Read Microsoft Word - Minutes MIT Expertmeeting 14012010.doc text version

Main observations of the expert meeting with prof. Bradford H. Hager and prof. Robert D. van der Hilst (MIT) about the technical review of the Bergermeer Seismicity Study, The Hague, 14-01-2010

Present: Hein Haak, mediator (Soil movement technical committee, Tcbb), Bernard Dost (KNMI). Jos De Groot , Cristel de Zwaan, Donald Voskuil, (Ministry of Economic Affairs) Jan van Herk, Hans de Waal, Hans Roest (State Supervision of Mines, SodM), Adri Plomp, Hans Haring, Alwin Hietbrink, Monique Klaver, Rick Lindhout (Bergen), Esther Oude Nijhuis, Corien Dijkstra, Hans Mulder (Alkmaar), Leo Bas (Heiloo), Eric Khodabux (prov. Noord-Holland), Peter van Langen (HHNK), Marius de Langen, Sieme Niks, Henk Beens Wouter Hubers (Gasalarm2), Jasper Vis, Wouter Botermans, Wim van Soest, (Taqa Energy BV), Ton Rijvovelt (MRD),

General remarks:

Prof. M.Nafi Toksöz is not present for personal reasons. Powerpoint slides of the presentation of MIT will become available for all participants.

Presentation of the findings of MIT.

After the welcome and opening words of the Chair, and the introduction of attendees, the formal part of the meeting started with a presentation by Prof Hager with a summary on the MIT review of the TNO report 2008-U-R1071/B as well as the MIT review of the TNO conclusions. (See PowerPoint slides.) Responses will then be given on the original questions of Gasalarm2, the Soil movement technical committee, and on a number of new questions. Prof Hager explains the context and scope of the MIT review of the TNO report. The focus of the MIT review is on seismic hazard, i.e. the probability of exceeding a given ground motion within a given time period at a given site. The probability distribution P (m,x,y.z,t) has been estimated using three approaches: a. Statistical analysis of limited data (empirical) b. Deterministic, geomechanical approach with fault area, slip and shear modulus c. Expert opinion. A possible explanation is presented for the mechanism of the observed reverse faulting (see 3-D figure). The re-activated fault in the Bergermeer field is called a "scissors-fault".Compaction takes place in the sandstone reservoir. As a result, aseismic shearing occurs where the sandstone is in contact with salt layers and as a result stress is transferred along the strike of the fault, concentrating in the pivot of the scissors where seismic events take place.

Summary of responses to earlier questions

Answers to the earlier questions of Gasalarm2 and the Soil movement committee are presented (see PowerPoint slides). In addition to these earlier answers on the effect of repressurization, a number of new elements are added by MIT. "The earthquakes that have already occurred may have reduced the stresses to near zero. Repressurization might increase the stresses on such faults." In addition to the earlier response to the questions by the Tcbb about the rate of stress changes, Prof. Hager adds that it is important `How fast you are injecting', in his opinion the cycling nature of the injection is not important.

Answers to new questions/discussion

The 1 % probability of an ML=3.9 event during the lifetime of the project given in the MIT report in response to a question by Gasalarm2 has to be seen as an operational definition of `a very small probability'. The estimation is based on an extrapolation of the log (N) vs magnitude plot for the Northern Netherlands, under a number of assumptions (e.g. neglecting uncertainties). This probability is much less than the value of 15% mentioned in the Gasalarm2 question. In response to a question of mr Hietbrink, prof. Hager answers that the lifetime of the project should be

1

corrected from 25 to 50 years. The stated probability has to be modified to a chance less than 2 % over the life of the project. There was a question about the statement in the review that a ML=3.9 event would be related to an intensity VI+. Prof Hager: "We used the Modified Mercalli Intensity scale, but on the European Scale (EMS) the intensity of a magnitude 3.9 event would indeed be closer to VII than to VI. (see attached figure of Magnitude vs. Intensity, courtesy Marius de Langen). Haak notes that the EMS scale should only be used when no accurate data are available. For the Bergermeer field there are accurate data for calibration as the seismic events have been recorded. In the `Rijksinpassingsplan' (RIP) an ML=3.9 event is described as `very light'. Question: "Is that a scientific qualification?" Answer by prof. Hager: "The evaluation of the EMS scale is not the topic of the MIT review". Dost adds: "Natural events are usually deeper than induced events. The descriptions of the impact of earthquakes are related to natural events. The induced events in the Netherlands occur at a depth of only 2-3 km. The impact is different from that of natural seismicity and concentrated in a smaller area. "Very light" is the right qualification for a natural earthquake, but for an induced earthquake this qualification is not preferred. Rick Lindhout asks a clarification on what 3.9 means for a natural earthquake. Dost explains that a deeper source radiates over a larger area, spreading the energy and the effects of the earthquake. Hietbrink concludes that an EMS Magnitude­Intensity relation is generally used for natural earthquakes and Rick Lindhout summarises for the record that the qualification `very light' is not appropriate for a ML=3.9 induced, shallow earthquake. Hietbrink asks about the likelihood that an ML=3.5 earthquake occurs in the field. Prof Hager answers that some relief of stresses could be expected while injecting gas in the field. Most experts would think that the fault structure stabilizes as a result. He himself belongs to the 10% of the experts that are not convinced about this because the earthquakes that occurred during production might have relieved the stresses. In that scenario, injecting gas would increase the stress. The question on the likelihood of an ML=3.5 event is repeated. A period of once every 15 years is mentioned for the region, based on the available cumulative frequency-magnitude relationship. Haak answers that no information is available on the (distribution of the) small earthquakes in Bergermeer. The lack of statistical information on smaller quakes in the Bergermeer field is subsequently discussed in more detail. Monique Klaver states: "we are looking for a sense of security for our citizens." Haak addresses the important point of monitoring. He expects a continuation of the events, which have been seen in the past. Hager expects that what has been calculated, is the worst case. What may well happen is that at higher gas pressures the stresses due to production tend to reduce. Lindhout is of the opinion that not enough data are available for extrapolation and drawing conclusions. Haak underlines that monitoring is an important issue. Beens makes the remark that monitoring could be `mustard after the meal.'

Continuation of discussion after short break.

Mr Dost remarks that a lot is known about the Bergermeer gas field. The four events are confined to a small part of the central fault in the middle of the field. Geomechanical models give similar results with respect to the maximum magnitudes. At the moment the most conservative (careful, prudent) approach is followed. Beens makes the remark that the Groningen field is not as mature yet as the Bergermeer field. Dost replies, that also gas fields such as Eleveld are taken into account, which are as mature as the Bergermeer field. Dost: different fields show a different behaviour, however when the events observed for different fields are added together, a consistent behaviour emerges. Mr van Herk makes some introductory remarks about the storage-plan and monitoring. State Supervision of Mines is an advisory body for the approval of storage plans. Conditions for approval may be related to the monitoring of seismicity and subsidence. The storage has to be operated in

2

accordance with the NEN norm for gas-storage 1918-2 1998. Van Herk points to the fact that Taqa has already obtained approval for a storage plan for a limited amount of gas, up to a reservoir pressure of 35 bar; Taqa is already injecting gas in the Bergermeer field. Taqa has asked approval for a second stage of injection with reservoir gas pressures up to a maximum of about 150 bar. This second phase (adaption of the storage plan) is studied. Van Herk expresses that most experts think, that due to the repressurization of the reservoir, the chance of seismicity may become lower. The importance of a good monitoring programme is underlined by the fact that there are some differences in opinion among experts about the behaviour of gas fields when they are repressurized. Questions asked are: What are the consequences of the observations of the monitoring? What should be accepted? Van Herk mentions that in the Dutch underground gas storage at Norg a fast repressurization did take place; however there is no significant level of seismicity. Only some minor events took place during depletion and injection. Monique Klaver noted that there was no seismicity before reinjection so that comparison with BGS is not appropriate. Mr Haak concludes again that the monitoring plays an important role. Mr de Langen presents a series of questions and concerns. He repeats the opinion that the geological/geomechnical model of TNO has too many limitations to predict the maximum possible magnitude. Prof Hager confirms that the best information to estimate the maximum magnitude of 3.9 comes from the geomechanical model like the fault area and maximum slip that can occur (not from statistics). Estimation of the probability is more difficult. The result is a very small probability, between 0,2 and 2% during the life of the project (much less than 15%). Prof. Hager estimates, as a worst case, that the probability for an ML=3,9 event is roughly comparable with that for the second stage of the production phase. Prof Hager: the max. magnitude could be described by ML=3.5 and the 3.9 can be seen as the upper bound at the 95% confidence interval (95 % limit at 2, = standard deviation). The 3.5 has to be seen as a calibration point. There is a rule of thumb to add 0.4 to the actually measured maximum magnitude. Mr. Beens mentions that the 3.9 from the KNMI is a 1 estimate. Prof Hager shows no disagreement with that figure based on a statistical analysis of KNMI (Monte Carlo method). Uncertainties are calibrated in the field with the events from 1994 and 2001. Estimates are made of the stress drop. Mr de Langen notes that differences in slip estimations are substantial. Dost points to the fact that the difference in slip estimations might be explained by e.g. aseismic creep. Mr de Langen asks for examples of the influence of the rate of pressure change on seismicity. Prof Hager confirms to have many examples and also information on the theoretical background. Prof Hager explains the following: earthquakes/seismic events may occur when stresses reach a certain threshold stress level. Stressing (injecting/producing) at higher rate will speed up this process. When we double the rate, the number of quakes may double in time. However only if a certain threshold stress value is exceeded. Mr De Langen presents two graph's with cumulative frequency-magnitude relationships, updated with recent data. Prof Hager is asked to give his opinion about the extrapolations. In the second (grey) graph an optimistic (A) and pessimistic (B) extrapolation is given. Prof Hager makes the remark that combining inhomogeneous datasets from different fields could be a problem. There are many ways to do such a statistical analysis. He mentions that in the grey graph the curve intersecting magnitude 3.7 at Ncu= 0 means that an ML=3.7 event should already have occurred at least once (100 =1). Such 3.7 event has not happened. The remark is made that, as this is based on the assumption of a Poisson distribution, the probability for the time period considered would be p=(1-1/e) = about 60 %. Dost makes the remark that the probability of a magnitude 3.5 event is better defined than the probability of a 3.9 event. There are no observational data for a 3.9 event. The proposed hypothesis of a more than 15 % chance of an earthquake, with 10 times stronger impact than 2001 (the question of Gasalarm2) can be rejected at a certain confidence interval.

3

In answer to a question about the use of Magnitude numbers of 2.4 and 2.7 in the MER-report, Prof. Hager refers to the text in the MIT study, where that question has already been answered. He does agree with the upper bound ML=3.9 as described in the TNO report. Mr Lindhout asks prof. Hager about living in the area close to an Underground Gas Storage. Prof Hager answers that, to illustrate his assessment, he would invest in a company that buys houses in the region and sells them after 5 years, when he expects that the concerns about seismic risk will have been greatly reduced. He would not protest if a UGS were to be built in his own neighbourhood. He acknowledges that possibly his neighbours, who are not experts in seismic hazard assessment, would do that. Mr Mulder asks about the need to move the gas storage to an area that is less sensitive for earthquakes or to design the installation for an ML=3,9 event. Prof Hager answers: when in an alternative reservoir all properties are equal, yes, it is better to avoid the risk. In response to the second question he replies that an engineer would overdesign the installation to be safe for an ML=3,9 event. Prof Hager is asked for the upper limit of a seismic event that he has seen for an existing storage project. Prof Hager answers that he is advising on a gas production project with a predicted maximum event of ML=4.5. Main conclusions, summarized by Dr. Haak at the end of the meeting. - Dr. Haak concludes that this has been a well-behaved meeting and he compliments all on this. - An upper bound of maximum magnitude ML=3,9 was derived with different methods. E.g. methods based on physics and the dimensions of the fault (generally preferred) and methods based on statistical arguments (which generally lead to more discussion). Because of the consistent outcome using the different methods, the ML=3,9 is a credible number as a maximum value for the Bergermeer field. - Statistical methods are inherently uncertain in predictive power but the probability of an ML=3,9 event is still seen as very small by MIT. - A period of 50 years should be taken into account for the project life, instead of 25 years. - The effect of the cycling and the temperature effects do not seem to be important. The temperature effects are insignificant when the injection takes place at a certain distance from the fault (about 150 m). No impact is expected from fatigue. The rate of stressing has a certain effect and references to research on this topic will be provided by prof. Hager. - Careful monitoring of the reservoir should be carried out and is recommended by MIT. Monitoring is part of the storage plan. Earthquakes and surface deformation are measured according to an approved measuring plan. The NEN norm 1918-2 1998 has to be followed. - A Magnitude 3,9 event would result in an intensity close to VII on the EMS scale. (this is a correction of the original statement in the MIT review: MMI VI+ becomes EMS VII). The relation between magnitude and Intensity is a noisy one. The meaning of the relation is different for (deep) natural earthquakes and (rather shallow) induced earthquakes. In general a shallow event results in a relatively larger impact in a smaller area around the epicentre. However, special soil conditions may give large effects in other area's as well. - TNO has used a 2-Dimensional geomechanical model. MIT concluded that a three dimensional model would be more appropriate. [Mr de Langen repeats that the geomechanical model by itself is not enough to estimate the maximum magnitude.] - There are two schools of thought on how the seismic risk is changing when a gas-reservoir is repressurized. School A: Injection is lowering the risk, because gradually inflating the reservoir will reduce the production-induced stresses in the reservoir-rock.

4

School B: The injection does not change the risk because the pressure increase during repressurization is substantial. (e.g. compared with the pressure drop between the events of 1994 and 2001). Prof Hager (in the position of the invited expert) states that he is a member of the second school of thinking, favoured by some 10% of the experts. Although he has reasonable doubts, he is not saying that the first school of thinking is wrong. He confirms that the rate of change in stress field will affect the frequency of events.

Final remarks

The mediator mr. Haak thanks prof. Hager and prof. van der Hilst for presenting the MIT review and answering so many questions. He thanks also all the participants who contributed to the constructive discussion in a balanced way. He mentions that a summary of the MIT-review is available in the Dutch language. The participants will get a copy of this summary.

5

Information

Microsoft Word - Minutes MIT Expertmeeting 14012010.doc

5 pages

Report File (DMCA)

Our content is added by our users. We aim to remove reported files within 1 working day. Please use this link to notify us:

Report this file as copyright or inappropriate

1300177


You might also be interested in

BETA
Microsoft Word - Minutes MIT Expertmeeting 14012010.doc