sc i a ‘Alonso, E. E., Gens, A. & Lloret, A. (2000). Géotechnique 50, No. 6, 645-656 Precompression design for secondary settlement reduction E. E. ALONSO", A. GENS* and A. LLORET* |A large water treatment plant is to be built on soft deltaic ‘deposits. Precompression has been selected as the method for achieving the required ground improvement, An instru- mented preload test has been carried out to obtain reliable information on precompression performance. Distributions of displacements throughout the foundation depth, obtained using sliding micrometer extensometers, have proved extrem ‘ly useful for identifying the mechanisms of behaviour con- trolling ground deformation. ‘The magnitudes of displacements are largely dependent on the overconsolidation state of the soil. As primary consolidation settlements take place rather quickly, they can be largely controlled by applying a preload over a limited period. The main design criterion therefore concerns secondary settlements. Labora- tory and field data indicate clearly that overconsolidating the soil, even in moderate amounts, significantly reduces the secondary compression rate. The performance of an unload- ing stage in the preload test provides crucial information in this regard. Therefore applying a preload surcharge larger than the final structure load is quite effective in controlling the magnitude of subsequent secondary settlements. The information collected during the site investigation and in the preload test provides the basis for the development of a Ground deformation model that ean be used for computing settlement histories. The model is used as a design tool for the final proposal of the precompression treatment required for the various structures of the plant. KEYWORDS: case history; compressbiliy; consolidation; eld ‘ground improvement; setlement Une grande usine de traitement des eaux doit tre construite sur des dépéts deltaiques tendres. On a choisi In méthode de précompression pour améliorer le sol de maniére adéquate. Nous avons fait un essai de précharge Instrumenté afin de récolter une information flable sur Ia performance de pré- ‘compression. Les répartitions des déplacements 4 Ia profon- deur de fondation, obtenues en utilisant des extensométres ‘micrometres i coulisse se sont révélées extrémement utiles pour identifier les mécanismes de comportement contrélant Ia déformation du scl. La magnitude des déplacements dé- ppend fortement de Tétat de surconsolidation du sol. Etant, donné que les tassements de consolidation primaires se pro- uisent assez rapidement, ils peuvent étre largement con- trdlés en appliquant une préeharge sur une période Umitée. Le principal eritere conceptuel porte donc sur les tassements secondaires. Les données obtenues sur le terrain et en labor- atoire indiquent clairement qu’en surconsolidant le sol, méme modérément, on réduit de maniére significative le taux de compression secondaire. La performance d'un stade de d&= ‘charge dans lessai de précharge donne une information cruciale 4 cet égard. Done, le fait d'appliquer une surcharge de précharge plus grande que la charge de structure finale est assez eflicace pour controler Ia magnitude des tassements secondaires subséquents. Linformation rassemblée pendent Vinvestigation du site et dans Pessal de précharge, sert de base au développement dun modéle de déformation du sol ‘qui pourra étre utilisé pour calculer les historiques de tasse- ment. Nous utilisons ce modéle comme outil conceptuel pour la proposition finale du traitement de précompression néces- saire pour les différentes structures de cette usine, ByTRODUCTION ‘The use of precompression to improve the foundation cheracter- isties of soft, fine-grained soils has been widely used in ‘geotechnical engineering. Generally, design for setlement re- duction has focused mainly on primary consolidation move- ‘ments using well-established soil mechanics principles and procedures (e.g. Aldrich, 1965; Johnson, 1970; Mitchell 1981, Stamatopoulos & Kotzias, 1983). Ifthe time to reach the end of primary consolidation is relatively short, primary consolidation Settlements can be largely eliminated by applying a preload over limited periods of time. In those cases, the time-dependent settlement of the ultimate structures is basically controlled by the secondary consolidation (drained creep) of the sci Although secondary settlements are certainly smaller than pri= ‘mary consolidation settlements, sometimes they may become & sigaificant design issue, especially when the structure is sensi tive to ground movements. ‘The construction of a large water treatment plant om soft eltaic deposits near Barcelona has required the design of a preloading operation with the objective of minimizing subse- quent settlements. The plant will occupy about 45000 m? close ‘to the mouth of the Llobregat River. Fig. 1 shows a pian view ‘of the current design. The various structures will generally apply moderate loads, but a number of them are quite sensitive to total and differential settlements. As the firm ground appears at great depths and the structures are very extensive, deep Manusenpt received 4 Febrery 2000; revised manuscript accepted 12 May 2000. Discussion on this paper closes 22 February 2001; for faher deal see pil 5 Technical Universi of Catsunya, Barcelona, Spain 64s foundations have been discarded in favour of ground improve- ‘ment by means of precompresion The plant is located inthe Liobregat deta region, the general geology’ of which is well known since it has been an area intensively occupied inthe past. Three main stratigraphic unis are distinguished. From top o bottom, they ae: (a) permeable Unit made up mainly of sands (upper aquifer); (0) 2 low permeability intermedate unit of sof clays and silts with Frequent granular parings; and (©) a deep pemeable unit composed of gravels and sands (lower aque). The intermedi ate aguitard effectively separates the upper aguifer fom the lower aquifer, making them independent from a hydrological point of view. Wheress the quality of the water inthe upper aquifer is Tow, the lower aquifer as been intensely exploited in the past. Piezometric levels 25 m below hydrosate have been measured in the deep aquifer, although water pressures have been steadily sing inthe last two decades owing to reductions in water extraction. As will be shown Iter, primary consolidation is i fact gute rapid, so secondary de'ormations are the msin source of cot- em regarding. firue plant setlements. This requires the development of a praical and reliable’ method to estimate secondary setlements aad to assess the effectiveness of preload ing in reducing them. Generally, the design of precompression to reduce furure seconiary setlemens has recived compar tively les anton lamilkowski tal, 1983; Magnan, 1998). ‘A number of authors (eg. Ladd, 1971; Mest, 1973; Kousofas et al, 1987; Yu & Frizi, 1994) have remarked on the signif. Cant feduction of secondary compression deformation when the soi is overconsolidated evento a modes degree indicating that the use of a preload surcharge lager than the final. structure Toad is an effective method for reding Secondary sexlementsALONSO, GENS AND LLORET fe reload test embankment Bjecrum (1972) analysed qualitatively the efect of a preloeding on subsequent secondary settlements using his conceptual creep ‘model. In this paper, precompression design will be based on the establishment of a direct relationship between overconsoliation ratio (OCR) and secondary compression coefficient, Cg. Such fan approach has some limitations from a theoretical point of view, but it provides a good base for achieving useful results in practice. Advantage will be taken of the field measurements fom a preload test in order t0 develop a ground deformation ‘model that can be used for prediction and design. The precise measurement of the distrbution of settlements in the ground will play a central role in this respect. An unloading stage is included in the preload test with the specific aim of observing directly the effecis of overconsolidation on secondary comprese sion deformations SITE INVESTIGATION AND GROUND CHARACTERIZATION The site investigation was carried out in stages. The first stage was directed towards achieving a general knowledge of the stratigraphic sequence of the area where the plant is to be built and of the properties of the materials encountered. It involved driling of eight deep boreholes (up to 80 m Jong) and six piezocone (CPTU) tests; 18 vibrating wire piezometers were also installed to observe the hydraulic conditions of the site ‘The second stage wes concentrated on the area where the preioad test was going to be carried out, and was focused ‘mainly on the determination of the specific ground properties requited for precompression design. Seven additional boreholes were drilled and four static penetration tests (two of them CPTU tests) were performed. A. cross-hole seismic test was also carried out 10 determine the elastic properties of the various stata, Detailed soil profiling was undertaken ffom the visual ‘observations of the soil samples and from the cone penetration tests, An extensive programme of laboratory tests was carried cout on the samples oblained ffom the boreholes. A third site fnvestigation stage covering in detail the whole of the plant has just been carried out, but the results are outside the scope of this paper. ‘The approximate ground sequence at the site of the preload testis depicted in Fig. 2, It consists of (a) 0-3 m. Upper silt. This material is of little significance as it will generally be excevated and removed to build the foundations of the various structures. (8) 3-10m. Sands (upper aquifer). Medium to fine sands, grey 1. Plan view ofthe water treatment plant: 1, anaeroble digesters; 2, primary caver; 3, activated sludge reactor; 4, secondary clarifiers + Settlement plates © Exenzomeler © crosetole est Boreas ScPrUtest cpr eet som 1 2g t Embankment Fig. 2. Soil profile end layout of the preload test and brown in colour; occasionally small sik inclusions are found, Most talues of SPT (coreted for dept) are ithe 20-30 range. {0-60 im ‘Clys and sits (imermedate aye). This Constnres the compressible sof ioe grained deposits that gjve se to cost stlements. They ae dark in colour, This fiyer contains a signifcamt numbers of sand and sity sand paring, the proportion of which dereses fom the fop of the lyer tovards the Botom. Infact iti nt usally feasible to ident) preci the int beeen hs ayer and the upper aqurer the wanston is quite gradual. AS depth ince, is aun of end tnd sy stn becomes Soy Sale The organi. mater content of the fine-grained. ©+ material ig low, about 0-7% in average snd never higher than_ 15% (@) 60-70 7m. Gravels and sands (lower aquifer). Sandy gravels predominate, although occasionally they sre replaced by Gense medium-size sands of limited thickness. This layer provides a quite rigid substratum to the previous materials 1nd can be considered incompressible in practice. Some of the geotechnical indices and parameters obtained in the site investigation are shown in Fig. 3. The fine-grained materials classify mainly as CL (low plasticity clays), and the ratio between Cy and C, is about 012 on average. The coefi- cient of consolidation has been determined both in the labora- tory (cedometer tests) and in the field (CPTU dissipation tests). The field velues are in general significantly higher, reflecting the presence of draining sublayers. Sometimes, however, large values of the coefficient of consolidation are obtained in the laboratory, corresponding to specimens with a high sind con- ‘A water table about Im above sea level is present in the Upper aquifer. The piezometer readings (Fig. 4) show that the stribution of pore pressure in the intermediate fine-grained layer is not fully hydrostatic; pore pressures fall below hydro- static as the lower aquifer is approached, reflecting the reduced piezometric levels there. Hydrostatic conditions prevail only up {0 a depth of approximately 15-20m, where the presence of granular materials is more widespread. ‘As indicated abave, an important soil behaviour feature for design isthe reduction of the secondary compression coefficient with overconsolidation ratio, A limited series of special oedo- ‘meter tests on the natural soil have been carried out. The tests ‘were performed as follows, First the specimens were consoli- dated (in steps) to their estimated in-situ preconsoidation stress rest, Aflervards an additional load was applied and left for a sufficient period of time, typically about 3 days, so that the secondary compression coeffieient could be reliably determined. ‘Then the applied vertical load was reduced in order to give the sample a certain degree of overconsolidation. The new value of the secondary compression coefficient was then determined fom the new slope of the vertical strain ploted against log time, Examples of the results obtained in three different samples are shown in Fig. Sa, Fig. Sb presen the results ofthe various tests in terms of the ratio between the value of C, in the coverconsclidated state and the value of the normally consoli- dated C. Note that the reduction of Cy is very sharp; small increases of overconsolidation ratio reduce secondary compres sion very significantly, The results obtained are in the same range as those determined by Ladd (1971). Fo Psa 2 4 ieee ° 2 1 ‘Ste estan @ Festeage 5 © Secend stage 7 ° ry ig, Excess pore water pressure with respect to. hydrostatic conditions measured in plezometers intalled during site nvestgn- tion PRELOAD TEST ‘Asis oflen the case in precompression weatments concerning large projects, a preload test was performed in order to obtain more ‘elable information for design. Apart from the usual objective of checking consolidation times under feld conditions, the preload test was aso devised to provide information onthe cffests of overconsolidstion and on the relative contributions of the various soil layers to the total setemeats. This Ted to the ieesurements of movenentsdroughout the sol colun and 9 the inclusion ofan unloading stage ‘The preload txt was performed by building an embankment 4 m high (effectively applying a distbuted load of 80 KPa) over an area of 80m 80m (Fig. 2), The constuction lased for two months, ffom the zad of October tothe end of December 1996. The test was monitored continuously for a period of two years, until December 1998. From 10 to 22 December 1998, the wes totum aes sen 10 20 30 40 50 eee 0 25 50 75 100 oS 12 34 5 6 Heo e eS os ca 09 of 0200 Tyr PP Ey ETT ST TEE aa et FE ae lel ° * bee a ja ] 2 = efoo " | | . . 2 | el ele 7 . & | * “ = e i, 7 re a dae 5 3 ™ = Ff Let? Pet Baa a ; al * . | le real ea Fig. 3, Soll properties determined ia the site Investigation48 ALONSO, GENS AND LLORET ‘Normal consaiaated stage fe o18-020Mes 4 03¢-o4omPa 9 094-o4emrs Coverconsotes stage © 029-o27hPs 8 049-0486P2 © o4e-044NPa i ¥ cove g ae 5 cose oro ore te Te Tae tee Tae Toes o Oedametrnts td (1971 eon preoaig es 7 Loo (571 mes rtaig tes 4 La 1971 fon tang tres) g So o ° 12 74 18 72 20 22 oon ) top 1'Sm of embankment was removed to check on the effets ‘of unloading on displacements, which were monitored for an ‘addtional eight months, c August 1999. At the end of August 1999, the first constuction operations started, disturbing. the pattems of displacements, and the preload test was concluded. To understand the setilement mechanisms properly it was necessary to measure precisely the distributions of displace- rents throughout the foundation soil. Two sliding micrometers (Kovari & Amsiad, 1982, 1983), capable of monitoring vertical isplacements at I'm intervals, were installed at the locations shown in Fig. 2. The micrometer type was specially selected s0 that it could measure large movements (up t0 10 em/m) with an adequate accuracy (0-03 mm/m) Adjacent to each sliding mi ‘crometer, four vibrating wire piezometers were installed at different depths, a8 shown in Fig. 2. In addition, vertical displacements af the base of the embankments were measured by. means of settlement plates Isid out in a grid of 20m % 20 m, For clarity, Fig. 2 indicates only the location of the settlement plates for which the results are given in this peper. The staring date for monitoring data is taken to be 21 October 1996. “The development of settlements at several points at the base of the embankment is shown in Fig. 6(a). The sequence of embsnkment loadinglunloading is also indicated. Naturally, maximum setlemerts are observed in the centre of the loaded area, reaching a value of about 23 cm. The associated variations of piezometric heads at thee different depths are shown in Fig 6(G). Considering the piezometers placed in the intermediate clay and silt layer idepths 18 m and 45 m), it is apparent thet the pore pressure response to loading is very suppressed, indi- cating a rapid. dissipation of pore water pressure. Primary ‘consolidation times are short and comparable to embankment construction times, The measurements at a depth of 63m correspond to a piezometer placed in the gravels and indicate the pore water pressure changes in the lower aquifer. It ean be observed that piezometic levels are not constant, This causesTe Ei ert te TS 125m tom cee examen cere ee N3 (85 m from centre) q “o Ne(so mom cents) “ens (eomtom cate) = tt ontannert ds. . f-__| | i i —— ssf é 3 10 : | — oes s 0 T Deph= 18m Ew e E Devt m t—1 = e ; 5 KRW i = 8 Pp ed z Dent = 63m hh i z i = 5 3 é = 6 0 a a ima: days ©) Fig. 6, (a) Vertical settements athe base of the embankment during the preload test. Embankment helght 1s indested some difficulties in the interpretation of measurements, fortu- nately limited to the deeper part of the intermediate layer. "The essential information for subsequent design is provided by the distributions of vertical displacements in depth (Fig. 7). ‘They are obtained from the readings of the two sliding micro- meters during the loading stage of the test. The measurements fare given in both differential and accumulated form. The results are very similar in the two extensometer ‘As expected. the contributions of the upper sit layer and the sand (upper aquifer) layer are small; most settlements occur in the intermediate layer where fine-grained materials are generally sicated, (0) Plezometric heads measured during the preload test at various depths. Embankment height predominant. Observation of setlement distributions within the intermediate levels yields very important information. Most deformations occur in the upper part of the layer, between 14 ‘and 201m deep, in spite of the fect that there the proportion of sranular, less compressible material is higher. In the deeper part of the layer the deformations are moderate or small, although the proportion of clay is very high, This reflects the fat that i the upper part of the layer the material is likely to be normally consolidated, because the pore pressures have been basically controlled by the upger aquifer, not subjected to extractions, fand have remained very close’ to sea level throughout. InDierenti placements: mv é 2 8 Py Deptn:m ° Accumuited doptacements mm ost 150200250 Dp m o —sarane | 82 rr one sonais6 19097 19007 — «ane © @ Fig. 7. (2) Distributions of diferential vertical displacements measured in extensometer 1. (b) Distributions of vertical displacements measured in extensometer 1. (€) Distributions of differential vertical displacements ‘measured in extensometer 2 (b) Disttbuions of vertieal displacements measured In extensometer 2. Date of frigin for displacement measurements: 23 Oct. 1996 contrast, in the deeper part of the layer, the material has become’ overvonsolidated by the past seduction of piezometric levels of the lower aquifer. ‘The extensometer data permit the analysis of the foundation considering it as a series of slices Im thick. If the vertical Stress increments are computed, it is possible to estimate the equivalent values of compression index every metre using the ‘observed differential displacements due to primary consolida- tion. The vertical stress increments caused by the embankment oad have been obtained from an elastic fnite element analysis using a realistic distibution of stiffness throughout the soil profile. The results for the locations of the two extensometers are shown in Fig. 8(2). The compression index values obtained in this way exhibit significant differences (Fig. 8(6)). In the ‘upper part of the layer compression indices are high, typical of raterials in a normally consolidated state. In the medium and eeper levels the values of compression index are much lower, characteristic of overcensolidated soils.PRECOMPRESSION DESIGN FOR SECONDARY SETTLEMENT REDUCTION 651 } . = i x : | ; eat of E raed afte « off} 2 tremens | : oo | S © eamenear? + 2 ae oO 20 40 60 2 100 My 008 oF O18 02 Vora! anes nerament Pa @ Fig. (a) Computed vertical stress increment due to the spplied preload surcharge. primary compresion indices with depth Differential vertical displacements corresponding to different ground depths are plotted in Fig. 9(a) against time in @ semi-log Scale. la this representation the evolution of settlements at long times is approximately linear, typical of secondary compression phenomena, It is also very interesting that, plotting the evolu- tion of strain rates with time, the field data appear to be a natural extrapolation from laboratory oedometer data to longer ‘After two years of observations an unloading stage was performed in the tet, reducing the applied stress from 80 kPa to SOKPa, Fig, 6(a) shows that, afer a small rebound, settle- ‘ments continued but at a smaller rate, The behaviour of the ground during unloading, however, is best examined by refer- fing to distributions of displacements (Fig. 10(a)). Now, the evelopment of vertical displacements is quite uniform through- cout the entire layer. For comparison, Fig. 10(b) is presented alongside. It contains the distnbution of vertical displacements for approximately one year before unloading (note the different displacement scale). The contrast is very clear. By providing ‘overconsolidation to the soil, the unloading stage has removed the large deformation peaks that existed and has resulted in a quite uniform settlement development occurring at a reduced fate, The same observation can be made by examining the results in terms of secondary compression coefficient, Ca. The values of Ca at each point can be obtained from observation of the individual deformation readings agsinst log time. As Fig. 11 shows, all the large values of C, have disappeared after unlcad- ing. The results for depths below 50m are not plotted because they ate affected by the reduction of water pressurés in the lower aquifer. GROUND DEFORMATION MODEL “The transfer of the results of the preload test 0 the design of the precompression treatment under a variety of ‘situations requires the development of a ground deformation model. For use in practice this model should be simple, but it must incorporate all relevant behavioural features. Given the large extent of the preload area in relation to the deformable ground eee, © (©) Distribution of thickness, a one-dimensional model has been deemed accep- ‘able. ‘The basic informatica required for a particular locaton is: (@) a detailed soil profile achieved by way of a close examination of the borehole samples and the CPTU records (®) an estimation of the initial value and the development of effective stresses and preconsolidation pressure throughout the soil column, In this section the model for the preload test zone will be escribed. It will be calibrated against field observations so that it can contribute 10 a more quantitative interpretation of the tes. The soil profile derived from the site investigation data is presented in Fig. 12. The most important pat is the distinction, In the intermediate layer, between free-draining sandy sublayers and the fine-grained intervals where transient consolidation phenomena occur. As Fig. 12 shows, the freedraining granular materials are far more widespread in the upper part of the intermediate layer. “The initial stress in the ground columa prior to the perform- ance of the test is shown in Fig. 13(3) together with the ‘estimated maximum past stress (preconsolidation pressure). This ‘maximum, past stress has been computed ffom the expected effects of the known 25m water level reduction of the lower aquifer, The effect is considered to extend until the granular materials become more predominant towards the upper part of the intermediate layer, where hydraulic conditions are likely to be controlled by the upper aquifer water levels. Therefore it was expected that the material has remained normally consoli- dated up 10 a depth of about 18-20 m. This is consistent with the estimations of primary consolidation indices shown in Fig 5(b). Subsequent charges in vertical effective stress during ‘embankment loading and unloading are computed from the clastic finite element analysis. The successive vertical stress distributions are plotted in Fig. 13(a). Neglecting possible age- ing and bonding effects, the distribution of overconsolidation ratios can now be conputed at various stages of the preload test. They are presented in Fig. 13(0). It can be observed that after embankment loading a large part of the intermediate layerTime: aye 2 "0. 1000 s € E : Dent a eno i " me 22278 seme nos = Oe 22479 we reer Fa ) hes pace to? pret i Fixe 5 B toe" bao tet reo Be a Ta te Tone ® oe Exon apt 0-22 m) oe cesormtaris OR = 1 a Sensor eth 2-8 m) Jo Oetimeterns (OCR «1:2 2 btarsonete (eth 35-4 Ae ndamatr tts (0A =2) a Exensomte(eth 7-£0) Fig.9 (a) Variation of diferent verte apacements at various depts with time.) Comparison of measured secondary compression strain rats measured in the laboratory and in the Bld is normally consolidated, so that large deformations both in primary and secondary ‘compression canbe expected there However, after unloading, all the foundation soil has at least some degree of overconsoldation Using. this estimation, the data from the extensometers can bbe used to relate the observed reduction in the value of the secondary compression coeflicient, Ca, with averconsolidation ratio. The results are shown in Fig. 14, where, in spite of some scatter, several interesting features can be noted. In the first place there is a strong reduction in the value of Cy with very moderate inereases of overconsolidation ratio. Therefore preloading beyond the final structure load should be quite effective in reducing secondary settlements. The reduction trend is quite similar to that obtained by Ladd (1971) for long prelosding times. The results obtsined from cedometer tests are also included in the same figure. They tend to plot above the field results, possibly reflecting the significantly. shorter preloading times. It is also worth noting that the points obtained in the loacing and unloading stages appear to follow the same trend. Therefore there is no need to distinguish between those two situations in the computation of secondary settlements,PRECOMPRESSION DESIGN FOR SECONDARY SETTLEMENT REDUCTION 653 ferential eplcement: navn ° 4 08 wo 10 20 ier iptace a — 8097 ranger @ Fig. 10, (a) Differential verdcal deplacements measured in ex measurements: 2 Det. 1998; (b) before unloading (date of origin 18 Exensemsart (betere unloading) (© estensemstr? (peter unloading) 1» Sxtensomete (eter ulosceg) ‘¢ Exteneometar 2 (ater unions) ae 2 ke 4 é e io og 10 Fig. 1, Observed values of the secondary compression coefficient, Cay beioge and after unloading ‘The ground deformation model is now constructed as fol- lows: (@) The soil column is divided into a number of individual intervals based on the soil profile obtained from site investigation. o someter 1: (a) after unlosdlng (date of origin for displacement Yor displacement measurements: 27 July 1997 (®) The development of effective vertical stresses and pre- consolidation pressures throughout the loading history is computed as indicated above. (6) The time history of settlements is calculated as the sum of the contributions of the vertical displacements for each soi interval. The total calculation time is divided into a large number of time increments, and the development of setlements is computed by means of o time-advancing scheme In the fine-grained soils the deformations are the sum of the immediate elastic defermation (computed from clastic expres sions), the primary consolidation deformation (computed using the oedometer parameters C. and C,) and the secondary com- pression deformation. Secondary compression strain occurring ina time interval (t,-1~ 4), at @ point with vertical coordinate 2, Atv (2), is computed as Aeu(a) = ColPelz, 9/04, t)Monti/t:-1) o where pes) and 04x, 1) are the preconsolidation stress and the effective vertical stess respectively at time f "The value of secondary compression depends on the over- consolidation ratio, OCR, according to the empirical expression derived from the preload test results (Fig. 14): Gq = 0:008{0-1 + 09 exp(—13(OCR — 1)}] @ For consistency with 3jerrum’ (1972) conceptual model, the peconsldaon press inne during sendy compres: (Pot pte-s)) = (4 eA) ® For the granular sublayers only instantaneous elastic deforms- sions and drained. creep deformations are computed, Their contribution to total setlement is smal.6st ALONSO, GENS AND LLORET Upper ait 7 MPa, =, C,= 00805 Send (upper scuter) E9738 MPa, 2,0, cay G.=016, 5, = 00095 ont, 620.008 (0.1 09emer) 2 28 sity sana Grave (er ate Fig. 12. Soil profile representing ground conditions in the prelosd ‘The ground deformation model can now be applied to the preload test. The parameters used are indisted in Fig. 12. Since the model has been calibrated using some of the preload test data, the close agreement between global sertlement computa- tions and observations is not especially significant (Fig. 15). It should be pointed out that the agreement also extends 10 the details of the settlement variations with depth. The model can therefore be used to provide a more quantitative insight into the results of the test. For instance, in the same Fig. 15, the contribution of primary and secondary consolidation setlements to fotal movements can be readily examined. The deformation ground model calibrated in this manner can be used as a tool for precompression design PRECOMPRESSION DESIGN The design of the precompression required for building the ‘water treatment plant is controlled by the magnitude of setle- ‘ments thatthe various structures ean safely undergo. The besic eiteria concern, therefore, the reduction of settlements expected {to occur after construction of the structure, ie. the time dependent seitements. Because of the planted construction schedule, preload times are limited to six months approximately, ‘Therefore the main design parameter t0 be determined is the height of the precompression embankment to be adopted for each particular sma, “The design process for each structure of the plant follows the following steps (@)_ adoption of s soil profile based on the detailed examination of the site investigation data (borehole cores and simples, CPTU results) (®) evaluation of the history of stresses and preconsolidation pressures taking into account all the stages of preloading, construction and operation (©) pplication cf the ground deformation model outlined above to compute the expected settlement history of the structure. Steps (a)~(c) are caried out for various heights of the prelead ‘embankment. From the analysis of the time-dependent setle- ments of the finished structure at various times, the optimum ‘embankment height for each case is selected As an example, the results of the caleulations carried out for the anaerobic digesters are presented, The location of those ssructures is shown in Fig. 1. They are the heaviest stuctares of the plant, and apoly a pressure on the ground of 1245 kPa. OF this total, 245 kPa corresponds to the self-weight of the struc- ture and the rest, IOOKPa, 10 the load applied when filing the tank Prior to construction, 1m of ground is excavated to house the foundation, The loading stquence considered for analysis includes the following: (@) application ef the precompression embankment Joed (vari- able) (®) preload maintained for six months (©) preload removal (2) ground exesvation (©) structure construction CO) filling of the stucture A subsequent emptying/filing of the tank after the one-year operation has also been included in the calculations. The computed history of displacements for four different preload embankment heights is shown in Fig. 16(a). The relevant variable for design is, however, the magnitude of the time- dependent settlements occurring after construction, They are plotted as a funcion of the height of preload embankment in Fig. 16(b). Ic can be noted that the additional reduction of settlements for embankment height values larger than 8m becomes small. Indeed, in this paricular case, a preload embenkment 8 m high has been adopted in the final design. ‘CONCLUDING REMARKS ‘When precompression is applied to a relatively free-draining ‘material, consolidation times ate short, and therefore primary setllements can te effectively controlled by using a prelosd equal to the final load over a limited period of time. The subsequent moversents associated with unloadingireloading will be smal, and will occur rapidly: In these cases, the main source of long-term displacements is likely to be secondary compres: sion settlements Using laboratory and field data gathered in relation to the precompression design for a large water treatment plant, it has been shown that preconsolidating the soil is an effective way to reduce secondary setlements to acceptable levels. Both labora- tory and field observations show that there is a sharp reduction in the value of the secondary compression coeficient with only small degrees of overconsolidation. The performance of an unloading stage a the end of the preload test has provided particularly useful data inthis respect. ‘The deployment of extensometer to obtain precise measure-PRECOMPRESSION DESIGN FOR SECONDARY SETTLEMENT REDUCTION 655 9 ia 7 f = Maximum past | T © Excancometer vot 4A lol sess before uneazing me Extonsomater =e ster unloading fosag 20 ‘Sess ater |___ - eesnsometer2: before unaadng roading = Exensomete2 ser unloading | T Dept: | ® I » | \ \ \ © of 04 06 OB wm 0 12 1 16 ‘assoc on o ° ig. 13. (a) Initial preconsoidaton stress and vertical stress distributions at various times ofthe preload test. (@) Distribution of overconsolidation ratio (OCR) with depth at various tmes of the preload tet 8 Cacometer tet Lad (1871 (enor prloscng times) Lag (1871) (medum prscaang tes) <--> Lae (1871) ong preleairg times) + yr 09 0ptt3 (1-2) +0 © Etensometet © Extansometer (ster unica) © Exensometer2 4 Bcensometer? ater uneacng) Salter: Fig. 14. Reduction of secondary compression coefficient with over= ‘consolation ratio derived from feld measurements. Oedometer test results have Been added for comparison ments of the displacement distributions at depth has provided very valuable information towards achieving a good understand- ing of the various _phenomens involved and their relative Importance. This information has provided the basis for the evelopment of a simple but comprehensive ground deformation 02 Extensoneter 1 S Cmguinesottoments Correuted prmaryectioments (Compted secondary setlements Messire sti 03 10 100 1008 Tine: days Fig. 15. Global settlements of the preload txt computed using the ground deformation mode. Observed settlements are ploted for comparison model that can be used for computing and predicting future settlements. The final result is a practical design tool firmly ‘based on the observed field behaviour of the ground. ACKNOWLEDGEMENTS ‘The authors acknowledge the support of Depurbaix S.A. and EMSSA in the development of this work Field work was carried out by Eurogeotéenica S.A. The authors are also grateful for useful discussions held withthe site engineer, Jordi Via656 ALONSO, GENS ANU LLUKEL Hatin "0 s00 t060” ¥800" 200028008000 8500 4000, ime: cays a 2 4 6 8 0 2 Height of ereankent o Fig. 16. (a) Settlement histories computed for the anaerobic digesters using diferent preload embankment heights, H. () Variation of ompoied secondary settlements at various times with preload embankment height, H [NOTATION (Water content PL Plasticity index _ Compression index Co Secondary compression coefiient Vertical sonoliation coecient £4 Horizontal conslidation coeFcient ot Ligud limit (74m Grain size smaller than 74 microns C,_ Swelling index REFERENCES ‘Alrich, H.R (1965) Precompression for support of shallow founds tons. Soi! Mech Found, Di, ASCE 81, No, 2, $20. ‘Bjerrum, L- (1972). Embankments on sot ground. Proceedings of the ‘eval conference on performance of earth and eartheuppored {uctures, Inia Va. 2, 1-54 Jamielkowak, M. Lancellota, R. & Wold, W. (1983), Precompression "ind speeding up consolidation, General tepor. Proc. Sub Europ. Conf Soll Meck. Found. Engng, Helsink 3, 1201-1226 Jotnson, S.J (1970), Precompresson for improving foundation sis, "Soil Mech. Found, Dis, ASCE 96, No. 1, 111148 Koutsofias, D, Cy Foot, R & Handiek, L, D. (1987). Geotechnical investigations ofthore Hong Kong, | Geoteck. Emons, Div, ASCE 113, No, 2, 87-195, Kevan, K.& Amaad, Ch. (1982). Anew method of measuring “formations in daphagin als and ples. Géotechnique 22, No.4, 4402-06. Kovar, Ke & Amstad, Ch (1982), Fundamentals of defomation meat ‘urements. Proceedings of International symposium on Feld meas troments in Geonechancs, Zurich, Vol. 1, 219-239. Ladd, © €. (1971) Setlment onchsis of cohesive soils, Research ‘Report R71-2. Cambeidge, MA: MFT. ‘Magnan, J (994), Methods 1 reduce the settlement of embesknens fon sof clay: review, In Vertical and horizontal diplacements of ‘Joundaions and embankments, ASCE Geotechscal Special Publice- ‘ion No, 40, Vl. 1, 77-81 Mest, G. (1973), Coeicient of secondary compression. JI Soil Mach ‘Found. Di, ASCE 98, No. 1, 123=137 ‘Mitcell, JM, (198). Soi improvement. Sate ofthe art repo. Pros. UGih int. Conf. Soil Mock and Found. Engng, Stcckbolm $, $09 5. Stamatopoules, A.C. & Kouas, PC. (1983). Setlementstime predic- Tiont in preloading. 1 Geoteck. Engng. Div, ASCE 109, No. 6 802-820 Ya, K Pf Frizi, RP (1994). Preloading organic sist init fare seilements In'Terscal ond horizontal displacements of fowrdations land embanirns, ASCE Geotechnical Special Publication No. 40, Vol. 1, 476180.