Some ceo:ec1nica >rover:jes was:e:aiincs acoons 0'inera by J. M. HUGHES" and D. WINDLH
IN BRITAIN the disaster at Aberfan in 1966 struction is referred to as the "fir tree" Fig. 2 shows a typical British lagoon dur- brought to light a general lack of aware- method and is described by Casagrande ing filling at Cadeby Colliery. Due to the ness of the problems concerning the sta- and Maclver (1970), also Kealy and Soder- intermittent nature of filling many areas in bility of spoil heaps and also a lack of burg (1969). Many lagoons were formed the lagoon are sometimes dry, whereas knowledge concerning the properties of in a similar manner in Britain and Eastern other areas are almost always under water. the material in these heaps. For the last USA before the Aberfan and Buffalo Creek It is considered good practice to move the century and more, waste from coal mining disasters, using coarse waste to build the inlet around the lagoon, thus forming operations was often just dumped indis- retaining banks out over the residue. beaches of fine waste around the sides of criminately, creating, in some cases, dang- Lagoons formed by this process will not be the lagoon. This keeps the surface water erous situations. discussed in this paper. away from the lagoon retaining banks. Recently coal mining has, in addition to producing coarse waste, produced con- siderable quantities of fine waste in the form of slurry and tailings which are often discarded by allowing them to settle in large ponds. This technique of disposing of fine wastes is common to many mineral extraction operations. The china clay in- dustry, for example, has to contend annu- ally with up to two million tons of fine discard in the form of a micaceous residue. Failures have occurred in several of these tailings lagoons with sometimes disastrous consequences. For example, the failure at Buffalo Creek in the USA in 1972, which has been described by Seals et al (1972) and Davies et al (1972), caused the loss of 118 lives. This awareness of the potential danger of these lagoon deposits and the need for enlarging and making more effici- ent use of the available space is one of the major concerns of the coal and china clay mining industries in Britain.
Method of forming tailings lagoons In the coal mining industry two kinds of waste are produced by present day opera- tions —coarse and fine discard. Coarse dis- card consists of material arising from driv- ing roadways and drifts, and also the coarse material separated out in the clean- ing and washing processes. Fig. 1. Fine micaceous residue from china c/ay processing being discharged into a The fine discard is made up of "slurry" settling lagoon and "tailings", where slurry is the fine mat- erial, mostly coal, remaining in suspension after the washing process and "tailings" is the reject material, often shale, from the froth flotation processes used for cleaning the fine coal. A similar situation exists in other mining industries. A coarse and a fine waste are produced simultaneously and the coarse waste is often used to impound the fine discard in waste lagoons. The fine discard is pumped in as a slurry and sediments out under water. Consequently the deposits are often extremely soft. Normally in Britain the banks of the lagoons are constructed with coarse dis- card or other suitable material. The fine waste is pumped in at one end of a lagoon which has an outlet at the other. In such lagoons, which are discussed below, the level of the outlet is slowly raised as the depth of sediment increases. In some areas of the world, where cli- gl Is matic conditions are suitable, the banks can be constructed with the sedimented material. The banks are built up slowly jvt-: - -.- with the soft sediments, which harden by desiccation. This technique of bank con-
"Auckland University, New Zealand (formerly Cambridge University, Engineering Department). licambridge University, Engineering Department. Fig. 2. Lagoon during filling at Cadeby January, 1976 23 Lagoon behaviour during and the inflow ceases, the clay size particles in pend on the daily mining operations. In the after filling the liquid settle over the whole lagoon, pro- coal mining industry the bands of clay are Due to the intermittent nature of the ducing an almost continuous layer of clay usually not more than 2-3cm thick, but filling and also to the process of sedimenta- over the entire lagoon. When filling recom- they can be separated by anything from tion within the lagoon, particles separate mences some of this layer will be scoured 1cm to 100cm. A cut Bm high through No. into distinct layers. At the inlet end, where away, but a substantial portion will remain. 9 lagoon at Cadeby showed (Fig. 3) layers the velocity of the fluid is high, the coarse Every time pumping ceases, a band or of clay extending the full length of the cut. particles sediment out first with progres- layer of fine particles is distributed over The banding can also be seen in the dried sively finer particles settling out further large sections of the lagoon; the thickness out 4in (102mm) dia sample (Fig. 4). away from the inlet (Borland 1971). When of these bands and their spacing will de- In a small lagoon, as the sediment con- solidates, more settlement occurs in the thicker deposits at the centre of the lagoon than at the outside, where the sediments lie over the toe of the surrounding bank. In many cases the effect of this settlement is to cause a dishing of all the layers, the effect of which is clearly visible at the sur- face when the lagoon is no longer used and can be as much as one metre in a 6m thick deposit. As mentioned earlier, it is considered good practice to form beaches around the sides of the lagoon and this also causes a dishing of the layers. The areas of fine waste that are not covered with water dry out to a limited extent and air enters the voids. On subsequent reflooding with water, a lot of air is expelled, but not all of it. This means that the material near the sides of the lagoon is often unsaturated. The thin layers of relatively impermeable clay act as barriers on which perched water tables can form. 'Y"i If an idealised cross-section is drawn through such a lagoon (Fig. 5) symmetrical about its centre line (although in practice A the beaches will be at their largest near the inlet or inlets and very much smaller elsewhere), it must consist of a perma- nently wet centre zone and an outer zone Fig. 3.Cutin lagoon deposits showing horizontal layering of unsaturated material containing perched water tables. Data, presented below, shows clear evidence of the different geo- technical properties of materials in the two different zones. The centre zone has the characteristics of a hydraulic fill and con- tains a finer particle fraction than the outer zone, which is not only composed of coar- ser particles but is also better drained, although it may contain perched-water tables. Piezometers inserted in the non-saturated outer zone will not indicate the true pore pressure if air is present. The results taken from two standpipes for lagoon No. 8 at Cadeby show this effect even with free standing surface water less than 5m » away a;-.-,-,. ANWIII4W (Fig. 6). A dashed line showing the piezo- Fig. 4. An air-dried sap/e of lagoon deposits metric pressure produced by a unique water table at the surface of the lagoon is also shown for comparison. The permea- ZONE CONTAINING PERMEABLE IMPERMEABLE bility of the thin bands of clay (which con- PERCHED GRANULAR CLAY SURFACE WATER TABLES LAYERS LAYERS WATER
0 I 6 7
HEAD OF WATER AT PIEZOMETER (m)
2- T Y ZO Il * OO PORE PRESSURE RETAINING ASSOCIATED WITH 3. HYDRAULICS FILL
/- STRENGTH 25 PORE PRESSURE ~ EOUILIBRIUM IN 5 CENTRAL ZONE PORE PRESSURES ASSOCIATED WITH 6- DRAINED ZONE STRENGTH OF A NORMALLY CONTAINING PERCHED WATER TABLES CONSOLIDATED SOIL 7- TION OF GRANULAR MATERIAL DEPTH (/TI) STRENGTH ALONG A HORIZONTAL SECTION X-X Fig. 6. Piezometer readings for a lagoon at Fig. 5. Idealised cross-section of a lagoon Cadeby (Sir William Halcrow & Partners) 24 Ground Engineering t - the moovement of water verticalli a isi lagoon the sedim t o e u ( ) on colliery lagoon residues.es. is trol y) Th's probabl sever I I n a ratio = 0.18 th thtofth o rti ( between sure at an y eve would be greater than proroducuces of — ydrostatic. When thee pore pressure m q ' Some desiccation wiwill occur in the par- reaches a wi e in a It can be seen tiall d d h e si'd e o the normal( consolidated condition. Th e servative lower ' la oon. T o con- stren g th of this central zone is controlled low a poor I ower solidatin sit. by the effective stress in th d' e s t h is stress is very low near th e sur- poorer a r 'a y replen- face the soil behaves virtually ass a liiqui;uid; Sm could be due t o the difficulties of trying rengt s en- hi h d oc- turbed samples. The strength of th e soil C count s o istur- i ecome slight( y in this zone might be predi icted from the bancece as the vane is inserted. A s t h ese nge in strength will ex ec ive vertical materials e exte yst ffi , it is a h f arse particles. The stres empton sim lifi e or even ' net effect of desiccation is to rein orce the ((1957), making the assum p tiion t h at the for a sin le la s o consolidated soili behavee aves as a normally consolidatedi a e done to make a cia . If e clay clay. This a pp roach should be a lower sible. would have many bo o e g o e agoon de- a y consolidated soil. poosit s as either desiccation w i II strengthen pr0 e ' Thec s air ried in the the sediments or the presenc f d yers ecame ex- nant( rti I il( ean a greater varied even within a partic I emely strong and very brittle. strength. us a comparison of lagoonn resresu I ts If the inlet is movede around the lagoon Thisis ccan be seen in Fig. 8 and 9, h y o limited value. Howevever a in order to form beaches at th 'd of th esesults from various'onslagoo are consider la g,oon, th e outer zone will cons f p ed together. Fig. 8 consisists of results conducted at various colliery lagoons by r ic hese will from collie for the do es. ' havee a t t th th th fiincr par-r Blidworth, Kinneil, G t orpe and Nati adeby, while Fig. 9 showss resur Its from im- Dr ve pro uced at t e undrained o china clay wast e 'a micaceous values 'ounded of izonta section of an residue'si ue) at Portworthy lagoon and Ma i e t e orm Pie lagoon. and 35 de . Co rs are so plotted in Figs. 8 and 9 is a line ' o Skempton's relationshipip oof mallymall consolidated under a ver li h
from Ma i e va ues (an old lagoon below the n .11 + 0003 Pl, wherere ir',,', hash will be of limited v e m e oe ometer dam, ' ' ing bank (inlet b g at th ere is a and thus m 8 centre (outlet area). static w bl h o of h e I agoon, dation will b do ica Th e strength of the central z (pasticity index) has beenn tak en permeabilit y of the materials. In pract'ce pend on th on o t e tobe20 p er cent from the results ofo tests much more ra piid consolidation rates would material. At the completion of fillining of t h e performed b y Wimpey Laboratories Ltd. bee expectedex due to the impimportantrt hori-
SHEAR STRENGTH (kN/m2) VANE SHEAR STRENGTH (kN/m>) 20 30 40 50 60 70 80 0 I 0 20 30 40 50 60 70 T J J REMOULDED k 2'O PEAK J J J J JJ J J
NLET REA J J JJ J
IO
12
l4 IO
OUTLET I6- AREA DEPTH —,Cu =OIB (m) 12- DEPTH
o'y'ig. (m )
7. Two vane resultss from Maggie Pie lagoon (Mander, Fi'g...8.Vane results from severeveral colliery waste lagoons (Wimpey Raikes & Marshall) Laboratories Ltd. J January, 1976 25 VANE SHEAR STRENGTH (kNI m *) SHEAR STRENGTH (kNlm~) 0 I0 20 30 40 50 60 10 20 30 40 50 60 I
k k k
k
k k k k k k 10 IO k k k k k k k kk k 12 12 k k k k k k k 14 14 k k 84
16— 16— DEPTH DEPTH Cu —, = 0 18 (m) (m) 6y
Fig.9. Vaneresult s from two china clay waste lagoons (Mender, Fi g... 10. Vane results from Portworthy lagoon (Mender, Raikes & Marshal/i Raikes & Marshalll
CADEBY COLLIERY UNDRAINED SHEAR STRENGTH (kNI m 20—I 2 Sm 0 5 10 15 20 25 30
IO PARNELL VANE 0 GEONOR VANE 10- x PRESSUREMETER
e Z v) 10 20 UJ CI
0 UJ 4. Sm x 30-
10
0 40-
10 50-
0 DEPTH I 2 3 4 5 6 (m) STRAIN
Fi . 11. Undra'g.. ined stress-strain curves for a lagoon at Cadeby Fig. 12. Com parison of pressuremeter and vane results 26 Ground Engineering zontal permeability in the granular layers. 0.11 + 0.0037PI be used as a lower bound. strength. The same research also indicates Further, with ideal undisturbed samples the In Fig. 12 are shown the results of two that leaching of the flocculants does not ap- results from these should show whether types of vane —a push-in Farnell vane and pear to cause sensitivity. More work must the soil has an unstable structure. a modified Geonor vane. It can be seen that be done before it can be catagorically The above information would indicate the Geonor vane indicates a higher strength stated that flocculants have no effect, par- that the material is not unusual. Pettibone than the Farnell vane. This is probably due ticularly in relation to other kinds of resi- and Kealy (1971) report the successful use to two reasons: (i) the Geonor vane is due, but initial results are promising. of mine tailings in constructing a road em- steadily jacked down into the ground A recent paper by Cabrera and Smalley bankment. The material was extremely soft whereas the Farnell vane has to be forced (1973), however, argues that a major fac- in places and a bulldozer was nearly lost. in, which can involve a lot of disturbance, tor contributing to the behaviour of quick However the dry material proved to be an and (ii) the modified Geonor vane mea- clays is the presence of a large proportion extremely satisfactory construction mat- sures the torque immediately above the of fine silt size particles of non-clay min- erial. vane and hence is likely to be more accu- erals. This type of particle is likely to be Undrained triaxial tests are particularly rate because allowance does not have to present in significant amounts in the clay difficult to carry out at the equivalent in be made for shaft friction. The values of layers of deposits of tailings or slurry. Thus situ stresses because the samples are very undrained shear strength produced from the thin bands of clay might be liable to soft and therefore difficult to handle. The the pressuremeter tests are higher than liquefaction in certain areas. samples are usually reconsolidated to a both sets of vane results. This is probably In certain lagoons such a failure would higher stress before the undrained tests are because the pressuremeter causes less dis- be unimportant because the bands of clay performed. The tests do not usually show turbance than either of the two vanes. are so thin that they do not affect the substantial loss in strength after peak. It is interesting to compare the stress- strength of the mass to a significant extent. However tests may not be particularly rele- strain curves in Fig. 11 with a stress-strain This will be particularly true if the inlet vant as the very soft soils will have been curve obtained with a similar instrument in position is constantly altered in order to disturbed during sampling, and this coup- a Norwegian quick clay by Hughes (1973) keep the finer particles away from the re- led with subsequent reconsolidation could and shown in Fig. 13. Also indicated in the taining banks, in which case any loss of well mask any indication of a potentially diagram is the peak shear strength recor- strength of the clay particles would be un- unstable condition. ded by the field vane at the same depth. important because it would be in the Therefore in view of these uncertainties, The sensitive Norwegian clay shows a middle of the mass of fine waste. in such the vane reduction in shear strength with situ tests as probably give large However, in all lagoons the liquefiable stress-strain curve shows a more accurate indication of the undrained strain. Only one zones of clay will not extend over the behaviour. any loss of shear strength of the lagoon whale-lagoon for the same reason, discus- In situ vane tests have produced ex- deposits and this is very much less than sed above, as the zones of coarse sedi- re- trernely variable results. Fig. 10 shows shown by the quick clay. ments in a loose condition will be locali- sults from in Portworthy three boreholes sed. If Cabrera and Smalley (1973) are lagoon, and illustrate a typical scatter. Des- Assessing the stability of the correct and the silt size particles can cause pite the large scatter a basic trend of lagoon deposit liquefaction problems, then the situation strength increasing with depth can be ob- In view of the problems which have oc- is very similar to that with coarser sedi- served. A desiccated surface crust with curred in foreign lagoons that have liquefied, ments, because appreciable sedimentation greater strength than the lower deposits one has to assess the likelihood of the sedi- will occur all the time and not just when also seems to be a common feature of ments losing strength. Two problems have the inflow ceases. many lagoons.. to be considered: (a) what happens to the Consequently site investigation on these The large scatter in vane results is due bands of coarse material, and (b) what waste lagoons must attempt to discover to the stratified nature of the deposits and happens to the bands of clay. if zones of loose material exist. Although also the zones of wet or dry material. Care The bands of coarse sediment could well it is not an ideal test the vane is must be taken to determine the actual probably be deposited in a loose state. Their relative the most suitable instrument with which effective vertical stress when using Skemp- density would depend, primarily, on the to search the deposits for loose and poten- C„ speed of the fiuid and the rate at which tially unstable areas. The vane can provide ton's = 0.11 0.0037PI. This can be — + the various sized particles sediment out of an almost continuous profile of strength o' the fluid. Therefore, as the velocity of the versus depth and is also suitable for per- seen in Figs. 8 and 9 where several of the fluid is decreasing from the inlet to the forming a large number of tests. vane results indicate lower strengths than outlet, regions of different densities and The critical areas could be assessed on would be predicted by this approach. These grain sizes would be distributed over the results may indicate that the material is their sensitivity. A large number of soft surface of the lagoon. Hence localised clayey silts in Britain show sensitivities of not yet fully consolidated in these areas. zones of very loose deposits could con- However there is some concern because at least 5. These soils provide satisfactory ceivably form, but the whole deposit foundations for buildings and embank- vane sensitivities often lie in the range of would not be in the same loose condition. 4-10. ments. On this basis, sensitivities of 5 do The dlfflculty is to locate these zones of undistur- not indicate a susceptibility to liquefaction. Pressuremeter tests, using an loose material and to assess how close and Vane sensitivities measured in Norwegian bed drilling technique (Wroth Hughes, their voids ratio is to the critical voids 1973 and 1974) have been conducted quick clays are normally well above 20. It ratio (Castro 1969). This is indeed a diffi- which is, of course, difficult to interpret vane in a lagoon at Cadeby Colliery cult task and probably not worth the effort sensitivities The sensitivities of 5-20. showed vane of 8-10. test involved to determine accurately the rele- results can be analysed a method, due However for a more accurate assess- by vant critical voids ratio and the in situ to Palmer (1972) or Baguelin et al (1972), ment an undisturbed stress-strain curve is voids ratio (Tavenas and La Rochelle, required. If this is done obtaining undis- and an undrained stress-strain curve de- 1972). by rived. turbed samples and testing them in the From recent research (Jowett and In these stress-strain curves, the stress laboratory then care must be taken that Chopra, 1971 and 1972) it appears that the o the samples are not so disturbed that they (ir, ) presence of flocculants in colliery waste refers to the shear stress have to be reconsolidated before testing. 2 does not seem to have any effect on its An alternative is to use the device for in composed of the radial stress, o„and the situ testing developed at Cambridge (Wroth and Hughes, 1973 and 1974) from circumferential stress, o,. The strain E is SHEAR STRESS the current displacement of the pressure- (SN/R~l which in situ stress-strain curves can be meter membrane divided by its initial SO- obtained. radius and is equivalent to the strain In certain cases it may be desirable to utilise the strength of the fine waste 4 sl 40 to in a triaxial test. Stress-strain either surcharge the lagoon deposits or 3 I economise on the design of the lagoon re- curves for four tests are shown in Fig. 11 20 taining banks. It can be desirable to sur- and are compared with nearby vane test charge the lagooned waste with either results in Fig. 12. The pressuremeter values coarse or fine waste in order to make a 0 I 2 3 4 5 . 4 7 of undrained shear strength support the STRAIN 4/o better use of the available tipping space C„ and also to produce a more environmen- suggestion that the relationship Fig. 13.Stress-strain curve for a tally attractive spoil heap.