A TECDOC onSeismic Soil Structure Interaction – IAEA draft TECDOC (Chapter 1-7 of the R0 Version))

COMMENTS BY REVIEWER
Reviewer:
Country/Organization: U.S. NRC Date: / RESOLUTION
JJJohnson (JJJ)
Comment No. / Page/Section / Comment / Accepted / Accepted, but modified as follows / Rejected / Reason for modification/rejection
1 / 14/1.2 / These objectives specifically indicate that the TECDOC will address SSI issues. Does on include hazard characterization nor the use of the SSC responses. / X / I am not sure whether this is only an observation or if two Objectives should be added stating that (i) hazard characterization for design and assessments that is a necessary element of SSI is treated; and (ii) the end result, i.e., in-structure forces and moments or stresses and in-structure response spectra, are treated to the extent feasible and relevant.
2 / 15/1.2 / Scope however deviates from the objectives by defining the use in BDB assessments’ such as SMA’s and SPRA’s / X / Bullets 4 and 5 on p.14 in the Objectives sub-section state that “design and assessment purposes”; we also state that assessments cover BDBE and response calculations for cases in which an actual earthquake ground motion has impacted the site.
3 / 15/1.3 / The hazard can be defined in deterministic or probabilistic space but the input motion used for SSI is unique for the input level of the hazard. The SSC design is deterministic based on that unique input. Suggest explaining this and discuss the treatment and levels of input hazard used in different countries in the Appendix. / X / Rewritten Chapters 2 and 4 and current to be updated Chapters 5 and 6 address the first part of this comment.
We are attempting to obtain the information concerning definition of seismic input from as many countries as possible and responsive. / X / The statement “the input motion used for SSI is unique for the input level of the hazard” is only correct for site specific motions when an actual GMRS or PBRS is selected. There are many steps prior to the selection that lead to different possible DBE and BDBE ground motion definitions within the state-of-practice and regulations. Also, time histories to be used in SSI analyses are not unique, etc.
4 / 20/2.1 / If methods for free field ground motion is the point of discussion then the other options to g should be included. / X / Chapters 2, 4, 5, 6 have been rewritten or modified; please review the next version.
5 / 21/2.2.1 / Why is it important to bring in the different scales of exploration? The site specific information need for SSI analysis is only constrained to the site area. / X / Geological and geophysical data provides information on uncertainty in soil stratigraphy and material properties.
6 / 23/2.3 / Now there is a trend to define hazard as UHSS (uniform Hazard Soil Spectra). A discussion non PSHA approach to ground motion definition need to be included to better explain the advantages of using PSHA type of approach. / Unfortunately, UHSS will need to be researched. Comment on DSHA is important – currently we point to IAEA SSG-9, but not is detail.
7 / 23/2.4 / Recommend that important concepts be captured in the text with a ref to ASCE 4-16 rather than directing the readers to a country specific document. / X / This is an IAEA document, which involves as many countries as wish to participate providing their state-of-practice through the questionnaire and reviews of the document. We all gain by engaging and understanding the procedures and processes in place for as many countries as feasible.
8 / 29/3.1 / The concepts used to determine 1-D or 3-D modelling for site configuration can be better visualized if the explanation is associate with graphics.
BJ to address. / X / BJ to address.
COMMENTS BY REVIEWER
Reviewer:
Country/Organization: U.S. NRCDate: / RESOLUTION
JJJohnson (JJJ)
Comment No. / Page/Section / Comment / Accepted / Accepted, but modified as follows / Rejected / Reason for modification/rejection
9 / 37/3.3.1 / It is recommended that the other soil models be associated with the software as has be done with Table 3-2 for SHAKE, FLUSH and SASSI. Applies to 1-D, 2-D and 3-D models.
This comment is similar to Peter Rangelow’s comment. AP to address. / AP to address.
10 / 116 thru 123/7.2 / Suggest that the expiations about the methods are followed by their limitations and advantages with some examples of their proper application. / X / Chapter 8.
11 / 123/7.3 / Suggest adding a discussion on why the sub-structuring method. What problems are overcome using variations of this method and the limitations of this method?. / X / JJJ to add.
12 / 137/7.4.2 thru 74. / There are a number of material and modelling issues discussed. It would be helpful if these were related to the appropriate segments of the SSI analysis showing how these effects are utilized in the analysis. / X / Chapter 8.
13 / 152/7.4.7 / 1. How has the effect of inclined waves been accounted for in the development of ground motion ?
2. and in the SSI analysis software available today? / X / 1. That is a very good question that requires input from seismologists and seismic hazard analysts that have developed elements of the PSHA and DSHA methodology. Have the researchers opined on whether non-vertically incident waves were important in the databases used for GMPE development. It is also one of the questions that we identified in Chapter 4 to inquire of the hazard analysts.
2. Assume sub-structuring approaches are dominant in today’s state-of-practice two of which are CLASSI and SASSI. For the assumptions in CLASSI (namely rigid foundation behaviour), non-vertically incident body waves and surface waves can be analyzed. Similarly for SASSI which is not limited to rigid foundation behaviour. We will attempt to demonstrate both (or at least one of these cases) and compare with other results.
3. In the past and currently, there is an explicit or implicit assumption that soil property variations and enveloping of results covers this issue. In the U.S., that has been the approach in ASCE 4-86, -98, -16. It was tested several decades ago and demonstrated to be a reasonable approach. Also, recent studies for high frequency ground motion have demonstrated that non-vertically incident waves assumed to be travelling at an apparent wave speed of 2-4km/s do not have a significant impact on response. (Note, this was for high frequency ground motion.)
14 / 152/7.4.7 / It is noted that there is a variability of the wave front with depth and across the horizontal dimension. How is this currently incorporated in the SSI analysis? / Section 7.4.7 is SMRs.
I do not see this text.
BJ to address. (JJJ can help.)
15 / 152/7.4.7 / If the depth of the surface waves two wavelengths deep, why would they not affect the current NPP foundations? How has these wave effects accounted in the SSI analysis / X / See response to Comment 13.
16 / 11/1.1 / 5.In general, more references need to be provided. For example, in Chapter 1 the document states “SSI analysis methodologies evolved over the 1970s and 1980s. Simplified soil spring approaches continued to be used in various contexts. More complete substructure methods emerged, specifically developments by Luco and colleagues (University of California, San Diego) and Roesset, Kausel and colleagues (Massachusetts Institute of Technology). The University of California, Berkeley Team (Seed, Lysmer and colleagues) developed several direct approaches to performing the SSI analyses (LUSH, ALUSH, FLUSH).”
References for key papers/reports by Roesset, Kausel, Seed, Lysmer, etc. should be provided. / X / See p. 11 “As one element of the U.S. Nuclear Regulatory Commission’s sponsored Seismic Safety Margin Research Program (SSMRP), the state of knowledge of SSI as of 1980 was well documented in a compendium Johnson [1-1] of contributions from key researchers (Luco; Roesset and Kausel; Seed, and Lysmer) and drew upon other researchers and practitioners as well (Veletsos, Chopra). This reference provided a framework for SSI over the 1980s and 1990s.
[1-1]Johnson, J.J., “Soil Structure Interaction: The Status of Current Analysis Methods and Research,” Lawrence Livermore National Laboratory (LLNL), UCRL-53011, NUREG/CR-1780, prepared for the U.S. Nuclear Regulatory Commission, Washington, D.C., 1981.
17 / The report states, “Although research in nonlinear methods has been performed and tools, such as Real ESSI, have been implemented for verification, validation, and testing on realistic physical situations, adoption in design or validation environments has yet to be done.”
Is it possible to reference other codes as well so this does not appear IAEA is endorsing a specific code? Also, some references to documentation Real ESSI and any other tool mentioned should be cited and included in the references. / X / 1. Bullet has been enhanced with LS-DYNA:
The methods implemented during these three decades and continuing to the present are linear or equivalent linear representations of the soil, structure, and interfaces. Although research and development in nonlinear methods has been performed and tools, such asReal ESSI, LS-DYNA and others, are available [J1]adoption in design or validation environments has yet to be done. (Maybe we can add other Nonlinear SSI analysis software)
2. BJ to add references.
18 / 16/1.3.2 / When citing US practice in Section 1.3.2, references to NPP performance goals should be provided. For example, CDF and LERF goals of ≤ 1E-4 and 1E-6 are given without any reference to where these numbers come from. This is another example that is associated with comment 16 / X / JJJ to add reference
19 / 19/2.1 / In Section 2.1, a more descriptive definition of the term “spatial variation of motion” should be provided / X / Chapter 2 rewritten, but this term is used in other Chapters as well and it seems to be self-explanatory.
20 / 21/2.2.1 / In Section 2.2.1 the document states, “This data is generally low strain data.” This is true for geophysical data, however, geotechnical data is not necessarily low strain data. / X / Refer to Chapter 3. AP to address.
21 / 21/2.2.2 / Should section 2.2.2 be called, for example, “Considerations for materials models selection”? / X / Chapter 2 rewritten – for soil properties, refer to Chapter 3.
22 / 24/2.5 / Section 2.5 begins with a statement, “See Section 2.3 for overall discussion of modelling.” Should the section provide a few introductory sentences about the scope of this section before referring to a prior section? / X / Chapter 2 rewritten
23 / 25/2.6 / In Section 2.6, “Partially embedded (less than all sides),” is not clear. Does it refer to the ratio of embedment depth to the height of the structure or that only part of the perimeter is embedded? / X / Part of the perimeter is embedded meaning there may be specific situations where only part of the foundation/partially embedded structure is adjoining soil (e.g., intake structure, building adjacent to another building)
24 / 28/3.0 / The document started using “shall” as if this were a code or standard. As a TECDOC, I don’t think the document should use code/standard language. / X / Shall should not be used.
25 / The authors reference the NERA project. More details are needed so that readers can access the specific information discussed in the NERA project. The project website is not sufficient to easily find the information that is being referenced. / X / AP to address.
26 / 29/3.2 / Consider changing the title of Section 3.2 to “Soil Behavior.” / X / AP to address.
COMMENTS BY REVIEWER
Reviewer:
Country/Organization: Date: / RESOLUTION
Comment No. / Page/Section / Comment / Accepted / Accepted, but modified as follows / Rejected / Reason for modification/rejection
27 / 32/3.3 / 16.Section 3.3 states that the threshold strain is typically 1E-4 to 5E-4. Ishihara (1996) suggests that equivalent linear models are applicable at shear strains on the order of 1E-3 to 1E-2. Maybe there has been a change in the consensus among practitioners since 1996. At a minimum, I think that the work by Ishihara should be cited and discussed along with the work by others that may suggest a smaller threshold strain. Work performed recently by Zalachoris and Rathje (2015) indicates that both 1D equivalent linear and non-linear site response analyses may overdamp high frequency motions. So, just because a threshold strain has been exceeded does not mean we are getting better results from the non-linear analysis.
Ishihara, K. (1996). “Soil Behavior in Earthquake Geotechnics,” Oxford Science Publications.
Zalachoris G. and Rathje E.M. (2015) “Evaluation of one-dimensional site response techniques using borehole arrays,” Journal of Geotechnical and Geoenvironmental Engineering, 141(12). / X / AP to address.
28 / 49/3.6 / Section 3.6 states, “For elastic and viscoelastic linear constitutive models, calibration usually does not pose any problem; the experimental data (elastic characteristics, G/Gmax and damping ratio curves) measured either in situ and/or in the laboratory are directly used as input data to the models. However, validation shall not be overlooked: results should be critically examined since, as indicated previously, those constitutive models are only valid for strains smaller than a given threshold v. If results of analyses indicate larger strains, then the constitutive models should be modified and nonlinear models should be advocated.”
a.I believe it is also common for practitioners to compare lab test results to published shear modulus degradation and damping curves (e.g. EPRI or Darendelli) and use the published curves.
b.From a theoretical standpoint, I agree that when you go beyond a threshold strain, non-linear models should be used. However, from a practical point of view, I am not sure recorded motions demonstrate that the non-linear analyses produce better estimates of ground motion (see Zalachoris and Rathje (2015)).
c.See also comment 27. / X / AP to address.
29 / 54/4.1 / In Section 4.1, there is an equation to calculate P(Y > y) is given. However, there is no introduction to or discussion of this equation. / X / Chapter 4 is rewritten and this equation has been removed.
30 / 56/Table 4-1 / In Table 4-1 in the Output column, there is a row for Magnitude-distance deaggregation. Consider adding epsilon to make this Magnitude-distance-epsilon deaggregation. / X / JJJ to add.
31 / 59/4.4 / In Section 4.4, GMPE is used for ground motion prediction equation. Consider also adding GMM for ground motion model. That is the terminology being used for NGA East because there will be no equations used for implementation of these ground motion models. / X / JJJ to add.
32 / 59/4.5 / Much of Section 4.5 is a repeat of Section 4.1. I don’t see the need for the redundancy. / X / Chapter 4 is rewritten. We agree.
33 / 61/5.1.1 / In Section 5.1.1 the document states, “Depth of source will also control magnitude.” I assume this is a typo. Seismic moment is a function of shear modulus, rupture area, and average displacement during rupture, and seismic moment is related to the magnitude. I anticipate that the intent of the statement is that deeper earthquakes result in lower levels of ground motion in comparison with shallow earthquakes.` / X / Not found in Version 509. I think “magnitude” was intended to “amplitude”. Text was from BJ.
34 / 69 / Sometimes there is document text between a figure and the figure caption (See Figures 5-3 and 5-4). / X / Editor.
35 / 70/5.3.2 / In Section 5.3.2, Numerical Earthquake Model, the document discusses large scale regional models that encompass the fault and geology in great detail with current capability of modeling motions up to 5 Hz. I think that there should be a clear distinction through the document where standards of practice are presented and then where state-of-the-art and areas where additional research are needed to bring the state-of-the-art into the state-of-the-practice. / X / Agreed. This is consistent with Philippe Renault’s comment and Sujit’s comment.
36 / 73 / Check equation numbering / X / BJ to check.
37 / 75/Eqn. 5.2 / The terms of Equation 5.2 are not explained in the text of the document / X / Formula for coherency. BJ to check.
38 / 75/5.4.2 / In Section 5.4.2 on Nuclear Power Plant (NPP) – specific applications the document states, “The peak spectral acceleration for all 12 sites is at 25 Hz.” This is in part due to the fact that spectral accelerations are only calculated at 7 frequencies. Consider stating that the actual peak may be at frequencies somewhat higher or lower. / X / Figure and description to be replaced by Figure from Philippe Renault’s paper which I have – JJJ to delete and replace. Commentary is good.
39 / 88/6.2 / In Section 6.2, the following bulleted item was not clear to me, “If the GMPEs are specified for a suite of natural frequencies, and implemented for the suite, then the resulting seismic hazard values or curves can define a site specific ground response spectra that adequately matches the site of interest. These seismic hazard values or curves could be probabilistically or deterministically defined.” / X / JJJ to check with AP on this bullet. (Looks correct to me)
40 / 91/6.2.2 / In Section 6.2.2, the document reads, “In all cases, when the DBE ground motion is defined at TOG, structures to be analyzed are modeled including embedment, and three deterministic soil profiles are used in the SSI analyses (best estimate, lower bound, and upper bound).”
I think there should be some additional discussion on how the best estimate, lower bound, and upper bound profiles are developed. / X / JJJ to add an approach taken in ASCE 4-16.
(Rewritten in response to comment from IRSN on only three soil profiles. Also, methods of deconvolution were added by AP.
41 / 91/fig 6-1 / I recommend additional discussion in the text regarding figure 6-1. Consider adding more practical discussion on practical ways of developing the FIRS, such as guidance in NRC’s ISG-17 and the references therein to NEI and Brookhaven National Labs. / X / JJJ to review and add/clarify.
42 / 91/6.3 / In Section 6.3 some discussion on present site response analysis is provided. The document mentions that response spectra are represented by time histories of events characterized by deaggregated events. It is also possible to directly develop Fourier amplitude spectra using a random vibration theory inversion technique. This technique for developing the input motion for an RVT site response analysis has been used in US practice. / X / JJJ to add – approach by Rathe et al.
43 / 97/6.3 / In the Convolution part of Section 6.3, the document states, “Normally, at least over the last five or more years, these probabilistic analyses are performed for a series of (up to) 60 earthquake simulations.”
Some clarification should be provided. Up to 60 earthquake simulations could be interpreted to be up to 60 time histories are used in the analyses or that 60 random profiles are generated for each base case velocity profile. Based on the wording in the document, I am not sure which meaning is intended. / X / JJJ to review and add/clarify.
44 / 98/6.3.3 / A portion of Section 6.3.3, 1D Models is a repeat of text from a prior section. Also in this section we read “Instead of Equivalent Linear Analysis approach, a 1D nonlinear approach is also possible. Here, 1D nonlinear material models are used (described in section 3.2). Nonlinear 1D models overcome biased estimates of site amplification, as they model stiffness in a more accurate wave. While modeling with nonlinear models is better than with ELA procedures …” From a practical perspective, I am not sure this statement is accurate (see Zalachoris and Rathje, 2015). If it is, please provide references which show that use of non-linear 1D models consistently match recorded ground motions better and with less bias in comparison with equivalent linear models. / X / BJ to address.
45 / Page 14,Section 1.2 (Objectives) / This section states:
This TECDOC provides a detailed treatise on SSI phenomena and analysis methods specifically for nuclear safety related facilities.
The purpose of the task is to review and critically assess the state-of-the-practice regarding soil-structure interaction methods. The emphasis is on the engineering practice, not on methods that are still at the research and development phase. The final goal is to provide practical guidance to the engineering teams performing this kind of analyses in present and in near future.
Comment: Consistent with the above, the report should augment the discussions regarding the state of practice with respect to linear SSI analysis. Further, the current discussionson aspects related to nonlinear time domain SSI analysis with special focus on soil behaviour, should be balanced with additional discussion on the respective implications on the structural response. / X / Chapter 8.
46 / Page 31/Section 3.2 / The discussion refers to Figure 3-1 as left and whereas the figure currently shows two plots top and bottom. (make consistent). / X / AP to address.
47 / Page 36/Section 3.3.1 / Refers to equation 0.1. Should refer to equation 3.1. / X / AP to address.
48 / Page 52/ Section 3.7 / Verify if Figure 3-12 “Left: Elastic Stiffness of a soil …” should be renumbered to Figure 3-13. Also correct text which refer to Figure 3-13 / X / AP to address.
49 / Page 98/Section 6.3.1 / Replace BDE by BDBE. / X / Done in Version 509.
50 / Page 99/Section 6.3.3 / Consider enhancing the description of the 1D models to describe cases for which is reasonable to use 1D equivalent linear approach (e.g,. cases where strain levels remain low and for level or gently sloping sites). Should this section describe using one-dimensional site response with SSI in 3 orthogonal directions? / X / AP/BJ/JJJ to address
51 / Page 101/Section 6.3.3 / The paragraph starting with “An alternative analytic procedure…” needs to be completed. Discuss the Exponential Window Method. / X / BJ to address.
52 / Page 106/Section 6.4.3 / Verify if the discussion should indicate that the scale factors are applied to the mean 1 e-4 rather than the mean 1e-5. / X / In the revision to ASCE 43-05 (in process), a Scale factor (SF) is defined contrary to the existing ASCE 43-05 where a Design factor was defined and applied. The SF multiplies the mean 10-5 hazard.
53 / Page 109/Section 6.5 / For the alternative of using a single time history, consider adding discussion on time histories generated to match a target design response spectra for all damping values used in the seismic response analysis (e.g. SRP 3.7.1 Option 1, Approach 1). / X / JJJ to add.
54 / Page 109/Section 6.5 / Verify if the second sentence in the bulleted paragraph discussing time history duration, should refer to “strong motion duration” in place of “duration.” / X / Done in Version 509.
55 / Page 110/Section 6.5 / This section indicates the calculation of PSD as conditional on the RS matching. Verify consistency with the latest version of SRP 3.7.1which updated this guidance to calculate the PSD always and demonstrate that no gaps in energy exist. / X / JJJ to check.
56 / Page 112/List of Section 6 References / The reference list is missing Reference 6-6 (Houston et al.), referred to on Page 111. / X / JJJ to add.
57 / Page 132/Section 7.3.4 / Regarding the discussion on the subtraction method and the extended (modified) subtraction method, consider providing comparisons of transfer functions obtained from these methods and also form the flexible volume method that demonstrate the similarities and differences in the results obtained from these methods. / To be discussed by team.
58 / Page 147/Section 7.4.4 / This section discussing contact modelling at the interfaces between structural foundation and soil/rock. Consider also discussing how the effects of separation or loss of contact between the structure (embedded portion of the structure and foundation mat) and the soil, are addressed/evaluated in linear elastic analyses. SRP 3.7.2 indicates that Perform sensitivity studies should be performed to assist in judging the adequacy of the final results. ASCE 4 provides additional guidance. Also consider comparing the aforementioned effects for both shallow and deeply embedded structures, in regards to structural response. / X / BJ to address.

Page 1 of 12