Construction Histoy Vol.10.

1944

The Failure of the Bouzey Dam in 1895

NORMAN A.F. SMITH

"TAank yo14 for the copy o/' "Cnsaiel-'' with you/- inreresting article un the Bouzey Darrr. It must huvc hern rm rrwfully jerry-built uffoir ~f the vertical hnnrl was so wholly uhsenr that it Iooked as if the mass filling the gup hud slid und $the mortar joints, incfudin,q the vertir.ril bond, wouldn'! *fund surh u paltry p l ~ I lUS 20 lhs. per sq. in. - uhmut twice rrs mrrch us u little boy's sucker!" (letter from Sir Benjamin Baker to Prof. W.C. Unwin in 1896)

Intruduction

The Boui-ey dam near Epioal in Easterr~France railed almost exactly 100 years ago, on Saturday 27th April 1895 at 5-45 in the rnurn~ng(Fig. 1). The ensuing flood-wave, pourlllg northwards along the valley of the Avitre (F1p. 7) to the Moselle, drowned 85 people and extensively damaged canal works, ralluay structures, bridges, villages and farms. fiy thc standards nf dam d~qasters ~t was a serious accidcnt although not so destructive of 11Fe ds the Dale Dyke darn failure' near Shcfheld In Mnrch 1864, which killed 244 people, and nothing like as terrible as the death-1011 of over 2200 in Jolinstown, Pennsylvania when the South Fork dam collapsed in M;ly 1889'. For comparison, 75 people were killed in the famous Tay Bridge rrililwi~ydisaster of 1878. There was another aspect of the gravity of the Bouzey frtil~tre,in tbe wider context ol' engineering safety and designer responsibility O F great qignilicance. The Bouzey darn, it was believed, had been desigrred rationally by the reasoned application of tnathematics to n

The Fatlure of the B o u ~ e ?Dam in 1895

structural problem. Those engineers In France, Brltain, Germany and the United States who had pioneered this hrst era o r d a m design were c o n r t d e n t t h a t n e w s l n n d n r d s in saftty of perronnancc and economy of construction were now to hand. T h a t they w e r e a s t o u n d e d a n d alarmed when a darn of the new age went su badly wrong 100 years ago is hardly surprising. T h i s is not to b e a " d i s a s t e r " paper. It i s about design and cnns ~ r u c t i o nand why structures fail. T h e r e f o r e , t o b e g i n w i t h , it i s neccssary to examine, nt least briefly, to the Ruuzey d a ~ n ' s the backgrou~~d short life.

Norman A. F. .';mith
H o w f a r such rudimentary calculations were applied to practical darn-builcling in thc r: * Y latc eighteenth and early nineteenth ceniuries has so Ihr been impossible to esttlblish. Sergc ~cliilvsky'maintained that dams of the period were designed by Belidur's concepts, to which was added the requireinent that the rario of p r e s s u r e to w e i g h t m u s l not e x c e e d t h e cocl-Rcienl ni l'rictiun a p p r o p r i a t e to tllc masonry itself and the connection between he dam artd its lioundation. What car1 be said is a that :I number of Frerlch canal darns o l the period - e.5. Lnmpy un the Canal du Midi; Vioreau, R o s m ~ l i a cand Glomel on the Canal de Nantes 5 h e s t ; and cIyulzilly and Groshois w i o n the Canal de Bourgogne, ranging in date from the 1780s to the 1840s' - have profiles liL which are cenainly not at oddh with the "nooverturning" rule hut that may be n o Inure than coincidental. Fig3: Bclidor'a pmblcm In 1 8 5 3 calnc a decisive s t e p . J.A.T.de Sazilly publ~sherihis pioneering st~tdyo r darn d c s i g n , "Sur un profile d'egale resistance proposC pour les murs de reservoirs d'e~iu".' This long and complex paper was a further French contribution to t l ~ c study u F retaining walls generally and was based on previous w o r k in t h e strength or materials by. ror e x a m p l e , C.L.M.H. N a v i e r a n d I.B. B6laoger. D e SazillySs irleas can be Ibllnwed, in their essentials, by referring to Fig. 4. It was a s ~ u m e d that a slraight gravity darn cuulrl be visi~alised as i l series of separiite sticcs o f unit thickness, and that any one A B could be treated independently o f all the othcrs. The Iin~itatior~s to these assumptior~s are considerable and ohvious. The height of a h dam vnfies across a valley and, pal-tiuularly near hi abutments, slices interact with each o t h e r a n d p r o v i d e additional resistance. H o w c v e r , t h e m o s t c r i t i c a l c a s e c a n be covered by annlysing the tallest slice and a s s u m i n g it to be u n s u p p o r t e d by its neighbours. De S a y i l l y s l a t e d i11 f u l l t h e t w o reclilirenients mentioned above: that it should be inlposr~blelor the section of darn above X Y to slide relative to the part below and that t h e line o f thrust representing W and P ~ h o u l dbe contained within XY so that the ~i~4: D~ ill^'^ ofequal rcsistimcem; the conlponent ul' water p ~ s r u r e i , ignored piece AEYX could not overturn. A huge

P~NAL

I Ii

The Background
M a s o n r y d , m s - g r a v ~ t y ,a r c h , arched and buttressed - have been hullt f r o m e a r l y t ~ m c ? while the E u r o p e a n t r a d i l i o n ha5 c l e a r l y dcfined and well attested orrgms ~n Fig ? : Tlle Bourn D.>mdnd 11.; envlrt>n\ the R o ~ n i l n Some R o m a n dams are a1111 In use as iire Metltevnl structures, albeit wilh the aid, in most cases, of considerable repairs and recot~structions. There ale big dnms of the sixteenth cenluq and later which, so f:';lr ns tror~ble-rl-ee service is concerned, have much twtter records, indeed enviable. Some Spanish dnms of the sixteenth :~ndseventeenth celituries are a particularly i~i~pressive lestimony to what could he achieved. and there are numerous exnrnpics of thoroughly successfu[ d i t n ~ sbuilt in Prance, Italy. G e r m a n y and Central Europe t h r u i ~ g h o u tthe whule ol' the early modern period. the beginnings of a tradition carried on wcll into the nineleenth ccntury. 1 mention a11 of fhis to make the point, rvhich if space permitted would bear cnnsiderable el:~boratiun,tlrnt evcn in an era before clums cou[d bc designed rationally. they could still k designcd very well. The old rnetllods (prc-1850) of proportioning dilrns, whatevcr tl?ose mcthods crlnlpriscd, were not unsaCe ever though thcy were undoubtediy uneconomic. Even in [he eighteenth uer~turyttje~,t was an emerging inte~.est in c:rlculatirrg tlre ability of a niasonry wall to resist overturning when loaded by some horizontal force. Charles C a u l o ~ n bh i d applied l ~ i ~ n s c to l f the particular problem of retaining walls bearing earth pressitre, w h i l c B,F.de Beliclur a d d r e s s e d t h e e s s e n c e of t h e darn p r o b l e m i n his consideriltion o f the behaviour of the walls of navigntion locks.' Relidor exanlines thc stability of a r e c i n n g ~ ~ l wall a r o f height h and width g, to be deter~nined. when resisting ;I depth uf water n, less that1 b (Fig. 3). I! is assumed that masonry is 5/? times as: l~eiivyas water. By qu;rting the overturning rnonzerlt due to the water pressure, P, and the restorir~g mument developed by the weight of the wall. W. Bclidur shows that y' must be at least 11.' /5b and in additinn, to achieve a safety factor of 1.5, he shows that y' has to be ?ti13 /lob.

49

The Failure of the Rouzey Dam i n 1895

advance was de Sazilly's consideration of stresse5. In effect he was extending t h e ideas o f N a v i e r and Betanger to dams hy treating each slice as a heavy vertical cantilever and assuming a linear distribution of stress on any horizontal w z r a 8 sm m section. When thc reservoir is empty the mE'iAES self-weight W I S taken to i n d u c e a I trapezoidal compressive slress diagram is at the water50 XYqp whose mi~xirr>um race. With the d a m loaded the m a x l m ~ r mstress will occur at the nirrace sincc the i ~ n p o s i ~ i o of n P induces bcnding stresseq across XY to i ~ ~ 10 ld the direct stress d u e to W. I n o t h e r words, the line of thrust moves fro111 l e h lo right as P increases. The calculation of t h e s e states o f stress was. Tor d e Sazilly. perfectly s t r d l g h ~ f o r w n r d a n d In o r d e r to Fig 5: The Fulcn.; Durn. 1x66 maxinlise the r n a s o n ~ y ' sperformance and economise on conslruclion costs he set a cot~dition.H e specified that tl~roughoutthe darn o n every section, at the poinl X reservoir empty, and at the poinl Y reservoir full, the r n a x i ~ n u n con~pressivr ~ stress that tlie material could withstand sl~ouldbe I-enlised. Hcnce the [ern1 "profile of equal resir;tnnce" used in his title. It means, oi course, that the two compressive stress diagrams, rcscrvoir full and reservoir empty. :)I-e mirror images o f each other. It also incans thal in Fig. 4 every horizontal section is subjected only to compression. nevcr ro tensiun: tlie shape of the prohle and de Sazilly's conditions see lo that, wcre npplied immediately. The Ful-ens d;un near St. Etienr~e. D c Sazilly's proposit~or~s genen~Hyreckoned to be Ihe first "n~odem"dam. was designcd and buill hetwecn 1858 and I Sfib (Fig. 5). 11s designer, F.X.P. Elnile Delocrc, in his c l i s c u s s ~ o of ~~ dam ~ design follows de Sazilly cxactly and their concepts were ndopted in~medintclyas the hnsis of European and American prnclice Ibr half a century. The next i~npol-t;~nr advance in the thcory of d;lnis came frurn Prolessor W.J.M. Kiinkine. In I870 he was consulted over rhr desien of thc Tansa d;ln~in India. In a iarnouq paper of I872 - one of the Inst and one OF the best u T his pl~blications"- he fou~tdin Favour oi the basic de Sazillion ideas h u ~ added new ancl crucial cr.mside~,ntions. Twu werc par-titularly crilical and for bi-evity we will contine ourselves tn t h a n here. Rrinkine stated tirnlly that nowhere on any horizontal section of a dam sl~ouldtensioil be allowed to dcvetop, rither at thc air-Lice, reservoir elnpty, or :1l Ihe watcr-face, reservo~r F~III. It is essential to re-e~nphasischere that i11 de Sazilly's protile of equal resislance. for the in ~ Figure 4. tlie "no-tension" condition is, in f a c ~ .met. style of cross-sectinn s l ~ o w t Iiowever. neitl~er de Sazilly nor h ~ disciples s positively spcll out the need to avoid tension. They recognised that it could occtlr and indeed, f o l l o w i n the earlicr worka or Nnvier, hlCry and BClanger, knew fl11l well that according to their theories, if the Corcc R, the I-esultiunt of P nrtd W, cut the section XY outside the "middlc-third" - i.e. acted with an eccentricity of greater than XYl6 -then tension was inevitable. In de Snzilly's paper thtre is a remark, at one and [he same time chilling and portentous,

N m m A. F. Smith
which slates, apropos tensile stresses, that it is allownhle to neglect "the force of cohesion i n the mortar, which is u n f a v o u r a b l e to resistance"." T h e o p i n i o n i s r e p e a l e d v e r b a t i m by Delocre. In short, the d c SazillyDelocrc view of tension in a masonry d m was not that ir should be avoided, but r a t h e r t h a t i t s c o n t r i b u t i o n lo struciural strength should simply hr discounlcd. Rankine knew better. And so it was that he brought into gravity dam design a new imperative: that there must be a strict observmncc o l the socalled "middle-third rule". The other of Rankine's observations c h a I l e n ~ e d t h e a s s u m p l i o n that ~naximum stresses did, in fact, occur on horizontul planes. Rnnkine puts it thus: -b"Thc direction in which the pressure Fig C: The haw rri;lngul;lr prnfile is exerletl amongst the particlcs close to either face of thc masonry is nrcess;lrily that oi a langenr lo that face; and, unless the FACEis vertical, the vertical pressure Found by means of the ordinary formula i s not the whole pressure, but only its vertical componeni: and the whole pressure exceeds the ve~tictllpressure in a ratin which becomes the greater. the greater the "batter", or dcvintion of the h c e from the verticnl"." True, and quite correclly sensed here i s that ~ n a x i m u r n171-inril~ul stresses occur in directions parallel to those of zero shear. The race of a darn, by definition, is not subject to any shearing stress and therefore the muxin~iirndirer! stress at the race of a dam must he on a plane at right-angles to that face. Moreover, its magnitude must excccd that o f the associ:~ted verticnl srress. Following Rankine's paper, a nunrber oYcon1ributor.s to [he theory ofdarns introduced variations on his criteria and by degrees the complexities of stress within n dam, including the role of shear, began to be assessed." As background to the Bouzey dam's design, which began in 1876, to its demise 20 ycars later, we can simplify the essential design concepts prevailing by considering the most b a s i c of all dam profiles, the triangular one showri in Fig: 6 . 11 is defined by the reliltionship hl=b2w, where w is the spcciiic gravity o r the masoi-lry. It is elclrlentary to eslablish a number of properties of this par1icul;lr form or profile. For example, il can be shown that on the typical section XY,reservoir elnpty, the co~npressive strcss diagram is itself triangular. If thc s a m e maximum compressive vertical stress is developed at Y. when the rcservuir is full, then the saiile triangular stress distribution, reversed, will obtain, as shuwn. This will be true at every level, and in a sense what we linve here is a profile of equal resista~~ce. However, anrl this is crucial, il is not, and cannot he, the profile of equal resistance because the r n t l x i n ~ ~ ~ working tn stress of the masonry cannot be realiscd ot every level. Approaching the c~,esr of a dam. up to and ahove maximum watcr level, stresses decrease progressively so that de Sazilly's theoretical profile is, to thnr extent, unattainable. O n the other hnnrl. the triangular protile automatically meet* Rankine's middle-third condition because the lines o f pressure, rull and empty, pass through the two mid-third points, respectively, at every level. Thus we have here rlre mosl straightforward

51

T h Failure of th Bourey Dam in 1895

N m n A.F. Smith
increasingly slender profiles. Hence the risk or t e n s i o n w a s i n c r e a s e d ; t e n s i o n , remember, whicb in some quarters there wns a willingness to discount. T h e maximum stresses allowed hnve n marked el'fecl on a dam's profile and Ibe height to which it can retain the simplified triangular shape we I ~ a v ebeen visualising. It reaches ils limits according to the range of numbers given iri the Table in Fig. 7. To have a higher dain not stressed, fronl of back, beyond the prescribed limits requires a base OF increasing width, and thus the more FnnliIiar shape of a high dam begins to emcrge, as in Fig. 8. This need to modify the basic triangular protile is c o m p o u n d e d b y a n o t h e r requirement, very relevntlt to B~ouxeyas we shitll see shonly. A dam with a pointed crest is not a very practical proposirion; i l is fig 9: Dam with widened c r c ~ ciificr~ltto build, vulnerable in use, and in any c a s e tails to provide a roadway T o r maintenance, access 01. traffic across the structure. As soon as the crest of gravity dam has some useful width, lines of thrust move and middle-third requirements arc offec~ed.Consider a width of road addcd to the top of a triangular profile as in Fig. 9. It is evidcnt that when the rcscrvoir is empty the mass o f the piece ABCYX, will, below n certain level, X Y say? ~ h i r lthe line of thrust jbrward and violale the middle-third rule on rhe air-ftrcr. This is very unlikely to be n problem but it is a fact. On the othcr hand, a very rcal issue arises when the reservuir i s full. The fact thal the top of the dam now has width means tltnt above a section such as mn, thc cenlroid of the profile is in any casc rnuch nearer the air-face. Thescfore the possibility of generatir~g tension at the rvnter-bce, at a point such as m, is greiilly increased when the reservoir i s Full. In other words, on the section mn, the line of thrust, R, is torced to the right by the addition of ABC. So it emerges that an urlexpccted properly of a prartrcnf profile is that it is particularly s~lsceplibleto unfavournble stress distributions tical- ~ h d top. In the old days it was two easy to, be unaware or Ibis, the not unnatural assumplion being that dams are vulnerable only in the depths where pressures are high. For a number of nineteenth century gravity dams - and Bouzey was a tragictllly outstnnding example - it is elementary to show (hat the middletllir-(1 rule was near[y or ac.ti.rul(y a*iolutedu n h ricar rhr cwsr. It was a dangc~,ous silualiun all too readily compounded by the fuurth factor w e are considcring. Inevitably designers and users alikc aosurne that a reservt,ir's maximum water-level is known, predictable and controllable. However, a potential dnngcr for any dam is that i n s u h i e n t spillway capacity, or exceptionally n p i d reservoir filling, or both, can produce an overload. It was e s ~ c i a l l y true u century or so ago. Reservoir bchaviour was, in its way, as problemaiicnl as dam behaviour. Indeed, the ratcs a t which reservoirs 611ed. and as vice versa. overflowed, were as I-evealingof rainfall strttistics and catchment area r~rnofr So it is not difflcutt to nppreciate that Inany a spillway was scriously under designed, and that a dam which wxq already stressed lo the limit near its cresl at normal maximum level,

Sgecific Gravity of Masonry

PrincipaP

Stress kg/cm2

H E I G H T

metres

20.00

66.7

61.5

57.1

Fig 7: Llrniri~~g lheifhts 01 trrangubr dam5 Ibr v;rriouu rpzc~ficprdvities ; ~ l dl d tFrrrn! rnruilnlln~principhl strcsws

profile fur a graviry dam. For the air-face, maximum principal stresses are easily calculated as heing equal lo psec20whcre p is the rnaxi~nulnvertical cornprcssjve stress nnd 9 the angle o f the race, Now this basic shape, and its behaviour when loaded, are s~rsceptibleto variations in ur least four parameters. all or whlch can have a txming on the desigrl process and n dam's safe operation. Fi~,stly, the density of masonry is a variable. Ceneridly it was not usual to use a stone whuse spccilic gravily w a lcss ~ than 2, while i r ~ thc case or a reaIly h e ~ v y ~ptcimen, it could he as hiph as 2.5 or a shade more: the masonry uf Lhe Vyrnwy dam iri Wnlcs was particularly dense at 2.6. Interestrngly enough, between these cxtremes of cletlsity, :I ZrianguIar profile is not rerluiwd to change all [ha! much, only from ahour 32 to 35 in the anglc of the dam. Secondly, and much Inore signilicant, is the znasonry's compressive s t ~ n g t h b , e it measured or estin~ated.The latter was il mill enough approach. To get stiirted on designs French engineers, such as Auguste Grneff, applied their nnalysec ro historic Spanish d;~rns a n d d e t e r m i n e d t h e prev:iiling maxlmurn stresses." T h e unswers were prwtty ~nixeri; r r o n ~as l u w as 6.5kg/c!n2 lu a s hlph as 14.5kplcln'. In lhe early days, and lo be safe. the lower linlits were chosen; the Fur,cns dam is s t ~ r s s e dto h.~kg/cm'. There was a horzus to this cautious approach to rnnxinlun~stresses. Because ir lends to a relatively w i d e d n n ~ x profile - and not a parliculilrly econo~nicone the middle-third rule is. forluitoilsly, likcly to be met. However. in t l ~ c i rdeterminillion to build more highly stressed dams, and hence o n e s w l ~ i c hmade m o r c e c o n o m i c u s e o f m a t e r i a l , as w e l l a s s a t i s f y i n g a c c r t a i n ambition to achieve "proper" desirn. European engineers generally were inclined, over the nruat wtdcn 10 w%trict ~i~S: ~~l~~ XY the ~ t r ICVCII e ~ ~ yeilrq, to allow h i g h c ~ co~npressivestresses in

hy

[ '%.J U! umoys arro aql s c UO!IS~S-SSOI~ ~ q,tusp Kaznon ayl Ipnq sy zuea.rlsdn aql uo wrrp aql j o aspq aql ~ o p IUS q SI?y311~1 SF UMOp 8~!43CaJ lilnq SEM I I R M ,lJO-lll3 B 'suo!~t?punojaql jo xnlcu padnssy at11 Jalunes

I[!"

\\r:

luea darnog

a q :]I ~ 3!g

jnq anlnn R $0 6 1 ! n n ~ 4q s a d s rrv puq map at11 pl!nq 01 pasn leilour p hruu~i!uralI.1, ' r u c l z IF q4!q .Jllcrln11da3xa I lnq 'UISZS ~ r4uol ! a~!nbL ~ ~SBM~j l aS y w !

'z

c paldrilalle a.molaa put! LTI!ZI?S a a .UO!~CIII!S uan$ s l q t ~ n l s ruoy l ~ al!lold ajes F u%!sal) 01 Jallt?m ]ua.rajj!p rr a11nb s! 11 :slahal walls put! i(1!1!qers .IOJa ~ y o d dSrr!ls!xa rlc ;~rLlcu 01~ - p~sm~ojlq%!t?~ls iClah!lqa~ pua - drl!yl auo s! 11,urnnip aq tsniu suo!]nu!]s!p .radoid p u t s u r a l q o ~ dI ~ I ~ J ~ aJ~J! n Ih ~ oh11 ~ ol ps!yddn aq uss suicp j o A.~oarlliluv .( yasl! u! ~narqo u a ST IOU 'suo!s!sap Jaqlo jo asuanl.rasuos E se L ~ u o I I paha!rpn ~ trallo h J a ~ seM .caln.r p.r!rl~-alpp!ru,, s,ao!ynsa JU a3uenJasqo .an!\s~adur!u%!sap B IOU SI!M .,uo!sual jng 'uS!sap j o sle!luas<a a111 SB ,,~LICIS!S>J

53113 W

$0 J a p m ayl J O Pu!t~larr~os p a a j plno3 p u c 'sa~clsarl 8 ~ paJaho3 1 I! 'sarlaw n!qna uo!~!ui L SDM kl!>t?dl?? S,J!flhlalal > I l l '1 88 1 (I! a)t!l ut?Eaq JTohlaca ayl jo 8 u ! ~ yPUB 0 8 8 ~ JO pua a111~c 1q3!~ palafduron s c wep ~ Aaznon a y ~ -paaprl! uS!sap .~ood k a n r! aq 01 ]nu par1.1n1 1nrlm 30 Ll!lnnas aql aarrcqua ol

= G y t

o s 01 palaa[qns aq plnoqc 'Alie!~!u! Isgal l e 'tucp aqi lnql JO '.131cm q311U1:o r papaat1 'a8cls 11:yI in lscal 1c ' a y j ~ ~ [nus3 l ~nrp pasu!huoxm q3!rlm saassneqa la sJuo6 sap InJ?rr?D [!asrlu> ay] j o 1rr~nrddn aql she[[ iou p!p s q q n d xncncJJ, sap alls!u!w at11 Q rlo!s!nap s t q L .1~1Z!aq ~ [qny saulsp a111 01 LUZ ppr: 01 pau!nlqo scm uolss!uuad 0881 ~ a q w a i d a u! ~ puv kaznoa 1% L~!l!q!ssod stql pas!r1303a~s~aau!Zua s,lsg,l ap ~ n r a qq b 'i1!3~drr3J ! O A J ~ S U! ~~ J S B ~ J ~ aBnq I I ! t! as!leal u31jo ue2 mlrp r! j o lyS!ay a v u! a r t ? a l m leu!ulou 'suJniar Su!ssaasu! JO me1 S u p d u ~ a c ~01 inarqns ~m 'Janamoq 'w!on>asax ' J a m ja WZ Inoqn plat1 aheq 01 snM urep papualrr! aq] ' 9 ~ U! ~ 'L~TVU!SI.IO 1

s!yl 01 Elam an.las 01 aiiu!luon palsal3 sarnlsn.lls all) prlc 'sa!.rlrInns le.lanas u! hl~nlilRal I>ana!qx SKM 11 .qq!ssod X~1mj.13d SFA~ sump Ll!ne.r3 'L~uoscur -qZ!q J O L I O I I ~ ~ . I I S I I O ~ prle 115!~3p punax all) 'slaqlo pue atr!ytlux ' s n o l a a 'Lll!zr~ ap j o Y J ~ M ayl S u ! m o l ~ o ~ .(I 'aJnl!q :\tug ayi p a p ~ n n ~ IIn!qM d ~~o!~nn~lsuona.r >n!suajxa plrr a~ueualr~!nu~ I n s l ~ ? a i aql ( Su!.tnp pup '~li?!qapIFII!Z!JO sch'>znoa do po!.~ad ~ q u! ] sulcp do 6 ~ o a q l 141 lnoq': apnru aq ol suo!lch.lasqo 1s.lauaS aalrll a t ! aJaqL m q a q luam irrqi asor11 pu!rr~Janau 'luapnni? daznng aql p n , u o l p f q n ! q ~suo!1e.laq!lap 1111 u! rrana pamp!suua t;r.na.lnilt?J ~ ~ q ~ayi 1 . lsyl j ssrrap!i\l or1 aq 01 sirlaas aJaql 'Isej UI -1ou sem I! Lllua~edd~? 111~ papaar] aq 01 Buirutm n uaaq aneq 1qnnqs1??qt?1.1 lualxa l c q ~ 01 puc 'awns arll Xlas~aaldaq rll sr:m Lazllog 1n aJnI!l:j j o trls!rieqaaru aql 'aas llerls am sv ,laha1 I n w akoqr uo!lnas U I I ? ~ ail1 S r ~ ! l u ~ n ~ a h o rr! a l l ~ s s x dJaeM Imuuzuoq aql pa1uam9nr: I! '133jja ul ,yaeJa all) U!EII!M ~ I I R ~ !4 U ~ !~ 1A 3~ st?^ 11 .uo!13v olu! altrv3 a301 Mau t? qn!qm ~o ] ] n s >r: ~ se 1a)sM $0 uo~lw~arlad
anlo] jj!ldn un

43!4M ~narold~ u ~ i i o d m X~mln~!urd ! e

SUM [anal JlonJasal l e t ~ o ~ l u . r a d o ~YUIJOU 311) prlr U I I ( : ~ n a u j o yjdap c ol p ? d d o l i a ~ oseh$ 1sar3 arl>L ,j13sl1

c,'a'?!AV J a q q aql 3u!tum~p Lq palnaln seM J!oAJasal X a z ~ ~ oayl n lrrql u u ~ t u n s s!yl Llddns 01 SFM 11 '(71 ' : ! f ) 11~uoucJIt) 40 a3'q1!h aq1 pue a8p!~q pro.1 9910 a111rlaa,\zlq wy01 I n j sun.r I ! : l e u ~ d g JO I S ~ M aql 01 ~ s n palenl!s ! sl I ~ , ap I Ieue3 a111 JO t10113asmaqlnos a111j o lahaI-l!wuins aqL '11!4a a m e w cl 3p ik-11~3 II!nq Xl)palln aql j o WYOZ Bu!s!l!rn Lq InoL jo lsaM palsautrog alaM urn1 at[L 'auogs Jah!x aqi rlo a.103 01 m0.g uopnas ulalnnos e pue ! L a s s l t c ~ ~ ol 1aA!(3 w o ~ J j . I a s n aJan!x ~ ~ pas!Ivuen dllerruassa 'ue!l>as rlJarpJou B : s l ~ v d lau!js!p OM1 JO I)als!suO3 lsg,] ap IsUeJ aqL ' 1 " ~u1ss111~-03urtr~ ayl j o agualthasuo3 s sr: uralsLs au!qx -alTasoW at11 S u o l ~ uo!l*s!hsu ylnos-quou J O ssol s6a3ucqB U I M O ~ ~ ~O J~ U! 8unBaq 1 STM %-EM 1~3 ap ~1 [Fur3 a u .U!EI!J~ 1 8 a q u! luaarn9.w p ~ sKvrnl~v~ay1 q i a i p Ouor 3u!p[tnq Inuna ~oruura v i l a p u n oj panu!luoD a n u c q aql
IIOM

',,

.-

-1
--13h91

amO,d

aluos OI lsaq any9

.op Lllensn k a q ~ 'Lep aql UOM 113!y~ 'asuas rlourulo3 pup anua!~adxa jo 1q5l 341 U! payporn Llsno!n!pnr 'sa.mpa2ord u41sap dals -Lq-da~sa p w L~an!~slal SPA\ I! Lpatutrln 111q 'uo!tn~os TnlauaS e Bu!sr[na~ III flqssamns slow SBM '~!Is!~I?~J Sulaq pun sr101ssaauo3 4u!y~tu ilq 'au~yunaydarr! X~1m~ynuraqln1l1 aql lo I)aueaq-lu!vj 2q1 JOJ ~ o u s! yloa l r a y ~ $0 Zu!p~a~ asap II :suorlrnha sno!Z!pord 01 peq Inq uralqold aq) 01 uo!lnlos l e y j e u a y l n w p3s!Iwau?8

J ! UO!~!PUO~

'w.n ]! ?e q m s 'Ke~1~!ds arp sem Xluo ]ON ' ~ J s / , w ( ) w se~ tIni1ur n! 'mnur!xl?ur sl! '40 LUFP 1111 ~ ~ ~ o l l l a n o uc p a ~ n p o l dCMOIJp o o l j a 3 n y pue SWJOIS J a u e jjoun] an!sseH 18s 1 U! ~ ! l > U ~ 0 : [ %!A) ~ Ulltp l2Jq.H 3 1 . [ l a j q adX) s ! q j o a-ln[!vj s n o w y y '~aqS!q u?Aa as!Jol laha1 lalnni a q l p ~ ~ o l n'smll!ds lc a l ~ n b a p e utin ~ a n p r: rr! STM +passnx!p SF

The Failure of the Bouzey Durn i n I895

Nmman A.F. Smith

Top water level was at 37 1.51~1 above datum o r 2 1 . 1 1 1 1 above the l'ounda~on level. As the ~cservorr Ibe~an to till, su ~ h dam c begall to leak: even at waler depth of less than 15m the leakage was up to 65 l/sec In D e c e m b e r o f 1882 t w o l i ~ s u r e smnterialised to add to the leakage They were put d o w n t o l o n g i t u d i n a l corltraction of rhe structure in cold weather. Come December 1883

I/
I
1 1

/L

Fig

12:

Pbn and eleuor~on of

llte f ~ l l o r ruf :

I384 (not lo scale)

the reservoir level had reached 15.5m, that is to say about two-thirds thc inaxrmum, and at this moment the water-supply to the reservoir was augmented very substantially by the completion of a 43km-long aqlteduct bringing Moselle water from Remiremont. Uy March 1884 rhc reservoir was 18.6n1 deep; the leaks hacl grown only a I l ~ c t i o n to 70 l/scc. The level rose a lurtI1e1-80cn1. .4nd then, o n the 14th of Mxrch, just before noon, 1351n of the dam slippcd forward into the V-formation indicated in Fig. 12. The ii~aximumdispl~cclncntwas 35cm and the hrenk occurred at Fo~lndatinn level (Fig. 13).If' By any reckoning this was a serious sir~lation and yet the asrotlishing thing is that not only were n o precautions taken, the reservoir level was actually allowed to rise n filrtller 20c1n, to 19.6111 of depth. In {his c o n d i t i o ~ ~ for , over I 8 months. [he Bouzey dam corltinuetl in use with watcr pouring through cracks at the rnte o r 232 l/sec. The reservoir was drainer1 in the autumn o r 188.5. Not surprisingly eroups of vertical cracks wei-e round at each end of the displi~cedsection and at its ceIltrc. They, quite evidently, were rhc result of bending frnctt~res.tension cracks in a broken "horizontal beam". Underneath the dam it was Found thul the cut-OFF wall had no nleans protected the poor rock i~nrnediatelyunder the structure. Fissured ant1 perlnenble bed$ reaching well into the reservoir nrsa had not beell sealcd Fig 13: The wture nf the frecture by the cul-ufi wnll, and the f o r m a t i o ~ ~ under s rhe dam were dislocated ant1 sntilrilted to a depth of 2-31n. Something which all Ihe plonetr allalysts of d i ~ m s hat1 thoughl about but never really believed c o ~ ~ happen, ld in fact had. A da111 had slid on i ~ foundations, s an effecl probably cxacerbntcd by a considel-able uplift presslrlt under the dam, although h a t factor W~LF not fully debated until a later date. Several Feirrilres of the 1884 accident and its aftermath are intripuirlg. Tlrat the broken dam was e x p c t e d to cuiry on for 18 months cannot. surely, have been a decision olengincers. It muxt retlecl the Canal de I'Cst's paritmount need lor water to s u u ~ i n traffic.'' And yet once the reservoir was drained, progress was very leisurely. Explorarion of the foundations occupied the yea[-$ 1885 and 188h; reports were prepared ant1 discussed in 1887 and 1889. There WCIY coriflicting opi~lions as to whar had happened and what to do next. Oddly, the

o n e thing which was never agreed s u r p r i s i n g l y g i v e n h o w m u c h trnlc e l a p s e d - was to d e m o l i s h l h ~ s illstarred structure and start again." The best chance to ensure the dnm's rulure 0 2 4 4 s t n was missed. One detects here, though METRFS t h e issue is n c v e r bronclled In l h e relevant literature, proround economic problems. It was a post-war period and t h e C a n a l de I'Esl was n e w and e x p e n s i v e . The u r g e to u s e t h e navigation and make it pay musl have \ beell powerful. Therefore the Bouzey darn was repaired. O n the upslream s i d e at t h e base m a s o n r y w a s used to s c a l t h e g a p belween the body O F the dam and the cut-aff wall, t h e w h o l e repair being generously covercd wirh puddled-clay. F o r lhe downstream s i d e there were s c v e r a l proposnls. The m o s t e x t r a ordinary w a s lo construct an earthen Pi8 IS: Tile Bouzey D;lm rebuilt Ithc old darn 1s hhuwn by thc b u l k to the full height of d ~ e dam; the broken Ime) best was Ibr n series uf la11 buttresses in n the dams of G ~ . o s h i sand Ch;~zilIy.'"In view of what the manner of the measures laken i was to happen in 1895 the decision 11ut to build "contreforts" was another chance missetl. In [he event, the chosen solutiwn was a massive buttress "toothed" into the original masonry and h a r i n g against a deep lcvel nburment. The intention was to prevent turther sliding. RUI nu strengthening was provided Tor the upper half of the dain and (ha1 was where the ultimate disaster was lurking. Drains were installed at thc base of the d m to take away any wwar percolating unrle~,neath.The verticnl cracks were filled with cement mortar or grout. The bure-hule user1 to investigate cwnditio~ison ihc lowel side of thc original dam was retained as a hasis for monitoring water-pressures beneath the modihed structure. The rebuill croscsection is shown in Fig. t4. On the 18th Novcmber 1889 the Bouzey reservoir began lo refill. The level passed 19.4m, the depth at which the accidenl of 1884 occurred, i A February 1890 and suhsequentiy the water level was relenllessly increased. On 15th May r l achieved a depth nf 2I.bm. jusl SOcm short of the mnximurn. I n hubsequent years the ~ n a x i r n ~ level l ~ n was always achieved. For 53 days in 1893 and for [ 67 in 1894 the water level fluct~lnted between 21.6m arid 22. l m. For the first hair n T t l ~ e1890s il must have seemed that rill wils well. The dam retained a hip11 to full water level without mishap. Tl~codolite n>easur~n~cnts"' along the d a ~ n indicated a mnxilnum deflection of ljmnl: leakage was o f lhe order of 70 lhcc.. signilicant hut nut enceplional, especially by Bouzey standards; the piezonietric levels were about lorn above foundation level. These various observations were not, as the years went by, maintained. The deflection measurement& were given up ill 1893 and replaced, totally ~rnsrttisfactorily, by visually sighting pegs. Thc piezorneter proved itself very easily damaged by ice so that after the Eavage wlnter of 1894-5 it was removed. This same fearsome winter is supposed to have signilic;~ntlyopened vertical fissures in the dam xs the wall contracted, a crack width of as much as 7m1n bcing lrcorded.

- - -

Th F~tilure of the Bouny V~tm in 1895

11 had bad f o u n d a l ~ o n s . ~ ', Moreover, agreed conc l u s ~ o n swere paramount ' . because, ns pointed out at the b e g ~ n n ~ n o g f thrs account. B u u z e y w a s '7 s u p p o s e d t o have b e e n d r s 1 , q ~ e d And ~f dam des~gn was to be trusted, it was v ~ t n lto comprehend how and fur wl~atreasnnq

Nmman A.F. Smirh

Fig IS: Ttle hilure of IRY5, r ~ c w c llrurn l opsrreom. The verricil scile i s 5 x the l~orizontal; the rrcpped l ~ n c shows. appmx~mntdy.[hc bul~oln of the cut-off wall

T o what extent accurite and frequent measurement m i g h ~ have predicted the final collapse wiIl never be known. S e v e ~ , a l il~rtividualsclaimed that towards rhe end a pronounced downstream curve o f the dam developed which could be detected by the naked eye. IF such coilld strike passe~,s-byit surely would have registered officinlly had it been measul-ed :iccurately. But it was nut and the Bouzcy diun failed.
The Failure

There were three witnesses to the collapse on the 27111 April 1895. Early that n ~ o l n i n g a workman callcd Thiriat was crnssing the dam and he stopped, 160m short uT Ihc spillway at the eastern enrl, the right bonk, to talk to a mas011 working below. As they talked the dam began lo break up. ZvI. Thiriat tlarl to j u ~ n povel- a l0cm crack io his dash lu safety. The mason he had bee11 chatting with w;ls the first victim of the rlisa~ler.Thc third eyewitness was the innkeeper M. Gihiri who was watering his horse in the village 01- Bouzcy, 300m downstream. Hc was able to provide a ~tsefula c c o ~ i n lof the failure sequence which compriscd two phases: an initial bl'enk near the cresl ir! thc middle vl" [he structure over a length or ?Om,followed by the main ct>llapseof 180m of the darn to a depth of ahout 10m. Both Thiriat and Gihin cnmrncntcd 011 the nvise. The failcrl dam is shown diagrnm~ni~ticiilly in Fig. 15. Kenctinns ro the disastcr came at several levels. On behalf or ari outraged public the pop ill:^ press clamuured for explanation, redress and the riamcs of those responsible. Louis Geislcr was quickly into print with his "15 V r ~ e sPt~otogioplliquesde In Vallte dc I'Aviere ap&s la Catastrophe de Bouzey", a photographic collec~ior!in the style of the cvcn better one covering the ruins of the Dale Dyke dam put togcther by the Manchestcr photographer, Jarncs Mudd. There was huge international coverage i r ~ thc engineering preqs in Prancc, Gcrmar~y,Grcnt Britain and the United States.?' Of course there was an official enquiry. conducted by French enpineers, 2nd tours of inspection by other experls two 01- whom were Prolessor W .C. Unwin of Imperial College and Dr.G.F. Deacon, designer of the Vyrnwy dam in Wales. The lattcrs' observations are among the most useful and objective ccrtninly in Er~glish, that the historian can consult." Not the l e a s l interesting feature aT the deliberations which took placc is the degree or disagreement as to what exactly had happenerl and how to explain it. It certairlly was nor enough to observe rhat thc dcsign uT the d i m was had and thnt

14: it,: ,\ C~~>V-IIII % I ~ W III~ b~mk I ~ I L ~ I ,I I~ IIIIUIO~~~IIIII 1,) The failed don1 itself is Prolchs~r~n W.L. U n w ~ n ) the placc to begin a briel account of what happened. Naturally there was a readiness to msockitc thc final collapse with the a c c ~ d e ~ of~ t 1884 ant1 the repairs of 1888-9. In reality no such connection could usefullq be made. T h e great nbittment hat1 not yielded and appemed to havc prevented a11 fu~ther movement oi the bnse or the dam. And in m y case the h i l u r e o i I R95 did uccur cniirelq. aho~*c the level of the remedrrtl works of six years before. I11 add~tion, the extremcs of the F<~il~tre of 1895 did not colncide with the vertical cracks at the lirnjts ot the movement of 1884 while the cenlral group of fisscu.es in the earlier acc~deni were conta~nedcompletely by thc final collapse. As Unwln pi11i t , "The fractrne of 1895 seemed to avoid, rather than to d c p n d on, the fi~sures u T 1884."24 T h e cross-section of the failed d a m i s instructive (Figs 15-1 7 ) . The break runs horirontally through about half the section and then drops nwny. Unwit~was struck by the fact that there was no evidence of failure by crushrng, that "The mnqs war torn uut along a nearly level plane of fracture ..... without any obvious cracking or splintering of rhe stones along the down-stream edge of t h e f r a c t ~ r r e .T l l c m a s s filling t h e gap nppe'ared r a t h e r to h a v e s l i d o r over1~rned.l'~' One s c h o o l of though1 w a s q u i t e convinced of ii shear failure. O f the three expcns appoirlted to the tribunal of enquiry, two advanced this view and they were s u p p o r t e d by t h e e x p e r t w i t n e s s o f t h e eminent M. Naurice Ltvy." In a sense they were probably correct. The mechanism or a d a m ' s failure is a very c o m p l e x nffair, n ' w h o l e s e r i e s of e f f e c t s occur in q u i c k s u c c e s s i o n , and the o n e which starts the chain is no1 necessarily the one associated w i t h t h e Final m o m e n t s o f c o l l a p s e . A t Pig 17: Tht 1ypic.l hope :~nd puritlon n l the hlled sectior! Bouzey the final stngc very likely was a shear

The Failure of

the

Bou~eyDam

it1

1895
failure on a surface sloping down towards the air-face. But the key question remeins: what B y initiated apply~ the ng t Failure h e analytical in the first techniques place'?

N m m A. F. Smith
i l i t h it lasted so long." The explanation has a on e number of elernenis Although the b a s ~ ctheory treals n Jam ah J. series of thin vert~cal"slices", in practice it is not; the slices interact one beside the and so a crack has to other i n resisting deforn~ation extend t o a considerable depth beforc all t h e r e s i s t a n c e i s finally o v e r c o m e . The p o i n t i s emphasisecl by the great length which gave wny at Bol~zey.AIso, because Bouzey was n "leaker", its face wns [rt~rntime to time treated to a waterproof coating. Presumably. Lherefare, the percolating water was inhibiled at intervals and sufficiently to prevent uplift pressures cqual to the full head of water. And then again the reservoir was not always FulI. A canal rvhich was busy in t h e s u m m e r p r e s u ~ n n b l ya l l o w e d tlie datn s o m e s e a s o n a l
respite.
Fig 1 % System of lorcer; i v ~ t h uplift

described earlier, the ones that were available l o the engineers of (he day, it i s easily shown bow completely the middle\ rhird rule was violated in the upper part of ! . + 5 . 3 nrr the dam. and that rhe greaiesi deviation was at the level of the Fracture. ActuaLly t h e compressive slresses induced at the air-face 5.9 were not destructive. The maxirnunl vertical stress was about 5 kdcm', t l ~ c cor~esponding principal slress parallel t the ahr-face being ! of the order 6 kglcm'. Such curnpiesslons were well w ~ t h ~ the n strength of the nlitsonry used and relalively low compared w ~ t h many other dams. By conlrast the situation at the water-face. rescwoir full, at a depth of loin 8.+ was tatal. Here the venical tensrle stress was about 1.4 kdcin', a rnerre hrgher ~t was still 1.1 kglc~n?.'' It seems very probable that Ernckjnp hot11 the masonry and the mortar could have Fig lH: S,,ccerr,ul r;,lra of strrsPtin kElcm') taken thesc stresses ever) thougl~ gootl design ;~cmh\ thc crilical secrion should not have required then1 Lo, particularly since neither was of first class quality. The cr~lciaidefect was lhat thc bond between the masonry and mortar was very poor and should not have k e n relied upon in any casc.'"t was quite incapable of resisting the tension which developed with a rull reservoir and so the Bouzey dnm cracketl. Rcfen-ing to the stress diagram I In Fig 18 we can show that if the tension o f 1.4 kgcm' was relieved by cracking, the crack would penetrate 2m and the next stress diagr;iin, 11. would then opply, the minimum stress nuw becoming zero and tllc new milximum being a cornpresaian of 5.9 kdcm'. Su far the darn is still not ovcr-stressetl, enactly as de Sitzilly. Delocre and their foIIowers were willing to allow. Huwever. ns soon as water pressure penetri~tedthe tensiun crack a new vcrticnl force cnmc into play, the tlplift force we were considering in the case o f Ihe Habril dam, cri~cketlirl tlmt case by n chronic hydraulic overload. The uplill U within the cracked Buuzcy dam significantly increased thc overturning moment and the magnitude ut the effect is readily calculr~ted (Fig 19). A1 n depth of 10 m 01- rvntcr we can now construct a [ h i d strcss diagram, 111 (Fig 18); it 3.3m length of thc section and, of corlrse, l t includes ;I new element applies tc the su~,viving o l tension - 3.8 kg/cm? at miximum - which, if it is retieved by cracking. allows an even l a g e r uplift Force and leaves a decreased length of section intact. It is hai-(irlly necessary to ernphasise the progressive nature of the filiturc now urtderwny, except to point out again that at the final stage, overturning about thc downstrea~nedge would rnost likely give way, if that is the phrase, to shear fracture. Such a Fracture would be along a surFace that Falls i~wily tuwards the air-hce, as indeed thc ruins of rhe d a n ~ confirmed (Fig 17). A similar surface of failul-e harl bee11observed in the Habra, dam. Such then is [tie explanation as to why the Bouzey dam failcd. And yet, givc11 that its profile w a so ~ poor that tension did occur, one wonders not so much why it failed but how
C

--!----,
1 ,

The Puzxle
The queslion has not been answered. Why did the Bouzey dam have such inadequate shapt. and one that allowed so much tension'? ATler dl, the basic n~echnnicsol' clam design are slraightlbrwar~land by 1876, when the project was begun, all the esseiitial elcments of the theory were in place including the middle-third rule. T o my knnwledge tlie designer of the Rouzey dam is never named. That lie could produce a profile which sinlply looks so appallingly inadequate is interesting in itself."' Earlier I noticerl that it wns c & \ y to overlook the vulnerability of the glnvity dam near its crest when l o n hut ~ the reservoir is fill1 or, as Hnbra was, overfull. Conceivably Bolrzey's proportions, . . not very high, we]-e ton readily regnrrled as ordinary. elenlentnry to design and posing no significant risk in practice. The reason why French engineers pursued the dcsign and construction of masonry dam s o tlssiduously was because, to quote M.1, Auguste G r ~ e f f apropos the 50-m Furens dam, ''We need not in t h i ~ case consider eorthen dams which nre of very doubtful qecurrly at il heiglrt of 20 metres; at 50 metres they would, or course, be quite out of the rluestioi~."" Always OF concern was this question o r height. But Bouzey w m nnt a high durn and therehre if, fol- that very reason, (he darll's designer was already complacent and ir, in addition, he was prepared, like dc SaziIly and Delucre, to allow tensile stresses and then discount them. eilher because he believer1 the n~aterial co~tldwithstand them or becausc it ,was assumed permissible to settle for a crncked srctiun, as Fig 18, then l3ouzey's design is explained. There is ii telling paragraph in tanglois rclating to the "calculs des inginieurs de 1876". He says "No observntiuns have come to light on the subject of upstrcam tension, u T which there is no consideration, ntither in the calculations of 1876 nor in those which p~rceded the reconstl.uction of IXSX-9."" There is OIIC o t l ~ e rreature o f Bouzey's ill-judged design to consider. T h e clam was proporticnerl, apparently, by the method of M.Boitvier. His procedure was much in vogue ot the end o r the nineteenth century and in a French ministei-iai circular a f 1897 wah even qpectfied as rnnndntory. But In wality Bouvier's method was ncvcr more than a device and its use nr Ro~rzeymay explain a good deal. Consider Flg 20 Boilvler, In a11 effort to allow that maximun~ stresres are not v e n ~ c a ones, l conhiclers the rest~ltantof P and W, the force R,

The Failure of the I3ouzey Drtm it1 189.5

as acting on the plune ZY rather than XY. T h e stress distribution on %Y, at right angles to R, is assumed linear. ss usu;~I. but the calculated stresses art. higher than those obtained by considering W acling on X Y . It is a device which is pttlusible but w i t h n ~ r rationale. ~ But w h n ~ is crucial is that whereas Bouvicr's inethorl gives an augrnentcd c o m p r c s ~ i v estress value for the air-k~cepoint Y, as intended, it gives no strcss value Tur the water-face ;it all, the point Z being in effect nowhere. As Langlois compl;~incd of Bouvier's neth hod , "it does not dekrrni~ieanything about what happens on the air-face."34 In shor~ a designer relying o n Uouvicr's nretliod will almost cerlninly design nn adequate air-Pace at the same time >is he ignores the water-face completely. 'She method is intrinsically a lechnique for dctcrmiiiing n peak stress at a point, not a stress distribution belween two points.

Norman A .F. Smith

, Inattcr of uplift and the difficulty of comprehending its There was. for e x a ~ n p l e the bchaviotrr in various circumstances. S o m e attention had already been given to uplift pressures in the design of such d a m s a s Vyrnwy, in Wales. and AIEeld, in A l ~ a c e . ' ~ Avoiding uplift altogether was oric possibility and in some quarters, fullowing a much misundel-stor>dFrenctr rninislerial circular of 1897," this Led tu dams h e ~ n g profiled so that the conlpressive stress at tlie water-face was grcater than the water pressure, the belief being ~ll:~ the t infiltration of cracks then became impossible. One result was w m e very broad profiles indeed. In fact the 1897 instructions were inrelrded to allow for uplift by redrlcing the effective densily 0 1 - Ihe masonry. They also drew atlention In the need to determine maxinium compressive stresses o n oblique joints, in effect maximum principill stresses, and to investigate the mngnitude and danger of shear stresses, snrnething about which the Bo~rzey failure had heen instrumental in promoting discussion. as we have seen. Thus, and this i?;the point. Souzey was central to the developing interest in the stress analysis of dams beyond the elemenlary concepis o i de Sazilly and Rankine. In France, Maurice LCvy was prorninen~ in pursing these increasingly sophisticated c o n c e p t s w h i l e in E n g l a n d an e q u i ~ l l y mathematical approach succeeded in triggering a firs! class controversy in which tI~cmain protagunisla were ProCes.soru Karl Pearson anrl W.C.Unwin." And then, in attempt to throw light on these intractable problems, r.ecolrr.sc was rr~adelo ~riudelsutilising. in the early experiments, s u c l ~ rather unsuitahle rnalerinls a s plasticine. gelatine and India-ruhher. Nevertheless, Tv1- all the crudeness of the techniques, it was the start of a very significant. ant1 at tinies a crucial, trend in twentieth century darn de~ign.'~' The failure of tlie Rnuzey dam, when all is said clnd done, stands at a cross-roads where the lirst phase of rittior~aldnnr design endccl, firat o f all in controversy but then in clarification. Most branches of civil engineering have rheit land-marks: For dam-building the Bouzey failitre was tru[y n key one?' Let m e quickly tinish Ilie story of the dam itself. It was rebuilt twice more. The first recot~structionwas ;I ~n:~kesliift a f h i r of 1901-2. According to Eellet'" it was only 6.4111 in 11eipl1t Sorrning a reservtrir or 1.5 million m' userul to the Canal de I 'Est no doubt, but a ;relit reduction in cilpncity all the snlne. Thc plrsent dam dates from 1939 and is of rock- fill construction, 27rn high and 5 0 4 m long. It o c c u p i e s exactly the site o f its ill-fated predecessor ul' which the only surviving part is the spillway.

Fig Ill: Bwvrcr'c nlc~hud

The Aftertnath
Uouzey failed at a crucial time and strange though it may seem to say so, it was not iiltogtt>cr a bad thing: much good came out of this catastropt~e.'~ For nearly I~nlf-a-ccntury there h d ttveloped a growing confidence, in dam dcsign but considerable variations in how to d o it, what assuniptions could and shotrld bc made, and what criteria lo apply. Arid this rcmember was in a pericd when dams were h i t l ; built in increasing ~ulmhers world-wide, for public waler-supply, irrigation and rhe newly devclopi~lgapp1ic:~hon to I~ydro-electric power (the Liouz,ey darn's use l'or a canal was actually rilthcr old fashioned). Moi.eovcr, dmnb~rilding was by no iileans the monopoly of European counlnes. kluch, and orten the mosl signilici~nt.was happening in Nol-tli Africa, India. 1t1e United Stares and Attstralia. That is For the thin arch where one finds Ihe very big dams, e.g. Aswan, and lhe emerging E:~sl~iun dams, e.p. in California and New South Wales. The need to Ful[y understand the events at Bourey was the~rforeof zre;~li m p r t a n u e and the nudielice awaiting enlight~i~ent was liirgc and tnternational. Close scrutiny of the failure disposed of il number of erroneocrs ur ulr~.eliableprupositions anrl and its rull cliscussio~~ e s ~ a b l i s h c dpositively that for safe d a m design ( a ) the no-tension c o n d i t i o ~ jwas a n imperative, (b) thc middle-third rule was a correct concept, essential to obherve, (c) a r n a x i ~ n t ~compressive m stress must nut bc exceedetl ant1 (d) n dam must nut be able to slide. Not Far behililt came some instr~rctivelessons to do with construction techniclues and the quality of malerinls. The Bouzey T~ilure not only summed up the essenlial dcvzlop~nents st) Ihr. It also helped rt, confirin the dirttc~ion of f u t ~ ~work. re New design problems and attitudes 10 them, alrcady it is true under consideratio11 anyway, were given [rest1 illumination and a rnuctj sharper focus.

U o r ~ ~ c . s l ~ o r:~Dr. tli~ N.A.F. ~~~~ Smith, c ~ 2 Pike Close, Marlow, Bucks SL7 2AX

References
1. Tecllnical details are in G.M.Binnie, Lar(v Vicroriun Water Eti:rrairreeta ( I98 I ) , pp.263277. A popular account is G. Amey. The C o l l q ~ s ofthe e Duke Dykr Durn 18h4 (1974). 2. A recent accrrunl is W. Frank, "The Cause of the Johnstown Flood", Civil Engineering, 58 (May I9X8), pp.63-66. 3. T w o general hislories of dams are: N.A.F.Smith, A Hisral:~of Dnms (1971) and N. J. Schnittcr, A I-lrslor..~ of Don1.s (Rotterdam. 1994). 4. B. F, dc Relidor, Arr.hitrc.rril,e H y d r a ~ r l i q uVol. ~ , 1, Part 2 (Paris 17501, pp.77-83. 5. In, for exnmplc , S. Leliavsky. I ~ - / - i ~ n / rrrtd i o n Hydr-u~flic U P . T ~ Vol. R ~ , 3 (London 1960),

pp.375-78. 6. 'These and other early French canal darns arc cavered in E. Wegmann, The Dcsi,qn nnd

The Failure of the Buuzey Dam in 1895

Colrsrrur.tion uf Dums ( N e w York 1907) and H. Bellet, Barrages en Muqonnel-ie er Mrrr:s h Rhervairs (Paris 1907). J.A.T.de Saziily, "Sur un type de pruRl d'egale resistance pruposC pour les murs de reservoirs d'eau", A11nu1e.s der Ponts cr Chnlrsskes, 2 (1853)- pp. 19 1-222. E. Delocrc, "Sur lii f o m e du prolil h adoptcr pour les grand.? barrages cn maiionnerie des ltservoirs", Annoles des I'ot~ts P I Chausstes. 2 r sein. ( 1566)- pp.212-272. W I. M.Knnkine, "Report on the Design nnd Conslruction of Masonry Darns", The Et~~~inr 33 r r(Jnnuliry. . 1872)- pp. 1-2. J.A.T.de Sazilly "Sur un lype de prolil d'ignle resistance proposC pour Ics murs de r6servoirs J'eau", p. 193. W.J.M. Rankine, "Design and Constructio~r of Masonry Darns", p. I . These developrncnts are reviewed by E. Wegman, Dc,si,qn und Constt.ur.fion nj'Darns Chaps. 11-IV and H. Bellc~, Bul-t.u,~es eft Maqonneric, Chnps.V-V1. A. Graeff- "Sur la Forme el le mode de construction du barrage du gouffre d'Enfer, sir Ic Furens, et des grmds barrages en geniral", Antroles r/r.s Punts e l Chiru.rsies, 2e sem. ( 1 Xhbj, pp.207-2 1 1. "La Rupture du Barraze de I'Habl-a. pr&s de Fcrri~nux".Lr G611ieCivil, XC11 (1928), No. 1 1 , pp.256-9; No. 12, pp.283-5: the author - denoted by "A.C." - may be And1.6 Coyne. See also E. Wegrnann. D e . ~ i ~ ullrl l f Consrrwrtion uf D U ~ pp.94-8 S and 1-1. Bellct, Buvrcr,yr,s rt1 Mul-ul~n~v.i~., pp.54-6. The tnost conlplcte general account of the Eouzey dam is Leon La~~glois, R~tptltreLIU B U I - ~ Id~ I <Rorr~cy ~P (Pnris 1898). These and other d e t n ~ [ 01s the Bouzcy dam's career are tnkcn from, in particular. L.l_nngluis and the report of a special cn~nrnission of the Corps des Ponts ct Challssies which was pilblishcd in 1895. Them is a copy irl t l ~ clibrary of the Instilutinn of Civil Engineers. It wils i~hslracterlby A.W.Binnie. "Faiture o f the Bouzey Dam". Proc. Inst. C ~ I .E . ~ I xC . , X X V (1895-6)- pp.461466. Tlws s:une report appears to he the basis of "t7nil~rl-e o f Ihe Great Dam nt Bouzcy". 3'11~ En<yirlrr~; 17 Jnnuluy IS96, pp.50- 1. The pint is noticed in "Historic Accjdents and Disaslers: No.XVII1. The Bouzey Dain", T ~ EP ~r~ince 4~Dccember -, 1941, p.45?. Langlois. Rrrprrrt.e drr R < I I - I - rle U ~B ~n.l c e y , pp.93-4 deals with c;~nalconsurr~ptionand there is LI lengthy co~npar:~tive sludy OF c;inals' water requircmcnts in Maurice Lkvy. O h s r l ~ ~ ~ u r is~rtu t ~ lu s C r r f r r ~ ~ - o /dr ?Ii~ R(JII:P,V. (!his n.rl. circa 18'36). T/1r E t ~ < y i r l4 ~~ Uece~nber ~-, 1942. p.453 notes [his oversight but does no1 rellect on the practicality of rebuilding. The dam bad alter all only been dihplaced. not destroyed. kI.Beller. RUI-1-osrsr11 M o r - u r ~ ~ ~ pp.37-9. r ~ ~ i r , The G ~ , o s b o i s d a n ~i \ the subject of M.Guenot. "La StabilirC rles Digues du Reservoir el du Contre-Riscrvoir de Grosbois ali~nentanl te C:lnal dc Bourgugne". Ann:iles des Ponts et Ch;~ussCeu..4 (19491, pp.4679 1. Su lar as I can determine (he Bouzey dilln ih the earliest to be instrumented in this way. The bibliography or the B o u ~ e yfablure is enormous. Idr too 1nuc11to record here. Serge Lelinvsky once refelred to the rnatly &I-titles written as being -'comparable to il sea o l ink": see his d~scussion to E.Gruner. "Dam Disasters", P ~ a rItrsr. . Civ. E I I , ~ 27 . . (1964)p.362. For :In ii~troductiontu the bihl~ng~,npl~y of the hilure there is ICOLD. L~ssorrs F ) . U Inrrn~ ~ I Irrtirlc~trs(Paris, 1973). p.29 ant1 A.O. Babb and T.W.R/lermel, C'rrtu/u,qof' Durtr Disnsrrls. Fuil1r1-os rrj~rl /tr~chidc,rr/.s, (bt'ashillgton D.C., 19681, pp.3 1-2. W.C.Unwin, "Thc Failulv of the Masol~i-y Dam at Bouzey, Frilnce" Co.s.rirr'sMtr,qu:irrc, XI ( 1896-7), pp.9- 19. G.F.Deacon , "Wi~ter Supply", Thlr E ~ r q ~ ~ I o p!~?d I i. o ;!(IIII [ 3th I~(.~~. Edition, Vol.28, p.403. PI-ofessorW.C.Unwirl frequently raked the Bouzey dniir issue at

Nmman A. F. Smith
meetings of the Civils. See, for examplc, his discussion to G.F.Deacon, 'The Vyrnwy Works Cur the Water-Supply of Liverpool", Proc I t ~ s t .Civ. Enfi., CXXVI (1895-6). pp.93-7 ant1 then to Sir John Ottley and A.W.Brightmore, "Expefimental Investigalions of the Stresscs in Masonry Dnms subjected to Water-Pressure", Prm. Inst. Civ. Eng., CLXXII (1907-8). pp.70-78. Unwin is worth quoting on this: "It had k e n said that the design of the dam was bad, though il was no1 wid in whal respect; and also lhat i t had a bad foundation. Thal was something like the mayor of town in France who had seventeen reason5 (or nut firing a royal salute, and Ihe first was that there was no powder". W.C.Unwin discklssion to Deacon. p.93. W.C.Unwin, Cassicr's Mrr:a;i~rr,p. 18. W.C.Unwin, C U J J ~ PMaguzine, I-'s pp. 18. H.Bellet. Ucrrrcrges err Mn~unt~erie, p.68. and Lkvy, Ohserrurior~s, pp. 10-23. And i~ can bc calculated that a zone of tension covered a hcigl~tof &out 6 mts above nnd berow lhe facture. Analyses of the accident nre replete with discussions about the strength of the rnateri:ils wed: see tanglois, Rupture du Barrage de bouxey, and Livy, Observntions, pp.7-22. The author of "Historic Accidents and Disasters", Drgirreer, 4 December 1942, p.455 makes much of the usc of lime rather than cernerit rrlurtar Phil~p A.Mnrley Parker puts i t in an interesting way when he says: "Neverlheless, these facts create a feeling not so much of astonishment at the dismter, as of confidence in a well, designed dam. We hnve a dam viulattng in every possible manner all the presentday principtes of sound construction. The foundation is bad, and the dam is not carried of down Tar enough; yet it only parlially Fdils." - Philip A.Morley Parker. The Cut~trul Wurrr, (19 13). p.393. Although one must be wary or ljinds~ghthere. In 1876, or tl~ereahuuts,the basis for appreciating a well proportioned dtlni was limited to only a decade or so's experience. A.Graeff. "Sur la Folme ct Ic mode de construction .......". p.196. L.Langlvis. R ~ r ( ) t i i ldl< . ~ BIII-I.Q<~C,P de Bouzey,. p.57. M.Uouvicr, "Calculs de rCsistance des grnnds barrages err rnaionncric", Altnrrles dcs Putlt~ef Chuusskrs. 2 ( 1875), pp. 173-205. Rouvier's rnethotl i s assessed by H.Rellet. Barrages en Maqonnerie, pp.46-9. L.Langlois. R L I ~ ~ Idu I I ~- (( I, I - I - u ile ~ ~ Bulazey, p.56. S.Lelinvsky, discussion to E.Gruner, "Dam Disasters", pp.362-3. Fur Vyrnwy see I.Dav~dson."George Deacon (1543- 1909) and the Vyrnwy Wurks", 7'rrr)lsat~1iurts oj'rhr Ncwr.umen Sociery, 59 ( I 987-88). pp.85-6. For Alfeld see S.Lelinvsky, Lllrl$t in Gruvit?. Durns, ( 19581, pp. 1-2. "CircuL~ire du Ministtre de I'Agriculture du 15 Juin 1897" is reproduced in H.Rcllel, Barrages en Maconnerie, pp.293-302. A.J.S.Pippard, "The Functions of Engineering Research in the University", Juur-. Inst. Ci],. EII,~ 33. . , (Feb.1950)- pp.265-284. And interestingty enough, the Bouzey dam itself was ireated 10 an experimental investigation long after the events described; see V. de Belaevsky, "Note sur la rupture du barrage du Bouzey", Le G{trie Civil. LXXXVl (March 1925). pp.239-40. This i~ the theme of "Lessons Learnt froni Daln Disasters", En~irleering,IS May 1964, p.683. H.Bellel, Burrrr,qes err M ( ~ ~ . u r n ~p.70. rie,

7.
-

8. 9.
10.

23.

I 1..
12. 13.

24.

25.
26.

27.
28.

14.

29.

15.
Eb.

30.
31. 32. 33.

17.

18. 19.

34. 35. 36.

37.

20. 2 t.

38. 39.

40.
41.

22.