Viscoelastic material curve fitting determines your material constants by relating your experimental data to the Prony series expansion for both the shear and bulk modulus of the hypoviscoelastic material option.
To perform curve fitting, you input the experimental data, define the order of Prony series expansion, perform nonlinear regression, view the curve-fitting results graphically, compare the experimental data, and write the fitted coefficients to the database as nonlinear data table commands for the subsequent finite elements analyses. You can fit shear modulus and/or bulk modulus and/or shift functions, along with discrete temperature dependencies for multiple data sets.
The steps for viscoelastic curve fitting are defined as follows:
| 1 | Step 1. Prepare Experimental Data | The experimental data must be a plain text file delimited by a space or a comma. |
| 2 | Step 2. Input the Data | Read in the experimental data as a plain text file. |
| 3 | Step 3. Select a Material Model Option | This includes Prony series expansion of shear and/or bulk moduli as well as shift function. The supported shift functions include WLF and TN. |
| 4 | Step 4. Initialize the Coefficients | Viscoelastic curve fitting is a nonlinear regression; the initial value of your coefficients is very important for a successful solution. |
| 5 | Step 5. Specify Control Parameters and Solve | Specify the error norm to be used, the solution control parameters, and perform the nonlinear regression. |
| 6 | Step 6. Plot the Experimental Data and Analyze | Graphically view the curve-fitting results. Review and verify the results by comparing them with the experimental data and the regression errors. If any factor is not acceptable, repeat steps 3 to 7 to obtain a new curve-fitting solution. |
| 7 | Step 7. Write Data to the TB Command | Write your curve-fitting results in TB command format to the database. |
Viscoelastic curve fitting allows for experimental data in both the time and frequency domains.
To input experimental data for viscoelastic materials in the time domain, the data
must be shear modulus and/or bulk modulus as a function of time and temperature. The
experimental data is named sdec or
bdec to distinguish it from other data types, such as
hyperelastic or creep, and to allow convenient analysis documentation.
To input experimental data for viscoelastic materials in the frequency domain, the
data must be storage modulus and loss modulus as a function of frequency, and
optionally temperature. The experimental data is named sdec
or bdec. The data must be represented in three columns, and a
/1,freq entry at the top of the experimental data file is
necessary to indicate that column 1 is frequency. Similar to time domain data, it is
possible to input data for shear (experiment type sdec) and
bulk behavior (experiment type bdec). Frequency values are
entered in cycles-per-second (which the program converts to
radians-per-unit-time).
Your viscoelastic test data must be a plain text file with headers to define the test data. The data file should be in table form, delimited by a space or a comma. The header defines the test data type and the temperature for your test data. For viscoelastic curve fitting with multiple temperatures, you can evaluate coefficients at each discrete temperature point and write it as a temperature-dependent Prony data table, or you can use the Williams-Landau-Ferry (WLF) or Tool-Narayanaswamy (TN) shift functions to account for the temperature dependency. (See Shift Functions in the Mechanical APDL Theory Reference.) A separate data file must be provided for each discrete temperature. The viscoelastic test data can be any of 4 data types, see Table 7.4: Viscoelastic Data Types and Abbreviations.
Table 7.4: Viscoelastic Data Types and Abbreviations
| Time | time |
| Shear modulus | smod |
| Bulk modulus | bmod |
| Temperature | temp |
The headers are used to describe the data types that characterize the test data columns or attributes of the data.
The following listing contains the appropriate headers, followed by the delimited data:
/temp,100 ! define temperature attribute 0.01 2992.53 1 2978.514207 2 2965.45541 4 2942.293214 6 2922.530649 8 2905.612202 10 2891.073456 20 2842.506984 40 2798.142793 60 2772.383729 80 2750.631843 100 2730.398114 200 2643.125432 400 2517.475394 600 2431.262053 800 2366.580897 1000 2313.955396 2000 2117.922594 4000 1833.734397 6000 1627.199197 8000 1470.6806 10000 1347.264527 20000 964.0141125 40000 586.1405449 60000 392.186777 80000 277.2706253 100000 202.0025278 200000 46.87056342 400000 2.669209118 600000 0.156653269 800000 0.0137224 1000000 0.005591539
You use the EADD option of the TBFT command to input your data files. The experimental data must be read in from a plain text file. The experimental data must be prepared as discussed in the previous section, and include both the header information and the formatted test data. Each file is viewed as a data set, and can be the complete set of experimental test data or a part of an experimental test data.
You can include several data sets, including tests performed at different temperatures. Although different data sets can have the same/or different temperature, each file can have only one temperature. Multiple temperature data sets must be input with multiple files.
Two types of data may be required for viscoelastic curve fitting, either shear modulus vs. time and/or bulk modulus vs. time. The data can also be a function of temperature, which can then be accounted for by either WLF or TN shift function.
TBFT,EADD,ID,Option1,Option2,Option3,Option4 ! input data
where:
ID = index corresponding to the material number |
Option1 = experimental data type, either sdec or
bdec |
Option2 = name of file containing experimental
data |
Option3 = file name extension |
Option4 = file directory |
The sdec coefficient refers to the shear modulus as a function of time. The bdec coefficient refers to the bulk modulus as a function of time.
The TBFT command provides the curve-fitting tools for viscoelastic material modeling. You represent your viscoelastic material behavior by a set of Prony-series expansions of shear and/or bulk moduli to characterize the shear and the bulk deformation of the material. You can also use the shift functions to characterize the material's temperature dependency.
First you define a case name to associate the set of coefficients for the Prony expansions with the shift functions that characterize the material behavior. You can use the case name to define several different options that characterize the same test data, and then to compare the curve-fitting results. To define the material model, you must first define a case name, and then specify the order of shear and bulk moduli and the type of the shift function(s), if required. You need to create an additional case to define different shear order, bulk order or shift options. Once you create a case, the number of shear terms, bulk terms, or shift options cannot be changed.
You define a viscoelastic material with the Prony series expansion by creating
a case and setting the order of shear modulus, bulk modulus, and shift options.
You create the case with the TBFT,FCASE command. The first
line will include FCASE,ID,NEW. Then you specify the
number of shear terms, bulk terms, and the shift function. The case is actually
created only after the option is issued. The following syntax examples and
argument descriptions illustrate the relationship of these activities:
TBFT,FCASE,ID,NEW,Option2,Option3
where:
ID = Index corresponding to the material number |
Option2 = PVHE (refers to Prony
Viscohypoelastic) |
Option3 = Case name |
TBFT,FADD,ID,CATEGORY,Option2,Option3
where:
ID = Index corresponding to the material number |
CATEGORY = VISCO |
Option2 = pshear or pbulk or shift |
Option3 = dependent on
Option2 as follows: |
Option2 = pshear or bulk, Option3 = NONE, or
1 to N
Option2 = shift, Option3 = NONE or TN or
WLF
TBFT,FCASE,ID,FINI
The initial values chosen for your coefficients determine the success of your viscoelastic curve-fitting operations. A complete model has (2*NG+1) + (2* Nk +1) + NS number of the coefficients, where:
NG is the order of the Prony series expansion of the shear modulus,
Nk is the order of the Prony series expansion of the bulk modulus, and
NS is the number of coefficients of the shift function. (NS = 2 for the TN option, and NS = 3 for the WLF option.)
The coefficients are ordered as shear terms first, then the bulk terms, and then the shift function. The coefficients are ordered as A0 G, A1 G, τ1 G, A2 G, τ2 G, … An G, and τn G for shear modulus, and A0 K, A1 K, τ1 K, A2 K, τ2 K, … An K, and τn K for bulk modulus.
A shift function must be used together with your shear and/or bulk modulus for temperature-dependent experimental data.
The default coefficient is set to 1, but it is good practice to redefine the initial values before solving. When initializing your coefficients, set An Ks to 1 and τn Ks to time values that are equally distributed in the log scale, spanning the data range from minimum to maximum time for the time domain data.
For example, consider the shear-decay versus time-data file. If the time values vary from 1 to 10000, and if you use third-order Prony, logical guesses for τ1 G, τ2 G and τ3 G that span this range could be τ1 G = 1, τ2 G = 100, and τ3 G = 10,000 (also (1), (10) and (10,000) or (1), (1,000) and (10,000), respectively).
For the frequency domain data, a trial-and-error process is necessary. Enter τ values at different orders of magnitude (1E-5,1E-3,1E-1,1E1,1E2. ...). The A (alpha values) values can be set to 1, and the least-squares solver generally converges to the best possible alpha values irrespective of the initial values.
The following equations describe the relationship between the Prony Coefficient (αn) and the corresponding coefficient generated in curve fitting (An). K and G are the shear modulus and bulk modulus at t = 0. This was done to keep all αn G and αn K values used in the TB tables positive.
αn K = (An K * An K) / K
αn G = (An G * An G) / G
A good guess for the WLF or TN parameter is the reference temperature you used
during your partial solve for shear and bulk. The index of the reference/base
temperatures is the sum of NumShear +
NumBulk + 1.
You can also fix (hold constant) your coefficients. You specify a value for a coefficient and keep it unchanged, while allowing the other coefficients to be operated on. You can then release the fixed coefficient later if desired. By default, all of the coefficients are free to vary.
You estimate coefficients for temperature-dependent data either by using the shift function or by setting the temperature-dependency flag and setting a reference temperature before solving for the coefficients. You can set the reference temperature only to values specified using the /temp,value header line in the experimental data.
You can also specify tref = all and initiate multiple solves to evaluate coefficients at all available discrete temperature values. In this case, for data at three temperatures (t1, t2, and t3), a single TBFT,SOLVE command initiates three separate solve operations at those three discrete temperature values, and generate data at three corresponding discrete temperatures.
With temperature dependency specified and the reference temperature set to a particular value, a TBFT,SOLVE command solves for coefficients only at that discrete temperature. To solve for coefficients at other temperatures, set the reference temperature to each of the desired discrete temperature values and solve again.
You can initialize the coefficients before or after activating temperature dependency. If the coefficients initialize before setting temperature dependency, the specified coefficients become the initial coefficients for all future solves for that particular model. These coefficients are, however, overridden when temperature dependency is active and another set of values is specified at a discrete temperature value. The curve-fitting tool looks for the initial coefficients at a particular temperature. If no coefficients are specified at discrete temperature values, the initial coefficients set before temperature dependency was activated are used.
TBFT,SET,ID,CASE,Option2,Option3,Option4,Option5
where:
ID = Index corresponding to the material number |
Option2 = Case name |
Option3 = (Blank--not applicable) |
Option4 = Index of coefficient |
Option5 = Value of coefficient |
TBFT,SET,1, myvisco1,,1,1.2 ! Initialize the first coefficient to 1.2 TBFT,SET,1, myvisco1,,2,1.5 ! Initialize the second coefficient to 1.5
Use the TBFT,FIX command to fix a coefficient to a value set by the TBFT, SET command or to release a previously fixed coefficient. By default, coefficients are not fixed.
TBFT,FIX,ID,CASE,Option2,Option3,Option4,Option5
where:
ID = Index corresponding to the material number |
Option2 = Case name |
Option3 = (Blank--not applicable) |
Option4 = Index of coefficient |
Option5 = 1 to fix, 0 to vary |
TBFT,FIX,1, myvisco1,,1,1.! Fix the first coefficient to a value set via TBFT,SET TBFT,FIX,1, myvisco1,,2,1 ! Fix the second coefficient to a value set via TBFT,SET
Temperature dependency uses Option4, and references
data files you entered with the appropriate temp header.
TBFT,SET,ID,VISCO,Option2,Option3,Option4,Option5
where:
ID = Index corresponding to the material number |
Option2 = Model name. See Table 7.4: Viscoelastic Data Types and Abbreviations (above) for the models available. |
Option3 = (Blank--not applicable) |
Option4 = tdep or tref |
Option5 = If Option 4
= tdep, then 1 turns temperature dependency on and 0 turns it off. If
Option 4 = tref, this value will be a
specific temperature, or ALL. |
Viscoelastic curve fitting is a nonlinear regression process. You can use either normalized or non-normalized error norm for the regression. The normalized error norm is the default regression option for calculating the error. This error norm generally gives better results than the non-normalized error norm, since the normalized error gives equal weight to all data points.
The solution control parameters for nonlinear regression include number of iterations, residual tolerance, and coefficient change tolerance. The solution stops when both residual tolerance of error norm and coefficient change tolerance is met or if the number of iterations criteria is met. The coefficients are updated when the solution is completed. In general it is very difficult to directly solve a complete case including coefficients of the shear modulus, the bulk modulus and the shift function. Three solver options including shear modulus only, bulk modulus, and shift function (or all) are provided to allow you to solve only Prony coefficients of the shear modulus, Prony coefficients of the bulk modulus, and coefficients of the shift function. In many cases, however, the coefficients of shift function can't be solved until the shear or bulk modulus are solved. It is normal for a solution to not converge at first, but to stop when the maximum iterations criteria is reached. At that point, you should examine the curve-fitting results and the solution history before proceeding any further. You can then adjust parameters and resolve the problem whenever it is necessary.
For viscoelastic curve fitting using the shift function, follow these three steps to perform the regression:
Solve the shear coefficients (if there are any):
Set the partial solve option using TBFT, SET,,,,,COMP, PSHEAR.
Set the reference temperature at which your partial solution will be performed using TBFT, SET,,,,,TREF, TX. Only data at temperature TX will be used to estimate shear coefficients
Solve
Solve the bulk coefficients (if there are any):
Set the partial solve option using TBFT, SET,,,,,COMP, PBULK.
Set the reference temperature at which your partial solution will be performed using TBFT, SET,,,,,TREF, TX. The reference temperature should be the same for both shear and bulk. Only data at temperature TX will be used to estimate shear coefficients
TBFT,SOLVE
Solve the shift function (or all) coefficients:
Set the partial solve option using TBFT, SET,,,,,COMP, PVHE.
TREF is not used when solving for all parameters. All temperature data is used to estimate the coefficients.
TBFT,SOLVE
If you do not use the shift function, discrete data must be supplied to account for temperature dependency. Follow these steps to perform the regression:
Set the temperature-dependency flag using the command TBFT, SET,,,,,TDEP,1.
Solve the shear coefficients (if there are any):
Set the partial solve option using TBFT, SET,,,,,COMP, PSHEAR.
Set the reference temperature at which your partial solution will be performed using TBFT, SET,,,,,TREF, TX. Only data at temperature TX will be used to estimate shear coefficients
Initialize the coefficients.
TBFT,SOLVE
Repeat the above steps for all temperatures.
Solve the bulk coefficients (if there are any):
Set the partial solve option using TBFT, SET,,,,,COMP, PBULK.
Set the reference temperature at which your partial solution will be performed using TBFT, SET,,,,,TREF, TX. Only data at temperature TX will be used to estimate the shear coefficients.
Initialize the coefficients.
TBFT,SOLVE
Repeat the above steps for all temperatures.
Alternatively, you can solve for both shear and bulk data at the same time.
Set the temperature dependency flag using the command TBFT, SET,,,,,TDEP,1.
Solve for ALL coefficients.
Set the partial solve option using TBFT, SET,,,,,COMP, PVHE.
Set the reference temperature at which your partial solution will be performed using TBFT, SET,,,,,TREF, TX. Only data at temperature TX will be used to estimate shear and bulk coefficients
Initialize the coefficients.
TBFT,SOLVE
Repeat the above steps for all of the desired temperatures.
The solution option command is:
TBFT,SET,ID,CASE,Option2,Option3,Option4,Option5
where:
ID = Index corresponding to the material number |
Option2 = Case name |
Option3 = (Blank -- not applicable) |
Option4 = comp |
Option5 = pshea (for Shear only) or pbulk (for
bulk only) |
The SOLVE option allows you to specify procedure types, tolerances, and number of iterations.
TBFT,SOLVE,ID,CASE,Option2,Option3,Option4, ..., Option7
where:
ID = Index corresponding to the material number |
Option2 = visco function name (See Table 7.4: Viscoelastic Data Types and Abbreviations.) |
Option3 = (Blank--not applicable) |
Option4 = Curve-fitting procedure: 0 =
non-normalized least squares, 1 = normalized least squares
(default) |
Option5 = Maximum number of iterations |
Option6 = Tolerance of residual changes |
Option7 = Tolerance of coefficient
changes |
Other solving parameters are available. See the TBFT command for details.
The best way to verify a good fit between your experimental data and the provided curves is to plot your curves and visually inspect them. The button provides direct means to plot the data.
All of your data is plotted as a function of column 1 for the X-axis. Columns 2 and above are each plotted in a separate graph, in a separate window, with the corresponding fitted data as a function of column 1. By default, all the shear data sets and/or the bulk data sets as well as the corresponding fitting results are plotted in two separated graphs. To view a specific data and its corresponding fitting result, click the right mouse button on the specific data set, and pick a desired option. You can also use the right mouse button to turn the legend and/or axis on and off. The scales can be also switched between log scale and regular scale. With the middle button, you can eliminate certain curves from each window's display in order to see the remaining data more clearly.
Right mouse button functions allow you to Zoom, Fit, Save Plot to File, View/Hide Legend, and View/Hide Grid. Two or more fitted functions can also be compared in the same plot. You can also see the number of points used to generate the plot, and also change the minimum X value and the maximum X value in a plot. Use the middle-mouse button (context sensitive) to hide a particular curve within a graph.
Reviewing your curve-fitting result graphically is the only way to ensure a good fit. After plotting the curve-fitting results, you can then review multiple plots and also verify the error norm/residual value that is printed in the curve-fitting window. This information helps you determine the quality of a curve fit and decide whether to accept the results. If not, you may want to go back to step 3, solve again by changing the order of the Prony series, redefining certain initial values of the coefficients, and possibly other control parameters. You can continue to use your original data, repeating step 3 through step 6 until you are satisfied with the solution.
After a successful curve fitting, the last step is to write the curve-fitting data as a Prony data table (the TB,PRONY command) to the database.
The curve-fitting tool (TB,FSET) calculates coefficients that differ from those shown in the TB,PRONY table. The curve-fitting tool uses the following equations to automatically convert the calculated coefficients to the TB,PRONY format:
Curve-Fitting Equation in Time Domain:
Prony Equation:
α i and τi (i = 1 to N) are the Prony coefficients (entered via the TB,PRONY) command. A0 is not entered in the Prony table.
Conversion Procedure:
Curve-Fitting Equation in Frequency Domain:
G’ is the storage modulus, G’’ is the loss modulus, Ω is the frequency in radians-per-unit-time, n is the order of the Prony model. A and τ are the fitted Prony parameters.
Conversion Procedure: