1. Empirical Rule of Thumb for incidence of RIS: "reservoir depth over one hundred meters and reservoir size larger than one cubic kilometer " Statistical validity not relevant for most site specific conditions.
2. Application of the Risk Prediction Method (RPM). Successful assessments for: Manicougan 3 & 5, LG2 & LG3, and Katse, Lesotho (1996). Location of RIS: - on existing discontinuities that could be reactivated Cohesive rock mass: reservoir load approximated by an infinite strip load area; stress concentration at the sides of the load. Most RIS events' location close to the shore of the impoundment and than less frequent bellow of the water storage.
2hp- Mitigation of the Global Hazard of Reservoir Induced-Seismicity.
Inclusion of RIS as part of the issues of Q 74, " Performance of Reservoirs," reflects the increased aspect of induced seismicity as an environmental concern of dam safety.
Along with other mitigation means available, mostly as passive mitigation procedures as risk identification, control and design of the Rate of Impoundment and preventive impoundment monitoring some active mitigation procedures were recommended by some participants to ISORIS'95. Probably the most efficient mean was sensed that for all new RI risk reservoirs, a training and review seminars to all professional associated to the risk sites. The seminar will provide update of RIS research elements and means to mitigate the induced seismicity hazard along with the efficient approaches of dealing with them environmental and public concern aspects. The efficiency of such approach emerged from a comprehensive half day seminar at the Three Gorges site organized as part of post symposium activities of ISORIS.
The definition of the RIS risk for the concerned sites is the old well-known rule of thumb on Identification of seismicity incidence: " if the dam is more than 100 meters high, or if it impounds more than one km3 of water, the risk is real and the possibilities of induced seismicity must be thoroughly investigated.". The site specific seminars must be funded as part of the project completion, and suggestions were made to be part of the environmental package associated to project financing.
Stresses concentration: Reference by the " rule of thumb "is of significance mostly from the statistical point and was confirmed in practice. Applied for the analysis of existing 300 VLD the occurrences of RIS is the 120 seismic reservoirs, close to the existing known incidence. Applied for the future estimated 130 new VLD reservoirs in construction and planning stage, the incidence risk of RIS may be evaluated with same ratios, which suggest about 35 projects where RIS risk would be of concern. This statistical probability of RIS incidence is also supported by the facts that the 130 new reservoirs planned for completion in the future will create higher local stress concentration on the earth crust. Higher stress concentrations are referred by the following average data, in which elements for existing 300 VLD are compared with the further 130 VLD:
| VLD | # of dams (avg)* | S CF** | Height | volume | area (meter) |
|---|---|---|---|---|---|
| Existing VLD | 293 | 116 | 24.7 | 956 | 25.8 |
| Future VLD | 129 | 147 | 11.8 | 365 | 32.6 |
Dealing with the increased risks of stress concentration on the crust which may be associated with increased hazards of Reservoir Induced Seismicity requires professional attention for implementation of environmental mitigation procedures for all new sites of new impoundments. Some most important tools are available as the Risk Prediction Model (RPM) endorsed at ICOLD 13th Congress, and the rest is steady application.
1. . T.W.Mermel, 1991, Lists of Very Large Dams, Published periodically in Journal of Water Power & Dam Construction, p.67-77, June 1991.
8. Approaches to the mitigation of Reservoir-Induced Seismicity hazards in Environmental Impact Assessment, Q60, R.40., ICOLD Proceedings 16th Congress San Francisco 1988, vol. 1, pp.637- 656, by Thomas Vladut,
1978. Initiation of the RIS World Survey: ICOLD Congress & JWP&DC. 1976. Review of French cases of RIS, EDF Paris, 1975. Conceptual Model of RIS: Ph.D. Thesis, U.of Bucharest, under Prof. R.Priscu. 1968. Public hearing and monitoring for Vidraru Arch Dam (h=165m) for increases of seismicity in a known seismic area.
by
Thomas Vladut Ph.D., P.Eng., Consultant
Reservoir Induced Seismicity (RIS) is a specific dam safety consideration and is associated with design, construction and operation of about 120 reservoirs. Reservoir impoundment can cause triggering of seismicity and this risk is analyzed as a reaction to reservoir filling and its implications on the dam safety concepts.
Any discussion on RIS could not avoid the environmental component of the challenging aspect of dam safety and consider to be one of the most widespread and controversial concerns outside of the engineering community, Facts on RIS are of necessity for rational explanation dealing with genuine environmental public concerns and other reservoir developmental concerns such as the lost land.
Historically, Reservoir Induced Seismicity (RIS) was considered an unintentional environmental phenomenon, which was noticed by people living in proximity to some reservoirs and was sensed without instruments. Some mines and petroleum production operations may produce similar induced seismicity. The first engineering project aborted on environmental grounds was the disposal of contaminated fluids in deep wells at the Rocky Mountain Arsenal in Denver, Colorado, USA (1966) as induced seismicity became a public concern.
The incidence of RIS parallels the development of dams. The rate of very large dams (VLD) development in the last sixty years, averages 4.3 very large dams per year, mean while RIS averages 1.7 cases per year and was associated to 120 reservoirs.
RIS concerns are traditionally about dam safety. The estimate of magnitude of induced seismicity is of significance for the final seismic design, particularly if it exceeds the site level of natural seismicity. From the physical stand point it is more important to estimate the location of the expected induced seismic event. RIS is generated at shallow depths and close to the dam structure. This may be more used for seismic design than natural events that may be much further away.
The International Symposium on Reservoir-Induced Seismicity (ISORIS'95) held in Beijing, China referred to 120 RIS cases [1] from 29 countries with 22 in China, 18 in USA and 12 in India.
Some examples of the upper boundary of RIS level on the Richter scale, are 6.3 at the Koyna dam, mining induced seismicity is 5.7, fluid disposal injection maximum level was 5.2 at the Denver Arsenal, 3.4 fluid withdrawal and geothermal processes three for enhanced oil recovery operations and recent Russian studies indicate seven for the Gazly natural gas fields.
RIS monitoring to date does not include peak ground acceleration, duration, or type of response spectra, frequency or dampening. The frequency of induced events is expressed by log N= A - bM, where N is the number of events of magnitude M or larger and A and b are constants. The most recognized value is the b constant for the site that is, usually less than one. Variation for regional site specific conditions are such as for Africa 0.47, Greece 0.5, and India 0.8, reflecting different continental stress conditions. Larger values are considered associated with induced activities (b>1) with extremes reaching to 2.0, as stress field may reflect higher stress concentrations.
The first observation of Reservoir Induced Seismicity was made by Evans in 1945 at the Hoover's dam reservoir. Evans related the seismic events to the variation of the water level of the reservoir. The impoundment was gradual and reached the 150-meter depth after three years and its maximum after six years. The seismicity increased in frequency, and occasionally in magnitude as the reservoir level increased. The major event occurred five years after filling the reservoir.
The induced-seismicity often appears as an energy release mechanism without further threat of increasing seismic risks.
RIS triggering and development can alter reservoir impoundment and operation, with each reservoir having its own individual pattern. Except for two well-known reservoirs, Koyna in India and Hoover Dam (USA) and to a certain degree the Marathon Lake in Greece, where RIS is a cyclic presence, most RIS reservoirs have the seismic incidence related to the impoundment sequence. An in depth analysis of the initial and steady state seismicity was the subject of a keynote presentation at ISORIS'95, by see reference]. History of RIS has two principal components: first, the initiation of RIS, and second the increased knowledge found through studies of RIS case histories. Summary of major events as shown below may be warranted to identify the major development milestones. These will help better understanding of the actual status of risks of induced seismicity.
Awareness and Understanding.
| Year | Delay in years | Reservoir Location (Dam) | Reservoir maximum depth (m) | volume km 3 | Magnitude [M] | Observations
| 1929
| 9
| Marathon
| 76
| 0.04
| 5.7
| n/a
| 1932
| 1
| Qued Fodda
| 101
| 0.22
| three
| n/a
| 1935
| 1
| Hoover
| 221
| 36.7
| 5.0
| Carder, D.S. 1945 relation of lake variation & seismicity
| 1959
| 3
| Hesinfenkiang
| 105
| 13.8
| 6
| n/a
| 1959
| 5
| Kariba
| 128
| 175
| six
| n/a
| 1962
| 3
| Koyna, India
| 103
| 4.75
| 6.3
| Dec.10/67 Damages to 230 km from the epicenter (200 lives, injuries over 1,500, 80% houses affected)
| 1962
| -
| Denver-3 well, Colorado
| 671
| 0.5
| 5.2
| Waste Fluid Injection, in disposal wells, Evans, 1965: correlation fluid disposal & seismicity
| 1965
| 3
| Kremasta
| 160
| 4.7
| 6.3
| n/a
| |
1968 - Rothes's "Fill A Lake, Start An Earthquake."
1967- 1970 - Randeley Oil Field, Colorado
First Controlled Earthquake Experiment, M=0.5-3.5
* This is not a full listing of known RIS, only some examples
Elements of reference: year of construction, delay of development after impoundment in years, H height of dam and V, reservoir volume in cubic Kilometer; M the magnitude on the Richter scale.
Year Reference Information Observation
1970 UNESCO Working Group on Seismicity of Large Reservoirs
1970 Gough & Gough Role of weight of reservoir on RIS
1973 Royal Society, London UK Colloquium Seismic Effects on Impounding
1975 Banff Canada International Symposium on Induced Seismicity
1976 USGS Open File report on RIS
1976 H.K. Gupta and B.K. Rastogi Dams and Earthquakes, Elsevier Reservoir-Induced Earthquakes, HKG 1990,
1979 ICOLD Q51, GT. Lane: World Survey (WS) on RIS Thirteen Congress New Delhi
1979 USGS Penrose Conference, on role of Pore Pressure Hubert & Morgenstern
1982 Application of World Survey for behavior differentiation Non RIS: Manicougan 5 (H=214m,V=35.7Km3, and RIS: Manicougan 3 (H=108m, V=0.35Km3, M=4.1)
1983 /86 Confirmation of behavior assessment La Grande 2 and La Grande 3 Quebec, Canada
1989 ICOLD, San Francisco RIS as an environmental issue, Q60, R40: Risk Prediction Model (RPM), result of 1979 World Survey. General Reporter; Q60: "Technology progress had been made related to the Induced seismicity including measurement and prediction techniques."
1994 NATO Workshop, Moscow Environmental and Ecological Problems: Earthquakes Induced by Underground Nuclear Explosions Springer-Verleg 1995
1995 ISORIS'95 November 1- 5, Beijing International Symposium on RIS. Proceedings available from IWHR: PO Box 360, Beijing 100044, China. Cost $ 70. or fax to 86-10-841-2316; November 1-5, Beijing, China, Proceedings Editor IWHR
* This is not a full reference citation list; for details see USCOLD, USA 1986: "Bibliography of Reservoir Induced Seismicity."
The principal risk element of seismicity is best identified by the old rule of thumb": if the dam is more than 100 meters high, or if it impounds more than one km3 of water, the risk is real and the possibilities of induced seismicity must be thoroughly investigated." The rule of thumb validity is of significance mostly from the statistical point and was confirmed in practice. Applied for the analysis of existing 300 VLD the occurrence of RIS is of eighty seismic reservoirs, close to the existing known incidence. Applied for the future estimated 130 new VLD reservoirs in construction and planning stage, the incidence risk of RIS may be evaluated with same ratios, which suggest about 35 projects where RIS risk would be a concern.
Another risk factor is the rock mass. Simplified expressions of geological risk relate the triggering mechanism to the existence of brittle bedrock that is subject to the fracture development.due to stored energy release. Genetic classifications of RIS suggest fracture development as specific and other causes as non specific as collapses of karst and soluble rocks and landslides, frost cracking. Geological risk assessment includes knowledge of existing discontinuities and states of stress of both in situ conditions and the modified condition due to the dam and reservoir. A simplified geomorphological approach based on the elastical parameter of the Poisson ratio presented in Vladut, 1991). RIS assessment for Qued Fodda impoundment in 1932 was made without instrumentation to provide seismic parameters. The elliptic approximation suggests that the reservoir be associated to one of narrowest valley similar to the other earliest RIS incidences as Bajna Basta, Vouglans, Camarilles, and Kastraki reservoir. The classification of reservoirs and lakes based on the elliptic approximation is provided (appendix 1) as background for environmental comparisons between reservoirs and natural lakes. The elliptic form provides comparison of conditions associated natural and manmade reservoirs, with different applications as the significance of lands submerged by reservoirs (table #A) included in appendix 1 to this presentation.
The potential to compare reservoirs with different sizes and shapes, provided the base for a world wide survey to compare and identify potential similarities and differences between RIS storage. The survey confirmed the general shallow nature of induced seismicity. For twenty reservoirs the average was about 2.2 Km, with an extreme low of 300 meters at Schlegais. The deepest RIS activity was 18 km. recorded at the Kremasta reservoir.
Incidence assessment for RIS is the result of observation, and was used as a determinant of different behavior for two reservoirs (Manic 3 and Manic 5) along the Manicougan River in Quebec, Canada. Observation of the seismic activity at LG2 (La Grande) reservoir confirmed the possibility of using the full stress path of the filling process as a means to regulate the rate of the impoundment. This would allow the design of a filling process with variable rates (ROI) as a potential way to mitigate the risk of induced seismicity.
Using this process for forecasting, an induced seismic behavior assessment for the next dam under construction, LG3 was made, with seismic behavior confirmed after five years of monitoring.
With the increased number of reservoirs, it is time to update the model from its initial definition that used only twenty known seismic reservoirs and most of the elements are available from the ISORIS proceedings. There are tens of reservoirs under development that fall in the RIS incidence domain but have impoundment water depth too small to be associated with induced-seismicity of significance. This implies that if potential risk is identified, a probabilistic approach using two parameters, water depth and volume, must be included to estimate the potential risk level. The difficulty is that a probabilistic approach is heavily dependent on the data base. Data for at least 80 reservoirs should be made available to today's dam designers, which is a significant improvement over previous statistical bases. Risk incidence analysis has been suggested for areas having similar geological conditions such as reservoirs in Brazil (State of Minas Gerais) and the karsts regions surrounding the Yangtze river, in proximity of major dams.
Monitoring is considered the practical and efficient engineering countermeasure against the risk of unknown induced seismicity and is the center concept for dam safety implications (CDSA, 1995). Identifications of seismic risks using the RIS incidence approach were also prepared for the Ertan dam in China, and the Katse reservoir, Lesotho. Validation of the assessment is not expected for several years. For the second reservoir some seismicity was noted during impounding in February 1996, about three months after filling.
Evaluation of the incidence of RIS often is associated to the empirical Rule of Thumb (reservoir depth above one hundred meters and volume over one kilometer cub of water). Statistical data on seismic reservoirs (using 66 case histories) show an average depth of 94.6 meters that is close to the treshold depth of one hundred meters depth. The indication on the average volume of seismic reservoirs is much larger (47.88 Kmc) well above the one cubic kilometer size of the impoundment. Only a third (34.8%) of seismic reservoirs have volumes larger than the one kilometer cub of water and depth equal or larger than the one hundred-meter treshold. Of significance are the facts that reservoirs with seismic events larger than five (ten storage) have both elements over the both indicators (for 90% of cases).
The general indications of the empirical rule have important exceptions both on the depth of the reservoir as well the size. Reservoirs with very low water depth as Cajuru in Brazil (volume of 0.2 kmc, with a delay of 17 years after impounding), Piasta, Italy with only 13 million cubic meter reservoir (water depth 93 m), and the before mentioned Marathon in Greece (67 m depth and volume of 41 million cubic meters ), and the Quinjin reservoir in China (50 m depth and 20 million volumes). Some participants at ISORIS visited and hard to believe proof of a tiny impoundment behind a cylindrical arch dam of about 10-12 m high and reservoir volume of 1-2 million cubic meter that triggered two sequences of seismic events, probably associated to the carstic environment.
Specific local crustal stress conditions are associated with the so called eight asesismic reservoirs; average depth 144.5 meter (range between 72 and 226 meters) and size of reservoirs averaging 8.8 kmc. , (range between 0.11 and 30 Kmc). This aseismic group has both indicators well above the empirical estimates.
The statistical indication by the empirical indicator is pertinent to the frequency of incidence of induced events, but the site specific indication could be misleading without a more detailed assessment. This contradiction is better referred on analyzing the twenty five-largest reservoirs ( volumes between 31 Kmc.at Keban reservoir, to the largest addition of 270 Kmc. at the Owen Falls). Reservoirs with lower heads than the hundred meters (ten sites) did not triggered seismicity. Only one notable exception LG3 (depth 93 meter which developed delayed weak seismicity (M=3.7). From the 15 reservoirs with water depths larger than the hundred meters only six were affected by RIS, which is less than 40% relevance.
A proposition was made that the available indentification of risk of RIS provided by the RPM (the Risk Prediction Model) be presented systematically as a training seminar on regional format and local for dams of paramount importance ie: Three Gorges and similar. Along with more in depth training of professionals a more systematic review may be undertaken by a specialized body like the ICOLD Speciality Committees.
The problem of the rate of impoundment (ROI) is probably the second most important issue of risk identification to designers, owners and contractors. No direct relationships between the ROI and size of seismic activity could be detected.
ROI has a high empirical component, it must be analyzed by the proposed stress path method, which still has only one application. This may provide an approach to the structure of critical loading rates, with the potential for mitigating larger induced earthquakes. Often RIS is mentioned as a neglected issue of dam practice. Considered as a general understood problem, succeed or fail on site specifics and the issue of RIS became an accepted risk inherent to dam construction. Inherent or not, the monitoring provides a relative site specific cushion. The question of the due diligence to avoid or to leave with induced activities is mostly an academic question. As with many "academic questions" of the environmental complexities, risks are more easily accepted than try to mitigate. Still for the RIS engineering assessment, even successful assessment procedures are available, most often the easy way, let monitor it, and deal with it when it happens is a common practice. This approach may be even cost effective. The unknown of induced seismicity includes also several cases of decreases of the local seismic activity. Probably induced seismicity is a neglected opportunity to use a full scale leaving laboratory for better understanding the earth crust behavior. CONCLUSIONS RIS is a social concern associated with new impoundments. The technical community, has to provide facts on RIS and are required to stipulate explanations' for genuine public concerns. A summary of RIS related concerns and issues concerning the dam safety are structured as a summary, included in table 3. [prefeered position for table 3; but could be at editors preference and suggestion]
Most engineering standards do not address the issue of risk of RIS adequately to offset the questionable historic response that the occurrence as unintentional. A solution may be provided in a form of guidelines [9] and to issue a challenge to all professionals to alter practices such as to manage risks as part of other environmental consideration of dam developments.
1. Comparisons of reservoirs and natural lakes. Fifteen types of forms are distinguished using two elements: shape (ratio of perpendicular diameters): n= 1-10 round, n= 10-100 elliptic, n= >100 very long, and the relative depth of reservoir (ratio of average depth and the smaller half diameter): - a= 1-10 very deep reservoir, - a= 10-100 deep reservoirs, - a= 100-1000 shallow reservoirs, - a= 1,000- 10,000 very shallow reservoirs, - a= > 10,000 extremely shallow lakes
These comparative approaches provide: - correlation between reservoir configuration and the delay of induced seismicity. - Alternate classification of lakes and reservoirs, which include limnological about one hundred types of water bodies.
a. Impoundment related activity, about 1.3 years. Configuration: - very long and shallow reservoir [n= 137 a= 184] - very long and deep reservoirs [n= 137 a= 1-100]
b. Delayed activity, about 4.3 years. Configuration: - Elliptic and shallow [n=16.5 a= 160]
Elements associated with different environmental concerns not related to induced seismicity are presented in two tables.
Assessment of the land impact is carried out through two components: - Power generated per unit of reservoir surface, MW/km2. ; - land required to produce a power unit, Km2./MW
| MW/Km2 | LAND USE | Km2/MW | EXAMPLES | POWER MW |
|---|---|---|---|---|
| 100 | exceptional lands use | < 0.01 | Seto, Japan | 2,320 MW |
| 100 - 10 | very good land use | 0.01 - 0.1 | Itapu, Brazil | 12,600 MW |
| 10 - 1 | good land use | 0.1 - 1 | LG2, Canada | 5,320 MW |
| < 1 | less efficient land use | > 1 | Akosombo, Ghana | 1,782 MW |
This page last updated on: Sunday, January 31, 1999