You are on page 1of 11

Dynamic Analysis - Response Spectra Page 1 of 11

Dynamic Analysis - Response Spectra

You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
mk:@MSITStore:C:\RISA\risa3dw.chm::/3D_2D_Only_Topics/Dynamic%20Analysis%2... 05-Feb-15
Dynamic Analysis - Response Spectra Page 2 of 11

Dynamic Analysis - Response Spectra


A response spectra analysis may be performed after the dynamic analysis to obtain forces, stresses and deflections. In
general, the response spectra analysis procedure is based on the assumption that the dynamic response of a structural
model can be approximated as a summation of the responses of the independent dynamic modes of the model.

To Perform a Response Spectra Analysis

1. Select Dynamics (Eigensolution / Response Spectra) from the Solve menu.


2. Set the Eigensolution parameters.
3. Select the desired Combination Method. Then use the checkboxes to indicate which directions you want to
perform your response spectra analysis.
4. Select the spectra to be used for each direction. Then specify the other parameters. For help on an item, click
and then click the item.

Note

 For a more thorough explanation of the Eigensolution options refer to Dynamic Analysis - Eigensolution.
 Upon the completion of the solution you are returned to the Frequencies and Participation spreadsheet
and the participation yielded by the RSA is listed. To view model results such as
forces/deflections/reactions you will need to create a load combination on the Load Combination
spreadsheet that includes the spectra results. See below.
 The Dynamic Solver option has been moved to the Global Parameters - Solution tab Advanced options.

To Include Response Spectra Analysis Results in a Load Combination

1. After running the response spectra analysis go to the desired combination on the Load Combination
spreadsheet.

You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
mk:@MSITStore:C:\RISA\risa3dw.chm::/3D_2D_Only_Topics/Dynamic%20Analysis%2... 05-Feb-15
Dynamic Analysis - Response Spectra Page 3 of 11

2. In the BLC column enter "SX", "SY", or "SZ" as the BLC entry (SX for the X direction RSA results, SY for the Y
direction RSA results, etc.).
3. To scale the spectral results enter the spectra-scaling factor in the Factor column.

Note

 You can include more than one spectra solution in a single load combination. If you do this you can also
have RISA-3D combine the multiple RSA results using an SRSS summation. To do this, set the "RSA SRSS"
flag for the combination to "+" or “-”. Use “+” if you want the summed RSA results (which will be all
positive) added to the other loads in the load combination. Use “-” if you want the summed results
subtracted.

Response Spectra
The response spectra represent the maximum response of any single degree of freedom (SDOF) system to a dynamic
base excitation. The usual application of this method is in seismic (earthquake) analysis. Earthquake time history
data is converted into a "response spectrum". With this response spectrum, it is possible to predict the maximum
response for any SDOF system. By "any SDOF system", it is meant a SDOF system with any natural frequency.
"Maximum response" means the maximum deflections, and thus, the maximum stresses for the system.

Response Spectra Analysis Procedure


In the response spectra analysis procedure, each of the model's modes is considered to be an independent SDOF
system. The maximum responses for each mode are calculated independently. These modal responses are then
combined to obtain the model's overall response to the applied spectra.
The response spectra method enjoys wide acceptance as an accurate method for predicting the response of any
structural model to any arbitrary base excitation, particularly earthquakes. Building codes require a dynamics based
procedure for some structures. The response spectra method satisfies this dynamics requirement. The response
spectra method is easier, faster and more accurate than the static procedure so there really isn't any reason to use the
static procedure.
If you wish to learn more about this method, an excellent reference is Structural Dynamics, Theory and
Computation by Dr. Mario Paz (1991, Van Nostrand Reinhold).

Frequencies Outside the Spectra


If a response spectra analysis is solved using modal frequency values that fall outside the range of the selected
spectra, RISA will extrapolate to obtain spectral values for the out-of-bounds frequency. If the modal frequency is
below the smallest defined spectral frequency, a spectral velocity will be used for the modal frequency that will result
in a constant Spectra Displacement from the smallest defined spectral frequency value. A constant spectral
displacement is used because modes in the “low” frequency range will tend to converge to the maximum ground
displacement. If the modal frequency is above the largest defined spectral frequency, a spectral velocity will be used
for the modal frequency that will result in a constant Spectra Acceleration from the largest defined spectral frequency
value. A constant spectral acceleration is used because modes in the “high” frequency range tend to converge to the
maximum ground acceleration (zero period acceleration).

You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
mk:@MSITStore:C:\RISA\risa3dw.chm::/3D_2D_Only_Topics/Dynamic%20Analysis%2... 05-Feb-15
Dynamic Analysis - Response Spectra Page 4 of 11

Mass Participation

The mass participation factors reported on the Frequencies Spreadsheet reflect how much each mode participated
in the Response Spectra Analysis solution. Remember that the RSA involves calculating separately the response for
each mode to the applied base excitation represented by the spectra. Here is where you can tell which modes are
important in which directions. Higher participation factors indicate more important modes. The participation factor
itself is the percent of the model's total dynamic mass that is deflecting in the shape described by the particular mode.
Thus, the sum of all the participation factors in a given direction can not exceed 100%.
The amount of participation for the mode may also reflect how much the mode moves in the direction of the spectra
application. For example, if the 1st mode represents movement in the global Y direction it won't participate much, if
at all, if the spectra is applied in the global X direction. You can isolate which modes are important in which directions
by examining the mass participation.
Note

 Usually for the RSA to be considered valid, the sum of the modal participation factors must equal or exceed
90%. If you do an RSA and the total participation is less than 90%, you need to return to the dynamic
solution and redo the dynamic analysis with more modes. If you are getting a lot of modes with little or no
participation see Dynamics Troubleshooting – Local Modes.
 You may also want to account for accidental torsion. See Modeling Accidental Torsion to learn how to do
this.
 Models with a large amount of mass lost into boundary conditions may have difficulty achieving 90% mass
participation. See Dynamics Modeling for more information.

Modal Combination Option


There are three choices for combining your modal results: CQC, SRSS, or Gupta. In general you will want to use either
CQC or Gupta. For models where you don’t expect much rigid response, you should use CQC. For models where the
rigid response could be important, you should use Gupta. An example of one type of model where rigid response
would be important is the analysis of shear wall structures. The SRSS method is offered in case you need to compare
results with the results from some older program that does not offer CQC or Gupta.
CQC stands for "Complete Quadratic Combination". A complete discussion of this method will not be offered here, but
if you are interested, a good reference on this method is Recommended Lateral Force Requirements and
Commentary, 1999, published by SEAOC (Structural Engineers Assoc. of Calif.). In general, the CQC is a superior
combination method because it accounts for modal coupling quite well.
The Gupta method is similar to the CQC method in that it also accounts for closely spaced modes. In addition, this
method also accounts for modal response that has “rigid content”. For structures with rigid elements, the modal
responses can have both rigid and periodic content. The rigid content from all modes is summed algebraically and
then combined via an SRSS combination with the periodic part which is combined with the CQC method. The Gupta
method is fully documented in the reference, Response Spectrum Method, by Ajaya Kumar Gupta (Published by CRC
Press, Inc., 1992).
The Gupta method defines lower ( f 1 ) and upper ( f 2 ) frequency bounds for modes containing both periodic and
rigid content. Modes that are below the lower bound are assumed to be 100% periodic. Modes that are above the
upper bound are assumed to be 100% rigid.

Unsigned (All Positive) Results


A response spectra analysis involves calculating forces and displacements for each mode individually and then

You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
mk:@MSITStore:C:\RISA\risa3dw.chm::/3D_2D_Only_Topics/Dynamic%20Analysis%2... 05-Feb-15
Dynamic Analysis - Response Spectra Page 5 of 11

combining these results. The problem is both combination methods offered (SRSS and CQC) use a summation of
squares approach that loses the sign. This means that all the results are positive, except reactions, which are all
negative as a result of the positive displacements.
Because the RSA results are unsigned you cannot directly add the results to other static loads in you model. One way
around this is to treat the RSA results as both positive and negative by manually providing the sign. Using two
combinations for each RSA result, one with a positive factor and the other with a negative factor you can capture the
maximum deflections, stresses and forces when combining with other loads. See Load Combinations with RSA Results
for an example.
The mass participation may indicate that a model is dominated by a single mode in a direction. You may base the
signs for the final combined RSA results on the signs for the RSA for this single dominant mode by checking the box
that says “Use Dominant Mode for Signage?”. When that option is selected then the Mode that that the highest mass
participation in that direction will be considered to be the dominant mode.

Other Options

Cut Off Frequency


This is the frequency used by the Gupta method to calculate the upper bound for modes having periodic and rigid
content. The “rigid frequency” is defined as “The minimum frequency at which the spectral acceleration becomes
approximately equal to the zero period acceleration (ZPA), and remains equal to the ZPA”. If nothing is entered in this
field, the last (highest) frequency in the selected response spectra will be used.

Damping Ratio
The damping ratio entered here is used in conjunction with the CQC and Gupta combination methods. This single
entry is used for all the modes included in the RSA, an accepted practice. A value of 5% is generally a good number to
use. Typical damping values are:
2% to 5% for welded steel
3% to 5% for concrete
5% to 7% for bolted steel, wood

Localized Modes
A common problem you may encounter in dynamic analysis is localized modes. These are modes where only a small
part of the model is vibrating and the rest of the model is not. Localized modes may not be obvious from looking at
the numeric mode shape results, but they can usually be spotted by animating the mode shape. See Plot Options -
Deflections to learn how to do this. If only a small part of the model is moving, this is probably a localized mode.
The problem with localized modes is that they can make it difficult to get enough mass participation in the response
spectra analysis, since these local modes don’t usually have much mass associated with them. Solving for a substantial
number of modes but getting very little or no mass participation would indicate that the modes being found are
localized modes.
If you have localized modes in your model, always try a Model Merge before you do anything else. See Model Merge
for more information. To get rid of localized modes not omitted with a merge the options are to adjust either the
mass or the stiffness of the model. For example, if your localized mode is an X-brace vibrating out of plane, you could
attach a spring (adjusting the stiffness) to the center of the X brace and restrain the brace. Alternatively you could
make the brace weightless and lump the mass at the end joints (adjusting the mass).
If you try the model merge and are still having trouble, see Dynamics Modeling for more help resolving local modes or
getting 90% mass participation for your model.

RSA Scaling Factor (Manual Scaling)


The most difficult part of the entire RSA procedure is normally calculating the scaling factor to be used when including
the RSA results in a load combination.

You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
mk:@MSITStore:C:\RISA\risa3dw.chm::/3D_2D_Only_Topics/Dynamic%20Analysis%2... 05-Feb-15
Dynamic Analysis - Response Spectra Page 6 of 11

The ASCE 7 uses a particular “shape” for it’s spectra (See Figure 11.4-1), but the parameters SDS and SD1 make it
specific to a particular site. However, the ASCE 7 imposes several requirements regarding the minimum design
values. ASCE 7-10 Section 12.9.4.1 specifies a modification factor, 0.85*V/Vt, that may be used to scale the response
spectra results to something less than or equal the base shear calculated using the static procedure (ASCE 7 Sect.
12.8).
The static base shear (V) is calculated using the equations in ASCE 7-10 Sect. 12.8.1
Note that there are limiting values for the static base shear in ASCE 7-10 equations 12.8-3 through 12.8-6.
Therefore, in order to calculate the proper scaling factor, we need to know what the unscaled RSA base shear (this is
called the Elastic Response Base Shear in the IBC) is, and we also need to calculate the value of "V" (static base shear).
The calculation of V isn't particularly difficult because the two values that present the biggest problem in this
calculation (T and W) are provided by RISA. To calculate the value of W, simply solve a load combination comprised
of the model seismic dead weight. This almost certainly will be the same load combination you used in the Dynamics
settings for the Load Combination for Mass. The vertical reaction total is your "W" value.
The T value is simply the period associated with the dominant mode for the direction of interest. For example, if
you're calculating the scaling factor for a Z-direction spectra, determine which mode gives you the highest
participation for the Z direction RSA. The period associated with that mode is your T value. Note that there are
limiting values for T, see ASCE 7-10 section 12.8.2.
Calculating the unscaled RSA base shear also is very straightforward. Just solve a load combination comprised of only
that RSA, with a factor of 1.
Example:

 Assuming we're looking at a Z direction RSA, enter “SZ” in the BLC field of the Load Combinations
spreadsheet and for the Factor enter "1". Leave all the other BLC fields blank.
 Solve the load combination and look at the Z direction reaction total. This total value is the unscaled RSA base
shear.
 To get the correct scaling factor, solve this equation:

Scale Factor = (V / Unscaled RSA base shear)


You would do this calculation to obtain the scaling factors for all the directions of interest (X, Y and/or Z). Unless the
model is symmetric the fundamental period for each direction is probably different. Be sure to use the proper value
for "T" for the direction being considered.
Note: The ASCE 7 has additional requirements for vertical seismic components. (See ASCE 7-10 section 12.4.2.2).

RSA Scaling Factor (Automatic Scaling)


The most difficult part of the entire RSA procedure is normally calculating the scaling factor to be used when including
the RSA results in a load combination. The Scaling Factor tool can be used to automatically calculate the RSA scaling
factor for each direction within your model.
The RSA scaling factors automatically default to 1.0 and can only be changed from within the Scaling Factor dialog.
The dialog may be opened by clicking the scale factor button, SF, on the Spreadsheet Toolbar. This button is only
available when the Load Combinations Spreadsheet is the active window.

Note

 Currently these factors are only calculated when you open this dialog and click the Calculate button.
Therefore, if you have previously calculated these and then made changes to your model, you will need to
come back into this dialog and re-click the Calculate buttons to update the scaling factors.

Base Shear

You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
mk:@MSITStore:C:\RISA\risa3dw.chm::/3D_2D_Only_Topics/Dynamic%20Analysis%2... 05-Feb-15
Dynamic Analysis - Response Spectra Page 7 of 11

The Base Shears section of the Spectra Scaling Factor dialog box reports to you the static base shear and the un-
scaled RSA base shear for each direction. If these values have not been calculated yet, then you may press the
Calculate button. This will launch the Seismic Load Generation window for the calculation of the Static Base Shear,
or the Dynamic Solution dialog for the calculation of the Unscaled Base Shear for your Response Spectra results.

Scaling Factor

Program Calculated

This section compares the ELF (Equivalent Lateral Force) Limit, as reduced by the Base Shear Multiplier, and the I/R
Limit. These values are explained in detail below.

 The Base Shear Multiplier is used in the calculation of the ELF Limit. This multiplier can be used to reduce the
calculated (unscaled) base shear below the base shear value obtained by the Equivalent Lateral Force Method.
For the 1997 UBC, this is covered in section 1631.5.4. For the 2000 IBC, this is covered in section 1618.7. For
the 2010 NBC, it is discussed in clause 4.1.8.12 (8). In the ASCE 7-10, it is discussed in section 12.9.4.1.
 The ELF Limit is the limiting scaling factor calculated per the Equivalent Lateral Force procedure:

 The I/R Limit simply takes the elastic dynamic base shear and reduces it to an inelastic base shear. This value
is calculated per the I (Importance Factor) and R (Response Modification Coefficient). You may edit these
values in the Seismic tab of Global Parameters.
 The Controlling Limit is the governing (larger) value of the ELF limit and I/R limit.

User Input

You may manually override the program calculated factor by selecting the User Input option and entering in an
alternate value.

Apply Checkbox

You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
mk:@MSITStore:C:\RISA\risa3dw.chm::/3D_2D_Only_Topics/Dynamic%20Analysis%2... 05-Feb-15
Dynamic Analysis - Response Spectra Page 8 of 11

The Apply SF to all RSA Load Combinations checkbox will apply the SF scale factors to all load combinations that
currently reference response spectra results. The final Load combinations will appear similar to those shown below:

Automatic Response Spectra Generation

1997 UBC
You may have the 1997 UBC spectra generated automatically by selecting "UBC 97, Parametric Design Spectra" for
your RSA. The Ca and Cv seismic coefficients are needed to calculate the values for the UBC ’97 spectra. See Figure 16-
3 in the UBC for the equations used to build the spectra. See Tables 16-Q and 16-R to obtain the Ca and Cv values. The
default values listed are for Seismic Zone 3, Soil Type “Se” (Soft Soil Profile). These values can be edited in the Seismic
tab of Global Parameters.

2000 IBC
You may have the 2000 IBC spectra generated automatically by selecting "IBC 2000, Parametric Design Spectra" for
your RSA. The SDS and SD1 seismic coefficients are needed to calculate the values for the IBC 2000 spectra. See Figure
1615.1.4 in the IBC for the equations used to build the spectra. See section 1615.1.3 to obtain the SDS and SD1
values.These values can be edited in the Seismic tab of Global Parameters.

2005 ASCE
You may have the 2005 ASCE spectra generated automatically by selecting "ASCE 2005, Parametric Design Spectra"
for your RSA. The S , S , and TL seismic coefficients are needed to calculate the values for the ASCE 2005 spectra.
DS D1
See Figure 11.4-1 in ASCE-7 2005 for the equations used to build the spectra. See section 11.4.4 to obtain the S and
DS
S values and Figures 22-15 thru 22-20 for the TL value. These values can be edited in the Seismic tab of Global
D1
Parameters.

2010 ASCE
You may have the 2010 ASCE spectra generated automatically by selecting "ASCE 2010, Parametric Design Spectra"
for your RSA. The S , S , and TL seismic coefficients are needed to calculate the values for the ASCE 2010 spectra.
DS D1
See Figure 11.4-1 in ASCE-7 2010 for the equations used to build the spectra. See section 11.4.4 to obtain the S and
DS
S values and Figures 22-12 thru 22-16 for the TL value. These values can be edited in the Seismic tab of Global
D1
Parameters.

2005 NBC
You may have the 2005 NBC spectra generated automatically by selecting "NBC 2005 Parametric Design Spectra" for
your RSA. The Site Class and the S values are needed to calculate the values for the NBC 2005 spectra. Please see
a
section 4.1.8.4(7) to obtain the S values and Table 4.1.8.4(A) for the Site Class. Please see section 4.1.8.4(7) for the
a
equations used to build the spectra.These values can be edited in the Seismic tab of Global Parameters.

2010 NBC

You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
mk:@MSITStore:C:\RISA\risa3dw.chm::/3D_2D_Only_Topics/Dynamic%20Analysis%2... 05-Feb-15
Dynamic Analysis - Response Spectra Page 9 of 11

You may have the 2010 NBC spectra generated automatically by selecting "NBC 2010 Parametric Design Spectra" for
your RSA. The Site Class and the S values are needed to calculate the values for the NBC 2010 spectra. Please see
a
section 4.1.8.4(7) to obtain the S values and Table 4.1.8.4(.A) for the Site Class. Please see section 4.1.8.4.(7) for the
a
equations used to build the spectra.These values can be edited in the Seismic tab of Global Parameters.

Adding and Editing Spectra


You may add your own spectra to the database and edit and delete them once they are created. You can add/edit
spectra data pairs in any configuration by choosing between Frequency or Period and between the three spectral
values. You may also choose to convert the configuration during editing. At least two data points must be defined.
Log interpolation is used to calculate spectra values that fall between entered points. Make sure that all of the modal
frequencies in your model are included within your spectra.

To Add or Edit a Spectra

1. On the Modify menu choose Response Spectra Library.


2. Select Add or Edit.
3. Select the format to be used and specify the parameters. For help on an item, click and then click the item.

Note

 Zero values are not allowed in the data.


 The spectra data is not currently stored with the RISA model file. Instead it is stored in the
RSPECT32.FIL database file located in the directory set using Tools - Preference - File Locations. If a file
with a custom spectra needs to be transferred to another computer, then this file must also be transferred
to the new computer.

Tripartite Response Spectra Plot


This plot is a convenient logarithmic representation of all the values of interest in the response spectra definition.

You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
mk:@MSITStore:C:\RISA\risa3dw.chm::/3D_2D_Only_Topics/Dynamic%20Analysis%2... 05-Feb-15
Dynamic Analysis - Response Spectra Page 10 of 11

These values are as follows:


Frequency (f)
Period (T)
Pseudo Velocity (Sv)
Pseudo Acceleration (Sa)
Pseudo Displacement (Sd)
The relationships between these values (for the undamped case) is as follows:
T = 1. / f
Sv = Sd * 2πf = Sa / 2πf
For the tripartite plot, the frequency values are plotted along the bottom with the reciprocal period values displayed
along the top. The ordinate axis plots the Sv values (labeled on the left side) and the diagonal axes plot the Sa (lower
left to upper right) and Sd (upper left to lower right) values.
The spectra data itself is represented with the thick red line. Therefore, to determine the Sv, Sa or Sd value for a
particular frequency or period, locate the desired period or frequency value along the abscissa axis and locate the
corresponding point on the spectra line. Use this point to read off the Sv, Sa and Sd values from their respective axes.
Remember, all the axes are logarithmic!

Single Spectra Plot


This plot is a Response spectra representation that is given in the building codes. The vertical axis can plot the spectra
using Pseudo-Acceleration, Pseudo-Velocity, or Pseudo Displacement on a vertical or logarithmic scale. The
horizontal axis will plot the Period or Frequency using a Linear or Logarithmic scale.

You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
mk:@MSITStore:C:\RISA\risa3dw.chm::/3D_2D_Only_Topics/Dynamic%20Analysis%2... 05-Feb-15
Dynamic Analysis - Response Spectra Page 11 of 11

These values are as follows:


Frequency (f)
Period (T)
Pseudo Velocity (Sv)
Pseudo Acceleration (Sa)
Pseudo Displacement (Sd)
The relationships between these values (for the undamped case) is as follows:
T = 1. / f
Sv = Sd * 2πf = Sa / 2πf
The spectra data itself is represented with the thick red line. Therefore, to determine the Sv, Sa or Sd value for a
particular frequency or period, locate the desired period or frequency value along the horizontal axis and locate the
corresponding point on the spectra line. Use this point to read off the Sv, Sa and Sd values from their vertical axis.

You created this PDF from an application that is not licensed to print to novaPDF printer (http://www.novapdf.com)
mk:@MSITStore:C:\RISA\risa3dw.chm::/3D_2D_Only_Topics/Dynamic%20Analysis%2... 05-Feb-15