Pavement base and subgrade compaction: build the structure under the mat

The base is the pavement, the surface just seals it

Pavement base and subgrade compaction is the work of densifying the soil and aggregate layers under the asphalt or concrete so they carry traffic without moving. This is the part of the job that decides how long the pavement lasts. The mat on top seals out water and gives you a riding surface. The structure under it carries the load down to the ground, spreading each wheel into a wider footprint at each layer so the soil at the bottom never sees more pressure than it can take.

Get the base right and a mediocre mat will live out its life. Get the base wrong and the best mat money can buy fails early, because it is flexing over a foundation that keeps moving under it. You see it as alligator cracking, rutting, and edge break, and the contractor gets blamed for the asphalt when the asphalt was never the problem.

Most pavement failure starts here, below grade, not on the surface. The cracks show up on top because that is where you can see them. The cause is almost always a subgrade or base that was soft, wet, under-compacted, or starved of drainage. So the work that matters most is the work nobody sees once the mat covers it.

Why does pavement fail from the base and not the surface?

Pavement fails from the base because the base carries the load and the surface only rides on it. When the wheel rolls over, the load spreads down through the layers and the soil at the bottom takes the smallest pressure of all, but only if every layer above it is stiff enough to spread the load instead of punching through. A soft layer anywhere in the stack concentrates the load instead of spreading it, and the layers above flex to fill the void.

Flexible pavement is exactly that. It bends under each axle and springs back. A well-built section bends a little and recovers. A section over a weak subgrade bends too far, and asphalt that bends too far, too many times, cracks from fatigue. That is the alligator pattern, interconnected cracks in the wheel paths, and it is a structural signal, not a surface one. The PCI survey calls it load distress for a reason.

The two killers below grade are under-compaction and water, and they work together. Soil that was never compacted to density settles under traffic. Soil that drains poorly softens when it gets wet and loses the strength it had dry. Either one turns a stiff foundation into a moving one, and the mat tells on it within a few seasons. For how that distress is read and rated after the fact, see the pavement condition assessment guide.

The pavement section: subgrade, subbase, base, and surface

A pavement section is a stack of layers, each stiffer than the one below it, that turns a concentrated wheel load into a pressure the natural soil can carry. From the bottom up the names are fairly consistent, though specs use them loosely, so read the section the project actually calls for.

The subgrade is the natural ground, cut or filled to grade and compacted. It is the foundation, and its strength sets how thick everything above it has to be. The subbase, when there is one, is a lower-cost granular layer that bridges a weak subgrade and adds drainage. The aggregate base is the dense, crushed-stone layer directly under the pavement that does most of the load spreading. On top sits the asphalt or the concrete, the wearing surface.

The thinner the structure below, the harder the surface has to work, and the surface is the most expensive layer per inch. That is the design trade. A strong subgrade lets you build a thinner, cheaper section. A weak subgrade forces a thicker base, a subbase, or stabilization to make up the difference. The compaction guide for the asphalt mat itself covers what happens above this line; everything in this guide is what happens below it.

Preparing the subgrade: strip, proof, undercut, condition

Subgrade prep starts with stripping the topsoil and any organic or unsuitable material off the cut. Topsoil holds organics that rot and compress, so it never stays under pavement. Strip down to a competent native soil, then bring the area to rough grade in cut or in compacted fill.

Then you find the soft spots before you build over them. That is what proof rolling does. A loaded truck runs the graded surface and the weak areas show themselves by pumping or rutting under the tires. Where the ground is soft, you undercut, dig out the bad material, and replace it with compacted structural fill or aggregate. How deep you go is a geotech call, not a guess, because a soft pocket can be a foot deep or it can be a sign of a saturated layer well below.

Moisture conditioning is the quiet half of the work. Soil that is too dry will not compact and soil that is too wet pumps and will not hold density. So you wet it with a water truck or dry it by discing and aerating until it sits near its optimum moisture content, then you roll it. On a clay site this conditioning is most of the schedule, because clay gives up or takes on water slowly and you cannot rush it. Skip the conditioning and you chase density you will never reach.

What is proof rolling?

Proof rolling is a field test where a heavy loaded vehicle is driven slowly across the prepared subgrade or base while an inspector watches for movement. It is not a density test. It is a deflection test you can see with your eyes, and its whole purpose is to reveal soft or unstable areas that a few scattered density tests would miss.

The common rig is a loaded tandem-axle dump truck, often cited in the range of a 20 ton gross load, though some specs call for a loaded scraper or a heavy rubber-tired roller. It moves at a slow walking pace, commonly in the 2 to 5 mph range, in overlapping passes so the whole area gets covered. The inspector and the foreman walk it and watch the ground under and behind the tires.

What you are looking for is pumping and rutting. Pumping is the ground flexing and water or fines working up under the load, a wave that moves ahead of the tire. Rutting is permanent deformation, the tire leaving a track that does not spring back. A stiff subgrade barely shows the tire. A failing one ruts, pumps, or quakes, and that area gets marked, undercut, and rebuilt, then proof rolled again. Verify the truck weight, speed, and pass pattern against the project specification, because the agency sets them.

What is a Proctor test, and what is optimum moisture?

A Proctor test is the lab test that establishes how dense a given soil can be compacted and at what moisture content it gets there. You take the soil, compact it at several moisture contents with a standard effort, and plot dry density against moisture. The curve peaks. That peak is the maximum dry density, and the moisture that produces it is the optimum moisture content. Every field density target is a percentage of that lab maximum, so the Proctor is the yardstick the whole job is measured against.

The shape of the curve is the physics. Too dry and the soil grains grind and bridge and will not pack. Add water and it lubricates the grains so they slide into a denser arrangement, up to the peak. Add more water past optimum and the water itself starts taking up space the grains should occupy, so density falls off again. That is why you compact near optimum, not soaking and not bone dry.

There are two versions and the difference matters. Standard Proctor, ASTM D698, uses a lighter rammer and less energy. Modified Proctor, ASTM D1557, uses a heavier rammer dropped farther for roughly 4.5 times the compactive energy, which reflects modern heavy compaction equipment. Modified gives a higher maximum dry density, commonly several percent higher, at a lower optimum moisture, commonly a point or two drier. So 95 percent of modified is a tougher target than 95 percent of standard. The spec must say which Proctor governs, and confusing the two is a real way to fail or argue over a test that was never apples to apples.

Maximum dry density (MDD)

The peak dry unit weight a soil reaches in the Proctor test, the 100 percent reference

Optimum moisture content (OMC)

The moisture content that produces the maximum dry density

Standard Proctor (ASTM D698)

The lower-energy compaction test, lighter rammer and shorter drop

Modified Proctor (ASTM D1557)

The higher-energy test, roughly 4.5 times the effort, higher MDD and lower OMC

Relative compaction

Field dry density as a percentage of the Proctor maximum dry density

What compaction do you need under pavement?

Aggregate base under pavement is commonly compacted to 95 percent of modified Proctor maximum dry density, and the subgrade target is often the same 95 percent in the top layer with a lower figure, frequently around 90 percent, allowed in deeper fill. Those are typical numbers. The project geotechnical report and the earthwork specification set the actual targets and say which Proctor they are measured against, so confirm them before you roll, never assume.

The reason base gets the tighter number is that it carries the most concentrated load and any settlement there shows up fastest at the surface. A common pattern is to allow standard Proctor for general subgrade and require modified Proctor for the granular subbase and base course, but specs vary by agency, so read yours. The figure is always relative compaction, a percentage of that soil's own lab maximum, not an absolute density you could carry from job to job.

Treat a test that lands a point or two above the minimum as your real margin, because field conditions move. A passing test in dry weather can read low after a rain wets the lift, and a borrow that changes character mid-job changes its own Proctor maximum. The number on the spec is the floor, not the goal. Build to comfortably clear it.

Field density testing: nuclear gauge, sand cone, and frequency

Field density testing measures the in-place dry density and moisture of a compacted lift so you can compare it to the Proctor maximum and call the lift passing or failing. The two long-standing methods are the nuclear density gauge and the sand cone, and a third, the lightweight deflectometer, is showing up on more specs as a stiffness check rather than a density one.

The nuclear density gauge, ASTM D6938, also referenced as AASHTO T 310, reads density and moisture in a couple of minutes by counting gamma and neutron radiation, which makes it the workhorse on production jobs where you need many tests fast. It is non-destructive and quick, but it is a licensed radioactive device, so it carries training, transport, and storage rules, and it wants a Proctor and an occasional sand-cone correlation to stay honest. The sand cone, ASTM D1556, is the slower reference method: you dig a small hole, weigh the soil you removed, and measure the hole's volume by filling it with calibrated sand. It is the tiebreaker when a nuclear result is in dispute.

Frequency is set by the spec, commonly as one test per a set area or per a set volume of each lift, with extra tests at suspect spots and at the start of a new material. The locations should be random and representative, not cherry-picked. A technician who tests only the spots that look good is producing a passing report on a failing lift, and the cores or the cracking find it out later.

Moisture control: compacting at optimum

Moisture is the lever that decides whether a lift reaches density or fights you all day. You compact near the optimum moisture content from the Proctor, commonly within a window of a couple of points around it that the spec sets. Inside that window the grains move and pack. Outside it, on either side, you lose density no matter how many passes you make.

Too dry and the soil will not compact. The grains have no lubrication, they bridge and resist, and the roller just rides over a loose lift that reads low. The fix is the water truck, applied and mixed in, then given a little time to spread through the lift before you roll. Too wet is the worse failure on cohesive soil. A clay above optimum pumps under the roller, the surface waves and shoves, and you cannot build density on a layer that is acting like pudding. The fix there is to disc it open and aerate it, let the sun and wind dry it back, which costs you time you rarely have.

On a clay subgrade, moisture is most of the battle and the weather runs your schedule. You cannot push wet clay to density and you cannot dry it on demand. Plan the earthwork around the forecast, protect a conditioned subgrade from rain with a sealed, sloped surface, and do not let a finished lift sit open and soak the night before you build on it.

The aggregate base: gradation, angularity, and fines

The aggregate base is a dense-graded crushed stone, which means a blend of sizes from a top size down through sand to a controlled amount of fines, designed to pack into a tight, interlocked mass with very little void left. Dense-graded is the point. The range of sizes lets small particles fill the gaps between large ones, so the layer ends up strong and relatively impermeable. The gradation band is in the spec, and a load that runs outside it does not perform the way the design assumed.

Angular, crushed stone beats rounded gravel here. Crushed faces lock against each other and resist shoving under load, where rounded river gravel rolls and rearranges and never develops the same internal friction. Many base specs require a minimum percentage of crushed or fractured faces for exactly this reason, so a cheaper rounded material is not a free substitution.

Fines are a balance you can get wrong in both directions. Some fines are needed to fill the voids and let the layer compact tight. Too many fines, especially plastic clay fines, and the base holds water, softens when wet, and loses the free-draining behavior the section was counting on. A clean open-graded base drains hard but has less stability; a dense-graded base is stronger but drains slower. Which one you want depends on whether the design is using the base as a drainage layer or as a load layer, so build the gradation the spec calls for and do not let the pit substitute.

Lift thickness and the roller that matches the soil

Lift thickness is the height of loose material you place before compacting, and it has a ceiling because compaction energy only reaches so deep. Place too thick and the bottom of the lift never sees enough energy to reach density, so it stays loose under a tight crust that tests fine on top and fails where you cannot reach it. A common compacted-lift figure for base and granular fill is in the range of 6 to 8 in, with the loose lift placed a little thicker to compact down into that, but the spec and the equipment set the real number, so verify it.

The roller has to match the soil. Granular base and sand compact best under a smooth-drum vibratory roller, where the vibration shakes the grains into a denser packing. Cohesive clay does not respond to vibration the same way; it wants a kneading, shearing action, which is what a padfoot or sheepsfoot roller delivers with its feet pushing into the lift and compacting it from the bottom up. Run a smooth drum on wet clay and you seal a slick surface over a soft lift. Run a sheepsfoot on clean stone and the feet just punch holes.

Passes are established, not assumed. You build a test strip, compact with a set roller pattern, and take density tests until you know how many passes at what amplitude bring that material to the target. Then that is the pattern, and the operator holds it across the job. Counting passes is cheaper than coring a lift you have already buried.

Subgrade types and how strength is measured

Subgrades fall into two broad camps that behave very differently. Granular subgrades, the sands and gravels, drain well, compact readily under vibration, and hold their strength wet or dry. Cohesive subgrades, the clays and clayey silts, are the hard ones. They are moisture-sensitive, slow to condition, and they lose strength fast when they get wet. Soft saturated clay is the classic problem subgrade, and a lot of undercut, stabilization, and schedule grief traces back to it.

Strength gets quantified so the pavement can be designed to it. The common measures in the United States are the California Bearing Ratio, the R-value, and the resilient modulus. The CBR rates the soil's strength against a crushed-stone standard, where a soft clay can read below 2 to 5 and a good granular soil reads much higher. The R-value is a related stability measure used by some agencies. The resilient modulus describes how stiffly the soil springs back under repeated wheel loading, and the AASHTO design framework leans on it. These correlate to one another loosely, so a design will name the one it used.

The number that matters is whatever the geotech reported for your site, because that value sized the section sitting above it. If the field subgrade is softer than the report assumed, the design is no longer valid and the section above it is under-built. That is a stop-and-call moment, not a keep-going one, because everything you place on a weaker-than-designed subgrade inherits the deficit.

Stabilizing a weak subgrade

When the subgrade is too weak to build on and undercutting it all is not practical, you stabilize it, meaning you improve the soil in place instead of hauling it off. The method depends on the soil and the problem, and the geotech specifies it, because the wrong treatment on the wrong soil makes things worse.

Chemical stabilization is common on cohesive soils. Lime works on clays: mixed in and given time to react, it dries the soil, raises its strength, and reduces its plasticity so it stops swelling and shrinking. Cement works on a broader range of soils and builds more strength but is less forgiving and can shrink and crack. The reaction and the cure time are real, so this is not a same-day fix; it wants mixing, moisture, compaction, and curing in sequence.

Mechanical stabilization uses geosynthetics. A geotextile separates a soft subgrade from the clean base above it so the soil does not pump up and contaminate the stone, which is a frequent quiet failure on wet sites. A geogrid does more: its apertures interlock with the aggregate and confine it, spreading load and letting you build a working platform over a subgrade that would otherwise rut under the first truck. On a genuinely soft site, a geogrid plus a thicker granular layer is often cheaper than excavating and replacing everything below. The design and the product are an engineering call, so build what the geotech detailed.

Drainage and the base: water is the enemy

Water is the single biggest driver of base and subgrade failure, and a well-built section can still fail if it cannot shed water. The mechanism is straight physics. Push a soil or a base past its optimum moisture and it softens, pore pressure builds under each wheel, and the layer loses the strength it had when it was drained. The same stone that carried the load dry turns unstable wet. So drainage is not a finishing detail. It is part of the structure.

The base earns its keep partly as a drainage layer, which is why dense-graded base is sloped to carry water to an outlet rather than trap it. On many designs the base is daylighted, run out to a slope or a ditch so water has a path to leave, or it is tied to an edge drain or underdrain that collects the water and pipes it away. A subgrade sloped to drain, commonly at a minimum grade the spec sets, keeps water from ponding under the pavement where it has nowhere to go.

The quiet failure is a base that holds water. A daylighted permeable layer that is too thin or too flat does not drain and just stores water against the subgrade, which is worse than no drainage layer at all. Plastic fines in the base do the same thing by plugging the voids. Keep the base draining, keep the subgrade sloped, give the water a real exit, and you take away the thing that turns a good section soft. Skip it and you own the failure a few wet seasons out.

Frost, freeze-thaw, and frost heave

In freezing climates the base has a second enemy, and it is the same water in a different state. Frost heave happens when a frost-susceptible soil, water, and freezing temperatures all show up together. The soil freezes from the top down, and as the freezing front advances it pulls up water from below, which forms ice lenses that swell the soil and heave the pavement. Then the thaw is worse than the freeze: the ice melts, the soil is saturated and trapped over still-frozen ground, and the subgrade goes to mush right when spring traffic loads it.

Silts are the worst offenders. A clean sand or gravel does not wick water well and a heavy clay freezes slowly, but a silt has both the small pores to draw water up and the permeability to feed the lenses, so it heaves hard. Clean granular bases are not frost-susceptible, which is one more reason to keep fines out of the base.

The defenses are familiar earthwork. Get the frost-susceptible soil out down to a depth the design sets, or break the water supply with drainage so there is nothing to feed the lenses, or carry a non-frost-susceptible section deep enough that the frost-sensitive layer never freezes. Frost depth and the required strategy are local and the geotech and the agency govern them, so do not carry a number from one region to another.

Density acceptance, test reports, and proving the base before paving

Acceptance is the step where the compacted layer becomes a documented, passing layer you are allowed to build on. Each lift gets density tested at the spec frequency, the results are compared to the Proctor maximum, and the layer is accepted only when the tests meet the target. The test report is the record: location, material, lift, Proctor reference, the measured density and moisture, the percentage, and pass or fail. That report is what defends the layer when something cracks later.

A failing test is not a paperwork problem, it is a rework problem. When a test fails, you find the extent of the soft area, recompact it, adjust the moisture if that was the cause, and retest until it passes. You do not average a fail away with nearby passes unless the spec explicitly allows a lot-based acceptance, and even then you respect the minimum. A single low test in a pattern of passes still means a soft spot that will telegraph through the mat.

The base before paving is a hold point, full stop. Do not pave over a subgrade or base that has not passed its density and its proof roll. Once the mat is down, every defect below it is buried and expensive, and the asphalt crew inherits a problem they did not make. The asphalt compaction guide assumes a sound, accepted base under it; this is the gate that makes that assumption true. Prove the base, document it, then call the paver.

Heavy-equipment yards and data center pads

Not every paved surface is a parking lot, and the heavy ones change the base. A laydown yard, a container terminal, a truck apron, or the equipment yard around an industrial site sees loads that dwarf a car: loaded trailers, reach stackers, cranes on outriggers, point loads that a standard parking-lot section was never sized for. The base under those areas wants more thickness, a tighter density, and often a stabilized or thicker subgrade, because the load reaches deeper and the soft layer that a car would never find is exactly what a loaded stacker punches through.

Data center and critical-facility sites add another reason to overbuild the base: the cost of differential settlement. A slab or a pavement that settles unevenly over a poorly compacted subgrade is a problem under a parking lot and a much bigger one under or beside a building full of equipment that cannot tolerate movement. The geotech report on these jobs usually drives tighter compaction, deeper proof rolling, and more density testing than a routine paving job, and the spec earns that rigor.

The principle does not change with the load, only the margin. The base still carries the load and the surface still seals it. On a heavy yard you simply have less room for the soft spot, the thin lift, or the test nobody took, because the consequence is larger and arrives sooner.

What to document

The compaction record is what answers the question when the pavement moves: was the base ever built right? A buried layer cannot be re-inspected, so the report is the only evidence, and on a disputed failure it is the difference between a defended job and an expensive guess.

Capture each test by location so it ties to a spot on the site, the material and which lift it was, the Proctor reference and whether standard or modified, the measured dry density and moisture, the relative compaction percentage, and the clear pass or fail. Keep the proof-roll observations and the undercut locations with it. When a lift fails and is reworked, record the retest, not just the first number, so the history shows the soft area was actually fixed and not papered over.

Common mistakes

• Paving over a soft or wet subgrade because the schedule was tight, then burying the problem under the mat.
• Skipping the proof roll and trusting a handful of density tests to catch every soft area.
• Placing lifts too thick, so the bottom never reaches density under a tight crust that tests fine on top.
• Compacting off optimum moisture, too dry to pack or too wet to hold, and chasing density you cannot reach.
• Confusing standard and modified Proctor, so the field target is measured against the wrong maximum.
• Running a smooth drum on clay or a sheepsfoot on clean stone instead of matching the roller to the soil.
• Letting plastic fines into the base or building a base that does not drain, so water softens it from below.
• Cherry-picking density test locations instead of testing random, representative spots.
• Taking no field density tests at all and accepting the base by eye.

Field checklist

Standards and references

The governing documents on any given job are the project geotechnical report and the earthwork specification. They set the compaction targets, name which Proctor applies, call the proof-roll rig and pattern, and set the test frequency. Everything below is the framework those documents draw from, and where they differ from the general practice here, they win.

For the lab moisture-density relationship, ASTM D698 is the standard Proctor and ASTM D1557 is the modified Proctor; the spec says which one the field density is measured against. For in-place density, ASTM D6938, also referenced as AASHTO T 310, covers the nuclear gauge method and ASTM D1556 covers the sand cone reference method. Pavement and subgrade design in the United States commonly works from the AASHTO framework, using subgrade strength measures like the California Bearing Ratio, the R-value, and the resilient modulus.

The Asphalt Institute and the state DOT or local agency provide the material gradations, the acceptance bands, and the regional rules for frost depth and drainage. Confirm the current edition and the local amendments before citing any of these on a submittal, and never carry a number from one agency's spec onto another agency's job.

Units, terms, and conversions

Earthwork uses a mix of names and units across the geotech report, the spec, and the test forms, so the same idea reads a few different ways on one job.

Dry density is given in pounds per cubic foot (pcf) in US practice and kilograms per cubic meter or grams per cubic centimeter in metric. Relative compaction is a percentage of the Proctor maximum dry density, sometimes written as percent compaction or percent of MDD. Moisture content is a percentage of dry soil weight, reported against the optimum moisture content. Lift thickness is in inches or millimeters, and base gradation is a percent passing each sieve size. Subgrade strength shows up as CBR (a percentage), R-value, or resilient modulus in psi or MPa.

Subgrade

The natural soil, cut or filled and compacted, that forms the foundation of the section

Aggregate base

The dense-graded crushed-stone layer directly under the pavement that spreads the load

Proof rolling

Driving a heavy loaded vehicle over the surface to reveal soft, pumping, or rutting areas

Relative compaction

In-place dry density as a percentage of the Proctor maximum dry density

Optimum moisture content (OMC)

The moisture content at which a soil reaches its maximum dry density

CBR

California Bearing Ratio, a subgrade strength measure relative to crushed stone

Undercut

Excavating soft or unsuitable material and replacing it with compacted structural fill

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• What temperature do you stop rolling asphalt?
• What is the pavement condition index (PCI)?
• What is crusher run or dense-graded aggregate?
• How thick should an asphalt driveway be?
• What is site grading?