PP Rahardjo & T Lunne (eds) In Situ Measurements and Case Histories, Bandung, Indonesia, 21, pp 649 654 Horizontal Cone Penetration Testing in Sand W Broere Geotechnical Laboratory, Delt University o Technology, The Netherlands AF van Tol Rotterdam Public Works, The Netherlands Geotechnical Laboratory, Delt University o Technology, The Netherlands ABSTRACT: The cone resistance and sleeve riction measured when a penetration test is executed in the horizontal direction dier rom those obtained when sounding in the traditional vertical direction In order to quantiy the dierences a number o tests has been executed in a 2m diameter rigid wall calibration chamber on sands with dierent densities and dierent gradations These tests have shown that the dierence between horizontal and vertical cone resistance is greatest or medium dense sands, but show little dependancy on the gradation o the sand The ratio o horizontal over vertical sleeve riction on the other hand shows little variation or dierent densities, but is strongly dependant on the coarseness o the sand 1 INTRODUCTION The cone penetration test (CPT) is traditionally executed in a downward vertical direction in order to obtain inormation about the soil properties and stratiication The interpretation o the obtained measurements is subsequently made using analytical and empirical models, as well as experience, which all implicitely or explicitely use the assumption that the penetration direction is vertical This is in general not regarded as a problem and the measures taken to limit the delection rom the vertical during penetration, or to measure this delection and correct or it, do not originate rom the possible problem in the interpretation o cone resistance Those measures are taken because the delection might lead to unacceptable errors in the depth registration and thereby the stratigraphy The introduction o mechanised tunnel boring in sot and strongly heterogeneous soils, as ound in many delta-areas over the world, has given rise to a need or more detailed inormation along the alignment o the tunnel At the same time the possibility has been created to obtain this inormation using cone penetration tests originating rom the tunnel boring machine in a horizontal orward direction, and it has been recognised that this could complement the inormation gained rom vertical CPTs Initiatives have also been unurled to execute horizontal CPTs through retaining walls o building pits to investigate the soil conditions below adjacent buildings, or through existing railway embankments and land ills, without disturbing the topside actvities Now the equipment used to perorm a vertical CPT can be easily converted to execute a horizontal cone penetration test (HCPT) The interpretation is not so easily converted however It has been recognised by Houlsby & Hitchman (1988) that the eective horizontal stress σ h is the controlling stress component in traditional CPT calibration chamber tests on sand In vertical CPT this is the stress that acts in the plane perpendicalur to the penetration direction, and in this plane can be considered a radially uniorm stress component In the case o HCPT the stress state perpendicular to the cone is not radially uniorm, as it will vary between σ h and the eective vertical stress σ v Combined with the act that most soils have been deposited in a layerwise manner, it is to be expected that the measurements obtained rom HCPT dier rom those in vertical CPT 11 HCPT experiences This dierence between horizontal and vertical CPT has already been observed in calibration chamber tests as well as in ield tests The calibration chamber tests by Broere & van Tol (1998) were perormed in a uniormly distributed ine sand and showed a horizontal cone resistance q H c greater than the vertical cone resistance q V c at the same depth The horizontal cone resistance was on average 12 times the vertical or medium densiied sands, whereas the ratio was closer to one or very loose or dense sands The horizontal riction ratio on the other hand was clearly lower than the vertical riction ratio, but showed little dependancy on the density o the sand A similar image can be derived rom the ield measurements by van Deen et al (1999) The main part o the tests was executed in sot clay and peat layers 1
PP Rahardjo & T Lunne (eds) In Situ Measurements and Case Histories, Bandung, Indonesia, 21, pp 649 654 instead o sands, but shows similar trends The horizontal cone resistance measured was greater than the vertical, even up to three times greater in clay And again the horizontal riction ratio was on average lower than the vertical And although on one hand these results strengthen the results rom the calibration chamber tests, they also indicate that there may be an inluence o the soil type 12 Reseach aims It has been recognised early on that the measurements rom CPT in sand depend not only on the stress level and (relative) density, but amongst others also on the grain size distribution o the material and the amount o ines present See or example the overview o soil classiication charts presented by Lunne et al (1997) or the comparison o numerous calibration chamber tests by Jamiolkowski (1988) It is also very likely that HCPT measurements show a similar dependancy on the gradation o sand, but this dependancy is not necessarily exactly equal Thereore the ratio o horizontal over vertical cone resistance (qc H/V ) and the ratio o horizontal over vertical sleeve riction ( H/V S ) may also depend on the soil type In order to investigate this behaviour a number o tests has been executed in the large rigid wall calibration chamber o Delt University o Technology (DUT) These tests have been limited to sand samples, as the preparation o large, homogeneous, cohesive samples in a calibration chamber is diicult and timeconsuming Three dierent sands have been used, which have been prepared at various densities Within each sample two vertical and one horizontal CPT has been made and the measurements at the depth o the HCPTh have been compared 2 THE CALIBRATION CHAMBER The DUT calibration chamber is a 2m diameter rigid wall calibration chamber, as sketched in igure 1 At the bottom o the tank a system o ilter drains connected to a pumping system is embedded in the sand This system allows luidisation o the sand bed within the tank A couple o vibrators aixed to the sides o the tank can be used to vibrate the entire tank and thereby densiy the sand To prepare a sample the sand is irst luidised and, when ully liquiied, the water is allowed to drain The vibrators are then used or a period o to 8 minutes, dependant on the relative densiy o the sand that is required Ater that the remaining water is allowed to drain and the density o the sample is determined by measuring the top level o the sand The preparation method used, luidisation instead o the more common pluviation, is one o the main dierences between the DUT calibration chamber and most other chambers The great advantage o this 323 78 24 23 ztop Vibrator Upper sounding lock Lower sounding lock 19 Figure 1 DUT calibration chamber Fluidisation system method is that the preparation o the sample is so labour-extensive; a new sample can be prepared each day, requiring only one man-hour actual work during this period The disadvantages are that the samples are somewhat less homogeneous over the height o the sample than pluviated samples, that due to the luidisation process ines may slowly be washed out o the sand used, and that the obtained sample is unsaturated but not dry, and that the level o saturation depends on the permeability o the sand and the draining time This last problem can be circumvented completely, as the chamber allows the testing o ully saturated samples as well, but this option has not been used in this test series The other major dierence is that the DUT chamber is a rigid wall calibration chamber, meaning that the lateral boundaries are sti and disallow any deormation o the sample This in contrast to the oten used lexible wall chambers, where a pressurised membrane is used at the lateral boundaries to keep a constant lateral pressure As is common the lower boundary is the sti chamber loor, whereas at the top an overburden load could be applied The sample is large enough however that reasonable stress levels are reached at the depth o the HCPT opening without an overburden The result o this combination o boundaries is that the DUT chamber alls somewhere between BC2 & 3 as given by Parkin (1988) as: 2
PP Rahardjo & T Lunne (eds) In Situ Measurements and Case Histories, Bandung, Indonesia, 21, pp 649 654 ine middle coarse sand ine middle coarse sand passed sieve (%) 1 9 8 7 6 5 4 3 2 1 1 2 3 5 1 2 5 1 2 sieve aperture (mm) Figure 2 Sieve curves or all sands Table 1 Minimal and maximal densities Sand e min e max 1 498 81 2 454 749 3 431 746 B1 σ v, σ h constant B2 ǫ v = ǫ h = B3 σ v constant, ǫ h = B4 σ h constant, ǫ v = This implies according to Salgado (1998) that the cone resistance q c measured in the tank is higher than would be measured in the same conditions in the ield At the chamber diameter-to-cone diameter ratio R d = 56 this is only a ew percent however and the eect can be neglected A urther special eature o the tank is o course the presence o a lock on the side wall o the tank, as sketched in igure 1 This lock allows a horizontal penetration to be made using a standard 35mm cone 3 THE SANDS Three dierent sands have been used, which dier in their gradation and coarseness The irst sand is a uniormely distributed ine sand o alluvial origin This batch o sand had been used beore in the same calibration chamber, as a result o which most ines had been washed out The second and third sand type has been obtained by mixing this alluvial sand in dierent proportions with a commercially available coarse river sand, which had been washed to remove all ines The sieve curves o the resulting sands are shown in igure 2 For these sands the minimal and maximal densities have been obtained by pouring dry sand through a unnel respectively vibrating and compacting a moist sample or an extended period o time The resulting e min and e max are listed in table 1 passed sieve (%) 1 9 8 7 6 5 4 3 2 1 1m depth 15m depth 2m depth 5 1 2 5 1 2 sieve aperture (mm) Figure 3 Sieve curves or sand 2 at dierent depths; inluence o segregation During the preparation o sands 2 and 3 a certain amount o segregation is observed due to the luidisation process, as the iner particles tend to loat upwards As a result the sand in which the HCPTs are made is somewhat coarser than would be derived rom igure 2 This can be seen in igure 3, where the sieve curves rom samples o sand 2 at three dierent depths are plotted The uppermost layer clearly diers rom the lower layers, although the lower layers do not dier strongly rom the overall sieve curve The correlation between vertical and horizontal CPT measurements is made or the 2m depth level only and the sand in this region does not show any local segregation inluences exhibits the same behaviour (not shown), although the dierences between the overall gradation curve and those o the lower segregated layers are even smaller than or sand 2 on the other hand is so uniorm that no segregation eects can be observed at all over the height o the tank The overall eect o this segregation is that the relative density D r calculated or samples o sand 1 has an estimated error o 2%, whereas or sands 2 and 3 this error is closer to 5% This error is o the same magnitude as in most calibration chamber tests and not considered a problem 4 OVERVIEW OF TEST SERIES For each o the three sands 1 samples have been prepared, with dierent vibration times and as a result dierent relative densities Within each sample two vertical penetrations were made ollowed by one horizontal, at locations as sketched in igure 4 Although sketched here in the same vertical plane, the path o the horizontal penetration lies in a plane at a 1 angle with the plane o the vertical penetrations As a result the path o the horizontal penetration passes the paths o the vertical penetrations at approximately 1cm 3
PP Rahardjo & T Lunne (eds) In Situ Measurements and Case Histories, Bandung, Indonesia, 21, pp 649 654 55 4 4 55 24 23 198 544 86 qc H /σ v 8 7 6 5 4 Figure 4 Locations o vertical and horizontal CPTs 3 2 18 16 14 12 1 8 6 4 2 qc H (MPa) 2 4 6 8 1 12 14 16 18 2 qc V (MPa) 1 5 15 2 1 1 2 3 4 5 6 7 8 q V c /σ v Figure 6 Horizontal vs vertical cone resistance 1 2 z (m) 15 x (m) s H /σ v 8 Figure 5 Example o horizontal and vertical cone resistance registration (, D r = 43) 7 6 An example o the resulting cone resistance registrations is given in igure 5 or a relative density D r = 43 The vertical depth has been plotted with the top o the tank as the reerence level z = The horizontal position is plotted relative to the inside o the tank wall x = From this igure it can be seen that at least at low densities there is no visible inluence o the vertical penetrations on the horizontal one At very high densities a limited reduction o the horizontal cone resistance due to the retraction o the vertical cones can be observed This eect can easily be corrected or, as the spatial inluence o this eect is limited and the horizontal cone resistance outside this zone is not inluenced For the locations o closest passage the cone resistance and sleeve riction have been determined The measurements have been normalised by the eective vertical stress σ v and are plotted as horizontal vs vertical measurement in igures 6 and 7 5 4 3 2 1 1 2 3 4 5 6 7 8 s V /σ v Figure 7 Horizontal vs vertical sleeve riction 4
PP Rahardjo & T Lunne (eds) In Situ Measurements and Case Histories, Bandung, Indonesia, 21, pp 649 654 q H/V c 2 H/V s 4 15 1 35 3 25 2 15 5 1 2 3 4 5 6 7 8 9 1 D r Figure 8 Ratio o horizontal over vertical cone resistance vs relative density 5 HORIZONTAL CONE RESISTANCE From igure 6 it can be seen that although there is some dispersion, on average the horizontal cone resistance is greater than the vertical This becomes even clearer when the ratio qc H/V is plotted vs relative density, as in igure 8 In that case the results correspond well with earlier indings by the authors (Broere & van Tol 1998) that horizontal cone resistance is slightly larger than vertical, although the observed trend that the ratio qc H/V is largest or intermediate densities is not substantiated by this data set When the separate sand types are considered there is no clear inluence on the cone resistance ratio The average o qc H/V per sand type increases slightly rom 15 or sand type 1 to 11 or sand type 3, but the dispersion and errors in the data are such that this is not a signiicant dierence As such there is no evidence o an inluence o the gradation or coarseness o the sand on the cone resistance ratio qc H/V 6 HORIZONTAL SLEEVE FRICTION In contrast to the cone resistance, the horizontal sleeve riction as well as the riction ratio show a clear inluence o the sand type This can already been seen in igure 7, but becomes even more evident when the the ratio o horizontal over vertical sleeve riction and the ratio o horizontal over vertical riction ratio are considered See igure 9 or a plot o sleeve rictions ratio vs relative density Although the dispersion in the data is large, especially or sand 1, there is a signiicant decrease o s H/V with increasing grain size d 5 The mean s H/V H/V s per sand type is equal to 128 or sand 1, 79 or sand 2 and 65 or sand 3 With an average cone resistances ratio greater than one, this trend is even more 1 5 1 2 3 4 5 6 7 8 9 1 D r Figure 9 Ratio o horizontal over vertical sleeve riction vs relative density R H/V 4 35 3 25 2 15 1 5 1 2 3 4 5 6 7 8 9 1 D r Figure 1 Ratio o horizontal over vertical riction ratio vs relative density pronounced when the riction ratios are considered (R H/V are 12, 77 and 6 resp) Given the distribution o the data and assuming that these ratios have a normal distribution around their means, there is a less than 1% chance that those dierence are caused by a dispersion in the data The averages or sand 1 decrease somewhat when the two sleeve riction ratios > 2 are neglected, to 111 and 11, but even in that case the dierences remain statistically signiicant This decrease in riction ratios ratio R H/V is caused by simultaneous changes in the separate riction ratios Figure 11 shows the horizontal riction ratio plotted against the vertical riction ratio From this igure it becomes clear that the relative decrease o R H/V between sand 1 and 2 is caused by a decrease o the horizontal riction ratio that is stronger than the decrease in the average vertical riction ratio The urther decrease between sand 2 and 3 on the other hand is caused by an increase o the average vertical riction ratio only 5
PP Rahardjo & T Lunne (eds) In Situ Measurements and Case Histories, Bandung, Indonesia, 21, pp 649 654 R H (%) 1 The cone resistance and sleeve riction measured in a horizontal cone penetration test dier rom those obtained with vertical tests The act that or medium dense sands the average value o q c measured in HCPT is 12 times the value measured in vertical CPT had been recognised already, but this observation had been made or a ine uniormely distributed sand only To investigate the inluence o the grain size distribution on HCPT measurements new tests have been perormed in three dierently graded ine to coarse sands Within each sand type only a limited number o tests has been perormed These tests show no inluence o the sand type on the ratio o horizontal over vertical cone resistance They do show however a strong inluence on the ratio o horizontal over vertical sleeve riction, as the ratio s H/V decreases signiicantly with increasing coarseness o the sand The same is true or the ratio o horizontal over vertical riction ratior H/V This decrease is caused by changes in the horizontal as well as the vertical riction ratio, and these changes are non-linear with the change in characteristic grain size or distribution No satisactory explanation o this dependancy on the sand type has been ound Given the limited amount o sand types tested the question whether such behaviour also occurs or dierent sand types, or or dierent soil types in general, remains unanswered The research shows however that in detailed soil analyses based on horizontal cone penetration the correlation charts developed or vertical CPT should be used with care 5 5 1 15 2 Figure 11 Horizontal vs vertical riction ratio R V (%) And although the inluence o the coarseness and gradation o the material on the sleeve riction is well known, the observed inluence on the horizontal sleeve riction and especially on the horizontal riction ratio remains surprising Ater all, the sands 2 and 3 exhibit a clearly dierent riction ratio, whereas they have the same horizontal riction ratio I this phenomena also occurs or other soil types, the characterisation o soils based on their horizontal riction ratio becomes more diicult and any calibration charts or vertical CPT should be used with extreme care in the detailed interpretation o HCPT 7 CONCLUSIONS ACKNOWLEDGEMENTS The authors wish to thank Mr KSMK Kosgoda or his help in perorming the tests REFERENCES Broere, W & AF van Tol 1998 Horizontal cone penetration testing In Robertson, PK & PW Mayne (eds), Geotehnical Site Characterization, Proc ISC 98, pp 989 994 Balkema Deen, JK van, G Greeuw, R van den Hondel, MTh van Staveren, FJM Hoesloot & B Vanhout 1999 Horizonal CPTs or reconnaissance beore the TBM ront In Barends, FBJ, J Lindenberg, HJ Luger, L de Quelerij & A Verruijt (eds), XII ECSMGE Geotechnical Engineering or Transportation Inrastructure, pp 223 23 Rotterdam, Balkema Houlsby, GT & R Hitchman 1988 Calibration chamber tests o a cone penetrometer in sand Géotechnique, 38(1):39 44 Jamiolkowski, M, VN Ghionna, R Lancelotta & E Pasqualini 1988 New correlations o penetration tests or design practice In Ruiter, J de (ed), Penetration Testing 1988, pp 263 296 Lunne, T, PK Robertson & JJM Powell 1997 Cone Penetration Testing in Geotechnical Practice London, Blackie Parkin, AK 1988 The calibration o cone penetrometers In Ruiter, J de (ed), Penetration Testing 1988, pp 221 243 Ruiter, J de (ed) 1988 Penetration Testing 1988, Rotterdam Balkema Salgado, R, JK Mitchell & M Jamiolkowski 1998 Calibration chamber size eects on penetration resistance in sand ASCE Journal o Geotechnical and Geoenvironmental Engineering, 124:878 888 6