From Hubbard to Marshall
By Gerald Huber, P.E.
Superpave, currently the most common method of asphalt mix design in North America, was developed in the early 1990s as part of the Strategic Highway Research Program. Superpave was not entirely new. The method draws upon history and incorporates new information. To understand current mix design, it is important to understand the development of mix design technology.
Early mix design methods
In 1890, E.G. Love published a series of articles on roads and paving. These articles were not technical but were similar to articles in current trade magazines. The articles contained ideas for designing a pavement. One article by F.V. Greene of the Barber Asphalt Paving Company was a specification for the construction of an asphalt pavement. Design technology was not discussed, but a recipe was given for the asphalt surface. A Barber wearing surface was specified as follows:
Asphalt cement 12 to 15%
Sand 70 to 83%
Pulverized carbonite of lime 5 to 15%
Mix was laid in two lifts. The first lift, called the cushion coat, contained 2 to 4 percent more asphalt and was compacted to a depth of one-half inch. The surface coat was made according to the specifications above. The lime was added cold to the hot (300º F) sand before the asphalt was mixed in. The quantity of lime was adjusted according to the properties of the sand. Proportions were adjusted based on visual observation of experienced personnel.
In 1905, Clifford Richardson, owner of the New York Testing Company published “The Modern Asphalt Pavement.” The 1912 second edition refers to many pavements built across the United States in the 1890s and 1900s. Richardson describes two types of asphalt mixes: surfacing mixtures and asphaltic concrete.
Surfacing mixture is a sand mix. Typical gradations are 100 percent passing the #10 sieve and 15 percent passing the #200 sieve. Asphalt contents are 9 to 14 percent. He discusses the ability of sand to carry asphalt and the calculation of the area of spherical particles. The asphalt content of these mixtures was determined by the “pat-paper test” (asphalt stain on paper), as shown in Figure 1.
In doing the pat-paper test, Richardson warns that the mixture must be sufficiently hot for the asphalt to be fluid. Cold mixtures are of no use and excessively hot ones may cause too much staining. Although he does not describe the test method in detail, the streaks on the paper suggest that the mixture is spilled onto the paper.
Asphaltic concrete is used for lower courses. Richardson warns that asphaltic concrete is not suitable as a surface layer on main streets but may be suitable for lesser streets. Horse’s shoes and hoofs ravel particles from the surface. In his opinion, the high asphalt content sand mix must be used to resist the impact of horseshoes.
Asphaltic concrete is more like current HMA. A cross section of asphaltic concrete is shown in Figure 2. Interestingly, the design of this mixture did not use the pat-paper test. Instead, Richardson calculates the voids in mineral aggregate. In fact, he refers to it as VMA.
Richardson describes adjusting the VMA to include the correct amount of asphalt. The gradation shown in the photo is similar to a pavement Richardson used in Michigan that was as follows:
1.5 inch 100%
1 inch 83.6%
½ inch 50.1%
¼ inch 40.3%
#8 36.8%
#200 5.2%
VMA 13.2%
Bitumen 7.4%
Under today’s specifications this mixture would be a 1.5-inch nominal maximum size mixture. It is a fine-graded mixture because the percent passing the primary control sieve (3/8-inch sieve which is not shown in the table) is above 40 percent. The VMA requirement of modern specifications is 11.0 percent that is 2.2 percent less than the VMA in Richardson’s mixture. This means the asphalt content would be about 0.9 percent lower than Richardson used.
Air voids are not calculated as part of Richardson’s mix design but he analyzed several pavements in his book and talks about the correct level of density as compared to the theoretical density. By calculation, the air voids are deduced to be about 2 percent. Note this is the in-place air voids. If air voids were higher, say 5 to 8 percent, Richardson commented that the pavements were unable to withstand thermal shock and would crack.
The key idea evolving from pavement design at the beginning of the 20th century was the concept of using an asphaltic concrete as the base layers with a sand asphalt mix as the surface.
Hubbard Field mix design
In the mid-1920s, Charles Hubbard and Frederick Field, with the newly created Asphalt Association (later the Asphalt Institute), developed a method of mix design called the Hubbard Field Method of Design. The Hubbard Field method was commonly used among state highway departments in the 1920s and 1930s although use continued on into the 1960s in some states.
Initially, the Hubbard Field method focused on the surfacing mixture, the sand asphalt-wearing course. Specimens were 2 inches in diameter and were compacted with a hand rammer.
A modified Hubbard-Field version was developed for asphalt concrete. It used 6-inch diameter specimens that were compacted with two different rammers. First 30 “heavy blows” were applied with the 2-inch rammer followed by 30 blows with a 5.75-inch rammer. The specimen was turned over and pushed to the opposite end of the mold. Again 30 blows of the 2-inch rammer were applied followed by 30 blows of a 5.75-inch rammer. The specimen was then placed in a compression machine and was loaded with a 10,000-pound load and was allowed to cool in a cold water bath under compression.
The Hubbard Field method built upon Richardson’s process. Specimens were made in the laboratory but instead of using a paper stain test, they developed an evaluation method to determine design asphalt content. Bulk specific gravity of the compacted specimens was measured. Maximum theoretical specific gravity was computed using aggregate bulk specific gravity (Note that asphalt absorption was therefore not accounted for.) Air voids were calculated as were voids in the aggregate skeleton (VMA by today’s terminology). So, the volumetric analysis was similar to the properties used today.
In addition to the volumetric analysis, the Hubbard Field method used a stability test where the compacted mix is squeezed through a ring slightly smaller than the specimen diameter. The peak load sustained before the mix started flowing through the orifice was called the Hubbard Field stability. In concept, this is identical to Marshall Stability where the specimen is loaded on its side and the peak load is the Marshall stability.
The Hubbard Field method selected asphalt content based on air voids and stability. Voids in the aggregate were evaluated to help adjust the mixture stability.
Hveem mix design
Early pavements in California were made using natural bitumen from the La Brea Tar pits located in the Los Angeles and Santa Barbara area. Although referred to as tar, these were actually natural asphalt seeps.
This asphalt was quite soft and was used in the role of penetrating macadam, in which it was sprayed on top of compacted open-graded aggregate, or it was used by mixing with gravel and making an oil mix.
In the 1920s, oil mix made with cutback asphalt was a common method of paving. It was mixed in windrows with the asphalt sprayed on top of a knocked-down windrow and mixed back and forth with a motor grader. Oil content was determined by eye, so an experienced person was needed to ensure that the mix had the proper brown color.
In 1927, Francis Hveem became a resident engineer in California, and having no experience with oil mixes, used the information about gradation with the paper stain test to evaluate asphalt content. He recognized this process was controlled by aggregate surface area and found a method to calculate surface area. He used surface area factors published in 1918 by a Canadian engineer, Captain L.N. Edwards, which were proposed for use in Portland cement concrete design.
Francis Hveem applied the design process used for oil mixes to hot mix asphalt. By 1932 he had developed a method to determine asphalt content based on surface area. He continued to make changes to the surface area factors and developed a test using motor oil to estimate asphalt absorption. The surface area factors in today’s Asphalt Institute manual MS-2 for Hveem mix design are those developed by Hveem for the California Department of Highways in the 1940s.
Hveem started developing a stability test. He recognized that mechanical strength of the mix was important and developed the Hveem stabilometer, which is a pseudo-triaxial test. A vertical load is applied to a confined specimen and the resulting horizontal pressure is measured. When asphalt content exceeds a threshold, the horizontal pressure increases, and Hveem used this property to discern stable and unstable pavements. Based on oil mixes, he developed threshold values for stability and applied them to HMA.
Hveem’s mix design philosophy is that sufficient asphalt binder is needed to satisfy aggregate absorption and to have a minimum film thickness on the surface of the aggregates. In order to carry load, the aggregates had to have a sliding resistance (measured by the Hveem stabilometer) and a minimum tensile strength to resist turning movement (measured by the cohesiometer). Stability and cohesion were influenced by the aggregate properties and the amount of asphalt binder. For durability, Hveem developed the swell test and moisture vapor sensitivity test to measure the reaction of the mix to water. The swell test used liquid water, and the vapor sensitivity test used moisture vapor. The effect on Hveem stability after conditioning was measured. Hveem found that thicker asphalt films had more resistance to moisture.
Air voids are not part of Hveem’s mix design system. He believed that film thickness and mechanical properties as described by stability were most important. In the 1980s or ‘90s, air voids were added as a consideration. Interestingly, if one looks at performance of HMA in the 1980s or early 1990s when rutting was a huge national problem and compares the general performance of Hveem mixes with Marshall mixes, a general statement could be made that Hveem pavements had lower asphalt contents and fatigue cracking was a major concern. It is not a coincidence that fatigue cracking research and beam fatigue is associated with research at the University of California Berkeley. In the Marshall states, fatigue cracking was not a predominant problem; rutting was the issue.
Marshall mix design
Bruce Marshall of the Mississippi Department of Highways developed Marshall mix design in the late 1930s to early 1940s. In 1943 Marshall approached the Corps of Engineers in Vicksburg, MS about using the Marshall method of design and was hired. The Corps adopted Marshall’s system in World War II for use on airfields. Post WW II, it was “civilianized” for use by state highway departments.
Marshall mix design is essentially an outgrowth of the Hubbard-Field method. The approach is similar although the practice was different. Hubbard-Field used two different sized rammers to compact samples. Marshall used one hammer and matched the compactor diameter to the mold diameter. Hubbard-Field had used a hand-tamp rammer. Marshall standardized the compaction energy applied by using a drop hammer.
Marshall included calculation of air voids from Hubbard-Field but not VMA. Instead, he used voids filled with asphalt as a criterion. In the 1950s, Norman McLeod advocated use of VMA in the mix design method. Presumably, he was aware of VMA in the Hubbard-Field method and believed it should apply to the Marshall method.
In the 1950s and 1960s, the Asphalt Institute was the de facto keeper of the Marshall standard and published it in “MS-2, A Manual of Mix Design Methods for Asphalt Concrete.” Although ASTM was the main home of the Marshall method (D-1889), the method was a reflection of MS-2. Even AASHTO, which adopted its own standard, mirrored MS-2. As a result ASTM and AASHTO had methods for Marshall mix design but the properties specified within them had been established by the Asphalt Institute from research and technical debates. Files at the Asphalt Institute contain letters and data from Marshall, who became a consultant after leaving the Corps, and McLeod, who worked for Imperial Oil in Canada.
Marshall was against inclusion of VMA; McLeod favored including it. McLeod’s most notable research papers on VMA are a 1956 Highway Research Board paper, a 1957 AAPT paper and a 1959 ASTM symposium paper. Other papers argued in favor of film thickness. Particularly, L.C. Krchma argued for film thickness in AAPT and Highway Research Board proceedings.
McLeod’s original work considered using one level of VMA for all mixes. This was later changed to a sliding scale based on aggregate nominal maximum particle size. The need for additional asphalt binder as mixture size became smaller was recognized, but there was no direct connection between surface area and VMA criteria.
In 1962, after much debate, the Asphalt Institute changed MS-2 to include VMA as a mix design criteria. AASHTO and ASTM changed their standards to reflect the Asphalt Institute revision.
The Marshall and Hveem mix design procedures served as the primary means of designing dense mixtures until the mid-90s, when the Superpave procedure was introduced.
Gerry Huber is the Associate Director of Research for Heritage Research Group.
Read the second part of the History of Mix Design