Asphalt Mix Design: The Basics for Buyers

Hot mix asphalt is an aggregate skeleton held together by a thin film of binder. By mass the proportions look lopsided: a dense-graded paving mixture is typically 94 to 96 % aggregate and 4 to 6 % binder, a description of ordinary practice rather than a specification limit. By volume the picture is quite different, and volume is the language the design is written in. Take a mixture at 5.0 % binder by total mass, with a compacted bulk specific gravity of 2.40 and a binder specific gravity of 1.03. The binder then occupies 0.050 × 2.40 / 1.03 = 0.117 of the compacted volume, or about 11.7 %. That one arithmetic step — mass to volume, through the specific gravity of the binder — is the hinge the whole discipline turns on, and it is why the Certificate of Analysis is a laboratory input and not merely a commercial formality.

The four things a mix design fixes

• The aggregate blend and its gradation. Which stockpiles, in what proportions, producing what percentage passing each sieve.
• The binder grade. Named by the project specification, or selected for the climate and the traffic. The design confirms it; it does not discover it.
• The binder content. A single figure, to one decimal place, that satisfies the volumetric criteria at the design compaction effort.
• The compaction effort those numbers were verified at. A blow count in the Marshall system, a gyration count in Superpave. Volumetric results quoted without the effort that produced them mean nothing.

The output document is the job mix formula, usually written JMF: the target gradation with its production tolerances, the target binder content with its tolerance, the binder grade, and the volumetric properties the plant is expected to reproduce. From approval onwards the JMF is a contract document. Production is measured against it and the paved work is accepted against it.

What a mix design is not

It is not a quality certificate for the binder. The design assumes the binder meets its specification; it does not verify that the cargo you bought matches its COA, and it will not rescue an off-specification binder. It is not transferable — a design belongs to a named set of stockpiles, at a named plant, with a named binder source. And it is not a performance prediction. Neither system in common use runs a test that models traffic on a pavement. Both control the composition of the mixture and rely on long correlation between composition and field behaviour.

Why the buyer has a stake in it

Because the JMF is the only thing that converts tonnes of asphalt into tonnes of bitumen. Ten thousand tonnes of mixture at a JMF binder content of 5.2 % is 520 tonnes of binder; at 4.8 % it is 480 tonnes. That forty-tonne difference is not a negotiating position between buyer and supplier. It is the result of a calculation about void space that the laboratory completed before anyone asked for a price. A buyer who understands where the number came from asks better questions, spots a weak design report, and stops wasting effort on the one line item that cannot be squeezed.

One point of vocabulary before going further, because it causes genuine confusion in export correspondence. In North American usage asphalt often means the binder itself. In British and most export usage, bitumen is the binder and asphalt is the finished mixture of binder and aggregate. Throughout this page, binder means bitumen and mix means the compacted mixture; the conventions are set out in full on bitumen vs asphalt terminology.

The Marshall method: compaction by impact, judged on stability and flow

The method dates from the 1940s and remains the reference procedure across a large part of the export paving market. A specimen 101.6 mm in diameter and 63.5 mm high is compacted by dropping a 4,536 g hammer through 457 mm onto each face — 35, 50 or 75 blows per face depending on the design traffic category. Specimen preparation is covered by ASTM D6926 and the test itself by ASTM D6927; the older designation ASTM D1559 has been withdrawn, so a specification still calling for D1559 is out of date. AASHTO T 245 and EN 12697-34 are the parallel procedures.

The specimen is brought to 60 °C in a water bath for 30 to 40 minutes, laid on its side in a curved breaking head and loaded at 50.8 mm per minute until it fails. Two numbers come out of that:

• Marshall stability — the maximum load the specimen carried, in newtons or pounds-force. It is an empirical index of resistance to deformation at one elevated temperature. It is not a fundamental engineering property, it does not convert into a pavement stress, and a very high stability is not a better result — it usually means a harsh, brittle mixture.
• Marshall flow — the vertical deformation of the specimen at the instant of maximum load, reported in units of 0.25 mm. High flow points to a mixture that is too plastic, generally from excess binder or too much rounded natural sand. Low flow with high stability points to brittleness.

The criteria most often quoted alongside the method are the Marshall design criteria published by the Asphalt Institute in MS-2, reproduced with local amendment in many national road specifications:

• Heavy traffic — 75 blows per face, stability minimum 8,006 N (1,800 lbf), flow 8 to 14 in units of 0.25 mm, VFA 65 to 75.
• Medium traffic — 50 blows per face, stability minimum 5,338 N (1,200 lbf), flow 8 to 16, VFA 65 to 78.
• Light traffic — 35 blows per face, stability minimum 3,336 N (750 lbf), flow 8 to 18, VFA 70 to 80.
• Design air voids 3 to 5 % at every traffic level, with minimum VMA set by nominal maximum aggregate size on the same scale used in the Superpave system.

Two honest qualifications. First, the current edition of MS-2 is built around Superpave and carries the Marshall material alongside it, so where these Marshall criteria are enforced today they are generally enforced through a national specification that adopted them, and the version binding on your project is the one in the project documents, not the figures above. Second, and far more important to a buyer: the Marshall procedure selects the binder content by preparing specimens at four or five binder contents in 0.5 % steps, plotting stability, flow, unit weight, air voids and VFA against binder content, taking the binder content that gives 4 % air voids, and then confirming that every other property is inside its limits at that binder content. The binder content is read off the air void curve. Stability and flow are checks on the answer, not the source of it. That is the most widely misunderstood point in the whole method.

Superpave: compaction by kneading, judged on volumetrics alone

Superpave emerged from the Strategic Highway Research Program in the early 1990s and replaced impact compaction with a gyratory compactor that kneads the mixture in a manner closer to a roller. The machine is specified in AASHTO T 312 and ASTM D6925: a 150 mm mould, ram pressure 600 ± 18 kPa, 30.0 ± 0.5 gyrations per minute, and the mould held at 1.16 ± 0.02 degrees measured internally. The earlier specification was 1.25 ± 0.02 degrees measured externally, which is worth knowing when comparing gyratory data of different vintages. Specimen height is logged at every gyration, so the compactor produces a densification curve rather than a single end point.

Compaction effort is tied to design traffic in equivalent single axle loads over a twenty-year design life. AASHTO R 35 and AASHTO M 323 give three gyration counts per traffic band:

• Below 0.3 million ESALs — N-initial 6, N-design 50, N-maximum 75.
• 0.3 to below 3 million ESALs — N-initial 7, N-design 75, N-maximum 115.
• 3 to below 30 million ESALs — N-initial 8, N-design 100, N-maximum 160.
• 30 million ESALs and above — N-initial 9, N-design 125, N-maximum 205.

The mixture is designed to arrive at exactly 4.0 % air voids at N-design. The other two counts are traps, and they reveal what the system is really worried about. Density at N-initial is capped — at 91.5 %, 90.5 % or 89.0 % of the maximum theoretical specific gravity, tightening as traffic rises — because a mixture already nearly compacted after six to nine gyrations is a tender mixture that shoves under the roller and ruts under traffic, usually because it is over-sanded or built on rounded aggregate. Density at N-maximum must remain at or below 98.0 %, because a mixture that can be pushed past 98 % in the laboratory will be pushed there by traffic, and at that point there is no air left for the binder to expand into on a hot day.

There is no strength test anywhere in Superpave volumetric design. Buyers arriving from a Marshall background look for a stability figure on the report and do not find one. Performance is addressed in two other places instead: in the selection of the performance-graded binder for the climate and the traffic, and in the aggregate consensus properties — coarse aggregate angularity to AASHTO T 335, fine aggregate angularity to AASHTO T 304, flat and elongated particles to ASTM D4791, and sand equivalent to AASHTO T 176. Those limits tighten with traffic and with proximity to the surface: at the highest traffic level within 100 mm of the surface, AASHTO M 323 requires 100 % of coarse particles with two or more fractured faces, fine aggregate angularity of at least 45, no more than 10 % flat and elongated particles at a 5:1 ratio, and a sand equivalent of at least 50.

Different routes, the same destination

Strip away the machinery and both systems do the same three things. They build an aggregate skeleton that carries load through stone-to-stone contact. They leave a controlled amount of room between the stones. They fill that room with enough binder to be durable, but not so much that the mixture loses stability when it is hot. Marshall adds an empirical strength check and gets its density from impact; Superpave uses a more realistic compaction mechanism, scales the effort to traffic more finely, and rests everything on volumetrics and binder selection. Designed competently under either system, the same aggregates will land at a broadly similar binder content — because the volume that has to be filled is a property of the stone, not of the machine that squeezed it.

The chain of reasoning, in order

Every link is fixed before the binder content appears at the end.

• The aggregate blend and the compaction effort are chosen first. Together they decide how tightly the stones pack, and therefore how much total void space exists between them once compacted. That space is the VMA.
• The specification imposes a minimum VMA for the nominal maximum aggregate size — 14.0 % at 12.5 mm, 13.0 % at 19.0 mm and so on under AASHTO M 323 — because a skeleton with too little room in it cannot hold a durable binder film however much binder you add.
• The specification also fixes the design air void content: 4.0 % at N-design under AASHTO M 323, and 3 to 5 % as the general design range for dense-graded mix.
• Whatever is in the VMA and is not air must be effective binder. If VMA is 14.0 % and air voids are 4.0 %, then effective binder occupies 10.0 % of the volume of the compacted mixture, and VFA is 10.0 / 14.0 = 71 %.
• Convert that volume back into a mass through the specific gravity of the binder, add whatever the aggregate absorbs into its pores, and the result is the total binder content that goes onto the purchase order.

At no point in that sequence does anybody express a preference. The binder content is simply the value that closes the equations. It is an output.

What happens arithmetically when you cut it

Binder volume and air volume are the same volume, which makes the arithmetic unforgiving. Using the figures from earlier — compacted bulk specific gravity 2.40, binder specific gravity 1.03 — removing 0.5 % of the mixture mass as binder removes 0.005 × 2.40 / 1.03 = 0.0117 of the mixture volume, that is 1.17 percentage points. The mixture does not shrink to compensate. That volume becomes air. Design air voids move from 4.0 % to roughly 5.2 %, and VFA drops from 71 % to about 63 % — below the AASHTO M 323 minimum at every traffic level. Nobody has overruled the mix designer. The mixture has simply left the specification.

Plant laboratories work to a rule of thumb that 0.1 % of binder content moves air voids by roughly 0.2 to 0.3 percentage points. That is a rule of thumb rather than a standard requirement, but it is consistent with the arithmetic above and it is worth carrying in your head, because it shows how little slack exists. A binder content two tenths of a percent low is not a rounding error; it is half a percentage point of air voids.

The failure mode is ravelling, and it arrives early

An under-asphalted mixture fails through its surface rather than its structure, which is why the damage appears long before anyone expects it. Four mechanisms compound:

• Thin binder films. The effective binder has to coat every particle in the blend, and the fine aggregate and filler carry the great majority of the surface area. Cut the effective binder and the film thins everywhere. The bond holding surface stones in place weakens, traffic plucks them out, and the surface starts shedding aggregate — ravelling. It typically shows in the wheel paths within the first winters, not at the end of design life.
• Permeability. Above a threshold, the air voids in a dense-graded mixture stop being isolated bubbles and connect into channels. NCHRP Report 531 examined this directly and found permeability governed jointly by air void level, nominal maximum aggregate size and lift thickness, with coarse-graded mixtures becoming permeable at lower air void contents than fine-graded ones. There is no single universal void figure at which a mixture turns permeable, which is precisely the problem: it depends on the mixture, and a dry mixture is likelier to cross it.
• Accelerated oxidation. A permeable mixture admits air throughout its depth rather than only at the surface, so the binder hardens through the whole layer. Penetration falls, stiffness rises, and the layer becomes brittle at exactly the moment it is already short of binder.
• Moisture damage. Water in the connected void structure attacks the binder-aggregate bond, and a thin film has less bond to lose. This is why AASHTO T 283 sits in the design procedure at all, and why anti-stripping treatment is a design decision rather than a site remedy.

The most-quoted field figure in this area comes from the research of Linden, Mahoney and Jackson (1989), who reported approximately a 10 % loss of pavement life for each 1 % increase in air voids above 7 %. That is a research finding widely cited in agency compaction specifications, not a standard requirement, and it was derived for in-place voids in the finished pavement. It is quoted here because it puts the commercial argument in one line: the arithmetic above showed how easily a design change of half a percent of binder becomes more than a full percentage point of air, and field compaction shortfalls add to that rather than cancelling it.

Over-asphalting fails just as certainly

The constraint is two-sided, and it is worth saying plainly because the argument above is not a case for more binder. Excess binder fills the VMA beyond the VFA limits, leaves too little air, and the mixture loses internal friction when hot. It ruts, it shoves under the roller, and binder migrates to the surface as flushing or bleeding, producing a slick black surface with reduced skid resistance. The Marshall system catches this as high flow and falling stability; Superpave catches it as excessive density at N-maximum. You would also be paying for binder that damages the pavement. A mix design is a narrow optimum, not a direction of travel.

Where the money actually is

If the binder bill is the problem, these are the honest levers, and none of them is the JMF binder content:

• The aggregate blend. A better-graded, less absorptive aggregate reaches the VMA minimum with less binder and absorbs less of what is added. Absorption is invisible on the purchase order and expensive in the tank.
• The mix type. Specifying stone mastic asphalt or an open-graded friction course where a dense-graded mixture would serve buys performance you may not need at a binder content one to two percentage points higher.
• Reclaimed asphalt pavement, where the specification permits it and the design accounts properly for the binder the RAP contributes.
• The binder purchase itself — grade, packing, Incoterm, lot size, timing and the quality of the documentation. This is the part that is genuinely negotiable, and it is where a buyer’s effort pays.

The JMF is a contract document

Production tolerance on binder content is a band of a few tenths of one percent around the JMF target, set by the paving specification rather than by the mix design, and it exists to absorb plant variability — not to authorise a deliberate offset. Running the plant at the bottom of the tolerance band as a matter of policy is a quiet way of building an under-asphalted pavement while every individual test still passes. Any real change to the binder content requires the mix designer to redo the volumetrics and the engineer to approve a revised JMF.

Grade selection is a climate and traffic decision, taken before the volumetrics

In the performance-graded system, the high-temperature grade comes from the seven-day maximum pavement design temperature and the low-temperature grade from the minimum, at a stated reliability. Traffic then adjusts the result: AASHTO M 323 directs the designer to increase the high-temperature grade by one grade for slow transient loading, by two grades for standing loading, and by one grade where design traffic reaches 30 million ESALs or more. In penetration-graded and viscosity-graded markets, the grade is named directly by the national road specification. Either way, the decision is made before the mix designer weighs out the first aggregate, and no volumetric result will change it.

The volumetric targets do not move with grade

The minimum VMA for a 12.5 mm mixture is 14.0 % whether the binder is PG 64-16 or PG 70-10, and the design air void target is 4.0 % either way. VMA is a property of the aggregate skeleton and the compaction effort; it is not a property of the binder. Two binders of similar specific gravity from a comparable source will therefore land the design at a broadly similar optimum binder content, and a buyer who expects a harder grade to reduce the binder bill will be disappointed.

That is not permission to substitute freely. Most paving specifications treat a change of binder source or grade as grounds for verifying the design again, and many require a new job mix formula. There are real reasons: absorption interacts with binder viscosity, compactability at a fixed gyration count can shift, and modified binders change the picture more substantially still. In stone mastic asphalt and open-graded mixtures the binder is doing structural work rather than simply gluing, so binder substitution there is a design question and not a procurement one.

Where the binder does enter directly: specific gravity

The specific gravity of the binder, Gb, appears in the calculation of the effective specific gravity of the aggregate, in the absorbed binder, in the effective binder content and hence in the dust proportion. It is the conversion factor between the mass you buy and the volume the design works in. It is measured by ASTM D70 or AASHTO T 228, a pycnometer method, and it must be reported with its reference temperature, because 15.6 °C and 25 °C are both in use and the values differ. A typical paving grade sits in the region of 1.01 to 1.06 at 25 °C, but a typical range is not a measurement: if the laboratory assumes a figure because the COA omitted it, the assumption propagates into VMA, into the absorbed binder and into the dust ratio, and nobody downstream can tell how far the numbers are out.

Where the binder enters again: mixing and compaction temperatures

The laboratory must mix and compact at defined temperatures, and so must the plant. The classical route is equiviscous, a criterion from the Asphalt Institute mix design method (MS-2) that Superpave practice adopted through AASHTO R 35: mixing at a binder viscosity of 0.17 ± 0.02 Pa·s and compaction at 0.28 ± 0.03 Pa·s, with the temperatures read from a viscosity-temperature relationship built from rotational viscosity measurements to AASHTO T 316 or ASTM D4402 and plotted per ASTM D2493. Note that these are viscosity targets, not temperatures: the temperature that satisfies them is a property of the particular batch. For an unmodified paving binder this works, and it is why a supplier is often asked for rotational viscosity at 135 °C and 165 °C rather than for a temperature range as such. The plant-side windows that follow from it are set out in the heating and temperature guide.

The equiviscous criterion must not be applied to modified binders. A polymer modified binder is not a simple Newtonian liquid at these temperatures, and forcing it onto the equiviscous criterion returns temperatures high enough to degrade both the polymer and the base binder. NCHRP Report 648 examined laboratory mixing and compaction temperatures for modified binders for this reason. For any modified grade, the supplier’s recommended mixing and compaction ranges govern, and they should arrive with the technical data sheet rather than being back-calculated by the laboratory. Separately, AASHTO M 320 and ASTM D6373 impose a handling ceiling of 3.0 Pa·s maximum at 135 °C on performance-graded binders, which exists so that the material can be pumped and handled safely at all.

Reclaimed asphalt pavement, and the one legitimate way to buy less binder

RAP carries aged binder, and to the extent that binder blends with the virgin binder it reduces the virgin tonnage required. This is the only route to a smaller binder order that does not damage the pavement, and it is a design decision rather than a purchasing one. The guidance in the appendix to AASHTO M 323 is tiered: below about 15 % RAP by mixture mass no change to the virgin binder grade is required; from roughly 15 to 25 % the virgin binder is selected one grade softer at both the high and the low temperature; above that, blending charts based on the recovered RAP binder properties are needed. Individual agencies set their own caps and their own rules, often lower. Ask the designer what RAP percentage the JMF assumes before you convert a mixture tonnage into a binder tonnage, because the two no longer scale together.

• Recycled asphalt pavement — why RAP moves the binder grade you must order, and the arithmetic for working out the virgin grade that lands the blend back on specification
• Warm mix asphalt — producing and compacting below conventional temperatures, and what that does to ageing, to the compaction window and to the binder you specify