The statistics file is a file that contains statistical data for properties used in the system. If you wish to perform a Sensitivity Analysis or a Monte Carlo Analysis, you will need a statistics file that contains data for each Property that you wish to vary in these analyses.
The file is a comma delimited file which contains one Property on each row. TRIM.FaTE needs information on which object the Property applies to, a CV and a distribution type. Below are the required columns for the file. The order of the columns is not important. TRIM.FaTE will find them if they are named correctly. You may have additional columns in the file to provide other information, but these columns must be in the file.
|Column Name||Data Type||Description|
|Property||Text||The name of the Property.|
|Chemical||Text||The Chemical for the Property, if any.|
|ObjectType||Text||The type of object in the system that this Property applies to (ie Compartment, Chemical, etc)|
|ObjectName||Text||The name of the object to which this Property applies.|
|Minimum||Numeric||The minimum allowed value for the Property.|
|Maximum||Numeric||The maximum allowed value for the Property.|
|CV||Numeric||The coefficient of variation for this Property.|
|Distribution||Text||The type of distribution that this Property has.|
|Dist_Param_1||Numeric||Optional parameter for distributions.|
|Dist_Param_2||Numeric||Optional parameter for distributions.|
|Dist_Param_3||Numeric||Optional parameter for distributions.|
When TRIM.FaTE looks for an entry in the statistics file for a specific Property, it looks at all of the lines that have that Property name in them. If a Chemical is specified, it looks at lines with that Chemical from within the lines that contain the Property name. Within these lines, it looks for the object type and name to find the most specific match. If it cannot find a specific match, it will look for a more general match. For example, if TRIM.FaTE is looking for a value for the Property BackgroundConcentration for Chromium in the Compartment SurfaceWater, it will look in lines that contain the BackgroundConcentration Property. It will then see if there are lines within this subset that contain Chromium. If it finds this, it will search these lines for the Compartment SurfaceWater. If it does not find Chromium, it will search all of the lines with the BackgroundConcentration Property and look for the Compartment SurfaceWater. If the Compartment SurfaceWater is not found, then TRIM.FaTE will return an error. It will also return an error if the Property name is not found.
The distributions that TRIM.FaTE supports are Uniform, Normal, Lognormal and Triangular. These are the only allowable values for that field. The Uniform distribution does not have a CV, so the value "na" should be placed in this column for all Uniform distributions. The CV will be calculated internally as (max - min) / ((max + min) * SQRT_3). The minimum and maximum are required for the Uniform and Triangular distributions.
There are also three extra columns where additional distribution information can be stored. One situation where this is needed is when a Property uses a lognormal distribution, but its value in the model is 0.0. The standard deviation for the lognormal distribution is calculated by multiplying the Property value (used as the mean) by the CV. If the value is 0.0, this gives a distribution with a 0 mean and a 0 standard deviation. Since a lognormal distribution cannot have a 0 mean, this would be an error. The solution is to provide a mean in column Dist_Param_1 and a standard deviation in column Dist_Param_2.