ae eaninc owero'sa—a suggested remedial measureby Professor H. C. VEDER"

Because the Leaning Tower of Pisa is sett-ling more every year, its position is pre-carious. In this articleii the structure of thesoil is described and the tilt. which beganshortly after construction was started. isexplained. The author's proposed remedialmeasures are directed at the causes of theinclined position of the tower i.e. the com-pression of a 30m thick layer of clay. Hesuggests compressing this clay layer withan external load on the side of the towerwhich is not leaning. The method couldbe cavied out by degrees and is reversi-ble, and the proposed technique wouldnot detrimentally affect tourism.

on the north side. When the builder sawthat the tower was leaning towards thenorth, he had the blocks moved to thesouth of the tower where they remainedtill the end of construction.

Because soil mechanics knowledge waslimited, it was long believed that thereason for the tilting could be found in the10m thick upper layer of sand '. It wasmaintained that this layer had settled, ormore particularly, that the fine particleshad been washed away.

More recent articles have rejected thistheory and maintain that the tower leansbecause the layer of clay was compressed.

By means of numerous borings the exactcourse of the surface of the clay could beestablished; it shows a depression exactlyunder the tower. The lowest point of thisdepression is eccentric to the surface ofthe foundation and is directed towards thesouth, i.e. in the direction of the inclina-tion. The deformation curve caused by thedistribution of stress resulting from thetilting corresponds almost exactly to theobserved shape of the depression.

The results of various calculationsapplying the theory of consolidation tothe clay layer were able to explain theaverage settlement as well as the inclined

THE TOWER OF PISA rests on a ring-likes h a I I o w foundation (outer diameter19.58m, inner 4.50m) and originally had itsfoundation less than 2m below ground sur-face.

In the course of 800 years the towerhas sunk about 2m and in the processhas rotated about 5.8 deg. towards thesouth so that there is now a difference of1.8m in height between the north andsouth side of the foundation. Whereas theoriginally projected bearing stress alreadyhad the quite high value of about 0.51N/mm', today the bearing stress amountsto a maximum of 0.961N/mm'nd a mini-mum of 0.067N/mm-".

The inclination of the tower which iscurrently about 1 in 10, has, in the courseof the past 50 years, increased by 1.5 percent. Annual settlement of the southernedge is about 1mm/year whereas thenorthern edge does not change.

Reasons for the Tower's indinationUnderneath the base of the foundation

there is an layer of silty sand approximate-ly Bm thick. Underlying this is anotherlayer about 30m thick consisting of alter-nating strata of clays and clayey silts.Beneath this stratum, to a virtually limitlessdepth, is an almost incompressible layerof sand. The ground water has a slightgrade and is less than 1m beneath thesurface.

Construction of the tower was begunin 1173. It is rarely known that a north-ward inclination was observed only a fewyears after construction was begun. Aftereight years, when four storeys had beenbuilt and the tower was 24.60m high, con-struction was suspended apparently be-cause of the then inexplicable tilting.

During the next 100 years the tower be-gan, strangely enough, leaning towardsthe south '. Then in 1272-1278 it wasdecided to build the tower to its presentheight of 54.58m despite the fact that itleaned.

The author believes this may be ex-plained in the following way. The con-struction material, namely the marbleblocks from Carrara, were originally stored

aHead of The Institute for Soil Mechanics, CivilEngineering and Rock Mechanics, TechnischenHochschule, Grat, Austria.

TSI,This article first appeared, in German, in Der

auingenieur, 50 (t975I, pp. 204-205.

38 Ground Engineering

4lal e,,I~ r,,''(

iilI 'a'q'r =en'liilr.-'~Q],IF > lg~

Fig. 1. The Leening Tower of Pise, from e 1782 print

9th Ser DIc~v

caI c~ oVts2cv V

Q v0Vv0—cv .9re

rn g

4th Storey

3rd Storey

2nd Storey

1st Storey

7th StoreY

6th Storey

5th Storey

4th Storey Inclination toward south

c0VOo re

y

Ire

O Q

0CUv

Completion of 1st storey

Works continuing up to themiddle of the 4th storey'~ ~start of foundations

Interruption to constructionwork, the first break lasting about 100

I 8 o o R 0 R

cv

t

ji

years ~8 o I

Resttmption snd completionof the structure—

Between these years work resumedup to the floor of the 8th storey

The second interruption lastedabout 90 years

I I

Ifs 8 O ONej g ttm

O O

tii il

I o SYears

Fig. The early stagesin the building of the tower

the tower could again start tilting after awhile '.

Clay with ~l,'iDUnlaroUSsilty layers

3 = Dp Corrosion-proof

3S,Q tie rods

SECTION A—B

N1.3 1.3PLAN

Fig. 3. Arrangement of tie-rods outside thetower's foundation

shown that require enclosing the founfoundation of the'h o ri o o hi

il t Io b Idd mustr la er of sand tower wi a

hi h II fo dilff g Th tho'ior aresulto t e o e to pie y s

anin rocess of the tower by meanscould have practically no in uence on

F oil mechanics standpoint it the leaning processtotal displacement of the tower '. From a soiuld be desirable to b oaden the o - of a

' 'd e preversible additional load on e p

a late as close as sure-free north side. ose to eh't etric I

Remedial measures so farsible to the existing foundation nng. but withou ou

d Idb I'dthe tower was spraye wi wr the tower —for example, using t e a ovet e cpo go

A I of hi Io d i ribsmall In any event, with these proce urea i

b ary to tightly enclose the of stress on the su ace omount of the injection mate 'ial between would be necessary odin soil, the tower temporari yI with a supporting framethe basin and the surroun i g ' bl; either technique would

dsin disa eared or wit ca es;

e'ut

the soil remained unchanged.There was no inffuence on t e

'yhe stability tower.

h appearance of theof the tower; it continued to tilt about In addition, t eId b considerably inarred for1mm/year. The failure oof this measure tower wou e

an movement of years wit t ese'these methods which, from the

—is.ssd i of tourism would be most un-Slty sand

'in the groundwater it did not stan point ofine partic es in e

desirable.cause the tower to lean.R rdin (b), if one wanted to-changeegar ing

soil,the deformation characteristics of the soi,4ff s of the clay near the highof view of soil mech- then the sti ness o e

h d r ssure (south side) should be in-onsible for the e ge pres uanics, two factors are respI: crease, or ed, th stiffness of the soil nearleaning of the tower, name y:d ressure (north side) should(a) the distribution of stress in the soil the low e ge p

d. Th d formation modulus ofdue to the load of the towe, g ',takin into be decrease . e eI be increased only with

'and line of stress the clay layer can e i

d'ff It unle additional load arei icuty un si of the a lied because the modulus is irreversi e

Sand ——''(b) the deformat on cha act n o pp

er of cia . A reduction o t e e o.-1Regarding (a), e o ggsu estions in the area o w

have been made for imp'

im rovin the present only exten to e sb h b done by removing sand locally,unfavourable distributioi n of stress by have to e one

wer and the line e.g. y was ing...b hing. This would of necessityd I h h 'd dof action of the force. Howeve,wever, all the lea to a se e

t r-rotation due to the loosening ofI h v the disadvantage that they a counter-rotation ueproposa s ave e- ear old de- volume'.

I th sand layer would beer in some way Because on y ed I ti thi irr ibl

dure would be very difffcult although set-anentl, either treate, regu a in

'n ' Idb o i kl . ThiThus the most obvious solution —t'on —that of tlement wou eiece b iece cedure wou d ave oI hav to be carried out very

h f d tio of th o hI — dd h i o diio o d

ffid 40m deep foun- near t e oun a'

dat o, econstructing i t e sa pnot feasible because most of the build- take p ace. n a i

'

in the rocess. tower's inc ine posi'

I I ) Id b ff d dFoundations on piles or csissons which the cay eyer w

January, 1976 39

responds to a uniformly distributed sur-charge of about 29.4 N/cm'. This createsa distribution of stress on the surface ofthe clay with a peak value of 12.7 N/cm'.The depression caused by settlementchanges from its present state somewhatas depicted in Fig. 4; this means a counter-rotation of almost 1 deg.

The anchors would be 50m long; 10mof this length, lying in the lower sandlayer would represent the bonded section.After the injected material had set, theanchors would be prestressed to abouthalf of the planned tensile force; i.e. 245kN each. An observation period of 10months would follow.

During this time the anchor forces mustbe regularly checked and, if necessary, re-stressed. After this observation period, theanchors would be prestressed to fullbearing capacity —490 kN each —depend-ing on whether observation results deem-ed this necessary. Subsequently, theanchors would be provided with an easilyaccessible covering which did not pro-trude over the ground surface and whichmade it possible to re-stress at any latertime.

One could also achieve the same resultswith various other methods of surcharg-ing. For example one such method wouldbe excavating trenches to about 3m abovethe clay layer on the north side of thetower. The trenches would be stabilisedwith bentonite slurry and subsequentlyfilled with scrap iron. Initially the trencheswould be arranged far apart in a pattern.Subsequently, the results would be ob-served for ten months. Deductions wouldthen be made regarding the effectivenessof this method and to ascertain whetherthese measures should be continued.

One could also use mercury in under-ground steel basins. Because this metalis very dangerous to humans, however,this method is not without its problems.

After about ten years the settling pro-cess should be almost complete. This canbe said with some degree of confidencebecause the tower began settling andleaning only a few years (approximatelyeight) after construction had been begun.The clay layer consolidates relativelyquickly because it has recurring intermedi-ate seams of rather permeable silty sand.

Advantages of the proposedmethod

It is not necessary to support the towerduring construction nor to drill through,enclose or weaken it with supportingstructures.

The tower itself is not touched.The soil immediately around the tower

remains unchanged, and there is no inter-ference with the ground water conditions.

The stabilisation measures are aimed ateliminating the cause of the inclined posi-tion; they are reversible and applicable indegrees; i.e. they can be adapted to therequirements indicated by continuous ob-servations.

Tourism would not be hampered in anyway and the stabilisation process as awhole is financially feasible.

Groundwaterlevel

X

Gay surface I Aaaumad aaunlaraotationi""""~~satdamaat dua to a

a> loading with 29.4N/arnIt—Settlement dua,l

la tOWer alaaar5 tea~at: /—tower + loadmg

m

06

05

B 1.0Ti

o 1.5

f 2.0

2.5

2.0

Bibliography1. Ricerche e studi. Bautenmisterium Rom (Minis-

try for Construction, Rome) 1971.Ssnpeoiesi "II campanile di Pisa". Pisa, 1959.

3. Tarzeghi, "Dih Ursschen der Schiefstellung desTurmes von Pisa". Der Bsuingenieur 15 (1934)p. 1.

4. Terecine, Proceedings of the 5th InternationalConference of Soil Mechanics snd FoundationEngineering, Paris. Vol. III. p. 212.

5. Schulze, Muhs, Bodenuntersuchungen fUr In-genicurbauten (1967) p. 662.

Fig. 4. Settlement profile before end afterpresrressing the soil

40 Ground Engineering

tained which is superimposed on the ex-isting stress to such an extent that theresulting stress distribution on the surfaceof the clay is more even and less eccen-tric. A further leaning of the tower can bestopped depending on the size of theapplied load and its distribution. It is evenpossible to eliminate somewhat the pre-sent tilt.

This corrective process can be startedslowly by first regulating the load, by con-trolling pore water pressure, and then stop-ping when, based on observations, the de-sired results can be exactly estimated.Subsequently, the process can even beaccelerated by using an "excess load."When a certain degree of consolidationhas been attained, this load can be par-tially removed to achieve a state of rest.

This procedure corresponds to Terz-aghi's "observation method", i.e. a step-by-step approach, controlled by observa-tions, to the end result.

The load could be applied to the sandlayer by using anchors (Fig. 3). A systemof reinforced concrete slabs would beplaced in a pattern about 1m under theground surface. In each of the slabswould be a tie rod which would be drilledthrough the layer of clay and which wouldextend to some depth into the lower sandlayer. In order to avoid de-stressing thesoil, drilling would be carried out usingbentonite suspension.

The anchors over their free length in theclay layer and in the upper sand layerwould be protected against corrosion bybeing enclosed in corrosion-proof tubesand by additionally injecting the space be-tween the tube and the tie rod. After in-jection, the anchors would be strainedwith tensile forces whose reaction loadthe soil. The resulting distribution of stresson the surface of the clay superimposesitself on the stress distribution resultingfrom the load of the tower.

Apart from the small drill holes for theanchors and for the construction of thecovering slab, the ground would not betouched. The tensile stress of the anchorscan be regulated easily to adapt them torequirements; the ground is not altered.

The stress distribution active on thebottom of the foundation under the toweris triangular and has a peak value of ap-proximately 94 N/cm'-'. This produces apeak value of 57 N/cms on the surfaceof the clay. This value is displaced south-wards about 5m from the perpendicularthrough the centre of the foundation. It isplanned to load an area of about 180ms(Fig. 3) with concrete slabs measuring1.20m x 1.20m. Each of these slabs isprovided with an anchor having a tensileforce of 490kN. The centres of the slabsform a pattern of squares 1.3m wide.

The load on the surface of the soil cor-

Pressuremeter(continued from page 31)but this was possibly an under-estimatebecause it was obtained by triaxial testand no mention was made of piston samp-ling. Assuming it to be correct, E = 220.

CSummary1. "Initial" failure occurs at a nett pressureequal to about 2.5 times the quick un-drained shear strength.2. "Cylindrical" failure occurs at a nettpressure equal to about 3.5 times the quickundrained shear strength.3. The table below compares values ofYoung's Modulus determined by the airbag penetration pressure meter with otherdeterminations on the same site and atShellhaven.

E

E(kN/m') C1. Penetration pressure meter 4255 2002. Trial embankment

(vertical movement) 2 346 1123. Trial embankment

(lateral movement)4. Triaxial cell (mean)5. Consolidation cell.6. Settlement of

oil tank at Shellhaven 2 160 220

Condusions1. Inevitably there must be some remould-ing of the soil in the immediate vicinity ofwhere the pressure meter has been pushedbelow the bottom of the borehole. How-ever, because the soil must remain in con-tact with the pressure meter, we considerthat this remoulded zone forms only aminute proportion of the total volume ofsoil stressed during a test and consequentlyhas a negligible effect on the determinationof the soils Young's Modulus.2. In comparing Young's Modulus obtainedby the various methods; the followingpoints are relevant:—(a) The triaxial specimens were "sculp-tured" from block samples taken withgreat care from trial pits. They thereforesuffered only a small amount of distur-bance with only a small loss in the valueof E.(b) The consolidation cell specimens weretaken from a 10in (254mm) dia. Roatingpiston sampie from a borehole. This wouldhave suffered more disturbance than thesamples in (a) above, with a consequentgreater loss of E.(c) The embankment observations areknown to include some plastic movement,although in the calculations the movementswere assumed to be elastic. This would, ofcourse, considerably reduce the value of E.The same observation probably applies tothe Shellhaven tank.3. The recorded values in the summary arequite in line with the above points. Dis-regarding the pressure meter result, onewould expect the true value of E to be alittle more than the 3250kN/m'or thetriaxial tests. It would appear, therefore,that the penetration pressure meter givesa fairly accurate measure of E.4. "Initial failure" of the clay occurs at anett pressure of about 2.5 times the shearstrength. "Cylindrical failure" occurs at anett pressure of about 3.5 times the shearstrength. The test is simpler and marginallyquicker than using the penetration vanebut the former entails a determination ofbulk densities in order to evaluate nettpressures.