QSPR-MS:: Probabilistic modeling of the migration

These tools use Pseudo Monte Carlo techniques to assess in almost real time the effects of very large combinations of the parameters. The distributions are chosen to be realistic of real conditions of use of food contact materials.
PROBABILISTIC MODELING OF THE DESORPTION OF PACKAGING SUBSTANCES How modeling can help exposure assessment	Please read our disclaimer before using these tools. The tools and servers running them are property of INRA and can be freely used with a valid account. Apply for an account if you need one.
A typical example with the support of the CPR Project Migration

Principles

This section presents tools for a quick evaluation of food contact materials in relation with different sources of uncertainty or variability. The main parameters are replaced by continuous statistical distributions. In details, an input X is described as the product of typical value, noted X_, and a dimensionless random contribution X*. The distributions of X* are normalized so that they have a mean equal to 1 and a variance equal to sX² (sX=shape factor). The whole methodology is described in Vitrac and Hayert, 2005, AIChE, 51(4), 1080-1095. Classical modeling, as used in Diffusion1D, is reduced to the simple case of sX = 0. When X = D_P, you can use the tools available in QSPR-MS, as the classification of molecules to generate an estimate of sD_P (e.g. class of molecules). The effect of sD_P alone on the desorption kinetic is described in this example.

Migrarisk assesses the contamination of foostuffs in with contact with monolayer food contact materials.
In the current implementation sBi and sK_P are set to 0 (non 0 values are only available for research purposes).

Exporisk assess the exposure of a population or of a subpopulation
This tool is only available for restricted users (please, contact us for further information).

Dimensionless example, effect of sD_P

The figure below describes the effect of the uncertainty on the diffusion coefficient on the dimensionless desoprtion kinetic.

v* is the dimensionless concentration C_P/C_Peq where C_Peq = 1/(L_FP+1/K_FP).

Fo is the dimensionless time, with Fo=D_P×t/l_P².

The result is a distribution of v* values according to the distribution of Fo values.
Three typical percentiles (10th, 50th and 100th) are plotted.
The model assuming sD_P=0 is plotted in bold line.