INTERNATIONAL SOCIETY FOR SOIL MECHANICS AND GEOTECHNICAL ENGINEERING This paper was downloaded from the Online Library of the International Society for Soil Mechanics and Geotechnical Engineering (ISSMGE). The library is available here: https://www.issmge.org/publications/online-library This is an open-access database that archives thousands of papers published under the Auspices of the ISSMGE and maintained by the Innovation and Development Committee of ISSMGE. Panel discussion: The prediction and the control of displacements around deep excavations in completely decomposed granite Débat de spécialistes: Prévision et contrôle des déformations autour d’une fouille profonde dans le granite complètement décomposé A. W. Malone - Geotechnical Engineering Office, Hong Kong, People's Republic of China C.W.W. Ng - Hong Kong University of Science and Technology, People's Republic of China J.W. Pappin - Ove Arup and Partners, Hong Kong, People's Republic of China S y n o p s is : Th is co ntrib utio n co v ers tw o su b jects: the co nstructio n m etho d s em p lo yed in H o ng Ko ng to keep d isp lacem ents aro und d eep excav atio ns to a m inim um and the pred ictio n o f d isp lacements using small strain mo d els. O f im p o rtance fo r the p red ictio n o f d isp lacem ents, it is sho w n that co m p letely d eco m p o sed granite o f the Ko w lo o n Pluto n exhibits sm all strain stiffness behav io ur ty p ical o f sed im entary so ils. In the case histo ry presented , finite elem ent co m p utatio ns w ith a small strain stiffness mo del gave go o d p red ictio ns fo r lateral w all d isp lacem ent but the v ertical settlem ent p red ictio ns w ere less g o o d , p o ssibly due to the effect o f remnant structure in the co m p letely d eco m p o sed granite no t allo w ed fo r in the m o d el. M in i m iz in g d i s p l a c e m e n t s p ro file o v erlain by thin layers o f lo o se sand y fill and m arine Exp erience in H o ng Ko ng sho w s that excav atio ns d eep b elo w the d ep o sits. SPT N -v alues increase fro m abo ut 10 at 5m depth to abo ut 100 at 30m depth. The gro undw ater table is abo ut 1.5m beneath the w atertable in co m p letely d eco m p o sed granite (i.e. m aterial that ‘ can gro und surface. Unlike m any sites in urban H o ng Ko ng , this is a be bro ken d o w n betw een the fing ers to its co nstituent grains and ‘ v irg in’ site, as it is believ ed that the ground beneath the site had no t slakes in w ater’ ) can cause large settlem ents o f the ad jacent gro und. exp erienced A t sites sensitiv e to d isp lacem ents, to p -d o w n excav atio n w ithin a co nstructio n d ew atering activ ities since it w as reclaim ed in the any substantial stress chang es resulting fro m co ncrete d iap hragm w all b o x is no w the fav o ured co nstructio n 1920s (Lui & Yau, 1995). A d jacent to the site there are sensitiv e m etho d fo r exc av atio n d ep ths g reater than aro und 12m. But structures, d am age to w hich had to b e av o id ed , and therefo re installatio n o f the d iaphragm w all need s to be carried o ut w ith care. p recautio ns w ere taken to lim it d isp lacem ents d uring b asem ent In the early d ays o f d iap hragm w alling in co m p letely d eco m p o sed co nstructio n. To achiev e an effectiv e gro und w ater cu t-o ff, a d eep g ranite in H o ng Ko ng , in the late 1970s and early 1980s, large d iaphragm w all w as installed , to ‘ eng ineering ro ckhead , Grad e 3 m o v em ents w ere co m m o nly reco rd ed d uring d iaphragm w all ro c k’ at 47-62m b elo w gro und lev el. Grad e 3 ro ck is d efined as co nstructio n (Fig u re 1). Practitio ners in H o ng Ko ng hav e since ‘ m aterial that canno t usually be bro ken by hand , is stained learnt that, to lim it d isp lacem ents, it is necessary to use sho rt trench througho ut and g iv es a dull ring w hen struck by ham m er’ . Curtain p anels, take care in excav ating and to m aintain the slurry p ressure gro uting w as carried to a d ep th o f so m e 10m beneath the w all. A w ell ab o v e the gro und w ater p ressure, i.e. rather m o re than that w atersto p w as installed b etw een trench p anels to 30m depth and a need ed fo r theo retical static stab ility . By this m eans, installatio n milled hyd ro fraise fo rmed o n p anel ed g es beneath this depth. Care settlem ents can b e lim ited to say less than abo ut 10m m . Bu t it w as taken in excav atio n o f the p anels, w hich w ere 4m in length, and rem ains a challeng e to keep d isp lacem ents due to excav atio n and this clearly p aid o f f as relativ ely sm all settlem ents o f the ad jacent d ew atering w ithin accep tab le lim its. A case histo ry w ill help to illustrate the m easures to be taken to m inim ize d isp lacem ents. The 26m deep excav atio n illustrated in Figure 2 w as carried o ut by to p-d o w n co nstructio n w ithin a co ncrete d iap hragm w all b o x 107m by 67m in plan (Lui & Yau, 1995). Gro und co nd itio ns at Drago n Centre, lo cated in an area o f d eep w eathering in the Sham Shui Po d istrict o f no rthern Ko w lo o n, co m p rise a d eep ly w eathered granite D istan ce fro m Excav atio n M ax. Tren ch D ep th 1.0 0.5 2.0 ■ 1.5 2.5 - c ST ' 0.1 Be ll C o m m o n o H o ng Ko ng Excav atio ns t ± Lo nd o n N PY ♦ Stud enterlu nd en ■ So u th C o v e Settlem ent M ax . Tre n c h D ep th Fig ure 1. Sum m ary o f m easured settlem ents caused by the installatio n o f co ncrete d iap hragm w alls, after Clo u gh & O ’ Ro urke (1990). 2325 gro und w ere reco rd ed during d iaphragm w all installatio n (< 10m m ), d ecay curv es and o ther m aterial p ro p erties w ere d eterm ined fro m co m p ared w ith tho se sho w n in Fig ure 1. Cam brid ge self-bo ring p ressurem eter tests carried out in co m p letely O n co m p letio n o f the diaphragm w all, a multiple w ell pumping test d eco m p o sed granite at a site in Ko w lo o n Bay , so m e 5km to the east w as carried out to test the effectiv eness o f the gro und w ater cu t-o ff. o f D rag o n Centre, and triaxial testing (p '-co nstant w ith lo cal strain A pum p ing test is no w stand ard p ractice in Ho ng Ko ng fo r m easu rem ent) o f so il fro m the sam e site (N g et al, 1998). excav atio ns d eep er than abo ut 8 - 10m at sensitiv e sites. D uring the co m p letely d eco m p o sed granite at Ko w lo o n Bay is a sim ilar so il to p um p ing test the g ro und w ater w ithin the d iap hragm w all b o x w as that at D rago n C entre, the sites b eing w ithin the sam e granite body d raw n d o w n to 24-30m b elo w gro und lev el. The asso ciated (a very unifo rm equigranular m ed ium -grained b io tite m o nzo granite red uctio n in head in stand p ipes installed in the ad jacent ground o f the Ko w lo o n Plu to n) and the so il sho w ing sim ilar fines co ntent o utsid e the b o x averaged 0.5m , ind icating that a go o d seal had been (15-30% ). achiev ed o v er m o st o f the site. A s a co nting ency m easure, recharg e stiffn ess (Fig u re 3). The S-shap ed curv e used in the mo d el is the w ells had been installed to m aintain w ater lev els, but these w ere no t m ean o f the m easured d ata, w ith Gmax/ p' set at 1500. used d uring the p um p ing test; so m e w ere o p erated The The m aterial exhib its p ro no unced strain-d ep end ent during The measured and co m p uted w all d isp lacem ents due to the excav atio n. During the w o rks, m easurem ents w ere taken o f surface pum p ing test are illustrated o n Fig ure 4. Inclino m eters 13,14 and 16 are at the m id -p o ints o f three sid es o f the d iagrap ham w all b o x. settlem ents and lateral w all d isp lacem ents (Lui & Yau, 1995). Larg e w all d isp lacem ents o ccurred d uring the p umping test (Figure A lo ng w ith the BR IC K pred ictio ns, an elastic M o hr Co ulo m b mo del 4) and these m ig ht hav e been less had the to p flo o r slab been cast p red ictio n is also illustrated . befo re carry ing o ut the p um p ing test. This p ro ced ure is no w being analysis, Yo u ng 's M o d ulus is assum ed equal to lx N (SP T ) v alue in intro d uced fo r sensitiv e sites. Fo r the elastic M o hr C o ulo m b M Pa, fo llo w ing the no rm al assum p tio n in Ho ng Ko ng . It is seen that the BR IC K p red ictio n is go o d but the elastic M o hr C o ulo m b m o d el o v er-p red icts. W all d isp lacem ents at the final excav atio n PR ED IC TIO N O F D ISPLA C EM EN TS to B5 level are sho w n in Figure 5. The BR IC K pred ictio n is again g o o d , in term s o f m agnitud e and curv ature, but the elastic M o hr N ow C o u lo m b m o d el is less g o o d . let us co m p are the o bserv ed d isp lacem ents w ith tho se co m p uted by sm all strain m o d elling . The seq uence o f o p eratio ns Fig ure 6 sho w s the asso ciated increm ental settlem ents due to excav atio n to B5 lev el o n survey mo d elled in the finite elem ent analysis co m p rised the p umping test, fo llo w ed by staged excav atio n to B5 lev el. The m o d elling used the SA FE finite elem ent p ro gram w ith Sim p so n’ s ‘ bricks o n string s’ soil Lateral D efle c tio n (m m ) m o d el (Sim p so n, 1992). This m o d el has the ab ility to cap ture the 100 10 effec t o f chang es in stress path and strain d ep end ent so il stiffness, 50 fro m a given mo dulus d ecay curv e (S-shap ed curv e). The mo dulus 0 -50 16 13 14 0 -10 -20 \VW '\\\\ Red u ced ' 30 Lev el (m PD ) -40 -50 ------- -60 ------- BR JC K .......... Me hr C o ulo m b Me asured (3 sectio ns) -70 Fig u re 4. Lateral d eflectio n o f the d iap hragm w all in the m ultiple w ell pumping test. Shear Strain (%) Fig ure 3a. Se lf b o ring p ressurem eter, Ko w lo o n Bay . Late ral D efle c tio n (m m ) 100 10 50 0 14 -50 16 13 0 / '.y -10 A -20 ( '^ L B 5 CT \ Red u ced ' 30 Lev el (m PD ) -40 ------- -60 ------- BR JC K ------- Me hr C o ulo rab ' Y\ M e asured Y\\ (3 >ections) \ \ -70 Shear Strain (%) Fig ure 3b . % -50 D rained co nstant p ’ triaxial tests w ith lo cal strain m easurem ents, Ko w lo o n Bay (N g et al, 1998). Fig u re 5. Lateral d eflectio n o f the d iap hragm w all d uring excav atio n to B5 lev el. 2326 D istan ce fro m Exc av atio n (m ) This hyp o thesis need s to be tested at o ther sites and by num erical exp eriments using a mo dulus d ecay curve representative o f so il w ith structure intact at sm all strain and suffering p ro g ressiv e d estructuring at larger strain, in a co ntinuum m o d el cap ab le o f hand ling p ro g ressiv e d estructuring. These o b serv atio ns, so il tests and co m p utatio ns hav e y et to be carried out. C o n c l u s io n s Regard ing co ntro l o f d isp lacem ent (co nclu sio ns fro m general p ractice): (i) ‘ to p -d o w n’ co nstructio n w ithin a d iaphragm w all b o x is favo ured fo r very d eep excav atio ns at sensitiv e sites, (ii) Fig ure 6. Settlem ents d uring excav atio n to B5 lev el. p recautio ns need to be taken d uring d iap hragm w all installatio n, and (iii) a so p histicated gro und w ater cu t-o ff sy stem is required . D istan ce fro m Excav atio n (m ) 0 20 40 60 80 100 120 14 Reg ard ing p red ictio n o f d isp lacem ent (co nclu sio ns fro m this case histo ry): (i) co m p letely d eco m p o sed granite o f the Ko w lo o n Pluto n sho w s small strain stiffness behav io ur typ ical o f sed imentary 10 so ils, y / /- / / (ii) the finite elem ent p red ictio ns w ith sm all strain stiffness w ere (iii) but less go o d fo r v ertical settlem ent, p o ssib ly due to the (iv ) the elastic M o hr Co ulo m b d isp lacem ent pred ictio ns w ere no t / go o d fo r lateral w all d isp lacem ent, ------ / M easured (3 sec tio ns) effect o f remnant structure no t allo w ed fo r in the m o d el, and ------ BR IC K ........ M o hr as go o d as the sm all strain p red ictio ns. C o u lo m b Settlem en t (m m ) Ac kn o w l ed g emen t s Fig u re 7. Settlem ents d uring the m ultip le w ell p umping test. The assistance o f Jam es Lui, Paul Yau and Je ssie Kw o ng in the lines no rm al to the w all at the p o sitio ns o f inclino m eters 13,14 and 16. p rep aratio n o f this co ntributio n is g ratefully ackno w led g ed . The m easured settlem ent tro ugh is narro w er and clo ser to the w all than m ig ht hav e b een exp ected fo r sed im entary sand s but the m axim um settlem ent is w ithin the range o f exp ected v alues, being Re f e r e n c e s abo ut 0.15% o f excav atio n d ep th and abo ut 60% o f m axim um w all m o vem ent. N either mo d el p red icts the settlem ent very w ell. Figure C lo ug h, G .W . and O ’ Ro u rke, T.D . (1990). Co nstructio n ind uced 7 sho w s settlem ents d uring the pumping test. Measured settlements m o v em ents o f in situ w alls. Proceedings o f ASCE Conference are surp risingly lo w , given a w all d eflectio n o f 50m m o r mo re. Bo th on Design and Performance o f Earth Retaining Structures. Cornell University, Geotechnical Special Publication 25, 439- p red ictio ns are po o r. The p o o r BR IC K settlem ent pred ictio n requires so m e exp lanatio n. W hy sho uld the BR IC K m o d el o v er-p red ict settlem ents but give 470. Lu i, J.Y .H . & Yau , P.K .F. (1995) The p erfo rm ance o f the d eep go o d p red ictio ns fo r w all m o v em ent fo r bo th the p umping test and basement fo r Drago n Centre. In Proceedings o f the 1994 Annual the excav atio n to B5 lev el? The Ko v alue at this site is exp ected to Seminar o f the Geotechnical Division o f the Hong Kong Institution o f Engineers, Instrumentation in Geotechnical Engineering, Ho ng Ko ng Institutio n o f Eng ineers, 183-201. be ro ughly 0.3 to 0.4, hence o ne co ntributo ry facto r may be stressind uced stiffness aniso tro p y . H o w ev er, further co m p utatio n suggests that this facto r is unable to fully exp lain the measurements. N g , C .W .W ., Sun, Y .F. and K.M . Lee (1998). Labo rato ry Therefo re the hy p o thesis is put fo rw ard that the stiff behav io ur o f m easurem ents o f sm all strain stiffness o f g ranitic sap ro lites. the ground in the activ e zo ne during the w ell p um ping test is due to Geotechnical Engineering, So utheast A sian G eo technical rem nant structure in the co m p letely d eco m p o sed granite. That is, So ciety (in press). structure p resent in the so il in the gro und but no t allo w ed fo r in the Sim p so n, B. (1992). The Thirty -seco nd Rankine Lecture : Retaining m o d el. C o m p letely d eco m p o sed g ranite has a rem arkably fragile structures : d isp lacem ent and d esign. Geotechnique, 42, N o . 4, structure. This m akes it d ifficult to take undisturbed sam ples and to 541-576. carry out insitu and labo rato ry so il testing w itho ut d amaging the soil structure. It is hy p o thesised that so il structure in the m ain bo d y o f gro und in the activ e zo ne surv iv ed w all installatio n (<10m m settlem ent) and the pum p ing test (up to 5 - 15m m extra settlem ent) but, ad jacent to the excav atio n, w as sub stantially b ro ken d o w n by staged excav atio n to B5 lev el (40m m extra settlem ent). The structure o f the so il w ithin the b o x is unlikely to hav e rem ained intact d uring p iling and d ew atering d uring the p umping tests (estim ated settlem ent ro ughly 75m m ). The resulting lo w stiffness is represented in the mo d el, w hich therefo re g iv es a go o d p red ictio n o f w all d isp lacem ent. 2327