Blog

It's a familiar scene on almost every earthwork site. The roller has made eight, ten, even twelve passes. The operator is tired, the schedule is slipping, and then the field density test comes back at 92% when the spec asks for 95% or 98%.

The usual reaction is to run the roller a few more times. Often that changes nothing, and sometimes it makes things worse.

Compaction doesn't fail because of too few passes. It fails because something in the system is off: the moisture, the layer thickness, the machine, the ground underneath or the test itself. This article walks through the seven most common causes, how to confirm each one on site, and how to fix it.

First, a quick reality check: passes are not the goal

Compaction is about reaching a target dry density, usually expressed as a percentage of Maximum Dry Density (MDD) from a Proctor test. Passes are only a means to get there.

Every soil and roller combination has a point of diminishing returns. After a certain number of passes, density barely increases. If you've reached that point and the test still fails, the problem lies elsewhere, so look at the seven causes below.

1. The moisture content is wrong

This is the most common cause of failed compaction.

Soil needs water to compact. Water lubricates the particles so they can slide into a tighter arrangement. But it has to be the right amount, close to the Optimum Moisture Content (OMC).

• Too dry: Particles resist movement. The soil looks fine on the surface but stays loose and lumpy below.
• Too wet: Water fills the voids and can't escape under the roller. The soil pumps, ripples, or feels spongy, and density stops improving no matter how many passes you make.

How to check on site

• Run a quick moisture test (speedy moisture meter, oven drying, or a field method as per your QC plan) before rolling, not after the failure.
• Do a simple hand squeeze test. Soil near OMC holds together when squeezed but crumbles when you press it with a finger. If water squeezes out, it's too wet. If it won't hold shape at all, it's too dry.

How to fix it

• Too dry: Add water with a tanker and mix it thoroughly using a grader or disc harrow. Spraying only the surface isn't enough.
• Too wet: Spread the layer thinner and let it aerate, turn it with a grader or harrow, or blend in drier material. In some cases lime treatment is the practical answer.

Aim for moisture within about ±2% of OMC, or whatever range your specification sets.

2. The layer is too thick

Roller energy fades with depth. If the loose lift is too thick, the top 100 mm gets well compacted while the bottom stays loose. A test taken at mid-depth then fails, even though the surface feels rock hard.

Typical loose lift thickness

• Granular soils and gravel: roughly 200–300 mm, depending on roller size
• Clayey soils with a padfoot or heavy roller: roughly 150–200 mm
• Small plate compactors and rammers: much thinner, often 100–150 mm

Always follow your project specification and trial results, since these figures vary with equipment.

How to check on site

Dig a small test pit after rolling. If you can see a clear density change between the top and the bottom of the layer, thickness is your problem.

How to fix it

Reduce the lift thickness and confirm it with a compaction trial strip before full-scale work begins. A trial strip is cheap compared with removing and redoing a failed section.

3. The roller doesn't suit the soil or the location

A roller that works beautifully on gravel can do very little on clay, and vice versa.

Granular soils respond well to vibration because it helps particles rearrange. Cohesive soils depend more on kneading and static weight, and heavy vibration can sometimes do little or even loosen the surface.

Confined areas such as trench backfill, pipe bedding and edges beside retaining walls are a frequent hidden weak point. A big roller simply can't reach them, so they get under-compacted. For these zones, use properly sized plate compactors and rammers rather than hoping a large roller will cover them.

How to fix it

Match the equipment to the soil and the work area, and don't stretch a single machine across jobs it wasn't built for.

4. The layer underneath is soft or yielding

You can't compact a layer well if the ground below it is soft. The roller's energy gets absorbed by the weak layer instead of densifying the soil above.

Typical signs:

• The surface deflects or "waves" under the roller
• The test passes at some locations and fails badly at others
• Rutting appears even when the top layer looks dry

This often happens over wet subgrade, old fill, organic soil or an unprepared base after rain.

How to check on site

Do a proof roll with a loaded truck or heavy roller and watch for deflection. Dynamic Cone Penetrometer (DCP) or plate load checks on the subgrade can confirm it.

How to fix it

Dig out and replace the soft zone, stabilise it (lime, cement or geogrid, depending on the design), or dry it out before placing the next layer. Compacting more on top of a weak base just wastes roller hours.

5. The roller is operated incorrectly

Even the right roller gives poor results if it's run the wrong way. Common operating mistakes:

• Speed too high. Fast rolling means less energy delivered per metre. Most compaction rollers work best at roughly 2–4 km/h.
• Poor overlap. Gaps between passes leave low-density strips. Overlap each pass by about a third of the drum width.
• Vibration on or off at the wrong time. Vibration helps granular soils but can over-stress wet or cohesive soils. Set amplitude and frequency to suit the material.
• No consistent pattern. Random rolling leaves some areas over-worked and others barely touched.
• Rolling stopped and started on the same spot. This can cause surface shearing and loosening.

How to fix it

Set a rolling pattern for each section, mark the lanes, record the passes, and brief operators on target speed and vibration settings. Intelligent or continuous compaction control systems, where available, take much of the guesswork out of this.

Machine condition matters too. Worn drum surfaces, weak vibration mechanisms or a roller that has lost vibratory output won't deliver the energy the spec assumes, so check your equipment regularly.

6. The soil itself is variable or doesn't match the lab sample

The Proctor test was done on one sample. The site rarely has one uniform soil.

Problems that show up in the field:

• Mixed material. Pockets of clay, silt and gravel in the same layer compact differently.
• Oversize particles. Large stones or lumps bridge across each other and leave voids around them.
• Segregation. Coarse material separates during hauling and dumping.
• Different borrow sources. The MDD and OMC change from one source to another.

If the MDD used for calculating compaction percentage doesn't represent what's actually on site, the test can fail even when the soil is well compacted, or pass when it shouldn't.

How to fix it

• Re-run Proctor tests whenever the material source or type changes.
• Remove or break up oversize material.
• Blend material properly before spreading.
• Keep the same borrow source within a layer wherever possible.

7. The test itself is the problem

Sometimes the compaction is fine and the test result is what's off. Before you tear anything up, check the testing side:

• Wrong reference density. Standard and Modified Proctor give very different MDD values. Confirm the right one is being used.
• Test depth. A test that samples the wrong depth, or includes loose surface material, gives a misleading number.
• Test timing. Surfaces dry out after rolling. Test too late and the top can appear loose or disturbed.
• Equipment errors. A nuclear density gauge out of calibration, or a sand cone with the wrong calibrated sand density, will give unreliable readings.
• Poor sampling spots. Testing only the best-looking areas hides weak ones. Testing only the worst areas does the opposite.

How to fix it

Make sure the method, reference density and frequency match the specification, calibrate instruments regularly, and cross-check a failing result with a second method or a retest nearby before ordering rework.

A quick diagnostic checklist for when a test fails

Before adding more passes, run through these questions in order:

• Is the moisture content within about ±2% of OMC?
• Is the loose lift thickness within what the trial strip proved?
• Is the roller type right for this soil?
• Is the layer below firm, with no deflection under a proof roll?
• Were speed, overlap and vibration settings correct?
• Does the soil match the Proctor sample being used as reference?
• Is the test method, calibration and location correct?

In most failed-compaction cases, at least one of these has an answer of "no".

Final thoughts

Repeated roller passes feel like action, but compaction is a system. Moisture, lift thickness, equipment, the ground below, operating technique, soil variability and testing all have to line up. Fix the weak link and the numbers usually follow, often with fewer passes than before.

The most cost-effective habit on any site is to do a compaction trial at the start. Prove the moisture, lift thickness, roller type and pass count that actually hit the target density, then repeat that recipe. It costs a little time up front and saves rework later.

For help selecting the right compaction equipment for your soil and site conditions, you can explore the range at Sona Vibrators.