Introduction Test roads, research, analytical studies, and, most importantly, the observed performance of pavements in service have served as the basis for concrete pavement design practices. A historical search reveals that the first controlled evaluation of concrete pavement performance was conducted in 1909. The Public Works Department of Detroit conducted what was probably the first pavement test track. It consisted of a set of steelshod shoes and heavy iron wheels mounted at opposite ends of a pole to simulate horse and wagon traffic of the day. This crude device revolved around a circular track containing sections of concrete, granite, creosote block, and cedar block. Based on this study, Wayne County, Michigan paved Woodward Avenue with concrete - "the first mile of rural concrete in the United States." Sixty more miles of concrete roads were built in Wayne County in the following two years. After 1916, concrete roads were being built 5 to 9 in. thick, but little was know about thickness requirements. During 1912 to 1923, the State of Illinois conducted the Bates Test Road. Using Old World War I trucks with wheel loads from 1,000 to 13,000 lbs., traffic was applied to test sections of different materials and thicknesses. The concrete sections of uniform cross-section were 4 to 9 in. thick while thickened-edge sections were 9-5-9 and 9-6-9 in., some with edge bars. The results showed that of 22 brick, 17 asphalt, and 24 concrete sections in the test, 1 brick, 3 asphalt, and 10 concrete sections satisfactorily withstood the truck loadings. As a result, several design formulations were developed for Illinois to use in building its first state highway system. Until 1922, many pavements had been built with no joints and a thickened center section in an attempt to prevent the formation of an erratic longitudinal crack that developed in 16- to 18-ft.-wide pavements. Results from the Bates Test Road led engineers to the use of a longitudinal center joint to eliminate longitudinal cracking. Design Procedures In the same time period, 1921-1922, a test road was built at Pittsburgh, California, to determine if steel reinforced pavements performed better than plain pavements. Here also, old World War I army trucks were used to traffic pavement sections half of which were plain and half reinforced. The results were inconclusive and the debate on reinforcement remains unsettled to this day. However, out of these early tests, simple equations relating pavement thickness to traffic loading emerged. The critical stress was at slab corners because of the narrow lanes, 9 ft. wide at the most, which resulted in truck wheels being located at pavement edges. Several corner cantilever equations were proposed separately by Goldbeck, Older, and Sheets based on the Bates and Pittsburgh test roads. These were the beginnings of so-called "mechanistic-empirical" design procedures (mechanistic - based on computed pavement response; empirical - calibrated to observe pavement performance). The 1930 PCA publication Design and Construction of Concrete Pavements adopted Older's equations involving computations of the load-induced corner stress, load, and slab thickness. A design thickness was selected to limit the stress to one-half of the concrete's modulus of rupture. Other pavement cross-sections in early use, in addition to the thickened center section mentioned above, were the uniform thickness section and the thickened edge section. The principle of the thickened edge section (edges 2 in. or 3 in. thicker than the center) was to compensate for the higher corner or edge stresses that occur due to wheel placement location. In 1912, the California Highway Commission adopted the thickened edge section as an alternate to the uniform thickness section and in 1921 it was made a standard for all concrete pavement construction. In 1920, in Maricopa County, Arizona, construction was started on a contract involving 141 miles of concrete pavement, all with thickened edges. In the late 20's and early 30's, almost all of the highways in this county were 6 in. at the middle with 8 or 9 in. at the edge; 10-7-10 in. pavements would later occur. The use of thickened pavements persisted to some degree through the early 1950's and then phased out as more pavements were built with increased lane widths of 11 or 12 ft. Westergaard's Edge The theory of pavement thickness design was advanced by the work of H.M Westergaard in 1926. He presented equations for determining stresses and deflections in concrete pavements due to loads applied at the interior of the slab and at free edges and corners. Factors for size and weight of loads, subgrade reaction, concrete thickness, modulus of elasticity, and Poisson's ratio were included. These equations, which permitted the determination of pavement thickness for any specified condition of loading, were used by engineers for many years. In the early 1930's, the Bureau of Public Roads conducted loading test on concrete pavement at Arlington, Virginia. In these tests, measurements of slab curling due to variations in pavement temperature, were made to provide a check on the Westergaard, Kelly, Spangler, and Pickett all developed slight modifications of the original Westergaard equations to provide theoretical results that were in closer agreement with measured stresses and deflections from the Arlington tests. PCA's 1933 thickness design procedure, Concrete Road Design; Simplified and Correlated with Traffic, was published. It was authored by Frank T. Sheets who, after serving with the Illinois Highway Department and working on the Bates test Road, joined PCA and became its president in the 1940's. Load stresses for moving wheel loads were computed by simple empirical equations representing a modification of the corner formula advanced by Older. These closely approximated the computations of Westergaard and the behavior of the Bates Test Road sections. The equations are given here because they were probably the first to be used in routine design practice (these were for pneumatic tires; for solid tires, stresses were increased by 25 percent): Case I - Protected Corners (smooth longitudinal edge bars) S = (1.92W/d^2) Case II - Unprotected Corners (no edge bars) S = (2.4W/d^2) where S = stress, psi W = wheel load, lbs. d = slab thickness, in. Computations were for relatively weak subgrade and thus Wetergaard's modulus of subgrade support was not directly incorporated. The 1933 PCA procedure was the first to introduce fatigue concepts, based on results from the Bates Test Road, where the number of wheel load repetitions causing slab failure was related to the computed stress level. Jointing Practices While new design equations were being developed to determine pavement thicknesses, changes were also underway in jointing practices. As previously mentioned, experience with early pavements indicated a need for a longitudinal center joint in pavements built with widths of approximately 12 ft. and over. The early rural pavements were built without transverse joints except at the end of the construction day. Blow-ups, which developed in some of these early pavements, led many engineers to the use of closely spaced transverse expansion joints in an attempt to relieve compression stresses that develop during hot summer days. A combination of transverse expansion and contraction joints with expansion joint spacings varying from 50 ft. to 120 ft. and contraction joint spacings from 15 to 60 ft. were used in most areas between the years 1925 to 1945. Poor performance on many projects built with these jointing combinations led to the construction of six experimental jointing projects. Undertaken in Oregon, California, Missouri, Minnesota, Michigan, and Kentucky, the projects demonstrated that closely spaced expansion joints were not required in concrete pavements built with normal aggregates under normal summertime temperatures when contraction joints are used at 15-ft. to 60-ft. spacings. They also showed that closely spaced expansion joints tended to progressively close up over the years, allowing greater opening to occur at the contraction joints. This progressive movement is no doubt caused by continued infiltration in open joints during the winter. As a result, design engineers have eliminated transverse expansion joints except at bridge abutments and some intersections. Skewed, Randomized Joints Other jointing practices were also being tried in the early years. Patents for skewed transverse joints were issued in 1906 and 1918. The structural advantage of skewed joints is that wheels on opposite sides of an axle do not cross the joint at the same time. This reduces the degree of joint faulting, makes for a smoother riding pavement. Engineers reasoned that, on the downstream traffic side of the joint, the obtuse angle of the skew provided a stronger section where it was needed. After building its first skewed joint roadway in 1932, the California Division of Highways deferred additional construction until its performance could be evaluated. Further delayed by World War II, it was not until 1951 that construction of projects with skewed joints was resumed in earnest. The Washington State Department of Highways had similar experiences starting in 1954 with both chevron and straight skewed joints. Benefits of the skewed practice for undoweled pavements. By 1975, 18 states reported skewed-joint use; 15 states using it in undoweled pavements and three states employing it in doweled pavements. A 1987 FHWA survey reports that 25 states specify skewed joints; about five of these use skewed joints in doweled pavements. Thirteen states use randomized joint spacings when skewed joints are specified. The use of randomized joints originated due to a problem with ride quality on pavements with undoweled joints constructed at smoothness level capabilities of the 1940's and 1950's. In the late 1950's, manufacturers of automobiles started receiving complaints of a peculiar phenomenon called "Freeway Hop" that was unique to concrete pavements on the California freeway systems. A study found that, with the suspension systems of certain models of large cars (a 1959 Buick was representative), a harmonic wave of objectionable vertical motion was induced by the interaction of small but repetitive road irregularities and tire-wheel imperfections; the cars were resonant to the repetitive 15-ft. joint spacings. Engineers suggested an irregular join-spacing sequence. As a result, randomized joints became the standard practice for major roadways in California. Sequences of 13, 19, 18, 12 ft. were used through the 1960's and 1970's. In the same time period, a dozen western states adopted a similar practice for their undoweled pavement designs, and four states extended the use to doweled pavements. Skewed joints accompanied the use of randomized joints. Experience in several states showed that the 18- and 19-ft. panels were too long - they sometimes developed intermediate cracks - so the spacings were changed. For example, California has used 12-, 15-, 13-, 14-ft. sequences in the last ten years or so. A 1987 FHWA survey reported that about 15 states employ the randomized joint feature. These include seven states that use it in undoweled pavements, three states that use it in plain, doweled pavements, and five states that use it in both undoweled and doweled pavements. Joint spacings and sequences vary from state to state; most have shortened their longest panels. Coinciding with much of the 20th century testing of jointed concrete pavements has been the evaluation of continuously reinforced slabs. The first continuously reinforced pavement was built by the Bureau of Public Roads in Maryland in the early 1920's and the second one was built in Indiana in 1939. The performance of the Indiana project and others built in Illinois, California and New Jersey around 1949 led to an increased interest in this design. By 1970 there were continuously reinforced pavements in service in approximately fifteen states. Post - 1950's Thickness Design Successive changes of the PCA design publications through 1951 resulted in the use of Pickett's equations which related load-unduced corner stress to thickness, wheel-load, tire contact area, and modulus of subgrade reaction. Fatigue relationships from the earlier 1933 version were still employed. The concept of "controlling wheel load" was introduced - the heaviest 100,000 anticipated wheel loads using a lane during the design life. As before, the design applied to both uniform thickness and thickened-edge cross-sections. The 1950's brought additional developments: many state agencies developed their own design procedures, Pickett and Ray prepared influence charts that could accurately analyze edge and interior stresses for multiple-wheel configurations (1951), and the AASHO Road Test commenced (1958-1960). The latter was the most extensive research on pavements ever conducted: $26 million was raised to build a test road in Ottawa, Illinois that would be traveled around the clock for two years on six different test loops. There were 12 different combinations of axles and many different thicknesses of asphalt and concrete, both plain and reinforced. Engineers developed much information including a serviceable index rating, which has been used ever since. This serviceability concept became an important tool of pavement management. Performance equations evolved from two years' testing; one for concrete and one for asphalt. These related the number of loads and pavement thickness to pavement serviceability. Based on the results of the AASHO Road Test and with the use of the Pickett-Ray influence charts, PCA revised its design procedure in 1966. The analysis modeled all four or eight wheels of a single or tandem axle crossing a pavement joint. The fatigue curve was modified based on new data; the previously used load impact factor was omitted because research had shown that static loads create greater stress than moving loads. The AASHTO Interim Guide for Design of Pavement Structures based on the results of the road test and an evaluation by the state highway agencies was published in 1972. (By that time, AASHO had changed its name to AASHTO, the American Association of State Highway and Transportation Officials). The concept of "Equivalent Single Axle Loads" introduced by engineers at the road test was included to simplify the handling of axle loads of mixed magnitudes. To extend the design of concrete to conditions other than those at the road test, Spangler's corner equation was incorporated to recognize different subgrade support conditions and different strengths of concrete. The design equations for both concrete and asphalt continue to form the basis of pavement design for many state highway agencies today. The AASHTO Interim Guide was revised slightly in 1981 by a modification in Chapter III relating to the safety factors used in the concrete design procedure. In 1984, PCA made extensive revisions in its design procedure represented by the publication Thickness Design for Concrete Highway and Street Pavements. In developing the procedure, a comprehensive analysis was made of concrete stresses and deflections by a finite-element computer program. The program modeled the conventional design factors of concrete properties, foundation support, and loadings, plus joint load transfer by dowels or aggregate interlock and concrete shoulder, for axle-load placements at slab interior, edge, joint, and corner. For the first time in any design procedure, the concept of erosion of pavement components was introduced based on concrete performance on major highways. The erosion criterion, based on the deflections caused by loads at slab corners and edges, is applied in addition to the stress-fatigue criterion. The erosion criterion recognizes that pavements can fail from excessive pumping, erosion of foundation, and joint faulting. The stress criterion recognizes that pavements can crack fatigue from excessive load repetitions. The criteria for the design procedures are based on the pavement design, performance, and research experience - including relationship to performance of pavements at the AASHO Road Test and to studies of the faulting of pavements. The AASHTO Guide for Design of Pavement Structures was published in 1986, replacing the 1982 Interim Guide. It retains the basic algorithms developed from the AASHO Road Test as used in the Interim Guide but has been expanded to include many new considerations such as reliability concepts, improved material characterization, drainage and environmental conditions, tied concrete shoulders or widened lanes, life cycle cost analysis, and pavement management considerations. New Generation Design Today, two extremely significant developments in the area of pavement design are underway. The first is the Long-Term Pavement Performance (LTPP) program of the Strategic Highway Research Program (SHRP). LTPP is a 20-year program (1987 to 2007) designed to deliver a steady stream of data to improve pavement design and rehabilitation techniques. The LTPP program is funded at $41 million for five years with recommendations under consideration for funding an additional 15 years. This is the largest pavement performance research effort in history. It is gathering data on the performance of a variety of in-service pavement types in a wide range of climate, traffic, and subgrade conditions. The specific objectives of the LTPP program are to: evaluate existing design methods develop improved design methodologies and strategies for the rehabilitation of existing pavements develop improved design equations for new and reconstructed pavements determine the effects of loading, environment, material properties and variability, construction quality, and maintenance levels on pavement distress and performance determine the effects of specific features on pavement performance establish a national long-term pavement data as to support SHRP objectives and future needs. The second significant development towards improving design technology is the increased attention and effort presently given to the development of "mechanistic" designs. Sophisticated procedures are now emerging that include improved and more extensive analysis of pavement responses to loads and environmental conditions. Much work is being done to establish realistic "transfer functions" (calibration to specific pavement distresses) and in the application of design reliability concepts. Some of the systems will include economic analysis and optimum strategies of rehabilitation and overlay techniques. A few state highway agencies have implemented such systems to some degree and others have them in a developmental stage. The greatest effort in the area of design is represented by the National Cooperative Highway Research Program Project No. 1-26, "Calibrated Mechanistic Structural Analysis Procedures for Pavements." This four year research program, conducted by the University of Illinois, is scheduled for completion in the summer of 1992. One of its principal objectives is to provide directions for the future revision of the AASHTO design procedure and state highway agencies. Through the years since 1923, pavement design has gradually, but steadily transformed to be closer to a science than an art. Analytical models have vastly improved and are still improving and extensive data banks on pavement performance are finally being built. Even so, successful pavement design will always largely depend on the good judgement and experience of the designer.