STATISTICALLY ORIENTED END-RESULT SPECIFICATIONS

38

NATIONAL COOPERATIVE HIGHWAY RESEARCH PROGRAM SYNTHESIS OF HIGHWAY PRACTICE

STATISTICALLY ORIENTED END-RESULT SPECIFICATIONS

0

TRANSPORTATION RESEARCH BOARD NATIONAL RESEARCH COUNCIL

0

0

TRANSPORTATION RESEARCH BOARD 1976

Officers HAROLD L. MICHAEL, Chairman ROBERT N. HUNTER, Vice Chairman W. N. CAREY, JR., Executive Director

Executive Committee

HENRIK E. STAFSETH, Executive Director, American Assn. of State Highway and Transportation Officials (ex officio) NORBERT T. TIEMANN, Federal Highway Administrator, U.S. Department of Transportation (ex officio) ROBERT E. PATRICELLI, Urban Mass Transportation Administrator, U.S. Department of Transportation (ex officio) ASAPH H. HALL, Federal Railroad Administrator, U.S. Department of Transportation (ex officio) HARVEY BROOKS, Chairman, Commission on Sociotechnical Systems, National Research Council (ex officio) MILTON PIKARSKY, Chairman of the Board, Chicago Regional Transportation Authority (ex officio, Past Chairman 1975) WARREN E. ALBERTS, Vice President (Systems Operations Services), United Airlines GEORGE H. ANDREWS, Vice President (Transportation Marketing) Srerdrnp and Parcel GRANT BASTIAN, State Highway Engineer, Nevada Department of Highways KURT W. BAUER, Executive Director, Southeastern Wisconsin Regional Planning Commission LANGHORNE BOND, Secretary, Illinois Department of Transportation MANUEL CARBALLO, Secretary of Health and Social Services, State of Wisconsin L. S. CRANE, President, Southern Railway System JAMES M. DAVEY, Consultant B. L. DEBERRY, Engineer-Director, Texas State Department of Highways and Public Transportation LOUIS J. GAMBACCINI, Vice President and General Manager, Port Authority Trans-Hudson Corporation HOWARD L. GAUTHIER, Professor of Geography, Ohio State University FRANK C. HERRINGER, General Manager, San Francisco Bay Area Rapid Transit District ANN R. HULL, Delegate, Maryland General Assembly ROBERT N. HUNTER, Chief Engineer, Missouri State Highway Commission PETER G. KOLTNOW, President, Highway Users Federation for Safety and Mobility A. SCHEFFER LANG, Assistant to the President, Association of American Railroads

BENJAMIN LAX, Director, Francis Bitter National Magnet Laboratory, Massachusetts Institute of Technology DANIEL McFADDEN, Professor of Economics, University of California

HAROLD L. MICHAEL, School of Civil Engineering, Purdue University THOMAS D. MORELAND, Commissioner, Georgia Department of Transportation J. PHILLIP RICHLEY, Vice President (Engineering and Construction), The Cafaro Company RAYMOND T. SCHULER, Commissioner, New York State Department of Transportation WILLIAM K. SMITH, Vice President (Transportation), General Mills

NATIONAL COOPERATIVE HIGHWAY RESEARCH PROGRAM

Transportation Research Board Executive Coinmittec

Subcommittee for the NCHRP

HAROLD L. MICHAEL, Purdue University (Chairman) ROBERT N. HUNTER, Missouri State Highway Commission HENRIK E. STAFSETH, Amer. Assn. of State Hwy. and Transp. Officials NORBERT T. TIEMANN, U.S. Department of Transportation HARVEY BROOKS, National Research Council

W. N. CAREY, JR., Transportation Research Board

Project Committee SP 20-5

RAY R. BIEGE, JR., Kansas Dept. of Transportation (Chairman) VERDI ADAM, Louisiana Department of Highways JACK FREIDENRICH, New Jersey Department of Transportation DAVID GEDNEY, Federal Highway Administration EDWARD J. HEINEN, Minnesota Department of Highways BRYANT MATHER, USAE Waterways Experiment Station THOMAS H. MAY, Pennsylvania Department of Transportation THEODORE F. MORF, Consultant EDWARD A. MUELLER, Jacksonville Transportation Authority

ORRIN RILEY, Howard, Needles, Tammen & Bergendofi REX C. LEATHERS, Federal Highway Administration

ROY C. EDGERTON, Transportation Research Board

Program Staff

Topic Panel on Statistically Oriented End-Result Specifications

ADAM, Louisiana Department of Highways D. Y. BOLLING, Federal Highway Administration C. S. HUGHES, III, Virginia Highway and Transportation Research

Council G. W. STEELE, West Virginia Department of Highways

C. R. SUNDQUIST, California Department of Transportation J. W. GUINNEE, Transportation Research Board

G. GUNDERMAN, Transportation Research Board

Consultants to Topic Panel

F. J. BOWERY, JR., and S. B. HUDSON, Woodward-Clyde

Consultants

K. W. HENDERSON, JR., Program Director

DAVID K. WITHEFORD, Assistant Program Director HARRY A. SMITH, Projects Engineer LOUIS M. M*cGREGOR, Administrative Engineer ROBERT E. SPICHER, Projects Engineer

JOHN E. BURKE, Projects Engineer (Retired) HERBERT P. ORLAND, Editor

R. IAN KINGHAM, Projects Engineer PATRICIA A. PETERS, Associate Editor ROBERT J. REILLY, Projects Engineer EDYTHE T. CRUMP, Assistant Editor

NATIONAL COOPERATIVE HIGHWAY RESEARCH PROGRAM 38 SYNTHESIS OF HIGHWAY PRACTICE


RESEARCH SPONSORED BY THE AMERICAN ASSOCIATION OF STATE HIGHWAY AND TRANSPORTATION OFFICIALS IN COOPERATION WITH THE FEDERAL HIGHWAY ADMINISTRATION

AREA OF INTEREST: CONSTRUCTION

TRANSPORTATION RESEARCH BOARD NATIONAL RESEARCH COUNCIL

WASHINGTON D.C. 1976


Systematic, well-designed research provides the most ef-fective approach to the solution of many problems facing highway administrators and engineers. Often, highway problems are of local interest and can best be studied by highway departments individually or in cooperation with their state universities and others. However, the accelerat-ing growth of highway transportation develops increasingly complex problems of wide interest to highway authorities. These problems are best studied through a coordinated program of cooperative research. In recognition of these needs, the highway administrators of the American Association of State Highway and Trans-portation Officials initiated in 1962 an objective national highway research program employing modern scientific techniques. This program is supported on a continuing basis by funds from participating member states of the Association and it receives the full cooperation and sup-port of the Federal Highway Administration, United States Department of Transportation.

The Transportation Research Board of the National Re-search Council was requested by the Association to admin-ister the research program because of the Board's recog-nized objectivity and understanding of modern research practices. The Board is uniquely suited for this purpose as: it maintains an extensive committee structure from which authorities on any highway transportation subject may be drawn; it possesses avenues of communications and cooperation with federal, state, and local governmental agencies, universities, and industry; its relationship to its parent organization, the National Academy of Sciences, a private, nonprofit institution, is an insurance of objectivity; it maintains a full-time research correlation staff of special-ists in highway transportation matters to bring the findings of research directly to those who are in a position to use them.

The program is developed on the basis of research needs identified by chief administrators of the highway and trans-portation departments and by committees of AASHTO. Each year, specific areas of research needs to be included in the program are proposed to the Academy and the Board by the American Association of State Highway and Trans-portation Officials. Research projects to fulfill these needs are defined by the Board, and qualified research agencies are selected from those that have submitted proposals. Ad-ministration and surveillance of research contracts are responsibilities of the Academy and its Transportation Research Board.

The needs for highway research are many, and the National Cooperative Highway Research Program can make signifi-cant contributions to the solution of highway transportation problems of mutual concern to many responsible groups. The program, however, is intended to complement rather than to substitute for or duplicate other highway research programs.

NCHRP Synthesis 38

Project 20-5 FY '74 (Topic 6-02) ISBN 0-309-0254-1 L. C. Catalog Card No. 76-51496

Price: $4.00

Notice

The project that is the subject of this report was a part of the National Cooperative Highway Research Program conducted by the Transportation Research Board with the approval of the Governing Board of the National Research Council, acting in behalf of the National Academy of Sciences. Such approval reflects the Governing Board's judgment that the program concerned is of national impor-tance and appropriate with respect to both the purposes and re-sources of the National Research Council. The members of the technical committee selected to monitor this project and to review this report were chosen for recognized scholarly competence and with due consideration for the balance of disciplines appropriate to the project. The opinions and con-clusions expressed or implied are those of the research agency that performed the research, and, while they have been accepted as appropriate by the technical committee, they are not necessarily those of the Transportation Research Board, the National Research Coun-cil, the National Academy of Sciences, or the program sponsors. Each report is reviewed and processed according to procedures established and monitored by the Report Review Committee of the National Academy of Sciences. Distribution of the report is ap-proved by the President of the Academy upon satisfactory comple-tion of the review process. The National Research Council is the principal operating agency of the National Academy of Sciences and the National Academy of Engineering, serving government and other organizations. The Transportation Research Board evolved from the 54-year-old High-way Research Board. The TRB incorporates all former HRB activities but also performs additional functions under a broader scope involving all modes of transportation and the interactions of

transportation with society.

Published reports of the


are available from:

Transportation Research Board National Academy of Sciences 2101 Constitution Avenue, N.W. Washington, D.C. 20418

(See last pages for list of published titles and prices)

Printed in the United States of America.

PREFACE There exists a vast storehouse of information relating to nearly every subject of concern to highway administrators and engineers. Much of it resulted from research and much from successful application of the engineering ideas of men faced with problems in their day-to-day work. Because there has been a lack of systematic means for bringing such useful information together and making it available to the entire highway fraternity, the American Association of State Highway and Trans-portation Officials has, through the mechanism of the National Cooperative Highway Research Program, authorized the Transportation Research Board to undertake a continuing project to search out and synthesize the useful knowledge from all pos-sible sources and to prepare documented reports on current practices in the subject areas of concern.

This synthesis series attempts to report on the various practices without in fact making specific recommendations as would be found in handbooks or design manuals. Nonetheless, these documents can serve similar purposes, for each is a compendium of the best knowledge available concerning, those measures found to , be the most successful in resolving specific problems. The extent to which they are utilized in this fashion will quite logically be tempered by the breadth of the user's knowledge in the particular problem area.

FOREVVORD This synthesis will be of special interest and usefulness to construction engineers and specification writers seeki.ng technical information on the use of statistically

L)Y Jtafl oriented acceptance plans for end-result specifications. Detailed information is Transportation presented on the contractor, s quality control systems, the contractmg agency s

Research Board acceptance procedures, and other factors to be considered in applying statistically oriented specifications. Current practice and the experiences of agencies using end-result specifications are outlined.

Administrators, engineers and researchers are faced continually with many highway problems on which much information already exists either in documented form or in terms of undocumented experience and practice. Unfortunately, this information often is fragmented, scattered, and unevaluated. As a consequence, full information on what has been learned about a problem frequently is not assembled in seeking a solution. Costly research findings may go unused, valuable experience may be overlooked, and due consideration may not be given to recom-mended practices for solving or alleviating the problem. In an effort to resolve this situation, a continuing NCHRP project, carried out by the Transportation Research Board as the research agency, has the objective of synthesizing and reporting on common highway problems—a synthesis being identified as a composition or com-bination of separate parts or elements so as to form a whole greater than the sum

of the separate parts. Reports from this endeavor constitute an NCHRP report series that collects and assembles the various forms of information into single concise documents pertaining to specific highway problems or sets of closely related problems.

in' recent years, statistically oriented end-result specifications (ERS) have attracted considerable attention in many transportation construction agencies. These specifications have been used for structures, pavements, embankments, and base courses. In addition, numerous components and individual materials are produced and accepted using these concepts. However, ERS are sometimes not used when they might be because technical information has not been conveniently available.

This report of the Transportation Research Board describes current technology for the application of ERS based on information from all sources, including some outside of the transportation field.

To develop this synthesis in a comprehensive manner and to ensure inclusion of significant knowledge, the Board analyzed available information assembled from numerous sources, including a large number of state highway and transporta-tion departments. A topic panel of experts in the subject area was established to guide the researchers in organizing and evaluating the collected data, and to review the final synthesis report.

This synthesis is an immediately useful document that records practices that were acceptable within the limitations of the knowledge available at the time of its preparation. As the processes of advancement continue, new knowledge can be expected to be added to that which is now at hand.

CONTENTS

SUMMARY

PART I

3 CHAPTER ONE Introduction

4 CHAPTER TWO Trends in Specifications for Materials and Con- struction

End-Result Specifications (ERS) Statistically Oriented Acceptance Plans for End-Result

Specifications

7 CHAPTER THREE Problems Associated with Statistically Ori- ented Specifications

Defining Good and Poor Material or Construction Tech- niques

Acceptable Buyer's and Seller's Risks Defining Lot Size and Testing Frequency Determining Equitable Reduction in Price Administrative Problems Human Factors Legal Factors Economic Factors and Cost Effectiveness

10 CHAPTER FOUR Contractor's and Producer's Quality Control Systems

General Requirements for Quality Control Systems Requirements for Certification

14 CHAPTER FIVE Contracting Agency Acceptance Procedures

Types of Acceptance Plans Use of Range in Acceptance Plans Random Sampling Rapid Test Methods Personnel

18 CHAPTER SIX Current Practices in Highway Agencies

U.S. Highway and Transportation Agencies Federal Highway Administration Port Authority of New York Foreign Transportation Agencies

32 CHAPTER SEVEN Comments from Trade Associations and Producers

34 GLOSSARY

36 REFERENCES

PART II

38 APPENDIX Features of End-Result Specifications Considered or Implemented by Various Agencies

ACKNOWLEDGMENTS

Special appreciation is expressed to Frank J. Bowery, Jr., and S. B. Hudson, of Woodward-Clyde Consultants of Rockville, Md., who were responsible for the collection of data and the preparation of the report.

Valuable assistance in the preparation of this synthesis was provided by the Topic Panel, consisting of Verdi Adam, As-sistant to Chief Engineer, Louisiana Department of Highways; Doyt Y. Bolling, Quality Assurance Specialist, Construction Operations Specifications Branch, Office of Highway Opera-tions, Federal Highway Administration; C. S. Hughes, III, Senior Highway Research Scientist, Virginia Highway and Transportation Research Council; Garland W. Steele, Director,

Materials Control, Soil and Testing Division, West Virginia Department of Highways; and Carl R. Sundquist, Senior Ma-terials and Research Engineer, California Department of Transportation.

John W. Guinnee, Engineer of Soils, Geology, and Founda-tions, and W. G. Gunderman, Engineer of Materials and Con-struction, both of the Transportation Research Board, assisted the Special Projects staff and the Topic Panel.

Information on current practice was provided by many high-way agencies. Their cooperation and assistance was most helpful.


SUMMARY The beginning of the trend toward statistical specifications for highway construction was probably the AASHO Road Test. During construction of this project, a suf-ficient number of unbiased test results were available to show the true variability of the results and their relationship to specifications. Since then a number of agen-cies have conducted studies on statistical concepts in highway specifications and many have used or experimented with statistically oriented end-result specifications.

End-result specifications (ERS) are used instead of "recipe" or method speci-fications. Under ERS, an agency monitors the contractor's control of the process that produces a material or item of construction (rather than directly controlling or inspecting the process itself) and accepts or rejects the end product. Thus, the end-result specification places the entire responsibility for quality control on the contractor. To be sure that the contractor's quality control is effective, the agency spot-checks the contractor's activities and makes sure that sampling is random, that tests are performed correctly, that all test results are recorded, and that documenta-tion is up to date. In addition, the agency must have a statistically oriented program of acceptance testing. Many specifications have a provision for reduction in con-tract unit prices when acceptance testing indicates that the end product does not fully comply with quality requirements.

Advantages of ERS to highway agencies include the proper allocation of responsibility for quality between the contractor and the agency, more complete records, statistically based acceptance decisions, and savings in engineering costs. Advantages to contractors include greater choice of materials and equipment and design of the most economical mixtures to meet specifications. A great benefit is lot-by-lot acceptance in most ERS so that the contractor can tell where he stands at all times.

Disadvantages to agencies include a general resistance to change by contrac-tors, the need to have more highly qualified agency personnel for spot-checking, a possible increase in administrative work initially, and the possibility that reduced price clauses are not sufficient to preclude submittal of inferior products.

The use of statistics is essential for effective quality assurance. Although most agencies have not adopted all of the elements of an ERS, many have incorporated some form of statistically oriented procedure.

There are two risks involved in any acceptance decision: rejection of accept-able material (seller's risk) and acceptance of rejectable material (buyer's risk). These risks can be reduced by increasing the number of measurements on which the acceptance decision is made.

A complete statistically oriented end-result specification should include lot size, point of sampling, method of random sampling, number of samples to be taken on each lot, method of test, target value of measured characteristic, realistic tolerances, and the action to be taken if quality requirements are not met.

Certain basic decisions must be made when drafting statistically oriented speci-

fications. Good- and poor-quality construction material or techniques should be defined through the establishment of realistic average measurements. In addition, the appropriate sizes of the seller's and buyer's risks should be set. Other decisions include establishing lot size and testing frequency, determining an equitable price reduction for construction material or techniques not meeting quality requirements, and retraining of personnel.

Agencies using ERS usually have some requirement for quality control by the contractor or producer. In addition, there may be a requirement that the plant be inspected by the agency and certified as acceptable. A minimum testing frequency may also be specified.

There are two general types of statistical acceptance plans: (a) the attributes sampling plan, which is most useful when the characteristic of interest (attribute) can not or need not be measured but can be accepted or rejected by visual inspec-tion; and (b) the variables acceptance plan, which uses both the average values and the variability of measurements (standard deviation or range) to determine accep-tance. If .the standard deviation is not convenient for field use because of the com-putations involved, it can be avoided by use of the range, or difference between the smallest and largest measurement in a group.

In order for ERS to work, all samples must be unbiased (that is, not influenced by opinion or judgment). To ensure unbiased sampling, random samples should be taken with the aid of a published random number table or other random method. Stratified random sampling (random sampling of sublots) may be used to avoid cluster samples.

Based on available information, 33 state highway agencies are using, planning to use, or have tried some form of statistically oriented end-result specifications. The status of these state specifications as well as the use of ERS in several foreign countries is discussed in Chapter Six. The state data are tabulated in an Appendix.

CHAPTER ONE

INTRODUCTION

The beginning of the trend toward statistically oriented END-RESULT SPECIFICATIONS (ERS) * for highway construction was probably the AASHO Road Test (1956-58). It was during the construction of this project that a suf-ficient number of unbiased test results of construction materials and techniques became available to expose the true variability of these results and their relationship to specifi-cations. Of course, there were precursors in this area. In the early 1940s Stanton Walker studied the variability of test results of the compressive strength of concrete and pub-lished guidelines for the overdesign required to meet specifications (1). In 1954 Odasz and Nafus published their findings on the variability of the physical test properties of asphaltic concrete (2).

In addition, since World War lithe military and most large manufacturing organizations have been using statisti-cal acceptance plans (3, 4). Other instances could be cited, but the point is that the significance of the relationship of the variability of measurements to highway specifications was not fully realized until after the startling experiences with construction control at the AASHO Road Tests. As a result of these experiences, Carey and Shook (5) con-cluded that "we really do not know the significance of many items included in our specifications, the true stan-dard or quality they are supposed to guarantee, nor the significance of the specification limits attached to them.

Sampling plans now being used are not adequate for estimating the true characteristics of materials or construc-tion items for which the specifications are written, and cer-tainly cannot guarantee 100 percent compliance to the specification limits."

The finding that the sampling procedures in use at that time (judgment sampling) did not result in 100 percent compliance with specification limits was emphasized by the Blatnik Committee's discovery in 1962 of many cases of nonconformance in accepted highway construction. The U.S. Congress threatened to pass laws making it a federal offense to "knowingly incorporate" nonspecification ma-terial in a highway project.

In 1963 the Bureau of Public Roads obtained under contract a report entitled "A Plan for Expediting the Use of Statistical Concepts in Highway Acceptance Specifica-tions" (6). Based on this report the bureau circulated to state highway agenices in 1965 a publication entitled "The Statistical Approach to Quality Control in Highway Con-struction" (7). This booklet contains explicit instructions for measuring current quality in a statistically valid man-ner and for determining the proportions of the total varia-bility of the quality measurements due to actual variability in the materials or variability caused by sampling or test-ing. Programs for most types of highway materials and

In this synthesis, important terms appear initially in small caps. These terms are defined in the Glossary.

construction are included as well as a computer program for determining the statistical parameters from the test values. Also in 1965, the National Cooperative Highway Research Program (NCHRP) issued NCHRP Report 17, "Development of Guidelines for Practical and Realistic Construction Specifications" (8).

In 1966, the bureau informally distributed to state high-way agencies copies of a "Futurized Revision of FP 61: Standard Specifications for Construction of Roads and Bridges on Federal Highway Projects" (9). This was not intended for immediate use but to show by example the changes that might be required by the adoption of statisti-cal concepts and by conversion to the end-result type of specification in which the contractor is responsible for the quality of the work and for conformance to specifications up to the point where the product, item of construction, or completed construction is submitted for acceptance. This futurized revision contains four statistical acceptance plans, instructions for random sampling, and a table of random numbers revised for practical use. [This table has been republished by the Asphalt Institute (10).]

As a result of bureau sponsorship, a number of states initiated research on the variability of measurements of the characteristics of materials and construction. In 1969 the Bureau of Public Roads published "Quality Assurance in Highway Construction," which summarized the results of this research (11).

In 1964 Mississippi incorporated what are believed to be the first statistically oriented specifications in their special provisions. To date, 37 highway or transportation agencies (U.S. or foreign) are known to have drafted some form of statistically oriented acceptance plans and/or end-result specifications.

In 1971 the Transportation Research Board (TRB) is-sued HRB Special Report 118 entitled "Quality Assurance and Acceptance Procedures" (12), which summarizes the state of the art of quality assurance and highway depart-ment needs in the area of quality assurance procedures. Part 1 of this report covers developments to date in the areas of quality assurance for aggregates, bituminous con-struction, concrete construction, construction practices, and general materials. Part 2 discusses highway department needs.

In Part 1, Bloem reported that "statistical procedures to monitor and regulate the quality of aggregates for highway work have been slow in gaining application." However, he stated that many aggregate producers were undertaking sta-tistical quality monitoring at their plants to ensure that the aggregate as furnished to the customer meets grading requirements. Hughes reported in HRB Special Report 118 that the change to statistically oriented specifications had been and would be slow because the highway industry is reluctant to abandon the traditional methods and specifica-tions that have been used for many years. He reported that

4

in 1970 approximately 25 highway agencies were in some stage of establishing the realistic variability of test results. Ten states had completed a version of statistically defensi-ble specifications, and four states were accepting material under such specifications. Smith (12) discussed the con-ventional specifications for portland cement concrete and the major advance that was made in Illinois by introduc-ing a statistical basis for the analysis, evaluation, and speci-fication of concrete strength. He listed nine points that have led to the questioning of the different approaches to control of concrete quality and discussed a number of developments that would lead to better procedures for quality in concrete. Verdi Adam (12) pointed out the need for development of suitable acceptance plans for some manufactured products and recommended use of the re-duced sampling plan of ASTM and AASHTO by the states.

In Part 2 of HRB Special Report 118, quality assurance is defined as a three-step process. ("1. What do we want? 2. How do we order it? 3. How do we deterimne that we

got what we want?") The classification of DEFECTS with respect to CRITICALiTy is discussed as well as the necessity of standardization of laboratories. The summary states that specifications must be realistic, economically controllable, and must define clearly the responsibility of the contractor and the owner in the area of QUALITY ASSURANCE and acceptance.

The purpose of this synthesis is to extend and amplify the concepts and findings of HRB Special Report 118 with respect to specifications for highway materials and con-struction and to show how they have been applied in those instances where current information is available. The syn-thesis should serve to enlighten new management and agencies not using statistically oriented end-result specifica-tions as to the advantages and disadvantages of statistically defensible acceptance plans. It should also provide con-tractors with a means of better understanding their respon-sibilities in an end-result-specification program and the advantages of such a program to them.

CHAPTER TWO

TRENDS IN SPECIFICATIONS FOR MATERIALS AND CONSTRUCTION

END-RESULT SPECIFICATIONS (ERS)

General

Instead of the recipe-type of specification that was formerly in use, some agencies have changed or are considering changing to the end-result specification used by the mili-tary and space agencies and by most large industries. Es-sentially, this means that instead of inspecting the process that produces a certain material or item of construction the agency monitors the contractor's control of the process and accepts or rejects the end product.

For example, in the case of portland cement concrete, the agency would no longer design the concrete mixture, specify the amount of cement per cubic yard, the water-cement ratio, the proportion of each size aggregate of specified gradation, the mixing equipment, the mixing time, and the minimum 28-day compressive strength that all test speci-mens are expected to exceed. During production, the agency would no longer test the gradation of stockpiles of aggregate, determine moisture contents, compute bin weights, check scales, monitor batch weighing and amount of water added, or write inspection tickets for each load of concrete. Instead, the agency would specify the quality of the materials to be used, the degree of quality control to be exercised by the contractor, the percentage of test re-sults required to exceed a specified limit, and the sampling and testing procedure on which acceptance would be based.

Responsibilities of Contractor or Producer

An end-result specification places the entire responsibility for QUALITY CONTROL on the contractor. Quality control includes all activities and considerations during the manu-facture of the product that are necessary to ensure that the product has the desired quality characteristics, both levels and tolerances. The specified degree of quality control may range from a simple statement that the contractor shall provide quality control to a detailed listing of minimum requirements for equipment, personnel, frequency of sam-pling, and required documentation. In any case, the as-sumption of quality control by the contractor or producer requires technically qualified personnel to design mixtures, make all required computations, oversee all elements of the production process, and make all necessary tests to ensure that the process is in control and that the end product will meet all specified requirements.

Agency Responsibility

The agency has entire responsibility for specifying quality assurance, those activities that ensure that the contractor's or producer's over-all quality control is effective. This re-sponsibility requires that the agency periodically spot-check the quality control activities of the contractor to see that sampling is random, that tests are performed correctly, that test results are reported, and that DOCUMENTATION is up to

5

date. Usually, this monitoring is supplemented by a pro-gram of acceptance testing. These tests must be expertly performed and be entirely UNBIASED. Sampling times or locations should not coincide with those of the contractor's or producer's quality control personnel and should be com-pletely randomized. In case of differences in results, the engineer's decision must govern, but it must be on firm ground to avoid claims. It is possible that after the re-liability of the contractor's or producer's QUALITY CONTROL

SYSTEM has been established, the amount of testing by the agency can be greatly reduced with consequent savings in cost and technical personnel.

Incentive Provisions

Under other specification systems, when acceptance sam-pling and testing indicate that the end product does not fully comply with quality requirements, the engineer must decide whether or not to reject the product or accept it as satis-factory on the basis of engineering judgment. Either de-cision could lead to difficulties. If there is provision in the specification for accepting the product at some stated per-centage of the contract price for the level of relative quality indicated by the acceptance plan, however, the engineer is relieved of the difficult decision. This part of the end-result specification is usually called the equitable reduction in price or incentive provision. The objective is not to cause the contractor undue hardship but to alert him to the neces-sity of increasing quality level to ensure against a loss of revenue and to save material that might otherwise have to be rejected and wasted.

Many engineers believe that a true incentive provision should also contain a bonus or an increase in contract price for construction material or techniques of a quality level exceeding specified requirements. This is not presently per-missible on federally financed work, however. Current in-centive provisions generally are in the form of reductions in price for construction material or techniques not in full compliance with specification requirements. West Virginia and Illinois have provided a bonus incentive on experi-mental projects in which federal funds were not involved; West Virginia also has a built-in incentive permitting re-duced cement factors for concrete having strengths sub-stantially exceeding requirements. Pennsylvania is con-templating a similar incentive provision.

Advantages and Disadvantages of

End-Result Specifications

The greatest advantage of ERS to state agencies is the actual placing of responsibility for materials and construc-tion quality on the contractor or producer. Other advan-tages are more complete, as-built records; STATISTICALLY

DEFENSIBLE ACCEPTANCE DECIsIoNS; and savings in engi-neering cost and technical personnel when all features of the ERS are fully implemented.

Advantages to contractors and producers stem from greater latitude in the choice of materials and equipment, and design of the most economical mixtures meeting the specified requirements. Perhaps the greatest benefit is due to the lot-by-lot acceptance procedures that are incorpo-

rated in most ERS. When lots are immediately accepted, conditionally accepted with a reduction in payment, or re-jected, contractors or producers understand their position. An enforced reduction in price is almost certain to attract the attention of management at higher levels. This gives management the opportunity to take corrective action be-fore large quantities of nonspecification material or con-struction are produced and avoids tie-up of capital when payment is held up due to failing tests.

Disadvantages to agencies appear to be initial resistance to end-result specifications by contractors and producers and possibly some increase in administrative work. Agen-cies performing the contractor quality-control activities as well as their own quality-assurance sampling and testing may experience an increase in workload because of the greater number of tests required. Virginia reports no in-crease in workload due to large lot size, however. Some agencies consider the possibility that reduced price clauses may not be sufficient to preclude submittal of inferior products. Spot-checking of contractor's or producer's quality-control systems may require more highly qualified personnel than those solely employed for inspection duties.

Some contractors and producers interviewed in a previous study (26) were of the opinion that the greatest advan-tage of ERS to them would be less risk of a shutdowp. They cited instances in which plants had been shut down or construction stopped as a result of a single failing test. They were in favor of greater latitude in the design of mixtures and anticipated a cost savings because of better use of available materials and assurance of correct yield of portland cement concrete mixtures. They also believed that having access to reliable test information would make better control of their product possible. Large organiza-tions believed that they could economize by having one quality-control team doing the sampling and testing at several plants instead of the agency practice of stationing several inspectors at each plant. Most contractors appre-ciated the advantages of lot-by-lot acceptance, particularly those who had payment held up on an entire project because of a few failing test results.

Other contractors and producers considered that the change to end-result specifications would have great dis-advantages. Small organizations could not afford to main-tain quality control technicians on a full-time basis when the prospect of successfully bidding contracts was un-certain; these organizations would have to arrange with a testing laboratory to do the work. Aggregate producers pointed out that their quality control would not be effective because they would have no control of the product after it left the plant and because the material would be accepted or rejected at point of use.

The Associated General Contractors, in cooperation with an AASHTO committee, recommended that (a) the con-struction of the highway system should continue to be a joint highway department-contractor effort; (b) the con-tractor should monitor work to assure himself of com-pliance with plans and specifications; (c) work need be tested only the number of times necessary to ensure a quality product; (d) in the interest of economy, duplica-tion of testing and inspection should be avoided to the

extent possible; and (e) highway departments continue to perform sampling and testing and retain their responsibility for quality control because most contractors perform only a limited volume of work in any area at one time.

Not all AASHTO members or producers subscribe to these recommendations. Many aggregate, concrete, and hot-mix producers think that a great deal of agency quality-control testing is a duplication of work they already per-form and would prefer to be held responsible for the product as delivered to the contractor. They feel, however, that the agency should maintain a quality assurance pro-gram. In general, the greatest objection to statistically oriented ERS by contractors and producers appears to be the provision for reduction in price, particularly when there is no corresponding bonus.

STATISTICALLY ORIENTED ACCEPTANCE PLANS FOR END-RESULT SPECIFICATIONS

The Role of Statistics in End-Result Specifications

That part of the science of STATISTICS that deals with the averages of measurements and the amount of variation from these averages is an essential element of end-result specifications. STATISTICAL CONTROL CHARTS provide the contractor with the information needed to maintain the re-quired quality uniformly. These charts are also a means of documentation that is easily understood by contractor and agency personnel. The use of statistics is also essential to effective quality assurance.

Deciding whether or not an end product is of acceptable quality is based on the variability of the measurements of the CHARACTERISTIC or characteristics that are chosen as an index of quality. The existence of this variability makes necessary the use of the statistical measure of variation known as the STANDARD DEVIATION to design ACCEPTANCE PLANS with realistic acceptance limits and to estimate the risks associated with statistically oriented acceptance pro-cedures based on these limits. Statistical methods can also be used to estimate the relative quality of the end product. The concept of STATISTICALLY RANDOM SAMPLING is the essential requirement in the use of statistically oriented acceptance procedures.

Statistically Oriented Acceptance Procedures

Most state agencies have not adopted all of the elements of a complete ERS, but many have incorporated some form of statistically oriented procedure in their specifications or special provisions. The accuracy of a single test is limited to the properties of a particular specimen or small amount of material that is tested regardless of the precision of the test method. Usually four or more measurements are made on UNBIASED RANDOM SAMPLES or at RANDOM LOCATIONS. Statistical methods employed to evaluate the measurements include simple averaging as well as plans based on the RANGE and on the NORMAL DISTRIBUTION, STUDENT'S DISTRIBUTION, and NONCENTRAL t DISTRIBUTION (19).

Buyer's and Seller's Risks Associated with Acceptance Procedures

There are always two risks involved in making any ac-ceptance decision (8). If the engineer rejects acceptable material, a TYPE I ERROR is made. The risk of making this error is called the SELLER'S RISK [or in statistical terms, ALPHA (co) RISK]. On the other hand, if the engineer accepts rejectable material, a TYPE II ERROR is made. The risk in-volved is called the BUYER'S RISK [or BETA (fi) RISK]. These risks can never be entirely avoided, but they can be greatly reduced. The size of the risks depends on the variability of the measurements on which the decision is based, on the number of statistically random locations or samples on which measurements are made, on the distance of the tolerance or acceptance limit from what would be con-sidered the average of measurements on nonacceptable material or construction (X'1,), and on the distance of this average from the average of what would be considered good material or construction (X'g).

In any case, both buyer's and seller's risks can be re-duced by increasing the number of measurements. In a given acceptance situation using a noncentral t acceptance plan in which the acceptance decision is based on four measurements and a constant standard deviation, the buyer's risk is 5 percent and the seller's risk is 10 percent. If the number of measurements is increased to 6, the buyer's risk is 2 percent and the seller's risk is 7 percent.

Essential Elements of a Complete, Statistically Oriented Lot Acceptance Plan

I. The specification should define the size of the LOT in terms of appropriate units of measure such as tons, square yards, or linear feet of lane. Lots are sometimes specified as being one day's production. This may lead to a large payment based on the results of a relatively few test results. This can be avoided by a limiting clause such as ". . . but not to exceed .

The point of sampling should be stated. Theoretically, concrete should be sampled at point of deposit in the forms and hot-mix samples should be taken behind the spreader. It is usually more practical to sample concrete at a point near placement in the forms and to sample hot-mix from the loaded trucks at the plant.

The method of random sampling should be stated. If STRATIFIED RANDOM SAMPLING is intended, the number and size of sublots should be given. The method of randomiz-ing times or locations, whether by use of a TABLE OF RANDOM NUMBERS or by other means, should be described.

The number of samples to be taken or the number of measurements to be made on each lot or sublot should be stated.

The method of test (AASHTO, ASTM, or agency) by which the material or construction will be evaluated should be stated.

The target or desired value of the measured charac-

teristic of the material or construction should be stated. In many cases this would be the JOB-MIX-FORMULA (JMF) value.

7. Realistic tolerances should be placed on the target value. In the cases where acceptance is based on average values, the tolerances should be based on at least twice the standard deviation (a-) of the measurements divided by the square root of the number of measurements (n) in the average plus some allowance (d) for target miss.

JMF±[

1

- — =+dJ

(1)

In other types of acceptance plans the elements of the

equation for calculating the acceptance number or quality index should be clearly defined.

8. The action to be taken in case the material or con-struction does not fully comply with the specified quality requirements should be given. This usually takes the form of a table of graduated reductions in price with some cut-off point beyond which the material or construction must be removed and replaced or, if agreed to by the engineer, may be left in place with only a token payment. A referee system may be provided whereby additional sampling and testing are performed. The results are averaged with part or all of the original tests and if the average falls outside process tolerances, the contractor is charged a specified amount for the additional work.

CHAPTER THREE

PROBLEMS ASSOCIATED WITH STATISTICALLY ORIENTED SPECIFICATIONS

DEFINING GOOD AND POOR MATERIAL OR

CONSTRUCTION TECHNIQUES

The initial step in drafting a complete, statistically based acceptance plan is to define good and poor material or construction techniques. In general, "good" means that the average quality is such that there is no question as to pur-chase at the full price. "Poor" means that the average quality is such that it is not acceptable for the intended use. The values assigned to the average of the quality charac-teristics of the two extremes are economically important. If a value unnecessarily restrictive for the intended use is assigned to the average of the measurements of good ma-terial ('g), the cost may be increased and scarce materials may be depleted. If the value assigned to the average measurements of poor material (X') is too great or too small, processing costs may be increased and usable material wasted.

For example, in a specification for a bituminous base course for a secondary road, a value of 2,000 might be assigned to .t for the Marshall stability acceptance plan. In some areas, this would probably mean that scarce crushed aggregate would have to be used rather than a readily available uncrushed gravel aggregate.

Because there are few guidelines based strictly on en-gineering considerations, there is a tendency to perpetuate historical acceptance limits in new specifications. In many instances, unbiased sampling has shown that these limits have not been and can not be consistently enforced (11,

13, 42, 43). Scarce materials can be saved by assigning REALISTIC VALUES to X g based on the unbiased sampling

of materials and of existing good construction.

The average of poor material (X') can sometimes be based on engineering or experimental DATA. When these data are not available, X'1, can be arbitrarily set at a mini-mum distance of approximately 2a- from X'g. If measure-ments are normally distributed, there is only a small probability that averages of groups of four or more mea-surements associated with X g would be confused with averages of groups of measurements associated with when is a minimum distance of 2a- from 'g•

ACCEPTABLE BUYER'S AND SELLER'S RISKS

When designing an acceptance plan, a decision must be made as io what are the appropriate sizes of the buyer's (A) risk (accepting unacceptable material or construction) and seller's (a) risk (rejecting acceptable material or construc-tion). There are no definite rules and few guidelines. The variables sampling plans of Military Specification MIL-STD-414 (4) are designed with a seller's (a) risk of 11 percent for very small lots of three to eight items and a 4 percent risk for lots of up to 220,000 items. The ac-ceptance plan shown in Note 21 of ASTM C 94 has a seller's (a) risk of 2 percent and a buyer's ($) risk of approximately 31 percent when acceptance is based on four test results and a risk of 21 percent when acceptance is based on five test results.

Acceptance plans based on the t and noncentral t distri- butions and having a distance of 2o- between and 'g have a buyer's ($) risk of about 5 percent and a seller's (a) risk of about 3 percent when acceptance is based on five measurements. These appear to be reasonable risks for the acceptance of most highway materials and construction techniques.

Some engineers believe that quality depends more on the degree to which reductions in price are enforced than on the acceptance rules. On this basis, acceptance plans could have a high buyer's () risk, which would require a re-duced number of measurements for the acceptance de-cision. Presumably the seller would voluntarily maintain a high level of quality to avoid loss of revenue.

Another approach to determining acceptable buyer's and seller's risks is to consider the criticality of the characteris-tic of the material or construction for which the acceptance plan is intended. One system of classifying critically, given in 1-!RB Special Report 118 (12), is as follows:

Critical—This defect will make the product danger-ous to use;

Major.—This defect will seriously impair perfor-mance of the item;

Minor.—This defect may impair performance but not seriously; and

Contractual—This defect is likely to have insignifi-cant effect on performance.

A suggested balance of acceptable risks adopted from reference 6 is given in Table 1. The balance of acceptable risks adopted by a particular agency should be determined to be acceptable to that agency.

DEFINING LOT SIZE AND TESTING FREQUENCY

In lot-by-lot acceptance plans, it is necessary to specify the size of a lot. A lot is defined as an amount of material or construction produced by essentially the same process. Theoretically, the number of samples taken to estimate the quality of a lot is independent of the size of the lot. If a cook has both one-gallon and five-gallon (3.8- and 19-litre) pots of soup on a stove and tests the flavor by taking a spoonful from the one-gallon pot, it would not be neces-sary to take five spoonsful from the five-gallon pot for the same purpose. In the case of highway materials or con-struction, large amounts are seldom of the same quality throughout. Unavoidable changes in raw materials, rates of production, or even weather conditions may have large effects on the measured characteristics from day to day or even hour to hour. Many specifications define a lot as one day's production. Some agencies define a lot as the quan-

TABLE I

ACCEPTABLE BUYER'S AND SELLER'S RISKS (from 6)

Classification Buyer's Seller s s of characteristics risk (s), % risk (ct), %

Critical 0.5 5.0

Major 5.0 1.0

Minor 10.0 0.5

Contractual 20.0 0.1

tity of material or construction represented by a stated num-ber of samples or test results. Another procedure is to define a lot as an approximate amount of material or construction.

There are economic factors to be considered in the se-lection of the lot size. If a lot is quite large and is rejected or a reduction in price is applied, there is a greater loss to the seller than if the lot was smaller. After a small lot is rejected or reduced in price, the seller has an opportunity to correct the deficiency before additional amounts of rejectable material or construction are produced.

It seems reasonable that the total number of measure-ments made on a lot should be related to the dollar value of the amount of material or construction. Sampling and testing can be considered to be a form of insurance against possible acceptance of poor material or construction. The amount of testing would be related to the value of the units of the lot and to the reduction in price if the lot was not in full compliance with the specified requirements. There are decision function equations available (20-21) that re-late the number of measurements to the variability of the measured characteristic, the cost of sampling, and the pos-sible loss. Unfortunately, the use of these equations in their present form would indicate a much larger total number of measurements or much smaller lots than would probably be acceptable to highway agencies. This may be because the factors are too large for the probability of loss and for the actual derivation of value assumed to be related to the possible loss. This is an area that needs research.

It does appear that the current practice of sometimes tak-ing only four or five samples or making four or five mea-surements on a lot is grossly inadequate when the dollar value of the lot affected by the acceptance decision is con-sidered. Both buyer's and seller's risks are decreased when the number of samples is increased. There is presently no rational basis for determining lot size other than to limit it to one day's production or to a quantity of material that can be assumed to be produced by the same process. Con-sideration should be given to the amount of reduction in price if the lot only slightly fails to meet requirements.

DETERMINING EQUITABLE REDUCTION IN PRICE

Most of the new statistically oriented specifications have some INCENTIVE PROvISIoN for a reduction in contract price for material or construction that is not fully in compliance with requirements. In some cases points are assessed for deficiencies according to the degree of lack of compliance; the reduction in price is determined by the total number of points. When the specification permits an estimate of the PERCENT of the lot that is WITHIN the specification limits or TOLERANCES (PWT), a GRADUATED REDUCTION IN CON-TRACT PRICE is related to the PWT. In any case the objec-tive is to provide an incentive to the contractor or producer to maintain quality well within the specified limits without imposing large reductions in price that would result in contingency items in future bids.

All of the present systems for determining the amount of reduction in price are arbitrary to some extent. It would be more rational if the amount of reduction were related to loss of performance (or failure to accomplish the de-

signer's intent) of the item to which the reduction is ap-plied. With the exception of a few items, however, the relationship between partial compliance with specified re-quirements and performance has not been established. West Virginia specifications contain a table of percentages of reduction in price for pavement of deficient thickness. These percentages express the ratio of the wheel-load-carrying capacity of a pavement to a pavement that has a specified thickness of 9 in. (230 mm). Rex (22) sug-gests the use of the AASHTO Guide for the Design of Flexible Pavement Structures as a method of determining penalties for a deficiency of thickness of asphalt pavements. Using a normal 3-in. (75-mm) surface course as an ex-ample, he estimates that the loss of serviceability in terms of equivalent axle loads of a 0.25-in. (6-mm) reduction in thickness would be 17 percent. A 0.50-in. (13-mm) deficiency would result in a loss of 33 percent serviceability.

Vesié and Saxena (23) derived an equation relating serviceability of concrete pavement to slab thickness, con-crete strength, and other factors. This equation suggests that other factors being equal, a 7-in. (175-mm) slab thick-ness has only one-half the pavement life of an 8-in. (200-mm) slab. Vesi6 and Saxena also found that a 10 percent reduction in concrete strength may reduce the pavement life 35 percent and that a reduction in strength of 20 per-cent may result in a 60 percent loss of serviceability. These relationships supply a rational basis for reduction in price for thickness and strength deficiencies of concrete pave-ment; however, specification of such reductions in price would probably result in contractor target values in excess of design values and consequently increased costs.

Rosenblueth, Esteva, and Damy (24) have developed an equation that relates the reduction in price for strength deficiencies to the dollar cost per pound per square inch and failure probabilities. This equation has not been fully investigated but appears to result in a reasonable gradua-tion of penalties. Shane (25) has developed a linear pen-alty function that relates the size of the penalty to concrete cost and the level of quality assurance. This function is being investigated to determine the practicality if actual between-lot and within-lot variabilities are used along with practical sample sizes. It remains to be demonstrated whether these equations have immediate application; at-tempts are being made to derive a formula for a graduated series of reductions in price that would be fair to both buyer and seller and would have practical application.

ADMINISTRATIVE PROBLEMS

Administrative problems of statistically oriented end-result specifications appear to be chiefly associated with increased paperwork and accounting at the project level. The con-verse may be true, however, if the procedures are well de-signed. Advanced or computerized systems can simplify preparing summary reports. Many of the acceptance plans require field computation of averages, ranges, or percent-ages within tolerance (PWT). If reductions in price are applied to individual lots, the project engineer must tabu-late the price paid for each lot to determine contract pay-ments. When lots have MULTIPLE DEFICIENCIES, the

computations may become complex, depending on the way that the multiple deficienices are related to the percent of contract price for the lots. Other complications arise when the intended number of measurements is not avail-able because of an unexpected shutdown.

Although under ideal conditions inspection personnel may be reduced because of the contractor's quality con-trol system, competent personnel must be trained to spot-check the system and to make judgments as to its efficacy.

Forms must be devised for recording and documenting acceptance data. It may be preferable for the agency to devise forms for the contractor's quality control personnel in order to ensure uniformity.

HUMAN FACTORS

Human factors include the reluctance of middle manage-ment to abandon old methods for new as well as their reluctance to use statistical methods to complement pro-fessional judgment. In the field, inspectors who have been trained to take REPRESENTATIVE SAMPLES and SECOND SAM-

PLES may be reluctant to record tests made on unbiased random samples. Others may be reluctant to accept such test results when they show greater variability and lower average levels than shown by previous data. The success of ERS depends largely on adequate education and in-doctrination at all levels of an organization.

LEGAL FACTORS

In theory, the contractor or producer has always been legally responsible for the quality of materials or construc-tion. To some extent, the responsibility has been shared in the past through the use of restrictive specifications. In addition, some state agencies designed mixtures and per-formed work that should have been done by the contrac-tor's quality control system. With ERS, the responsibilities of the contractor and agency are sharply defined minimizing adverse legal proceedings on the part of the contractor or producer. In a complete ERS, the acceptance procedures should be detailed and should set forth the sampling point, sampling plan, and test methods on which acceptance will be based. If these procedures are strictly followed by the agency, all contractors can be assured of being treated fairly and equitably.

ECONOMIC FACTORS AND COST EFFECTIVENESS

A survey made in 1970 (26) indicates that when a state agency adopts acceptance testing, there are appreciable cost-benefits in quality control to the contractor. The dol-lar cost of such a system is estimated to be about 20 per-cent less than that of traditional procedures. The economic benefits in terms of the degree of quality assurance obtained by this system when compared to the existing system are estimated to be even more favorable.

Statistical quality control is not generally practiced in the highway industry, unlike other large industries. The prob-able costs of such a system were estimated by projection of current costs of contractors and producers who are cur-rently maintaining voluntary testing programs. Estimates

10

based on data obtained from 49 concerns in 15 states indi-cate that the cost of an acceptable degree of quality con-trol of highway materials or construction by the contractor or producer would average about 4 percent of contract price. The current total engineering expenditures by state agencies on federal-aid projects approaches 10 percent of contract price. Reported engineering costs on rural free-way projects are 13 to 10 percent of total construction costs (46). A study by the Federal Highway Administration based on record sampling indicates that the cost of the independent assurance sampling program (progress and final record sampling) amounts to 12 percent of total en-gineering costs. Other testing costs • must- be included to obtain the total cost. With reference to manufacturing in-

dustries, an editorial in Quality Management and Engineer-ing (48) states that "at least two large corporations that have gained many years of experience in this area have set a figure of 4% of net sales billed as a goal for total cor-porate costs. However, these people emphasize that this figure may be as much as 200% high or 400% low de-pending on the specific plant or division involved." In general, comparable quality cost figures for large industries are not available (27).

It is generally agreed, however, that the actual cost of inspection, sampling and testing of highway materials, and construction under traditional procedures has not been determined for many cases.

CHAPTER FOUR

CONTRACTOR'S AND PRODUCER'S QUALITY CONTROL SYSTEMS

GENERAL

The business of constructing highways differs from other large industries in the United States in at least one im-portant respect, that is, de facto assumption of responsi-bility for quality of the product by the purchasers. Nearly all other large manufacturers of end-use items or compo-nents have accepted the philosophy that they have an in-herent responsibility for quality and have installed quality control departments that assume an important role in man-agement. In the case of suppliers to the U.S. Department of Defense or National Aeronautics and Space Administra-tion, an approved quality program to ensure compliance

with the requirements of the contract is mandatory for the contractor and any subcontractors. Many of the detailed requirements for the quality program and inspection sys-tem are given in two Military Specifications: MIL-Q-9858A, and MIL-I-45208A.

Under these specifications, the producer controls manu-facture so that materials or products offered for acceptance will consistently maintain some agreed-upon quality level. The purchaser specifies what is desired but does not inspect the manufacturing operation. The purchaser, however, may inspect the manufacturer's quality control system to make sure that it is adequate and may perform acceptance sam-pling and testing on the completed product at the point of

acceptance in accordance with well-defined procedures. There is no question as to the manufacturer's or producer's responsibility for full compliance with the contract re-quirements.

Until recently, the assignment of responsibility for quality and satisfactory performance in the highway industry has not been as sharply defined. The requirements of Sub-section 104.01 of the AASHTO Guide Specifications for Highway Construction and similar requirements of state highway construction contracts were interpreted by the attorney general of Colorado (28) as placing the responsi-bility for quality control on the contractor. This responsi-bility applies not only to the contractor's work and that performed by subcontractors but also to the quality con-trol of materials purchased from suppliers. To some ex-tent, state agenices have specified the materials to be used, their proportions, the equipment and methods used in processing, and the equipment and methods of incorporat-ing the final product into the pavement or structure. The end product or construction may then be tested to establish compliance with specification requirements. In case of noncompliance, the product or construction, in theory, is subject to removal and replacement without additional compensation to the contractor. The representatives of the purchasing agency who have been intimately involved in the production or construction may retain some moral responsibility, however, for the end result and this may affect their acceptance decisions.

REQUIREMENTS FOR QUALITY CONTROL SYSTEMS

Agencies that have changed or are in the process of chang-ing to ERS usually include in specifications some require-ment for quality control by the contractor or producer. This requirement may be a simple statement that the con-tractor or producer shall install a quality control system or it may be a detailed list of requirements. The futurized version of FP-61 (9) suggests that:

The contractor shall be responsible for the complete supervision, performance, and completion of all work in accordance with the original approved or revised draw-ings, specifications, special requirements, and contract. The contractor shall provide and maintain an inspection system acceptable to the engineer, for the quality control of all construction and all products not accepted on the basis of certification.

The contractor shall provide and maintain an inspection system which will provide reasonable assurance that all materials, products, and completed construction submitted to [the agency] for acceptance conform to contract re-quirements whether manufactured or processed by the contractor, or procured from sub-contractors or vendors. The contractor shall perform or have performed the in-spections and tests required to substantiate product con-formance to plans, drawings, specifications, and contract requirements and shall also perform or have performed all inspections and tests otherwise required by the contract. The contractor's inspection system shall be documented and shall be available for review by the engineer prior to initiation of production and throughout the life of the con-tract. The engineer will furnish written notice of the

11

acceptability or nonacceptability of the inspection system. The contractor shall notify the engineer in writing of any change to his inspection system. The inspection system or changes thereto will be subject to disapproval if it does not provide sufficient protection against submission of nonconforming materials, products, or construction.

The contractor's inspection and testing procedures shall be prescribed by clear, complete and current instructions. The instructions shall assure inspection and test of ma-terials, work in process and completed construction as re-quired by the specification. In addition, criteria that will provide reasonable assurance of acceptance of submitted lots shall be included.

Inspection and test results shall be promptly recorded on approved forms or charts which shall be kept complete and shall be available at all times to the engineer during the performance of the work.

The contractor shall take prompt action to correct assignable conditions which have resulted or could result in the submission to [the agency] of materials, products and completed construction which do not conform to the requirements of the specification.

The contractor shall provide and maintain measuring and testing devices necessary to assure that supplies con-form to the technical requirements. In order to assure continued accuracy, these devices shall be calibrated at established intervals against certified standards. When re-quired, the contractor's measuring and testing equipment shall be made available for use by the engineer to de-termine conformance of materials, products or completed construction with contract requirements. In addition, con-tractor's personnel shall be made available for operation of such devices and for verification of their accuracy and condition.

Process control procedures shall be an integral part of the inspection system, where applicable to site operations.

West Virginia and Pennsylvania implemented contractor process control requirements based on the foregoing. West Virginia has supplemented these with specific minimum requirements for hot-mix bituminous concrete plants as given in Table 2. Pennsylvania has included similar guidelines in their specifications.

REQUIREMENTS FOR CERTIFICATION

In addition to the aforementioned requirements, specifica-tions may include a requirement that the plant be certified acceptable following an inspection by agency personnel and capable of producing a uniform product meeting specifica-tion requirements. An agency inspection generally includes the testing equipment and facilities used by the contractor's quality control personnel. The contractor's personnel may be certified as to their capability to perform the required tests, to make necessary computations, and to document results, as evidenced by examination or experience.

A minimum testing frequency is generally specified based on the expected variability of the measurements. Rapid test methods other than those used by the agency for acceptance purposes may be used. General requirements for docu-mentation of test results in a ledger are sometimes outlined. The construction and up-to-date maintenance of control charts may be specifically required. One concept of the contractor's process control activities is shown in Figure 1, which has been adapted from reference 26.

12

TABLE 2

CONTRACTOR'S PROCESS CONTROL REQUIREMENTS (26)

BITUMINOUS CONCRETE

Process Control Requirement Frequency*

A. ALL TYPES OF PLANTS

1. Cold bins

Determine aggregate gradation of each bin. As required Determine gate calibration chart for each bin. As required

C. Determine gate settings of each bin to assure compliance with plant-mix formula (PMF). As required

2. Hot bins

Determine aggregate gradation of each bin. 2 per day Determine theoretical combined grading. 2 per day

3. Bituminous mixture

Ross count. Per specifications Aggregate gradation. Per specifications

C. Percent of bitumen. 2 per day

B. WEIGH BATCH INCREMENT TYPE PLANT

L Batch weights

Determine percent used and weight (lb) of each bin to assure compliance with PMF. As required Determine weight (lb) of bituminous material used. Daily

C. CONTINUOUS VOLUMETRIC PROPORTIONING PLANT

1. Hot bins

Determine gate calibration chart for each bin. As required Determine gate settings of each bin to assure compliance with PMF. S As required

2. Bituminous material

a. Determine gallons per revolution or gallons per minute to assure compliance with PMF.

D. WEIGH SCALES AND ASPHALT PUMPS

Calibrate scales and pumps. Prior to start of job

Check calibration of scales and pumps. Weekly

*Frequency for process control will vary with the size and type of aggregate or mixture and the batch-to-batch variability of the item.

Contractors Acceptable (a)

Process Risk of Rejection Determine

__________ Characteristic

t to be Measured for Each Item

Determine State Agency

Process Control Specifications Limits i Quality Contro

Producers and Suppliers

Materials and Mixtures

Construction Operations I

Reprocessed

Non-Conf orm ing OL11 Material -

Wasted

ano testing

Procedures

Reports and Control

Charts

Sampling Plans

Process Sampled

Field and Laboratory Measurements and Tests

t Calibration

of Measuring and Testing

Equipment

Test Reports Contractors Control Group

Compares Results

vs.

Control Charts Control limits

Process Process Outside

Within Control Control Limits

Limits

Material, - Construction

Submitted to State Agency for Acceptance Testing

I Accepted I ( Accepted J

Rejected I at Reduced

Figure I. Contractor's process quality control (26).

14

CHAPTER FIVE

CONTRACTING AGENCY ACCEPTANCE PROCEDURES

TYPES OF ACCEPTANCE PLANS

General

Statistical acceptance plans are of two general types. The ATTRIBUTES SAMPLING PLAN is most useful when the char-acteristic or attribute of interest can not be measured or does not have to be. measured and the unit can be classi-fied as acceptable or defective by visual inspection. An attributes sampling plan can also be used when single measurements of the attribute are not normally distributed. The most general use of such a plan is in a go or no-go situation, such as when a unit of concrete pipe is either satisfactory or defective with respect to the presence of spalls or exposed reinforcing, or with respect to failure to meet a strength test. Particles of aggregate either have or do not have fractured faces and either meet or do not meet certain requirements for hardness. The number of irregu-larities in a concrete pavement slab either is less than or greater than a permissible limit.

The second type of acceptance plan, a VARIABLES SAM-

PLING PLAN, is most useful when the characteristics of a material or item of construction can be measured. Ac-ceptance depends on both the average value and variability of the measurements. The variability of the measurements is determined by their standard deviation or range. Vari-ables plans can be of two types: in one plan, the standard deviation is known or can be assumed to be known; in the other, the standard deviation is computed directly from the sample measurements. To simplify computations, the range can be substituted for the standard deviation.

When either an acceptance plan based on variables or a plan based on an attribute is appropriate, the choice can depend on several considerations. An attributes sampling plan requires practically no computations and is adaptable to control charting. The usual inspection process is to sub-ject each item in the sample to a rapid visual examination or to use a simple gage to determine whether or not a cer-tain dimension meets specified requirements. No elaborate testing or measuring equipment is needed and compara-tively little time is required for the inspection of a large number of items. Moreover it is often possible to note the presence or absence of two or more types of defects dur-ing a single inspection. To aid in the choice of a suitable buyer's or seller's risk, many tables and OPERATING CHAR-

ACTERISTICS CURVES are available (3). The great disadvantage of an attributes sampling plan is

that much available information is not obtained. Because the purpose of the inspection is simply to classify an item as good or bad, the inspection reports do not show the average level and variability of a characteristic. There is no indication of the corrective action that can or should be taken.

In general, a single-attribute sampling plan is much less efficient than a sampling plan based on variations in mea-surements. To obtain a certain buyer's or seller's risk, the

number of samples needed for an attributes sampling plan may be 30 percent greater than the number needed for a plan based on the distribution of variables. For these rea-sons, there are comparatively few examples of the use of attributes acceptance plans in connection with highway construction.

Attributes Acceptance Plans

Attributes acceptance sampling plans for thickness of con-crete pavement have been proposed for use by the New York State Department of Transportation (29); an at-tributes plan is described in NCHRP Report 168 (30). AASHTO M 164-74, Section 7, "Quality Assurance of Mechanical Requirements," includes a type of attributes sampling plan. The manufacturer is required to test a stated number of specimens from each lot depending on the size of the lot. A copy of the inspection test report is to be furnished to the purchaser on request.

Variables Acceptance Plans

Standard Deviation Known

Acceptance plans based on sample averages are comparable in simplicity to attributes plans. Acceptance limits for sample plans are based on the assumptions that the stan-dard deviation (o) of a large number of measurements is known and that the standard deviation [or the COEFFICIENT

OF VARIATION (V) derived from the standard deviation of a small group of measurements] representing a lot has the assumed value. The ACCEPTANCE LIMIT is determined by multiplying the known or assumed standard deviation (a) (or coefficient of variation) by a PROBABILITY FACTOR (k) and applying the result to the target value. For acceptance, the average (X,,) of a small group of measurements is required to equal or exceed a lower acceptance limit (L) or be less than an upper acceptance limit (U), or both. In equation form, accept if

(2)

In the case of plus and minus tolerances on the JMF or target value, the acceptance rule is, accept if

X,JMF+ka- (3a)

and if

X,,JMF — ka- (3b)

The acceptance plan for concrete on the basis of strength of ASTM C 94-71 (31) is, in part, based on averages of consecutive strength tests as given in the Table of Note 21 of the specification. The average is required to meet or exceed a value obtained by multiplying the specified strength by a factor given in the Table. These factors were obtained by the use of the equation (26, Eq. 2)

15

F= icr fcr V(2.055) (4)

f'c V

in which

icr = required average strength (psi); = specified design strength (psi);

n = number of consecutive test results; V = coefficient of variation expressed as a decimal;

and 2.055 = probability factor for 2 percent seller's risk.

This is using an assumed value of the coefficient of varia-tion (V) of 15 percent and a 2 percent seller's risk based on the standard normal variable value of 2.055.

One objection to this type of acceptance plan is that the standard deviation or coefficient of variation of small groups of measurements is seldom equal to the assumed value. In actual practice the buyer's (fi) risk and seller's () risk associated with this acceptance procedure are largely indeterminate because of the basic assumption that the coefficient of variation (V) remains constant and is known; however, this assumption may not be correct.

A more serious objection, however, is the large buyer's risk given in Table 3 (26), which is based on the assumed conditions. It can be seen that the buyer's ($) risk is so large that it may be unacceptable when the number of tests (n) is less than five.

Standard Deviation Unknown

When the standard deviation is not known, it can be esti-mated from the measurements on the sample. In this case the lower or upper acceptance limits (L or U) are usually determined by engineering considerations or by the limits of variability of measurements from the average of good material ('g). For acceptance, the average of n measure-ments (.,,) is required to be at least ks from L or U. In equation form, accept if:

(X,, - L) ks (5a)

or if

(U—X',,) ks (Sb)

in which

L or U = a specified limit; = the average of a small group of measurements;

s = the standard deviation of the small group of measurements; and

k = the factor that determines the statistical prob-ability of acceptance.

For example, if L = 3,000, then k',, would have to be ks greater than 3,000. If k = 1.00 and normal distribution is assumed, there would be a buyer's risk of about 16 percent of accepting material with an average value of less than L.

AASHTO M 242-73 (32) includes an acceptance limit of this type in the form

. 8 =L+1.07s (6)

in which X, is ..the computed allowable average of three to five strength tests of reinforced concrete pipe, L is the

specified strength, s is the standard deviation computed from the results of the three to five strength measurements, and 1.07 is a probability factor.

In the variables acceptance plans previously discussed, the buyer's ($) risk has been based on the assumption that the sample measurements had a normal distribution. To be statistically correct the Student's t distribution must be used for this purpose when the standard deviation is not known (33). One of the differences between the Student's r dis-tribution and the normal distribution is that with the distribution the probabilities of acceptance and rejection depend on the number of measurements from which the SAMPLE STANDARD DEVIATION (s) is computed.

For example, concrete having an average compressive strength of 2,500 psi (17 200 kPa) is not acceptable and a plan is to be designed with a buyer's ($) risk so that there is only one chance in 20 (or 5 percent) of accepting concrete of this quality. Acceptance is to be based on the average of the results from four concrete cylinder tests.

Then, assuming normal distribution, the acceptance rule would be: accept if

.2500+ 1.645—f— (7a)

Using the t distribution it would be: accept if

2500 + 2.353 (7b)

It can be seen that there is a much smaller actual risk of accepting concrete with an average strength of 2,500 psi if acceptance is based on the t distribution instead of on the normal distribution when the standard deviation is not known.

In this acceptance procedure, the buyer's (,8) risk of accepting concrete having an unacceptable average of 2,500 psi was fixed at 5 percent. As in the previous examples, nothing is stated about the seller's (ce) risk of having material with an acceptable average ('g) rejected. Obviously if an acceptance plan entails a high seller's risk

TABLE 3

BUYER'S () RISK ASSOCIATED WITH ASTM C 94-71

seller's risk (a) = 2 percent f 3000

V 5 10 15 20 25

far 3210 3440 3720 4030 4220

8 (%) n Acceptance Limit (i) for X

78 1 2880 2730 2570 2380 2150

60 2 2970 2940 2900 2860 2810

44 3 3020 3030 3050 3080 3110

31 4 3040 3090 3140 3210 3280

21 5 3060 3130 3200 3290 3400

14 6 3070 3150 3250 3360 3490

16

its use will eventually lead to large contingency items in future bids. When the standard deviation can not be as-sumed to be known, it is not possible to fix both the buyer's and seller's risks. If the buyer's risk is fixed, however, it is possible to estimate the seller's risk under stated condi-tions by use of the tables of noncentral t (19, 34).

The size of the seller's risk is determined by the number of measurements (n) made on the lot, the buyer's risk, the estimate of the POPULATION STANDARD DEVIATION (a-'), and the distance (Kp) between the average of poor material (',) and the average of good material ('g). This dis-tance is measured in multiples of the estimate of the population standard deviation.

If n = 4, the stated value of = 3,500 psi (24 000 kPa), the value of X'g = 4,500 psi (31 000 kPa), the size of the buyer's risk = 5 percent, the estimated value of the standard deviation (a') = 500 psi (3 400 kPa), and Kp = 2.0, from the noncentral t tables the seller's risk is about 10 percent. If the value of the standard deviation (o-) 660 psi (4 500 kPa), the seller's risk would be about 28 percent whereas if the value of the standard deviation = 400 psi (2 700 kPa), the seller's risk would be only about 2 percent.

Using the tables of I and noncentral t, statistically de-fensible acceptance procedures can be designed that will have acceptable buyer's and seller's risks (18). -

USE OF RANGE IN ACCEPTANCE PLANS

Substituting Range (R) for Sample Standard Deviation (s)

Acceptance procedures using the sample standard devia-tion (s) may not be suitable for field use because of the computations involved. A variety of inexpensive, portable calculators can compute the sample. standard deviation quickly, however. In addition, these computations can be avoided by use of the range (R). For small groups of 4 or 5 measurements, the range or difference between the small-est and largest measurementin the group has approximately 96 percent efficiency as compared to 100 percent efficiency of the sample standard deviation computed from the group of measurements (35).

Substituting R for s in acceptance procedures is accom-plished by the use of the constant, c (36) or d21 (37). Values of these constants for small groups of measurements are as follows:

n C

1.910 4 2.234 5 2.474 6 2.670 7 2.830 R = cs

ts (8a)

Vn

can be converted to

X fl —X 1) +----R (8b) c'/n

Using values from the previous example with n = 4 and with a 5 percent buyer's risk, the acceptance rule would be: accept if

2.353s X4 -2500+ -

'/4

2500+ 1.18s

2500+ 1.18

R4 C

2500 + 1.182.234

R4

- 2500 2500 + 0.53 R4 or

x4 0.53

R4

If tic Vn = k, a simple table of acceptance constants can be constructed for different buyer's risks and sample sizes (Table 4).

Acceptance Plans for Estimating PWT and Reduction in Price

The constant k is directly related to the QUALITY INDEX (Q) and has the same numerical value. For values of k other than the minimum required for acceptance, the percent within tolerance (PWT) can be estimated by reference to a Q TABLE (8, 19). For example, when n = 4, the Q table shows that when Q = 0.53 the PWT of the material is about 90 percent. However, if the value of Q is only 0.43, then PWT = 82 and only about 82 percent of the material would be within tolerance. Thus, an acceptance procedure of the form accept if

XflXJ k (9a)

R,

or

)Xfl (9b)

R,,

provides not only a definite limit for acceptance or rejec-tion but through the Q table affords a way of establishing graduated reductions in contract price related to the PWT of material not in full compliance with specified require-ments.

One way of arriving at the amount of reduction in con-tract price (used by Colorado, Alaska, and Wyoming) is to write the acceptance equation in the form

P = (X,, - U + kR,JF (lOa)

or Using this approach, an acceptance procedure of the

form, accept if P = (L - X, + kR,)F (lOb)

17

in which P is the percent reduction in contract price, U and L are upper and lower tolerance limits, and F is a price reduction factor for a particular characteristic.

RANDOM SAMPLING

ERS simply does not work unless all samples taken to de-termine acceptability are unbiased. Any exercise of judg-ment as to whether or not the sample will produce a good, failing, or average test result nullifies the assumptions on which a statistically oriented specification is based.

The best way to ensure unbiased sampling is to pre-determine the sampling points or sampling times by the use of a random number table. Tables published in text-books are somewhat inconvenient, but simplified tables are available for area sampling (9-10). When these numbers are used to locate sampling points in a large lot, the loca-tions can be clustered in a relatively small part of the lot. To avoid this condition, RANDOM NUMBERS are usually used in what is called stratified random sampling. The lot is divided into SUBLOTS and an equal number of random locations found for each sublot.

In determining the density of newly constructed asphalt pavements, there is usually a general pattern of high density in the center of the traffic lane and a lower density in the free edges of the lane. If random sampling points are lo-cated over the entire width of the lane, the transverse variability increases the standard deviation of the measure-ments and leads to uncertainty as to the actual average density. It may be preferable to use a type of stratified random sampling with random sampling points located along a line parallel to and approximately 2 or 3 ft (0.6 or 0.9 m) from the free edge.

Improvised methods (such as randomly drawing num-bered slips from a hat) may be used when sampling from a process, as for example in the production of a hot-mix plant. Systematic sampling in which samples are taken at the same time each day is not acceptable for use in quality control or with statistical specifications. Additional infor-mation on random sampling can be obtained from ASTM E 105, "Probability Sampling of Materials."

RAPID TEST METHODS

Although not a part of ERS, rapid test methods are becom-ing essential in acceptance procedures and quality control. The use of these methods makes an increased number of tests practical and economical, and acceptance decisions can be made quickly. Nuclear methods of measuring the density of asphalt pavement behind the roller provide con-tractors with a quality control tool for ensuring compliance with specified density requirements and for optimizing the use of personnel and equipment.

Currently 39 states are using nuclear systems for speci-fication control of the moisture and density of soils and aggregates. Nuclear equipment is being used by 29 states for specification control of the density of bituminous con-

crete; 15 states are engaged in nonspecification or research use of nuclear systems for measuring asphalt content (14).

In addition to nuclear equipment for density determina-tions, rapid methods include a nuclear cement-content mea-suring system for plastic concrete (38), a pycnometer method of determining the asphalt content of hot-mixed paving mixtures (39, 40, 44), and a method of prediction of the potential 28-day strength of concrete from the results of early tests (16, 45). There are many others, some of which are under development by the Federal Highway Administration and state highway agencies.

PERSONNEL

The transfer of certain quality control responsibilities to the producer or contractor has created a shortage of qualified technicians in some areas. This shortage has been reduced in some states in various ways. For several years, West Virginia has conducted training programs for CERTIFIED

TECHNICIANS in concrete and hot-mix plant control. New Jersey has jointly financed and conducted training courses for state and contractor personnel.

Trained agency personnel should be required to monitor the contractor's and producer's quality control systems. Ideally, when confidence has been established as to the accuracy and reliability of these systems, agency testing personnel can be greatly reduced. In addition, the frequent contact of contractor technical personnel and state agency monitors can result in more realistic specification require-ments if it is evident that tolerances for some characteristics are actually unreasonable.

The Pennsylvania Department of Transportation, in co-operation with the Pennsylvania State University, is con-ducting training courses related to the understanding and use of statistically oriented specifications for department and contractor personnel. Virginia Highway Research Council conducted the first statistical seminar and has presented courses in statistical quality control methods in concrete pavement construction for the American Concrete Paving Association. Other agencies have conducted or are planning to conduct similar seminars or training.

TABLE 4

ACCEPTANCE FACTOR k: BUYER'S RISK OF AC-CEPTING V,'

n 1% 5% 10% 20%

4 1.021? 0.531? 0.37R 0.221?

0.681? 0.381? 0.28R 0.17R

0.51R 0.31R 0.231? 0.141?

0.421? 0.26R 0.191? 0.12R

18

CHAPTER SIX

CURRENT PRACTICES IN HIGHWAY AGENCIES

U.S. HIGHWAY AND TRANSPORTATION AGENCIES

Based on available information, 33 states are using, plan-ning to use, or have tried some form of statistically oriented end-result specifications. Eighteen states are not using and not planning to use this type of specification.

Although many states have relaxed equipment require-ments and require some degree of quality control by the contractor, the ERS adopted by most states are essentially statistical concepts applied to acceptance procedures. These are end-result specifications (ERS) to the extent that ma-terial or construction is accepted, rejected, or accepted at a reduced price on the basis of the quality of the completed product or item of construction.

The following compilation of the use of ERS in highway agencies is based on information received in response to a Transportation Research Board letter of request for as-sistance and from other available sources. Because of space limitations the abstracts are necessarily incomplete and are primarily intended to show the statistical approach in the various state agency specifications. Features of ERS con-sidered or implemented by various agencies are tabulated in the Appendix.

Alabama

Prior to 1967, Alabama conducted a large amount of re-search to determine parameters for asphaltic and portland cement concrete, aggregates, compacted bases and subbases, and compacted embankments. They have not used and are not presently planning to use ERS.

Alaska

Alaska has a statistically oriented acceptance plan in sup-plemental specifications that applies to asphaltic-concrete plant-mix pavement. The plan governs acceptance with respect to gradation, oil (asphalt) content, and pavement density. A lot is the amount of material produced and placed in one day; normally five samples are taken from each lot. Lots are evaluated for acceptance and price reduction by use of the equation

P=(X,,+aR—T)F ( ha)

or

P= (T1 +aR—)F (lib)

These equations are similar to those used by Colorado and Wyoming. Values of a are given for sample sizes of from one to five and F-values are given for percentages passing sieves, compaction, and asphalt content by extraction. If the total of the P-value is less than 3, the material is ac-cepted at contract price. If between 3 and 25, the engineer may require correction or may accept the material at a reduced price. If the total P-value is greater than 25, the

engineer may require removal and replacement, corrective action, or the material to be left in place with no payment.

The specifications are used for both producer and con-tractor quality control and have been accepted very well by contractors. The initial reaction of construction and ma-terial district personnel has been most favorable. Primary advantages cited are the placement of acceptance criteria on a quantitative basis and the establishment of a uniform procedure for applying price adjustments to usable ma-terials that may be slightly out of specifications. There has been no significant influence on cost-effectiveness; the num-ber of inspection personnel has not changed although the amount of testing has increased slightly.

Arizona

Arizona has a statistical acceptance plan for asphaltic con-crete that has been in use since 1973. The lot size is 2,500 tons (2 300 metric tons) divided into five sublots. Random samples are taken from the mat after placement but before rolling in each sublot. Acceptance with respect to asphalt content and gradation is based on the average of the results of five samples. Provision is made for the use of a lesser number of test results if necessary.

The acceptance schedule lists pay factors from 1.00 to 0.40 for deviations of the mean of the lot for 1, 2, 3, 4, and 5 tests. Mixture characteristics covered are asphalt content and percent passing the No. 200 (75-sm), 40 (425-sm), 8 (2.36-mm), 4 (4.75-mm), and/8-in. (9.5-mm) sieves, and the ½-in. (12.5-mm) and larger screens.

The specifications apply only to the acceptance of the contractor's product. The number of inspection personnel or amount of testing has not been reduced.

Arkansas

Arkansas has not used ERS because of resistance by local contractors but will continue to promote the concept in certain areas of highway construction.

California

California has made limited use of statistically oriented end-result specifications in the areas of relative density of treated and untreated soils and aggregates, purchased con-crete with specified 28-day strength, final thickness of PCC pavement, net cement content of cement-treated base, and contraction joints in PCC pavement. Quality control speci-fications apply to both contractor and producer; however, the contractor is held primarily responsible for producer quality assurance except where department inspectors re-lease material from manufacturing plants.

Most specifications are based on the relation of a moving average of five measurements to specified tolerances. A reduction in price applies only to concrete pavement de-

19

ficient in thickness. For the most part, producers and con-tractors have accepted these specifications with a minimum of objections.

Contractors previously wanted more end-product speci-fications and their own quality control. In view of the in-creased sampling required and other problems, however, contractors are no longer strong advocates of end-result specifications. The Associated General Contractors in co-operation with an AASHTO committee recommended that respective state highway departments continue to perform sampling and testing. California contractors have generally followed this recommendation.

Several producers of aggregate, ready-mix concrete, and asphaltic concrete believe that much of the department's quality control testing is a duplication of work they al-ready perform. They would prefer to be held responsible for the product as delivered to the contractor. The pro-ducers believe, however, that the department should main-tain a quality assurance program.

For the most part, the department has been pleased with the end-result specifications, but there has not been over-whelming support and progress has been slow. If test meth-ods improve and more rapid test methods to analyze ma-terials in place become available, California may adopt more end-result specifications. California is attempting to develop test methods that will indicate the quality of the product as soon as it is in place or at least within 24 hours after incorporation into construction. Other research proj-ects related to specifications and quality assurance are under way or planned for the near future.

6. Colorado

Since 1969, Colorado has had a statistical acceptance speci-fication used in practically all projects. The equation for determining acceptance includes a term (F), which estab-lishes the percentage reduction in contract price for ma-terial or construction not in full compliance with specifica-tion requirements:

The formula, P = (. + aR - T,,)F, will (12a) be used if a maximum limit only is speci-fied or when the average of the several test values is above the mid piont of the speci-fication band or above the job-mix formula value. The formula, P= (T L +aR—,,)F, will (12b) be used if a minimum limit only is speci-fied or when the average of the several test values is below the mid point of the speci-fication band or below the job-mix formula value.

Where

P is the percent of reduction in contract price,

X,, is the average of the several test values from samples taken from the lot, with n indicating the number of values,

a is a variable factor to be used as n changes according to the following:

when n is 3, a = 0.45; n is 4, a = 0.38; n is 5, a = 0.33; n is 6, a = 0.30; and n is 7, a=0.28.

R is the difference between the highest and lowest values in the group of several test results from the lot,

T,, is the upper or maximum tolerance limit permitted by the specifications,

T1 is the lower or minimum tolerance limit permitted by the specifications, and

F is the price reduction factor to be applied for each element as shown in the table included in the specification.

These equations have previously been used with a = 0.33, n = 5, but have been slightly revised with n = 3 to 7 for use on future work. A lot is the quantity represented by the 3 to 7 samples. A method of identifying suspect values or outliers is provided.

F-values are tabulated for a large variety of tests. These range in size from 0.01 for metal tensile strength to 25.0 for asphalt content.

There is no reduction in contract price when P is nega-tive or less than 3. Under the revised specifications, the percent reduction in price is applied to a fraction of the contract price ranging from 0.10 for soil compaction to 0.70 for asphalt content of paving mixtures.

A lot is normally the quantity represented by five sam-ples. The maximum quantity represented by each sample is set forth in the department's standard sampling and test-ing schedules.

A system of stratified random sampling is used based on a table of random numbers.

The specifications are used for acceptance by the division of highways. Contractors can use the information for quality control. The initial reaction to a statistical accept-ance specification was very negative. Contractors now readily accept the specifications and indicate a strong de-sire for continued use. There has not been any significant change in bid prices, inspection personnel, or amount of testing.

7. Connecticut

Connecticut is in the process of field-testing ERS for two-course bituminous concrete. A field-testing program simu-lating the application of the specification has been carried out on two paving jobs and there are plans for two more. The data obtained are for informational purposes only; incentive provisions have not been enacted and, thus, pay-ment to the contractor is not affected.

The simulated specification requires that the engineer maintain control charts for bitumen content and for each sieve size shown in the job-mix formula. The centerlines of the charts are the JMF values. The control limits for indi-vidual tests are based on Table II of ASTM E 178, "Recom-mended Practice for Dealing with Outlying Observations"; results falling outside these limits are not included in the daily averages. Target density for compaction is based on ten nuclear density measurements on a control strip that must have an average Marshall density of 95 percent. Sub-

20

lot size is approximately 500 yd2 (418 m2 ) and one ac-ceptance test is made on each sublot. At least 15 tests must be made on each day. Test results showing less than 96 per-cent of target density require recompaction of the sublot. The lot size for gradation and bitumen content is the num-ber of tons produced in one day. Four or preferably five samples are taken from each lot. Lot size for surface tolerance is also one day's production.

The acceptance schedule provides for reducing the pay-ment to 95 and 90 percent for bitumen content when the average of four tests deviates more than 0.25 percent (per-centage points) from the JMF or 0.32 percent for the average of five check tests. Similar reductions in price are given for four sieve sizes, for average compaction less than 96 percent of target value, and for percent of pavement exceeding 2.0 percent of deviation ¼ in. to 10 ft (6 mm to 3 m).

Producer and contractor reactions have been agreeable, but incentive provisions have not yet been enacted. Con-necticut believes that ERS shows promise.

Delaware

Delaware has not used and currently does not plan to use ERS. There is no research in progress on ERS.

Florida

Florida has been active in the statistical treatment of his-torical test data since 1965. This has led to the investiga-tion of test methods; to the development of tentative pro-cedures for random sampling; to a tentative department policy on contractor's quality control systems; to tentative quality control and acceptance procedures for portland ce-ment concrete, coarse aggregate, and reinforcing steel; and to tentative asphaltic-concrete-pavement quality assurance requirements.

Statistically oriented ERS have been implemented for asphalt cement. ERS for asphaltic concrete have been simulated on four projects; ERS for structural concrete have been simulated on two projects. Statistical parameters of tests on lime rock for base construction are being de-termined. A study has been completed of a method of predicting potential 28-day strength from the results of early tests.

As part of the training program, the ERS for asphalt cement and the contractor's quality control requirements have been presented and discussed with contractors at three statewide conferences.

Georgia

Georgia has used ERS in the "Special Provisions for Port-land Cement Concrete and for Hot-Mix Asphaltic Concrete Construction." The specified size of a lot of concrete pave-ment is 5,334 yd 2 (4460 m2 ) and the lot is divided into three sublots. Two sets of two cylinders each are cast from the concrete from each sublot. One set for early tests is cured in accordance with ASTM C 684-73T, Method A. The other set for 28-day test is cured by the standard method of AASHTO T 23. Paving concrete is accepted on

the basis of early strength. The average of the three sets of early-strength cylinders must equal or exceed 3,000 + 0.18R psi (20.6 + 0.18R MPa) for acceptance. If the early tests fail to meet this criteria, three sets of 28-day cylinders are retained. If the average 28-day strength is less than 3,000 + 0.18R psi, the pay factor is reduced to 0.95 for concrete of strength equal to or exceeding 3,000 - 0.07R psi (20.6-0.07R MPa) and to 0.70 for concrete of strength equal to or exceeding 3,000 - 0.30R (20.6 - 0.30R MPa). The engineer may order removal of concrete not meeting the requirements for the 0.70 pay factor or may allow it to remain in place with a pay factor of 0.50.

The target density for asphaltic concrete is determined by a CONTROL STRIP, which must have at least 92 percent of VOIDLESS DENSITY based on APPARENT SPECIFIC GRAVITY

of the aggregates. For aggregates having an absorption greater than one percent, the target density must equal at least 96 percent of voidless density.

The contractor is required to submit a JMF along with eight specified items of related information.

A lot of asphaltic concrete is normally one day's pro-duction; four tests are made on each lot. For contracts greater than 2,000 tons (1 800 metric tons), the average of the four tests is required to be within tabulated limits for a pay factor of 1.00 for asphalt content and the percent-ages passing designated sieves. A graduated scale of re-duction in price is provided for lots not meeting the criteria. Criteria are provided for asphalt content determined by extraction and by digital printout of asphalt weight for each batch. Provision is made in the tables for test results less than four in number.

When two or more pay factors are less than 1.00, the adjusted payment is based on the lowest pay factor. To ensure uniformity, the range of the tests made on each lot is required to be less than stated limits for asphalt content by extraction and for percentages passing certain sieves. If the range tolerance is exceeded, a pay factor of 0.55 is applied. If the average density of the compacted pavement in a lot is not equal to or greater than 97.5 percent of target density, a graduated reduction in pay factors is applied. The surface-tolerance acceptance schedule provides for a graduated reduction in pay factors for lots exceeding 0.75 percent for new construction and 1.00 percent for resurfacing of tested length based on a 1/4 -in. (6-mm) deviation from a 15-ft (4.5-m) rolling straightedge pro-vided by the contractor.

Hawaii

Hawaii uses ERS for the density of compacted asphaltic concrete pavement. The contractor is required to build a control strip one-paver-width wide and having an area of approximately 400 yd (330 m2 ). A test site is located near the center of the length of the control strip and nuclear density measurements are made at this site after each roller pass to establish the density growth curve. After the in-crease in density is 1.0 pcf or less (16 kg/m3 or less), 10 nuclear-density measurements are made at random lo-cations over the entire strip. The average of these 10 measurements is the target density for future lots.

21

The standard deviation of the ten control-strip measure-ments is computed and the average control-strip density less 1.64 times the control-strip standard deviation becomes the lower control limit for density. If any individual density test falls below this limit, the contractor is required to rework the sublot represented by the test.

A minimum of five density tests are made during each production day. The standard deviation of these tests is pooled with the standard deviation of the control-strip measurements and the pooled value is used in a t TEST OF

MEANS to determine if the production-day average density is equal to or greater than the control-strip density. If it is not, the pavement may be accepted at 92 percent of contract price.

On the two projects in which ERS has been used, con-tractors favored its use. Hawaii has reserved assessment of ERS pending further use.

Idaho

Idaho has recently adopted a computer-oriented program of statistical control of asphalt hot mixes. Maximum tol-erances (plus or minus) have been established for both a single test and the average of 2, 3, 4, or 5 tests of certain characteristics. These characteristics are the percent asphalt content and the percentages passing the ½-in. (12.5-mm), 3/8 -in. (9.5-mm), No. 4 (4.75-mm), 8 (2.36-mm), 16 (1.18-mm), 30 (600-1 m), 50 (300-m), and 200 (75-jim) sieves. Individual test results and the moving average of five test results are plotted on a control chart posted in the field laboratory at the plant site.

Both individual and moving average must conform to the applicable criteria in order for the material to be accepted. Quality control is maintained by rejecting out-of-specifica-tion material instead of by a price reduction system.

Contractors are on record as opposing a price reduction system.

Illinois

Illinois revised Section 406, "Bituminous Concrete Binder and Surface Courses, Special" of the standard specifications for road and bridge construction to a statistically oriented specification in 1973. Most equipment restrictions have been eliminated. A lot is one day's production of one type of mixture. The contractor must submit for approval a job-mix formula based on the department's design methods.

Five random samples are taken from each lot for ex-traction tests from material behind the paver. Acceptance or conditional acceptance is based on the arithmetic aver-age of the five test results for percent retained on stated sieves and for asphalt content. Tolerances are given for payment of 100 percent down to 40 percent.

Density of compacted pavement is determined by test of five cores taken at random locations in each lot. The required density is 93 percent of voidless density as deter-mined by a high-pressure air meter. Compliance is deter-mined for each lot by use of the equation

results of computations of percent of MAXIMUM DENSITY

made on the five random cores taken from a lot. A quality level of 93 or greater is required for 100 percent payment. A schedule of reductions in contract price is provided for lots having a quality level of less than 93. The acceptance equation includes a buyer's risk of about 10 percent of accepting pavement having an actual density less than 93 percent of voidless density.

The average thickness of the lot is estimated from three measurements from each of the five cores from each lot. A tolerance of ¼ in. (6 mm) from plan thickness is allowed. Lots having an average thickness with a de-ficiency exceeding the tolerance are subject to a reduction in contract price.

Producers feel that this type of specification increases responsibility. Contractors are sometimes skeptical and have not used ERS to its fullest economic advantage. The agency feels that satisfactory quality control can be ob-tained with ERS if appropriate acceptance parameters are used.

Indiana

Indiana has developed ERS for construction of bituminous pavements and simulated its application on several selected projects to determine whether such acceptance procedures are suitable. Under the simulated conditions all acceptance testing is performed by commission personnel other than those who perform the standard control testing.

The contractor is responsible for maintaining control of the paving mixtures; a minimum of two samples of the mixture must be taken from each lot. A lot is one day's production of a particular mixture. Samples are to be taken by stratified random sampling with a minimum of one sam-ple in the morning and one in the afternoon. When the results of individual tests or lot averages approach or ex-ceed tabulated tolerances for gradation or bitumen content, the contractor must take corrective action. Acceptance of mixtures is determined on the basis of extraction test re-sults on five random samples from each lot. Samples are taken from trucks at the plant. The contract price for mixtures is adjusted for variations from allowable toler-ances in accordance with tabulated values. When more than one adjustment is made on a lot, the minimum per-centage of contract price governs.

Maximum density is determined from a TEST STRIP. If the average density of a lot is less than 98 percent of the maximum density, a graduated table of reduction in price is used to determine payment.

Surface smoothness is determined by use of a 16-ft (4.9-m) rolling straightedge. If more than one percent of the total length tested does not comply with surface smooth-ness requirements, an adjustment in contract price applies.

Because ERS has not been applied, contractor comments are based on speculation and Indiana can not yet estimate the success of ERS.

Iowa

- 0.275R = Quality Level (13) Iowa uses a statistically oriented acceptance plan for den-sity and thickness of compacted Class A subbase, asphalt- where X, is the average and R is the range of the five

22

treated base, aphaltic concrete base, and Type A asphaltic concrete. In general, acceptance is based on a quality index (QI), which is calculated by the use of the equation

QI - average value - (specified value - tolerance) (14) - largest value - smallest value

Five to seven measurements are used and QI-values of from 0.30 to 0.40 corresponding to about 80 percent within tolerance are required for acceptance without reduction in price.

Contractor and/or producer reactions have been ex-cellent; department reaction has been the same. This type of specification is being evaluated for other materials and project control applications.

Kansas

Kansas has not used ERS but is interested in other states' experiences. In general, the state administration as well as most producers and contractors do not favor adoption of ERS.

Kentucky

Kentucky has not developed any programs related to ERS but serious thought is being given to this concept.

Louisiana

Louisiana has implemented end-result or statistically ori-ented specifications for both asphaltic and portland cement concrete in their special provisions and are in the process of developing embankment and base-course specifications.

The specifications for asphalt cement, cutbacks, and emulsions list the customary AASHTO or ASTM require-ments and limiting values which, if exceeded, result in an adjusted contract unit price of 99 percent of the asphaltic concrete or 80 percent of the shipment of asphalt cement. When test results are such that a penalty would result from more than one of the test values, only the price adjustment for the greatest reduction is applied. If the asphaltic ma-terial exceeds other stated limits, either payment is with-held or the material is removed and replaced.

The contractor must assume full responsibility for the quality control of asphalt paving mixtures and must have a certified asphaltic-concrete technician at the plant. Mix-tures are accepted in lots equal to one day's production. A stratified random sampling plan is used so that two samples are taken in the morning and two in the afternoon. The time of sampling is determined by the use of random numbers. Provision is made for cases in which less than four samples can be obtained. Adjustments in price are made for pavement not in full compliance with require-ments for Marshall stability, roadway density, and surface tolerance. The lower percent of contract price is used for final adjustment for material deficient in both stability and density.

The lot size for pavement compaction is one day's pro-duction divided into five sublots. One density test is made on each sublot and acceptance limits are given for indi-vidual tests and the average of five tests. A graduated table of reduction in bid price is provided for percent of linear

feet of pavement exceeding surface tolerances of 1/8 and i46-in. (3- and 1.5-mm) deviation from a 10-ft (3-rn) rolling straightedge.

The contractor is required to furnish the source of ma-terials and the mix design for portland cement concrete on a form provided by the department. The contractor has full responsibility for the quality control of concrete mixes. A certified concrete technician must be present at the plant or at the job site whenever work is in progress. A lot of structural concrete is an identifiable pour not exceeding 200 yd3 (153 m3), For small items a lot is an identifiable pour completed in one day.

A lot of concrete pavement is equivalent to 3,000 linear ft (900 m) of single-lane construction. Surface testing of a lot is accomplished with a 10-ft rolling straightedge. Tolerance is ½ in.; not more than four percent of con-tinuously reinforced pavement or eight percent of jointed pavement should exceed this tolerance. Deviations of ¼ to ½ in. (6 to 12 mm) must be corrected by grinding; pavements with ½-in, deviation must be removed and re-placed. Five cores are taken from each lot and when the average thickness deficiency is more than 0.10, a graduated adjustment in price is applied. When the average com-pressive strength of cores 28 days or more is not greater than 4,000 psi (27.6 MPa) for non-air-entrained concrete and 3,600 psi (24.8 MPa) for air-entrained concrete a graduated adjustment in price is applied. When a penalty applies to both thickness and strength, only the adjustment requiring the greater reduction applies.

Two random batches are sampled from each lot of struc-tural concrete and three cylinders for a 28-to-3 1-day test are made from each batch. Minimum strengths of from 3,300 to 4,200 (23 to 30 MPa) are specified for different classes of concrete. Concrete not obtaining these average strengths for the lot is subject to a graduated reduction in price. Other minimum strengths apply to the average strengths of single batches. Concrete not obtaining these strengths is investigated and if the concrete is allowed to remain in place, payment is based on the average compressive strength for the lot. When the average compressive strength for the lot is less than stated minimum values, an investigation is made. If the concrete is allowed to remain in place, pay-ment for the lot is based on 50 percent of contract price. Nonstatistical control charts are maintained for the per-centages of concrete aggregate passing all specified sieves. Nonstatistical control charts showing specification limits are also maintained for asphalt content, percentages of aggregate passing JMF sieves, temperature, and percent crushed concrete.

Both the department and contracting agencies are pleased with the results obtained by the use of ERS. A reduction in amount of testing and required number of personnel has definitely been obtained. Department opinion is that ap-preciable financial benefit has been achieved.

Maine

Maine simulated use of ERS for asphalt paving in 1969-70. In 1972 an ERS contract was written that did not include a bonus feature and was withdrawn at the contractor's request.

23

Maryland

Maryland has not fully instituted ERS but has incorporated some of the features in acceptance of portland cement, in portland cement concrete, and in the price adjustment sys-tem for asphalt content when off-test samples are reported.

Massachusetts

Massachusetts does not use ERS.

22. Michigan

Michigan does not use ERS; however, current trial projects using this concept could lead to implementation of this approach to quality assurance.

Minnesota

Minnesota developed an ERS acceptance plan for asphalt pavement in 1973. Acceptance is based on comparison of construction densities with the average of 10 nuclear-density measurements on a control strip. The average must be at least 96 percent of Marshall density.

Lot size is normally one day's production of hot-mix and each lot is divided into five sublots of approximately equal area. One nuclear-density test is made on each sublot. Each sublot density is divided by the control-strip density to obtain a percent relative density. The five relative-density percentages are averaged to obtain a mean relative density and the range of the five percentages is found. The quality level is then calculated using the equation:

Quality Level = mean relative density - 0.60R (15)

A quality level of 95.5 or greater is required for 100 per-cent payment. A lower quality level results in a reduction in contract price for the lot. This type of statistical ac-ceptance plan is apparently based on a buyer's risk of about 1.5 percent of accepting lots having 95.5 percent of control-strip density.

An ERS for acceptance of bituminous mixtures is being developed. Acceptance with respect to percentages passing sieves and asphalt content is based on average values cor-rected for the number of tests. Average values outside of tolerances result in a graduated reduced payment.

Two statistically orinted end-result specifications for granular base have been developed: one for the aggregate gradation and the other for compaction.

The plan for aggregate gradation uses a lot size of 7,000 yd3 (5 400 m3) divided into four segments. Samples from each segment are tested separately and the percent-ages passing stated sieves are averaged. If the average values are outside given tolerances, either the bid price is reduced or corrective action is required.

Acceptance of aggregate base course with respect to compaction is referenced to the average of 10 nuclear-density measurements made at random locations on a con-trol strip. The standard deviation is comptued and used along with the average to provide an estimate of the num-ber of measurements to be made on lots represented by the control strip. The equation used is a variation of ASTM E 122 (48):

[0.03X2cr

N= ] (16)

In theory, this equation provides the number of measure-ments (N) on future lots that result in an average (X) that is correct within plus or minus three percent, 95 percent of the time. For example, if the average of the nuclear mea-surements on the control strip was 140 pcf (2.2 X 10 kg/rn3) and the standard deviation was 5.0 pcf (80 kg/ m3), six measurements would be required on each lot of approximately 1,000 ft (300 m) in length of constructed roadbed. If the average of the density measurements is

greater than the control strip average minus 2cr! Vn the lot is accepted. If an acceptable average density can not be obtained, a new control strip is constructed.

The reception of the ERS for asphalt pavement has been very favorable. Although the department was overstaffed during the transition period, in about two years a reduction of inspection personnel as well as total engineering cost per project is expected. It appears that a more uniform product than in the past is being obtained as a result of ERS.

Contractors' reactions to granular base ERS have been mixed. At this time bid prices have reflected an increase in cost, but it is expected that this increase will be reversed when ERS begins to come into greater use.

There has been a greater uniformity in product and construction as well as improved quality.

Mississippi

Mississippi has used statistically oriented specifications in their special provisions since 1964. Definite lot sizes are specified for most types of construction and the method of random sampling is included in the standard operating procedures. Tolerances are in terms of units of deviation (UD). A UD is a proportional part of the over-all variance inherent in results obtained from sampling and testing ma-terials or work produced under realistically controlled con-struction conditions and having adequate quality charac-teristics. For plant-mixed pavement mixtures, values of the UD are specified for the percentages passing specified sieves and for the percent asphalt content. For each lot the UD-values for asphalt content, gradation, stability, and density are determined by specified procedures. The pay factor for a UD-value not exceeding 0 is 100 percent of the price bid. Larger values of UD relate to reduced percentages. The final pay factor is obtained by adding. the four percentages and dividing by four.

Producer or contractor reaction to this type of specifica-tion has been favorable. Mississippi considers their speci-fications to be very satisfactory.

Missouri

Missouri has not used and is not planning to use ERS.

Montana

Montana has not used and has no immediate plans for ERS.

24

Nebraska

Nebraska has a statistically oriented acceptance plan for asphaltic concrete in their special provisions that has been used on 1971, 1972, and 1973 projects. The contractor must submit a job-mix formula, but target value for asphalt content is determined by the materials and test engineer. Control charts are maintained by the project engineer for bitumen content, No. 10 (2.00-mm) sieve, No. 200 75-gm) sieve, and difference between No. 50 (300-.im) and No. 200 sieves. Each content chart consists of three plots: individual test results, cumulative average-of-lot test results, and moving range of three most current test results. Control limits are specified for all charts.

The contractor is responsible for monitoring all charts and making adjustments if plots indicate that continued production would result in a reduction in price.

Lot size is 1,000 tons (900 metric tons) for the first lot and 2,500 tons (2 300 metric tons) for following lots. Normally, five samples are taken from each lot. Pay fac-tors of 1.00 to 0.70 are based on the average of 3, 4, or 5 test results and the amount of deviation from the job-mix formula for bitumen content and gradation. Density of compacted pavement has a pay factor of 1.00 when the average of five samples is 90 percent or greater than void-less density and 0.50 for 89.9 percent or less. Percent of contract price for pavement deficient in thickness is based on performance statistics developed at the AASHO Road Test. The range is from 100 percent for pavement deficient 3 percent of plan thickness to 64.8' percent for pavement 8 percent deficient.

Comments obtained by questionnaire from 29 engineers and 6 contractors indicate that contractors have no major difficulty with the statistically oriented acceptance plans. Cost is about $0.10 higher per ton ($0.11 per metric ton) or $3,000 to $5,000 per project, but greater latitude in mix formula sometimes may offset higher cost. Four of the six contractors believe that the disadvantages outweigh advan-tages, however.

There is no indication that bidding prices are markedly higher for contracts with the incentive provision. On one project the number of state personnel was reduced by one, but there is a difference of opinion regarding the possibility of the reduction. The amount of testing required is about the same, but it is extremely important that field personnel be well trained and skilled in testing. A materials and tests division study indicates that asphaltic concrete produced under incentive provisions is more uniform. Engineers with experience in the administration of incentive contracts favor their use. The Department of Roads plans to continue ex-perimentation with the use of the statistically oriented ac-ceptance plans with incentive provisions for asphaltic con-crete work and, possibly, expand their use.

Nevada

Nevada does not use ERS except for compaction of plant-mix bituminous pavement where acceptance is based on the average of five tests performed at randomly selected loca-tions within a test strip. The area of the test strips is ap-proximately 2,800 yd2 (2 300 m2 ) when the thickness of

the lift is 3 in. (75 mm) or more and approximately 4,200 yd2 (3 500 m2) when the thickness is less than 3 in.

The average of the five measurements must be at least 98 percent of the average density of a control strip. Indi-vidual measurements must exceed 95 percent. If required density is not obtained, compactive effort must be continued.

Control strips are of at least 400 yd2 (330 m2 ) in area. Maximum density is the average of ten measurements at random locations. All density measurements are made with a nuclear gauge. Reductions in price are specified for par-tial compliance in several areas. Liquidated damages are assessed on the degree of noncompliance in the areas of PCC pavement thickness, concrete cOmpressive strength, and asphalt products. No change has been noted in bid price, amount of testing, or number of inspection personnel.

New Hampshire

New Hampshire does not use ERS.

30. New Jersey

New Jersey has used ERS for bituminous concrete produc-tion for six years. The producer is responsible for the finished mixtures and the contractor is responsible for per-centage of air voids in the compacted pavement. The con-tractor is required to submit a job-mix formula along with three Marshall specimens having the proposed composition to be used to verify the maximum specific gravity, design air voids, stability, and flow for each asphaltic concrete mixture.

The producer is required to perform quality control test-ing to keep the mix within the specified tolerances. Av-erages of five measurements of the produced mixture must meet tabulated tolerances for average and range of per-centages passing sieve.

Acceptance testing of bituminous mixtures is to be per-formed by the producer's quality control technician under the supervision of the engineer.

Provisions are made for adjustments in contract price for lots not in conformance with requirements with respect to job-mix formula and range in asphalt content or aggregate passing the No. 8 (2.36-mm), No. 50 (300-pm), or No. 200 (75-sm) sieves. When more than one adjust-ment of contract price applies to a lot, only the greatest adjustment is used. Tables of reduction in price are shown for nonconformance with requirements for stability and air voids. Lot size is approximately 5,000 yd2 (4 180 m2). Compliance with thickness requirements is determined by measurements on 15 cores from each lot, five of which are taken from each of three sublots. A QL-value is calculated by the equation

QL (average thickness - thickness acceptance limit)

(17)

where k3 is the average of the three ranges of five mea-surements on each of the three sublots. The buyer's risk is about 5 percent. A table of reductions in price is pro-vided for lots having a QL-value of less than 0.36.

New Jersey has just recently begun to use computer

25

simulation to aid in the development and testing of ERS specifications. This technique enables the department to see how the specification will perform under a variety of conditions; potential problems can be spotted and cor-rected before the specification is adopted for field use. For example, if a reduced pay schedule is to be used, it is pos-sible to determine the over-all expected pay factor asso-ciated with any specific quality level of work. In this man-ner, a specification can be developed that is not only fairer to all parties concerned but one that is statistically defensible should the need arise.

As a result of work in this area, a training manual on computer simulation will be included as part of Federal Demonstration Project No. 42, "Highway Quality Assur-ance—Process Control and Acceptance Plans."

Producers and contractors have accepted this system with some reservations, particularly in the area of air void penal-ties. New Jersey believes that ERS has been successful. Sample testing by the central laboratory has been reduced about 60 percent by ERS.

New Mexico

New Mexico is not planning to use ERS and has no current or proposed research in this area.

New York

New York does not use ERS. The department made a survey of asphaltic-concrete density specifications in 1972-73 that included all 50 states and the District of Co-lumbia. At that time 84 percent of the agencies polled (43 agencies) indicated that they had density requirements for bituminous concrete. In addition, two others indicated they were experimenting with the use of such requirements. Of the 43 with density requirements, approximately half (21) either had or were considering the use of statistically based specifications and two-thirds of these 21 (13 agen-cies) had or were considering provision for reduced payment.

Of the 43 agencies that specified density, 63 percent (27) used or were considering using nuclear testing; 79 percent (34) based their requirements totally or partially on labora-tory densities; and 91 percent (39) tested totally or par-tially during construction. Only ten agencies made use of control strips in developing density requirements and only 17 tested for density after completion of paving.

North Carolina

North Carolina has a statistical acceptance specification for aggregate base course. Two random samples of each lot of 1,000 or 2,000 tons (900 or 1 800 metric tons) of aggregate are averaged to determine acceptability of the aggregate gradation; a reduced unit price is used for material outside of allowable tolerances. Random sampling of lots is also used for acceptance of aggregate stabilization and soil base course.

North Dakota

North Dakota has used ERS for three years. Acceptance samples for hot bituminous pavement are taken from haul-

ing units that are randomly selected. Normally four sam-ples are taken from each day's production and tested for asphalt content and gradation. Lots are accepted at an adjusted unit price when the test results deviate from the JMF outside the stated tolerances for four sieves and for asphalt content. Pay factors range from 100 to 90 percent.

The ERS has been generally well received by contractors. North Dakota is enthusiastic about the success of the specification for the following reasons:

It recognizes variability in materials and workmanship. It places the responsibility for quality of the final prod-

uct on the contractor—where, in the department's opinion, it rightfully belongs.

It establishes an incentive to do good work and pe-nalizes those who perform substandard work. This has the effect of eliminating subsidization of the contractor who historically has performed substandard work under the prescribed "methods and equipment" specification.

North Dakota has completed research on the variability of compaction of roadway embankments and has experi-mented with a statistical acceptance plan in which the lot is acceptable if

XrL+RQ (18)

where X is the average of density measurements at five locations determined by the use of random numbers, R is the difference between the largest and smallest measure-ments in the group of five test results, L is the target value for the percent compaction, and Q is the quality index. A value of 0.36 has been selected for Q, which in this case is equivalent to an estimated 80 percent of future tests ex-ceeding the value of L. The lot size is normally one day's production. Lots failing to meet the quality index require-ment must be reworked by the contractor.

This specification has not been implemented in practice. The current special provisions for quality control of hot bituminous pavement with respect to gradation and asphalt content of the paving mixture discharged into the truck specify a lot size as one day's production. Normally four samples are taken from each lot and the lot quality is estimated from the average of the four test results. The deviation of this average from the JMF is compared with tolerances shown in tables for the lot payment schedule. Provision is made for evaluating less than four tests. The tables, tolerances, and pay schedules are similar to those proposed by FHWA Region 15, but the tolerances for asphalt content are much wider than those shown in the Region 15 tables.

The lot size for compacted pavement is the number of tons placed in one production day on mainline pavement. Sublot size is 1,500 yd2 (1 250 m2 ) and one density test is normally made on each sublot. If an individual test result falls below the required minimum, the contractor is re-quired to recompact the sublot and another test is made at a random location. The average of this test result and of the original test result represents the density of the sublot. At least five tests must be made on each production day of mainline pavement regardless of the size of the lot. If

26

the average density of the mainline pavement does not equal or exceed the required percentage of Marshall density, the lot is accepted at an adjusted unit price.

Ohio

Ohio does not use ERS but uses ASTM C 183 for the sampling of cement.

Oklahoma

Oklahoma has not used and is presently not planning to use ERS. Several studies have been made indicating that ERS would not appreciably benefit the construction program.

Oregon

Oregon does not use a true ERS; however, acceptance of aggregate gradation is based on average values and a re-duced unit price is used for material outside of specified tolerances.

Pennsylvania

Pennsylvania has developed a statistically oriented specifi-cation for bituminous concrete that is presently a special provision to the standard specifications (Form 408). The contractor is responsible for the quality of the construction and the materials used. The contractor must provide and maintain a quality control system that provides reasonable assurance that all materials, products, and completed con-struction conform to contract requirements. Suggested guidelines for a contractor's control system are provided.

The contractor is required to submit a JMF and the gradation of the aggregates must be within the approved JMF tolerances. If three successive tests show that the percentage passing any sieve is outside the tolerance for that sieve the contractor must make appropriate adjust-ments or the engineer can refuse to accept further pro-duction of the completed mixture.

The bituminous mixture is accepted at the plant on a lot-to-lot basis. A lot is 3,000 tons (2 700 metric tons) divided into five approximately equal sublots. One random sample is taken from each sublot. If less than three sam-ples can be obtained from a partially completed lot, they are combined with those from the last lot. If three or more samples are obtained from a lot they are used as the num-ber of tests for that lot. Compliance with JMF tolerances for asphalt content are determined by computing the quality index (Q). A graduated scale of reduction in price is pro-vided for lots having less than 90 percent within tolerance as indicated by the quality index. One Marshall test (3 specimens) is required from each sublot. If three con-secutive tests for any property fail to meet specified re-quirements, the contractor is required to take immediate corrective action.

A lot of completed pavement is 5,000 linear ft (1 500 m) of paving lane or 6,700 yd2 (5 600 m2 ), whichever is less. Each lot is divided into five approximately equal sublots. One nuclear-density test is taken at a random location on each sublot; no test is taken within 2 ft (0.6 m) of the edge of the pavement. The quality index (Q) calculated from

the five tests must indicate that at least 85 percent of the material has been compacted to 96 percent of control-strip density. A table of graduated reductions in price is provided for lots not meeting this requirement.

This specification was used in six resurfacing projects as a special provision. Contractor reaction was generally fa-vorable. The price of bituminous concrete was not in-creased and the amount of testing has remained Constant except for an increase in nuclear-density testing.

Pennsylvania has designed a work plan for the imple-mentation of ERS given in Table 5.

Rhode Island

Rhode Island has a statistically oriented acceptance plan for portland cement that combines a requirement for a certifi-cate of compliance from the manufacturer with a random sampling procedure. Each shipment or truckload is sam-pled by a state materials inspector. The total number of shipments in a week is considered to be a lot; one sample selected by the use of random numbers represents the lot.

If the sample fails to meet the specified quality require-ments, two additional samples are tested and based on these results the lot is either rejected or accepted. If rejected, the degree to which the structure has been affected is assessed and a decision made as to the removal of the structure or an adjustment in price. The manufacturer is placed on a suspect list and cement from this source is frequently tested. If there is a second failure to comply with requirements, the manufacturer is not permitted to supply cement for a period of up to three months or until evidence is submitted that corrective action has been taken.

Rhode Island considers their specification to be successful.

South Carolina

South Carolina is currently using statistically oriented speci-fications for quality control and acceptance of hot-laid sand-asphalt base, and asphalt binder and surface-course mixtures. A lot is considered to be the tonnage produced each production day. Each lot is divided into five approxi-mately equal sublots. Normally five samples are taken from each lot. Provision is made for the use of a reduced num-ber of samples if necessary. Binder course materials are sampled from the truck at the plant. Sand-asphalt and surface-course mixtures are either sampled from behind the paver or at the plant. A detailed random sampling plan is used in either case.

Acceptance is normally based on the average of five test results; however, control limits and pay factors are pro-vided for a lesser number of tests. When the average of the day's production test results for a particular sieve or for asphalt content are outside the appropriate limits, an ad-justment is made in the unit price for the lot. The percent of bid price for the lot is determined from a table showing deviations from the JMF for I to 5 tests and percentages of bid price from 100 to 90. For lots that are out of tolerance on two or more sieves, the unit bid price is adjusted from the one that is out of tolerance resulting in the largest reduction. For lots that are out of tolerance on both asphalt content and gradation, the adjusted unit price is determined by multiplying the contract bid price

TABLE 5

PENNSYLVANIA WORK PLAN FOR IMPLEMENTATION OF STATISTICALLY ORIENTED END-RESULT SPECIFICATIONS

Winter Winter Winter Winter Winter Winter 1910 1971 1972 1973

73-74 74 14-15

1975 75-76

$916 16-77 $977

77-78 $979

78-79

Phase I Phase I Phase IA Phase IA Phase IA

SThTISTICAL Sel.cIed Elementary Elementary Elementary Elementary Elementary

TRAINING BMTR BMTR-Const. BMTR-Const. Districts Districts Districts

Districts Districts Schooi&C.r SCtIOO1&CIrI. Schcd8Cert

Follow-Up Follow-Up QUALITY Redraft Training

Training Training ASSURANCE InU By BMTR B Implement

BMTR BMTR - PROGRAM July let. Districts

Districts Districts

Gather Stat. Adjust Spec. Adjust. Spec. Phase III Phase III

BITUMINOUS Data 1967 Require

More Data Write let.

Field 6 Trial Phase II Trial Phase II implement

Follow-Up Follow-Up

CONCRETE Technicians simulatift Projects Training Projects Training Training Training & Test Draft

Districts Districts Districts Districts Equipment

PORTLAND CEMENT : 1967 Rewrite Spec Phase Ii Phase III Phase ill

CONCPETE Require lit Spec. Requite Gather 1st. Spec. Simulate Training Implement

Follow-Up Follow-Up

PAVNO A STRUCTURES Technicians

Draft Technicians,

Dale Draft In Field Districts Training Training

(INCuJDINO AGO) a Test Equipment

Equipment AgQ Pnts

Adjust. Spec. Districts Districts

D.vstopmenl Require iii. Spec. Trial Phase Ii Phase III Phase Ill

BASE COURSES New Test, Technicians Draft B Rewrite Training Follow-Up Follow-Up INCLUDING a Testing Field Spec.

Projects Districts

Implement Training Training

SUBBASE Equipmsnt Simulate Each 1 pe

Adjust. Spec. Districts Districts

Requite 1.1. Spec. Modify Adjust Spec. Phase lit

Technicians Draft & Rewrite 6 Trial Spec. For. 6 Trial Phase II Implement

Follow - Up SOIL 9 Testiml Field Spec. Projects Nest Trial Projects Training Training

Equipment Simulate Project Districts District

Trial Trial lit Spec. Phase ill Phase ill - PAVEMENT

SMOOTHNESS Eieluation8Ohtalnlng DistrIct Use District use Draft

Phase Ii Implement Follow-Up Follow-Up

Equipment Equipment Equipment Training Training Training ALL TYPES Only Only Districts Districts Districts

PC Gather Data Field Field Simul. Final Spec. Phone II

Phase ill Training

Phaae Ili Training NCHRP

PAVEMENT - - - - Write Spec. Simulation Rewrite Spec. Training_ - - - Districts- Districts - - - DEPTH Gather -

lit. Spec. Simulate Aipec. - - - sflll

tistiowi Draft Field Phase Ii implement Follow-Up

BASES - Data Training Training

Phase I - Elementary Statistical Training Course Phase IA - Elementary Statistical Training Course - Inspector level Phase II - Statistical Course And Application New Spec.- Thai Materiel . Phase ill - Follow-Up Training New Spec. —4

28

in series by the percentage of payment for asphalt content and the least percentage of payment for gradation.

Contractor reaction to the specification has been favor-able. Neither the number of inspection personnel nor the bid price has been increased. The amount of testing may be slightly greater.

South Dakota

South Dakota has not yet used ERS. A study of quality control of base course by statistical methods was conducted in 1966, but there has been no follow-up on the findings.

Tennessee

Tennessee does not use ERS or statistical quality control.

Texas

Texas has not used and has no immediate plans to use ERS.

44. Utah

Utah has a modified form of ERS for base course and plant-mix seal coat. Acceptance of base course aggregate with respect to gradation is based on the average of five gradation tests each made on a random sample from a lot consisting of the number of tons placed each production day. If the average of the percent passing a designated sieve is outside the specified tolerance, a reduction in price is effected.

Target density for compaction is determined by construc-tion of control strips. Average density is based on 10 nuclear measurements made at random locations. Test lots are equal to the number of tons compacted each production day. Lots are divided into sublots of about 1,600 yd2 (1 300 ma). One density test is made at a random loca-tion within each sublot. In order for the material to be accepted, the average of all density tests made on a lot must be not less than 98 percent of target density and no indi-vidual test can be less than 95 percent. Acceptance of plant-mix bituminous seal coat with respect to gradation is based on the average of the results of five gradation tests. The five tests are taken on a random basis from a lot con-sisting of the number of tons placed each construction day. If the average of the percentage passing each sieve is out-side stated tolerances, the lot is subject to a reduced pay factor. Control charts are maintained by the engineer.

Utah has added an ERS for concrete pavement in which statistical methods are used for acceptance of thickness and of compressive strength.

Vermont

Vermont adopted a supplemental specification for quality assurance for paving in 1972 that has been used on proj-ects of at least 25,000 tons (23 000 metric tons) of bi-tuminous concrete pavement. The engineer maintains con-trol charts for bitumen content and for each sieve desig-nated in the JMF. Each control chart consists of three plots: the individual test results, the cumulative averages

of the day's test results, and the moving range of the three most recent test results. The control chart central lines are the target values of the JMF; the control chart limits are given in a table that is a duplication of that published in reports of FHWA Region 15. The acceptance schedule and pay factor table for the adjusted unit price for ma-terial having test values outside of tolerances has also been adopted from Region 15.

The lot size is one day's production; five random samples are normally taken from each lot.

The density of each day's production of compacted pave-ment is determined from the average of all tests made with a portable nuclear-density measurement device. Not less than five tests are made at random locations within sublots of approximately 400 yd2 (330 m2 ) of paving. Target density is determined from control strips of approximately 400 yd2. The average density of the control strip as de-termined from core samples must be at least 95 percent of laboratory 50-blow-compacted Marshall specimens. If the average of the density tests made during a production day does not equal or exceed 98 percent of target density, the adjusted unit price for the material is based on a pay factor of 0.92.

Surface roughness of the pavement is determined by the use of an FHWA roughometer or Mays Ride Meter. If the roughness exceeds given tolerances, a pay factor is deter-mined from a table. If more than one pay factor for grada-tion, bitumen content, compaction, or surface tolerance, is less than one, the total quantity of material produced that day is successively multiplied by all factors to determine an adjusted tonnage. Payment is made using the adjusted tonnage as the accepted quantity at the contract unit price per ton.

Producer reactions are generally favorable. Previous ex-perience does not indicate that the program has affected the bid price to any appreciable extent. An additional employ-ee has been hired to perform nuclear density tests. In general, an increase in quality of pavement has not been experienced. Department representatives point out that their program lacks control because of the 100 percent payment on materials that varies both positively and nega-tively from the JMF when mathematically averaged.

Virginia

For several years, Virginia has been using ERS in the special provisions for subbase and aggregate base courses and for Section 212, bituminous concrete (41). These specifications focus on producer quality control and con-tractor responsibility.

The specifications for subbases and bases include tol-erances for percentages passing individual sieves and for Atterberg limits. Material is sampled and accepted at the central mixing plant. Lot size is normally 2,000 tons (1 800 metric tons) and acceptance is passed on the av-erage of the results of four tests on each lot. Provision is made for the use of a reduced number of tests if necessary. Adjustment of unit bid price is based on points, depending on each one percent that the gradation on a particular sieve is out of tolerance and on the percentages that the liquid limit and plasticity index are out of tolerance. A similar

29

adjustment system applies to the Atterberg limits. If the total adjustment for a 2,000-ton lot is more than 25 points, the failing material must be removed from the road. If the total adjustment is 25 points or less, the unit price is re-duced one percent for each adjustment point for the ton-nage represented by the sample or samples. Individual test results and lot averages are plotted on control charts. Standard deviations are computed from the total number of test results; adjustment points are assigned if the varia-bility exceeds tabulated values.

The bituminous-concrete specification requires that the contractor furnish a job-mix formula, suitable field labora-tory, and listed testing equipment. Material is normally sampled, tested, and accepted at the plant by agency per-sonnel. Lot size is normally 2,000 tons (1 800 metric tons), but may be increased to 4,000 tons (3 600 metric tons). Acceptance for gradation and asphalt content is based on results of four tests on stratified random samples. Pro-vision is made for the use of a reduced number of samples if necessary. The average of the tests is required to be within stated tolerances applied to the JMF. Different tol-erances are shown for extraction tests and for asphalt con-tent documented by records. Adjustment of the unit bid price is based on adjustment points assigned to test aver-ages outside of tolerance for percentages passing particular sieves and for asphalt content outside the stated tolerance.

If the total of the adjustment points is more than 25, the material must be removed from the road. If less than 25, the unit bid price is reduced one percent for each adjust-ment point. The standard deviation of the total number of test results is computed and adjustment points are assigned if the variability exceeds stated values. Individual test re-sults and lot averages are plotted on control charts. A referee system is provided in case of a dispute (i.e., if one of the four samples taken from a lot appears to be ques-tionable, the contractor or the engineer may request that this sample be disregarded; five additional random sam-ples are then taken and the test results of the eight samples are averaged).

If the contractor questions the average of the four origi-nal test results, additional testing may be requested. In this case, four additional random samples are taken from the lot and the eight test results are averaged.

In either case the resulting average is compared with the process tolerance for the average of eight tests. This is equal to the process tolerance for the average of four tests divided by 1.4. If the average of the eight test results is outside the process tolerance for the average of eight tests, the unit bid price is adjusted and the contractor pays for the additional tests at the rate of five times the bid price per ton of material per sample.

Both Virginia and its producers are satisfied with the specification.

Washington

Washington does not use ERS or statistical quality control

48. West Virginia

West Virginia has fully implemented ERS in the special provisions for two demonstration projects. One project in-

cluded concrete and embankment and the other included hot-mix and crushed-aggregate base course. Contractors were required to provide and maintain a quality control system and were entirely responsible for all materials and construction. The contractor's quality control requirements were specified. Lots of hot mix consisted of four sublots. Each sublot was approximately 400 tons (360 metric tons) of mixture. Acceptance was based on one sample from each sublot taken from a randomly selected truck at the plant.

Based on the four random samples the percentage of extracted aggregate passing each specified sieve must be within the specified tolerance at least 80 percent of the time when determined by West Virginia Statistical Acceptance Plan A. If this requirement is not met, production is dis-continued until corrections are made by the contractor.

Based on the same four random samples as used for the gradation test, the asphalt content of individual samples must be within ±0.40 percent of the plant-mix formula (PMF) and the average of results of tests on four samples must not differ from the PMF by more than 0.6 - 0.45R. Lots not meeting the latter criterion are subject to a graduated reduction in price.

Each completed layer or course of asphalt pavement is tested for acceptance with respect to compaction. A lot is one day's construction and is divided into five sublots. One random sample is taken from each sublot and the percent within tolerance is determined by Acceptance Plan A. Target density is obtained by the construction of a control strip. Acceptance is based on West Virginia Acceptance Plan A.

Smoothness of completed pavement is determined with a BPR-type roughometer or the equivalent. Pavement hav-ing a BPR roughness of less than 81 in/mile (1 300 mm/ km) is accepted at 100 percent of contract price. Contract price is reduced in graduated steps to 70 percent for 114 in., mile (1 800 mm/km). Pavement with roughness greater than 114 must be overlaid or replaced.

Pavement is accepted with respect to thickness in lots of approximately 5,000 linear ft (1 500 m) of each two lanes of construction. Each lot is divided into five approximately equal subkts and one core is taken from each sublot at random locations. The pavement thickness is considered acceptable if both of the following conditions are met:

The average of the five cores is equal to or greater than the specified thickness; and

When determined by West Virginia Acceptance Plan A, at least 80 percent of the item has a thickness greater than 85 percent of the specified thickness.

If either of these conditions is met, an additional two cores are taken from each sublot and these cores together with the original five cores are evaluated. The pavement thickness is then considered acceptable if both of the following conditions are met:

The average thickness of the fifteen cores is equal to or greater than the specified thickness; and

When determined by West Virginia Acceptance Plan

30

A, at least 80 percent of the item has a thickness greater than 85 percent of the specified thickness.

If either of both of these criteria are not met, the entire lot must be resurfaced.

West Virginia has had several years' experience with the implementation of statistical end-result specifications in sup-plemental specifications for crushed-aggregate base course, hot-laid bituminous concrete, portland cement concrete pavement, and structural concrete. There is an incentive provision for portland cement concrete in the form of a possible reduction in cement for concrete having a com-pressive strength equal to or greater than the design strength plus a statistically derived value.

Wisconsin

Wisconsin does not use ERS. ERS specifications for bi-tuminous surfacing have been simulated on two projects, however. The specifications were found workable but would require increased testing and manpower under the simulated conditions. Contractors are reluctant to use ERS.

Wyoming

Wyoming has drafted ERS requirements for special pro-visions essentially the same as those specified by Alaska (Eqs. ha and lib) and Colorado (Eqs. 12a and 12b) and has let two contracts for construction in the current season. Acceptance of lots is based on the equation

or

in which

P = percent of reduction in contract price; = average of all the test values from samples taken

from the lot; R = difference between the highest and lowest values in

the group of test results from the lot; Tu = upper or maximum tolerance limit permitted by

the specification; T1, = lower minimum tolerance limit permitted by the

specification; a= Q-value for 78 PWT (Ref. 19); and F = price reduction factor.

F-factors were derived as follows: A distance outside the specification limits was established for each element be-yond which the material was considered to be unacceptable. This point was also considered to be the 25 percent price-reduction point. This distance outside the specification limits plus the estimated ½ R for the element were then divided into 25 percent to obtain the factor. Although a price adjustment could theoretically be applied when all values are within specifications, the special provision pro-vides that the work or materials not be evaluated for price adjustment except when deviations from specifications oc-cur on one or more of the individual tests for the lot.

This specification assumes a buyer's risk of about 7 per-

cent of accepting measurements with an actual value of = TU or TL when n = 5.

Lots normally consist of five sublots. The size of sublots for different materials is given in a table. The F-factors are computed from a base that is equivalent to the mean of unacceptable material and are assigned tabular values for different types of measurements.

Bituminous mixtures are sampled by taking a portion from each of the four quarters of a loaded truck. The portions are mixed and the test portion is obtained by quartering or by the use of a sample splitter.

All sampling is done on a random basis.

FEDERAL HIGHWAY ADMINISTRATION

The Federal-Aid Highway Program Manual (Vol. 6, Ch. 4, Section 2, Subsection 10, dated 5-27-75) calls for the estab-lishment of a program to attain the widespread use of quality assurance techniques in highway construction by 1980. Modern quality assurance includes the activities of process control and acceptance sampling and testing and assures through systematic evaluation that the production process control job is being done effectively.

The Office of Research and Development launched a Quality Assurance Program in 1963. In recent years, there has been more emphasison developing statistically based specifications, rapid testing methods, and quality-assurance training materials. Current implementation of the Quality Assurance Program is designed to cover four areas: pro-motion, training, demonstration and trial applications, and implementation and use.

The National Highway Institute offers a pilot course on statistical quality control of highway construction. It pro-vides the states with a basic curriculum with which to implement in-house quality-assurance training for both industry and state personnel.

PORT AUTHORITY OF NEW YORK

The Port Authority of New York and New Jersey does not presently use ERS, but is gathering information on con-crete suppliers in order to obtain data that will establish a realistic coefficient of variation for concrete plants in the New York area. Plans are to use this data in preparing a statistically based concrete specification for both highway and building concrete.

It is anticipated that obtaining the data will require one year. During this period the New Jersey Department of Transportation experience will be observed.

FOREIGN TRANSPORTATION AGENCIES

1. Alberta, Canada

The Alberta Transportation Department reports that their construction specifications are end result, but that the mea-surement of this end result is not on a statistically oriented basis. Some quality-control test procedures lend themselves to a statistical approach; however, the sample size, etc. is not included as part of the contract specification.

A possible exception is the procedure for measurement and payment for portland cement concrete pavement. Ma-

31

jor pavement lanes of specified thickness are divided into units 3,000 yd2 (2 500 m2). An initial random sampling of five cores is taken from each unit and if the average of these cores is deficient in thickness by more than 0,3 in. (7.6 mm), five additional random cores are taken. The average of the ten cores, subject to some restrictions, is the basis for a table of graduated reductions in price.

Manitoba, Canada

The Manitoba Department of Highways does not at present use statistically oriented specifications. Statistics are used in the development of specifications and for control pur-poses but not as a part of the contracts. The objection to ERS is that it requires more technically trained field per-sonnel. A shift to ERS would probably result in a shift of technical personnel from the agency to the contractors and would require some modification of the specifications with respect to payment. Some movement in this direction is anticipated but not for some time.

New Brunswick, Canada

New Brunswick does not have ERS and has decided to retainits current system for the time being.

Nova Scotia, Canada

The Nova Scotia Department of Highways has not used ERS and does not have plans to do so in the foreseeable future.

Ontario, Canada

Ontario has not tried ERS and is not planning to do so in the immediate future.

Saskatchewan, Canada

Saskatchewan highway specifications are currently a mix-ture of end-result and procedural requirements but tend toward end-result specifications. There is little use of the statistical approach except for the use of averages in a few instances; there are no incentive provisions.

An investigation has been made of the normal variability of asphalt content and gradation percentages of asphaltic paving mixtures with a view toward developing realistic tolerances. A computer program has been used to produce a histogram and control-chart data of test results.

South Africa

South Africa has quite comprehensive statistically oriented ERS developed by the National Institute for Road Re-search. Features include definite sample sizes of from 4 to 8 test portions, normal lot size of a day's production, sub-lots equal to the number of test portions, random sampling by use of random number tables, statistically derived ac-ceptance and rejection limits, standardized buyer's and seller's risks, and provision for reduced payment for condi-tional acceptance with payments determined by a mathe-matical formula.

To determine whether or not a lot is acceptable, the mean test result (,) and the standard deviation (s) of the test results for the particular lot are calculated. In the case of a characteristic with a lower specification limit (L8 ) the following limits are calculated:

Acceptance Limit L:t = L8 + K8 's (20a)

Rejection Limit L r = L8 + Kr 'S (20b)

L, ACCEPT

Lr < L, CONDITIONALLY ACCEPT

X,, <Lr REJECT

The contractor's risk of being wrongly assessed has been standardized and is equal to five percent at the acceptance limit and one per cent at the rejection limit. When the sample mean falls within the conditional acceptance range, the contract bid price is multiplied by the following reduc-tion factor:

Lower Limit Specification

S F - - 50

[L 1 - 2L. + (21)

X

In the specifications a table is given that specifies the value of L. (and/or L',), the number of samples (n), as well as the constants (K8 and Kr) for each characteristic. The value s is equal to the actual sample standard deviation as determined from the test results on that particular lot.

On some contracts the value of F is specified as F = 0.90 for relative density and bitumen content and 0.954 for thickness of asphalt surfacing. When two or more charac-teristics have payment reduction factors, the factors are multiplied consecutively to determine the payment reduc-tion factor to be applied to the lot.

As of June 1975, two contracts were being controlled by ERS and three other contracts were being prepared. A 6 percent increase in bid price for asphalt paving has been the only increase in bids. This is believed to be due to the past practice of virtually always accepting the product at full price even if the test results showed the material should be rejected.

Clients and consultants using ERS have liked it. Con-tractors are a bit hesitant, mostly because of a lack of understanding and uncertainty as to how the system will affect their profit margin.

A laboratory program has been set up to improve the reproducibility and repeatability of the test results. The main impediment is the lack of rapid nondestructive test methods as well as the lack of tests with lower testing variabilities.

Netherlands

The Netherlands Highways and Waterways Administration has an end-result specification (1975) that requires the contractor to ensure that the materials used for the prepa-ration of asphalt, sand-cement, or surface dressings fulfill applicable requirements by examination or by a certificate of origin. The contractor must determine the gradation of each consignment of aggregate, the penetration of the bitumen, and the bitumen rating of mineral fillers. The

32

composition of the paving mixture to be used must be deter-mined by the contractor before the start of production; the composition of the mixture that meets Marshall require-ments must be recorded in writing. The contractor must maintain routine checks of the thickness of layers, the degree of composition, percent air voids, and Marshall values. Contractor records must be available to the client. The evenness of the pavement should be checked regularly with a 3-m (10-ft) straightedge. The contractor must take two cores for each 2 000-rn2 (2,500-yd2 ) section and five cores taken from each section at points designated by the client.

Tests of evenness and skid resistance are made by the client; tests for thickness, bitumen content, percent air voids, and degree of compaction are made on the cores. Acceptance is based on the results of tests made on one core or on the average of results of tests on two associated cores. A table of graduated penalty reductions in terms of guilders per 2 000 m2 (dollars/2,500 yd2 ) is provided for deficiency of thickness and for deviations greater than and less than specified limits for asphalt content and percent voids. Deficiencies in evenness and skid resistance must be corrected. The contractor must guarantee the pavement for a period of three years.

CHAPTER SEVEN

COMMENTS FROM TRADE ASSOCIATIONS AND PRODUCERS

To obtain the reactions of industry to statistically oriented end-result specifications comments were requested from:

American Concrete Pavement Association (ACPA); National Asphalt Pavement Association (NAPA); National Crushed Stone Association (NCSA); National Sand and Gravel Association (NSGA); National Slag Association (NSA); and Lone Star Industries, Inc. (Producer) (LSI).

The comments received have been abridged and are iden-tified by the acronym of the organization. The abridged comments do not necessarily reflect the policy of the organization or of their. individual members.

Question 1. Should the contractor or producer assume entire responsibility for the end product?

ACPA - Yes, provided that equipment is developed for rapid testing and acceptance, that political influence is not a factor, and that restrictive specifications are removed.

NAPA -Yes, provided that decisions are made on a lot-to-lot basis within 24 hours so that the contractor can adjust the process in time to do some good. Contractor's process control tests should be specified so that all bids are based on the same set-up.

NCSA - Membership comments are not available; in our estimation, however, contractor responsibility is not forthcoming for a long time.

NSGA - This is a desirable objective that should lead to numerous improvements. There are a great number of instances, however, in which this would be very difficult if not impossible to accomplish. The concrete industry is made up of many small producers who may never have the ability to justify the expense of quality control programs.

The 1971 Bureau of Census survey indicates that 20 percent of the companies employ less than 4 people, 40 percent less than 8, 73 percent less than 20, and 93 percent less than 50.

NSA - It is believed to be impractical for the contrac-tor or producer to assume the entire responsibility for an end product.

LSI - Statistically based quality limits that permit us to guarantee meeting the customer's specifications have been in effect for many years.

Question 2. Should statistically derived acceptance limits that allow for a small proportion of low test results due to normal variation be used?

ACPA - Yes, provided that the statistical specification is based on acceptable historical data under present con-struction methods and designs and that the measurements have normal distribution.

NAPA - Yes, provided that the terminology is "toler-ances which control variability within acceptable limits."

NCSA - Yes. This type of acceptance limit is definitely supported.

NSGA - Yes. In general the industry does support this type of acceptance limit. The principal problem is that sophisticated plans that attempt to control consumers' and producers' risks are not widely understood. The key to the problem is education.

NSA - Yes, because variation is inherent in any prod-uct. The amount of variation permitted should vary in-versely with the price of the product or construction.

LSI - LSI is not a specifying agency.

Question 3. Should there be an adjustment in price for a greater than normal nu,nber of failing tests or for ,naterial not in full compliance with specified requirements?

33

ACPA -Yes; however it should be reasonable, be re-lated to pavement life, and possibly increase within a reasonable limit for superior work.

NAPA - Yes, provided that adjustments for small vari-ations are small and yet enough to provide incentive to the contractor.

NCSA - Yes; however, adjustments in bid prices must be made with care. The penalty should be severe enough to prevent a contractor from substituting cheap local ma-terial when the intended ingredient is a quality processed material.

NSGA - No. Penalty clauses are not supported by ready-mixed concrete producers. The fact that the penalty is based on the in-place bid price, which may be five or ten times the cost of the concrete as delivered, is a serious matter for the concrete producer. Tests made by other than the producer may not be made in accordance with all of the requirements of standard methods.

NSA - No direct comment. LSI - No direct comment.

Question 4. What is the probable cost-effectiveness, par-ticularly with regard to the increase in bid price, when the contractor or producer is required to make all process con-trol tests to ensure that material or construction will meet specifications when submitted for acceptance?

ACPA - This depends on the interpretation of the re-quirements. On one project, the bid price for quality con-trol had a range of 500 percent, apparently because of a lack of understanding of what was required. Contractors with experience say that process control costs them very little after they become 'familiar with it; initially, however, they believed that the cost was high. Costs will come down with experience and time.

NAPA - In the initial phases (probably four to five years) there will be no reduction in costs. Bid prices will increase to reflect costs of process control testing. Auto-mation of process control testing may eventually reduce cost to less than testing by an agency. When restrictive material, equipment, and construction specifications are re-moved, some ingenuity may produce reductions in material and production costs that will be reflected in bid prices.

NCSA - If contractors or producers are required in all cases to equip themselves to perform all process control tests, the ultimate cost may be greater. However, if the states or similar agencies furnish the equipment that might be used on a number of jobs (as, for example, in aggregate plants in Virginia), the ultimate cost would probably be reduced.

NSGA - Statistically oriented specifications and penalty clauses would probably increase the bid price of ready-mixed concrete by a substantial amount. Balanced against this would be the reduced cost of testing and inspection. In addition, there would be improved product quality and a decrease in the cost of settling disputes that arise under current specifications. If the proposed changes can be shown to reduce costs they would be accepted by the industry and by specifiers.

NSA - It is believed that, as in the experience of most

states moving toward statistically based specifications, the initial bid prices would be inflated until the contractor or contractor-producer realizes that minimizing the variation of his product or process minimizes penalties. A decrease in bid prices could also be expected when the contractor becomes fully cognizant of the specifications, thereby gain-ing a certain edge over competitors.

LSI - The end-result specifications applied by agencies in our area include a demand for excess testing by the producer and contractor. Some of the tests specified for acceptance of materials are not suitable for use in produc-tion control.

Question 5. Has your association conducted or sponsored training courses in statistical terms and concepts?

ACPA - Yes. Training courses on statistical specifica-tions are conducted; however, contractors are not often aware of the need for that training. This is realized only when the lack of it hits them in the pocketbook.

NAPA - No; however, the association has prepared and distributed two reports intended to familiarize members with statistical terms and concepts.

NCSA - No, however, sales marketing courses attempt to explain the concept of statistically oriented specifications, standard deviations, coefficients of variation, and similar terms.

NSGA - Yes. Short courses for industry teach compu-tation of standard deviation and its application to computa-tion of overdesign (for compressive strength of concrete, generally along the lines of ACI 318, ACI 301, or ASTM C 94). There is no statistical training course exclusively devoted to these concepts.

NSA - No training courses are conducted and none are planned for the immediate future. Training in the use of statistical quality control is included in orientation/training seminars, however.

LSI - Not applicable to producer.

Question 6. Do you train technicians in making physical tests and computations?

ACPA - Yes. Training courses for technicians are conducted.

NAPA - No, however a series of slide-tape presenta-tions for use in training laboratory technicians in process control testing has been prepared that includes:

Duties of a technician; Sampling and testing aggregates; Job-mix formula and blending aggregates; Specific gravity of fine and coarse aggregates; Setting up the cold feed; and Setting hot-bin weights.

A supplement on testing finished products is being prepared. NCSA - No training courses are conducted. NSGA - Yes. An extensive training and certification

course for industry technicians is conducted that covers general concrete technology and includes tests and certifica-tion as a concrete technician or concrete production tech-

34

nician by the National Ready Mixed Concrete Association. NSA - Yes. Periodically instruction is provided on the

proper techniques for physical testing required for use with statistical concepts.

LSI - Not applicable to producer.

Question 7. What is your association's general assessment of the success of end-result specifications?

ACPA -The success to date is marginal because of a lack of education and frequent misunderstanding by con-tractors. Those who understand end-result specifications are not communicating to the people on the job who are actually doing the work. In addition, some specifications are not realistic. Both contractors and specifiers must work together to develop good statistical specifications.

NAPA - Member reaction varies depending on the na-ture of the specifications.

South Carolina uses a statistical specification but per-forms all the process control testing as well as acceptance testing. The members there consider it successful.

In New Jersey, where the producer performs acceptance testing, there was initially some resistance and resentment. There is general acceptance now, however. Member re-action seems to be mixed; some consider it a success, others do not.

Louisiana is the only state that has operated a true end-result specification for several years. Member reaction generally seems to be positive.

NCSA - No direct comment. NSGA - It is not believed that in the ready-mixed con-

crete area this particular specification has been widely ac-cepted or particularly successful. For certain other types of material, such as bituminous mixtures, acceptance ap-pears to be more easily accomplished.

NSA - This type of specification has much merit; how-ever, major problems are expected with respect to relation-ships among the user, contractor, and aggregate supplier. Mishandling of aggregate by the contractor, over which the aggregate supplier has no control, can result in contamina-tion with impurities and segregation that can affect the quality of the end product and perhaps result in penalties. Even though the supplier may keep detá'tled records of the quality of the material furnished, it is our opinion that many legal actions will develop. This is a major weakness of end-result specifications as far as the aggregate industry is concerned. It would seem that conditions similar to this

will result also with other materials furnished to a project and that this question must be dealt with before end-result specifications will be workable.

LSI - Not applicable.

Question 8. Does your association have research proposed or under way that is related to this type of specification?

ACPA - No research is proposed or under way. NAPA - No research is under way. In our long-range

program, we have a task to develop the details of a typical quality assurance program (e.g., type and number of tests to be conducted, number of personnel needed and their skills, testing equipment needed).

NCSA - No direct comment. NSGA - No direct comment. NSA - No direct comment. LSI - Not applicable.

Question 9. Is any particular research needed?

ACPA - The method of measuring pavement depth by cores needs investigation. Pavement depth does not have a normal distribution. This should be examined in more detail in addition to the method by which the core samples are taken.

NAPA - The research needed to establish end-result specifications for hot-mix asphalt pavement has been done. Implementation is all that is required if a state wants to adopt end-result specifications.

NSGA - The Federal Highway Administration quality assurance-quality control demonstration program with ten-tative guide specifications has been helpful and is a neces-sary part of the implementation of any new specification. Pennsylvania's program is exemplary; it includes (a) sta-tistical training for both the highway department and pro-ducer personnel, (b) model specifications, (c) historical statistically oriented data, and (d) phased implementation of the whole system (including field trials and provision for re-evaluation before final specifications are issued). The development of rapid test methods and field trials of the proposed methods are needed.

NSA - No direct comment. LSI - A study needs to be made to deterimne the best

practical tests for materials acceptance. The study should include a look at the relationship of control test results and finished product quality.

GLOSSARY

ACCEPTANCE LIMIT (L OR U). A number that is the largest or smallest value of measurements of a characteristic of acceptable material.

ACCEPTANCE PLAN. An agreed-upon method of taking and

making measurements on a sample for the purpose of determining the acceptability of a lot of material or construction.

ALPHA (ci) RISK. See SELLER'S RISK.

35

APPARENT SPECIFIC GRAVITY. The ratio of the weight in air to the unit volume of the impermeable portion of a solid material.

ATTRIBUTES SAMPLING PLAN. A plan in which quality is measured by noting the presence or absence of some characteristic (attribute) and counting how many units do or do not possess this characteristic.

BETA ($) RISK. See BUYER'S RISK.

BUYER'S ($) RISK. The risk of accepting unacceptable material or construction.

CERTIFIED TECHNICIAN. A technician certified by some agency as proficient in performing certain duties.

CHARACTERISTIC. A measurable property of a material, product, or item of construction.

COEFFICIENT OF VARIATION (V). The ratio of the vari-ability of measurements about an average (expressed as the standard deviation) to the average about which the variation occurs (expressed as a percent).

o-(lOO)

x CONTROL STRIP. A short section of a pavement course that

has been compacted to the highest density obtainable with the equipment used for the rest of the construc-tion of the course. The maximum density obtained is used as a reference when determining the density of the rest of the course.

CRITICALITY. The classification of various factors of speci-fications by the degree to which they affect safety, performance, or durability.

DATA. Measurements collected for a planned purpose and suitable for the inference of conclusions.

DEFECT. A failure to meet a requirement with respect to a single quality characteristic.

DOCUMENTATION. Proof in the form of detailed records or charts supporting the effectiveness of a quality control system.

END-RESULT SPECIFICATION. A specification that places the entire responsibility on the contractor or producer for supplying an item of construction or material of specified quality.

GRADUATED REDUCTION IN CONTRACT PRICE. Reductions that are small for small deviations from specified requirements but increase rapidly as the size of devia-tions increases.

INCENTIVE PROVISION. A table of reductions in price for material or construction not in full compliance with specified requirements. In some cases, an increased price is included for material or construction exceed-ing requirements.

JOBMIX FORMULA (JMF). The percentage of each ma-terial in a mixture intended for a particular use; may include mixing temperature of bituminous mixtures.

LOT. An isolated quantity of material from a single source; a measured amount of construction assumed to be produced by the same process.

MAXIMUM DENSITY. The highest density that can be ob-tained under stated conditions.

MULTIPLE DEFICIENCIES. More than one defect in a unit of product with respect to the characteristic(s) under consideration.

n. Symbol for the number of measurements in a group or subgroup.

NORMAL DISTRIBUTION, STUDENT'S t DISTRIBUTION, and NONCENTRAL f DISTRIBUTION. Arrangements of data described by mathematical formulas that give the frequency of occurrence of individual measurements.

OPERATING CHARACTERISTICS CURVE (oC CURVE). A graphic presentation of a sampling plan showing the relation-ship between the quality of a lot and the probability of its acceptance or rejection.

PERCENT WITHIN TOLERANCE (PWT). The estimated per-centage of measurements that can be expected to fall above a lower limit, beneath an upper limit, or between upper and lower limits.

POPULATION STANDARD DEVIATION (o OR o'). The standard deviation of all possible measurements of a particular characteristic of a large amount of material.

PROBABILITY FACTOR. A tabular value which, when multi-plied by the standard deviation of a group of measure-ments, indicates the frequency or probability of oc-currence of measurements within a specified range.

Q TABLE. A table of quality indices (Q) related to per-centage of measurements exceeding a stated limit when Q is computed from groups of measurements of dif-ferent sizes.

QUALITY ASSURANCE. The activities that have to do with making sure that the quality of a product is what it should be.

QUALITY CONTROL. The activities that have to do with mak-ing the quality of a product what it should be.

QUALITY CONTROL SYSTEM. The over-all system that en-sures a product of specified quality including docu-mentation supporting its effectiveness.

QUALITY INDEX (Q). A derived value that indicates the estimated percentage of measurements that will exceed a stated limit.

RANDOM LOCATIONS. Sampling locations determined by the use of a table of random numbers.

RANDOM NUMBER. A number selected entirely by chance as from a table of random sampling numbers.

RANGE. The difference between the smallest and largest measurement in a group of measurements.

REALISTIC VALUES. Values that can be expected to be obtained under normal conditions.

REPRESENTATIVE SAMPLES. Samples taken on the basis of the judgment of the sampler that they represent some level of quality.

SAMPLE STANDARD DEVIATION (s). The standard deviation of a small group of measurements made on samples or portions of a sample.

SECOND SAMPLES. A sample taken when the initial sample indicates that the material is defective.

SELLER'S (a) RISK. The risk of having acceptable material rejected.

STANDARD DEVIATION (a- OR s). A measure of variability that can be calculated from the differences between individual measurements in a group and their average.

36

STATISTICAL CONTROL CHARTS. Graphical charts with sta-tistically derived control limits and plotted values of measurements for a series of measurements or the averages of groups of measurements.

STATISTICALLY DEFENSIBLE ACCEPTANCE DECISIONS. De-cisions based on data obtained by random sampling. Such decisions have been evaluated by the use of statistical procedures.

STATISTICALLY RANDOM SAMPLING. Sampling at times or locations determinel in advance by the use of a table of random numbers.

STATISTICS. The science that deals with the treatment and analysis of numerical data.

STRATIFIED RANDOM SAMPLING. Random sampling of sub-lots.

SUBLOTS. Subdivisions of a lot. TABLE OF RANDOM NUMBERS. A table arranged so that

every digit has an equal chance of occurrence. (tee) TEST. A method of testing a hypothesis regarding

means. Usually used to decide whether or not there is a meaningful difference between two average values.

TEST STRIP. See CONTROL STRIP.

TYPE I ERROR. A decision made on the basis of sampling to reject a lot of acceptable quality.

TYPE II ERROR. A decision made on the basis of sampling to accept a lot of unacceptable quality.

UNBIASED. Not influenced by opinion or judgment. UNBIASED RANDOM SAMPLES. Samples taken at times or

locations chosen by a method not influenced by opinion or judgment.

VARIABLES SAMPLING PLAN. A sampling plan based on sample measurements such as the average or average and standard deviation of the sample.

VOIDLESS DENSITY. The density obtained mathematically by dividing the weight of a unit volume of material by its specific gravity.

Is greater than or equal to. Is less than or equal to.

a < x < b means x is greater than a but less than b.

REFERENCES

I. "Control of Quality of Ready Mixed Concrete." Pub-lication No. 44, NatI. Ready Mixed Concrete Assn., 3rded. (1953). ODASZ, F. B., JR., and NAFUS, D. R., "Statistical Quality Control Applied to an Asphalt Mixing Plant." Proc. Assn. Asphalt Paving Tech., Vol. 23 (1954) pp. 78-96. "Sampling Procedures and Tables for Inspection by At-tributes." Military Standard, MIL-STD-105D (Apr. 29, 1963). "Sampling Procedures and Tables for Inspection by Variables for Percent Defective." Military Standard, MIL-STD-414 (June 11, 1957). CAREY, W. N., JR., and SHooK, J. R., "The Need for Change in Control Procedures." Nati. Conf. Statisti-cal Quality Control Method. Hwy. and Airfield Con-struction, Charlottesville, Va. (May 1966). "A Plan for Expediting the Use of Statistical Con-cepts in Highway Acceptance Specifications." Miller-Warden Assoc. for the Bureau of Public Roads, Ofc. Res. and Develop. (Aug. 1963). "The Statistical Approach to Quality Control in High-way Construction." Bureau of Public Roads (Apr. 1965). "Development of Guidelines for Practical and Realis-tic Construction Specifications." NCHRP Report 17 (1965) 109 pp.

"Futurized Revision FP-6 I: Standard Specifications for Construction of Roads and Bridges on Federal Highway Projects." Miller-Warden Assoc. for the Bureau of Public Roads (Dec. 1965). "Asphalt Overlays and Pavement Rehabilitation." Manual Ser. No. 17, The Asphalt Institute (Nov. 1969) pp. 99-108. "Quality Assurance in Highway Construction." Fed-eral Highway Administration [Reprinted from Pub. Roads, Vol. 35, Nos. 6-11 (1969)]. "Quality Assurance and Acceptance Procedures." HRB Special Report 118 (1971). SHooK, J. F., "Construction Materials Control for the AASHO Road Test." Jrnl. Soil Mechanics and Foun-dation Div., Proc. Amer. Soc. Civ. Eng., Paper 2211, Vol. 85, No. SM5 (Oct. 1959). "Results of the 1973 Questionnaire on Highway Ap-plications of Nuclear Techniques." Hwy. Res. Circ. No. 159 (July 1974). "Rapid Test Methods for Determination of Bitumen Content in Bituminous Mixtures." Special Tech. Publ. 461, Amer. Soc. Test, Materials (1969). HUDSON, S. B., and STEELE, G. W., "Prediction of Potential Strength of Concrete from the Results of Early Tests." Hwy. Res. Record No. 370 (1971) pp. 25-35. "Making, Accelerated Curing, and Testing of Con-

37

Crete Compression Test Specimens." ASTM C 684-74. 1.1 STEELE, G. W., HUDSON, S. B., and VAN Tn., C. J.,

"The Statistical Approach to Realistic Highway Speci-fications." Proc. Quality Assurance Workshop, spon-sored by Federal Highway Administration, Washing-ton (Oct. 1968) pp. 35-58.

19. HUDSON, S. B., "Handbook of Applications of Sta-tistical Concepts to the Highway Industry, Part II, Acceptance Plans." PB-220 595, Nati. Tech. Infor. Serv.

20, MACE, A. E., Sample-Size Determination. Reinhold (1964). LAMMI, T., "On Statistical Quality Control of Con-crete." English ed., The State Inst. Tech. Res., Helsinki (1963). REX, H. M., "Effect on the Road of Variations in Thickness." Proc. Assn. Asphalt Paving Tech., Vol. 33 (1964) pp. 10-17. VEsI, A. S., and SAXENA, S. K., "Analyses of Struc-tural Behavior of AASHO Road Test Rigid Pave-ments." NCHRP Report 97 (1970). ROSENBLUETH, E., ESTEVA, L., and DAMY, J. E., "Bonus and Penalty in Acceptance Criteria for Con-crete." ACI Title No. 71-3, JrnL Amer. Concrete Inst. (Sept. 1974) pp. 466-72. SHANE, R. M., "Controlling Concrete Compressive Strength by a Penalty Function." AC! Digest Paper Title No. 70-5, Jrnl. Amer. Concrete Inst. (Jan. 1973) pp. 28-30. "The Economic Feasibility of the Application of Sta-tistical Concepts and Methods to the Control and Ac-ceptance of Highway Materials and Construction." Materials Research and Develop., Inc. for the Federal Highway Administration, unpub. (1970). Quality Progress (Apr. 1971). DUNBAR, D. W., "Legal Responsibilities for Quality Control in Highway Construction." Western Builder (Aug. 1966) pp. 51-7. DiCocco, J. B., and BELLAIR, P. J., "Acceptance Sampling Plans for Rigid Pavement Thickness." Re-search Rept. 70-11, Engr. Res. and Develop. Bur., New York State Dept. of Transportation (Apr. 1971). WEBER, W. G., JR., GREY, R. L., and CADY, P. D., "Rapid Measurement of Concrete Pavement Thick-ness and Reinforcement Location-Field Evaluation of Nondestructive Systems." NCHRP Report 168 (1976) 63 pp. "Standard Specification for Ready-Mixed Concrete." ASTM C 94-71. "Standard Specification for Reinforced Concrete

D-Load Culvert, Storm Drain, and Sewer Pipe." AASHTO M 242-73. LEwis, E. V., Statistical Analysis. Van Nostrand (1963). "New Tables of the Non-Central t Distribution." ARL 63-19 Aeronaut. Res. Lab, for the Dept. of Com-merce (1963). DIXON, W. J., and MASSEY, F. J., JR., Introduction to Statistical Analysis. McGraw-Hill, 2nd ed. (1957). "Mathematical and Statistical Principles Underlying Military Standard 414." Ofc. Asst. Secy. Defense (Supply and Logistics) (1958). DUNCAN, A. J., Quality Control and Industrial Sta-tistics. Richard D. Irwin, Inc., revised ed. (1959). "A Radioisotope Backscatter Gauge for Measuring Cement Content of Plastic Concrete." FHWA-RD-73-48, Federal Highway Administration, Ofc. Res. and Develop. (1973).

STEELE, G. W., and HUDSON, S. B., "A Pycnometer Test Procedure for Determining Asphalt Content of Paving Mixtures." Spec. Tech. Publ. 461 (1969) pp. 67-85. KANDHAL, P. S., KOEHLER, W. C., and WENGER, M. E., "Rapid Determination of Asphalt ,. Content Using Pennsylvania Pycnometer." Hwy. Res. Record No. 468 (1973) pp. 89-99. "Road and Bridge Specifications, Section 304: Con-struction of Density Control Strips." Virginia Dept. of Highways (1970). "Evaluation of Construction Control Procedures-In-terim Report." NCHRP Report 34 (1967). "Evaluation of Construction Control Procedures-Ag-gregate Gradation Variations and Effects." NCHRP Report 69 (1969). WARDEN, W. B., "Evaluation of a Large Sized Pyc-nometer for Maximum Specific Gravity Determina-tions." Proc. Assn. Asphalt Paving Tech., Vol. 43 (1974) pp. 491-534. HUDSON, S. B., and STEELE, G. W., "Developments in the Prediction of Potential Strength of Concrete from the Results of Early Tests." Trans. Res. Record No. 558 (1975) pp. 1-12. "Modern Highway News Report." Rural and Urban Roads (Jan. 1975) p. 5. "Recommended Practice for Choice of Sample Size to Estimate the Average Quality of a Lot or Process." ASTM E 122. Quality Management and Engineering (Oct. 1975) p. 5.

APPENDIX

FEATURES. OF END-RESULT SPECIFICATIONS CONSIDERED OR IMPLEMENTED BY VARIOUS AGENCIES

Agency Contractor Agency Random Lot-by-Lot Acceptance based on Other Compaction Rapid Graduated Not or Quality Sampling Acceptance Statistical Referenced Test Unit-Price Considering

Producer Control Average Quality I

Acceptance to Control Methods Reduction ERS Qual. Con. Values Index Criteria Strip

Alabama . X

Alaska a c g a c g a c g a c g

Arizona. ag ag ag

Arkansas X

California c f t cit . c t

Colorado a c g f acgf a c g f a c g f a c g f

Connecticut a g c s c a c g c c a c g S

Delaware . X

Florida a c g f a c g f a c g f a c g f a c g f c c a c g f

Georgia a c f acgf a c g f s acg f . f c . acgfst

Hawaii c c c c c

Idaho ag ag ag ag

Illinois ag a c g a c g t a c g c a c g t

Indiana ag a c g s a c g a c g a c g s

Iowa Ct ct ct

Kansas X

Kentucky X

Louisiana a f g a c f g c f s c f s cfs

Agency Contractor Agency Random Lot-by-Lot Acceptance based on Other Compaction Rapid Graduated Not

or Quality Sampling Acceptance Statistical Referenced Test Unit-Price Considering

Producer Control Average Quality Acceptance to Control Methods Reduction ERS

Qual. Con. Values Index Criteria Strip

Maine a c g s t a c g s t a c g s t acgst

Maryland f f f a

Massachusetts X

Michigan c g c g c g c c c g

Minnesota a c g a c g a c g a c g c c c a c g

Mississippi a c g a c g acg a c g a c g

Missouri X

Montana X

Nebraska ag a c g t a c g t a c g t

Nevada c ct c c c ft

New Hampshire X

New Jersey ag agt ag ag t a c g t

New Mexico X

New York X

North Carolina g g g g,

North Dakota a c g a c g a g c a g

Ohio - X

Oklahoma . X

Key: a=asphalt content; c=compaction (density); f=compressive strength; g= aggregate gradation; pportland cement; s=smoothness; t=thickness.

Agency Contractor Agency Random Lot-by-Lot Acceptance based on Other Compaction Rapid Graduated Not or Quality Sampling Acceptance Statistical Referenced Test Unit-Price Considering

Producer Control Average Quality Acceptance to Control Methods Reduction ERS Quol. Con. Values Index Criteria Strip

Oregon g g

Pennsylvania a g c a c a c g a c c c a c

Rhode Island p p p p

SouthCarolina ag ag ag ag ag

South Dakota c c c c

Tennessee x

Texas x

Utah a a c g t a c g f t a c g f t c c acgft

Vermont a c g a c g a c g s a c g c c a c g s

Virginia a g a g a c g a g c c a g

Washington x

West Virginia acfgst acgfst acfgst g a c f s t c c acfgst

Wisconsin x

Wyoming a c g a c g acg a c g c a c g

Key: a=asphalt content; c=compaction (density); f=compressive strength; g= aggregate gradation; p=portland cement; s=smoothness; t=thickness.

Published reports of the Rep.

NATIONAL COOPERATIVE HIGHWAY RESEARCH PROGRAM No. Title

20 Economic Study of Roadway Lighting (Proj. 5-4), are available from: 77 p., $3.20

Transportation Research Board 21 Detecting Variations in Load-Carrying Capacity of

National Academy of Sciences Flexible Pavements (Proj. 1-5), 30 p., $1.40

2101 Constitution Avenue 22 Factors Influencing Flexible Pavement Performance

Washington D.C. 20418 (Proj. 1-3(2)), 69 p., $2.60 23 Methods for Reducing Corrosion of Reinforcing

Steel (Proj. 6-4), 22 p., $1.40 R ep. 24 Urban Travel Patterns for Airports, Shopping Cen- No. T ite ' ters, and Industrial Plants (Proj. 7-1), 116 p., -* A Critical Review of Literature Treating Methods of $5.20

Identifying Aggregates Subject to Destructive Volume 25 Potential Uses of Sonic and Ultrasonic Devices in Change When Frozen in Concrete and a Proposed Highway Construction (Proj. 10-7), 48 p., $2.00 Program of Research—Intermediate Report (Proj. 26 Development of Uniform Procedures for Establishing 4-3(2)), 81 p., $1.80 Construction Equipment Rental Rates (Proj. 13-1),

1 Evaluation of Methods of Replacement of Deterio- 33 p., $1.60 rated Concrete in Structures (Proj. 6-8), 56 p., 27 Physical Factors Influencing Resistance of Concrete $2.80 to Deicing Agents (Proj. 6-5), 41 p., $2.00

2 An Introduction to Guidelines for Satellite Studies of 28 Surveillance Methods and Ways and Means of Corn- Pavement Performance (Proj. 1-1), 19 p., $1.80 municating with Drivers (Proj. 3-2), 66 p., $2.60

2A Guidelines for Satellite Studies of Pavement Per- 29 Digital-Computer-Controlled Traffic Signal System formance, 85 p.+9 figs., 26 tables, 4 app., $3.00 for a Small City (Proj. 3-2), 82 p., $4.00

3 Improved Criteria for Traffic Signals at Individual 30 Extension of AASHO Road Test Performance Con- Intersections—Interim Report (Proj. 3-5), 36 p., cepts (Proj. 1-4(2)), 33 p., $1.60 $1.60 31 A Review of Transportation Aspects of Land-Use

4 Non-Chemical Methods of Snow and Ice Control on Control (Proj. 8-5), 41 p., $2.00 Highway Structures (Proj. 6-2), 74 p., $3.20 32 Improved Criteria for Traffic Signals at Individual

5 Effects of Different Methods of Stockpiling Aggre- Intersections (Proj. 3-5), 134 p., $5.00 gates—Interim Report (Proj. 10-3), 48 p., $2.00 33 Values of Time Savings of Commercial Vehicles

6 Means of Locating and Communicating with Dis- (Proj. 2-4), 74 p., $3.60 abled Vehicles—Interim Report (Proj. 3-4), 56 p. 34 Evaluation of Construction Control Procedures— $3.20 Interim Report (Proj. 10-2), 117 p., $5.00

7 Comparison of Different Methods of Measuring 35 Prediction of Flexible Pavement Deflections from Pavement Condition—Interim Report (Proj. 1-2), Laboratory Repeated-Load Tests (Proj. 1-3(3)), 29 p., $1.80 117 p., $5.00

8 Synthetic Aggregates for Highway Construction 36 Highway Guardrails—A Review of Current Practice (Proj. 4-4), 13 p., $1.00 (Proj. 15-1), 33 p., $1.60

9 Traffic Surveillance and Means of Communicating 37 Tentative Skid-Resistance Requirements for Main with Drivers—Interim Report (Proj. 3-2), 28 p., Rural Highways (Proj. 1-7), 80 p., $3.60 $1.60 38 Evaluation of Pavement Joint and Crack Sealing Ma-

10 Theoretical Analysis of Structural Behavior of Road terials and Practices (Proj. 9-3), 40 p., $2.00 Test Flexible Pavements (Proj. 1-4), 31 p., $2.80 39 Factors Involved in the Design of Asphaltic Pave-

11 Effect of Control Devices on Traffic Operations— ment Surfaces (Proj. 1-8), 112 p., $5.00 Interim Report (Proj. 3-6), 107 p., $5.80 40 Means of Locating Disabled or Stopped Vehicles

12 Identification of Aggregates Causing Poor Concrete (Proj. 3-4(1)), 40 p., $2.00 Performance When Frozen—Interim Report (Proj. 41 Effect of Control Devices on Traffic Operations 4-3(1)), 47 p., $3.00 (Proj. 3-6), 83 p., $3.60

13 Running Cost of Motor Vehicles as Affected by High- 42 Interstate Highway Maintenance Requirements and way Design—Interim Report (Proj. 2-5), 43 P., Unit Maintenance Expenditure Index (Proj. 14-1), $2.80 144 p., $5.60

14 Density and Moisture Content Measurements by 43 Density and Moisture Content Measurements by Nuclear Methods—Interim Report (Proj. 105), Nuclear Methods (Proj. 10-5), 38 p., $2.00

15 32 p., Identification of Concrete Aggregates Exhibiting

44 Traffic Attraction of Rural Outdoor Recreational

Frost Susceptibility—Interim Report (Proj. 4-3(2)) Areas (Proj. 7-2) n - L

8 $1.40 ' 66 p $4 00 45 Development of Improved Pavement Marking Ma-

16 Protective Coatings to Prevent Deterioration of Con- terials—Laboratory Phase (Proj. 5-5), 24 p., crete by Deicing Chemicals (Proj. 6-3), 21 p.,

$1.40

$1.60 46 Effects of Different Methods of Stockpiling and 17 Development of Guidelines for Practical and Realis- Handling Aggregates (Proj. 10-3), 102 p.,

tic Construction Specifications (Proj. 10-1), 109 p., $4.60

$6.00 47 Accident Rates as Related to Design Elements of 18 Community Consequences of Highway Improvement Rural Highways (Proj. 2-3), 173 p., $6.40

(Proj. 2-2), 37 p., $2.80 48 Factors and Trends in Trip Lengths (Proj. 7-4), 19 Economical and Effective Deicing Agents for Use on 70 p., $3.20

Highway Structures (Proj. 6-1), 19 p., $1.20 49 National Survey of Transportation Attitudes and Behavior—Phase I Summary Report (Proj. 20-4),

Highway Research Board Speciai Report 80. 71 p., $3.20

Rep. Rep. No. Title No. Title 50 Factors Influencing Safety at Highway-Rail Grade 76 Detecting Seasonal Changes in Load-Carrying Ca-

Crossings (Proj. 3-8), 113 p., $5.20 pabilities of Flexible Pavements (Proj. 1-5(2)), 51 Sensing and Communication Between Vehicles (Proj. 37 p., $2.00

3-3), 105 p., $5.00 77 Development of Design Criteria for Safer Luminaire 52 Measurement of Pavement Thickness by Rapid and Supports (Proj. 15-6), 82 p., $3.80

Nondestructive Methods (Proj. 10-6), 82 p., 78 Highway Noise—Measurement, Simulation, and $3.80 Mixed Reactions (Proj. 3-7), 78 p., $3.20

53 Multiple Use of Lands Within Highway Rights-of- 79 Development of Improved Methods for Reduction of Way (Proj. 7-6), 68 p., $3.20 Traffic Accidents (Proj. 17-1), 163 p., $6.40

54 Location, Selection, and Maintenance of Highway 80 Oversize-Overweight Permit Operation on State High- Guardrails and Median Barriers (Proj. 15-1(2)), ways (Proj. 2-10), 120 p., $5.20 63 p., $2.60 81 Moving Behavior and Residential Choice—A Na-

55 Research Needs in Highway Transportation (Proj. tional Survey (Proj. 8-6), 129 p., $5.60 20-2), 66 p., $2.80 82 National Survey of Transportation Attitudes and

56 Scenic Easements—Legal, Administrative, and Valua- Behavior—Phase II Analysis Report (Proj. 20-4), tion Problems and Procedures (Proj. 11-3), 174 p., 89 p., $4.00 $6.40 83 Distribution of Wheel Loads on Highway Bridges

57 Factors Influencing Modal Trip Assignment (Proj. (Proj. 12-2), 56 p., $2.80 8-2), 78 p., $3.20 84 Analysis and Projection of Research on Traffic

58 Comparative Analysis of Traffic Assignment Tech- Surveillance, Communication, and Control (Proj. niques with Actual Highway Use (Proj. 7-5), 85 p., 3-9), 48 p., $2.40 $3.60 85 Development of Formed-in-Place Wet Reflective

59 Standard Measurements for Satellite Road Test Pro- Markers (Proj. 5-5), 28 p., $1.80 gram (Proj. 1-6), 78 p., $3.20 86 Tentative Service Requirements for Bridge Rail Sys-

60 Effects of Illumination on Operating Characteristics tenis (Proj. 12-8), 62 p., $3.20 of Freeways (Proj. 5-2) 148 p., $6.00 87 Rules of Discovery and Disclosure in Highway Con-

61 Evaluation of Studded Tires—Performance Data and demnation Proceedings (Proj. 11-1(5)), 28 p., Pavement Wear Measurement (Proj. 1-9), 66 p., $2.00 $3.00 88 Recognition of Benefits to Remainder Property in

62 Urban Travel Patterns for Hospitals, Universities, Highway Valuation Cases (Proj. 11-1(2)), 24 p., Office Buildings, and Capitols (Proj. 7-1), 144 p.,

$2.00 $5.60 89 Factors, Trends, and Guidelines Related to Trip

63 Economics of Design Standards for Low-Volume Length (Proj. 7-4), 59 p., $3.20 Rural Roads (Proj. 2-6), 93 p., $4.00 90 Protection of Steel in Prestressed Concrete Bridges

64 Motorists' Needs and Services on Interstate Highways (Proj. 12-5), 86 p., $4.00 (Proj. 7-7), 88 p., $3.60 91 Effects of Deicing Salts on Water Quality and Biota

65 One-Cycle Slow-Freeze Test for Evaluating Aggre- —Literature Review and Recommended Research

gate Performance in Frozen Concrete (Proj. 4-3(1)), (Proj. 16-1), 70 p., $3.20

21 p., $1.40 92 Valuation and Condemnation of Special Purpose 66 Identification of Frost-Susceptible Particles in Con- Properties (Proj. 11-1(6)), 47 p., $2.60

crete Aggregates (Proj. 4-3(2)), 62 p., $2.80 93 Guidelines for Medial and Marginal Access Control 67 Relation of Asphalt Rheological Properties to Pave- on Major Roadways (Proj. 3-13), 147 p.,

ment Durability (Proj. 9-1), 45 p., $2.0 $6.20 68 Application of Vehicle Operating Characteristics to 94 Valuation and Condemnation Problems Involving

Geometric Design and Traffic Operations (Proj. 3 Trade Fixtures (Proj. 11-1(9)), 22 p., $1.80 10), 38 p., $2.00 95 Highway Fog (Proj. 5-6), 48 p., $2.40

69 Evaluation of Construction Control Procedures— 96 Strategies for the Evaluation of Alternative Trans- Aggregate Gradation Variations and Effects (Proj. portation Plans (Proj. 8-4), 111 p., $5.40 10-2A), 58 p., $2.80 97 Analysis of Structural Behavior of AASHO Road

70 Social and Economic Factors Affecting Intercity Test Rigid Pavements (Proj. 1-4(1)A), 35 p., Travel (Proj. 8-1), 68 p., $3.00 $2.60

71 Analytical Study of Weighing Methods for Highway 98 Tests for Evaluating Degradation of Base Course Vehicles in Motion (Proj. 7-3), 63 p., $2.80 Aggregates (Proj. 4-2), 98 p. $5.00

72 Theory and Practice in Inverse Condemnation for 99 Visual Requirements in Night Driving (Proj. 5-3), Five Representative States (Proj. 11-2), 44 p., 38 p., $2.60 $2.20 100 Research Needs Relating to Performance of Aggre-

73 Improved Criteria for Traffic Signal Systems on gates in Highway Construction (Proj. 4-8), 68 p., Urban Arterials (Proj. 3-5/ 1), 55 p., $2.80 $3.40

74 Protective Coatings for Highway Structural Steel 101 Effect of Stress on Freeze-Thaw Durability of Con- (Proj. 4-6), 64 p., $2.80 crete Bridge Decks (Proj. 6-9), 70 p., $3.60

74A Protective Coatings for Highway Structural Steel— 102 Effect of Weldments on the Fatigue Strength of Steel Literature Survey (Proj. 4-6), 275 p., $8.00 Beams (Proj. 12-7), 114 p., $5.40

74B Protective Coatings for Highway Structural Steel— 103 Rapid Test Methods for Field Control of Highway Current Highway Practices (Proj. 4-6), 102 p., Construction (Proj. 10-4), 89 p., $5.00 $4.00 104 Rules of Compensability and Valuation Evidence

75 Effect of Highway Landscape Development on for Highway Land Acquisition (Proj. 11-I), Nearby Property (Proj. 2-9), 82 p., $3.60 77 p., $4.40

Rep. Rep. No. Title No. Title

105 Dynamic Pavement Loads of Heavy Highway Vehi- 133 Procedures for Estimating Highway User Costs, Air des (Proj. 15-5), 94 p., $5.00 Pollution, and Noise Effects (Proj. 7-8), 127 p.,

106 Revibration of Retarded Concrete for Continuous $5.60 Bridge Decks (Proj. 18-1), 67 p., $3.40 134 Damages Due to Drainage, Runoff, Blasting, and

107 New Approaches to Compensation for. Residential Slides (Proj. 11-1(8)), 23 p., $2.80 Takings (Proj. 11-1(10)), 27 p., $2.40 135 Promising Replacements for Conventional Aggregates

108 Tentative Design Procedure for Riprap-Lined Chan for Highway Use (Proj. 4-10), 53 p., $3.60 nels (Proj. 15-2), 75 p., $4.00 136 Estimating Peak Runoff Rates from Ungaged Small

109 Elastomeric Bearing Research (Proj. 12-9), 53 p., Rural Watersheds (Proj. 15-4), 85 p., $4.60 $3.00 137 Roadside Development—Evaluation of Research

110 Optimizing Street Operations Through Traffic Regu- (Proj. 16-2), 78 p., $4.20 lations and Control (Proj. 3-11), 100 p., $4.40 138 Instrumentation for Measurement of Moisture-

111 Running Costs of Motor Vehicles as Affected by Literature Review and Recommended Research Road Design and Traffic (Proj. 2-5A and 2-7), (Proj. 21-1), 60 p., $4.00

97 p., $5.20 139 Flexible Pavement Design and Management—Sys- 112 Junkyard Valuation—Salvage Industry Appraisal tems Formulation (Proj. 1-10), 64 p., $4.40

Principles Applicable to Highway Beautification 140 Flexible Pavement Design and Management—Ma- (Proj. 11-3(2)), 41 p., $2.60 terials Characterization (Proj. 1-10), 118 p.,

113 Optimizing Flow on Existing Street Networks (Proj. $5.60 3-14), 414 p., $15.60 141 Changes in Legal Vehicle Weights and Dimensions-

114 Effects of Proposed Highway Improvements on Prop- Some Economic Effects on Highways (Proj. 19-3), erty Values (Proj. 11-1(1)), 42p., $2.60 184 p., $8.40

115 Guardrail Performance and Design (Proj. 15-1(2)), 142 Valuation of Air Space (Proj. 11-5), 48 p.,

70 p., $3.60 $4.00 116 Structural Analysis and Design of Pipe Culverts 143 Bus Use of Highways—State of the Art (Proj. 8-10),

(Proj. 15-3), 155 p., $6.40 406 p., $16.00 117 Highway Noise—A Design Guide for Highway En- 144 Highway Noise—A Field Evaluation of Traffic Noise

gineers (Proj. 3-7), 79 p., $4.60 Reduction Measures (Proj. 3-7), 80 p., $4.40

118 Location, Selection, and Maintenance of Highway 145 Improving Traffic Operations and Safety at Exit Gore

Traffic Barriers (Proj. 15-1(2)), 96 p., $5.20 146 Areas (Proj. 3-17) 120 p., $6.00 Alternative Multimodal Passenger Transportation

119 Control of Highway Advertising Signs—Some Legal Systems—Comparative Economic Analysis (Proj. Problems (Proj. 11-3(1)), 72 p., $3.60 8-9), 68 p., $4.00

120 Data Requirements for Metropolitan Transportation 147 Fatigue Strength of Steel Beams with Welded Stiff- Planning (Proj. 8-7), 90 p., $4.80 eners and Attachments (Proj. 12-7), 85 p.,

121 Protection of Highway Utility (Proj. 8-5), 115 p., $4.80 $5.60 148 Roadside Safety Improvement Programs on Freeways

122 Summary and Evaluation of Economic Consequences —A Cost-Effectiveness Priority Approach (Proj. 20- of Highway Improvements (Proj. 2-11), 324 p., 7), 64 p., $4.00 $13.60 149 Bridge Rail Design—Factors, Trends, and Guidelines

123 Development of Information Requirements and Transmission Techniques for Highway Users (Proj. 150

(Proj. 12-8), 49 p., $4.00 Effect of Curb Geometry and Location on Vehicle

3-12), 239 p., $9.60 Behavior (Proj. 20-7), 88 p., $4.80

124 Improved Criteria for Traffic Signal Systems in 151 Locked-Wheel Pavement Skid Tester Correlation and

Urban Networks (Proj. 3-5), 86 p., $4.80 Calibration Techniques (Proj. 1-12(2)), 100 p., 125 Optimization of Density and Moisture Content Mea- $6.00

surements by Nuclear Methods (Proj. 10-5A), 152 Warrants for Highway Lighting (Proj. 5-8), 117

126 86 p., $4.40 Divergencies in Right-of-Way Valuation (Proj. 11- 153

p., $6.40 Recommended Procedures for Vehicle Crash Testing

4), 57 p., $3.00 of Highway Appurtenances (Proj. 22-2), 19 p., 127 Snow Removal and Ice Control Techniques at Inter-

changes (Proj. 6-10), 90 p., $5.20 154 $3.20 Determining Pavement Skid-Resistance Requirements

128 Evaluation of AASHO Interim Guides for Design at Intersections and Braking Sites (Proj. 1-12), 64

of Pavement Structures (Proj. 1-11), 111 p., $5.60 155

p., $4.40 Bus Use of Highways—Planning and Design Guide-

129 Guardrail Crash Test Evaluation—New Concepts lines (Proj. 8-10), 161 p., $7.60

and End Designs (Proj. 15-1(2)), 89 p., 156 Transportation Decision-Making—A Guide to Social

$4.80 and Environmental Considerations (Proj. 8-8(3)),

130 Roadway Delineation Systems (Proj. 5-7), 349 p., 135 p., $7.20

$14.00 157 Crash Cushions of Waste Materials (Proj. 20-7),

131 Performance Budgeting System for Highway Main- 73 p., $4.80 tenance Management (Proj. 19-2(4)), 213 p., 158 Selection of Safe Roadside Cross Sections (Proj.

$8.40 20-7), 57 p., $4.40 132 Relationships Between Physiographic Units and 159 Weaving Areas—Design and Analysis (Proj. 3-15),

Highway Design Factors (Proj. 1-3(1)), 161 p., 119 p., $6.40 $7.20

Rep. No. Title No. Title 9 Pavement Rehabilitation—Materials and Techniques

160 Flexible Pavement Design and Management—Sys- 10 (Proj. 20-5, Topic 8), 41 P., $2.80 Recruiting, Training, and Retaining Maintenance and

tems Approach Implementation (Proj. 1-bA), Equipment Personnel (Proj. 20-5, Topic 10), 35 p., 54 p., $4.00 $2.80

161 Techniques for Reducing Roadway Occupancy Dur- 11 Development of Management Capability (Proj. 20-5, ing Routine Maintenance Activities (Proj. 14-2), Topic 12), Sop., $3.20 55 p., $4.40 12 Telecommunications Systems for Highway Admin-

162 Methods for Evaluating Highway Safety Improve- istration and Operations (Proj. 20-5, Topic 3-03), ments (Proj. 17-2A), 150 p., $7.40 29 p., $2.80

163 Design of Bent Caps for Concrete Box-Girder Bridges 13 Radio Spectrum Frequency Management (Proj. 20-5, (Proj. 12-10), 124 p., $6.80 Topic 3-03), 32 p., $2.80

164 Fatigue Strength of High-Yield Reinforcing Bars 14 Skid Resistance (Proj. 20-5, Topic 7), 66 p., (Proj. 4-7), 90 p., $5.60 $4.00

165 Waterproof Membranes for Protection of Concrete 15 Statewide Transportation Planning—Needs and Re- Bridge Decks—Laboratory Phase (Proj. 12-11), quirements (Proj. 20-5, Topic 3-02), 41 p., 70 p. $4.80 $3.60

166 Waste Materials as Potential Replacements for High- 16 Continuously Reinforced Concrete Pavement (Proj. way Aggregates (Proj. 4-1A), 94 P., $5.60 20-5, Topic 3-08), 23 p., $2.80

167 Transportation Planning for Small Urban Areas 17 Pavement Traffic Marking—Materials and Applica- (Proj. 8-7A), 71 p., $4.80 tion Affecting Serviceability (Proj. 20-5, Topic 3-

168 Rapid Measurement of Concrete Pavement Thickness 05), 44 p., $3.60 and Reinforcement Location—Field Evaluation of 18 Erosion Control on Highway Construction (Proj. Nondestructive Systems (Proj. 10-8), 63 p., 20-5, Topic 4-01), 52 p., $4.00 $4.80 19 Design, Construction, and Maintenance of PCC

169 Peak-Period Traffic Congestion—Options for Cur- Pavement Joints (Proj. 20-5, Topic 3-04), 40 p., rent Programs (Proj. 7-10), 65 p., $4.80 $3.60

170 Effects of Deicing Salts on Plant Biota and Soils— 20 Rest Areas (Proj. 20-5, Topic 4-04), 38 p.. Experimental Phase (Proj. 16-1), 88 p., $5.60 $3.60

171 Highway Fog—Visibility Measures and Guidance 21 Highway Location Reference Methods (Proj. 20-5, Systems (Proj. 5-6A) 40 p., $4.00 Topic 4-06), 30 p., $3.20

172 Density Standards for Field Compaction of Granular 22 Maintenance Management of Traffic Signal Equip- Bases and Subbases (Proj. 4-8(2)), 73 p., $4.80 ment and Systems (Proj. 20-5, Topic 4-03) 41 p.,

• $4.00 23 Getting Research Findings into Practice (Proj. 20-5,

Topic 11) 24 p., $3.20 24 Minimizing Deicing Chemical Use (Proj. 20-5,

Topic 4-02), 58 p., $4.00 - 25 Reconditioning High-Volume Freeways in Urban

Areas (Proj. 20-5, Topic 5-01), 56 p., $4.00 26 Roadway Design in Seasonal Frost Areas (Proj. 20-5,

Topic 3-07), 104 p., $6.00 27 PCC Pavements for Low-Volume Roads and City

Streets (Proj. 20-5, Topic 5-06), 31 p., $3.60 28 Partial-Lane Pavement Widening (Proj. 20-5, Topic

5-05), 30 p., $3.20 29 Treatment of Soft Foundations for Highway Em-

bankments (Proj. 20-5, Topic 4-09), 25 p., $3.20

30 Bituminous Emulsions for Highway Pavements (Proj.

Synthesis of Highway Practice 20-5, Topic 6-10), 76p., $4.80 31 Highway Tunnel Operations (Proj. 20-5, Topic 5-08),

No. Title 29 p., $3.20

1 Traffic Control for Freeway Maintenance (Proj. 20-5, 32 Effects of Studded Tires (Proj. 20-5, Topic 5-13),

Topic 1), 47 p., $2.20 46 p., $4.00

2 Bridge Approach Design and Construction Practices 33 Acquisition and Use of Geotechnical Information

(Proj. 20-5, Topic 2), 30 p., $2.00 (Proj. 20-5, Topic 5-03), 40 p., $4.00

3 Traffic-Safe and Hydraulically Efficient Drainage 34 Policies for Accommodation of Utilities on Highway

Practice (Proj. 20-5, Topic 4), 38 p., $2.20 Rights-of-Way (Proj. 20-5, Topic 6-03), 22 p., 4 Concrete Bridge Deck Durability (Proj. 20-5, Topic $3.20

3), 28 p., $2.20 35 Design and Control of Freeway Off-Ramp Terminals 5 Scour at Bridge Waterways (Proj. 20-5, Topic 5), (Proj. 20-5, Topic 5-02), 61 P., $4.40

37 p., $2.40 36 Instrumentation and Equipment for Testing Highway 6 Principles of Project Scheduling and Monitoring Materials, Products, and Performance (Proj. 20-5,

(Proj. 20-5, Topic 6), 43 p., $2.40 Topic 6-01), 70 p., $4.80 7 Motorist Aid Systems (Proj. 20-5, Topic 3-01), 37 Lime-Fly-Ash-Stabilized Bases and Subbases (Proj.

28 P., $2.40 20-5, Topic 6-06), 66 P., $4.80 8 Construction of Embankments (Proj. 20-5, Topic 9), 38 Statistically Oriented End-Result Specifications (Proj.

38 P., $2.40 20-5, Topic 6-02), 40 P., $4.00

THE TRANSPORTATION RESEARCH BOARD is an agency of the National Research Council, which serves the National Academy of Sciences and the National Academy of Engineering. The Board's purpose is to stimulate research concerning the nature and performance of transportation systems, to disseminate information that the research produces, and to encourage the application of appropriate research findings. The Board's program is carried out by more than 150 committees and task forces composed of more than 1,800 administrators, engineers, social scientists, and educators who serve without compensation. The program is supported by state transportation and highway departments, the U.S. Department of Transportation, and other organizations interested in the development of transportation.

The Transportation Research Board operates within the Commission on Sociotech-nical Systems of the National Research Council. The Council was organized in 1916 at the request of President Woodrow Wilson as an agency of the National Academy of Sciences to enable the broad community of scientists and engineers to associate their efforts with those of the Academy membership. Members of the Council are appointed by the president of the Academy and are drawn from academic, industrial, and govern-mental organizations throughout the United States.

The National Academy of Sciences was established by a congressional act of incorpo-ration signed by President Abraham Lincoln on March 3, 1863, to further science and its use for the general welfare by bringing together the most qualified individuals to deal with scientific and technological problems of broad significance. It is a private, honorary organization of more than 1,000 scientists elected on the basis of outstanding contribu-tions to knowledge and is supported by private and public funds. Under the terms of its congressional charter, the Academy is called upon to act as an official—yet indepen-dent—advisor to the federal government in any matter of science and technology, although it is not a government agency and its activities are not limited to those on behalf of the government.

To share in the tasks of furthering science and engineering and of advising the federal government, the National Academy of Engineering was established on December 5, 1964, under the authority of the act of incorporation of the National Academy of Sciences. Its advisory activities are closely coordinated with those of the National Academy of Sciences, but it is independent and autonomous in its organization and election of members.

TRANSPORTATION RESEARCH BOARD National Research Council

2101 ConstItutIon Avenue, N.W.

Washngton, D.C. 20418

ADDRESS CORRECTION REQUESTED

NON.PROFIT ORG.

U.S. POSTAGE

PAID

WASHINGTON, D.C.

PERMIT NO. 42970

N '0

• (J).N

Cq

H

(JDCN. 'H Z—• C N: 0

Lfl 0'

0 V) 0 0H

O HPs

Date post:	04-Jan-2022
Category:	Documents
Upload:	others
View:	5 times
Download:	0 times

STATISTICALLY ORIENTED END-RESULT SPECIFICATIONS

Documents