Effect of Vertical Confinement on Dynamic Cone Penetrometer Strength Values in Pavement and Subgrade Evaluations

Transcription

1 TRANSPORTATOfV RESEARCH RECORD 1473 Effect of Vertical Confinement on Dynamic Cone Penetrometer Strength Vales in Pavement and Sbgrade Evalations MOSHE LIVNEH, ILAN ISHAI, AND NOAM A. LIVNEH The Dynamic Cone Penetrometer (DCP) has become one of the most sefl testing devices in pavement evalation in Israel and other parts of.the world. Conseqently, a reliable tre meaning and inerpretation of the reslts are needed. Research work dedicated to stdying the effects of vertical confinement on the DCP strength vales of the granlar pavement layers and sbgrade is smmaried. Specifically, for major effects were stdied: vertical confinement of granlar layers, vertical confinement of cohesive layers, vertical confinement of rigid strc-. tral layers, and the effect of vertical confinement of pper asphalt layers on the DCP strength vales of the granlar layers below them. Based on engineering analysis and experimental testing in the laboratory and in the field, the following conclsions were made. (a) No vertical confinement effect by rigid pavement strctre, the pper granlar layers, or the pper cohesive layers on the DCP strength vales of lower cohesive sbgrade layers was fond. Any differences in the reslts between the confined and nconfined DCP vales can be explained by the friction that developed in the rod dring tilted penetration. (b) However, vertical confinement effects by pper asphalt layers on the DCP vales of the granlar pavement layers exist. Since this is the tre effect of the pavement strctre, any DCP measrement for pavement evalation prposes shold be performed throgh a narrow boring in the asphalt layers and not after removal of a wide strip of asphalt. Generally, these confinement effects decrease the DCP vales, and. ths increase the strctral strength measred. These confinement and friction effects, which can be evalated qantitatively, shold be taken into consideration when sing the DCP method to evalate existing pavements. drilling of a small-diameter hole in the asphalt strctre. These isses are considered to be related to vertical confinement. Since the DCP has become one of the most sefl testing devices in pavement evalation in Israel, a reliable tre meaning of the reslts and an interpretation of the reslts were reqired. Conseqently, research work was dedicated to stdying the effects of vertical confinement on the DCP vales. Specifically, the following effects were stdied: I. Effect of vertical confinement of granlar strctral layers on clay and silt sbgrade DCP vales; 2. Effect of vertical confinement of cohesive layers on clay sbgrade DCP vales; 3. Effect of vertical confinement of rigid strctral layers (an allasphaltic strctre) on clay sbgrade DCP vales; and 4. Effect of the vertical co.nfinement of pper asphaltic layers on the DCP vales of the granlar layers below them. The stdy of these effects is important for se of the DCP in the reliable evalation of existing pavements. This paper presents the theoretical backgrond for vertical confinement, as well as an analysis of the reslts obtained from both laboratory and field tests designed to stdy the effects mentioned earlier. Dring the last decade, the Dynamic Cone Penetrometer (DCP) test has been increasingly sed in many parts of the world for pavement and sbgrade evalation throgh its relationship with the in sit California Bearing Ratio (CBR). This is becase of its economy, simplicity, and capability of providing rapid measrements of the in sit strengths of sbgrades and pavement layers withot excavation of the existing pavement, as in the in sit CBR test (1-9). Extensive work with DCP testing and the experience gained in Israel have shown that, in addition to the CBR-DCP correlation, some other factors that have an inflence, sch as vertical confinements, shold be taken into consideration (J ). For example, a qestion occasionally arises, that is, whether DCP reslts, obtained from the sbgrade by means of the rod's penetration throgh the strctral layers, are identical to the reslts obtained from the same sbgrade after removal of the pavement strctral layers. In other words, are the sb grade DCP reslts affected by the presence of the flexible strctre? Similarly, the qestion can be applied to an all-asphalt pavement whose sbgrade was examined following Transportation Research Institte, Technion, Haifa 32, Israel. DCP TEST IN PAVEMENT EVALUATION The DCP, as developed in Soth Africa (J 1), consists of a steel rod with a cone at one end. It is driven into the pavement or the sbgrade with a sliding hammer, and the material resistance to penetration is measred in terms of millimeters per blow. The cone is angled at 3 degrees, with the larger diameter of the cone being 2 mm. The hammer weighs 8 kg, an the dropping sliding height is 575 mm (Figre 1). The DCP was originally designed and sed to determine the strength profile of the flexible pavement strctre and sbgrade (12-14). Usally, pavement testing at a given point involves the extrsion of a 4-in. circlar core from the top asphalt layers only and penetrating the DCP from the top of the base corse layer down to the reqired pavement or sbgrade layer. The properties of the asphalt layers are directly evalated in the laboratory by a proper mechanical test (resilient modls, diametrical test, splitting test, Marshall test, or others). The pavement parameters are continosly measred and recorded with depth by the DCP. Immediately at the test's conclsion, the shallow 4-in. hole is easily filled with either portland cement concrete (reglar or fast cring) or a proper cold asphaltic mixtre. In case of only sbgrade evalation for pavement

2 2 TRANSPORTATION RESEARCH RECORD 1473 NUMBER OF BLOWS HAMMER (NIL HAMMER MASS = 8 kg.) TWO PARTS SCREW TOGETHER MEASURING TAPE jf l6mm TEMPERED CONE JI 2omm FIGURE 2 DCP test otpt. FIGURE 1 General details of Soth African DCP. design prposes, the DCP is penetrated dowri from the top of the natral soil or compacted sbgrade. Dring testing the nmber of blows verss depth is recorded. The DCP vale is defined as the slope of the blows-verss-depth crve (in millimeters per blow) at a given linear depth segment (Figre 2). CORRELATION BETWEEN DCP AND CBR To be able to relate DCP vales to the strctral parameter of the pavement nder the local pavement design and evalation technology, an extensive controlled laboratory and field test was carried ot to correlate DCP to CBR (1,15). In the laboratory conventional and DCP tests were performed on a wide range of ndistrbed and compacted fine-grained soil samples with and withot satration: Compacted granlar soils were tested in flexible molds with variable controlled lateral pressres. Field tests were made on natral and compacted layers representing a wide range of potential pavement and sbgrade materials. Pavement evalation tests were also performed for pavement and material evalation and for correlation with pavement condition. The correlative laboratory and field testing program reslted in a qantitative relationship between the CBR of the material and its DCP vale as follows (Figre 3): log CBR = (log DCP)1. 5 (R 2 >.95) (1) where the DCP is the penetration ratio in millimeters per blow. This relationship, which was initially formlated in 1985, was based on 56 comparative test reslts. Later, this correlation was checked as data accmlated over several years (1). Finally, on the basis of 135 tests the total eqation, which was based on old and new data, is formlated as follows (Figre 3): log CBR = (log DCP)1. 5 (R 2 =.876, N = 135) (2) From a practical standpoint, both Eqations 1 and 2 yield almost identical reslts (1). Several other agencies and researchers arond the world have also tried to develop correlations between DCP and CBR vales (3,8,11,16). Webster et al. (8) compared some of these correlations (Figre 4). It is evident that general agreement was reached between 1 r:i 1 ci:::i DCP, mmblow FIGURE 3 Relationship between CBR vales and DCP vales (1,15).

3 Livneh eta!. 3 1 c:::=====r:==:::= :::c:::c=i=r:::r:::i:::===1===i====:::i==i:==i===i:=i=1=+========i r:i t--t-t-t--+--t t j::q DCP, mmblow WES DATA LIVNEH (1987) HARRISON (1987) VAN VUUREN (1969) KLEYN (1975) FIGURE 4 Universal data for relationship between DCP and CBR (8). the varios sorces of information. On the basis of these reslts the following eqation was selected as the best correlation: log CBR = (log DCP) (3) in which the ratio DI B is larger than 4 to 1. For the DCP device this ratio means depths of more than abot 1 cm. In this context it is important to show that the state of deep fondation is defined by Meyerhof (19) as follows: It can be seen that a niversal correlation exists between the DCP and CBR for a wide range of pavement and sb grade materials, testing conditions, and technologies. D = 4VNq,B where (4) (5) ENGINEERING BACKGROUND OF VERTICAL CONFINEMENT EFFECT In the context of the basic bearing capacity approach, a plastic failre mechanism can be adopted to describe soil behavior nder coneshaped penetrometers. This approach was adopted by Livneh (17) and Livneh and Greenstein (J 8) to find the effects of lateral pressre on CBR vales in granlar materials. Similarly, few analytical or nmerical soltions for wedge and cone-shaped penetrometers, which accont for both cone apex-angle and roghness, are given in the literatre (19-21). Drgnogl and Mitchell (2) proposed three types of failre mechanisms relevant to deep fondations that can also be adopted for penetration tests (Figre 5). On the basis of actal test reslts with variable apex-angles and cone roghness, they conclded that the failre mechanisms described in Figre 5(b) and 5(c) are inappropriate, whereas the failre mechanism represented in Figre 5(a) closely represents the actal failre srface assoiated with wedge penetration. On the other hand, Meyerhof ( 19) provided soltions for both cohesive and cohesionless sqils for certain conditions assming the failre mechanism shown in Figre 5(b), whereas Nowatki and Karafiath (21) sed a finite-difference techniqe for limited conditions and obtained some penetration resistance vales for cohesionless soils assming the failre mechanism shown in Figre 5(c). It shold be noted that this mechanism is only possible in the sitation and D = the fondation depth, B = the width or diameter of the fondation, and <!> = the material's internal angle of friction. Here, for cohesive soils (where<!> is eqal to ), Eqations 4 and 5 lead to the ratio of DIB eqal to 4, whereas for cohesionless soils (where<!> is> ) a deep fondation is defined at a depth of D greater than 4B. I I f I I I I I I I lqf 1 f T D (a) (b) (c) FIGURE 5 Different failre mechanisms proposed for penetration test (2). -

4 4 TRANSPORTATION RESEARCH RECORD 1473 To investigate the vertical confinement effect one shold look at the Teraghi's basic bearing capacity formla. Under this concept the penetration resistance q can be defined as follows: where q = the nit tip resistance, C = the nit cohesion, -y = the material density, D = the depth of the penetrometer tip, Nn N,., Nq = Teraghi's bearing capacity coefficients (Figre 6), and Kc, K,. = the shape factors. The qestion then asked is whether DCP test reslts obtained at a depth of 1 cm and more below the clay sbgrade, when the test is condcted in the clay sbgrade after drilling into the pavement, are different from the reslts obtained from a similar sbgrade at a depth of 1 cm and more below the srface, when the test is condcted following the removal of a wide strip of pavement. In other words are DCP test reslts obtained at a depth of 1 cm and more still sbject to the possible effects of vertical confinement? The answer according to Meyerhof (19) is negative when cohesive sbgrade is concerned. If one looks at Figre 6 (which presents the bearing capacity coefficients as a fnction of the angle of internal friction), it can be seen that for <I> eqal to, the vale of Ne remains constant, commencing from a depth eqal or greater than that expressed in Eqation 4. Brinch-Hansen (22) also provide a negative answer since, commencing from a certain vale of D, the increase in Ne as a conseqence of the contined increase in D is negligible. This is reflected in the following eqation:.35 de = 1. + B D (1 + 7tg 4 <!>) >O CT I () 1 -- Strip ( D<.B) --Sqare (D<B) -- - Pile CQ8>4-1)..._ 1 Lt: Ne v > v!::: "",,,' Ne <,,.. v, <!) a: < IJJ CD V'" N'. '. v, 1 v N-g I VJ '... v I, J,,, I I I ANGLE OF INTERNAL FRICTION, JO. I, I, A, J. ff, :-7,,. IW' v ' FIGURE 6 Bearing capacity factors for spread and pile fondations (16). (6) (7) where de is the increase factor of Ne depending on the fondation depth. Theoretically, vertical confinement is mainly possible becase of the bearing capadty.component -ydnq (Eqation 6). In cohesive soils, when <I> is eqal to, Nq is eqal to 1., and ths, this component is qite negligible. Hence, it can be seen that for clay sbgrades the vales of both Ne and Nq in Eqation 6 eliminate the effect of vertical confinement in deep penetration measrements of the DCP. On the other hand, for granlar materials (where <I> is> ), both Nq and Ne possess significant vales that increase with depth. Therefore, the vertical confinement of the pper layers will be significant and will increase with the depth of the DCP tip. This sitation of a very deep fondation is similar to a position in which the fondation is not as deep bt the soil srface is bond by a rigid strctre (as, for example, in an all-asphalt strctre). On the basis of the preceding discssion, it can be conclded that the DCP reslts obtained from a penetration depth of 1 cm or more from the srface of the clay sbgrade will be only negligibly affected by the presence of a rigid strctre on the sbgrade's srface or will be totally naffected. These conclsions are derived from the engineering analysis, and they are compared with the empirical testing described later in this paper. EFFECT OF CONFINEMENT BY. GRANULAR STRUCTURE The effect of confinement by the granlar strctre on the sbgrade' s DCP.vales was stdied on the basis of an analysis of actal DCP test data (1). The analysis presented in this section is based on 11 DCP test locations. The DCP test was condcted following the drilling of asphalt cores at these points. The sbgrade DCP vales obtained by this method are referred to as the confined vales, that is, DCPcon These vales are compared with the DCP vales derived from the sbgrade vales described earlier following removal of all strctral layers (i.e., after the digging of test pits). The DCP vales obtained from the exposed sbgrade are referred to as the nconfined vales, that is, DCP nc The test reslts obtained are presented in Figre 7. It can immediately be seen from Figre 7 that the two.test poplations are not identical, as also verified statistically. Ths, for example, a statistical t-test of the reslts showed that the mean nconfined DCP (x) is larger than the mean confined DCP ()i) at a confidence level of 95 percent. Moreover, a linear y = a + bx type of regression analysis leads to the following reslts: a= 21.3 (Sa= 6.42), b =.13 (Sb=.158), R 2 =.45, and CT 2 = where Sa and Sb = standard deviations for a and b, respectively; CT 2 = mean sqare error (MSE); and R = correlation coefficient. A statistical F-test of this regression leads to the conclsion that at a confidence level of 95 percent, the regression is not significant or, in other words, that there is no correlation between x vales and y vales. The analysis according to Grbbs (23) leads to the conclsion that the systematic error related to measring x vales is sig-

5 Livneh eta!. 5 i:l'; :: Cl 3 e; 2 Cl 1 a Clay li!i Silt 1 2 Constrained regression throgh graph origin DCP UNCONFINED, mm I blow FIGURE 7 DCP test reslts for sbgrade beneath granlar strctre. nificantly different from the systematic error related to measring y vales. At the same time, the precisions of the readings in both tests (as expressed by the standard deviation of the random error) can be considered identical. This analysis proves once again that the two test poplations are not identical, and ths indicates the possible existence of vertical confinement. Now, attention shold be given to the two silt points in Figre 7. Since in the silt 4> is > on the basis of the discssion in the engineering backgrond, it is conceivable that the two test points in the silt are indeed related to the phenomenon of vertical confinement, whereas there is no reason for the confinement effect to take place in the rest of the test points, which are in clay. For this reason the character of the two poplations shold be reexamined following exclsion of the two silt test points (marked separately in Figre 7). The hypothesis that the two silt points do not belong to the poplation was examined by similar linear regression withot the two points described earlier. The regression led to the following parameters: ca a= 12.1 (Sa= 3.199), b =.418 (Sb=.86), R 2 =.771, and ' 2 = 9.5 An F-test of this regression leads to the conclsion that at a confidence level of 95 percent the model is significant and a linear correlation between x and y does exist. The conclsion is that the two points do not belong to the same poplation, becase withot them there is a significant linear correlation. Finally, it can be assmed that the difference between the two poplations (x,y) is a reslt of the friction created in the rod dring tilted penetration of the DCP rod throgh the granlar material (as will be discssed later or of the friction created in the rod dring a collapse of the granlar material on the rod srface dring penetration. A constrained regression throgh the graph origin was condcted to obtain the extent of the friction's effect on the reslts, and the following correlation was obtained: y =. 73x. This leads to the following relationship: DCPnc = 1.34 DCPcon (8) 6 In other words when the DCP test is condcted throgh granlar layers, an error of 34 percent magnitde may occr. In this specific case the DCP vales of the clay sbgrade shold be increased by abot 34 percent to obtain the tre redced strength of the material beneath the granlar strctre. Finally, it shold be mentioned that the effect of friction created in the rod (either by tilted penetration or by material collapse) can be qantitatively evalated by torqe measrement in the DCP device dring penetration intervals. This work is being done and_ will be reported in the near ftre. EFFECT OF CONFINEMENT BY CLAY LAYERS It is also important to examine whether condcting the DCP test on the srface of the clay sbgrade and condcting the DCP test throgh this sb grade (commencing from a certain depth) lead to identical reslts. The reslts of the 27 tests are presented in Figre 8. The method was to condct a DCP test on the srface and then dig a.5-m-deep test pit and. condct the DCP test again, commencing from the pit bottom downward. The DCP vale in the first test, which corresponds to a depth of.5 to. 7 from the soil srface, is compared with a DCP vale in the second test, which corresponds to a depth of. to.2 from the pit bottom. The first test provides the vale of the confined DCP (the x vales), and the sec;ond test provides the vale of the nconfined DCP (they vale). To test whether the two test poplations are identical, a statistical analysis similar to that described earlier was performed. According to the statistical t-test reslts, both poplations are identical. However, this reslt shold be accepted with cation becase there is no assrance that in the test poplation (derived from varios sites with a wide range of strengths) the condition reqiring that the standard deviation from the mean does not change with the change in the nmber of sites is indeed obtained. 6 i:l'; 5 :: 4 Cl 3 a Line of eqality a 2 :::> ej Cl 1 a DCP CONFINED, mm I blow FIGURES Correlation between confined and nconfined DCP within clay sbgrades. a a

6 6 TRANSPORTATION RESEARCH RECORD 1473 A regression analysis of the linear y = ax + b type leads to the following reslts: a= 2.86 (Sa= 3.75), b =.98 (Sb=.16), R 2 =.59, and cr 2 = Examination of the confidence interval for the parameters a and b at a confidence level of 95 percent leads to the acceptance of the hypothesis that a eqals ero and b eqals 1, that is, that there is no additive or mltiplicative bias. Hence, the calclated vales for a and b, 2.86 and.98, respectively, are not significant. An analysis by the method of Grbbs (23) also indicates that both the systematic rror and the standard deviation of the one poplation's random error are identical to those of the other poplation. The final conclsion is that despite the variations shown in Figre 8, it seems that the two poplations are identical. In other words the effects of the penetrating rod's srface friction are not felt in this case. This is logical, since in clay material (in contrast to granlar material) the chances of preserving the gap between the sides of the hole created with the cone's penetration and the penetrating rod's srface, whose radis is smaller, are better. EFFECT OF CONFINEMENT BY RIGID STRUCTURE An examination of the DCP vales of a clay sbgrade confined beneath a 5-cm-thick all-asphaltic pavement has yielded deviant strength reslts compared with the CBR vales obtained by direct testing. The DCP test was condcted in the sbgrade following drilling of the core for the entire depth of the all-asphalt strctre. The res.lts are presented in Figre 9. When the regression is restrained throgh the graph ero of Figre 9, the eqation is: CBRv = 1.75 CBRs (9) or DCPs = 1.55 DCPv (1) where CBRv is the CBR vale calclated from the DCP test condcted throgh the all-asphalt strctre, and CBRs is the direct CBR vale. To tet whether the increase in strength preented in Eqations 9 and 1 is the reslt of vertical confinement cased by the rigid allasphalt strctre, a special series of tests was condcted in the laboratory as described in the next paragraph. In the test series clay samples (liqid liniit, 66 percent; plasticity index, 43 percent} were compacted by 12 blows (4 samples) and 26 blows (18 samples) in the standard CBR mold. The moistre contents were higher than the plasticity limit. These clay samples were not satrated with water. Each pair of samples was tested with the DCP instrment both withot vertical pressre (free top srface) and with the srface restrained by means of a thick metal plate with a 3-cm-diameter hole. At the end of the test it was clearly observed that when the sample was nrestrained there were vertical movements and cracking arond the area of penetration, whereas in the case of the restrained samples, the pper soil srface remained smooth and ncracked and withot any vertical movement. All of the varios statistical tests indicate that restraining the samples does not affect the DCP reslts. This can also be seen from the plotted reslts presented in Figre 1. The regression model leads to the following parameters: a= 1.26 (Sa= 3.51), b = 1.1 (Sb=.114), R 2 =.9, and cr 2 = 8.73 For a confidence level of 95 percent, a eqals ero and b eqals 1. This analysis does not follow the trend of reslts presented in Figre 9. The explanation for this can be given by assming that friction forces do develop along the penetrating rod as a reslt of nonvertical penetration rather than becase of the artificial restraining of the clay in the laboratory or its restraint by the all-asphalt pavement at the site. To test this assmption a series of tests was condcted in which the rod was inserted both vertically and at an angle of 15 degrees.. U::r;: Q e: ::r:: E- or.i) 2 ::s ::::> IXl 1 15 o- ::::> 1 E- < ::r:: E- E- ::r:: IXl Q ll.. b fl) fl) 4= 1 Constrained regression throgh graph origin DD 5 >o 1 E- IXl E- E_ DIRECT CBR VALUE IN % DCP RESTRAINED SURFACE, mm I blow FIGURE9 CBR reslts from clay sbgrade beneath 5-cmthick all-asphalt strctre. FIGURE 1 DCPvales. Effects of restrained sample srface on

7 Livneh et al. 7 The statistical analysis indicates that there is a difference between the two tests. The mean CBRv vale for the vertical penetration is 6.6 percent, whereas the CBR, vale for penetration at a 15-degree angle is 8.9 percent. A constrained regression throgh the origin prodces the following correlation (Figre 11): CBRv =.78 CBRr (11) The meaning of this above regression is that the DCP vale decreases (the strength increases) when the penetrating rod is inserted at an angle. This fact can be sed to explain the reslts presented in Figre 9, since in many cases nder actal deep DCP penetration, the rod tends to tilt at an angle ofp to 15 degrees. It shold be noted that the tilted penetration can be avoided by penetrating the DCP rod throgh a vertical spporting frame. This is accomplished by the reglar DCP device or by an atomated DCP device (24) :: 12 1 I 8 6 :::> 4 2 EFFECT OF CONFINEMENT BY ASPHALT LAYERS The effect of confinement by the asphalt layers on the granlar pavement layers was examined by condcting DCP tests on the base corse materials after drilling an asphalt core and comparing the reslts with those obtained from performance of the DCP test on these materials following the removal of a wide asphalt strip from the same spot tested previosly. In contrast to the case described in the preceding section, one can expect here an increase in the normal DCP vale compared with the confined DCP vales in granlar materials, since the bearing capacity factor, Nvq is far greater than 1. Indeed, the reslts obtained indicate the following correlation, as can also be seen in Figre 12: DCPnc = 1.84 DCPcon (12) The difference between the DCP vales is statistically significant, and ths, the asphalt does indeed have a confining effect on the state - 5 :i..j :!:! 1 < DCP AT A PERPENDICULAR PENETRATION, mmblow FIGURE 11 DCP vales from vertical penetration and penetration at a 15-degree angle DCP CONFINED, mm I blow FIGURE 12 Effects of asphalt confinement on DCP vales in granlar material. of the granlar material. Since this effect has been shown to exist, it is now appropriate that the strength vales of the granlar material in the CBR or DCP field tests shold be determined only after the drilling of asphalt cores and not after the removal of a wide strip of asphalt. SUMMARY The DCP has become one of the most sefl testing devices in pavement evalation in Israel and other parts of the world. Conseqently, a reliable tre meaning of the reslts and an interpretation of the reslts were needed. This paper smmaried research work that was dedicated to stdying the effects of vertical confinement on the DCP vales of the sbgrade and the granlar pavement layers. Specifically, for major effects were stdied: vertical confinement of granlar layers, vertical confinement of cohesive layers, vertical confinement of rigid strctral layers, and the effect of vertical confinement of pper asphalt layers on the DCP vales of the granlar layers beneath them. On the basis of the theoretical analysis, experimental testing in the laboratory and in the field, and statistical analysis, the following conclsions were made. 1. No vertical confinement effect by rigid pavement strctre or by pper cohesive layers on the DCP vales (or strength) of lower cohesive sbgrade layers exists. Also, no vertical confinement effect by the pper granlar layers on the DCP vales of the cohesive sb grade beneath them exists..any difference in the reslts between the confined and nconfined DCP vales in the rigid strctre case or in the case of the granlar layers can be explained by the friction that developed in the DCP rod by tilted penetration or by a collapse of the granlar material on the rod srface dring reglar penetration. This friction effect, which is also a fnction of the ratio between the cone tip and the rod diameters, can be qantitatively evalated by torqe measrements in the DCP device dring penetration intervals.

8 8 TRANSPORTATION RESEARCH RECORD A vertical confinement effect by the pper asphaltic layers on the DCP vales of the granlar pavement layers does exist. Generally, these confinement effects case a decrease in the DCP vales, ths increasing the strctral strength measred. Since this is the tre effect of the pavement strctre, any DCP measrements for pavement evalation prposes shold be performed throgh a narrow boring in the asphalt layers and not after removal of a wide strip of asphalt. 3. The vertical confinement effects and the friction effects described earlier, which can be qantitatively evalated, shold be taken into consideration when sing the DCP method in pavement and sbgrade evalation. ACKNOWLEDGMENTS This work was condcted as part of a comprehensive project on the carrying capacity of nsrfaced rnways financed by the U.S. Air Force Engineering & Services Center, Tyndall Air Force Base, Florida. The athors thank that organiation for the sponsorship and aid. Thanks are also de to the Israel Air Force, the Technion Soils & Roads Testing Laboratory, and also to A. Aines and F. Hirosh for their technical and administrative assistance. REFERENCES 1. Kleyn, E.G., J. H. Maree, and.p. F. Savage. The Application of a Portable Pavement Dynamic Cone Penetrometer to Determine In-Sit Bearing Properties of Road Pavement Layers & Sbgrades in Soth Africa. Proc., 2nd Eropean Symposim on Penetration Testing, Amsterdam, May Smith, R. B., and D. N. Pratt. Field Stdy of In-Sit California Bearing Ratio and Dynamic Penetrometer Testing for Road Sb grade Investigation. Astralian Road Sbgrade Investigation. Astralian Road Sbgrade Investigation, Astralian Road Research oard, Vol. 13, No. 4, p Harrison, J. A. Correlation Between the California Bearing Ratio and the Dynamic Cone Penetrometer Strength Measrements of Soils..Proc., Institte of Civil Engineers, Part 2.83, Technical Note 463. Institte of Civil Engineers, Operating Instrnientationfor the TRRL Dynamic Cone Penetrometer. Information Note. Transport and Road Research Laboratory, Crowthorne, Berkshire, United Kingdom, Livneh, M., and I. Ishai. Pavement and Material Evalation by a Dynamic Cone Penetrometer. Proc., 6th International Conference on the Strctral Design of Asphalt Pavements, Ann Arbor, Mich. Jly 1987, pp Ch, K. M. Determination of CBR and Elastic Modls of Soils Using a Portable Pavement Dynamic Cone Penetrometer. Proc.,!st International Symposim on Penetration Testing-ISOPT-1, Orlando, Fla., March Bncher, M. S., and D. J. Christiansen. USAF New Contingency SoilsPavement Testing Van. ASCE Special Pblication on Road and Airport Pavement Response Monitoring Systems. ASCE, 1991, pp Webster, S. L., R.H. Gra, and T. P. Williams. Description and Application of Dal Mass Dynamic Cone Penetrometer. Instrction Report GL U.S. Army Engineers Waterways Experimental Station, May de Beer, M. Developments in the Failre Criteria of Soth African Mechanistic Design Procedre for Asphalt Pavements. Proc., 7th International Conference on Asphalt Pavements, Vol. 3, 1992, pp Livneh, M., I. Ishai, and N. A. Livneh. Carrying Capacity of Unsrfaced Rnways for Low Volme Aircraft Traffic. Phase III. Application of the Dynamic Cone Penetrometer. Preliminary Report. Transportation Research Institte, Technion, Haifa, Israel, April Kleyn, E.G. The Use of the Dynamic Cone Penetrometer (DCP). Report 274. Transvaal Roads Department, Kleyn, E.G., and M. J. J. Van Neerden. Using DCP Sondings to Optimie Pavement Rehabilitation. Presented at Annal Transportation Convention, Soth Africa, Kleyn, E. G., and P. F. Savage. The Application of the Pavement DCP to Determine the Bearing Properties and Performance of Road Pavements. Report L Transvaal Roads _Department, Soth Africa, Kleyn, E.G., and P. F. Savage. The Application of the Pavement DCP to Determine the Bearing Properties and Performance of Road Pavements. Proc., International Symposim on Bearing Capacity of Roads & Airfields, Trondheim, Norway, Livneh, M. Correlation Between the DCP and CBR Vales. Pblication Transportation Research Institte, Technion, Haifa, Israel, Van Vren, D. J. Rapid Determination of CBR with the Portable Dynamic Cone Penetrometer. The Rhodesian Engineer, Vol. 7, No. 5, 1969, pp Livneh, M. The CBR Test with Lateral Pressres. Proc., 3rd Asian Regional Conference on Soil Mechanics and Fondation Engineering, Vol. 1, 1967, pp Livneh, M., and J. Greenstein. A Modified CBR Test for Granlar Materials. Geotechnical Testing Jornal, ASTM, Vol. 1, No. 3, 1978, pp Meyerhof, G. G. The Ultimate Bearing Capacity of Wedge-Shaped Fondations. Proc., 5th International Conference on Soils Mechanics & Fondation Engineering, Vol. 2, Drnnogl, H. T., and J. K. Mitchell. Inflence of Penetrometer Characteristics on Static Penetration Resistance. Proc., Eropean Symposim on Penetration Testing, Stockholm, Sweden; Nowatki, E. A., and L. L. Karafiath. The Effect of Cone Angle on Penetration Resistance. In Highway Research Record 45, HRB, National Research Concil, Washington, D.C., Brinch-Hansen, J. A General Formla for Bearing Capacity. Blletin 11. Danish Geotechnical Institte, Copenhagen, Denmark, Grbbs, F. E. Errors of Measrements, Precision, Accracy and the Statistical Comparison of Measrement Instrments. Technometrics, Vol. 15, No. 1, Feb. 1973, pp Livneh, M., I. Ishai, and N. A. Livneh. Atomated DCP vs. Manal DCP in Carrying Capacity of Unsrfaced Rnways. Project Transportation Research Institte, Technion, Haifa, Israel, 1992, 42 p. Pblication of this paper sponsored by Committee on Strength and Deformation Characteristics of Pavement Sections.