The RESP command generates a displacement, velocity, or acceleration response spectrum from a given displacement or acceleration time-history. The response spectrum can then be specified in a single-point or multiple point spectrum analysis (ANTYPE,SPECTR with SPOPT,SRSS or MPRS) to calculate the overall response of a structure. For theory details about the response spectrum calculation, see POST26 - Response Spectrum Generator (RESP) in the Mechanical APDL Theory Reference.

RESP requires two previously defined variables: one containing frequency values for the response spectrum (field LFTAB) and the other containing the time-history (field LDTAB). The frequency values in LFTAB represent not only the abscissa of the response spectrum curve, but also the frequencies of the one-degree-of-freedom oscillators used to generate the response spectrum. You can create the LFTAB variable using either the FILLDATA command or the DATA command.

The displacement time-history values in LDTAB result from a full or mode-superposition transient dynamic analysis. You can create the LDTAB variable via the DATA command (if the time-history is on a file) or the NSOL command. A numerical time-integration scheme is used to calculate the response spectrum.

An example of response spectrum generation based on an acceleration time-history following a full or mode-superposition transient analysis follows. Note that the expansion pass is required to obtain the acceleration time-history in a mode-superposition analysis.

/post26

my_node = node(0,0,10)
nsol,2,my_node,ACC,Z,az10   !Define variable 2 as acceleration along Z at my_node
plvar,2

filldata,3,,,,0,0.25        !Define variable 3 as values from 0 with 0.25 increment (default end value)

resp,4,3,2,3,0.03,,,,1      !Acceleration response spectrum generation (specType=3) is stored in 
                            !variable 4 based on frequency values in variable 3 and acceleration 
                            !time-history (inputType=1) in variable 2 
                                  
plvar,4
finish

An example of response spectrum generation based on a displacement time-history following a mode-superposition transient analysis follows. Note that the expansion pass is not needed in this case.

/post26

file,,rdsp                  !Time-history is on the reduced displacement (RDSP) file

my_node = node(0,0,10)
nsol,2,my_node,U,Y,uy10     !Define variable 2 as displacement along Y at my_node
plvar,2

filldat,3,,,,0, 0.25        !Define variable 3 as values from 0 with 0.25 increment (default end value)

resp,4,3,2,1                !Displacement response spectrum (specType=1)is stored in variable 4 based  
                            !on frequency values in variable 3 and displacement time-history (default 
                            !inputType) in variable 2
                          
plvar,4 
finish

If you are working with noisy results data, you may want to smooth the response. Smoothing may allow for better understanding or visualization of the response by smoothing out local fluctuations while preserving the global characteristics of the response. The time-history smooth operation allows fitting a n^th order polynomial to the actual response.

This operation can be used only on static or transient results. Complex data cannot be fitted.

Four arrays are required for smoothing data. The first two contain the noisy data from the independent and the dependent variables, respectively; the second two will contain the smoothed data (after smoothing takes place) from the independent and dependent variables, respectively. You must always create the first two vectors (*DIM) and fill these vectors with the noisy data (VGET) before smoothing the data. If you are working in interactive mode, the program automatically creates the third and fourth vector, but if you are working in batch mode, you must also create these vectors (*DIM) before smoothing the data.

When the arrays have been created, issue the SMOOTH command. You can opt to smooth all or some of the data points via the command’s DATAP argument, and you can specify the fitting order for the smoothed curve via the FITPT argument. DATAP defaults to all points, and FITPT defaults to one-half of the data points. To plot the results, you can choose to plot unsmoothed, smoothed, or both sets of data.