NATIONAL COOPERATIVE HIGHWAY RESEARCH PRO6RAM REPORT 117 HIGHWAY NOISE A DESIGN GUIDE FOR HIGHWAY ENGINEERS
HIGHWAY RESEARCH BOARD NATIONAL RESEARCH COUNCIL NATIONAL ACADEMY OF SCIENCES -NATIONAL ACADEMY OF ENGINEERING HIGHWAY RESEARCH BOARD 1971 Officers CHARLES E. SHUMATE, Chairman ALAN M. VOORHEES, First Vice Chairman WILLIAM L. GARRISON, Second Vice Chairman W. N. CAREY, JR., Executive Director
Executive Committee F. C. TURNER, Federal Highway Administrator, U. S. Department of Transportation (ex officio) A. E. JOHNSON, Executive Director, American Association of State Highway Officials (ex officio) ERNST WEBER, Chairman, Division of Engineering, National Research Council (ex officio) OSCAR T. MARZKE, Vice President, Fundamental Research, U. S. Steel Corporation (ex officio, Past Chairman, 1969) D. GRANT MICKLE, President, Highway Users Federation for Safety and Mobility (ex officio, Past Chairman, 1970) CHARLES A. BLESSING, Director, Detroit City Planning Commission HENDRIK W. BODE, Professor of Systems Engineering, Harvard University JAY W. BROWN, Director of Road Operations, Florida Department of Transportation W. J. BURMEISTER, State Highway Engineer, Wisconsin Department of Transportation HOWARD A. COLEMAN, Consultant, Missouri Portland Cement Company HARMER E. DAVIS, Director, Institute of Transportation and Traffic Engineering, University of California WILLIAM L. GARRISON, Professor of Environmental Engineering, University of Pittsburgh GEORGE E. HOLBROOK, E. I. du Pont de Nemours and Company EUGENE M. JOHNSON, President, The Asphalt Institute A. SCHEFFER LANG, Department of Civil Engineering, Massachusetts Institute of Technology JOHN A. LEGARRA, State Highway Engineer and Chief of Division, California Division of Highways WILLIAM A. McCONNELL, Director, Operations Office, Engineering Staff, Ford Motor Company JOHN J. McKETTA, Department of Chemical Engineering, University of Texas J. B. McMORRAN, Consultant JOHN T. MIDDLETON, Acting Commissioner, National Air Pollution Control Administration R. L. PEYTON, Assistant State Highway Director, State Highway Commission of Kansas MILTON PIKARSKY, Commissioner of Public Works, Chicago, Illinois CHARLES E. SHUMATE, Executive Director-Chief Engineer, Colorado Department of Highways DAVID H. STEVENS, Chairman, Maine State Highway Commission ALAN M. VOORHEES, Alan M. Voorhees and Associates
NATIONAL COOPERATIVE HIGHWAY RESEARCH PROGRAM Advisory Committee CHARLES E. SHUMATE, Colorado Department of Highways (Chairman) ALAN M. VOORHEES, Alan M. Voorhees and Associates WILLIAM L. GARRISON, University of Pittsburgh F. C. TURNER, U. S. Department of Transportation A. E. JOHNSON, American Association of State Highway Officials ERNST WEBER, National Research Council OSCAR T. MARZKE, United States Steel Corporation D. GRANT MICKLE, Highway Users Federation for Safety and Mobility W. N. CAREY, JR., Highway Research Board
General Field of Traffic Area of Operations and Control Advisory Panel G3-7 ALGER F. MALO, City of Detroit (Chairman) JOHN E. BAERWALD, University of Illinois WESLEY R. BELLIS, Consultant FRED W. HURD, The Pennsylvania State University ADOLF D. MAY, JR., University of California HAROLD L. MICHAEL, Purdue University KARL MOSKOWITZ, California Division of Highways WILBUR H. SIMONSON, Consultant WAYNE V. yOLK, Wisconsin Department of Transportation WILLIAM W. WOLMAN, Federal Highway Administration EDWARD A. MUELLER, Florida Department of Transportation
Program Staff W. HENDERSON, JR., Program Director M. MAcGREGOR, Administrative Engineer W. L. WILLIAMS, Projects Engineer W. C. GRAEUB, Projects Engineer HERBERT P. ORLAND, Editor J. R. NOVAK, Projects Engineer ROSEMARY S. MAPES, Editor H. A. SMITH, Projects Engineer CATHERINE B. CARLSTON, Editorial Assistant S . DH-404 Idaho Dept. of Hhwoys .. STANDARD. . COMPUTATION SHEET Description
Location P . . .. , Sheet , of
e (3 • 5 , . or e
Tr ,. . . . . 4fTh /ei'tertT.
Q 4b l LIKe ( I
rr
re ro br h- . b<. •.k
T kw-b1 L_ .000Z
e< go a orrLcIb to ° ( dc-e
mT p L the (- k IT
IT A to
ãJ) -
. de
Computed by Date -.ID 1 Project No. .
Checked by , Dote ______File or Book No.
Designed , . ' Checked DH - 404 3-67 STANDARD. COMPUTATION SHEET Idaho Dept. of Highways
Description - A 1 •Q
Location F Sheet of -
6 'P .toSrcdP.
.-6o .8333 079
I 3 I LID
- Zo I 61l1
0 o ..s000 4.00 .Qf
tIll 3869 i -
L!O 360
50 -7E ZZZ2 U 5 0 6 s
60 67 6 00 778
3611 r389 7Z0 S 57
70 3SE I II
75 U1667 08333 - 1079
055% lB 0 I. 55
SI '-ioo o5ooO ZO o I3 01
:H.S6
63 6I!I T , lU 10
03333 30 0 j 77
65 0?7?S 360
07-2Z 5 0 i S3
U833 017 600 1776
qo o LJ
89 0055% 10 I Z2S Computed by AFel e tj Date 11-I0 71 Project No. Checked by Dote ______File or Book No. Designed Checked '-7
DH-4043-67 STANDARD COMPUTATION SHEET Idaho Dept. of Highways DesdrIption cie (e eJ a ' Location N1 HR P Rep-, Sheet______of - 10k) IUO 9 e iBO
3 (coo 17.73 V
H S.O L-...... .. ..- 5 r .
0
Sts - - -
30 60 7'76
6.s - r
3.60 . .
3-OP . -.
1.80 V - ------'-.--•-- r- ••-- - 0-- i ------...... 50
ii-ió.• . .. I..I . . - It - -1.
-- Computed by i n Date 1 10 / I Froect No. Checked by Date ______File or Book No. Designed Checked 0
-60 •-10 20 10 60 60 e d9rees
sern - f;e el rev\ - A aI 0 £10 60 50 loo 0 e oIq- •q
IO 103 V7?ll
-- &.o osh oofl-qSj
O 105 Z0%1 I
R S4a4e j
NATIONAL COOPERATIVE HIGHWAY RESEARCH PROGRAM REPORT 117
HIGHWAY NOISE A DESIGN GUIDE FOR HIGHWAY ENGINEERS
COLIN G. GORDON. WILLIAM J. GALLOWAY, B. ANDREW KUGLER, AND DANIEL L. NELSON BOLT BERANEK AND NEWMAN LOS ANGELES, CALIFORNIA
RESEARCH SPONSORED BY THE AMERICAN ASSOCIATION OF STATE HIGHWAY OFFICIALS IN COOPERATION WITH THE FEDERAL HIGHWAY ADMINISTRATION
AREAS OF INTEREST: HIGHWAY DESIGN ROAD USER CHARACTERISTICS URBAN COMMUNITY VALUES
HIGHWAY RESEARCH BOARD DIVISION OF ENGINEERING NATIONAL RESEARCH COUNCIL NATIONAL ACADEMY OF SCIENCES- NATIONAL ACADEMY OF ENGINEERING 1971 NATIONAL COOPERATIVE HIGHWAY RESEARCH PROGRAM
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 NCHRP Report 117 their state universities and others. However, the accelerat- ing growth of highway transportation develops increasingly Project 3-7 FY '67 complex problems of wide interest to highway authorities. ISBN 0-309-01907-9 These problems are best studied through a coordinated L. C. Catalog Card No. 72-169279 program of cooperative research.
In recognition of these needs, the highway administrators - Price $4.60 of the American Association of State Highway 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 re- ceives the full cooperation and support of the Federal Highway Administration, United States Department of Transportation. The Highway Research Board of the National Academy of Sciences-National Research Council was requested by This report is one of a series of reports issued from a continuing research program conducted under a three-way agreement entered the Association to administer the research program because into in June 1962 by and among the National Academy of Sciences- of the Board's recognized objectivity and understanding of National Research Council, the American Association of State High- modern research practices. The Board is uniquely suited way Officials, and the Federal Highway Administration. Individual fiscal agreements are executed annually by the Academy-Research for this purpose as: it maintains an extensive committee Council, the Federal Highway Administration, and participating structure from which authorities on any highway transpor- state highway departments, members of the American Association tation subject may be drawn; it possesses avenues of com- of State Highway Officials. munications and cooperation with federal, state, and local This report was prepared by the contracting research agency. It has been reviewed by the appropriate Advisory Panel for clarity, docu- governmental agencies, universities, and industry; its rela- mentation, and fulfillment of the contract. It has been accepted by tionship to its parent organization, the National Academy the Highway Research Board and published in the interest of of Sciences, a private, nonprofit institution, is an insurance effective dissemination of findings and their application in the for- mulation of policies, procedures, and practices in the subject problem of objectvity; it maintains a full-time research correlation area. staff of specialists in highway transportation matters to The opinions and conclusions expressed or implied in these reports bring the findings of research directly to those who are in are those of the research agencies that performed the research. They a position to use them. are not necessarily those of the Highway Research Board, the Na- tional Academy of Sciences, the Federal Highway Administration, The program is developed on the basis of research needs the American Association of State Highway Officials, nor of the identified by chief administrators of the highway depart- individual states participating in the Program. ments and by committees of AASHO. Each year, specific areas of research needs to be included in the program are proposed to the Academy and the Board by the American Published reports of the Association of State Highway Officials. Research projects to fulfill these needs are defined by the Board, and qualified NATIONAL COOPERATIVE HIGHWAY RESEARCH PROGRAM research agencies are selected from those that have sub- are available from: mitted proposals. Administration and surveillance of re- search contracts are responsibilities of the Academy and Highway Research Board its Highway Research Board. National Academy of Sciences The needs for highway research are many, and the 2101 Constitution Avenue National Cooperative Highway Research Program can Washington, D.C. 20418 make significant contributions to the solution of highway transportation problems of mutual concern to many re- sponsible groups. The program, however, is intended to (See last pages for list of published titles and prices) complement rather than to substitute for or duplicate other highway research programs.
FORlV RD This report will be of special interest to highway design engineers, highway planners, architects, tire manufacturers, automobile manufacturers, legislators, By Staff and other officials who have to deal with the problems of traffic noise. The research Highway Research Board presents information that will allow engineers and architects to predict noise levels expected from a new highway facility. By comparing these .predicted noise levels against recommended noise design criteria, the impact of highway-generated noise on the community can be estimated. The noise evaluation technique is presented by means of a series of examples and includes a "cookbook"-type manual. The recommended noise design criteria are based on task interference considerations of speech and sleep. It should be noted that these recommended noise design criteria (noise standards) are tentative and subject to change as additional research is undertaken.
Questions related to highway noise levels and their effects on users of adjacent property arise frequently in the planning and design of highway improvements, particularly in urban areas. It is important to have means of evaluating probable noise levels adjacent to highways so that noise may be considered in the design of highway features and reduced through legislative actions or enforcement of vehicle regulations, or by design changes. It was with these thoughts in mind that this research was initiated in 1963. How will the introduction of a new highway influence the noise environment? How acceptable will people living or working near the highway find this new environment? These questions are of increasing importance today, as both the number of highways and the number of vehicles on the highways increase. To enable the highway engineer to answer these questions, Bolt Beranek and Newman has developed a "Design Guide for Traffic Noise Prediction." The Design Guide, intended to be a design "cookbook" rather than an acoustical textbook, is com- pletely self-contained and makes generous use of coordinated worksheets, tables, and figures. In this way, the highway designer, with no experience in acoustics is at once able to use the Design Guide quickly and effectively. The Design Guide estimate of highway noise levels is built on information readily available to the highway designer. This information includes the char- acteristics of the traffic flow (traffic density, vehicle types, and speeds), the charac- teristics of the roadway (elevation or depression, surface roughness, steepness), and the characteristics of the observer (location, height, intervening barriers or planting). On the basis of these inputs, the designer obtains estimates of an average noise level and a peak noise level. The differences between the estimated highway noise levels and the criteria determine the impact of the highway on the neighborhood. When the estimated levels lie above the criteria, individual and even group reactions may occur, depending on how much the estimates exceed the criteria. If necessary, the designer can repeat the Design Guide procedure, introducing such changes as roadway relocation or depression or barriers in order to satisfy the criteria. The Design Guide thus not only equips the highway designer to take the noise of a new highway into consideration, but also shows how various alternate designs will produce different degrees of impact on the neighborhood environment. This effort was by no means designed to solve all of the problems in the field of highway noise, and additional highway noise research will be conducted under the NCHRP. Previously completed NCHRP studies involving highway noise are presented in NCHRP Report 78, "Highway Noise—Measurement, Simulation, and Mixed Reactions," and NCHRP Report 75, "Effect of Highway Landscape Development on Nearby Property"; legal problems involving highway noise are given in NCHRP Project 11-1(7), "Valuation and Compensability of Noise, Pollution, and Other Environmental Factors." Of the many facets of the highway noise problem, future studies should assess the effects of the temporal variations of highway noise on speech communications, sleep, learning, and mental health. In addition to the research contained herein, a "highway noise" tape record- ing has been produced to assist engineers in their understanding of how different noise levels are heard, what the significance is of changing noise levels by various amounts, and how motor vehicle noise varies with traffic flow conditions. Loan copies of the tape recording, "Illustrative Recording of Traffic Noise," are avail- able on request to the NCHRP.
CONTENTS
I SUMMARY
PART I 2 CHAPTER ONE Introduction and Research Approach Definition of Symbols
3 CHAPTER TWO Findings Model of Traffic Noise Adjustments Criteria for Highway Noise
31 CHAPTER THREE Applications
31 CHAPTER FOUR Conclusions and Suggestions for Future Research Highway Traffic Noise Calculations Demonstration Project—Barriers and Propagation of Traffic Noise Attenuation Due to Structures and Intervening Buildings Tire-Roadway Noise Effects of Time-Varying Noise on Speech, Sleep, and Annoyance
33 REFERENCES
PART II
35 APPENDIX A Design Guide for Traffic Noise Prediction
75 APPENDIX B Illustrative Recording of Traffic Noise ACKNOWLEDGMENTS The study reported herein was conducted by Bolt Beranek undertaken jointly by Mr. Gordon, Dr. Galloway, and Mr. and Newman, Acoustical Consultants, in connection with Kugler. NCHRP Project 3-7. Cohn G. Gordon and Dwight E. Stewart Ferguson and Ronald Burns assisted in many of the Bishop acted as Co-Principal Investigators with major assistance field measurements of traffic noise undertaken during the study. from Dr. William J. Galloway, B. Andrew Kugler, and Daniel The measurement, analysis, and interpretation of stop-and-go L. Nelson. traffic was planned and directed by Charles Dietrich. David Lubman contributed to the statistical interpretations involved Analytic studies of traffic noise prediction procedures and in comparing traffic noise with ground noise and with speed. the review of noise criteria were largely accomplished by Particular thanks are due to Karl Pearsons, Dr. Sanford Messrs. Gordon and Nelson. The development of detailed Fidehl, and Brian Curtis for their efforts in preparing the design guide calculation procedures, tables, and graphs was highway noise demonstration recording. HIGHWAY NOISE
A DESIGN GUIDE FOR HIGHWAY ENGINEERS
SUMMARY Traffic noise is becoming an increasingly important consideration in the urban en- vironment. Currently, traffic noise is the predominant and most widespread source of urban noise, and it is clear that in future urban planning the design and routing of new highways should include traffic noise as one of the parameters. This study has revolved about the need for a Design Guide that will allow highway engineers to include highway noise in the design of future highways. This report summarizes the technical considerations used in the calculation procedures and the design criteria given in the Design Guide that is part of this report (Appen- dix A). The report discusses and compares different analytical and experimentally derived models of traffic noise, and describes the model used in the Design Guide. The report also describes the sources of information and technical approaches used in determining the noise level adjustments for finite element length, acoustical bar- riers, elevating or depressing the roadway, gradients and different road surface con- ditions, and the presence of intervening buildings or foliage between the observer and the noise source. Finally, the report describes and compares several approaches to the selection of criteria for traffic noise. These include criteria based on comparison of intruding noise levels with background noise levels, task interference considerations such as sleep and speech interference, and annoyance. The Composite Noise Rating (CNR) procedures where community response estimates are made based on the character- istics of the intruding noise source, consideration of background noise levels, and assessment of some community factors were also investigated in conjunction with the criteria study. The final criteria selected for the design guide are primarily based on the consideration of existing background noise levels and on speech interference criteria. The report concludes by considering various future research needs. The Design Guide developed, based on this report, identifies an the variables necessary for traffic noise prediction in terms of roadway parameters familiar to highway designers, the intention being to provide an easy-to-use design "cookbook." These parameters are then identified for a particular traffic situation and, through a simple procedure, transformed into noise level estimates through the use of charts and tables. Special work sheets are provided to facilitate this procedure. The evaluation of the particular traffic situation is achieved by comparing the estimated traffic noise levels against design criteria. The design criteria provide "maximum" levels for a variety of situations and for both inside and outside use. An "Illustrative Recording of Traffic Noise" is provided to serve as an illus- trative accompaniment to the Design Guide. The text of the recording appears as Appendix B. CHAPTER ONE INTRODUCTION AND RESEARCH APPROACH
Traffic noise is becoming an increasingly important parame- of the principles involved in their development. For this ter in the urban environment (1). Studies in several major reason this report includes a practical Design Guide for American and European cities have shown that despite the highway noise entitled "Design Guide for Traffic Noise noise produced by aircraft, surface traffic (automobiles, Prediction" (Appendix A). This Design Guide is the basic buses, trucks, motorcycles) is the predominant and most tool that has been developed in the study. Its purpose is to widespread source of noise. Even in industrial areas the allow the engineer to predict the noise environment of a dominant source is often the traffic related to the industrial highway design and to make decisions concerning the ac- ceptability of this environment without recourse to other activity. As traffic activities increase, so urban (and suburban) than highway-related parameters. noise levels are generally on the increase. A recent survey A further accompaniment to this report is the "Illustra- of community noise (2) established the statistic that, in the tive Recording of Traffic Noise." In this recording some selected geographic areas, noise levels were apparently rising basic acoustic terminology is introduced and illustrated. The at a rate close to 1 decibel (dB) per year. This increase, ways in which a total traffic noise situation derives from a one might suppose, is directly attributable to the increasing succession of individual vehicles and the influences of traffic population on the roads. If this postulate is correct, various roadway and building designs on traffic noise are then it is clear that future urban planning that includes noise demonstrated. Finally, the recording demonstrates some as a parameter must pay close attention to traffic flows aspects of the impact of traffic noise on speech intelligibility. within the urban area. The text of the recording is given in Appendix B. Three general principles of noise control can be identi- The purpose of this report is to supply some indepth dis- fied as potentially applicable to the traffic noise problem. cussion of the procedures and data used in the Design The first of these is the direct control of the noise of the Guide. individual vehicle. The two major categories of motor ve- Chapter Two describes the various methods of predicting hicle that are acoustically significant are automobiles and and modeling traffic noise that were studied in the course diesel trucks. Automobiles, although individually quiet, exist of the program. The method used in the Design Guide is in such numbers as to make their total noise contribution developed and discussed. In addition, the types of adjust- significant. ments that are applied to the basic traffic noise prediction The second general principle of noise control concerns to account for road geometry and terrain effects (shielding, traffic routing. Through traffic can bypass populated areas; etc.) are discussed. The sources of data used in the Design the provision of limited-access highways can effectively re- Guide are given. Finally, various criteria by which the sub- duce the vehicle population on surface streets. Unfortu- jective effects of traffic noise on people and communitics nately, however, almost any highway routing must affect can be assessed are presented and discussed. The bases of some people. Highways often attract residential and semi- the criteria used in the Design Guide are given. residential developments around them, and so it must be Chapter Three indicates the aids developed for applica- anticipated that highways in non-populated areas may even- tion of the findings to highway design. In Chapter Four, tually find themselves the cause of environmental noise ,some conclusions deriving from the study are presented, and problems. future work is suggested. The third general principle of traffic noise control is that the highway design itself has an impact on traffic noise; DEFINITION OF SYMBOLS therefore, the highway engineer can exert some degree of control over the noise environment generated by his Al = articulation index. creation. C = speed of sound in air. This study is concerned with the last two principles and it D = distance measured between observer and near- revolves around the need for a design tool that will allow est point to center line of roadway, in feet. the highway engineer to include the noise environment as DB = distance measured between observer position one of the parameters in his highway design. Thus, this and barrier, in feet. study is concerned with formulating those aspects of traffic De = distance measured between observer and cut flow and highway design that influence traffic noise and of roadway, in feet. relating these to the environmental needs of those people = distance measured between observer and who might be exposed to highway noise. Of overriding equivalent lane of roadway, in feet. concern in this study has been the need to develop these = distance measured between observer and the formulations so that they can be used and appreciated by DF center of the farthest lane of roadway, in feet. the highway engineer without any in-depth understanding D = distance measured between observer and cen- S = speed measured as the average speed of ve- terof hicular flow, in miles per hour. = distance parameter, measured between ob- SA = speed measured as above, for automobiles. server and shoulder of roadway, in feet. ST = speed measured as above, for trucks. U = level that describes effects of voice level and T = spacing between consecutive vehicles, in feet. speaker-listener distance. V = vehicle volume, in vehicles per hour. H = height parameter, in feet. VA = vehicle volume parameter (in vehicles per L = length measured along a finite roadway ele- hour), for automobiles only. ment, in feet. VT = Vehicle volume parameter (in vehicles per La = instantaneous traffic noise level measured in hour), for trucks only. dBA. X, Y, Z = random variable. Laio = sound level that is exceeded 10% of the time. Xo, Y0 = mean level. La50 = sound level that is exceeded 50% of the time. o = angle measured as included angle between ob- Lp = peak traffic noise level measured in dBA. server and roadway element. N = number of traffic lanes on roadway. H = source power per unit length. p = acoustic pressure. p = traffic density, in vehicles per mile. (p2 ) = mean square acoustic pressure. Poc = characteristic impedance of air. P = width parameter, measured from outside to a- = standard. outside lane on roadway, in feet. a-2 = variance.
CHAPTER TWO
FINDINGS
MODEL OF TRAFFIC NOISE The following sections discuss examples of some of the methods used in developing highway noise models, compare A vital part of the Design Guide is the methodology the results of these analyses, and present the bases for the whereby the noise environment produced by a specified method of predicting traffic noise that is used in the Design traffic situation is predicted. This section presents the stud- Guide. ies and thoughts behind the prediction methodology so that concepts and limitations of the methodology can be clearly appreciated. Requirements Past efforts to describe quantitatively the noise generatcd At the onset it is important to define just what the output by freely flowing traffic have taken two basic approaches: of a prediction method must consist of. From measured acoustical data obtained from arn- First, the noise output should be specified in terms of an pling traffic flows, attempts have been made to find mathe- easily measurable physical descriptor of the noise. Past matical expressions that most nearly describe the observed studies of response (3, 4) to traffic noise have shown that noise levels as a function of the several traffic flow parame- several different scales of physical noise measurement cor- ters (e.g., speed, vehicle flow rate, and observation distance). relate well with subjective response to noise. Such scales From knowledge of the acoustic power generated by include various versions of noisiness, loudness, and A- typical individual vehicles, noise models have been derived weighted sound pressure level. The latter scale is particu- by superimposing the sources that constitute the total flow. larly useful because A-levels can be read directly from any Three techniques used in this regard are: (I) assuming the precision sound level meter. Because the "dBA" level is total acoustic power due to all vehicles to be distributed regarded as (3) "statistically indistinguishable from the best uniformly along a continuous line and thus constituting an psychological derived measures in its reliability as a pre- acoustic line source of known acoustic power per unit dictor of human response to traffic noise" it is increasingly length; (2) considering the vehicles to be uniformly spaced being specified in national standards, and recommended in discrete sources along a hypothetical single-lane-equivalent international practice, for traffic noise studies. It is the roadway; and (3) using a Monte Carlo simulation pro- physical measure, therefore, used in this study. cedure, and deriving the statistical expectation of noise A second requirement for the prediction method con- produced by a randomly occurring flow of vehicles along cerns the way in which it accounts for the statistical varia- a hypothetical single-lane-equivalent roadway. tion in traffic noise from moment to moment. Early in this study it was decided that a useful prediction method must of the statistical distributions are controlled by individual define at least two points on the "statistical time distribu- noisy vehicles and by background (ambient) noise sources, tion" curve of the noise levels, and the points selected were respectively. Furthermore, for sufficiently long measure- the 50% level (L50), defined as that level exceeded 50% ment sampling times, it was found that noise level distribu- of the time, and the 10% level (L10), defined as that level tions are statistically "normal" with time for the flow rates exceeded 10% of the time. between 400 and 2,250 vph observed in the study. A third requirement relates to the characterization of the The effects of average vehicle speed, S, were studied. It sources of traffic noise—the vehicles themselves. On most was found that for fixed values of the ratio, V/ D, in which highways the vehicle population can be divided into two V is the flow of vehicles per hour past the observation acoustically significant groups. The first group can be de- position and D is the distance from highway center line to fined as "automobiles" and consists of gasoline-fueled ve- the observer, noise levels varied as 30 log S. Because the hicles having the general operating characteristics of pas- noise levels due to a moving column of noise sources can senger cars. The second group covers heavy diesel-fueled be shown to vary as 11S, the source strength of the indi- trucks such as tractor/trailer combinations and tank trucks. vidual vehicles in the Johnson and Saunders model is seen Certainly there are some vehicles that do not clearly lie to vary as the fourth power of speed. This is further within either of these two categories. Their numbers, how- discussed later. ever, are sufficiently small as to render them acoustically Taking account of the several factors just mentioned, insignificant—for highway design. Johnson and Saunders have produced the following em- Thus, the prediction method must be able to handle pirical expression for predicting the mean noise levels, in traffic situations involving different mixes of automobiles dBA, due to traffic flow on a level roadway: and trucks, each traveling at a different average road speed. L50 =3.5+101ogV—101ogD+3OlogS (1) Thus, it was decided early in the study to develop a pre- diction method for each vehicle group separately and to In the data from which Eq. 1 was derived the over-all arrive at the composite traffic level by summing the vehicle composition was normalized to include 20% com- contribution from each group. mercial vehicles such as trucks. Thus, the empirical ex- pression is assumed valid for this mixture of vehicle types, Empirically Derived Model varying ±1 dB with a mix ranging from 0% to 40% trucks. The objective in the empirically derived model is to infer The speed range considered was from 33 to 55 mph, with from measured data the parametric dependences that relate the constant in Eq. 1 obtained by normalizing to 40 mph. traffic noise levels to the physical measures of the traffic The small effect of trucks on the mean noise levels is not flow. Empirically derived models of traffic noise are limited at all consistent with observations in the United States (3). to the extent that data are available for the appropriate These results may be compared with Galloway's linear- range of traffic conditions and to the extent that effects not ized approximation to the Monte Carlo simulation of traffic related to traffic noise (spurious noise sources, attenuation, noise [described by Galloway et al. (3) and in the follow- etc.) can be discerned in the data. ing]. This approximation expresses the mean noise level in An important study of freely flowing traffic noise was dBA as: produced in Great Britain in 1967. Johnson and Saunders L5o=20+ 10logV-10logD+20logS (2) (5) undertook a series of roadside noise measurements be- tween 1963 and 1965 and, on the basis of the resulting data Comparing the numerical values obtained from the two and theoretical considerations, derived a model of the noise equations for D = 100 ft, S = 40 mph, and V = 1,200 vph, for freely flowing traffic. Efforts were made to ensure that one obtains 62.3 dBA and 62.8 dBA, respectively. This is measured data did not include anomalous effects due to indeed a strikingly close agreement if the small effect of the shielding, reflective surfaces, and the like. Noise levels, in truck-passenger car mix in Johnson and Saunders data is dBA, observed under various conditions of traffic flow, considered not representative of situations in the U.S. It velocity, and observation distance were examined with a is estimated that for a 20% mix of diesel trucks the average statistical distribution analyzer to determine the time dis- noise level would be 8 dBA higher than that predicted tribution of noise levels. Different parts of the level dis- previously (3). tributions were studied as the several traffic parameters varied. Analytical Models As might be expected, the results of the study show that Continuous Line Source Model at distances close to the roadway, and/or at sufficiently low traffic densities, it is the noise of individual vehicles that is The simplest model for freely flowing traffic noise assumes observed, and noise levels decrease at 6 dB per doubling of that vehicles are spaced sufficiently close together that the distance as expected for discrete (point) noise sources. At total acoustic power can be considered evenly distributed greater distances from the roadway and/or higher traffic along the equivalent center line of the roadway; Thus, the densities, the noise due to individual vehicles tends to smear source strength is described in terms of constant power per out into a line source from which levels (mean and peak) unit length. To a first approximation the line source is decrease at 3 dB per doubling of distance. assumed to be infinite in length and the expression for the Another result of the Johnson and Saunders study is that, acoustic pressure, (p2 ), at the observation point (Fig. 1) as would be expected, the highest and lowest noise levels due to an element, dx, of the line source, as follows:
Figure 1. Schematic configuration for derivation of line source noise model.
model was found to be accurately described by a Gaussian d6 (3) (p2)= (_podH)2i-D time-level distribution, the single line source model has no in which II is the source power per unit length, D is the statistical time distribution of levels. If one is interested perpendicular distance from the observation point to the not only in the mean levels that are presumably described by Eq. 5, but also, say, in the 10% or 90% levels, some roadway centerline, and p0c is the characteristic impedance of air. It should be noted from Eq. 3 that the contribution measure of a statistical time distribution must be introduced, to the mean-square acoustic pressure is a linear function of which in reality results from the truly discrete nature of the the subtended angle, d9. Segments of the line source that vehicles that constitute the traffic flow. correspond to equal angles of intercept contribute equally to the total mean-square acoustic pressure. Discrete Source Model Integrating Eq. 3 from limits of —7r/2 to +7r/2, cor- As mentioned, Johnson and Saunders have derived in con- responding to an infinitely long line source, the total junction with their empirical model a simple theoretical acoustic pressure is given by model of highway noise based on assumed uniform spacing p0c111 of identical vehicles, all traveling at the same average speed (4) D 2..- on a single-lane straight roadway. The spacing between consecutive vehicles is taken to be T, the average speed is Considering the vehicles to be automobiles and using the 5, and the perpendicular distance from the roadway to the automobile source power suggested by Galloway et al. (3), observation point is D. the noise level due to an infinite line source can be expressed The acoustic intensity observed is the summation of an in the following manner: infinite series of terms, each of which is proportional to the L50 =20+ 10logV-10logD+20logS (5) inverse square of the distance of the individual sources from the observation point. Thus, which is identical to Eq. 2. Thus, a simple continuous line source model is sufficient to estimate the average traffic 1 =-° (6a) noise level if the flow rate is of the order of 1,000 vph or 1=+ D2 + (St+nT) 2 more. It should be noted that by assuming a line source in which t is time from an arbitrary reference and n is a free of even vehicle distribution per unit length, the noise levels index to identify the individual vehicles. described are steady with time. Whereas the empirical This summation can be replaced by the single term, rl
sinh(2,rrD/T) L50 =4O log S— lO log D+ lO log p (6b) 3dBA (14) [ cosh(2D/T) - cos (2St/T)] -I- 10 log [tanh(1.19 X 10 3pD) ] + K Substituting this into Eq. 3, the intensity can be defined For sufficiently large values of the quantity DI T (pD> (within a factor representing the individual source power) —1,200), the hyperbolic tangent term approaches unity and as a function of time. The value of t is arbitrarily set equal the logarithm of the hyperbolic tangent vanishes. The high- to zero when a particular vehicle (n = 0) is at nearest density flow regime then tends to the same dependence as approach to the observation point. derived in the empirical model (see Eq. 1): Clearly, the maximum intensity is obtained when the L50 =K3 ±l01ogV-101ogD+30logS (15) cosine term in the denominator of Eq. 6b equals +1 (ve- hicle at nearest approach) and the intensity has its mini- The time-varying nature of the cosine term in Eq. 6b mum value when the cosine term equals —1 (two nearest indicates a variation in noise level around a mean value, as vehicles equidistant from the observation point). The 50% compared to the steady level prediction of the simple line level (i.e., the level exceeded for 50% of the time) coincides source of Eq. 5. A probability density function for the with the cosine term equal to 0. The mean level can then be noise level distribution described by Eq. 6b could be devel- written as oped. The periodicity introduced by the cosine term, how- ever, would generate a non-Gaussian distribution that is L10 — lO log _.tanh(21rD/T)] (7) not observed in the data obtained in traffic noise measure- DT I ments. In practice, therefore, only the mean level predicted and the maximum level can be expressed as by this approach is of use.
Lmax - 10 log Icoth (1rD/T)] (8) Simulation Model The analytical models described previously have several Noting that approximations can be made to the hyperbolic drawbacks. First, traffic flow is not characterized by uni- tangent and cotangent functions over certain ranges of the form spacing of vehicles. Second, the absorption of sound arguments, one sees that for DIT> ¼ (pD = 1,320 in the atmosphere is a function of frequency and distance, ft-veh/ mile) neither being accounted for in the analytical models. Third, the identification of a "siiigle-laiie equivalent" for L50 —10 log (l/DT) (9) multi-lane highways is justifiable only after examination of the effect of assuming multiple lanes first in the analysis. and for D/T < Yi (pD <440 ft-veh/mile) Fourth, the simple models do not allow for mixture of L50 —lO log (11T2 ) (10) various vehicle classes based on the noise output of the different types of vehicles. Fifth, the statistical distribution 2,640 ft-veh/mile) Similarly, for DIT> ½ (pD> of noise levels as a function of time cannot be realistically Lmax ~ 10 log (1IDT) (11) obtained from a deterministic model. Galloway et al. (3) used a simulation model to account and, for DIT < 1/6 (pD < 880 ft-veh/ mile) for these factors in developing a model of noise levels produced by freely flowing traffic. The model assumes a log (1ID2 ) (12) L.a. ~ 10 random distribution of vehicles distributed along a highway Note that to this point the results of the discrete source of any number of lanes. The noise characteristics of each model are independent of the type of vehicle, with all vehicle class are described in terms of octave frequency vehicles being assumed identical. band sound pressure levels at a given reference distance. To introduce the power dependence of each individual The simulation consists of summing the noise levels pro- noise source, Johnson and Saunders have indicated that duced at a specified observation point by a Poisson distri- each automobile generates acoustic power proportional to bution of vehicles having an average flow rate of 111. By the fourth power of the vehicle's speed (5). This velocity repeating the process a number of times, each time ran- dependence is assumed "on the basis of a statistical sam- domly selecting the vehicle distribution, but maintaining the pling of traffic cruising on an uninterrupted highway and average flow rate constant, histograms of the noise level as assumed representative of average vehicle behavior." Thus, a function of time are generated for a particular set of each term of the summation in Eq. 6a should contain the average flow rates, lane configurations, and vehicle mixes. factor K1S4, in which K1 is a constant. The mean noise The histograms allow computations of both mean noise level would then show the dependence, or levels and measures of the distribution function (e.g., standard deviation) for the various traffic flow parameters. L50 =4O log S— 101ogD — lO log T A detailed discussion of this approach is available by + 10 log [tanh(.-)] + K2dBA (13) Galloway et al. (3). It should be noted that the model successively superimposes vehicles until the addition of one more unit causes the summation to increase by no more in which K2 is a constant derived from experimental data. Because the headway distance, T, is inversely proportional than some specified amount (e.g., ½dB). In this sense, to the number of vehicles per mile, p, Eq. 13 can be re- the computer model truncates the roadway at some distance written as beyond which the individual traffic noise sources are in- WI significant with regard to the total effect of all closer technique is that noise levels can be estimated for traflic vehicles. The computer simulation, then, never describes flow conditions that may not be presently available or easily an infinite array of vehicles but always sees a finite road- accessible for measurement. TrallIc mixes, road geometries, way length. flow densities, and velocities can be varied at will. Some typical data derived from the simulation model are shown in Figure 2. As part of the study by Galloway et al. Comparison of Models the results of the computer model were verified by com- All the different models described previously contain terms parison with measured data; agreement was found to be that are linearly dependent on flow velocity, V, and dis- excellent (3, Fig. 1). An important asset of the simulation tance normal to the roadway, D. These terms describe the
80 0
75
70
65
0
60
1, 55
50
PASSE NGER CARS ONLY 20mph O 35mph 50mph o 65mph
40 - _
35 1 1 10 100 1000 Density in Vehicles per Mile of Roadway Figure 2. Traffic noise levels from computer simulation 100 ft from single-lane equivalent flow. line source characteristic of highway noise. The discrete At very high traffic densities, the computer model source model of Johnson and Saunders, Eq. 14, also in- implies that mean noise levels increase with density at a cludes the term 10 log [tanh (1.19 X 10 3pD ) 1, which ef- rate that is less than linear (i.e., more nearly 6 log p than fectively adjusts the line source representation to account 10 logp). for discrepancies between the lumped and distributed source Simulated noise levels decrease with distance at a rate descriptions. This effect is most marked at low densities that is greater than the linear expression 10 log D. The and! or small observation distances and approximations to computer-generated mean noise levels for distance of 100, the hyperbolic tangent function were introduced previously. 300, and 1,000 ft from the roadway more nearly approxi- At higher values of the product, pD, the hyperbolic tangent mate 15 log D than 10 log D behavior. term vanishes and the discrete source model approaches the An explanation for the behavior of the simulation result line source description. lies in the method by which the simulation model truncates A significant theoretical difference between the Johnson the infinite linear array of vehicles assumed in the other and Saunders results and the other models is the velocity formulations. The computer randomly positions successive dependence used. In their results, the effect on noise level four-vehicle units along the roadway in accordance with of individual vehicle velocity, obtained from random sam- the design traffic mix and traffic flow density. This process ples, was chosen to vary with the fourth power of velocity. terminates when the addition of another unit increases the In this work passenger-car noise varies with the third cumulative sound pressure level by less than a specified power of velocity, based on controlled measurements of the amount. A simple analysis that parallels the mechanics noise produced by individual vehicles. performed by the computer suggests that, as the traffic It might be argued that the acoustic power generated by density increases and noise levels increase, the length of the individual passenger vehicle should be proportional to highway that is effectively included in the view of the the power expended by the vehicle, thus providing some computer to obtain a stable noise distribution decreases. physical justification for selecting the cubic dependence on One can examine the practicality of the foregoing result. velocity. It is likely that the actual behavior observed is Considering the distance dependence of simulated traffic related to the predominance of exhaust, engine, or tire- noise levels, one notes that in the calculation for a finite roadway source mechanisms. Further, the importance of length line source, the noise level is proportional to the each mechanism probably varies not only with the type of angle, O where vehicle, but also with the speed range of intetesss. This is notably important as regards trucks because these vehicles O=arctan (L/2D) (16) tend to operate at nearly constant engine speeds and use L being the line length shown in Figure 1. Expanding the gear changes to regulate velocity. In a practical sense, the arctangent function in powers of the argument, (L!2D), difference between third- and fourth-power velocity laws is for L greater than 2D, it is seen that only when the value not very significant in the numerical results for the velocity of L!2 is much greater than D can the angle be approxi- range of practical interest. This is shown by the excellent mated by the first term of the expansion, /2. In this case agreement between the models described in the example the result is equivalent to the infinite line source result of given previously in "Empirically Derived Model." Eq. 13. The simple analysis of the finite line source further Because the velocity term in each of the models is a indicates that distance behavior should vary from 10 log D measure of the acoustic power produced by the individual to 20 log D as the ratio, L/2D, varies from very large to vehicles that comprise the traffic flow, the parametric models very small as compared to unity, respectively. The simu- given previously can be generalized to any type of vehicle lation results given by Galloway et al. (3) are closely by replacing the velocity terms of models 1, 3, or 4 by the described by the term 15 log D. The question then is, appropriate description for the type of source concerned. "Is the distance effect a realistic phenomenon or is it an In the development of the estimation procedure for use in idiosyncrasy of the computer procedure?" A series of measurements were obtained to compare with the simulation the Design Guide, it is assumed that automobiles generate results. The distance dependence characteristics in the two acoustic power proportional to the third power of their cases agree very closely. This leads one to believe that, in speed and that truck noise levels are independent of speed real life, roadways with which one is usually concerned are over the major portion of their operating range. indeed truncated at such lengths that the near-field approxi- Comparing now the results of the computer simulation mation (one-term expansion) to the arctangent function of• model with the empirical and theoretical models, very good Eq. 13 is not realized. As an example, one might look for agreement is obtained between the line source model and the configuration in which the noise level calculated by the the computer simulation output over the intermediate range two-term arctangent expansion is 1 dB below the infinite of traffic densities and for fixed observation distance, D. line value. This condition is obtained when L = SD, or the At lower traffic densities, the simulated levels are accurately double angle, 20, equals 136°. described by the discrete source analytical model. The in- It seems likely that in many practical situations the creasing slope of the plots in Figure 2 with decreasing length of roadway that effectively contributes to the noise density is accounted for by the hyperbolic tangent term of levels observed at a particular point is less than five times the analytical model. the observation distance. Lt is thus appropriate to expect Two interesting effects appear in the computer model noise levels to fall off at more than 3 dB per.doubling ..of that do not agree with the analytical and empirical models: distance from the roadway. This effect is both geometric
(i.e., shielding of extremities of the roadway by buildings final constant, as the means of predicting, for the Design and terrain) and physical in terms of the reduction in noise Guide, the average noise levels of automobile traffic. The levels produced by distant sources as a result of air absorp- modified equation is tion of sound. L50 = lO log p - l5 log D + 30 logS On the basis of the explanation for the distance behavior +10 log [tanh(1.19 X 103pD)] + 29dBA (19) demonstrated by the simulation model and on the strength of measured data, it is believed it is more appropriate to or, in terms of the more conventional automobile volume describe highway noise levels by 15 log D rather than by flow, V (vehicles per hour), This adaptation should apply only to the traffic 10 log D. L50 = 10 log V— 15 logD +1ogS flow regime over which the line source description is appro- 10 3VD/S) I + 29dBA (20) priate. The corresponding effect of line truncation with + 10 log [tanh(l.19 X regard to the discrete source noise model is to limit the This equation is plotted in Figure 3 for a fixed distance of summation of Eq. 6a to finite values of n. It is gratifying 100 ft from the traffic lane and for different values of aver- to note that the results of the computer simulation model age automobile speed, SA. This figure appears as Figure have brought attention to certain aspects of traffic noise that B-3 in the Design Guide (Appendix A). * are observed in reality, but that have seemingly escaped It should be noted that the foregoing model considers all consideration in previous analytical modeling techniques. traffic to be traveling on a single lane at a distance, D, from Incorporation of these modifications into the model of the observation point. In a later section the method of traffic noise for use in the Design Guide are discussed in extending the model to account for the distribution of the the following. traffic population on a number of parallel lanes is discussed.
Automobile Traffic Prediction Truck Traffic Prediction It has been noted that over a wide range of traffic flow Rather than extend the automobile noise prediction method situations the computer simulation results of Galloway et al. to account for different (non-zero) percentages of heavy (3) show good agreement with measured data. With the trucks, it is more convenient to consider trucks as an en- exception of the decreased density dependence at very high tirely different population of vehicles an4 tQ develop sepa- traffic densitics, it seems appropriate to model the noise of rate prediction curves that apply only to trucks. The real automobile traffic for closeness of fit to these simulated traffic situation that is a superposition of the automobile and levels. truck populations is then established simply by adding The problem then is to bring about agreement between logarithmically the predicted noise levels of each popula- the analytical models and the simulation model for various tion, because each population is statistically independent of values of traffic density and velocity at a standard distance the other. (in this case, 100 ft). For values of pD > 600 ft-veh/ mile With regard to noise levels due to truck traffic, many one finds that the line source model described by of the physical arguments and basic models are identical to the development of the automobile noise prediction L50 =101ogp—l5logD+3OIOgS+29dBA (17) schemes. The analytical model of Johnson and Saunders is is in excellent accord with the curves of Figure 2. For developed independently of the type of source vehicles, with values of pD <600 ft-veh/ mile, very good fit to the curves the source characteristics inserted. The empirical model of of Figure 2 obtains from Johnson and Saunders is based on mixed automobile and truck traffic flows of 20% truck composition; therefore, L50 = l0 log p - l5 log D + 30 logS this model is probably not particularly applicable to the + lO log [tanh(1.19X 10 3pD)]+31dBA (18) study of noise due to trucks alone. The line source model which derives from the analytical model for uniformly developed with regard to automobile traffic includes third- spaced discrete noise sources whose source power and power velocity dependence of the individual sources that generalized spectrum are given by Galloway et al. (3, compose the line source. Clearly, the line source model, Fig. B-7). Note that when the product, pD, exceeds 1,320 Eq. 17, and the discrete source model, Eq. 18, readily lend ft-veh/ mile the logarithm of the hyperbolic tangent term themselves to modification for the generalized truck source approaches zero and the parametric dependence of Eq. 18 characteristics. is identical to the line source model. The difference between As noted previously and as discussed by Galloway et al. high-density noise levels predicted by the two methods is a (3), truck noise levels are nearly independent of velocity. constant 2 dB. Comparing the low-density (pD < 600) Furthermore, it is observed that the noise levels due to a levels described by Eq. 18 with the computer simulation, truck at 50 ft exceed the noise due to the average auto- one finds best agreement at 50-mph velocity. At higher mobile, at the same distance and traveling at 50 mph, by velocities the analytical model overestimates the computer about 15 dBA. One can, therefore, modify the models by results by 1 to 2 dB; at speeds less than 50 mph the ana- dropping the 30 log S term from each formula and finding lytical results are slightly lower than the simulated noise an appropriate constant by comparison with the estimated levels. automobile noise levels at fixed values of p and for values Because the generally important range of traffic densities of S and D of 50 mph and 50 ft, respectively. is described by the condition, pD> 600 ft-veh/mile, the For truck traffic the general expression for the mean researchers have selected Eq. 18, with an adjustment of the noise level can be written vo? +ha+, io LOS S 1-o pi+ 3 $ ue'j here 10
80 50 40 30 70
SA mph
60
40 DN = 100 FT
20 Hourly Auto Volume, VA - vph Figure 3. Plot of Io for automobiles as a function of volume flow and average speed.
L50 =101ogp—l5logD+K (21) totes to the same parametric dependence as the line source model at sufficiently high values of the product, pD, there Evaluating this at D = 50 ft, S = 50 mph, and equating to being a 2-dB difference between the levels predicted by the Eq. 17 plus 15 dB, the value of the constant, K, is found two different methods at high densities. to be 95 dBA. Thus, the level of truck traffic noise is In the Design Guide the modified equation that follows predicted by is used to predict the mean noise levels due to truck traffic only: L50 = 101ogp — 15logD+95dBA (22) L50 '= 101ogp — lSlogD By a similar procedure, one can relate the discrete source + 10log[tanh(1.19X 10pD)]+95dBA (24) noise model for trucks to the similar expression for auto traffic to find the general equation or, in terms of the truck volume flow, V,
L50 = 101ogp — 15logD L50 = 101ogV — 101ogS — lSlogD +10log[tanh(1.19X 10 3pD)]+97dBA (23) + 10log[tanh(1.19X 10VDIS)]+95dBA (25) Because no data are available for noise levels due to This equation is plotted in Figure 4 for a fixed distance of 100% truck traffic, no direct check is possible to verify the 100 ft from the traffic lane and for different values of results of the foregoing equations. By analogy with the average truck speed, S. Note the apparent paradox that situation for automobile traffic, it might be anticipated that truck traffic noise at fixed volume flow decreases with in- Eq. 22 would best describe high-density truck flow (where creasing vehicle speed. This is accounted for by the fact the line source model is appropriate to significant vehicle that truck traffic noise is a function of vehicle density only overlap), whereas the discrete source model should better —distance being constant—and that, for fixed volume flow, fit observations of low-density truck traffic noise. As in the density decreases as average speed increases. Figure 4 case of automobile noise, the discrete source model asymp- appears as Figure B-4 in the Design Guide.
11
100
90
80
DN = 100 FT 60
20 30 50 40 50 60 70
I I I I I III! I I I I I I liii 10 100 1000 10,000 Hourly Truck Volume, VT - vph Figure 4. Plot of L. for trucks as a function of volume flow and average speed.
Effect of Road Width and Observation Distance must be taken of the width of the roadway occupied by traffic. So far, all the models and prediction equations have as- Account must also be taken in the real situation of sumed that the traffic is located at a distance, D, from the changes in the distance between the observation position observation position. In real life, of course, a highway may and the roadway. This matter is discussed earlier; suffice it have many lanes, and the width of the roadway may be here to say that a 15 log (DE) relation has been selected such that the observation distance from the near lane is as representing most closely the rate at which levels change significantly different from the observation distance from with distance from a roadway in real life. the farthest lane. The two influences of roadway width (or number of This problem can be overcome by using the concept of lanes) and observation distance are taken into account by the "single-lane-equivalent" distance. This is defined as applying to Figures 3 and 4 the corrections shown in that imaginary lane on which the total traffic flow can be Figure 5. The relevant observation distance is taken to be assumed located to be acoustically identical to the real-life the distance to the center line of the near lane (DN). This situation. This distance is denoted by DE in this report and seems to be appropriately convenient. Figure 5 appears as it can be shown to approximate closely to the geometric Figure B-5 in the Design Guide. mean of the distances from the observation point to the Two special effects are not taken into account in Figure nearest (Dr) and farthest F) lane center lines, respec- (D 5. Firstly, for sufficiently low-density traffic flow (pD < tively; i.e., 1,320 ft-veh/mile), distance behavior of mean noise levels D, but also on DE =VDN DF (26) is dependent not only on the term, 15 log the term, 10 log [tanh(l.19 X 10 3pD)}. For values of The prediction curves of Figures 3 and 4 in assuming all pD <420 ft-veh/ mile, it can be shown that the hyperbolic the traffic to be located on a single lane imply the use, tangent tends to the value, 1.19 X 10 3pD, which effectively therefore, of the single-lane-equivalent concept and, in reduces the distance dependence to —5 log D. This result applying these curves to a real traffic situation, account is not shown in that the reduction in slope of any one of 12 Mill IN on IUllhlUI11111I _JHHhIRIIIIlIIIUUIIIIIII __uiiuiouIauuIIIllhIUuIuIuII UiiUUIIIIIIIhUhIIIIII "Ui!iiiIIIIIIllhRUIIIIIII I INI, uIIIIIIINlNhIIH F Ro dw.y Width milli I ,uiiiiiiIuiIIuvsuaIaIIuI __•iilIIIIIUUUIIIIIIIIIIIII iiiiIUIIIIIIIIi!IRIIlIIIII Mill ONhhhhhhh1hI millillinsom llillooLillillII ._!11111I!11111111111u111Il Figure 5. Distance adjustment to account for observer-near lane distance and width of roadway.
the curves of Figure 5 would be related to the product, pD, is also required. Early in this study it was decided that the rather than simply distance alone. noisier aspects of the traffic environment could be ade- Secondly, it can be shown that sufficiently close to quately defined using the 10% level, L10 (i.e., that level a multi-lane highway the single-lane-equivalent concept which is exceeded for 10% of the time). In this section breaks down in that levels do not fall off as from a finite the method by which L10 can be estimated is described. or infinite line source but rather are influenced by the Using the theoretical model of Johnson and Saunders the depth-of-field of the distributed source. The net result of 10% intensity level can be evaluated by suitably setting the this phenomenon is to further reduce distance dependence value of 27rSt/ T in Eq. 6b. The appropriate value of the of the noise levels. Unlike the geometric effect involving cosine term is 0.951. The difference between the 10% and the traffic flow density, this latter modification to the 50% levels can, therefore, be written D (assumed) 15 log D behavior is a function solely of the distance from the source and the true physical extent cosh(l.19 X 10pD) L10 —L50 (27) (depth) of the source. — 10 log cosh(1.19 X 10pD) —0.951
Computing the 10% Level Because this model is based on a regular array. of moving vehicles it must be expected to be in some error insofar as Knowledge of the average traffic noise level is not, in itself, the "tails" of the statistical time distribution are involved. necessarily sufficient if one is to define environmental ac- ceptability. Some knowledge of the environmental peaks At low values of pD, however, where there is no "overlap" 13
of source influence, Eq. 27 should begin to approximate decibel adjustment to give the 10% level for both auto- the truth. mobiles and trucks. A source of data concerning the 10% traffic noise level Step 5: Using the usual decibel addition techniques, is to be found in Galloway et al. (3) from the simulation combine the 50% levels and the 10% levels for auto- model results; standard deviations for 100% automobile mobiles and trucks to obtain the 50% and 10% levels of traffic at different distances and volume flows are shown the composite vehicle population. (3, Fig. 6a). These data collapse to a single curve if plotted against the parameter pD (ft-veh/ mile). If it is assumed ADJUSTMENTS that the statistical time distribution is approximately nor- mal, then the difference between the 10% and 50% levels The basic traffic noise model assumes a straight, infinitely is found by multiplying the standard deviation by 1.28. long roadway lying at grade on a flat, level terrain. Apart This approach has the two following weaknesses: from the flow parameters the only variables included so far are roadway width (number of lanes) and observer-near 7 1. The assumption of distribution normality is not cor- lane separation. rect at low values of pD. To accommodate such real-life idiosyncrasies as curves, 2. At high values of pD the standard deviation might not junctions, gradients, cross-sectional changes, and non-level Vertical Adjustment fraction; an acoustic barrier does not create an absolute acoustic "shadow" on the side of the barrier remotc from The vertical displacement of an at-grade roadway to a posi- the source—diffraction of the sound at the top edge of the tion below (depressed) or above (elevated) ground level barrier "spills" sound energy into the shadow zone. provides a certain degree of acoustic shielding of the Consider the schematic diagrams in Figure 7 in which vehicular noise sources from the observation position. the elevated and depressed highway configurations are• The theoretical and experimental performance of acoustic represented by a source and an observer separated by a shields or barriers has been discussed by several authors knife edge. In the case of the elevated configuration, (a), 9). The basic principle involved is that of dif- (1, 3, 8, the position of the knife edge corresponds to the outer edge of the shoulder; in the case of the depressed configuration, (b), the knife edge corresponds with the intersect of the cut-slope with the terrain. In (c) the generalized geometry is shown represented by dimension X, Y and 7. TABLE 1 In a recent theoretical and experimental paper by ADJUSTMENT FOR INCREASED NOISE LEVEL Maekawa (8), the performance of semi-infinite acoustic OF TRUCKS ON GRADIENTS shields as a function of those parameters was carefully GRADIENT ADJUSTMENT discussed and validated in the laboratory. The basic design (%) (cIa) curve deriving from Maekawa's study is shown in Figure 8. It relates the excess attenuation of a semi-infinite barrier 3to4 +2 to the parameter 8(X + Y - Z) and the wavelength of 5to6 +3 the radiated sound; however, Maekawa's model applied +5 only to a single source-receiver distance. To obtain design data for practical use in highway design, Galloway (3) ,Theuence of gradients of 2% or less is considered to be neglible. 12 10 ___II!IIiiIIIIIIII__ 11111111 11111111 ___IIINI!!iIIII 11111111 _11111111 ___IIIIIIhhiI!iiii!IIIIIII_ 11111111 IIIIIHIIIIIIIIIiIIEINiIIIIHI104 1 o6 102 - 103 io VD/S In fs.t - v.hlcks/mIIe Figure 6. Derivation of 10 - Lo for homogeneous traffic population. ç k 15 performed measurements of noise produced by traffic under Source conditions of elevated and depressed grades. Based on these measurements Maekawa's curves were modified to be linear, as shown in Figure 8. This curve forms the basis I Observer for the adjustments for vertical configuration used in the Design Guide under the following assumptions; The design curve is applicable to a lane of vehicles (line source) as well as to a single vehicle (point source). All traffic can be assumed concentrated on a single Sch.matic of Elevated Highway lane whose position is that of the single-lane equivalent as described in "Effect of Road Width and Observation Distance." The effective wavelength of traffic noise is about 2 to Observer 3 ft (i.e., the numerical attenuation of the dBA traffic noise lies close to the attenuation provided at a sound frequency /1 of about 500 Hz). Source / Automobile noise sources are located on or close to the road surface. The effective location of truck noise sources is somewhere between the road surface and the exhaust stack exit. b) Schematic of Depressed Highway The foregoing assumptions have not been validated experi- mentally but represent reasonable assumptions. On the basis of the design curves of Figure 8, Figures 9 and 10 have been developed. In these figures the attenua- tions for elevated and depressed highways, respectively, are shown as a function of certain parameters. These curves are Source V felt to represent adequately the acoustical performance of z -_.Observer the respective geometries as regards automobile traffic noise. They are limited to a 15-dBA attenuation due to refraction are arbitrarily reduced by 5 dB for trucks, due to theJigher location of truck noise source. c) Generalized Geometry of Acoustic Barrier Shielding Adjustment Figure 7. Acoustic shielding geometrics. Three types of shielding are considered: (1) shielding by roadside barriers, (2) shielding by structures, and (3) shielding by plantings on the terrain between the roadway no 4v'uc.k re4iov c0 r b,r'rcr's and the observation position. see p5O This adjustment, as in the elevated-depressed case, is also Shielding by Barriers limited to —15 dBA, for the reasons mentioned previously. If the barrier does not shield the entire roadway element The attenuating influence of a solid, acoustically opaque under consideration, its effectiveness is reduced. A method roadside barrier has at, its origin the same physical princi- of evaluating the finite barrier adjustment is presented in the ples that were discussed with regard to the shielding influ- Design Guide. ence of elevated and depressed configurations. Thus, the design curve of Figure 8 is applicable and the same assump- Shielding by Structures tions concerning the source that are made in the section "Vertical Adjustment" are valid. The presence of buildings or structures on the terrain be- The resulting design plot is shown in Figure 11. In tween roadway and observer can have a significant shielding developing this plot it is assumed that the barrier is long effect. Sound penetration studies (14) indicate that this enough to effectively hide the total length of noise-producing shielding is effective for the first two or three rows of houses roadway. The degradation of performance caused by cur- and remains constant thereafter. Although no precise mea- tailing the length of a barrier depends on: surements are available in this area, a typical value of The potential performance of the barrier as set by its 3 to 5 dB per row of houses can be used (33). This attenua- height, noise source, and location relative to the observer. tion should not exceed a maximum of 10 dB and should be The total angle subtended at the observer by the applied only in cases where no direct line-of-sight exists unshielded portion of the highway. between observer and source. 16 -50 III 1111111111 11111111 ° —30iii ____ C 1111111111 iuuoi 4. C 0 E 1111111 11111111 IIlii!HhiiiIIIII UIHi!jiPaIiiI1iiIDesign Curve —10 11111 0 jjlflIiiliIIIII 11111111 11111111 0 2 a/A Figure 8. Attenuation by acoustic shielding. EQUIVALENT LANE OBSERVER 10 I 1111111 I 1111111 I__II _I 1111111 IAIIIllI__I__II __I__IUlIIIINii!!!II -20 IIUIHI I I. 4.0 6.0 8 - — H2 / D5 Figure 9. Adjustment for elevated roadway. 17 EQUIVALENT OBSERVER LANE I Z DE 10 __•__huh __I 1111111__I__I _I 1111111 I1IHhhI__I__Iohm, I_IRIIUINii!!!IU -ST.- 1111111 • •iuuir a - a (D - Dc) Figure 10. Adjustment for depressed roadway. EQUIVALENT OBSERVER LANE [nH I DEDB 10 I •IIIIN I_1111111 I_I 0 0 _I_Ilh_I IluuhhI_I C I 4 -1 0 MENEM11111 _• IIIMU_I_ii IL_ -,n I 1111111 I IIIIIIFi- I - DB) Figure 11. Adjustment for roadside barriers. 18 Shielding by Landscaping Surface Adjustment In general, planting either at the roadside or the terrain The influence of road surface on traffic noise is discussed between the roadway and the observer position has little at some length by Galloway et al. (3). Based on this work influence on the propagation of traffic noise (3). Bushes, the adjustments of Table 2 have been developed to account trees, and similar foliage can attenuate sound only if the for deviations from "normal" surface conditions. These growth is dense enough and the depth of the foliage is great, adjustments are applicable to both automobiles and trucks. as some experimental measurements (11, 12, 13) indicate. A single line of trees or a depth of trees of less than Interrupted Traffic Flows Adjustment 50 to 100 ft will provide little actual attenuation; rather, the When traffic flow is interrupted, as by a STOP sign or signal, visual isolation may reduce a person's awareness of the noise it must be anticipated that the noise characteristics of that to a greater extent. traffic flow will be somewhat different from the same traffic A design value of 5-dB noise reduction for every 100 ft flow operating under a freely flowing condition. Unfortu- of planting depth may be used if these trees are at least nately at this time there is no clearly identifiable method of 15 ft tall and sufficiently dense so that no visual path be- modeling this type of traffic flow, and interrupted flows have tween observer and roadway exists. This attenuation should received sparse treatment in the literature. A modeling tech- not exceed 10 dB at a maximum. nique that might be suggested is a derivative of the tech- nique used for predicting the noise from freely flowing traffic. In the freely flowing model, the sound power is related TABLE 2 to mechanical power expended by the vehicle at constant CLASSIFICATION OF ROAD SURFACE speed in overcoming rolling friction and aerodynamic drag. AS IT RELATES TO SURFACE INFLUENCE The same logic might suggest that the noise output of a ON VEHICLE NOISE vehicle at variable (stop-and-start) speed could be related ADJUST- to the mechanical power expended in this mode of opera- SURFACE MENT tion. The vehicle power in such operation is primarily TYPE DESCRIPTION (dB) expended on changing the vehicle speed, so a number of Smooth Very smooth, seal-coated parameters related to speed irregularity suggest themselves. asphalt pavement —5 Among these are the variance of the velocity, o, the Normal Moderately rough asphalt and variance of the acceleration, o- (or the standard deviation, concrete surface 0 Rough Rough asphalt pavement with a, which interestingly and confusingly enough is known to large voids ½ in. or larger the traffic flow theorists as "acceleration noise"), and in diameter, grooved concrete +5 weighted sums or mixed measures such as a a2 V(t). All of _- Cross Traffic Volume Small Compared Main Highway — — — — — — — — — — — — — — — — — — — — — — — — — — — / ------— — — — — — — — — — — — — — — — — 5Oft 5Oft i i 2 1.5miles (No Traffic Exits) 'I Figure 12. Site of measurement study for interrupted flow effects. 19 these measures can be derived experimentally from a simple TABLE 3 velocity record taken with the "floating-car" technique. LEVEL ADJUSTMENT FOR INTERRUPTED FLOW To develop a better quantitative feel for the influence of flow interruption upon traffic noise, a series of measure- ADJUSTMENT (dB) VEHICLE ments was made on a section of roadway that permitted TYPE measurement of identical traffic flows under interrupted and L59 L10 non-interrupted conditions, respectively. Figure 12 shows Auto 0 +2 a plan of the roadway situation, and the two measurement Truck .0 + locations. Two microphone locations are shown: Position 1 is 50 ft from the center of the near lane, close to a signalized intersection. The intersecting roadway carries only a small volume of vehicles (relative to the main highway); micro- phone Position 2 is also 50 ft from the center of the near following considerations to specifying maximum desirable lane but located about 1.5 miles farther along the main traffic noise levels: highway. The length of highway between the intersection and the second microphone location is free from traffic The relation of highway noise to the ambient. exits and so the traffic volume and composition measured Task interference as associated with speech, sleep, at Position 2 is identical to that measured at Position 1. learning, and other on-going activities. Only the operating mode and speed of the vehicles are General annoyance (i.e., subjective satisfaction or different. dissatisfaction with the environment). The results of these studies are shown in Figures 13 and In selecting these approaches it is assumed that allowable 14, representing different measurement days. The data traffic noise levels will lie below those levels that are di- shown are the statistical time distributions, plotted on prob- rectly physiologically harmful (i.e., levels causing hearing ability paper, obtained from 20- to 30-min magnetic tape loss or extreme startle reaction, for example). recordings of the traffic. The recordings were not made The selection of noise criteria involves three separate simultaneously but within 45 min of each other. The solid elements: (1) the choice of the physical measure or rating curve represents the data obtained at microphone Posi- scale for the noise, (2) the choice of a procedure for tion 2 under the free-flow traffic condition. The dashed evaluating the effect of the noise, and (3) the selection. of line represents the data obtained at the signalized inter- numerical values to achieve the desired environment. Gallo- section Position 2. way et al. (3) examined the first element, choice of scale, On both measurement days the volume flows were close in detail. The results of that investigation led to the recom- to 4,000 vph. Speeds were estimated at close to 50 mph. mendation that traffic noise be described in terms of A- On the first measurement dam (Test 1) the percentage of weighted sound pressure level expressed in dBA. In this heavy trucks was counted at about 2%. A truck count was section, the second two elements, procedure and quantitative not obtained on the second measurement day (Test 2). values for criteria, are discussed. Both tests show the expected result that stop-and-start operating conditions increase the slope of the statistical time Relation To Existing Ambient distribution of traffic noise. Further, it is noted that the Highway noise is considered as an intrusion with respect to mean (50%) noise level is not significantly changed; thus, the ambient levels that existed or would exist in the absence the mean noise level for interrupted flow can be computed of the highway and its associated traffic. on the basis of the vehicle operating speed that would It has often been common practice to require that any prevail if the cause of flow interruption was removed. new noise intruding on an environment be controlled only Taking account of these data and considering the likely to an extent compatible with the existing ambient. When additional impact of interrupted flow conditions on 100% the new sound is broadband in nature, as in the case of truck traffic, the corrections for interrupted flow are given traffic noise, with no time or frequency characteristics that in Table 3. The quantitative corrections for interrupted flow clearly identify it, the intruding levels can be higher relative are felt to be sufficiently small as to reduce the immediate to the ambient than if the new source of sound contains need for a model that represents the interrupted flow situa- pure-tone components or has intermittent time properties. tion in detail. Thus, Table 3 is used in the Design Guide. The clear consequence of a philosophy that permits each new noise intrusion to equal the existing noise environment CRITERIA FOR HIGHWAY NOISE is, of course, an upward creep of the environmental noise levels in 3-dB steps. Although such a situation is not par- In the development of highway noise criteria, the research- ticularly desirable, the process may not elicit too much ers have considered values and methodology previously response from the affected communities—until, of course, derived on several different, although not unrelated prem- such time as the environment becomes unacceptable from ises, and have arrived at criteria that they believe should the viewpoint of task interference or general annoyance. be used as design goals in the highway planning process. The purpose here is to examine the extent to which in- The principal consideration underlying all such criteria is trusion of traffic noise above the ambient is permissible some measure of subjective (human) response to noise. without incurring an unreasonable penalty in terms of sub- These responses, in turn, are reflected in one or more of the jective response. Fundamental to this approach is an under- 98 98 95 95 '4 \\ '4 — — — — — Interrupted Flow (1) ------Interrupted Flow (1) Free Flow (2) all Free Flow (2) '4 80 80 -o 70 5 2 0.5 0.5 0.2 0.2 Al Al 55 60 65 70 75 80 85 90 95 55 60 65 70 75 80 85 90 95 Noise Level In dBA Noise Level in dBA Figure 13. Results of interrupted flow study—Test 1. Figure 14. Results of interrupted flow study—Test 2. 21 standing of the correlation between objective measures of some value judgment must be made concerning the maxi- the noise and probable subjective reactions. To this end, mum permissible noise level. several general principles deriving from past experience in One concept of highway noise criteria based on this community noise studies are cited. intrusion-versus-ambient approach states simply that the It is generally changes in the environment rather than mean traffic noise levels (dBA) should not exceed the range absolute levels that precipitate public reaction. This notion, of mean ambient levels (measured or estimated) character- of course, does not extend to very high levels, but is prob- istic of the type of area of concern. Some typical ambient ably valid over a large range of situations that presently levels are shown in Figure 15. occur for traffic noise. The assumption is that ambient Intermittent excesses due, for instance, to passing trucks levels that have existed for some time have come to be should not exceed the range of ambient levels by more than accepted as satisfactory or are considered beyond the con- 10 dBA. It is noted that extraordinary events such as very trol of public reaction. No doubt, symbiotic relationships close truck passages or low aircraft flyovers may produce develop with time whereby people acclimatize to their en- levels that exceed ambient levels by 25 to 35 dBA; although vironment, provided exposure is not too severe. There is people may be acclimatized to such events, if they occur evidence to indicate that just the opposite may obtain when frequently, individual or public complaints may be precipi- levels are too high. At any rate, it is desirable that altera- tated. Thus, the 10-dBA excursion allowed for infrequent tion of the subjective acoustic environment be avoided or events by the foregoing criterion is conservative. Excesses minimized. of less than 5 dBA above the ambient range probably will Next, it might be postulated that the magnitude of change cause no significant complaints, whereas intermittent peaks will depend on the degree of intrusion that the new source of 5 to 10 dBA are only marginally acceptable. Ten- to bears to the ambient, and this of course will depend on the 15-dBA excesses may signal a potentially serious problem, time-varying characteristics of both the new source and the and peaks greater than 15 dBA or so would probably initiate existing ambient. strong individual or concerted public action. Traffic noise is most meaningfully described as a "statisti- Several comments can be made concerning the foregoing cal time distribution" of levels. Three useful values of the scheme for highway noise criteria. First, it is premised on distribution are the levels that exceed for 10%, 50%, and the assumption that existing ambient levels are, not likely to 90% of the time. the ambient itself commonly derives change in the future. Second, the resulting criteria for from traffic sources and thus is also characterized by a dis- traffic noise will prove acceptable only to the extent that the tribution with time; so, it is necessary in assessing the degree present environment is judged acceptable; further attenua- of intrusiveness of traffic noise to superimpose two random tion of traffic noise, however, is wasted effort as no benefit variables, each described by its own statistical properties. will accrue to the area. The criteria developed in the fore- Typically, noise due to such random processes as traffic flow going manner assure that relatively quiet areas remain quiet has a nearly Gaussian time distribution, and the probability while noisy areas require less stringent noise control mea- density representation is symmetrical about the mean value. sures. Some upward creeping of the over-all environment Knowing, for instance, the 10% and 90% levels of both the is allowed by this procedure. Finally, it should be noted existing ambient and the intruding highway noise, one can that application of these criteria is not everywhere possible. mathematically ascertain the percentage of the time for In very quiet areas, it is simply not possible to avoid alter- ing the environment, and in these cases it is necessary to which the intrusion will be above or below the ambient, set realistic design limits on the noise levels consistent with assuming well-behaved statistical properties of both signals. annoyance- and task-related requirements. In many cases, Experience has shown that large variations in acceptable introduction of a highway will significantly raise the am- noise levels result from different individual subjective eval- bient levels, and no reasonable amount of noise control will uations of highways. Indeed, evidence of this phenomenon prevent the change from being measurable, noticeable, and suggests that it is meaningless to design for other than the even objectionable. Whenever the amenity afforded by the average response with some margin of safety. In general, margin between the existing ambient and the maximum people of higher socio-economic status demonstrate greater permissible level is compromised, public reaction probably sensitivity to highway noise, and property owners are more will ensue. conscious of the deleterious effects of noise on property Sawley and Gordon (14) have developed guidelines for values than are apartment dwellers. The adjustment for the predicting community response to a steady intrusive noise subjective "meaning" of intruding highway noise is difficult when the intruding level is related to the statistical distribu- to estimate and is beyond the scope of desired generality. tion of the ambient. These results are given in Table 4. A significant part of community response to noise is When the intrusion is characterized by a statistical time- determined by conditions that prevail inside of dwellings level distribution, one can establish analytically the sub- and other occupied buildings. Some account must be taken jective exposure on the basis of percentage time audibility. of the inside levels as well as those outdoors. Knowing the The acoustic interpretation of the mathematics is rather noise reduction properties of typical building constructions, complicated except in simple cases. one can determine what indoor noise levels result from Given the mean levels (X0, Y0) and standard deviations outdoor sources and deduce the amount of intrusion. (o, o) of two normally distributed random variables, Inevitably, the situation will arise in which the ambient X (highway noise) and Y (ambient), the following ques- noise level is exceeded by the intrusive contribution and tions might be asked: For what percentage of the time will PIN i] 80 "DOWNTOWN" COMMERCL&L AREAS WITH HEAVY TRAFFIC INDUSTRIAL AREAS 70 -o C COMMERCIAL AREAS LIGHT TRAFFIC -60 I 50 URBAN RESIDENTIAL EM AREA (DAYTIME) QUIET SUBURBAN AREA (NIGHTTIME) 30 Figure 15. Typical continuous background noise levels. the signal, X, exceed the ambient, Y? For what relative 1 P(Z>Q)=—J e 212du (28) positions of the two distributions will X exceed Y by a. dBA for a given percentage of the time? To answer the first question, one needs to know the in which probability that Z = X - Y is greater than zero. This is u= (Z—Z)/o, obtained by evaluating =xo — Yo (29) Scra= Vo.2+o.2 and P(Z> 0) is tabulated for various values of the upper TABLE 4 limit of integration. EXPECTED COMMUNITY RESPONSE ON BASIS Typically, the standard deviations, a-,, and are nearly OF COMPARISON OF INTRUDING NOISE LEVEL equal. Further, experience has shown that a realistic value TO STATISTICAL DISTRIBUTION OF AMBIENT of o (and o- ,) is 5 dBA; thus, o = 7 dBA. NOISE LEVEL Eq. 28 dictates then that for X to be greater than Y for 10% of the time (i.e., the noise is audible for 10% of the LEVEL OF INTRUDING NOISE (IL) time) the condition on the respective mean levels is X0 = RELATIVE TO THE AMBIENT NOISE EXPECTED COMMUNITY — ( 1.28)(1.41)o= Y0 —9 dBA LEVEL (AL) RESPONSE Y0 - 1.288cr = Y0 Similarly: for 50% audibility, X0 = Y0; for 90% audi- IL AL No observed reaction <50% bility, X0 = Y0 + 9 dBA; and for 99% audibility, X0 = 50% AL < IL < 10% AL From no observed reaction to sporadic complaints Y0 + 16 dBA. 10% AL a From Sawley and Gordon (14). X5 = Y0 + 26 dBA, respectively. 23 Now, if the guidelines of Table 4 are interpreted in terms TABLE 5 of the implied audibility, one finds that for the (assumed) EXPECTED COMMUNITY RESPONSE ON BASIS equal statistical properties of the traffic noise and ambient OF COMPARISON OF MEAN LEVELS OF INTRUDING distributions, with the common standard deviation equal to NOISE AND AMBIENT NOISE 5 dBA, anticipated community response can be related following the Composite Noise Rating (CNR) approach EXPECTED COMMUNITY to the mean levels as given in Table 5. NOISE LEVEL RESPONSE It is interesting to note that the mean level of an intruding X, < Y0 No observed reaction source with a statistical time distribution may be higher than Y.