FORMULA DESIGN (TASTEL+)

STATISTICS

This package analyses instrumental data (Production and physical and chemical criteria, ...), with product comparison graphics, analysis of variance, ... Statistical analysis evidences also correlation with SENSORY ANALYSIS.

Product analysis with instrumental results

Operating process management by the production data follow up with the sensory non conformity links

Evidence of the formula optimum in regard with a sensory analysis product profile target.

Analysis Preparation

The Systeme suggests to enter the product names, sample names, then the variable names.

Data Entry

Data Entry is equivalent to a spreadsheet as displayed on the previous screen, or data can be imported using various data import formats.

Calculation

Data analysis is very large and various with:

Data description, with various distribution statistics with means, standard deviation, confident intervals, …

Data structure analysis with Analysis of Variance including normalisation studies, and mean comparison tests (LSD, Newman&Keuls, Duncan…), with graphs, and thus, in complete or incomplete factors.

Multidimensional Analysis with principal Component Analysis, for example, or classifications as Ascendant Hierarchical Analysis, ...

Links between Sensory Analysis description and Instrumental Analysis are possible with joint multidimensional Analysis…

Quality Control: Production and Sensory Analysis

A joint analysis between production data and Sensory Analysis on the same samples in time analysis will provide indicators and correlations between production parameters and changes on Sensory aspects. Sensory aspects could be defined by visual, olfactory or gustatory descriptors.

This analysis will be then able to precise the main production points on which to focus in order to avoid any change on the product sensory profiles.

Production data follow up

This analysis displays watching and control limits based on the production data follow up

These limits will be useful to define the tolerance limits to accept or to reject new production batches.

You can display the various distribution indicators:

Joint Analysis: Recipes and Sensory Analysis

  A first step is to enter the recipe data of 5 to 6 or more products.

  Second, is to get a Sensory product profile for each of these products

  Then, a joint data calculation will identify correlations and links between these two types of data (Recipe in one hand, and Sensory, in the other hand).

  It will be then possible by introducing a new desired sensory product profile to know the calculated recipe of this new product.

A very simple example below can evidence links between recipe data: Amount in Hony, Sugar, fruits, chocolate and pH with Sensory descriptors as caramelised taste, fruity, bitter, sweet, or then chocolate.

The system will then model a new product: Chocolate yogurt (Yaourt chocolat)

This analysis has for aim to gather and to link Sensory Analysis and Recipe Data in order to facilitate and to speed the product development operations.

The recipe data can be extracted automatically from the other system called RECIPE developed by the Hamilton Grant Company.

MIX PLANS

The system suggests Experimental design according the mix types (I,II,III,IV) and the element number with their levels, in order to issue optimum formulation graphics.

Technical optimisation of a mix formulation in comparison of an experimental answer: stability, viscosity, texture, preference, ...

Price optimisation by a different mix of lower price elements.

Mix definition and Experimental Design

Entering the mix components, up to 4 maximum and their constraints (lower level, and upper level, in the mix analysis field), allow the definition of the mix, but also implies the mix type: . Type I, if doesn’t exist constraints on the components: From 0% to 100%, and no solvent . Type II, if some minimum levels are higher than 0%, and no solvent . Type III, if some minimum levels are higher than 0%, some maximum levels are lower than 100%, and no solvent . Type IV, if most of the mix is constituted of solvent (higher than 90 %, for example).

At the end of the component data entry, and the saving with a mix number, it is then possible to take knowledge of the suggested experimental design.

Data entry of the experimental data

The data entry of the experimental data is asked according the suggested design, in order to test the model validity (linear model, quadratic model, other...).

After validation, results are displayed with a answer surface graph and the model coefficients.

FORMULAS OPTIMISATIONS

Starting of formulation critical factors to get a ‘target’ answer: stability target, sensory product profile, or also any other objective, the system suggests experimental design. Graphics display answer surfaces in order to identify optimums, and the experimental model.

Identification of factor links for an experimental answer

Experimental model to search the best technical mix or the best price for a processus or a formulation.

This module allows to enter very easily the main factors influencing experimental results in order to suggest an experimental design, and consequently to identify the laws at the basis of the process to optimise.

Then, experimental design are automatically available to prepare experiments: A three complete factors model with linear and/or quadratic model is displayed on the previous screen.

According protocol optimisations, experiences with 2 to 15 factors can be calculated.

Example data entry with the three factors previously quoted is doing as follows:

A difference value between experimental result and calculated value is asked to test the model validity according various possible laws.

Calculation can then graph iso-answer curves according the different considered factors:

Additional results are displayed as model coefficients, for example, if this one is accepted.

COST OPTIMISATION

Based on the ingredient costs, the integration limits, and the making constraints, the System will suggest the least expensive formula meeting these points.

Cost optimisation by an optimum arrangement of ingredients

It is possible to enter ingredient names and their costs. Then, considering the integration limits of these ingredients, and according to the various product constraints, the System will suggest the best solution at the lowest price.

Starting from a simple example:

A big delicatessen company needs to develop its sausages production and to improve its raw materials:

Its R&D Department informs it is possible to put:

It is also known to not excess 27% up to 33% of total fats in the finished products to meet the marketing and legal specifications.

The question is then to know the optimum ingredient percentages to produce this sausage at the best price, in the definite limits of fats

Data entry is esily done as follows:

Results are then the following ones:

Steps between the lower and the upper fats limits can be displayed in numerical results or in graphs:

So, we can observed that the minimum raw materials cost can vary from 1,24 up to 1,10 per kg, according the fats constraints, the System is giving also the percentages of each ingredient.

The System can optimise, with sophisticated techniques called Simplex method, raw material costs, with changes in percentages or by substitution of ingredients.