When asphalt cores fall below the specification, the cause more often lies in the paving day than in the test: the mix cooled before rolling, the lift was too thin for the aggregate, a joint edge was left unsupported, or the reference density did not represent the mix. This guide explains how the degree of compaction of asphalt is defined, why it matters and how low results can be avoided. The test methods are covered in our guide to asphalt coring and hot-mix testing; for soils, where the Proctor test is the reference, see sand cone test and degree of compaction.

Two references on two scales

Every core result starts with the bulk density of the core, determined in the laboratory to TS EN 12697-6:2020. It is compared with one of two references, and the specification must say which:

Percentage of the Marshall density. Degree of compaction (%) = core bulk density / bulk density of the Marshall specimens × 100. The specimens are compacted to TS EN 12697-30:2019 with the number of blows the specification sets, from the approved mix design or, better, from hot mix sampled on the day.
Percentage of the maximum density (air voids). The maximum density is that of the mix with no air voids, determined to TS EN 12697-5. Air voids (%) = (maximum density − bulk density) / maximum density × 100, as defined in TS EN 12697-8.

The two scales differ because a Marshall specimen is not void-free. For example, if the Marshall specimens contain 4% air voids, their density is 96% of the maximum density, and a core at 97% of the Marshall density is at about 0.97 × 96 ≈ 93% of the maximum density, that is, about 7% air voids. The same core can pass a “97% Marshall” limit and still sit close to an air-void limit. Our reports state the reference, its source and its value; where the specification uses the maximum density, the value to be used must be agreed in advance.

Reference	What 100% means	Can a result exceed 100%?
Marshall density	Density reached with the laboratory compaction effort	Yes, if site compaction exceeds the laboratory effort or the reference is unrepresentative
Maximum density	A mix with no air voids	No; it points to a wrong reference or a test error

Why density matters

High, connected air voids let water and air into the layer. Water weakens the bond between binder and aggregate, air ages the binder faster, the surface starts to ravel and fatigue life drops. Published reference points, each valid in its own context:

NCHRP Report 531 (Brown, Hainin, Cooley and Hurley, 2004) found in-place air voids to be the most significant factor in permeability and recommended about 6–7% or lower.
Roberts et al., Hot Mix Asphalt Materials, Mixture Design, and Construction (2nd ed., NAPA, 1996), state that dense-graded mixes should neither exceed 8% nor fall below 3% air voids during their service life; too few voids invite rutting and bleeding.
Linden, Mahoney and Jackson (1989) proposed a rule of thumb that each 1% of air voids above about 7% costs roughly 10% of pavement life: one study’s estimate, to be used with caution.

Temperature, lift thickness and aggregate size

Asphalt can only be compacted while the binder is fluid enough for the aggregate to move. A low delivery temperature, a delay before the first roller pass, wind and a cold or damp base all shorten that window, notably on winter days in Cyprus. Thin layers cool fastest; NCHRP Report 531 notes that for thicker sections cooling is typically not a problem. Minimum temperatures come from the specification and the binder supplier, and the contractor records them.

For lift thickness, NCHRP Report 531 recommends at least 3 times the nominal maximum aggregate size (NMAS) for fine-graded mixes and at least 4 times for coarse-graded mixes and SMA. For an EN mix, the upper sieve size D in the designation (for example AC 16 under TS EN 13108-1) is a working proxy for the NMAS, although the two are not defined identically. A 40 mm layer of a coarse-graded AC 16 is only 2.5 times D: below the recommended ratio, and harder to compact.

Rolling, edges and joints

Breakdown rolling close behind the paver. The first passes, while the mix is hottest, do most of the work.
A rolling pattern set on a trial section. Passes, roller types and their order are best fixed on a trial section checked with cores.
Care with vibration. On thin lifts or a cooled mat, vibration can crush aggregate or loosen the layer instead of compacting it; many specifications restrict it.
Edges and longitudinal joints. An unsupported edge spreads under the roller, and the cold side of a joint cools before the next lane is laid. These are usually the lowest-density points of a pavement, so joint compaction needs its own method.

Causes in the mix and in the reference

Low density is not always a rolling problem. Grading or binder content away from the job mix formula (JMF), checked to ASTM D2172/D2172M-26 and TS EN 12697-2:2025, changes how the mix compacts; a mix short of binder, for example, is stiffer and harder to compact. Segregated coarse patches, often at the end of a lorry load or at the paver wings, have an open texture and low density. A mix with modified binder needs higher temperatures and an adapted rolling pattern.

A result above 100% of the Marshall density is a warning, not a bonus. Usually the reference does not represent the mix: the specimens came from another day or mix, were compacted at the wrong temperature, or the mix has changed since the design. Against the maximum density, a result above 100% is physically impossible.

How cores are chosen and tested

Random or targeted points. Acceptance cores are taken at random points within each lot defined by the specification. Cores at a visible defect or on a longitudinal joint answer different questions and are reported separately.
The right bulk density procedure. TS EN 12697-6:2020 offers dry, saturated surface-dry (SSD), sealed-specimen and dimensional procedures, and its Annex A guides the choice by void content and surface texture (the 2020 edition limits the dry procedure to specimens with up to 7% voids). On open-textured or high-void cores, water drains from surface-connected voids during the SSD procedure, so the density comes out too high; the sealed procedure is then appropriate. The report states the procedure.
Sound cores only. A cracked or deformed core, or one whose layers have separated, gives no valid density; we note it and recommend a replacement. Cores cut with water are dried to constant mass, and each layer is sawn off and tested separately.

When cores fail

One low core does not make a failed road, and one good core does not clear it. When results fall below the limit:

define the extent with additional cores around the weak point and across the lot, to separate local defects (joint, edge, segregation) from a systematic problem;
check the reference: which density, from which day and mix, and whether the same bulk density procedure was used;
read thickness and density together, as thin areas are harder to compact;
on the next paving days, take hot-mix samples for a daily Marshall reference, as described in our guide to asphalt coring and hot-mix testing.

What follows, whether conditional acceptance, further investigation or remedial work, is decided by the authority and the supervising engineer under the contract.

How we can help

We take asphalt cores with water-cooled diamond drills, measure layer thickness to TS EN 12697-36:2022 and bulk density to TS EN 12697-6:2020. During paving we sample hot mix to TS EN 12697-27:2018, compact Marshall specimens to TS EN 12697-30:2019 as a daily reference, and check binder content and grading against the JMF. Every report carries a verification code (see how to read and verify a test report). The scope is on our asphalt coring and hot-mix testing service page; send us your specification and paving programme through our contact page.

Frequently asked questions

Our cores reach 97% of Marshall, yet the air voids are said to be too high. How?

Both can be true: a Marshall specimen already contains voids, so 97% of the Marshall density can mean about 7% air voids or more (see the example above). Check which reference and limit the specification uses.

How many cores are needed?

The specification sets the frequency and lot size. In a dispute, enough cores are needed to define the extent of the problem; we report each core individually.

Can low density be fixed by rolling again?

Only while the mix is still hot enough to compact. Rolling a cooled layer adds little density and can damage it; any further measure is for the authority and the supervising engineer.