processing of time histories - civil engineeringbartlett/cveen7330/processing of time...

Course Information○

UDOT Ground Response Analyses Report (p. 34 - 51)○

Reading Assignment

Caltranstimehistory.pdf○

Other Materials Material

Install Seismosignal software on your computer○

Install Seismomatch software on your computer○

Homework Assignment #2

You are to develop the input acceleration time histories for the seismic evaluation of the city office building Hurricane, Utah. Using the information provided in this lecture do the following:

Using the USGS mapping for 2008, determine the controlling earthquake in terms of magnitude (M) and distance from the seismic source to the project site (R) for this area. The deaggregation information can be found at:

1.

https://geohazards.usgs.gov/deaggint/2008/

Show how you use this information to determine the M and R values that are appropriate for the subsequent steps in this homework (10 points). Select from the PEER strong motion database, one candidate time history that is consistent with the controlling earthquake.

Use ASCE 7-05 to develop a design target spectrum for this site for site class C conditions. State all assumptions that you made in developing the target spectrum (20 points).

2.

Adjust the target spectrum developed in problem 2 for near fault effects (i.e., fault directivity). Adjust the target spectrum as described in this lecture for the fault normal component only (10 points).

3.

© Steven F. Bartlett, 2014

Processing of Time HistoriesMonday, January 13, 2014 2:32 PM

Processing of Time Histories

https://geohazards.usgs.gov/deaggint/2008/


Geological mapping has shown that the Hurricane fault is a significant seismic source: https://earthquake.usgs.gov/hazards/qfaults/kml.php Using this link, you can download the Google earth kml file required to locate the Hurricane Fault in Google earth. Use the link for Holocene to Latest Pleistocene found on this page. This Hurricane fault is a normal fault. Using the selected time history in problem 1, develop 1 set of spectrum-compatible horizontal acceleration time histories for analyses of the dam for potential rupture of this fault system. This set should consist two components: fault normal component and a fault parallel component. Use Seismosignal and Seismomatch to process the time histories and match them to the target spectrum. This processing should include, rotation, filtering, spectral matching and baseline correction, as appropriate (20 points).

4.

Please provide the following plots for the fault normal and fault parallel components of ground motion.

plot of both components of time history selected in problem 1○

plot of both components of rotated time histories○

plot of spectrally matched time histories (fault normal and fault parallel components)

○

plot of target spectrum versus matched spectrum for both components

○

plot of match and non-matched time histories superimposed on each other

○

plot of baseline-corrected, spectrally-matched time histories for the fault normal and fault parallel directions

○

Processing of Time Histories (cont.)Monday, January 13, 2014 2:32 PM


https://earthquake.usgs.gov/hazards/qfaults/kml.php

Recommendations for selecting of candidate time histories for spectral matching and ground response analyses

The candidate time histories for the analyses should come from earthquakes that have earthquake magnitude and source-to-site distance that are approximately the same as the controlling earthquake magnitude and source distance associated with the proposed site. The controlling earthquake is that earthquake and its associated fault that has the largest contribution to the seismic hazard for a given site. This information can be determined from the candidate site by a deagregation of the probabilistic seismic hazard analysis (PHSA) using the tools from the following website:

1.

http://eqint.cr.usgs.gov/deaggint/

The candidate time histories for spectral matching should be selected from the National Geophysical Data Center (NGDC), U.S.G.S., PEER and California Strong Ground Motion Instrumentation Program, or other appropriate strong motion databases. We found that the PEER web site was particularly useful because its records had been already pre-processed for engineering evaluations.

2.

http://peer.berkeley.edu/smcat/

The earthquake magnitude, M, of candidate time histories should be within +0.5 M of the controlling fault magnitude for the proposed site. For example, a site with a controlling earthquake magnitude of 7.0 should have candidate time histories selected from earthquakes with M between 6.5 and 7.5.

3.


RecommendationsMonday, January 13, 2014 2:32 PM


http://eqint.cr.usgs.gov/deaggint/

http://peer.berkeley.edu/smcat/


a. R < 15 kmb. 15 < R < 30 kmc. 30 < R < 50 kmd. R > 50 km

In addition to earthquake magnitude, it is important that the candidate time histories have the appropriate source-to-site distance. This criterion is often difficult to meet for moderate to large earthquake that are close to the seismic source because there is only a handful of appropriate records. To aid in determining the appropriate distance for the candidate time history, we propose dividing the source-to-site distance into the following four categories:

4.

We recommend that the candidate time histories be selected for the appropriate M and from events that fall within the same source-distance category. For example, if the controlling source distance for the design event is 20 km, then candidate time histories should be selected from source distances that fall between 15 and 30 km,

5.

Whenever possible, we recommend the selection of candidate time histories from the appropriate tectonic regimes. For Utah, the tectonic regime is extensional.

6.

Whenever possible, we recommend that the candidate time histories have peak ground acceleration (PGA), peak ground velocity (PGV) and peak ground displacement (PGD) with minus 25 percent and plus 50 percent of the target spectral values (CALTRANS 1996a). This will allow the spectral matching process to be completed with less difficulty. In addition, the spectral matching process will not introduce as large of change in the spectral content of the matched time history.

7.

It is recommended that the selection include at least 3 and as many as 7 time histories for the ground response analyses. The number of time histories to be used in nonlinear dynamic analyses should take in account the dependence of the response on the time domain characteristics of the time history (e.g., duration, pulse shape, pulse sequencing) and its spectral response content. ASCE 4-98 recommends that at least 3 independent time histories be used for non-linear analyses.

8.

Recommendations (cont.)Monday, January 13, 2014 2:32 PM



Note that the increase spectral values to account for directivity begins at 0.5 seconds.

The maximum value of a 1.2 factor or 20 percent increase is reached at a period of 2.0 s and continues at 1.2 for the remaining part of the spectrum.

Based on recommendations from CALTRANS

Adjustment of Spectrum for Near Fault EffectsMonday, January 13, 2014 2:32 PM



The candidate acceleration time histories should be rotated to find their principal components and the principal component used for spectral matching. If the candidate time histories have been selected to represent near-field motions having strong velocity pulses in the fault-normal component, it is important the horizontal components of these motions be transformed into their principal components so that these align with the direction of fault directivity. The major and minor principal components are the directions that best correlate with the fault-normal and fault-parallel directions. To accomplish this, the horizontal motion of the two recorded components, ax(t) and ay(t) are transformed into a new set of orthogonal axes x’ and y’ as shown in the below figure.

The transformed accelerations in the x’ and y’ directions are calculated from:

ax’(t) = ax(t) cos q1 + ay(t) sin q1

ay’(t) = -ax(t) sin q1 + ay(t) cos q1

Rotation of Time HistoriesWednesday, January 08, 2014 2:32 PM



The principal components are found by minimizing the covariance between ax’(t) and ay’(t). The covariance is calculated from:

Substituting x' and y' for x and y, respectively, in the above equation yields the corresponding relations that define the covariance of components ax'(t) and ay'(t).

When the above function is minimized (found to be zero), the corresponding rotation angle (theta1) defines the orientation of the major and minor principal components. The major principal component is the component with the highest accelerations and should be used to represent the ground motion in the direction of the fault directivity. (Sometimes is also useful to examine the magnitude of the pulses in the corresponding velocity time histories to determine which of the two components represent the major component.

For example, the figure on the next page shows the unrotated 1987 Superstition Hills acceleration time history. The covariance between the ax(t) and ay(t) is minimized at theta1 angle of 25 degrees counterclockwise). At this angle, the rotated 135 degree component becomes the major principal component (i.e., the principal component is found at an azimuth of 95 degrees). The rotated time history are also shown on the subsequent page. Note that the peak acceleration has increased in each of the rotated time history in the major principal component direction. The Excel spreadsheet (rotation.xls) was used to perform the rotations and is included on the course website.

Rotation of Time Histories (cont.)Wednesday, January 08, 2014 2:32 PM



Rotation of Time Histories (cont.)Wednesday, January 08, 2014 2:32 PM



Rotation of Time Histories (cont.)Sunday, August 14, 2011 3:32 PM



If filtering is required, this will remove any additional unwanted noise in the candidate time history. This can be reduced through the use of filters at both high and low pass frequencies. The BAP manual (1992) suggests that high frequency noise (i.e., between 30 and 50 Hz) may originate in several ways: (1) from earthquake-induced vibrations in equipment close to the recorder, (2) from an unexpected higher-mode oscillation in the mechanical transducer, (3) or from the inability of the automatic trace-following digitizer to cope with an unclear photographic trace. The BAP manual suggests that unless it can be verified that high-frequency content is in fact useful earthquake input, the high frequencies should be filtered out. The use of a high and low pass filtering removes unwanted noise and produces a frequency range over which the recorded signal of the earthquake ground motion significantly exceeds the noise level. Generally it is recommended that an anti-aliasing filter such as a Butterworth filter should be used rather than an abrupt cut-off frequency that is used by the program SHAKE.

Butterworth filtering can be accomplished in Seismosignal.

If the candidate time history has not been filtered, then this is done prior to the spectral matching process. (Note this step is not required for records from the PEER website, because filtering has already been done.)

It is recommend to use a low pass Butterworth filter to remove frequencies greater than 15 Hz from the rotated acceleration time histories prior to spectral matching. It is also recommended to use a high pass Butterworth filter for frequencies less than 0.14 Hz (T = 7.0s) as recommended by Geomatrix (1999). The high and low pass filters are included Seismosignal.

See next page

FilteringWednesday, January 08, 2014 2:32 PM



Screen shot from Seismosignal

Note that a Butterworth filter has been applied and the filter configuration is set as a bandpass. The order is a 4th order filter and freq 1 is set at 0.07 Hz and freq 2 is set at 15 Hz.

Note the effects of filtering and baseline correction on the time history (blue line = unfiltered, uncorrected, grey line = filtered, corrected)

Filtering (cont.)Wednesday, January 08, 2014 2:32 PM



Spectral Matching

Spectral Matching creates high-quality design ground motion time histories by taking actual earthquake accelerograms and adjusting them to match a target response spectrum. These time histories are used by structural engineers in non-linear analyses of the dynamic response of buildings and soil structures to earthquake ground shaking.

The ground motion time histories used in analyses need to accurately reflect a design level of safety and have realistic time-dependent characteristics.

Pasted from <http://www.ez-frisk.com/Tech/SpectralMatching/Spectral.html>

Pasted from <http://www.stanford.edu/~bakerjw/research/spectral_matching.html>

Spectral MatchingSunday, August 14, 2011 3:32 PM


http://www.ez-frisk.com/Tech/SpectralMatching/Spectral.html

http://www.stanford.edu/~bakerjw/research/spectral_matching.html


http://www.seismosoft.com/en/SeismoMatch.aspx

Program is available at:

Simultaneous matching of a number of accelerograms, and then creation of a mean matched spectrum whose maximum misfit respects a pre-defined tolerance

•

Possibility of using this software in combination with records selection tools and records appropriateness verification algorithms to define adequate suites of records for nonlinear dynamic analysis of new or existing structures

•

Capability of reading single accelerograms defined in both single- or multiple-values per line formats (the two most popular formats used by strong-motion databases) or of reading a number of accelerograms at the same time (if they are defined in the single-value per line format)

•

Creation of the target spectrum by following Eurocode 8 rules, by computing the spectrum of a specific accelerogram or by simply loading a user-defined spectrum

•

SeismoMatch is an application capable of adjusting earthquake accelerograms to match a specific target response spectrum, using the wavelets algorithm proposed by Abrahamson [1992] and Hancock et al. [2006]. Some of its features are:

Elastic response spectra and pseudo-spectra •Overdamped and constant-ductility inelastic response spectra •Root-mean-square (RMS) of acceleration, velocity and displacement •Arias (Ia) and characteristic (Ic) intensities •Cumulative absolute velocity (CAV) and specific energy density (SED) •Acceleration (ASI) and velocity (VSI) spectrum intensity •Housner intensity •Sustained maximum acceleration (SMA) and velocity (SMV) •Effective design acceleration (EDA) •Predominant (Tp) •Significant duration•

The following strong-motion parameters are then computed for the matched accelerograms:

Pasted from <http://www.seismosoft.com/en/SeismoMatch.aspx>

Spectral Matching using SeisomatchSunday, August 14, 2011 3:32 PM





Unmatched Kobe Record (candidate record for spectral matching)

Acceleration Response Spectrum (Kobe Record - 5 percent damped)

Note that this is acceleration time history is one that has been provided by Seismomatch in its default folder. We will use it as a candidate time history for this example, even though it may not be strictly applicable for a real site and a real design case.

Spectral Matching in SeismomatchSunday, August 14, 2011 3:32 PM



This represents the design spectrum at the surface for a given site. This design spectrum is referred to as the target spectrum and it is often determined using methodologies such as:

Code-based design procedures such as ASCE 7-051.Attenuation relations2.Probabilistic seismic hazard analyses (PHSA)3.

T (s) SA (g)

0 0.590.01 0.5950.02 0.6050.03 0.6420.05 0.7310.075 0.8570.1 1.0010.15 1.150.2 1.2190.25 1.2440.3 1.2490.4 1.2220.5 1.2120.75 1.091 0.9751.5 0.7682 0.6223 0.434 0.3245 0.269




Importing the target spectrum within Seismomatch using the Load Spectrum from file option. This is target spectrum is the same as the design spectrum given previously.

Comparison of the Kobe response spectrum with the target spectrum. Note that the Kobe record is has lower amplitudes for all periods. The goal of spectral matching is to increase the amplitude of this record so it more closely matches the target spectrum.




Results from the spectral matching.




Recording and processing of time history can introduce drift in the record. This drift is noticed in the displacement time history which has been accentuated by the double integration process of the acceleration time history.

blue line = base line corrected time history, grey line = uncorrected time history

When applying the baseline correction, a quadratic correction is usually sufficient, as shown below; however inspection of the uncorrected displacement time history can be used to determine the best function (constant, linear, quadratic, cubic) for correcting the time history.

Note that a quadratic correction is usually appropriate for cases where an acceleration time history has been double integrated to calculate a displacement time history. An error in the acceleration time histories creates a linear drift when integrated to a velocity time history and a quadratic error when integrated again to a displacement time history.

Baseline correctionWednesday, January 08, 2014 2:32 PM



BlankSunday, August 14, 2011 3:32 PM


processing of time histories - civil engineeringbartlett/cveen7330/processing of time...

Documents