydraulic conductivity: Soil or rock property governing Darcy flow.

Flownet: Network of flow lines and equipotential lines.

Uplift: Water pressure force acting upward beneath a structure.

Seepage force: Hydraulic body force equal to gradient times unit weight of water.

Exit gradient: Hydraulic gradient where seepage emerges from the ground.

Critical gradient: Gradient at which submerged soil particles are theoretically suspended.

Piping: Progressive internal erosion associated with seepage.

Suffusion: Migration of finer particles through a coarser soil skeleton.

Filter: Permeable material designed to retain soil while passing water.

Cut-off: Low-permeability barrier that lengthens or restricts seepage.

Grout curtain: Grouted zone in rock intended to reduce hydraulic conductivity.

Relief well: Well used to lower pore pressure or uplift.

Basal heave: Upward failure of an excavation base, commonly in soft clay.

Dewatering: Controlled removal or pressure reduction of groundwater.

Piezometer: Instrument measuring pore-water pressure or hydraulic head.

Anisotropy: Directional variation of hydraulic or mechanical properties.

Review Questions

Short Answer Questions

1. State Darcy’s law and identify its principal assumptions.

2. Explain why uplift affects sliding resistance.

3. Distinguish backward erosion from suffusion.

4. What are the two competing functions of a granular filter?

5. Why can a partial cut-off increase local exit gradients?

Long Answer Questions

1. Derive the flownet discharge expression for a two-dimensional homogeneous isotropic foundation and discuss its limitations.

2. Explain the mechanisms of internal erosion and develop a field and laboratory investigation plan.

3. Compare cut-offs, grout curtains, drainage blankets and relief wells for seepage control.

4. Develop a postgraduate stability-assessment procedure for a hydraulic-structure foundation including bearing, sliding, uplift and basal heave.

5. Discuss the assumptions and limitations of analytical well-flow equations in dewatering design.

Analytical / Application Questions

1. A foundation has \(k=3\times10^{-5}\,m/s\), head loss 5 m, four flow channels and ten potential drops. Calculate discharge per metre width and explain two reasons the estimate may require correction.

2. For saturated soil with \(G_s=2.70\) and \(e=0.70\), calculate the critical gradient. If the measured exit gradient is 0.60, calculate the gradient-based factor of safety and explain why it is not a complete internal-erosion assessment.

3. A structure has downward load 2,400 kN/m, uplift 900 kN/m, horizontal load 500 kN/m, \(\mu=0.50\), and reliable passive resistance 40 kN/m. Calculate the preliminary sliding factor of safety and identify the additional checks required before design approval.

Multiple Choice Questions with Answers

Question 1

For a homogeneous isotropic flownet, discharge per unit width is represented by which expression?

A. q = kH(Nd/Nf)

B. q = kH(Nf/Nd)

C. q = H(Nf/Nd)/k

D. q = k(Nd/Nf)/H

Correct Answer: B. q = kH(Nf/Nd)

Question 2

For saturated soil with Gs = 2.65 and e = 0.65, the critical hydraulic gradient is:

A. 0.40

B. 0.65

C. 1.00

D. 1.65

Correct Answer: C. 1.00

Question 3

A principal effect of uplift beneath a hydraulic structure is to:

A. Increase effective vertical load and friction

B. Reduce effective vertical load and frictional resistance

C. Eliminate all horizontal water pressure

D. Increase soil unit weight

Correct Answer: B. Reduce effective vertical load and frictional resistance

Question 4

Which internal-erosion mechanism involves finer particles migrating through the constraining framework of coarser particles?

A. Suffusion

B. Overturning

C. Consolidation

D. Liquefaction by vibration only

Correct Answer: A. Suffusion

Question 5

A flownet has k = 2 × 10^-5 m/s, H = 6 m, Nf = 4 and Nd = 12. The discharge per metre width is:

A. 1.0 × 10^-5 m3/(s·m)

B. 2.0 × 10^-5 m3/(s·m)

C. 4.0 × 10^-5 m3/(s·m)

D. 1.44 × 10^-4 m3/(s·m)

Correct Answer: C. 4.0 × 10^-5 m3/(s·m)

Question 6

The main purpose of a properly designed granular filter is to:

A. Block all water flow

B. Retain protected soil while allowing controlled drainage

C. Increase uplift pressure

D. Replace every stability analysis

Correct Answer: B. Retain protected soil while allowing controlled drainage

Question 7

The unconfined radial-flow well equation assumes, among other conditions:

A. A fully penetrating well in a homogeneous aquifer

B. Zero hydraulic conductivity

C. Only turbulent flow through rock joints

D. An impermeable aquifer with no saturated thickness

Correct Answer: A. A fully penetrating well in a homogeneous aquifer

Question 8

If downward load is 1,900 kN/m, uplift is 870 kN/m, horizontal load is 420 kN/m, μ = 0.55 and passive resistance is 60 kN/m, the preliminary sliding factor of safety is approximately:

A. 0.98

B. 1.20

C. 1.49

D. 2.35

Correct Answer: C. 1.49

Question 9

A partial cut-off that does not reach a low-permeability stratum may:

A. Guarantee zero seepage

B. Redirect flow beneath its tip and concentrate gradients

C. Remove the need for filters

D. Always reduce pore pressure everywhere

Correct Answer: B. Redirect flow beneath its tip and concentrate gradients

Question 10

Which measurement is most directly useful for tracking foundation pore-pressure response?

A. Concrete colour

B. Piezometric head

C. Air temperature alone

D. Aggregate shape only

Correct Answer: B. Piezometric head