A Quarter Century of Geotechnical Research

Chapter 2: Foundation Engineering Project

Prior to the establishment of this research program, foundation engineering guidelines were mostly empirical and barely sufficient for most highway applications. The result was an inability to accurately predict the performance of foundation systems, which, in turn, led to very conservative design procedures as well as an occasional foundation failure. A comprehensive research plan was initiated in the late 1970's to develop improved design and construction guidelines for building safer and more cost-effective bridge foundations. Engineering improvements were expected to reduce the cost of these foundations and stretch the highway dollar to buy more bridges that will last longer.

An analysis of future needs for highway bridge foundations was completed under this project to provide essential planning information for conducting effective research on bridge foundations. Estimates of the number of new bridges to be constructed and those that would need to be replaced or rehabilitated during the remainder of this century were made in 5-year increments using data from FHWA bridge inventory and inspection reports. An analysis of FHWA foundation management review reports was also made to determine typical foundation types used in each State and likely to be used on future construction. The analysis was also supplemented with personal interviews with selected State and FHWA bridge engineers from various regions of the country (1).

Results of these analyses indicated that more than 100,000 bridges would be constructed, replaced, or rehabilitated in the United States during the last 20 years of this century. Approximately 20,000 new bridges would be built and more than 15,000 existing bridges had deficient foundations. Many of the satisfactory foundations would also have to be replaced because of retrofit problems caused by replacing the superstructure. A large number of reusable foundations required special design and construction procedures that needed to be developed. It was also noted that more than two-thirds of these bridges were likely to be supported on piles, one-fourth of them on spread footings, and the rest of them on drilled shafts or other types of foundations (1).

2.1 Background

Because the safety and cost-effectiveness of bridge foundation systems is of major importance, an appropriate research program to develop better design tools has long been advocated by both structural and geotechnical engineers. Foundations represent about 30 percent of the cost of highway bridges in typical applications; however, this cost can be even higher where bridges are built near or on difficult soil conditions. The total annual expenditure of public funds for bridge construction is conservatively estimated to be more than $2 billion, which means that foundations are costing more than half a billion dollars per year.

Because the predominant type of foundation system used in the highway industry is piles, the first priority for this project was the efficient design of piles. Research was also needed to develop new concepts for reducing our excessive reliance on pile foundations. In situations where deep foundations are required, the overall cost can be significantly reduced by using smaller piles and fewer of them as a result of design economies developed under this research project.

The indiscriminate use of pile foundations was partly due to a lack of confidence in the safety and dependability of drilled shafts and shallow foundations, but mainly due to unreasonably restrictive deformation and movement constraints. Therefore, the establishment of rational criteria for tolerable movements was a necessary first step toward improving foundation design (see section 2.5).

Assuming a 25 percent reduction in the use of piles can be accomplished by substituting cheaper drilled shaft or shallow foundation units, the use of piles will still account for much more than half the foundation systems used for highway structures. Therefore, the continued investigation of pile-soil behavior will provide cost-effective improvements. One of the most incongruous aspects of foundation design is the overwhelming use of a system (friction piles) that is the least understood.

2.2 Objectives

The overall project objective was to develop improved foundation design and construction methods for highway bridges. This objective was approached through efforts directed toward accomplishment of the following four separate, but related goals:

1. Determine the structural consequences of foundation movements and define the limits of tolerable movements for use in the design of efficient foundation systems.
   
   
2. Establish a fundamental understanding of the behavior of soils during and after foundation installation for use in analytical predictive methods.
   
   
3. Improve the efficiency, reliability, and use of less expensive shallow foundations.
   
   
4. Develop more accurate concepts and guides for the design and construction of deep foundations.

2.3 Scope

Research efforts in this project were directed toward developing new predictive techniques for foundation design, better criteria for tolerable movements, and permissible stress levels in piles and drilled shafts, as well as improved construction control techniques. Each component of the total foundation system was studied individually and each total system was evaluated as a whole. The behavior of the foundation systems was also examined under both working and ultimate stresses.

2.4 Project Description

The major research efforts were organized into five tasks:

1. Structural Consequences of Foundation Movements – Research was conducted to develop improved criteria for tolerable foundation movements. A practical limit to the amount of foundation movement that can be tolerated without serious structural or functional distress by various types of highway bridges was defined and established through structural analysis and supporting field data.
   
   
2. Predicting the Behavior of Piles and Foundation Soils Under Structural Loads – This task involved the development of improved analytical predictive techniques for the design of piles. Special emphasis was placed on the study of load transfer between a pile and the surrounding soil to include both single piles and groups of piles. An important associated objective of this task was to determine the effects on soil properties of installing the pile and other remote stress applications such as negative skin friction. The predictive techniques were to consider the "before" driving case (in situ soil properties), the "during" driving case (the changes in soil properties brought about by the installation method), and the "after" driving case (soil recovery with time).
   
   Another major emphasis was on full-scale load tests on piles and groups of piles to develop correlative relationships between the performance of an individual pile and a group of piles. The corresponding values of bearing capacity and settlement were of primary concern; however, valuable information concerning pile spacing, sequence of driving, and efficient shape of the group were also obtained. Load test studies were also performed on drilled shafts.
   
   Another important associated objective of this task was the development of rational criteria for allowable stresses in piles to replace the blanket stress level codes that were being used in the highway industry. The criteria were developed in terms of driving stresses and static loading, as well as environmental exposure and soil conditions (see section 2.6.5).
   
   
3. Improved Design and Construction Techniques for Drilled Shaft Foundations  – Research under this task investigated the behavior and efficiency of drilled shafts as a suitable alternate for pile foundations in a similar manner to the Task 2 approach for piles, except with less emphasis on groups and more emphasis on construction problems such as free-fall of concrete heights, nondestructive evaluation techniques, and the use of drilled shafts in intermediate geo-materials such as soft rocks, hardpan, and glacial till.
   
   
4. Innovative Load Test Methods – Research under this task was to develop economic load test alternatives for bridge foundations. Systems that were evaluated under this investigation were Statnamic, Smartpile, Osterberg load cell, and several dynamic load test systems.
   
   
5. Improved Design for Shallow Foundations – Research under this task was to investigate the behavior and efficiency of shallow foundations, with emphasis on demonstrating the reliability and cost-effectiveness of spread footings. A survey of the problem areas, current practices, and the performance records of highway bridges supported on spread footings was conducted to demonstrate that the use of spread footings in many cases can be made a more acceptable and attractive alternative to expensive pile foundations.
   
   An important associated objective of this task was to determine the enhancement potential of various soil and site improvement techniques for reducing the reliance on pile foundations. Research was conducted to evaluate methods for improving a marginal foundation site to permit the use of shallow foundation systems in design situations where the soil or site conditions would normally dictate the use of piles.

The Foundations Project was the "flagship" project of the FHWA Geotechnology Research Program. It could and probably will be a separate report by itself; in fact, each of the five main tasks could also be a separate report. The following six sections of this chapter provide a summary of the major highlights of this project.

 

2.5 Structural Consequences of Foundation Movements

During the development of the FHWA research project on foundations, it was determined that one of the major reasons that bridge foundations were over-designed had its roots in the lack of rational criteria for tolerable movements of the superstructure. The American Association of State Highway and Transportation Officials (AASHTO) design code at that time was relatively silent in this area, and basically promoted a "zero" settlement design procedure. This resulted in a very high cost of obtaining little or no settlement in terms of utilizing the soil load-carrying capacity because it is almost impossible to achieve this objective. The cost was often extremely high and very wasteful.

Foundation movements under highway bridges (figures 1 and 2) had occasionally been predicted and measured; however, there had been very few investigations of the effect these movements would have on the safety and serviceability of the bridge structure. As a result, tolerable foundation movement had not been established beyond the conceptual stage, and much controversy among engineers was prevalent.

In addition to the disagreement on the definition of tolerable settlement, there was also disagreement on the accuracy of settlement predictions. As a result, many engineers still prescribed strict settlement constraints to guard against erroneous predictions. In many cases, the precautions were unrealistic and underestimated the accuracy of the prediction techniques, which are normally accurate within 10 to 20 percent on average and are rarely more than 50 percent off the measured value.

It was noted in earlier studies that using piles did not guarantee that there would be little or no settlement, unless the piles were founded on hard rock. The nature of load transfer from the pile to the soil surrounding (side friction) and beneath the pile (tip bearing) requires some movement of the pile to mobilize the load-bearing capacity of the soil material. The relative movement required to mobilize maximum frictional strength is approximately 6.35 mm (0.25 in), while the displacement required to mobilize the soil's shear strength under the pile tip is usually much larger than that value. If the allowable settlement is restricted to a smaller value than what is required to mobilize tip resistance, the load-carrying capacity of the pile is artificially reduced to the level of side friction support. The support capacity available at the pile tip then becomes an additional safety factor.

FIGURE 2.  Bridge girder jammed against abutment and excessive rocker tilt.

Overly restrictive settlement constraints can reduce significantly the supporting capacity of friction piles. To comply with the prevalent requirements of zero or near zero settlement, the foundation engineer must reinforce his or her design by adding more piles, using longer piles, or increasing the diameter of the piles. The conservative effect of this design philosophy is evident by examining any load-settlement curve from a pile load test. A slight increase in the allowable settlement value can impact significantly the allowable design load for a pile foundation, especially within the straight line portion of the curve.

A relaxation of the stringent movement criteria was also expected to impact significantly the use of spread footings, whose major drawback is the risk of some settlement. Because the economic breakpoint in the normal design procedure for spread footings centers around 25 mm (1 in) of settlement, the majority of bridge engineers were reluctant to use this less expensive method of foundation support. As a result, the highway industry has been accused of having numerous "buried treasures" beneath bridge abutments and piers, because piles were used instead of spread footings or because more and/or larger piles were used than necessary.

Research performed under this project indicated that bridge structures can withstand a reasonable amount of settlement, and the amount varies according to the span arrangement (simple versus continuous) and length of span as well as other design variables. This indicates that the AASHTO blanket criterion approach was inappropriate and the zero settlement design approach was too conservative.

2.5.1 Preliminary Studies - Prior to the initiation of a comprehensive research study by FHWA, a series of smaller studies that were conducted in Ohio, Washington, Connecticut, Canada, and by FHWA staff were evaluated to help develop a detailed research plan. Studies done for the building industry were also included in the evaluation. A study of the measured movement data and corresponding damage assessments were very enlightening, but insufficient to establish rational tolerable movement guidelines. It was also noted that the building codes in many major U.S. cities actually permitted larger settlements than AASHTO bridge specifications, even though building elements, such as glass doors and windows, elevator shafts, utility lines, and brittle wall panels, are more sensitive to foundation settlement than bridge elements.

Although no well defined set of criteria was generally agreed upon, all of the researchers agreed that the development of a rational set of tolerable movement criteria for bridges was a high priority research need. Everyone then turned to FHWA's Office of Research to conduct a major effort to solve this problem.

2.5.2 FHWA Tolerable Movement Study -  A comprehensive analysis of the data from the earlier studies that evaluated field measurement surveys was conducted as part of an overall analytical investigation of the effect of different magnitudes of settlement on the potential level of distress produced in a wide variety of steel and concrete bridge structures of different span lengths and stiffnesses. A total of 314 bridges in the United States and Canada was analyzed to determine if the measured foundation movements and corresponding damage assessments could yield sufficient insight toward the development of rational criteria for tolerance to movement. The data by themselves proved to be insufficient, but when combined with the structural analysis studies, and an appraisal of existing design specifications and practice, a rational set of criteria was developed.

The researchers determined that functional distress is more difficult to assess than structural distress because of its subjectivity. Functional distress is defined under this study as damage to the architectural elements or a reduction in ride quality. Architectural damage is less severe than structural damage that affects the integrity of a main supporting element of the bridge; however, architectural damage is usually more visible and causes an annoying or insecure feeling on the part of the motorist. It is also referred to as "cosmetic damage" and includes cracking or misalignment of bridge railings, curbs, decks, abutment wing walls, and damage to light poles and utility lines. The deterioration of ride quality involves the "bump" at the end of the bridge and other roadway unevenness associated with bridge foundation settlement.

The specific results of this study have shown that, depending on type of spans, length and stiffness of spans, and the type of construction material, many highway bridges can tolerate significant magnitudes of total and differential vertical settlement without becoming seriously over-stressed, sustaining serious structural damage, or suffering impaired riding quality. In particular, it was found that a longitudinal angular distortion (differential settlement/span length) of 0.004 would most likely be tolerable for continuous bridges of both steel and concrete, while a value of angular distortion of 0.005 would be a more suitable limit for simply supported bridges.

For continuous steel bridges, differential settlements of 25 mm (1 in) or more would be intolerable for span lengths up to 15.2 m (50 ft) because of the significant increase in stresses caused by these settlements. However, for span lengths between 30.5 and 61.0 m (100 and 200 ft), the stress increases caused by differential settlements up to 76 mm (3 in) were quite modest, and for spans longer than 61.0 m (200 ft), the stress increases caused by 76 mm (3 in) differential settlements were negligible.

A basic design procedure was developed that permits a systems approach for designing the superstructure and the foundation system. This design procedure incorporates the tolerable movement guidelines that are based on strength and serviceability criteria which, in turn, are based on limiting longitudinal angular distortion, horizontal movements of abutments, and deck cracking (2,3).

2.5.3 Publications and Implementation Items - Research and development activities for the establishment of tolerable movement criteria were very successful. The FHWA R& D reports documented the efforts that were made to develop the rational criteria and presented detailed recommendations on how the new criteria should be used in typical bridge and foundation engineering scenarios. A technology sharing report was developed to help practitioners implement the new criteria in standard design situations (4). Workshops and presentations were also held to aid the implementation process. A new article was developed for the AASHTO Bridge Specifications (see chapter 7).

2.6 Pile Foundations

The first difficult problem confronting the designer of a foundation is to establish whether or not the site conditions are such that piles should be used. The most common case is that in which the upper soil strata are too compressible or generally too weak to support heavy vertical reactions transmitted by the superstructure. In this instance, piles serve as extensions of columns or piers to carry the loads to a deeper, more rigid stratum such as rock (point-bearing piles). If such a rigid stratum does not exist within a reasonable depth, the loads must be gradually transferred by friction along the pile shaft. Scour and the relative inability of spread footings to transfer inclined, horizontal, or uplift forces and overturning moments also require the use of piles in many instances.

Another problem facing the pile designer is choosing from among the 100 or more different kinds of piles. There are many variations in materials, configurations, and installation techniques. Guidelines for selecting the best pile for various situations can be found in one or more of FHWA's guidance manuals (5, 6, 7) and the general literature.

After a particular kind of pile has been selected for use, the designer then must determine the number, length, and size of the piles required. Simple guidelines were not available at the start of this research program to design and analyze piles for the various situations that can occur in bridge foundations; however, improved pile design guides are now available in various FHWA references listed at the end of this report. The ultimate load that can be supported by a certain kind of pile depends on the strength of the soil and/or pile material. Usually the ultimate load is determined by soil failure; however, the pile itself may fail if forced to penetrate difficult site conditions such as dense soil or rock.

Calculating the ultimate load on the basis of soil failure is one of the most difficult problems confronting the foundation designer. Because available theories for determining the ultimate load of a pile and a group of piles were not accurate enough to provide economical foundation designs, research was conducted to define the complex mechanics of load transfer between pile and soil and among the various piles within a pile group. Load transfer in piles and pile groups was measured in the laboratory and field to develop and refine analytical models of the pile behavior process to provide more economical design procedures.

Assuming the pile material is not over-stressed, the ultimate load capacity of a pile is equal to the sum of two major soil components — point resistance and side friction. The amount of support contributed by each component varies according to the soil properties and pile dimensions. These resistances can be calculated through mathematical relationships; however, the required input data are difficult and expensive to obtain. As a result, less expensive index testing usually is performed to obtain approximate values for estimating the resistances.

Pile capacity usually is predicted using static and dynamic analysis procedures. On large, expensive structures, pile load tests often are used to verify the design loads. Recently completed FHWA experimental studies of instrumented piles load tested to failure have increased our understanding of load-transfer behavior of single piles and pile groups; however, more tests are needed to confirm the new prediction methods.

A mathematical model of pile group behavior was deemed to be a valuable analytical tool that systematically can convey engineering experiences from one site to another. Several mathematical models had been developed, including one by FHWA; however, none had been validated adequately because of the lack of precise field data on pile group behavior. This was especially true of new foundation design methods based on finite element analysis. To date FHWA has performed most, if not all, of the pile group load tests to failure. These results have been used to validate and refine the FHWA pile group design model called PILGP (8).

2.6.1 Areas of Emphasis -  Because of the high reliance on driven piles, a large portion of FHWA funds for foundation research was directed at improved design and construction procedures for driven piling. FHWA pile research was divided into the following three major areas of emphasis:

• The interaction between a single pile and the surrounding soil to develop accurate prediction methods for pile capacity.
  
  
• The behavior of groups of piles to determine appropriate efficiency factors that must be applied to predictions based on single pile design.
  
  
• Pile materials and driving systems.

 

2.6.2 Single Piles in Clay - Field load tests on single piles in soft clays have provided experimental data to evaluate a promising pile capacity predictive technique based on the general effective stress method. The effective stress method for predicting pile capacity was validated and its accuracy and general applicability were improved. Figure 3 shows a full-scale pipe pile being instrumented in the laboratory prior to field installation.

The accuracy of the method was limited by the assumptions made to describe the changes in effective stress between pile driving and subsequent loading. Major uncertainties arise in attempted modeling of the effective stresses after the disturbed soil has reconsolidated. Current analysis techniques were considered applicable to full displacement piles driven into normally consolidated clays (9).

FIGURE 3.  Lab instrumentation of a pipe pile for field load testing in clay.		FIGURE 4.  Field instrumentation of a pipe pile for field load test in sand.

2.6.3 Single Piles in Sand - Previous methods for predicting the behavior of driven piles in sand gave widely different results because of several sources of error. Variability of the soil and methods used to obtain the design parameters also contributed to the different results as did some of the simplifying assumptions made in developing the theoretical methods.

An FHWA research study on the behavior of piles in cohensionless soils involved the analysis of data on instrumented piles (figure 4) tested to failure under vertical loads. The data, collected through an extensive literature search, consisted of 35 pile load tests at 10 different sites. The piles were of various kinds, lengths, and diameters and included steel pipe and H-piles and prestressed concrete and timber piles. Average pile diameter was 0.4 m (1.27 ft), and average pile length was 15.1 m(49.6 ft ).

Statistical analyses determined the vertical and horizontal variability of the soil at each case history site. Load transfer characteristics were analyzed for each pile without consideration of residual driving stresses. In those few cases where residual stress data were available, the load transfer characteristics were reanalyzed to learn the effect of residual driving stresses. It was found that residual stresses play an important role in pile design and that proper consideration of residual stresses can result in shorter pile lengths for driven piles in sand (10).

2.6.4 Allowable Stresses in Piles -  In addition to the ability of soil or rock to carry the load transferred from a pile, the load capacity of the pile also is important. The load capacity of a pile is governed by its structural strength and, to a lesser extent, by the surrounding environmental conditions. The structural strength is a function of the allowable stress levels that apply to the particular pile material and the cross-sectional area of the pile.

To provide a factor of safety against failure (figures 5 and 6), allowable stress levels normally are specified as a percentage of the peak strength value of the pile material (for example, steel, concrete, or timber). Allowable stress levels for piles vary significantly because of different building codes in different jurisdictions. Significant controversy had arisen concerning the allowable stress levels used in the highway industry at the time of this research project (circa 1980) because the choice of allowable stress values greatly affects the dimensional analysis and thus the economy of the foundation system. Another factor to be considered was that competition between materials producers was keen and significantly affected by changes in the allowable stress codes.

FIGURE 5.  Concrete piles damaged by difficult driving conditions.

FIGURE 6.  Steel H-piles badly damaged by hard driving.

A comprehensive investigation of the design of piles as structural members was performed to define and establish, through structural analysis and supporting field data, a rational guideline or design methodology for determining allowable stress levels for pile design codes used in highway bridge foundations. These methods must take into account not only the material properties of the pile itself, but also the individual effects of long-term loads, driving stresses and driveability, imperfections in form or material, and various environmental conditions that tend to reduce pile capacity.

As part of the overall review, the investigators studied the codes and specifications of an array of national and foreign organizations and code bodies. The basis for each code was reviewed and corroborative data were assembled to develop improved values of allowable stresses. The claims of material suppliers for increased allowable stress levels also were evaluated and documented. Some of the claims were found to be overstated, and all of them ignored at least one or more of the important factors that govern the structural strength of a pile.

In general, it was recommended that allowable stress values in use at that time be decreased for concrete and timber piles and increased for steel piles. The new procedures for determining appropriate allowable stress levels on a case-by-case basis according to the major factors governing structural strength of piles are a significant improvement over the previously unsubstantiated blanket stress levels. Reasons for each suggested change to then current methods for determining allowable stress values are documented in the FHWA report (11).

2.6.5 Performance of Pile Driving Systems  - Proper construction control of pile driving operations is as essential as good design practices. Construction control is more difficult for piles than for spread footings because the excavation and construction of footings can be observed. The construction control for pile driving involves checking the pile materials and the installation equipment. The pile inspector can visually check many requirements; however, some of the most important checks require instrumentation. The high cost of pile foundations prompted FHWA to seek more efficient pile driving systems as well as more efficient design methods.

Pile driving technology has evolved from an archaic system of pounding a "stick" in the ground with a heavy mass to a sophisticated system of installing long, slender structural elements in a well-defined soil mass. In the previous era it didn't matter how efficient the pile driving system was because the only objective was to pound the stick into the ground. However, the vertical advance of a pile under a given hammer blow can be used as a measure of the pile's bearing capacity. Thus the hammer takes on a second function, and doubles as a piece of testing equipment.

A comprehensive study of available pile driving systems and corresponding measurement techniques was completed under this research program. In addition to the direct measurements of output, researchers investigated the usefulness and practicality of installing gauges (figure 7) and other instrumentation on the pile driving system to determine performance rating, spot the cause of erratic or inadequate operation, and evaluate the significance of erratic behavior in terms of performance capability.

The researchers developed three devices to evaluate the field performance of pile hammers. The ram velocity measurement (RVM) apparatus uses radar to record ram velocity as a function of time for conversion to kinetic energy for driving. The study also developed a system that uses piezoelectric accelerometers to measure pile hammer impact velocity. This system is referred to as AMS (for Acceleration Monitoring System). Both the RVM and AMS are expensive and complicated devices for routine use on pile driving projects. A simpler device, the Saximeter II, was developed for FHWA to aid the pile inspector in log keeping and monitoring pile driving performance. It electronically measures blows per minute and converts time duration to the corresponding fall height (stroke) of the ram, and converts stroke to the potential energy.

An inspection manual was also developed to aid construction engineers in evaluating the performance of pile driving systems (12). It can be used to ascertain that the pile hammer conforms to certain minimum standards, and to record observations on hammer and driving system behavior. The manual explains the theory of operation and inspection procedures for various hammer types, describes all components of the driving systems, includes a glossary of pile driving construction terms, and provides summary sheets and forms to aid the inspector in recording the pile installation process. A slide-tape show was also developed to provide a visual rendering of the inspection manual for instructional purposes.

2.6.6 Pile Wave Equation Technology -  For many years, pile design used pile driving formulae that were established to relate dynamic driving forces to available pile bearing capacity. The original dynamic formula was developed more than a hundred years ago and was based on a simple energy balance between the kinetic energy of the ram at impact and the resulting work done on the soil. The concept assumed a pure Newtonian impact with no energy loss.

A close examination of the pile driving process recently disclosed that the concept of a Newtonian impact does not apply. It was also noted that when the hammer strikes the pile, a force pulse is created that travels down the pile in a wave shape. The amplitude of the pile wave decays before reaching the pile tip because of system damping properties; however, a portion of the force in the wave reaches the tip and forces the pile to penetrate the soil.

Before the demise of the dynamic driving formulae and before the wave equation analysis was fully developed, pile designers relied on static pile analyses to predict pile capacity and determine pile sizes and lengths. The success of this method depended greatly on the accuracy of boring data and the engineer's ability to properly classify zones in the subsoil with regard to relative pile support capability. Static analysis methods will always be very useful because they are much more accurate than the old driving formulae; however, the value of wave equation type methods has been steadily increasing.

The wave equation method is a computer based analysis that was developed from the classical wave theory, which models wave propagation in a slender rod subjected to an applied force at one end. Modifications to the classical theory are necessary to account for changes in the traveling wave form due to pile and soil properties. If the wave is completely dissipated by the pile material and soil properties before reaching the pile tip, no penetration will occur. This can be due to either too small a hammer for the pile, or too small a pile cross-sectional area specified for the length being driven. Wave equation analysis can detect both of these scenarios before they happen in the field, plus it can also predict pile damage if too large a hammer is selected.

Soil data input requires both an understanding of site-specific soil properties and the effects of pile driving on those properties. These dynamic properties are known as damping and quake and are roughly correlatable with soil type. These properties are best determined by experienced geotechnical engineers. Research was needed to provide better methods to determine soil damping and quake values from laboratory and/or in situ soil testing equipment to better predict pile capacity.

2.6.6.1 Pile Capacity Prediction  - In response to this need, FHWA initiated a contract research study to perform an in-depth assessment of current techniques and potential methods for determining soil quake and damping input parameters to the wave equation computer analysis program. The contractor was required to develop a data base of pile foundation sites containing research-quality soils data, pile driving records, load test information, and Pile Driving Analyzer (PDA) data. These data were to be used to make correlations among quake and damping factors, Wave Equation Analysis Program (WEAP) capacity predictions, and soil test data. It was also a requirement to locate 6 to 10 actual projects where load testing and dynamic measurements were made, plus good soil data had to be available to perform the required correlations to evaluate the newly developed procedures and propose additional modifications, if necessary.

The following is a reprinting of the report's abstract:

"Research has been conducted on the potential improvement of dynamic wave equation analysis methodology using in-situ soil testing techniques. As a basis for this investigation, the literature was reviewed and a summary was compiled of efforts made to date on the development of models and associated parameters for pile driving analysis. Furthermore a data base was developed containing more than 150 cases of test piles with research quality data on static load tests, dynamic restrike tests, soil information, driving system data and installation records. One hundred data base cases were subjected to correlation studies using the CAse Pile Wave Analysis Program (CAPWAP) and various static analysis methods. This work yielded dynamic soil model parameters which did not indicate a specific relationship with soil grain size" (13).

2.6.6.2 Simplified Capacity Predictions -  Static load testing (figure 8) to failure is probably the best method available to determine the actual static capacity of a pile; however, these tests are expensive and time consuming. As a result, they are not routinely conducted. In many cases a load test to twice the design load is conducted to verify safety factors and save money on the cost to extend the loads to the failure range. These tests give some measure of assurance to the project design team, but do little to advance the knowledge of how to design a particular pile in a particular soil. The simple act of striking a pile with a heavy hammer can also be thought of as an instantaneous load test to failure; because in order for the pile to penetrate further into the soil, the soil must fail under the driving forces. In other words, pile driving is actually a very fast load test under each hammer blow. With the right type of instrumentation, the engineer can take advantage of these failure measurements and use the information to make a prediction of potential static capacity in the field during pile driving operations.

This concept was proposed to FHWA by the developer for further study. The idea to develop a simplified method based on energy balance between the total energy delivered to the pile and the work done by the pile/soil system was examined under a research contract study in the early 1990's. This method, called the "Energy Approach", uses the calculated transferred energy and maximum pile displacement values from the measured data, together with the field blow counts as input parameters to calculate the pile capacity.

To verify and refine the new method, two large data sets were retrieved from the FHWA Deep Foundations Load Test Data Base (see section 5.2). One set contains 208 dynamic measurement cases on 120 piles monitored during driving and followed by a static load test to failure. These cases reflect various combinations of soil-pile driving scenarios. The other set contains data on 403 piles monitored during driving, but without static load test data.

The Energy Approach method was found to be very accurate in predicting the pile capacities that were measured in the static load tests. These estimates were also found to be more accurate than the sophisticated office methods commonly used in engineering practice. This comparison was especially true for records obtained at the end of initial driving. This approach can be used in place of or as an independent check of the office methods (14).

2.6.7 Micropiles - Micropile technology was conceived in Italy in the early 1950's to fill the need for an economical and versatile foundation system that could be used to underpin, repair or retrofit badly damaged infrastructure elements in war-torn Italian cities. It was introduced in the United States about 20 years later; however, it did not catch on near as much as it did in Europe, especially, in the late 1980's and early 1990's. In addition to static applications, significant growth in the use of this technology has occurred in seismic and slope Stabilization applications.

All of these uses are directly applicable to highway projects, which caught the eye and interest of FHWA geotechnical engineers. Two major earthquakes in California and one in Japan also stirred interest in the use of micropiles for seismic retrofit of highway bridge foundations. At the same time, the French Highway Administration began a major research project to investigate the basic mechanisms and engineering characteristics of micropile systems.

Micropiles are defined as small-diameter, drilled piles composed of placed or injected grout with some form of steel reinforcement in the center of the grout to resist the bulk of the design load. The central reinforcing element is either a high-strength steel bar or tube that is secured in the grout that is injected under high pressure to improve bonding with the surrounding soil. Micropiles can be installed through virtually any ground condition and at any inclination. Modern construction techniques keep noise and vibration to a minimum, and they can be constructed under limited headroom conditions to within a few millimeters of adjacent properties. Figure 9 shows an early model micropile that was exhumed for inspection purposes.

In 1992 FHWA was invited to become a cooperative research partner in the French micropile project.

The major objective of this cooperative research project was to establish reliable engineering guidelines and safe design methods for the use of micropiles in the reinforcement and Stabilization of foundations and slopes.

It was anticipated that the development of reliable engineering guidelines, combined with site monitoring for field performance assessment of micropile systems, would significantly increase the engineer's confidence in the technique and, thereby, greatly enhance its expansion to new fields of application.

The initial investigations included a comprehensive review and critical assessment of available information on current state of the practice, research case studies, site performance monitoring, quality control issues, and any comparisons or analyses of current codes of practice. Several ongoing micropile construction projects were also instrumented and monitored to provide researchers with additional case histories to study and evaluate. The major physical experimental tasks involved centrifuge modeling and related analytical/numerical simulations, full-scale testing, and field monitoring, which were designed to study the engineering behavior of individual micropiles and micropile groups and/or systems under axial and transverse load response modes (figure 10). Buckling, corrosion, and seismic aspects were also investigated under actual working conditions as well as at failure under ultimate loading.

FIGURE 10.  Load testing of micropiles in California.

2.6.7.1 State-of-the-Practice Review - In 1993 FHWA initiated a contract to produce a comprehensive review document that would delineate all of the known information about micropile technology that could be gathered from various organizations around the world. The contractor put together an international group of experts to evaluate the information obtained in the surveys. The review report (15) provides a comprehensive introduction to micropile technology, and assisted FHWA in developing an engineering guidelines manual (7) containing reliable design guides, construction specifications, and quality-control procedures for the wide spectrum of micropile applications.

2.6.7.2 Seismic Behavior of Micropiles - In 1995, FHWA initiated a contract research study to develop improved seismic design methods for micropile systems used in new construction in earthquake zones, seismic retrofitting of bridge foundations, vertical excavation retainment systems, and slope Stabilization. Laboratory centrifugal model tests were conducted to correlate with numerical model studies on various micropile systems (isolated piles, groups, and networks of piles) to evaluate their behavior under axial, lateral, and combined loadings in selected engineering applications. Shaking table tests were also conducted at Christchurch University in New Zealand as part of the overall study to improve seismic design methods. This study is scheduled to be completed after this report is printed.

2.6.8 Publications and Implementation Items -  Research and development activities for pile foundations were very successful. Results were documented in numerous reports, and several technology sharing reports were issued to help implement the findings. Several workshops and symposia were held to promote the new guidelines and prediction methodologies that were developed under the research program. Several new articles were added to the AASHTO Bridge Specifications and many others were updated based on the FHWA pile foundation research results (see chapter 7).

2.7 Pile Groups

The state of the art for pile foundation design at the beginning of this project did not include an accurate method for relating the ultimate bearing capacity and settlement behavior of a single pile to the behavior of a group of closely spaced piles. To develop such a method, it was necessary to conduct field load tests to failure of full-scale pile groups to obtain field data that were useful in interpreting fundamental phenomena that control the behavior of groups of driven piles.

Older methods of pile group design treated the group as a collection of individual piles requiring an adjustment factor. Predicting the behavior of pile groups required correction of the load capacity of the individual piles in the group for the interaction effects transmitted through the soil mass. How this should be done was never certain, and it was recognized that it would probably be different in cohesive as contrasted to granular soils.

In 1980, FHWA initiated a research program to investigate pile group behavior through carefully performed experiments. First, existing mathematical models used to design pile groups were identified and evaluated. From this evaluation, the "hybrid model" was selected and used to analyze a proposed full-scale pile group to be load tested.

The hybrid model, a load-deformation model, reasonably predicts load versus settlement, load transfer patterns, and load distribution to pile heads. The model was especially helpful in designing the instrumentation system for the full-scale pile group load test. The field data acquired from the load tests were used to refine the hybrid method, which led to the FHWA PILGP1 computer program, a modification and refinement of the hybrid model (see section 2.7.3).

2.7.1 Pile Groups in Clay - To develop the field data required to verify and refine PILGP1, 11 instrumented steel pipe piles were driven into a very stiff, saturated, over-consolidated clay soil at the University of Houston, Texas, campus. The outside diameter of the piles was 273 mm (10.75 in), with a wall thickness of 9.27 mm (0.365 in). The piles were driven closed-ended 13.1 m (43 ft) deep. Nine piles were driven in a 3 x 3 square array on a spacing of three pile diameters. The two remaining piles were driven apart from the group to serve as control piles.

Each of the 11 piles was instrumented with full bridge strain gauge transducers and mechanical telltales to monitor load transfer from the piles to the surrounding soil. Four of the group piles and one control pile were instrumented with piezometers and lateral total pressure cells. The surrounding soil also was instrumented with piezometers and vertical ground movement devices. Electronic load cells measured applied loads, and settlements were measured at each pile head. Three-dimensional translations were measured on the massive concrete pile cap that was suspended off the ground.

The load testing program consisted of 11 compression and 6 uplift (tension) tests, all of which were carried to ultimate failure loads. The control piles were load tested in compression at three time intervals to study the effect of soil setup. Each control pile also was tested in uplift. The nine-pile group (figure 11) was tested in compression atthree time intervals to assess setup characteristics of the group. Four of the group piles were tested in uplift; however, these tests were preceded by compression tests on two smaller groups of four and five piles. The five-pile subgroup was formed by cutting away the four corner piles, leaving the four edge piles and the center pile. The four-pile subgroup was formed by removing the center pile (8,9).

FIGURE 11.  Pile group load test in clay.

2.7.2 Pile Groups in Sand -  A second pile group load test study was performed on a group of eight timber piles in sands at the Locks and Dam No. 26 near Alton, Illinois, as part of an evaluation of several rehabilitation schemes for the distressed locks and dam structures. The timber piles were instrumented to measure load transfer and deformation up to and including the failure load. The acquired field data were used to evaluate and refine the PILGP1 program (16).

The third pile group study involved load testing five single piles and a group of five piles in sand at a test site in Hunters Point, California (figure 12). The load test results were also used to evaluate PILGP1 and a simplified procedure developed under previous studies of the behavior of single piles and pile groups. Data were obtained on the comparative behavior of piles of different types with the same relative geometry. Residual stresses, load transfer, and load-settlement characteristics were measured and used to evaluate the effectiveness of the new pile design and analysis procedures. An isolated control pile and three of the five group piles were fully instrumented, to obtain the load-response data for the analysis. Soil and ground water conditions were evaluated with standard and state-of-the-art equipment, such as Standard Penetration Test, Dutch Cone, Stepped Bladed Vane, Dilatometer, and Pressuremeter testing devices (17). Figure 13 shows a closeup view of the test pile group and two soil instrumentation boreholes.

After each of the first and third pile group load test studies, a pile group prediction symposium was held to discuss the test results and evaluate the most popular pile group design methods in use at that time (18,19,20).

FIGURE 12.  Reaction frame for pile group load test in sand.		FIGURE 13.  Instrumentation bore holes adjacent to pile group.

2.7.3 Predictive Model for Pile Group Design and Analysis (PILGP1) - A comprehensive computer-aided design method called PILGP1 was developed for FHWA at the University of Houston for the design and analysis of pile groups. It is a load deformation model that predicts load versus settlement behavior, load transfer patterns, and load distribution to pile heads. The model designs the pile group as an interactive element rather than as a collection of individual piles requiring an adjustment factor. It was developed from the results of the first full-scale field load test program on single piles and pile groups. Two additional full-scale field load test programs were later completed to further refine and validate the computer model. The current version of the model (PILGP2) is being tested by several consulting firms, States, and Federal agencies. Continued testing, evaluation, and refinement of the model is currently under way.

The development and verification of PILGP1 was based primarily on the full-scale field tests of pile group behavior under both working loads and failure conditions. Because of the many variables involved, numerous full-scale field tests needed to be conducted to provide a statistically meaningful data base. However, the costs involved in full-scale field testing significantly restricted the number of tests that could be conducted. The alternatives to full-scale field testing are model field testing, laboratory model studies, and centrifuge model testing.

2.7.4 In-Service Monitoring of Pile Groups -  The three full-scale pile group load tests to failure produced very accurate and valuable data; however, they were very expensive to conduct. In addition to the pile group load tests to failure, four full-scale field projects were initiated with FHWA research funding to observe pile group behavior under working loads. The short- and long-term behavior of the in-service piles was compared with analytical predictions made by PILGP1.

One of the projects was located on the Natchez Trace Parkway (NTP) in Mississippi, where a pile-supported bridge abutment was instrumented to obtain load transfer data on a group of six steel piles (12 HP 53) in soft clay and silt. Samples of load/settlement curve data are shown in figure 14. Another project was located at Fort Belvoir, Virginia, where soil and pile conditions were similar to those at the Mississippi project. Instrument readings were taken weekly during the first year at each site, monthly during the second year, and quarterly for several years.

The third site, at the Mocks Bottom over-crossing in Portland, Oregon, near Swan Island, is underlain by a thick compressible clayey silt deposit. High down-drag loads were expected because of the approach embankment loads on the compressible soil. A bitumen coating was used to reduce down-drag loads by about 90 percent. Although the bitumen-coated piles cost about 15 percent more than the uncoated piles, fewer piles were required. Pile instrumentation included settlement and load transfer monitoring.

The fourth site was located at the West Seattle Freeway in Washington where a group of twelve 310-mm- (12-in-) diameter concrete piles supports a pier in medium-dense sands. The bridge pier and pile cap were instrumented to measure the amount of load transferred to the pile cap, and each pile was instrumented to measure load transfer from the pile cap to the top of the piles. Three piles were instrumented for load transfer along the entire pile length of 30.5 m (100 ft).

2.7.5 Model Testing - The high cost, measured in both time and money, of obtaining high quality data from full-scale load tests on single piles and pile groups, led to the FHWA staff study to determine if accurate data could be obtained by conducting model tests in simulated ground conditions.

Using carefully controlled large test pits located at FHWA's Turner-Fairbank Highway Research Center (TFHRC), more than 200 model tests on single piles and pile groups have been completed on several different types of piles in sand and clay soils. These data are being compared with the carefully controlled centrifugal and full-scale single pile and pile group load test data. This work is continuing as more variables are studied, and results are entered in the load test data base to be used for verifying and refining the PILGP1 model.

An investigation of scale effects is an important part of this study to determine if the model test data can be productively correlated with the load test data of the full-scale field tests. The establishment of appropriate scale factors allowed more model tests to be substituted for expensive full-scale tests. Small-scale models permit numerous parametric studies at a reasonable cost, and allow soils and other conditions to be carefully controlled.

The first series of laboratory model tests was patterned after the timber pile field study at Alton, Illinois. The sandy soil at the Alton test site was matched as closely as possible at TFHRC. Model load tests were run on single piles and pile groups (figure 15) at 1/20, 1/15, 1/10, and 1/3 full scale. As a minimum, three load tests were performed for each scale. Each pile was instrumented with strain gauges to measure load transfer.

FIGURE 15.  Load test on small-scale model pile group in test tank.

The second series of laboratory model tests was patterned after the full-scale load test on steel pipe piles in clay. Model tests were run at 1/15, 1/10, 1/6, and 1/4 full scale. The laboratory model tests on the 1/20, 1/15, and 1/10 scales were performed in a steel tank 1.5 m (5 ft) in diameter and 1.5 m (5 ft) deep. Because the 1/6, 1/4, and 1/3 scale models were too large to be tested in the laboratory test mold, outdoor test pits were constructed at the TFHRC site (figure 16). The third series of tests was patterned after the full-scale pile group load test to failure at Hunters Point, California.

FIGURE 16.  Load test on large-scale model pile group in test pit.

The scaling factors identified in this study were used to establish relationships between load deformation behavior of reduced-scale and full-scale piles and pile groups. These small-scale tests provided a significant amount of data to validate PILGP1 at less cost than full-scale field testing.

2.7.6 Centrifuge Model Testing - Small-scale models permit parametric studies at reasonable cost and allow soils and other conditions to be carefully controlled; however, it is difficult to achieve similitude between corresponding stresses and strains in the model and prototype. The response to load of a small pile and a large pile cannot be modeled by any simple, direct relationship derived by ordinary dimensional analysis. The question of scale effects must be resolved before any useful relationships for pile design can be developed.

Models of large, heavy structures where gravity is a principal loading factor are not effective indicators of prototype behavior because the state of stress in the model caused by self-weight will be much lower than in the prototype. If the model can be placed in an artificially high gravitational field, the state of stress limitation can be counteracted almost entirely. A centrifuge apparatus provides the necessary accelerated gravity rate to load test the model under simulated gravitational forces. However, to accurately measure stresses and strains, the centrifuge must be able to accommodate a model that is large enough to handle the required instrumentation. The larger centrifuge capacity provides more accuracy in direct modeling of large prototype structures. A pilot study validated the feasibility of using centrifuge techniques for corroborating the PILGP1 mathematical model.

In a larger study, the centrifuge was used to test models of the full-scale pile groups that were load tested to failure in the previously described field studies. The combination of centrifuge model testing and small-scale laboratory testing of conventional models at TFHRC provided valuable physical data to establish relationships between pile groups of varying scales in the same environment.

The larger, more comprehensive centrifuge testing program on single piles and pile groups was performed in sand and clay soils. The results were used to test and verify assumed similitude relationships between scaled models and full-scale prototypes. The load deflection and load transfer data were used to predict full-scale performance and compared favorably with both measured results and predictions made by computer generated results. The experimental procedures developed and the verification established in the program has encouraged the investigation of the many factors that influence pile performance under controlled laboratory settings. The centrifuge technique should also be useful and very cost-effective in establishing the predicted behavior and the sensitivity to design changes of pile foundations for large projects (21).

2.7.7 Lateral Loads on Pile Groups - Several analysis and design methods for pile groups under lateral loading have been proposed, but none had been validated using carefully performed field experiments on full-scale structures. The theory of elasticity and a number of "efficiency" formulas often were used as analytical methods for lateral load design. One of the most promising methods is called the Poulos-Focht-Koch (PFK) procedure. This method was evaluated through field load tests. Data from the load tests also were used to develop an improved method of analysis.

The main objectives of the field load tests were to provide high-quality field data on the performance of a full-scale pile group under lateral loading and to compare measured response with that predicted by the PFK method. The load test was performed at the FHWA pile group test site at the University of Houston, Texas, campus using the same piles and some of the instrumentation used in the vertical load tests previously described.

The results of these tests were also evaluated under a cooperative study with the U.S. Army Corps of Engineers that was performed at the University of Houston site. After the lateral load tests in clay were completed, the researchers replaced the top 3 m (10 ft) of clay with sand backfill and repeated the load tests. The test results were used to refine and validate the predictive models (22).

2.7.8 Publications and Implementation Items -  Research and development activities for pile group foundations were very successful. Numerous quality reports were generated from an unprecedented program of research into an area where few have ventured and none to the same extent. Textbooks and specification codes have incorporated the results of these tests, and the state of the art was advanced significantly.

2.8 Drilled Shafts

A drilled shaft is a machine-excavated circular hole in soil and/or rock that is filled with concrete and reinforcing steel to support heavy structural loads in single or multiple units. Vertical loads are resisted by both the base area of the shaft and in side friction which can be very significant because the concrete is cast wet and cures directly against the soil forming the walls of the borehole. They are sometimes socketed in rock, and steel casing is sometimes required for hole Stabilization that may or may not be removed. Horizontal load is resisted by the shaft in horizontal bearing against the surrounding soil or rock.

Drilled shafts have many advantages that set them apart from piles and spread footings. In fact each foundation element has certain pros and cons that have to be weighed one against the other when deciding on which system to use in a particular design situation. Each system also has several disadvantages that have to be evaluated for each design situation. To give confidence as to performance with respect to the intended task, any foundation element requires both a reliable method of construction and a standard to define acceptance after construction.

In this regard, drilled shafts were considered to be less reliable than others because of the uncertainty of the effects of construction on the actual service behavior, and the limited knowledge of either reliable quality-control tests to locate and evaluate defects or inexpensive load test procedures. Even if there are only occasional failures, they highlight the variables and unknowns present when working underground, particularly in water-bearing and potentially caving soils. This results in a lower risk tolerance for a single or double shaft supported pier compared with multiple pile supported foundations.

Because drilled shafts for many design situations offer higher capacities with potentially better economics than driven piles, the FHWA has spent considerable time and money in research and development of improved design and construction guidelines for drilled shafts. Other advantages include less noise and vibration during construction and the ability to go very deep to avoid scour problems. A major research program was designed to evaluate existing nondestructive testing techniques for identifying defects and/or results of adverse downhole conditions that impact the load transfer/settlement behavior of drilled shafts. It was also planned to develop rational acceptance criteria for defective drilled shafts on the basis of quality control during construction, plus a field test program to verify the research findings that included a search for more economical methods to test shaft capacity with and without defects.

2.8.1 Nondestructive Evaluation (NDE) of Drilled Shafts - In July 1988, FHWA initiated a contract research study to examine drilled shafts for the effect of defects on performance, and to develop acceptance criteria for use by construction engineers to accept, reject, or modify a newly constructed drilled shaft. The study included the construction of 20 drilled shafts with and without defects for different soil sites located in California and Texas (figure 17). The shafts were constructed using different techniques: dry construction, and wet construction using drilling water, controlled bentonite slurry, and controlled polymer slurry. Five instrumented shafts were statically load tested (figure 18) to determine the effect of the man-made defects on shaft performance, and all shafts were dynamically load tested to correlate with static results. All shafts were tested non-destructively using both surface reflection and direct transmission techniques to determine their effectiveness in identifying defects and/or the results of adverse down-hole conditions that impact the load/settlement behavior of the shafts; results were summarized and evaluated in the report. The allowable defect criteria developed consider the design basis, the ratio of design stress to a maximum code allowable, the type of stress, the level of quality control, and the risk tolerance.

FIGURE 18.  Vertical load test of defective drilled shaft.

In addition to the establishment of acceptance criteria, a rating guideline for implementing special integrity testing and a decision tree were developed to guide the engineer or decision maker through construction of a drilled shaft project that includes nondestructive testing. The report also includes detailed information on four promising low-strain NDE techniques (sonic echo, impulse response, cross-hole sonic logging, and gamma-gamma logging), which can be used as part of the quality control procedure, plus information on three alternative load test procedures that show high potential for being economical substitutes for expensive static load tests (see section 2.10).

The sonic echo (SE) and impulse response (IR) test methods use impact energy applied to the top of the shaft to generate energy waves down the shaft and return to a receiver that measures the vibration response. These methods are not able to locate deep defects and they aren't able to detect the size and location of the defects. Smaller defects lying below a larger defect are easily masked from above. Cross-hole sonic logging (CSL) and gamma-gamma logging (GGL) overcome these problems by being downhole methods that pass ultrasonic or radiation waves through the concrete between source and receiver probes in a water-filled tube or hole pair as the probe cables are pulled from the bottom back to the surface over a depth measurement wheel. These methods test the quality of the concrete lying between a pair of tubes. Four tubes installed in a shaft before concreting gives sufficient coverage to adequately inspect a shaft for defects. Figure 19 shows an end view of a rebar cage with four NDE access tubes.

Results clearly show that the CSL method is superior. The GGL method requires a radiation source and requires PVC tubes because steel is not compatible with the radioactive materials. This method is also sensitive to defects close to the test tubes, and doesn't have the same range away from the installed tubes that CSL has. In addition, CSL is generally much faster to perform and does not use radioactive materials (23). Figure 20 shows a technician performing a surface deflection test on one of the defective shafts.

FIGURE 19.  Instrumented drilled shaft showing man-made defect and NDE access tubes.

FIGURE 20.  Low-strain impact test with instrumented hammer and geophone sensor.

2.8.2 Drilled Shafts in Intermediate Geomaterials -  Current design methods for drilled shafts in soil or competent rock are reasonably well-founded; however, comparatively little effort has been expended to determine design parameters for intermediate materials such as shales, claystones, and marls. In September 1991, FHWA signed a second contract to develop, test, and recommend criteria for determining appropriate exploration, in-situ testing, and the necessary inputs to the load transfer function or other design methods for drilled shafts in intermediate geomaterials.

The contract consisted of essentially three tasks:

1. The first task required a literature search and discovery of case histories involving drilled shafts. Of particular interest were projects that employed analytical or semi-analytical methods based on soil mechanics principles and load-transfer function concepts, as well as locations where load test data existed.
   
   
2. Using existing exploratory and sampling techniques that were known to reliably characterize the geomaterials and quantify design material parameters, a preliminary set of criteria for exploration, in-situ testing, and load transfer design mechanisms was developed.
   
   
3. Using this pilot test program, specific plans,specifications, and estimates for conducting research tests were developed for locations where state or private projects included load testing of drilled shafts. The contractor performed soil characterization tests and predictions of shaft behavior, including instrumenting selected shafts for load-transfer analysis. Upon completion of these projects, the pilot criteria and design methodology were revised or modified to better predict drilled shaft behavior.

Data were collected from projects in Australia, Georgia, Texas, Massachusetts, and Kentucky. These data have been used to develop the criteria for exploration and sampling of soil materials to use for developing new load transfer functions for the design of drilled shafts in shales, tills, and other materials between soils and hard rock. A field test program was also designed to validate the new design methods (24).

2.8.3 Free Fall of Concrete in Drilled Shafts  - This study developed information and data that were used to justify increased depths of free-fall placement of concrete into properly constructed clean and dry shafts without meaningful loss of strength or segregation of the concrete aggregate. The report of findings has had a significant economic impact on the drilled shaft industry.

Prior to this study, the question of whether the free-fall of concrete over long distances adversely affects the concrete strength and integrity in drilled shafts persisted in the minds of many engineers and construction inspectors. This question persisted despite past efforts to answer it and dispel the concern that concrete does not segregate during free-fall at any height, and that free-fall placement can be accomplished without adverse effect on the concrete.

To accomplish the research goals in a cost-effective way, four 18-m- (60-ft-) long, 0.9-m- (3-ft-) diameter shafts evenly spaced and tangent to a central 1.5-m- (5-ft-) diameter access shaft were constructed. The four test shafts were divided into six 3-m (10-ft) sections with one of four different concrete mixes placed in each section. The slump, maximum aggregate size, and placement procedures were also varied. The low slump mixes were also placed with and without super plasticizer. The three placement procedures were free-fall central drop with careful control to be sure that the concrete didn't strike the rebar cage, free-fall sloppy drop with effort actually made to see that the free-falling concrete did hit the rebar cage, and tremie placement with a tremie pipe extended all the way to the concrete placement level.

The results of the research were very conclusive and positive. Construction codes have been changed in many cities, counties, and States, accordingly. More specifically, the results were as follows:

None of the lifts placed by central drop free-fall procedures within the research program exhibited any signs of aggregate segregation. The design strengths of all centrally dropped lifts varied from 13 percent less to 20 percent more than the reference cylinder strengths. All of the strengths were well above the intended design strength.

Due to the small variation in the compressive core strength and lack of aggregate segregation, no definitive effect of slump, aggregate size, height of drop, depth of fluid pressure, or addition of super plasticizer was discerned.

Surprisingly, in six out of seven direct comparisons made between sloppy drop and central drop placement procedures, the sloppy drop methods actually resulted in higher average compressive core strengths than equivalent central drop procedures. Also, no segregation of aggregate was noted for any of the sloppy drop mixes placed. Thus, on the basis of this research, it is concluded that striking the rebar cage or the side of the shaft does not have a detrimental effect on the strength or integrity of the concrete.

Due to the high strengths and lack of segregation that were apparent in all of the sloppy drop lifts, the effects of aggregate size, slump, height of drop, height of fluid pressure, and addition of super plasticizer did not appear to affect the results in a meaningful way for the well-designed concrete mixes. Even though sloppy drop procedures were not found to affect the strength or segregation of the concrete, it is not intended that contractors should begin to place concrete in a haphazard fashion. The sloppy drop procedure adversely affected the placement of the rebar cage and also caused additional concrete contamination as a result of traveling down the soil sides of the shaft. In all cases, the shafts were fully formed and no honeycombing, voiding or exposed rebar was evident (25).

2.8.4 Load and Resistance Factor Design of Drilled Shafts - In early 1997, FHWA co-funded a research study on "Resistance Factors for Drilled Shafts with Minor Defects" with the Association of Drilled Shafts Contractors (ADSC) and a number of SHA's including California, Florida, Illinois, Minnesota, Montana, North Carolina, Pennsylvania, and South Dakota. The objective was to develop design techniques to account for sub-detectable minor defects that can occur during construction. The analytical study and most of the experimental work was done by the University of Houston at its National Geotechnical Experimentation Sites (NGES) facility. Some of the future testing will be done by FHWA staff at the TFHRC.

It is proposed to derive resistance factors (or "workmanship factors") for drilled shafts through analytical modeling that will be calibrated by performing field loading tests to structural failure on drilled shafts with selected defects. The load testing will primarily be lateral because the loss of moment resistance in the lateral mode of loading is potentially more severe than loss of axial capacity for drilled shafts with steel percentages normally used in highway bridge construction.

A computer program called LPILE Plus was modified to perform accurate and realistic modeling of the influence of minor defects on drilled shaft behavior. The modified version allows for the development of an independent stress-strain curve within the defect area. It also allows for a reduction in the tensile strength of both the defective and nondefective portions of the shaft section. A defect may be located in either the compressive or tensile portion of the shaft. It is also possible to include more than one defect in the same section. Six field lateral loading tests were conducted to verify and calibrate the modified LPILE model.

Lateral load tests were conducted by jacking a test shaft against a reaction frame consisting of a wide flange steel girder supported by two reaction shafts. A manual jacking system was used to jack the test shaft against the reaction frame. Loads were measured with a load cell and dial gauges were used to monitor lateral deflections of the test shafts. All test shafts were constructed under wet conditions using a polymer slurry.

The test shafts were then extracted and examined carefully to determine the location of the plastic hinges, and to caliper the exact dimensions along the shaft axes. The results of the load tests will be used to verify or to calibrate the modified version of LPILE. This modeling will be eventually used to develop recommendations for resistance factors for drilled shafts under combined axial and lateral loading. Preliminary recommendations will be developed in this phase of the research, while more general and refined recommendations will be developed in the following phases if the results of the preliminary phase justify further study.

Because typical drilled shaft defects are usually found at the boundaries between soil and concrete, they can have effects on both the structural performance of the drilled shafts and the soil-structure interaction. To isolate the effect of the structural defects from the variation in the soil resistance, structural laboratory tests will be conducted on slightly defective shafts out of the soil at the TFHRC, if the preliminary field test results appear to justify the effort. The results of this testing program will also be used to verify the method in the modified version of LPILE that is used to model the changes in cross-section and stress-strain behavior within the concrete drilled shaft.

Eight drilled shafts will be tested in the TFHRC laboratory experiment. Both the diameter and steel reinforcement of the laboratory experiment will be identical to those of the field load tests. The laboratory test shafts will include defects essentially similar to those included in the field tests. To investigate the effect of rebar corrosion, the diameter of the rebars of two additional defective shafts will be reduced to one-third of the original size within the defect. Six slightly defective drilled shafts will be tested under pure bending moment conditions. One of the other two shafts will be tested under combined shear and bending moment, and the last shaft will have no defects (reference shaft) and will be tested under pure bending moment conditions.

If the results of the load tests at TFHRC and NGES-UH prove successful, additional field loading tests will be conducted on slightly defective shafts over a larger range of soil conditions and for a larger range of defect types sufficient to provide recommendations for the AASHTO code committees. It is also possible that the testing will be extended to include fixed-headed shafts because the preliminary work only looked at free-headed shafts.

It is expected that these results will help to solve the problem or source of conflict between some (usually owners) who interpret any anomaly that shows up on NDE signatures as a defect and others (usually contractors) who believe that all anomalies are not necessarily defects that require attention. Serious defects require further probing, immediate repair, or penalty charges assessed against the contractor; however, minor defects should be accounted for in design through the derivation of appropriate resistance factors.

2.8.5 Publications and Implementation Items  - Research and development activities for drilled shafts were very successful. Several quality reports were generated and appropriate specifications and codes were revised (see chapter 7) to reflect the positive results and improved procedures developed under the research program. The FHWA design manual and instructional workshops were also updated to reflect the advancements and contributions of this program.

2.9 Spread Footings

Foundation engineering is one of the oldest professions of mankind. The major decision in selecting a foundation system has always been whether to use shallow or deep foundations to support a superstructure. Foundation support is achieved by transferring the loads to the ground materials without incurring excessive settlements or distortions. Shallow foundations (spread footings or mats) normally are less expensive to construct than deep foundations (piles, drilled shafts, or caissons). However, surface soils usually are less capable of supporting heavy loads than deep soil deposits or bedrock. Deep foundations provide extra security against bearing capacity failure or settlement. A greater risk of unexpected engineering and contractual problems exists, however, during deep foundation installation. Engineering and cost analyses are necessary to determine the proper foundation system for major structures, such as bridges, buildings, dams, and nuclear power plants.

The use of spread footings to support highway bridges in the United States varies widely between States and appears to be far below the optimum level. A recent survey by FHWA determined that most States use pile foundations to support the majority of their bridges. The extensive use of piles to support U.S. highway bridges is contrasted by the extensive use of spread footings to support highway bridges in some foreign countries. For example, highway bridges in England seldom are founded on piles despite severe subsidence problems from coal mining. Another contrast can be seen in the building industry where spread footings and mat foundations are used quite extensively even though building elements (for example, doors, windows, elevator shafts, and utilities) have less tolerance to settlement than bridges.

In the United States, many tons (megagrams) of pile materials and large sums of money have been devoted to using pile foundations where spread footings might have been appropriate. The extensive use of piles in highway bridge foundations may have been encouraged by the AASHTO Bridge Specifications, which at one time stated that "piling shall be considered when footings cannot, at reasonable expense, be founded on rock or other solid foundation material" (circa 1990).

Some of the reasons that piles were preferred over spread footings as a foundation for highway bridges include the following:

• The lack of well-documented performance evaluation data.
  
  
• The lack of rational tolerable movement criteria for bridges.
  
  
• Skepticism and uncertainties concerning the potential cost-savings and the accuracy of settlement predictions for spread footings.
  
  
• Skepticism and uncertainties concerning the quality of fill or natural ground below the spread footings.

The advent of reinforced concrete in the early 1900's and recent improvements in excavation technology have increased greatly the appeal of spread footings for bridge foundations. Modern soil mechanics and improved methods of site investigation and laboratory testing have improved the accuracy of settlement and bearing capacity predictions. Also, compaction control and improved grading procedures have minimized spread footings being founded on weak and/or compressible soils. Finally, special ground improvement techniques such as soil reinforcement, stone columns and dynamic compaction have increased the attractiveness and applicability of spread footings.

Many highway agencies have a policy not to use spread footings on cohesive soils because of the concern for bearing capacity failure and/or excessive settlement. This is a conservative approach because some cohesive soil deposits (especially over-consolidated clays) can support heavy bridge loads without distress resulting. Although bearing capacity on sands is not a problem, some highway agencies do not use spread footings on cohesionless soils because of the concern for excessive settlement. This also is a conservative policy because a spread footing on sand usually will provide satis-factory support because consolidation of sands usually is minimal and occurs rapidly. Most of the settlement occurs before the sensitive superstructure elements are erected.

The use of spread footings on compacted fill also is infrequent. Although a properly compacted fill often is stronger and more stable than natural ground and easily able to support a spread footing, designers often use spread footings on in-situ soils and avoid their use on prepared fills. Large settlements from fills usually can be traced to older, nonuniform fills that may have been constructed of poor soils or uncompacted waste materials dumped on unprepared natural ground surfaces. However, the use of random, uncontrolled fill in modern highway construction, especially near bridges, is readily avoidable. The Connecticut and Washington State transportation departments frequently support bridge abutments on spread footings on compacted fill.

A foundation system must be functional as well as safe. There is a wide degree of engineering performance between an unyielding support system and one that fails. Persistent maintenance problems and failures of noncritical elements (such as parapet walls and joints) are expensive to correct and should be avoided if peculiar to certain systems, situations, or methodologies. To improve the design process, engineers should correlate functional distress (bumps, cracks, and misalignments) with system characteristics (abutment type, soil type, superstructure type, and amount and kind of movement) to determine where spread footings are and are not appropriate.

2.9.1 Performance Evaluation Studies -  To increase the number of documented case studies of spread footing performance, FHWA staff, in cooperation with the Washington State Department of Transportation (WASHDOT), evaluated the performance of numerous highway bridge abutments supported by spread footings on compacted fill. During this review, the structural condition of 148 highway bridges throughout Washington State was visually inspected. The approach pavements and other bridge appurtenances also were inspected for damage or distress that could be attributed to the use of spread footings on compacted fill.

On the basis of this review and detailed investigations of the foundation movement of 28 selected bridges, it was concluded that spread footings can provide a satisfactory alternative to piles, especially when high embankments of good-quality borrow materials are constructed over satisfactory foundation soils. None of the bridges investigated in Washington displayed any safety problems or serious functional distress; all bridges were in good condition (26).

On the basis of the results of the Washington Department of Transportation study, FHWA decided to expand the idea of spread footing evaluations by initiating a comprehensive field investigation of 10 new bridges being constructed in 5 northeastern states. Under this research contract, a long-term study of the settlement performance of 24 spread footings supported on sand was completed to provide a reliable data base for engineering evaluation. Figure 21 illustrates the instrumentation monitoring plan for this study.

The results of the northeastern bridge studies served to confirm the performance results of the WASHDOT study and added to the value of the FHWA data base. The researchers were also able to locate 10 good-quality case histories in the literature for use in the data base. Many cases were identified but most were found to be lacking sufficient settlement or soils data to be included in the study. It was also noted that all of the case histories in the data base represented actual projects where loads were limited to working stress levels. A definite need existed to obtain settlement and soils data at failure loads to provide a larger and more comprehensive statistical basis to judge the effectiveness of settlement predictions and the satisfactory performance of spread footings on sand (27).

FIGURE 21.  Schematic of instrumentation plan for spread footing performance evaluation study.

A follow-on study was initiated to develop a rational prediction method and a user-friendly data base with prediction and correlations modules for use in designing spread footings on sand. As part of this study, the contractor built five large model footings on sand at the Texas A&M University's National Geotechnical Experimentation Site. Each footing was carefully instrumented and load tested (figure 22) to failure (28). A prediction symposium was also held to discuss the results of the load test program and to evaluate the most popular settlement prediction methods in use at that time (29).

FIGURE 22.  Load testing of large model spread footing.

2.9.2 FHWA Staff Research on Spread Footings  - A series of smaller model load tests was concurrently conducted at TFHRC to evaluate the effects of a high water table, depth of embedment, size and shape effects, and the use of geosynthetic reinforcing elements within the sand beneath the footings (see section 3.5.3 for complete details). A total of more than 100 model footings was instrumented and load tested to failure. The test data were added to the data base (see section 5.2.2). All of the footing load tests were performed by following the ASTM Guidelines (D1194-72-1993).

A minimum of five nuclear density and moisture content readings were taken at each 0.3-m lift of sand during placement operations to characterize the soil fill. After filling the test pit, in-situ soil tests were done to complete the soil characterization work.

Results show that embedment has a very significant positive influence on the performance of footings of the same size, even when the water table is close to the bottom of the footing. Placement of the footing base at a depth equal to the width of the footing increases the ultimate bearing capacity by three to five times the value of a footing resting on the ground surface.

The depth of the water table also affects the ultimate bearing capacity of the footing. Those footings resting on sand with a high water table performed much worse than those where the water table was lowered some distance below the base of the footing. Bearing capacity values increased by a factor of two when the water table was lowered from the surface to a depth equal to the width of the footing. Also, as expected, the bearing capacity increased with increasing width of the footing and relative density. A comparison was made between the measured results of each footing load test and the predicted values for bearing capacity using the results of each in-situ test (30).

2.9.3 Dynamic Testing of Footings -  At both the TFHRC and Texas A&M University (TAMU) sites, a separate but coordinated study was conducted to evaluate a novel idea to develop a dynamic measurement system that can be used to quickly determine a soil stiffness modulus beneath a footing in order to ascertain if the natural ground or man-made fill is of sufficient quality to adequately support the imposed loads on the footing. An NDE test, termed the Wave Activated Stiffness test, was proposed to be a more efficient, less costly method of checking for bearing capacity and settlement adequacy than conventional static load testing or performance observations. These are "after-the-fact" methods as opposed to the "before-the-fact" dynamic methods that are done during construction, just after the footing is built and before loads are applied.

This nondestructive impact test is called the WAK test for short, rather than WAS because of the common use of "K" to symbolize soil stiffness in most design equations, coupled with the fact that the impact of the instrumented hammer when it strikes the footing sounds like a loud "whack." The instrumentation consists of a force transducer attached to the tip of the sledgehammer and geophones placed at the diagonal corners of the footing to record the footing-soil response to each impact blow. The force transducer provides the input to the system (impact force) and the geophones yield its response (velocity of the footing).

The impact of the hammer causes the footing and a bulb of soil directly beneath the footing to vibrate. The velocity of this vibration is measured by the geophones. The theory is based on a simplified, linear system model of the soil-footing assembly. This model consists of a mass, a spring, and a dashpot, which is frequently used to describe the vertical vibration of soil-structure systems. The compression wave that travels down a steel pile when struck by a heavy hammer is another example of this modeling tool.

Twenty-two WAK tests were conducted at the FHWA TFHRC site and 16 at the TAMU site. Conventional static load tests were also performed on each footing where a WAK test was done. All results indicate good agreement between static load test stiffness and the dynamic stiffness from the WAK test. Each of the footings was tested four times with the WAK equipment; twice before and twice after the static load tests were conducted.

A plot of dynamic stiffness results against static stiffness results for the entire series of footing tests yielded a graphical relationship that allows one value to be picked off the plot if the other one is known or determined by testing. Therefore, a dynamic result can be used to approximate a static stiffness parameter K (which represents the slope of the initial part of the static load settlement curve) without actually running an expensive static load test.

The manufactured K value can be used to estimate the supporting capacity during construction when there is still time to correct any problems. This type of test can also provide a small measure of comfort to design engineers who fear that construction operations may not actually provide a solid foundation base beneath the footing. Such a test will help to identify poor compaction problems and other possible construction deficiencies that could occur when spread footings are used in lieu of piles (28).

2.9.4 Publications and Implementation Items

Research and development activities for spread footings were very successful. The research reports documented the testing and evaluation results and presented recommendations and supporting information for using spread footings in lieu of deep foundation elements in situations where foundation soils could safely transmit bridge loadings without damaged settlements resulting. Technology transfer items were developed to inform practitioners about the research findings and new ideas being promoted. The AASHTO Bridge Specifications were revised as a direct result of this research project on spread footings (see chapter 7).

2.10 Foundation Load Testing

During the past 20 years of bridge foundation testing and evaluation, FHWA researchers have learned a lot about what works and what doesn't work. Hundreds of load tests have been conducted at TFHRC by staff personnel, and hundreds more have been funded by FHWA in field and laboratory projects. Numerous innovative improvements have resulted from these efforts, which, in turn, have led to advancements in the testing techniques and in foundation engineering design and analysis.

All of the early work and most of the later work was done by static load testing, which is considered by most engineers to be the most reliable way to determine the bearing capacity and load/settlement behavior of a foundation element or groups of elements. It is, however, also very time consuming, cumbersome, and expensive to perform, especially on large elements or groups. Safety is also an important issue because of the sheer massiveness of the required loading systems. As previously discussed, dynamic pile driving methods have become a common alternative to predict the bearing capacity of piles, but these methods are empirical, their accuracy varies considerably, and the trend toward much larger and deeper elements reduces their applicability. Also, they do not provide direct measurements, they induce high accelerations, and the load/settlement behavior is controlled by the action of stress waves.

In an effort to overcome the practical difficulties of both conventional static and dynamic load testing, FHWA initiated a series of research studies to evaluate several promising load test methods that were developed by U.S. and European inventors. None of these methods came with sufficient documentation for standardized test procedures or for the interpretation of data produced by these tests. The following sections summarize the efforts to evaluate and refine several of these innovative methods.

2.10.1 Statnamic Load Testing

The term "Statnamic" is a combination of the two words — "static and dynamic" — because it was neither one nor the other, but somewhere in between. Upon closer study it is seen to be closer to being quasistatic than dynamic, especially now in its latest version where there is no dynamic impact. In Statnamic testing a high-energy, fast-burning solid fuel is ignited within a pressure chamber to act as an upward propellant force. As the fuel pressure increases, an upward force is exerted on a set of reaction weights while an equal and opposite force pushes downward on the foundation element. Loading increases until it reaches a maximum, and then it is vented to control the unloading cycle. Figure 23 schematically illustrates the statnamic concept.

In the original version the weights would come toppling down on the pile or shaft after the fuel had completely burned off, and possibly damaged the foundation element being tested. This problem was solved by placing sand in an outer, concentric container that allowed the sand to flow downward (while the weights were pushed up and off the foundation) and cushion the blow from the falling weights (dynamic impact). The extra time to set up the sand container, and cleanup for retesting created a different kind of problem that was later solved by the development of a catch mechanism and the pressure venting system. A calibrated load cell located between the piston and the pile top is used to measure the applied load and a photovoltaic laser sensor records displacement during

 

the test. The assembly and take down is very straightforward. All components are easily handled with a small hoisting rig. This system is capable of producing a given force or load using only 5 percent of the mass required in an equivalent static test. The original development in 1988 began with a 0.1 MN test device; however, current test devices are capable of producing loads up to 30 MN. Furthermore, because the direction of force is along the cylinder assembly, loading is perfectly axial or horizontal if the apparatus is shifted 90 degrees to perform a lateral load test. A batter pile test configuration can also be adopted if desired. The relatively slow application and release of compressive forces eliminates negative tensile forces, compressing the pile and soil as a single unit.

Throughout loading, load and displacement signals are digitized and sent to a raw voltage data file. After the event, the raw signal voltages are converted to load and displacement values using factory calibration values. Load-displacement graphs are presented immediately to on-site engineers for quick evaluations. Supplementary graphs of velocity and acceleration versus time can also be generated with simple post-processing commands. All data are stored for future analysis and reference.

During the 1990's FHWA has performed or funded many Statnamic tests and correlation studies with conventional static load tests to develop standardized testing procedures and improved data interpretation methods. Many SHA's and other domestic and foreign agencies are now conducting their own evaluation programs to further expand the data base of case histories and performance studies. Measurements made with the Statnamic device owned by FHWA compared very well with static load tests performed by TFHRC researchers on spread footings and pile groups in the research test pits (see appendix A). Measurements on full-scale piles and drilled shafts on other FHWA sponsored research projects have also compared well with static load test results. Figure 24 shows a test at FHWA's foundation load test facility.

FIGURE 24.  Vertical statnamic load test on model pile group at TFHRC.

In late 1996, FHWA participated in a research program with the Utah DOT on a major rebuilding project on I-15 in Salt Lake City. The research involved the lateral behavior of a nine-pile group that was tested in both a free and fixed head condition. A 14 MN device was used on the fully instrumented group. FHWA also funded a similar test program at the Auburn University NGES during the same time period (see chapter 5).

2.10.2 Osterberg Cell Load Testing

The Osterberg Cell is another innovative technique that has been evaluated by FHWA during the 1990's to determine its applicability and potential for reducing time and costs of performing foundation load testing, especially for drilled shafts. The drilled shaft designer must face both the problems of predicting subsurface soil and rock strength and compressibility characteristics, and the difficulty of estimating the impact of construction installation technique on the completed shaft. Model testing and laboratory investigations don't provide sufficient insight to assess the complex interaction between soil (especially intermediate geomaterials) and the concrete shaft. The Osterberg Cell, commonly called the "O-cell", is well suited for this problem and has provided an attractive short duration alternative for testing drilled shafts (31).

The O-cell is a hydraulically driven, high-capacity sacrificial jack-like device that is installed at the bottom of the reinforcement cage (figure 25) of a drilled shaft or at the tip of a driven pile. Unlike the conventional static and dynamic load tests that apply a

compressive force at the top of the pile or shaft, the O-cell loads the unit in compression, but from its bottom end. It requires no overhead reaction frame, dead load, or other external system. As the O-cell is pressurized and expands, the shaft and the soil provide reaction to the applied loads. The end bearing support provides reaction for the side friction along the shaft, and vice versa, until reaching the capacity of the cell or either of the two support components. O-cell tests automatically separate the end bearing support component from side friction values. Thus, an O-cell test load placed at the bottom of the shaft has twice the testing effectiveness of that same load placed at the top of the shaft. Production shafts can also be tested, if the device is grouted after testing is completed.

Movements of the foundation element during an O-cell test are measured by electronic gauges connected to a computerized data acquisition system. Linear vibrating wire displacement transducers (LVWDT's) are attached to the bottom plate of the O-cell. Telltales are used to measure both the compression of the test pile and the upward movement of the top of the O-cell. The downward movement of the bottom plate is obtained by subtracting the upward movement of the top of the O-cell from the total extension of the O-cell as determined by the LVWDT's. The upward movement of the top of the test pile or shaft is measured by digital gauges mounted on a reference beam. The loading mechanism has evolved from the original bellows-type expansion cell to the current piston-type jack. However, the piston extends downward instead of upward like that of a conventional load test.

The O-cell test offers a number of obvious advantages over conventional load testing, including economy, high load capacity, safety, reduced work area, and the ability to separate the end-bearing and side-friction components. Disadvantages include the need for advance installation, sacrificial expendability, unsuitability for certain types of piles, and the balanced component limitation, i.e., the test capacity is limited to twice the capacity of the support component reaching ultimate first. It has been noted that the majority of load tests on drilled shafts are now being done with the O-cell.

Boston Engineers were the first to use the O-cell in a practical application in 1987 on a bridge site near Boston, Massachusetts and later that same year at Rochester, New York. After testing, they recovered the cell and used it on another test site. In 1988, two more tests were performed with FHWA research funds at a bridge in Port Orange, Florida. More than 200 additional tests have been performed with O-cells in the United States on piles and drilled shafts during the 1990's, including some lateral load tests.

The Minnesota DOT recently completed a major load-testing research program featuring the use of O-cell devices in both axial and lateral load-testing modes. The shafts were 58 m deep and 1.3 m in diameter. One of the test shafts was instrumented with Sister Bar strain gauges to develop a better understanding of the load transfer distribution. The first-ever embedded O-cell lateral load test was on a second test shaft on this same project.

Other interesting applications in FHWA-sponsored projects include two previous record- setting performances in Massachusetts and Kentucky on highway bridge projects. The Owensboro, Kentucky record of 53.4 MN (6000 tons) was soon broken by the Boston Central Artery project record of 55.9 MN (6,280 tons), which was easily broken by the Florida project. The site conditions on Boston's Southeast Expressway (I-93) were treacherous because the test shaft was drilled between the two fast lanes inside two jersey barriers only 2 m apart. The testing was completed without obstructing or disrupting traffic flow.

No job appears to be too big or too small for the use of an O-cell. Several world records for load testing have been set recently, including the current world record for total load of 135 MN (more than 15,000 tons) set in Florida in 1997. Whether inside a building or under a bridge with limited access or low headroom, or within a cofferdam in the middle of a river, O-cell testing can perform well. It can be used to isolate portions of a pile or shaft by installing multiple cells, and is fast becoming a favorite of engineers and contractors because of its speed of installation and cost advantages.

2.10.3 Dynamic Drop Weight Load Testing

The use of a heavy weight (90 KN) falling from various heights (0.3 to 5 m) is another method to predict the load capacity of a pile or drilled shaft (figure 26). The foundation element is instrumented with strain gauges and accelerometers to measure the force and impact velocity of the stress wave generated by striking the element. The measured data are correlated to driving resistance to predict load capacity. An electric theodolite is used to measure the settlement associated with each hammer blow. The friction in the hammer leads

and the cushion assembly cause the actual energy delivered to the pile or shaft to be much less than the free-fall energy of the ram. This is accounted for by the instrumentation gauges similar to the Pile Driving Analyzer equipment used during pile driving operations to monitor resistance to driving. A service crane is required on site to perform the tests.

Three different methods (one U.S. and two European ) were evaluated on several FHWA research test sites to determine the applicability of this type of testing to check pile or shaft capacity. The GRL method was developed at Case Western Reserve University in Cleveland, Ohio, by the same people who developed the PDA, WEAP and CAPWAP. The "SIMBAT" method was developed by the French government's Center for Experimental Research on Buildings and Public Works (CEBTP). The TNO method was developed by the Dutch government at their national research center in the Netherlands. The Europeans have amassed a large data base of static/dynamic test result correlations. As a result, these methods have become established practice in several European countries, but have been slow to catch on in the United States because of the greater success of the O-cell and statnamic methods. Also, test results on FHWA projects have not measured up to expectations thus far, usually over-predicting the measured capacities. Further investigation is planned to complete the evaluations.