U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations
This magazine is an archived publication and may contain dated technical, contact, and link information.
|Publication Number: Date: July/August 1998|
Issue No: Vol. 62 No. 1
Date: July/August 1998
The following is adapted from a paper prepared for the Sixth International Purdue Conference on Concrete Pavement Design and Materials for High Performance, Nov. 18-21, 1997. The entire paper was published in the conference Proceedings, which is available from Purdue University, and in the May 1998 issue of Concrete International, the official magazine of the American Concrete Institute.
As designers planning for the future, we must continually look back to where we have been - both the mistakes made and the lessons learned. It is amazing how often we appear to have "reinvented the wheel" or how we have duplicated experiences around the country and around the world. Duplication, however, is not without value because it provides verification that supports the logic of our design philosophy.
One unfortunate aspect of our computer generation is that the new engineering graduates have neither the opportunity nor the time to trace the chronology of pavement design evolution. Copies of reports written before 1977 are difficult to find. Most were printed in very limited numbers and have either been discarded with old files or have vanished into the boxes of old retirees. Most of the important basic research and development on pavements predates the age of computerization. The valuable experimental works and writings of Friberg, Teller, Bradbury, Westergaard, Childs, VanBreeman, and others are rarely mentioned in today's technical papers. Their works are described in more detail in the American Concrete Institute's Monograph No. 7 on Better Concrete Pavement Serviceability by E.A. Finney.1
Many people believe that the history of concrete pavements began in 1894 with the placement in Bellefontaine, Ohio. That pavement is still in use, and the American Concrete Pavement Association recently memorialized its builder, George Bartholomew, on the pavement's centennial. But, according to Blanchard's American Highway Engineers' Handbook of 1919, in 1879 in Scotland, a concrete was used with portland cement for binding. "The surface was very good, but when the road commenced to break, it went to pieces very fast."2
Blanchard goes on to say that the first portland cement concrete (PCC) pavement in the United States was put down in 1893 on South Fitzhugh Street in Rochester, N.Y., by J. Y. McClintock, Monroe county engineer. This was a section of portland cement grouted macadam, a forerunner of the modern concrete pavement of the Hassam type. The cost of this pavement was $1 per square yard (per 0.84 square meters).2
However, the pavement soon deteriorated. As McClintock described, "The piece of pavement laid developed irregular temperature cracks, and on one portion of it where the hacks stood in the shade of the court house, the horses would drill holes with their feet in kicking off flies, so that it soon became a question of how the pavement could be maintained. ... It was some 2Â½ years after the pavement was laid ... that it was deemed wise by the city authorities to cover the new portion of the roadway with asphalt."2
I see no reason to try to correct history. We can be content with knowing that the Bellefontaine pavement was the first long-lasting PCC pavement, and we can let the asphalt promoters revel in the Rochester pavement becoming in 1896 the first overlay of PCC pavement.
The first slabs were about 6 inches (150 millimeters) of uniform thickness and usually about 6 to 8 feet squared (1.8 to 2.4 meters on each side) or the dimensions that were compatible with the mixer capacity. As better concrete construction equipment was developed, slabs got longer and wider. Because joint edges became chipped and faulted, they were soon minimized to create ribbons of unjointed concrete that cracked transversely. As the width increased to handle two lanes of traffic, longitudinal cracking became prevalent. Soon thickened centerline, or keel-section, pavements were being tried.
In 1909, Wayne County, Mich., conducted a test of various surfaces that were being used - brick, granite, wood blocks, and concrete. They used a circular track with a "Paving Determinator," which consisted of an iron-rimmed wheel on one end of a 20-foot (6-meter) pole and steel horseshoes on the other end. As a result of the test, Wayne County built the first mile of rural pavement for automobiles. (As an aside, it is ironic that in 1995 the Pennsylvania Department of Transportation (DOT) published a report on tests conducted on a similar 16-ft [5-m] circular track. The purpose was to evaluate the damage mechanism, amount of damage, and the repair and prevention of damage from horses and buggies. They determined that they have 1,900 lane-miles [3,000 lane-kilometers] of damage that costs them $1 million to $3 million per year.3)
The Report of the 1914 National Conference on Concrete Roadbuilding contained over 260 pages of guidelines on all aspects of concrete pavement design and construction.2
Around 1917, dowels were used for the first time in Virginia.4 This led to the evolution of many different configurations of slab cross sections, jointing, and reinforcement schemes.
In 1921 and 1922, the Pittsburg, Calif., Road Test was conducted. It used surplus army trucks with solid tires to traverse the instrumented slabs of various configurations and reinforcement schemes.5
In 1922 and 1923, the Bates Road Test in Illinois subjected 78 different pavement sections to truck traffic. It showed the benefit of thickened edges and longitudinal centerline joints in reducing the amount of slab cracking. In addition, the superiority of concrete over brick and asphalt pavements was demonstrated, and the tests led to the first thickness equation (Older formula for corners) for concrete slabs.6
The results of the Pittsburg and the Bates road tests showed the advantages of using mesh that held the cracks together as the slabs were tested to destruction. Eventually, this reasoning was used to justify a 1- or 2-in (25- or 50-mm) decrease in concrete thickness for adequately reinforced slabs.7 (I must add that this justification was for low-speed roadways on which, upon failure, the reinforced slabs became articulated and remained passable.)
Over the years from the Bates Road Test to the late 1950s, the Bureau of Public Roads conducted many detailed measurements of pavement slab properties (moisture and thermal gradients, slab deflections under load, impact, load transfer devices, subgrade friction, etc.). These studies were published in Public Roads and were integrated by Westergaard and others to form our early slab design procedures. This type of work is still sponsored by the Federal Highway Administration (FHWA) today.
In 1950 and 1951, the Bureau of Public Roads (now FHWA) with the Highway Research Board (now the Transportation Research Board), several states, truck manufacturers, and other highway-related industries conducted Road Test One - MD just south of Washington, D.C. An existing 1.1 mi (1.8 km) of two-lane highway was carefully inventoried, instrumented, and traversed by 1,000 trucks per day. The results showed the value of good load transfer between slabs, the effects of speed and axle weights, and the problems caused by pumping. It produced the first dynamic wheel equivalence factors.8
By the mid-1950s, continuously reinforced concrete pavements (CRCP) started to gain in popularity because the design offered the benefit of eliminating joint distress. First looked at in 1923 by the Bureau of Public Roads, it was followed by the Stilesville project in 1938, Vandalia in 1947, and then many experimental miles in Maryland, Pennsylvania, and other states in the early 1950s. The cost of the steel in CRCP was expensive, and so, to be competitive, CRCP was built 1 to 2 in (25 to 50 mm) thinner, leading to premature distress. (The justification perpetuated for using thinner structural slabs are many but are primarily related to comparisons of deflections with jointed pavements that had poor load transfer between slabs.)
Also in the 1950s, the slipform paver came into use. It reduced paving trains from 100 workers down to about 25. Also the economics changed in that materials became cheap and labor was more expensive. Hence, this led to a return to uniform thicknesses that could be easily placed by the early slipforms. By the mid-60s, the last states dropped thickened edge pavements.9
The AASHO (American Association of State Highway Officials) Road Test was conducted at Ottawa, just south of Chicago, from 1958 to 1960. Six loops of pavement were traversed by controlled truck traffic as part of a statistical factorial design. The construction control at this test was a demonstration of all that had been learned about reducing the variabilities inherent in concrete production and pavement construction. This $27 million experiment yielded the best information ever developed on pavement, including the AASHTO (American Association of State Highway and Transportation Officials) Pavement Design Procedures based on pavement serviceability and performance concepts. Among the many findings were the demonstration of the value of properly graded granular subbases and properly doweled joints.10
Unfortunately, the road test construction techniques used in 1958 were already being made obsolete! The interstate construction era began in 1956 and gave great impetus to slipform paving technology. Emphasis shifted to speed of construction, which led to the compromise of good concreting technology.
Prestressed concrete was introduced in the late 1940s and was used first in airport pavements. About 1959, two-way prestressed slabs were used at Biggs military airfield in Texas. The 24-in (610-mm) plain pavement was replaced with 9-in (230-mm) post-tensioned slabs. Unfortunately, the fear of the unknown, the need to use more skilled labor, and the reluctance of mile-a-day slipform contractors to embrace this unproven technology have held this concrete-saving technology back. About a dozen highways with prestressed concrete pavements of various designs were built in the United States between 1970 and 1990.
A considerable amount of research and development has been done since the AASHO Road Test, and it is too prolific to list. Much of this is available in reports by FHWA, the Portland Cement Association (PCA), and the Transportation Research Board, among other sources. In addition to the many studies using accelerated testing facilities, a tremendous amount of data is being collected through the Long-Term Pavement Performance Program (LTPP) studies. Other continuing significant efforts are the Minnesota Test Road and the WesTrack (presently testing flexible pavements) experiments.
Considerable research continues on rigid pavements. Much of it consists of developing better information for inputs to pavement management systems, comparing the performance of alternative designs under dynamic loads, finding solutions to durability problems, and developing more economical ways of recycling/reconstructing old pavements.
About 1970, the University of Texas had a long-running study to use computer technology to analyze the dynamic behavior of pavements. At that time, Dr. Ron Hudson said that someday he would be able to conduct an AASHO Road Test in the computer and would never need to conduct a field test. I believe that on the computer technology side, we are approaching the possibility of reaching such a goal. On the data input and theory side, we still have a need to improve many of the basic relationships that we use. We also need to be aware of the tolerances with which we measure or quantify the data we use. Over time, details on measurement methods and variability are lost, and the results are generalized to the point where only averages are considered. Some examples of variability that confound our experimental results follow.
It is convenient to look at the AASHO Road Test data because the results were so well-published. At the AASHO Road Test, the elastic modulus of subgrade reactions k gross (obtained with 30-in- [760-mm-] diameter plate) in units of pounds per cubic inch (pci) is taken to be an average 60 pci (16 kPa/mm). Actually, the values of k gross on the subgrade ranged from 28 to 56 pci (7.6 to 15.2 kPa/mm) over the 1Â½ years of the study with no measurements made over the winters. The values over the subbase ranged from 45 to 80 pci (12.2 to 21 kPa/mm).
A project in New York serves as another example of an actual highway construction project. Jim Bryden of the New York DOT measured the k values repeatedly on the Catskill-Cairo Test Road. It is a 7.5-mi- (12-km-) long, four-lane divided highway. He stated, "Measured modulus values range from 100 to over 2500 pci [27 to 680 kPa/mm]." In his Table 12 for the plate-bearing test on the "granular subbase, the average k was 830 pci [225 kPa/mm] with a standard deviation of 888 pci [242 kPa/mm], based on 63 tests." Other data items show just as much variability. Bryden concluded: "Several factors probably contribute to the wide scattering of values obtained from the test, the most obvious being non-uniformity and discontinuities in the subgrade. The lower variation in modulus values of the subgrade cut sections supports this hypothesis. Another factor is variation in groundwater level and soil moisture content. Since only 3 to 6 tests are run each day, these could vary considerably during the entire test cycle, changing the modulus values. Also, warping and curling effects may be important for values measured on the pavement surface."11
The above information is noteworthy because many agencies do not conduct plate-loading tests - they assume values - and few people have conducted as many tests as Bryden. Even if the plate test is not the best measurement to represent the foundation support, the variability exhibited along the route in a repetitive series of seasonal tests would probably exist in any other test methods.
Modulus of Rupture
Many pavement designers believe that pavements fail in flexure, and that the true measure of strength is the modulus of rupture. Unfortunately, the modulus of rupture (MR) is not a unique number; the outcome depends on the method of the test. The PCA used to publish a chart that showed the relationships between modulus of rupture test methods.12 The comparative results of the three methods - cantilever loading (for a 30-in [760-mm] span), center loading, and _ point loading - are based on uniform conditions of moisture and temperature that do not occur in nature. Furthermore, the results are dependent on span length. This begs the question: What is the strength (MR) of a three-dimensional slab that is continuously supported (as contrasted to _ point support), with both moisture and thermal gradients (as contrasted to uniformly conditioned), and with an infinite two dimensions (as contrasted to a 30-in-long beam). Bengt Friberg proved that a slab-on-grade had a moisture gradient (wet bottom) that produced a compression in the bottom that, away from the slab ends, produced a residual compression of about 250 psi (1.7 MPa) or more.13 This means that a wheel load placed on the surface must first overcome the residual compression before the concrete goes into tension! The residual compression provides a significant increase in resistance to loading that otherwise might produce cracking.
It is well proven that with durable concrete, pavement distress is caused by the magnitude and frequency of vehicle loads. But what is the fatigue strength of concrete? As discussed above, it is difficult to quantify the flexural strength of the concrete in a three-dimensional slab. This difficulty is further compounded by the problems associated with determining how many loads will cause fatigue failure.
Craig Ballinger in Effect of Load Variations of the Flexural Fatigue Strength of Plain Concrete gives some perspective on the subject. He tested air-dried specimens of various lengths up to 64 in (1.6 m) with _ point loading and used a multiple correlation analysis to obtain a regression equation.14 One must ask: (1) How much stress does a heavy load actually cause in a concrete beam (slab)? and (2) What is the flexural strength of the concrete beam (slab) when it is loaded so that we can calculate the "percent of ultimate strength" consumed? If questions one and two can be answered, Ballinger found that "The Miner hypothesis appears to represent the cumulative damage effects from variations in fatigue loading in a reasonable manner."
Some pavement designers assume "average concrete" properties in their calculations without any information about which aggregates, cement, pozzolans, or mixture proportions that the con- tractor will use later on the job. Concrete properties of particular importance to pavement design are: E (Modulus of Elasticity), strength, thermal expansion, shrinkage, creep, heat generation, and durability (physical and chemical reactivity). A good pavement designer should also be a concrete expert. Some facts to be kept in mind are:
In the United States, there are 118 cement plants, each producing a variety of "unique" products under broad specifications. From personal experience on a research project, five Type I cements from different plants ranged in 28-day strength from 2738 to 4975 psi (19 to 34 MPa).
There are 420 coal-burning plants in this country, and 28 percent of their fly ash is acceptable for use in concrete in accordance with ASTM C618. These products react differently with various cements, and the result is dependent on the quantities used. This is particularly important with respect to alkali aggregate reaction and sulfate resistance (and, possibly, delayed ettringite formation).
Thousands of aggregate sources are available for use. Unfortunately, aggregate is not an inert filling. In addition to some aggregates reacting with the cementious materials, there are other characteristics that can cause problems.
It is important to recognize that these properties also vary with strength of the concrete. The variables are so great that it's imperative that the job mix be pretested to verify its properties and to measure its durability properties. On the other hand, there is great risk in letting the contractor switch cement sources (or other ingredient) without verifying the new properties.
What constitutes pavement failure? Is it a structural crack or a series of cracks and quantifiable distress measures? Or is it a function of rideability (smoothness)?
At the AASHO Road Test, there were two distinctive failure modes. The very thin pavements failed with continuous edge pumping that caused edge cracking that coalesced into a longitudinal edge crack. The thicker pavements failed by joint pumping that caused transverse cracking starting particularly in the traffic leave side of the joints. The data from both were averaged together in the road test analysis to develop a performance equation. Even so, of the 84 pavement test sections greater than 8 in (200 mm) in thickness, only seven sections had a serviceability index of less than 4.0 at the end of the testing. In fact, only three sections could actually be considered as having failed. Hence, one can conclude that even though the AASHO data is the best that we have, it hardly predicts failure of the thicknesses of pavement that are now being built (greater than 8 in). Additionally, at the road test, there were no punchthroughs (shear failure) such as those produced at the Pittsburg Road Test under steel wheels, nor were there other types of environmentally induced failures such as blow-ups, CRCP punchouts, and so forth.
Another weakness of the data from the AASHO Road Test is that there was no relationship developed between the axle loads and concrete strengths. Strength was incorporated into the design equations by substituting the Spangler stress equation into the road test relationship. The stress equation is based on the elastic relationship up until a crack forms. Unfortunately, the road test equation is a dynamic function of serviceability (rideability), and it can be argued that the two relationships are incompatible.
One also needs to look at the effects of uncontrolled variables (environment) on pavement performance. A good example is the Road Test One - MD where controlled testing during July and August produced negligible damage. In September, the area had very heavy rains. In August, eight joints were pumping compared to 20 and 28 in September and October, respectively. The edge pumping was 162 ft (50 m) in August, 462 ft (140 m) in September, and 380 ft (116 m) in October after the heavy rain.
In the preceding section, I tried to raise some questions about the weak assumptions that underlie the models and equations we use to determine the thickness of our pavements. In lieu of placing our emphasis (and our faith) in the accuracy of the equation, I propose we shift our emphasis to ensuring the quality of the product we build.
In 1977, I proposed a design concept called PAST-PIF, which stands for Pick A Slab Thickness - Protect It Forever. The process consists of a "belt and suspenders" operation in that, like in a space capsule, each component has a purpose and each has a backup:
One basic tenet is that the pavement is built as designed and specified. The concrete is made from pretested materials that are brought together in a well-proportioned mixture that was demonstrated to have the same properties that the designer assumed. Similarly, if the designer is using a 40-year design life, then he must ensure that the hardware will protect the corners for 40 years. That is, the dowels cannot corrode, disintegrate, lock up, or develop looseness that will render them useless in 10 years. According to Westergaard's equation, a 10-in- (254-mm-) thick slab that loses its dowels should have been designed as a 16-in- (406-mm-) thick slab! The design of all components must be balanced in that they will all last for the assumed design life. Similarly, the concrete must last 40 years without deteriorating from chemical or physical reactions before that age. Hence, much material testing, construction control, and quality assurance are required in the PAST-PIF concept.
With a brief background on PCC pavement history and a look at what we don't know behind us, I now want to look toward the future. What challenges face us and what are our research needs? Although many innovations have been proposed over the years - such as self-stressing concrete pavements, prefabricated component pavements, prestressed pavements, and others - few of the ideas have been marketing successes. Hence, the following thoughts are concerned more with the construction process of our more standard designs. Broadly, they deal with: (1) making pavements more economical to construct, (2) speeding up the construction process to reduce traffic delays, and (3) providing more safeguards so that the pavements have a better chance of serving their design lives without premature distress. It should be noted that this approach is similar to that being proposed by the American Concrete Pavement Association.15
On-Grade Ultrasonic Mixers
In the 1960s, Ohio State University experimented with ultrasonic concrete mixers in which the water thoroughly wetted the aggregate as it moved through a pipeline subjected to ultrasonic frequencies. Such a mixer would not need a rotating drum. The mixer on grade could pick up the aggregates from a windrow, and a slurry could be fed through an umbilical hose into the mixer. The mixture could be extruded on the grade. Much faster setting mixtures could be used because of zero haul times.
Self-leveling mixtures are already being used for sub-floors. Vibrators and their associated problems during construction (broken vibrators, vibrator trails, etc.) would be eliminated.
The extruded ribbon of concrete could be "instantly internally heated" to initiate the setting so that finishing, jointing, texturing, and curing could be completed in the trailing forms. There would be no need to come back later for joint sawing. Work is underway at the Advanced Cement-Based Materials Center at Northwestern University.
Most paving mixtures contain adequate mixing water to hydrate the cement if the moisture is not allowed to evaporate. It should be possible to develop an oil, polymer, or other compound that would rise to the finished concrete surface and effectively seal the surface against evaporation. R.K. Dhir recently published some test results on self-curing mixtures.16
Durable Concrete Without Entrained Air
Entraining proper air in concrete is difficult and requires an inordinate amount of care, control, and testing. It has been demonstrated that internally sealed (wax bead) concrete, polymer-impregnated concrete, and to some extent, latex-modified concrete become impermeable to moisture and are inherently durable when subjected to freeze-thaw exposure. If an inexpensive way could be developed using admixtures (oil within coatings, like small capsules that time-release their contents) to render hardened concrete impermeable, concrete could be made more durable in a fail-safe manner without air tests, losing strength, moisture gradients and associated warping, shrinkage, and chemical activity.
The incorporation of the above items into one paving operation could produce a pavement that would meet Fast Track criteria. The dowels and tiebars would be vibrated in, and the joint grooves would be formed into the extruded concrete. No subsequent operations would be required behind the slipform operation.
High-strength concrete is already being used in rapid-cure patches. The high cement contents cause high temperatures that result in thermal contraction problems. Presently, other than early opening, there are no advantages to using higher strengths in pavements. Such concrete is expensive, and if higher strength pavements are to be competitive, ways must be found to minimize the amount of the expensive concrete. The French developed a two-layer extruded slipform operation that might be used by encapsulating normal concrete within the protective high-strength concrete. Other more economical shapes might also be considered, such as slabs cast with internal voids, or beam and slab configurations although we have no data on deflections, water movement, friction, curling, and warping of unusual slab configurations. Also, jointing technology would be needed. A vigorous research study would be needed to make the use of 10,000-psi (69-MPa ) concrete more efficient for structural pavements.
Ultra-High-Strength Concrete for Continuous Pavements
Just as continuously welded rails are used, it should be possible to construct a continuous ribbon of concrete that would withstand a temperature range of 100 F (55 C). A tensile strength of about 2500 psi (17 MPa) would be needed, which might be possible with a compressive strength of about 25,000 psi (172 MPa) (plus a factor of safety). This could be accomplished with polymer-impregnation if a field process could be developed. Alternatively, for comparison, a laboratory strength of about 106,000 psi (731 MPa) has been attained with portland cement. Special concretes are presently being used in the 25,000-psi (172-MPa) range based on a reactive powder process.17 The strength must be attained about 18 hours before the cooling concrete begins to contract. Of course, such continuous ribbons of ultra-high-strength concretes will experience about 2 in (50 mm) of movement at the ends, making special anchors or joints necessary.
We must continue to build on the wealth of available research on pavements even though much of the work precedes the computer revolution and one must look hard to find the information. This older experimental work was meticulously done despite the lack of modern electronics. If one has the opportunity to search the files, one can often find precedents for today's "innovations," such as variable thicknesses, stainless dowels, beam and slab construction, etc.
PAST-PIF shifts the emphasis from slab thickness to concentrating on seeing that all design assumptions are met, that the pavement is built as the designer intended with long-lasting materials, and that the pavement is protected and maintained to fulfill the design assumptions. Because most premature distress is materials related, the designer must play the role of a materials engineer in pretesting the job materials.
Finally, the research needs for the future are being looked at, primarily from the materials and construction point of view. If PCC pavement construction is to stay competitive, ways must be found to place concrete more economically, with less delay to the traffic, and in a way that the pavements provide more assurance of a maintenance-free design life. One-pass paving is needed with "triggered" fast-setting concrete that is self-leveling, self-curing, durable, and without entrained air so that all paving operations can be completed within the trailing forms. The use of high-strength concrete, if it is to be economical, will probably require new slab configurations that are untested. Ultra-high-strength concretes might be used similar to the continuous steel rails of the railroads.
Thomas J. Pasko Jr. retired as the director of advanced research for FHWA on Aug. 1, 1997, after 36 years of service with the agency. He received both his bachelor's and master's degrees in civil engineering from Pennsylvania State University and completed additional graduate-level courses at Cornell University. He is a licensed professional engineer in Pennsylvania. He is a Fellow of the American Concrete Institute, a past member of the ACI board and Technical Activities Committee, and a past chairman of the ACI Pavements Committee.