Ground Rupture in the Baldwin Hills

23 April 1971, Volume 172, Number 3981 SCIESNCE

On the day after the failure, it was apparent that major offset had occurred along what was to become known as the "Reservoir fault," the west side of the fault having moved relatively downward with respect to the east side (Fig. Ground Ruptuire 1). The offset was sufficient to crack the lining all the way across the floor of the and surface in the Baldwin Hi] is reservoir, displacement of as much as 6 inches (15 centi- meters) was created. The crack line was Injection of fluids into the ground for oil recovrey punctuated by several ragged, cavelike and waste disposal triggers surface faultiing. sinkholes, which marked the points of water entry that had led, possibly over a period of hours or days, to the deteri- Douglas H. Hamilton and Richard L. Mee han oration and eventual destruction of the reservoir foundation.

Responsibility for the disaster has on August 10, 2008 never been formally fixed. Insurance On the Saturday afternoon of 14 pears as a slight bucklinig of road pave- carriers for the Los Angeles City De- December 1963, water burst through ment on the far side ccf the reservoir partment of Water and Power, con- the foundation and earth dam of the basin and thence beconnes a faint, dis- structor, owner, and operator of the Baldwin Hills Reservoir, a hilltop water continuous break in the ground surface, reservoir, promptly paid for flood dam- storage facility located in metropolitan which trails off south of the reservoir ages. A report on the results of an in- Los Angeles. The contents of the reser- into the brush-covered and excavation- vestigation carried out by the State of voir, some 250 million gallons of treat- scarred terrain of the Inglewood oil California's Department of Water Re- ed water that had filled the artificial, field. sources (2) provided a taut narrative www.sciencemag.org 20-acre clay- and asphalt-lined basin to Since 1963 geologist: s and engineers of events on the day of the disaster and a depth of 70 feet, emptied within hours have been intensively ii nvestigating this a cool appraisal of failure as the re- onto the communities below the Bald- crack and several similLar ones nearby, sult of an unfortunate combination of win Hills, inundated a square mile of all known to be surfac4 e expressions of physical factors: "Sitting on the flank residences with mud and debris, and deep, near-vertical faultts of Pleistocene of the sensitive Newport-Inglewood damaged or destroyed 277 homes (1). or greater age. For thi ere is no doubt fault system with its associated tectonic

Fortunately for those in the path of that the disaster occurri ed as a result of restlessness, at the rim of a rapidly de- Downloaded from the flood wave, indications of immi- displacement along faulIts in the uncon- pressing subsidence basin, on a founda- nent failure had been observed bv a solidated sediments th,at underlie the tion adversely influenced by water, this reservoir caretaker several hours be- reservoir. These displaicements led to reservoir was called upon to do more fore the final breach occurred; even so, rupture of the protecttive clay lining than it was able to do." police evacuation teams had barely suf- that covers the floor )f the bowllike Two lawsuits filed in 1966 by the ficient time to clear the area. Conse- structure, which had b4een constructed city and its insurers against the oil quences of the disaster were minimal in 1951. Ironically, the I10-foot-thick (3 companies active in the Inglewood oil compared with what would have oc- meter) lining and its underdrain had field at the time of dam failure charged curred if no warning had been pro- been especially designed and constructed that the oil field operations had led to vided, but they included five lives lost, to isolate the soft, saindy foundation the events directly associated with $12 million in property damage, and rock from leakage of Fthe reservoir breaching of the dam. These suits were loss of the reservoir itself. water, thus providing w]hat was thought settled out of court for nearly $3.9 The remains of the Baldwin Hills to be a margin of safety against piping, million dollars, thus disposing of the Reservoir stand empty today, the north- a process characterized by gradual de- immediate financial issues that arose ern rim of the bowllike structure hav- velopment of eroded s;ubsurface cavi- from the ground rupturing beneath the ing been gashed from crest to founda- ties and channels. WhaLt the designers reservoir. Origin of the ruptures is a tion by the escaping water (see cover). evidently had not thouglht possible were question that remains unadjudicated. A linear crack issuing from the base offsets along one or moire of the buried Mr. Hamilton is a geologist and Mr. Meehan of this gap can be traced across the faults great enough to 4destroy the lin- is a civil engineer; both are partners in the con- sulting firm of Earth Sciences Associates, Palo asphalt floor of the reservoir. It reap- ing. Alto, California. 23 APRIL 1971 333 Geologic Setting production and water flooding. The bar- rier effect of many subsidiary faults in The Baldwin Hills, site of the Ingle- the east block may be local, however, wood oil field and the failed reservoir, and may not extend the entire length form part of an interrupted chain of of any given fault. low hills that rise in striking contrast to Although oil production has been the surrounding flat terrain of the Los primarily through solution gas drive, a Angeles basin (Fig. 2). Processes of both peripheral water drive exists along the folding and faulting have contributed northeast margin of the east block and to the uplift of this chain (3, 4). In the also along the west margin of the west Baldwin Hills, the primary anticlinal block. The edgewater condition of the fold structure has been much modified east block encroaches from the area of by faulting, especially by lateral and artesian formation water in the north- dip-slip displacement along the Ingle- erly extension of the permeable oil sand wood fault, which bisects the hills. horizons, beyond the area of petroleum Many unnamed subsidiary faults, ap- entrapment where the groundwater for- parently related genetically to the Ingle- merly existed at or near hydrostatic wood fault, are present throughout the pressure. hills and are especially numerous in the vicinity of the reservoir. Moody and Hill (5), among others, Exploitation of the Oil Field consider the Inglewood anticline to have developed as a drag fold during defor- The Inglewood oil field was discov- mation of the thick, plastic sedimen- Fig. 1. Baldwin Hills Reservoir. Seepage ered in 1924. It was explored and de- tary section overlying the Newport-In- through the ruptured reservoir lining (fore- veloped so rapidly that its period of glewood zone of basement right-lateral ground) undermined and then breached greatest yield occurred within the first transcurrent faulting. The anticlinal the Baldwin Hills dam embankment 3 years of production. Although the folding began and continued while ma- (background). The relatively downward fluid in the oil field displacement of the ground surface to the pressure reservoir rine sedimentation was still on. on August 10, 2008 going left of the crack is clearly visible in the was at hydrostatic levels under preex- as evidenced by thinning and lateral dis- photograph. [Wide World Photo] ploitation conditions (570 pounds per continuities in the stratigraphy of the square inch in the east block), fluid overlying section of late Tertiary and pressures measured in wells declined Quaternary strata. As movement con- horizontal displacement on the Reser- through the years to about 50 pounds tinued along the Newport-Inglewood voir fault was inferred on the basis of per square inch in the early 1950's (2, zone, the deformation exceeded the ca- observed striations on the fault surface. 7, 8). Castle and Yerkes (4) have sug- pacity of the overlying strata to adjust However, the significant vertical offset gested, however, that blocks of higher by folding, and failure was propagated component that is evident along these pressure may have persisted after the through the section as normal and faults clearly places them in' the cate- general pressure declined, owing to fault www.sciencemag.org strike-slip faulting. The surface strain gory of normal (gravity) faults. compartmentalization of the reservoir. pattern associated with this tectonic de- In any case, production and develop- formation (6) is illustrated in Fig. 3, ment, mainly by "infill" drilling be- and the subsurface structure is shown Inglewood Oil Field tween wells, has continued steadily to in Fig. 4. the present. The near-parallel, north-striking faults The Inglewood oil field occupies an A significant modification of the ex- that splay outward from the Inglewood irregularly oval area that extends diag- traction program was started in 1954, Downloaded from fault south of the Baldwin Hills Reser- onally across the trend of the hills along when the Standard Oil Company ini- voir (Fig. 3) were probably formed as the axis of the faulted Inglewood anti- tiated a pilot "water-flood" program of an array of tear faults developed in re- cline (Fig. 3). The field is localized by secondary recovery in its east block sponse to strike-slip displacement along the anticlinal structure and by the dis- leases. The history and technical de- the dominant Inglewood fault. The tribution of its sedimentary strata. The velopment of this program were well Reservoir fault is a member of this extensive breaking associated with the described by Oefelein and Walker in family of steeply dipping faults, all of Inglewood fault has disrupted the orig- 1963 (9). They also presented detailed which intersect the Inglewood fault inal fold structure, so that it now has information on the structure and stratig- at depths of no more than a few thou- the form of a series of slices translated raphy of the main producing horizon, sand feet. successively downward toward an axial known as the Vickers zone, of the In- The apparent sense of displacement trough along the west side of the Ingle- glewood field east block. The pilot water on these secondary faults is largely ver- wood fault (Fig. 4). The Inglewood flood involved the injection of brine in tical. At the time of construction of the fault forms a structural break between selected intervals from two arrays of Baldwin Hills Reservoir, vertical strati- the east and west blocks of the field and four wells each, generally at pressure graphic separation of as much as 26 effectively isolates the two producing gradients in the range of 0.5 to 0.9 feet was noted at surface excavations blocks hydraulically from each other. pound per square inch per foot of across what became known as the Res- Some subsidiary faults within both depth. ervoir fault; lesser apparent vertical dis- blocks are also fluid barriers, as was Results of the pilot program were placement was recorded for several re- confirmed by changes in reservoir pres- sufficiently encouraging to prompt a lated faults beneath the reservoir. Some sure with time and in response to both full-scale program in 1957. The east 334 SCIENCE, VOL. 172 block secondary recovery program has injected fluid seem to h,lve increased aIs Subsidence and Ground Rupture been expanded by increments since its greater numbers of injection wells, op- in the Baldwin Hills inception, with a general pattern of in- erating under increasing pressure, have creasing numbers of injector wells and been brought into service. Such prob- Although the phenomenon of "earth increasing volume and pressure of in- lems frequentlv have occurred after crack" ground rupturing may not have jection (Fig. 5). Although the program significant raising of injection pressure been definitively observed in the Bald- has been successful, it has involved in specific wells, and some have been win Hills prior to 1957, the existence many technical difficulties, including severe enough to cause abandonment of a more widespread and larger-scale maintenance problems in the wells (7). of the injection well. However, tech- form of ground surface movement was l'roblem.s such as loss of major amounts niques of sealing damaged intervals and recognized by engineers of the Los of injected Iluid in narrow intervals, repairing wells have often been effec- Angeles Department of Water and Pow- sudden increases in formation "take" tive in restoring wells to injection after er as early as 1943, when comparison with concomitant loss of injection pres- each episode of failure. and the overall of leveling surveys indicated that eleva- sure, casing failures, "breakthroughs" injection programn has been progressively tion changes were taking place in the to producillg wells, and surface leaks of expanded. area (10). The ground subsidence was on August 10, 2008 www.sciencemag.org Downloaded from

Sca/e o 5 /0 Ki/ometers Fig. 2. Goneralized location map, Los Angeles basin. 23 APRII 1971 335 of concern to the department because Fig. 6. Table 1 summarizes the se- operation of the Inglewood oil field; of its plans to construct and operate a quence of ground-rupturing events. and (ii) deformation largely or wholly reservoir, but it was not known by the The principal features of contem- of tectonic origin. community at large. Not until 1955 porary ground subsidence and surface Tectonic theory. This theory is based were sufficient data available to estab- rupture are clearly established. They on records of recent surface rupturing lish that the changes in elevation de- have been thoroughly documented by along known "active" faults and on in- fined a bowl-shaped area of subsidence the U.S. Geological Survey (4), the Los dications that the sense of the con- (the outline of which appeared to coin- Angeles Department of Water and temporary movement appears to be cide with the outline of the Inglewood Power (10, 11), the State of California indistinguishable from that of prior dis- oil field) (10). In 1957, 6 years after Department of Water Resources (2), placements. It is argued that vertical the reservoir had been put into opera- and by several consultants (12, 13) and offset along the family of faults in the tion, the beginning of surface cracking other investigators. Some interpreta- east block has occurred in recent geo- and faulting in the vicinity of the inter- tions have differed slightly from others logic time as a result of lateral dis- section of Stocker and La Brea streets, as to the exact pattern, magnitude, and placement along the Inglewood fault or southeast of the reservoir, attracted the history of this deformation. Explana- as part of the process of anticlinal fold- attention of public officials. tions for the origin of the deformation ing, or both, and that it may be rea- The spatial relations among patterns have, in contrast, been sharply segre- sonably assumed that the tectonic en- of ground deformation in the Baldwin gated into two categories: (i) response vironment that created these structures, Hills, preexisting geologic structural fea- to lowering of fluid pressure, fluid with- relatively recently and rapidly, persists tures, and the area of production in the drawal and injection, and related phe- today. The tectonic theory is offered by Inglewood oil field are illustrated in nomena and activities associated with some geologists as an explanation of on August 10, 2008 www.sciencemag.org Downloaded from

Fig. 3. Geologic map of the Baldwin Hills. 336 SCIENCE, VOL. 172 the subsidence in the Baldwin Hills as the Inglewood fault or nearby major anticline show the ground to be moving well as the cause of the ground rup- faults in the Baldwin Hills or adjacent toward the anticlinal axis (and radially ture in the east block area. area. toward the center of the subsidence The tectonic theory is open to ob- Continued upward warping of the bowl) (Fig. 6). Moreover, offset of the jections of several kinds. First, and Baldwin Hills might well lead to the folded formations by the Inglewood most important, displacement along sub- observed normal faulting, but detailed fault demonstrates that the major epi- sidiary faults generally occurs in re- analysis of survey leveling data by sodes of folding must have predated sponse to a relatively greater displace- Hayes, Castle, Leps, and others indi- much of the faulting. This suggests that ment along the principal fault, which in cates that little or no such regional faulting has superseded anticlinal fold- the Baldwin Hills is clearly the Ingle- uplift has occurred during the past ing as the dominant style of tectonic wood fault. Yet relatively large offsets half-century for which data are avail- deformation in the Baldwin Hills. have occurred on the subsidiary faults able. Further, the margins of a near- Finally, the shape of the subsidence in the east block, whereas no evidence surface, upwarping, anticlinal fold bowl does not correspond to the shape has been produced to indicate related should move outward, away from the of the "central graben," as would be creep, rupture, or crustal strain that fold axis, but surveys of four bench expected if the subsidence were caused would give rise to displacement along marks situated on the flanks of the by further downfaulting.

West Block Central Grob en East Block A' Reservoir fault Fault f I Activoted 1952-63,1966 Activated1/968

Inglewood Fm. f (Plelstcone) I; / PA % i A I f -t I -1 I x 4 n /-..Se LewlI I I I -r 1f Xr1 I on August 10, 2008

Pico Formation (Pliocene)

,j.' 0'.0 I I a, % 1- X- I AL------L--" I I'L,AU 1000 feet .-.~7 ,.09EL,.I.41:..s...... ;.,- ...l. IL~- .. ;....-.Z www.sciencemag.org , .. .. . -- f.-..-- *......

/ P ! tVcke( . ..;...... To. .. ; s ...... s, .' aCS Zone aXl 805ds ...... X ...... /",'''''''.-''.',,' ''' 0-' "*0', ...... ''''''''-''''-'''-'';'- ;' X .''..' ::' ', ':; ','. .'. '...... /, .. j X a4~~~~~~~~~~~~~~~~~ezzi V .E .-,,,, ..,', *. _:- .-'' Downloaded from B1 # ...i.|...... :; 't, ' *'. t; SC236 .. ,, .. ... *: ./. ... ;..... i...... I .f_ ; , .L 2000 f.:t ...... ,.. ,ec.s j Note - Geologic section is from California DWR ...... Bulletin ...... 0., 'Investigation of Failure, Boldwin Hl'ils Reservoir"(bef No.2). Some Eost Block fault projections reinterpreted by D.H.Hamilton,1970 EXPLANVATORY AWEY ...... Injection ...... intervals from Ref No. 7 ...... ;s . 0 ; . - a3000 ef Location of Section A-A' snown on Figure 3

...... Exposeid interval af .. _ ... . a ...... Dr well Scale

0 We/I No. 5 00 fOOOi'eet 0 300 meters (Htorlzontal vertical) Fig. 4. Geologic structure section through the Baldwin Hills-Inglewood area and the Inglewood oil field. 23 APRIL 1971 337 Table 1. Sequence of ground rupturing along earth cracks in the Baldwin Hills, 1957-68. rim of the subsidence area, but almost For location of cracks, see Fig. 6. all the rupturing has occurred in one Crack Occurred Cumu- limited sector of the bowl circumfer- No. fiort Length lative ence (Fig. 6). observed length namename ~~~~~~(date) (feet) (et 1 May 1957 2,600 2,600 Influence of Secondary 3, 4 Jan. 1958 1,900 4,500 Recovery Operations 2 Mar. 1958 600 5,100 5a Before 1961 400 5,500 Recognition of these two serious ob- S Feb.-Mar. 1963 400 5,900 jections, and further recognition that la 1962-1963 (?) 800 6,700 the procedure of exploitation in the 8 Feb. 1963 400 7,100 Inglewood field changed from simple Wilson's fault I* Aug.-Dec. 1963 2,000 9,100 Wilson's fault V Aug.-Dec. 1963 800 9,900 extraction to extraction combined with Wilson's fault I 1968 Recurrent movement injection, has led to modification of the (2,400) subsidence theory. As here presented, it Wilson's fault V 1968 Recurrent movement eliminates the objections noted above, Hamilton's fault f 1968 1,400 11,300 is supported by striking correlation of * Wilson's fault I is now identified as the Reservoir fault. water-flood events and episodes of fault movement leading to ground rupture, and provides a satisfactory mechanical Subsidence theory. Advocates of the faulting suddenly begin to occur as late basis for the observed history of ground subsidence theory argue that the 6 feet as 1957? rupture. The revised theory may be re- or more of ground subsidence in the The other possible objection to the ferred to as the "subsidence fluid injec- Inglewood field is clearly related in subsidence explanation is the local dis- tion" theory. In substance, it specifies space and time to the net withdrawal tribution of ground rupturing. The that injection of fluid into the ground of some 67,000 acre-feet (14) of oil, downward drag and stretching presum- under pressures only slightly greater water, and sand from shallow produc- ably would be distributed around the than normal hydrostatic pressure will

ing horizons, a contention supported by reduce and may, in areas subject to on August 10, 2008 theoretical mechanics and by analogous normal faulting, eliminate shearing re- experience in other areas (4). Almost ' Hydrostatic pressure sistance along potential failure planes. all the observed surface rupturing has When this activity is carried out in occurred on the edge of the subsidence ground earlier affected by faulting, and 6 tX December 1957 bowl; stretching on the rim of the sag- where additional stresses have been set 4 je ging ground surface is an obvious and 2 A lj E - up by differential subsidence, reduction mechanically proved consequence of in shear strength will give rise to move- . . . . i subsidence [demonstrated by Grant u. 0.4. 0.5 0.6 0.7 0.8 0.9 1.0 1./ ments that will be concentrated along a (15), Deere (16), Lee and Strauss (17), preferred surfaces of weakness such as www.sciencemag.org 6 and others], and tension cracking of the December /959 preexisting faults. According to this the- near-surface materials is one result (il- 4 ory, fault activation and consequent lustrated diagrammatically in Fig. 7). 2 a 0 l i ground rupturing should develop pref- 0X Subsidence also exerts a downward 0 erentially in an area being affected by .0 drag on the sediments that rim the sub- 4 0.5 0.6 0.70.80a 1.o /./ fluid injection. This prediction is veri- (U 8 sidence bowl, and, where faults or other 0) fied by the localization in space (Fig. 8) 0) 6 established surfaces of weakness are and time (Fig. 9) of earth-crack ground Downloaded from December 196/ present, accumulating elastic strain may u'.. 4 rupturing in the Baldwin Hills in and

be relieved by upward "popping" of the 2 Iw adjacent to the area of fluid injection ground on the edge of the bowl. -0 for secondary recovery, which began Two objections might be leveled at 0.4 05 t 9/.0 /./ almost simultaneously with the initia- the subsidence theory as an explanation a tion of the "full-scale" water-flood pro- 0 of ground rupture in the east block. a)- gram. The correlation of fluid injection .0 First, ground rupturing was not directly events and surface rupturing is even observed prior to 1957, and the only z- more striking when it is examined in positive evidence of drag or tension- detail. induced failure before 1957 is a record One of the first of the full-scale in- of continuous extension of the rupture jector wells was placed in operation in caused by movement since 1952 of the May 1957. In that same month, a fault, Reservoir fault under the Baldwin Hills the probable subsurface projection of Reservoir inspection gallery. However, which lies within or very near the in- most of the ground subsidence between jection interval of this well, was acti- the early 1920's and 1963 had already vated and became the first recognized occurred by 1957, and the rate of sub- earth crack in the east block area. In- sidence in the east block had slowed jection was started in 21 additional markedly in the late 1950's and early Fig. 5. Injection history for Vickers east wells in that area between 1957 and 1960's. Why then should intense surface block, 1957-63. 1963 (Figs. 5 and 8), during which 338 SCIENCE, VOL. 172 time at least eight more faults were movement, upon which are superim- ential subsidence with little or no effect activated. The relation of injection in- posed individual jumps that reflect of pressure injection. Then, in late tervals to the east block subsurface events of larger fault movement. These 1957, a jump in the crack extension structure is illustrated by the geologic stages correspond to episodes of surface was followed by a doubling in the rate cross section in Fig. 4. This section rebound or uplift, identified by Leps of extension and also by the first epi- passes through two of the east block (13) from records of comparative level- sode of surface rebound. This level of faults, each of which had been related ing of bench marks at and south of the activity then continued for 31/2 years, to surface rupturing by 1969. reservoir. until 1961. At that time, three more The history of contemporary move- When the history of movement along injector wells were activated in the area. ment along the Reservoir fault (beneath the Reservoir fault is compared with the This increase in injection was followed the east side of the Baldwin Hills Res- operational history of injector wells lo- several weeks later by another fault ervoir basin) is known from monthly cated in the vicinity of the fault's jump and a further increase-a tripling readings taken on strain gauges extend- southerly subsurface projection, the evi- -in the rate of crack extension. Mid- ing across a crack in the concrete in- dent influence of pressure fluid injection and late 1963 was a period marked by spection gallery beneath the reservoir on fault activity can be recognized. The intense activity in injection operations. and across the fault. The opening of earliest stage of crack extension began Injection was started (or restarted) in the crack, which reflected displacement prior to the initiation of the water-flood four additional wells close to the reser- along the underlying fault, began in program and continued until just after voir, and severe and unusual operation- 1952 and was monitored until the time the third of the first three injectors al problems were occurring in those of reservoir failure in 1963. The history along the fault was brought into service. wells (now nine) near the fault. Uncon- of crack activity (Fig. 10) shows three This stage may therefore represent trolled loss of fluid occurred in five in- stages of progressively accelerating ground movement attributable to differ- jectors. One well appears to have been on August 10, 2008 www.sciencemag.org Downloaded from

Fig. 6. Contemporary surface deformation, Baldwin Hilis-Inglewood area. 23 APRIL 1971 339 Tnso therefore distinctly related to the ear- Tension~ ~ ~ ~ lier episodes of surface rebound in the Tensionpresio area. In considering the possible cause- and-effect relations among injector activity, subsurface problems, and fault Subsidence movements, it seems significant that all recorded episodes of fault movement Fig. 7. Diagrammatic representation of fault activation attributed to subsidence. since 1957 have occurred after one or more of the following: initiation of injection in nearby wells, increases of pinched or sheared off at depth. In ent of 0.73 pound per square inch per injection pressure, or problems such as May, 6 months before the reservoir foot at a depth of 1200 feet, the top dropping of fluid pressure concomitant failure, brine was observed seeping of the injection interval. Late in 1963, with increases of fluid take, loss of fluid from cracks in the ground surface an episode (or episodes) of movement in narrow zones, and so on. The se- south of the reservoir and on the trace occurred along the Reservoir fault. quence of events suggests that the in- of the Reservoir fault. Wellhead injec- which culminated in failure of the res- jection caused or contributed to the tion pressures during 1963 were erratic, ervoir. This fault movement, though movement. but pressures in excess of 400 pounds apparently normal in sense of displace- As is evident from Fig. 7, most of per square inch were common; 400 ment, actually involved upward move- the ground cracking occurred in and pounds per square inch creates a gradi- ment of the footwall block and was beyond the outer tnargin of the injec-

Waste brine injector wells on August 10, 2008

North \G \ www.sciencemag.org

Ge\x5~ \ Area of horizontal R fault

sfce compression servoir fult

0 Downloaded from

0 O Q\r°'

r Scalefo Stocker- SclorVolumeolueonjeceof9 lEtdr) WeLaaterBrea area \ / I o ~~~~~~~grouind ruptures 100,000 Barrels 00 0 4,000,000 Barrels

K)4,000,000 Barrels

(IBarrel = 159 Liters)\.\

Scale

0 0I- - - 2000W feete 0 500meters

Fig. 8. Zones of ground rupture and cumulative amounts of injected water, Baldwin Hills-Inglewood oil field, through 1963. 340 SCIENCE, VOL. 172 tion area. The explanation for this con- Fluid Injection Causes Faulting tests. No laboratory testing of the Ingle- dition lies in the three-dimensional wood oil sands has been performed, to geometry of the faults, and !also in the The mechanical behavior of systems our knowledge, but an a posteriori the- relation of the area of surface rupture that consist of solid particles with oretical analysis of subsidence in the to the area of ground surface extension fluid-filled voids has long been investi- Inglewood oil field has been made by and compression induced by subsidence. gated experimentally by workers in the Castle and Yerkes (4), who used rea- Both projection of fault plane dips mea- field of soil mechanics, who have de- sonable compressibility data for the oil sured at the surface and reference to rived mechanical principles that have sands. Their analysis indicates that in- the cross section constructed from sub- been of great practical use in the design creased effective stresses accompanying surface data (Fig. 4) indicate that most of civil engineering works. Changes in fluid withdrawal and decreasing fluid of the east block faults dip to the west. volume and in strength of soil and rock pressure will lead to subsidence of the Consequently, fault planes in the im- materials with fluid-filled voids have same order of magnitude as that shown mediate vicinity of injection intervals been shown to be fundamentally re- in Fig. 6. reach the ground surface, 800 to 1500 lated to change in stress within the Subsurface fluid injection has been feet above, some hundreds of feet to solid skeleton. An increase in stress in practiced less widely than fluid with- the east. Faulting initiated at depth and the solid skeleton produces a decrease drawal, and theoretical and empirical propagated to the ground surface would in its volume, owing to elastic deforma- understanding of the mechanical effects therefore result in surface rupturing tion of the skeleton and, in the case of of high-pressure fluid injection has lag- hundreds of feet to the east of the sub- uncemented assemblages, to closer pack- ged accordingly. Such injection into the surface point of origin. Also, reference ing of particles. For liquid-saturated subsurface has many practical applica- to the map of surface deformation (Fig. systems, stress within the solid skeleton tions, among the most important of 6) shows that the surface rupture has is generally expressed in terms of ef- which is water (and steam) flooding of occurred in the zone of ground surface fective stress if (18), given by the ex- oil-producing zones, with the object of extension but that it dies out as the pression enhancing production by maintaining faults enter the zone of surface com- hydraulic pressure and oil to- of=a- u(1) flushing pression. The surface compression evi- ward producing wells. This process is dently "clamps" the fault surfaces where a is the total stress acting across known as secondary recovery. together, whereas the extension is asso- some plane within the system and u Secondary recovery is now widely on August 10, 2008 ciated with a pulling apart. is the fluid pressure in the region of the practiced by the oil industry, and an The only active ground rupturing plane. This relation is more complex advanced technology has been devel- known to us outside the east block area for a solid-liquid-gas system, but it oped to service these operations for in the Baldwin Hills is a north-south holds approximately if the liquid and production enhancement. The economic zone of cracking at the north margin gas pressures are approximately equal. feasibility of production in many older of the Hills (Fig. 8). This cracking re- It follows from the effective stress prin- oil fields now depends on this tech- portedly began to develop about 10 ciple that a decrease in fluid pressure nology. years ago, well before the time of any produces a like increase in effective The mechanical effects of high-pres- secondary recovery operations in that stress and therefore a decrease in vol- sure fluid injection on the volume of www.sciencemag.org vicinity. However, the rupturing is very ume of the solid skeleton. sediments may be described by the prin- near two fluid injection waste disposal The principle of effective stress pro- ciple of effective stress in a fashion cor- wells, which probably have affected the vides a reliable theoretical basis for the responding to that for withdrawal. It is strength of the subsurface in the same prediction of settlements caused by apparent from Eq. 1 that an increase fashion as the pressure fluid injection changes in either total stresses or fluid in fluid pressure, induced perhaps by for secondary recovery of oil. pressures, if the compressibility of the pumping fluid into the ground at depth,

Although water flooding was under solid phase is known from laboratory would be accompanied by an equal de- Downloaded from way in 1970 throughout most of the crease of the quantity described as the Inglewood oil field, no ground ruptur- effective stress. This would result, in ing other than that just noted has been accordance with experimental results, recognized in the field's central area, cbegin exploitelion in an expansion of the soil element sub- or around its north, west, and south /00 jected to the increase in fluid pressure. margins; it still continues, however, in Pressure injection over a wide area the east block area. The absence of might then be expected to result in a 50Kscale fault activation within other parts of floodfull inj cti\water rise of the overlying ground surface, or, the field can be explained by the oo o I /wnijection Injection in the case of formations that had pre- "clamping" effect of surface compres- viously been compressed through reduc- sion in the ground there. mar- The west sequence not obseeved w-/ tion of fluid pressure, restoration of gin of the field is probably in surface former higher pressures could result in extension; unlike the faults in the east rebound of the compressed sediments. block, however, most of those in the 5,000 Results of laboratory tests suggest that west block die out upward at points the amount of rebound may be roughly beneath the ground 0 surface. Conse- E /920. 1930 /940 /950 1960 /970 equal to or, under the less usual but quently, even if fault activation has Q ~~~~~~~Year predictable situation of rebound from occurred at depth, it evidently has not Fig. 9. Time relationship of ground rup- virgin compression, substantially less propagated upward through the un- turing and east block injection, Baldwin than the compression accompanying a faulted shallower ground. Hills and Inglewood oil field. like decrease in fluid pressure. 23 APRIL 1971 341 /95/ /952 /953 /954 /955 /956 /957 -13958 I/ 1961

-.45s;x _ Active injection -425 En E3l Cn Intermittent injection (Detal/s not known) -400

-.375

-.350 A .325 Crack eXtension

300

.275

.250

~ 225

.200

-.175 21 EpIp sode X I r Episode rEpIsodelMI _ 0 Stage of gradual movement no jumps Stage of accelerated movement Final pre- faillure sta0e- B t.125 Pre - in/ectlon stage Early low volume and pressure High volume High volume ond

pressure, on August 10, 2008 irregular behavior .1 0 0 2 6 start _ , f rorn 237 -.075 .075 ' E EniDE-0 5t" 2/65t""Startr r ~~~~~~~~~~~~~~~~~~ ~~13 ~~~~Start -5 Start /M~ 5 .050 Start as Start 224124 Start =1 E n _214 ,.025 Start E - 422 .000 I I I I :1 www.sciencemag.org

a.00 ..Sg 1 :: t Downloaded from

faceuplift east of.---.,,,,.,Reservoir fault (ide d by. M. Leps)- andstagesofcrackextensionactivity.(C)Stagesofinjl......

acivity0. (D)tenimesof opseratonrofal idcratedi injpectior wellseloae nearditHellReservoirfaul.sE Volumedi ofm flui inecteode in wells located near the Reservoir fault.

342 SCIENCE, VOL. 172 Ground heave attributable to injec- a fluid stress component, equal to the tion-induced volumetric expansion of A rr assumed hydrostatic fluid pressure of substrata apparently has not generally 690 pounds per square inch at a Cr depth given rise to secondary problems, per- |-Cm0.6 of 1600 feet, and an effective stress haps because injection programs usually component that, under the assumed involve relatively small quantities or, state of limiting equilibrium, is about not infrequently, are carried out . in , one-third of the vertical effective stress, I _~~~~~~~~. 113 areas simultaneously or previously af- 0r. _-& T3 (r, or 303 pounds per square inch. fected by subsidence and perhaps lack- Reduction of fluid pressure increases ing structures such as reservoirs that both major and minor principal stresses, could be seriously damaged. However, produces compression of the element, another side effect of high-pressure in- B and, by reduction of the ratio of prin- jection is the weakening of substrata cipal stresses, "backs away" from the that inevitably accompanies artificially 1-En Failure envelope for JO. + critical state of shearing equilibrium. induced increases in fluid pressure. This c) sandy sediments / Increase of fluid pressure has an op- will occur in both cemented and un- Tn posite, effect. If the fluid (a Mohr circle destabilizing cemented rock materials; it can be more at failure pressure is raised beyond the natural easily visualized in the case of unce- co hydrostatic condition, the principal mented sediments such as those in the stress ratio will tend to increase, a con- Baldwin Hills. dition not compatible with the pre- The shear strength of unconsolidated Normal stress, a viously defined state of limiting equi- sediments is mainly frictional and can Fig. 11. Typical failure conditions in un- librium; shearing will then occur and be related to the quantity previously de- cemented sediments: (A) as determined will continue until some readjustment scribed as the effective stress. This can by laboratory tests; (B) Mohr failure cri- of stresses restores equilibrium. be shown experimentally by confining terion. If normal faulting had occurred in a sand specimen in a box (or flexible the distant geologic past but a readjust- membrane) and then causing the speci- ment of stresses had since provided a on August 10, 2008 men to fail by shearing. The shear inch. The hydrostatic fluid pressure margin of safety against reactivation of stress on the failure plane at failure gradient is 0.43 pound per square inch existing normal faults, and if subsidence foot of or 690 will be, for a typical sandy sediment, per depth, pounds per were not contributing to disequilibrium, about 0.6 times the normal stress on square inch at a depth of 1600 feet. then the initial stress condition in the From Eq. 1, effective vertical stress is the failure plane (Fig. 1lA). This rela- east block area might be somewhat dif- tion holds in a simple solid-fluid sys- then 1600 less 690, or 910 pounds per ferent from the postulated condition of square inch. Horizontal stress includes tem, if the effective stress is taken as limiting equilibrium. However, unless a the normal stress. It can also be dem- recent and major change in orientation onstrated theoretically and experimen- or magnitude of principal stresses in the www.sciencemag.org tally that the relation between major vicinity of the east block is postulated, and minor principal stresses (effective it is reasonable to assume that the ratio stresses) at failure for the same ma- of horizontal to vertical effective terial will be approximately as shown stresses across the recently active, nor- in Fig. llB. mal fault planes would be no greater Subsurface stresses in relatively un- than about 0.5, a typical ratio obtaining consolidated sediments under condi- in "normally consolidated" sediments Downloaded from tions of normal (gravity) faulting are under tectonically quiescent conditions. approximately as shown on Fig. 1 lB. In this case the initial horizontal effec- The major principal stress is vertical; tive stress would be 455 pounds per expressed in terms of effective stress, it square inch for the example in Fig. 12, is equal to the pressure exerted by the and a fluid pressure increase from 690 weight of the overburden less any fluid to 915 pounds per square inch (gradi- pressure. The faulting is by definition a 690 ent increase from 0.43 to 0.57) would shear failure of the material, from be required to reestablish the limiting which it may be deduced that the hori- principal stress ratio of one-third; zontal effective stress is at or near one- hence, faulting would be activated in third the vertical effective stress. the affected such a Fluid ground by pressure Figure 12 illustrates the fluid pres- pressure increase. It would thus appear that only :] (pounds per square sure on a hypothetical system at a inch) a small increase in fluid pressure (a depth of 1600 feet-the depth of in- gradient of 0.6 or less) should be suffi- tensive injection activity in the east cient to at least local shear Effective pressure trigger block-and initially in a state of incipi- failure in the east block, even if the ent (pounds per square normal faulting. A geostatic pres- inch) material was not stressed to the point sure 1 gradient of pound per square Fig. 12. Hypothetical natural stress condi- of failure by subsidence or active nor- inch per foot is assumed; thus, total tions in element at a depth of 1600 feet. mal faulting. vertical stress is 1600 pounds per square Nornal faulting is imminent. The relation of fluid pressure to for- 23 APRIL 1971 343 mation fracturing has been treated by entries of fluid at the tops of injection quences of injection of fluid into the Hubbert and Willis (19), who show that- intervals (leading in at least one inci- ground, a practice that is becoming in- hydraulic fracturing will occur in tec- dent to discharge of injection fluid at creasingly widespread not only in sec. tonically relaxed regions at injection the ground surface), all indicate that ondary oil recovery operations but as a pressure gradients of less than 1. In the injection pressures- were high enough means of industrial waste disposal and above example, the reduction of hori- to markedly reduce, or, in some areas, groundwater management. zontal effective stress (455 pounds per locally to eliminate the shear strength Experience in the Baldwin Hills sug- square inch) to zero should result in along preexisting fault surfaces and, gests that, although fluid injection op- fracturing along vertical or near-ver- also locally, to "fracture" the formation erations may be carried out for bene- tical planes and would occur at an in- hydraulically and to create new frac- ficial purposes, the effects of such jection pressure gradient of 0.62. As ture surfaces. The weakened zones injection on the geologic fabric can be demonstrated above, this hypothetical must have been confined initially to serious and far-reaching. fracture condition is statically inadmis- the immediate vicinity of the injection sible, given the shear failure criterion wells, but subsequently, as new injec- References and Notes adopted for the material, without read- tors were put into operation, relatively 1. Metric equivalents: 1 gallon = 3.785 liters; 1 acre = 0.404 hectare; 1 foot = 0.3 meter; 1 justment of stresses, which could arise larger volumes of material became af- square mile = 2.590 square kilometers. only through shear deformation. The fected. In the case of the Reservoir 2. "Investigation of Failure-Baldwin Hills Re- servoir" (California Department of Water Re- history of difficulties with injectors in fault, significant shear displacement sources, Sacramento, 1964). the east block and comparison of in- probably had occurred along much of 3. R. F. Yerkes, T. H. McCulloh, J. E. Schoell- hamer, J. G. Vedder, "Geology of the Los jection gradients with fracture gradients the break by mid-1963, and, at the end Angeles Basin-An Introduction," U.S. Geol. in similar tectonic environments leave Surv. Prof. Pap. 420-A (1965). of that year, rupture propagating to 4. R. 0. Castle and R. F. Yerkes, "Recent Sur- little doubt that fracturing could (and the ground surface was of sufficient face Movements in the Baldwin Hills, Los Angeles County, California," U.S. Geol. Surv. did) occur in the east block at injection magnitude to shear the clay lining of Open File Rep. (1969). gradients well below 1.0. Fracturing the Baldwin Hills Reservoir and cause 5. J. D. Moody and M. J. Hill, Geol. Soc. Amer. Bull. 67, No. 9, 1207 (1956). would and doubtless did result in ex- its failure. 6. R. 0. Castle, "Geologic Map of the Baldwin tension of the effects of a single injector Hills Area, California," U.S. Geol. Surv. Open File Rep. (1960). to points hundreds or thousands of feet 7. Unpublished 1969 operating data of the away from the injection interval, thus Conclusions Inglewood oil field.

8. Metric unit: 1 pound per square inch = 0.0703 on August 10, 2008 exposing large volumes of sediment to kilogram per square centimeter. the elevated injection pressures soon, or 9. F. H. Oefelein and J. W. Walker, J. Petrol. That the earth-crack ground ruptur- Technol. 1964, 509 (May 1964). perhaps immediately, *after the raising ing of the Baldwin Hills was genetically 10. From unpublished data of S. A. Hayes (1943 of pressure at the wellheads. Moreover, and 1955) [cited in (2)]. related to high-pressure injection of 11. From unpublished data of F. J. Walley (1963) fracturing must have been accompanied fluid into the previously faulted and [cited in (2)]. 12. T. M. Leps and F. J. Walley, unpublished by shear deformation, unless initial hor- subsidence-stressed subsurface seems data. izontal and vertical stresses in the injec- established beyond reasonable doubt. 13. T. M. Leps, unpublished reports and per- sonal communications. tion zones were equal, which was cer- The fault activation appears to be a 14. Metric unit: 1 acre-foot = 1233.5 cubic not meters. tainly the case in the recently near-surface manifestation of stress- www.sciencemag.org 15. U. S. Grant, IV, "Subsidence of the Wil- faulted sediments comprising the east relief faulting triggered by fluid injec- mington Oil Field, California," Calif. Div. block. tion, a mechanism identified as being Mines Geol. Bull. 170 (1954), vol. 1, chap. 10, P. 19. In the Baldwin Hills east block area, responsible for the 1962-65 Denver 16. D. U. Deere, Pa. State Univ. Miner. Ind. injection pressure gradients of as much earthquakes and for generation of small Exp. Sta. Bull. 76, 59 (1961). 17. K L. Lee and M. E. Strauss, paper presented as 0.9 were employed, and relatively earthquakes at the Rangely oil field in at the International Symposium on Land an western Colorado (20). Subsidence, Unesco, Tokyo, 1969. large bodies of ground, including 18. K. Terzaghi, Theoretical Soil Mechanics extensive reach of the Reservoir fault, These examples of fault activation (Wiley, New York, 1943).

19. M. K. Hubbert and D. G. Willis, Trans. Downloaded from were subjected to pressure gradients through the response of stressed ground AIME 210, 153 (1957). greater than 0.7. The history of casing to artificially induced increases in sub- 20. J. H. Healy, W. W. Rubey, D. T. Griggs, C. B. Rayleigh, Science 161, 1301 (1968). failures in injection wells, sudden injec- surface fluid pressure demonstrate some 21. We thank T. M. Leps, R. H. Jahns, and tion pressure losses, uncontrolled major of the mechanically predictable conse- others for suggestions and criticism.

344 SCIENCE, VOL. 172