Groundwater and Darcy's Law

Groundwater creeps through pores, and only the pores carry it.

Before this Lesson

Last winter, a highway crew piled road salt on bare ground, and every rain since has carried chloride into the sand below. The town's supply well is 800 ft away. Monitoring wells by the pile and by the town well show the water standing 4.0 ft lower at the town well.

Chloride doesn't react or break down. It goes wherever the water goes, as fast as the water goes. Is it headed for the well, and if so, does it arrive in months, years, or centuries?

Head in the ground

A well is a pipe set into saturated ground, with a slotted screen at the bottom. Water rises in it until its weight balances the pressure at the screen, so a well is a piezometer, like the standpipes in The Energy Equation. Its water level marks the hydraulic head:

h=Pγ+zh = \frac{P}{\gamma} + z

where zz is the screen's elevation and P/γP/\gamma is how far the water rises above it. The velocity head is missing for a reason. Even at a brisk 1 ft/day, v2/2gv^2/2g is about 2×10−122 \times 10^{-12} ft, a tiny fraction of the width of an atom. So the head is just the water level in a well, as an elevation, and the energy equation's rule holds: water flows from higher head to lower, losing the difference to friction.

That answers the town's first question. The water stands 4.0 ft higher at the pile, so it's flowing toward the well.

Aquifers and aquitards

An aquifer is a saturated layer that can deliver useful amounts of water, such as sand or gravel. A layer that barely passes water, such as clay, is an aquitard.

An unconfined aquifer has no lid. Its top is the water table, where the water is at atmospheric pressure, and a well in it fills to the water table.

A confined aquifer is sealed above by an aquitard, and below by another aquitard or bedrock, and stays full to its top. Its water usually entered where the layer comes to the surface on higher ground, so it's under pressure, like a buried main fed from a hilltop tank. A well drilled through the aquitard fills to above the aquifer's top, and that level, mapped across the aquifer, is the piezometric surface. Water standing above the top of its aquifer is what artesian means, so problems often call a confined aquifer an artesian one. Where the piezometric surface stands above the ground, a well overflows on its own: a flowing artesian well.

Recharge area A: water-table well B: artesian well C: flowing artesian well Piezometric surface Ground surface Water table Unconfined aquifer Aquitard Confined aquifer Impermeable bedrock
An unconfined aquifer over a confined one
Wells B and C tap the confined aquifer, so their water rises to the piezometric surface.

The aquitard keeps the two waters apart, so a confined aquifer's piezometric surface needn't match the water table above it. The Handbook draws both aquifer types beside its well equations, labeling the aquitard and the piezometric surface, but doesn't define them in words.

Checkpoint

A boring passes through 15 ft of sand that is saturated below a depth of 6 ft, then 30 ft of stiff clay, then 50 ft of gravel resting on bedrock. A well screened only in the gravel fills to 2 ft below the ground surface. Which statements are true?

A.
The sand is an unconfined aquifer, with its water table 6 ft below the ground.
B.
The water in the gravel well stands at the level of the sand's water table.
C.
The clay is an aquitard.
D.
The gravel well is a flowing artesian well.
E.
The gravel is a confined aquifer under artesian conditions.

Darcy's law

Picture water working through sand: countless passages a fraction of a millimeter across, twisting around the grains. In passages that narrow, all the water is close to some grain, so drag is the whole story, with no eddies and no speed worth mentioning. Drag that slow grows in step with the speed. In a pipe, doubling the flow roughly quadruples the head loss. In sand, it only doubles it.

Henry Darcy measured this in 1856, testing sand filters for the water supply of Dijon, France. The flow rose with the head lost across the sand and with its area:

Q=KiAi=h1−h2LQ = KiA \qquad i = \frac{h_1 - h_2}{L}

The hydraulic gradient ii is the head lost per unit length of flow path. The hydraulic conductivity KK, in units of a velocity, says how readily the soil passes water. The Handbook's Darcy's Law section writes Q=−KA (dh/dx)Q = -KA\,(dh/dx), the minus sign saying the flow runs toward falling head.

For an aquifer, ii is the difference in head between two wells over their spacing along the flow. AA is the slice the water crosses, perpendicular to the flow: the thickness bb times the width WW. The plan area, LL times WW, is the floor the water travels along, not the doorway it passes through. In an unconfined aquifer, bb is the saturated thickness.

Datum h1 h2 L Q b W A = bW
Flow through a block of confined aquifer between two wells

Match the units of KK before multiplying. A day has 86,400 s, which the Handbook's conversion table leaves out, so 1 cm/s is 864 m/day, or about 2,835 ft/day.

Worked Example Flow through a confined aquifer
1 of 3

A confined sand aquifer is 40 ft thick and has a hydraulic conductivity of 120 ft/day. Two observation wells 1,500 ft apart, on a line along the flow, have water levels at elevations 372.0 ft and 366.0 ft. How much water flows through a 2,000 ft wide section of the aquifer?

Only the pores carry the flow

Divide Darcy's law by the area and you get a velocity, q=Q/A=Kiq = Q/A = Ki, which the Handbook calls the specific discharge, or Darcy velocity. It's a bookkeeping speed: the flow spread over the whole section, as if the grains weren't there. But the grains fill most of the section, and the water gets through only the connected pores between them. That open share is the effective porosity nen_e, and the same flow through a smaller opening has to move faster. The water itself travels at the seepage velocity:

v=qne=Kinev = \frac{q}{n_e} = \frac{Ki}{n_e}
Darcy velocity Seepage velocity q = Q/A v = q/ne Q Q
The same flow, spread over the section and squeezed through the pores

With nen_e around 0.2 to 0.3, that's three to five times the Darcy velocity. Effective porosity counts only the pores that carry flow, so it's less than the total porosity, n=Vv/Vn = V_v/V, from Soil as Three Phases. The Handbook's Darcy's Law section writes it as a plain nn and calls vv the average seepage velocity, but its Hydraulic Conductivity/Permeability section, just before, writes the Darcy velocity as v=Kiv = Ki. Go by the name, not the letter.

A conservative contaminant, one that doesn't react, decay, or stick to the grains, rides with the water. Its travel time over a distance LL is t=L/vt = L/v. Using the Darcy velocity stretches that by 1/ne1/n_e, and the contaminant arrives well before the forecast.

Worked Example When the salt reaches the town well
1 of 2

Back to the salt pile. The sand under the field has a hydraulic conductivity of 40 ft/day and an effective porosity of 0.25, and the water table drops 4.0 ft over the 800 ft from the pile to the town well. How long does the chloride take to reach the well?

What sets K

Why does sand pass water readily and clay barely at all? Not pore space: a clay's porosity is often 40 to 70 percent, against 25 to 50 for a sand. It's pore size. In slow flow the drag comes from the walls, and at a fixed gradient the speed through a passage goes roughly as the square of its width. Pores scale with grains, so a clay particle a thousand times smaller than a sand grain gives a KK about a million times smaller. Mixed soils follow their small pores: silt in a sand clogs the gaps and pulls its KK down.

The Handbook prints no table of typical values, though its section on Break-Through Time for Leachate to Penetrate a Clay Liner mentions clays of 10−610^{-6} to 10−810^{-8} cm/s. Standard references give rough ranges like these, which real soils overlap:

Clean gravel Clean sand Silt Clay 10−10 10−9 10−8 10−7 10−6 10−5 10−4 10−3 10−2 10−1 100 101 102 Hydraulic conductivity, K (cm/s)
Rough ranges of hydraulic conductivity

A typical gravel and a typical clay sit about eight powers of ten apart. At the same gradient, the gravel passes as much water in a second as the clay passes in three years.

Checkpoint
1 of 2

Borings at a site find four layers: silty sand, clay, clean gravel, and clean fine sand. Which list runs from the highest typical hydraulic conductivity to the lowest?

A.
Clay, silty sand, fine sand, clean gravel
B.
Clean gravel, fine sand, clay, silty sand
C.
Clean gravel, fine sand, silty sand, clay
D.
Clean gravel, silty sand, fine sand, clay

Measuring K in the lab

With a range that wide, KK gets measured, not guessed. Geotechnical problems often call it the coefficient of permeability, kk. A lab permeameter pushes water through a sample of length LL and area AA, and the soil decides which test to use.

Sands and gravels pass water freely, so the constant-head test holds the head difference hh across the sample fixed, with an overflowing tank upstream and a fixed outlet downstream, and collects what comes out.

Sample A L Constant level h Volume Q collected in time te
A constant-head permeameter

That's Darcy's law with the flow written as a volume over a time, and i=h/Li = h/L. The Handbook's Hydraulic Conductivity/Permeability section gives

K=QiAteK = \frac{Q}{iAt_e}

where QQ is the total volume collected, not a rate, and tet_e is the elapsed time.

Silts and clays pass so little water that a cup might take days to fill, so the falling-head test watches a water level instead. A narrow standpipe of area aa feeds the sample. Because aa is small next to AA, a tiny flow makes a big drop: the standpipe is a magnifying glass on the flow.

Sample A L t = 0 t = te a h1 h2
A falling-head permeameter

As the level falls, the push weakens and the flow slows. With the level a height hh above the outlet, the sample passes K(h/L)AK(h/L)A, which is exactly what drains from the standpipe:

−a dhdt=K hL A-a\,\frac{dh}{dt} = K\,\frac{h}{L}\,A

The level falls at a rate proportional to the head that's left, an exponential decay. Integrating from h1h_1 at t=0t = 0 to h2h_2 at tet_e:

K=aLAte ln⁡h1h2K = \frac{aL}{At_e}\,\ln\frac{h_1}{h_2}

The Handbook prints the same result as K=2.303 (aL/Ate)log⁡10(h1/h2)K = 2.303\,(aL/At_e)\log_{10}(h_1/h_2), and its symbol list calls the sample length xx. Use ln⁡\ln alone or log⁡10\log_{10} with the 2.303, and keep the standpipe's aa on top and the sample's AA below.

Checkpoint
1 of 2

A constant-head test runs on a sand sample 15 cm long and 10 cm in diameter, with a head difference of 40 cm across it. In 2.0 min, 750 mL of water is collected. What is the sample's hydraulic conductivity?

cm/s

Steady Flow to Wells applies Darcy's law to water converging on a pumped well from every side.

Recap

  • Head is h=P/γ+zh = P/\gamma + z, the water level in a well as an elevation, and water flows from higher head to lower.
  • An unconfined aquifer's top is the water table. A confined aquifer lies between aquitards, and its wells rise to the piezometric surface: artesian above the aquifer's top, flowing artesian above the ground.
  • Q=KiAQ = KiA, with i=(h1−h2)/Li = (h_1 - h_2)/L and A=bWA = bW perpendicular to the flow, in matching units.
  • The water moves through the pores at v=q/nev = q/n_e, faster than the Darcy velocity q=Kiq = Ki, and travel time is L/vL/v.
  • KK follows pore size: gravel, sand, silt, clay. A constant-head test gives K=Q/(iAte)K = Q/(iAt_e), and a falling-head test K=(aL/Ate)ln⁡(h1/h2)K = (aL/At_e)\ln(h_1/h_2).