TRANSPORTATION RESEARCH RECORD 1336 57 Comparison of and Computed Drilled Shaft Capacities Based on Utah Load Tests RICHARD PRICE, KYLE M. ROLLINS, AND EDWARD KEANE Fourteen load tests were performed at a number of bridge sites in Utah in an effort to reduce foundation costs. Soil profiles generally consisted of cohesionless materials. Failure loads were determined from load-settlement data using four methods, in- cluding Davisson's method, double tangent method, 1 in. crite- rion, and the Hirany and Kulhawy's method. Hirany and Kulhawy's method typically failure loads 30 percent higher on the average than Davisson's method. Load capacities were computed using equations proposed by Reese and O'Neill/FHWA, Meyerhof, and Reese et al. Good agreement between measured (Davisson's method) and computed capacities was generally found for the FHWA and Meyerhof methods, but the Reese et al. method significantly underpredicted capacity. The design methods were generally more than 25 percent conservative in comparison with the Hirany and Kulhawy failure load interpretation. In three cases involving dense gravels and weathered shale the design equations resulted in computed capacities that were only 20 to 30 percent of the measured failure load. In an effort to reduce construction costs, the Utah Depart- ment of Transportation has performed a number of load tests on drilled shafts. Since the measured capacities were generally higher than estimates based on design equations, it was pos- sible to increase design loads for each shaft. The costs of load testing proved to be insignificant when compared with savings that were realized by reducing the size and number of drilled shafts needed. It is estimated that the testing program resulted in cost savings of more than $890,000 in the first 2 years. Unfortunately, load testing cannot be performed for each project, and reliance must be placed on load prediction equa- tions. Reasonable capacity prediction equations are necessary to provide safe yet economical shaft foundations for highway structures. OBJECTIVES AND SCOPE OF INVESTIGATION Since the inception of the Utah load testing, Reese and O'Neill (1) prepared a drilled shaft handbook with new design rec- ommendations for nationwide use. This work was prepared for the Federal Highway Administration in cooperation with the Association of Drilled Shaft Contractors. In addition, a new procedure for interpreting axial load test data was pro- posed by Hirany and Kulhawy (2). R. Price, RB&G Engineering, P.O. Box 711, Provo, Utah 84603. K. M. Rollins, Civil Engineering Department, Brigham Young Uni- versity, 368 CB, Provo, Utah 84602. E. Keane, Materials and Re- search Division, Utah Department of Transportation, 4500 S. 2700 W., Salt Lake City, Utah 84119. It was determined that a review and summary of the existing load test information would be valuable for (a) establishing a computer data base on the performance of drilled shafts in Utah that could be updated with future tests, (b) evaluating the variation in load capacity on the basis of various load test interpretation procedures, (c) determining the suitability of various design equations for Utah soil conditions, and ( d) making recommendations for future load testing procedures. Since the load test data set was not used in the development of any of the design methods, it provides an independent check on the various design methods proposed for use. Load test data can easily be compared with calculated values based on design methods. By comparing load test data with these methods, one can better determine which method is more applicable for a given circumstance. In this study, four methods were used for determining fail- ure load from the load versus settlement test data. The pro- cedures are Davisson's Method (3), the double tangent method (4,5), Hirany and Kulhawy's Method (2), and Terzaghi's sim- ple procedure (6), which defined failure at a settlement of 1 in. The results from each interpretation procedure were com- pared with three prediction methods. The load prediction equations included those proposed by Reese and O'Neill (1), Meyerhof (7), and Reese et al. (8). DESCRIPTION OF DATA BASE CHARACTERISTICS A total of 14 load tests were available for evaluation, and additional tests are presently under way. In each case the function of the foundation was to support a bridge or overpass. Geotechnical Conditions The subsoils at the test sites generally consisted of granular materials deposited by stream flow along with some man- made fill. Clay layers were encountered in only two of the tests. Subsoils encountered at these sites ranged from very loose to very dense sand and gravel with some silt and cobbles. SPT N values ranged from 10 to 80 but were typically between 20 and 40 blows/ft. A fairly typical boring log in the alluvial deposits is shown in Figure 1. Most of the shaft tips were founded in soil; however, in three cases, the shaft tips were