PT102177B - METHODS OF CONTROL AND FORMULATION OF THE COLOR OF WINE MIXTURES BY SPECTROCOLORIMMETRICAL PREDICTION - Google Patents

Abstract

A quality wine must maintain the characteristics of colour, aroma and flavour over different production batches and year after year. As climatic influences vary, it is impossible to obtain a wine with constant characteristics, even though it originates from the same vine. The solution therefore consists in the blending of wines. Starting with different batches of wine produced at different harvests, a combination of these batches is produced so as to obtain the wine with the desired characteristics. One of the important characteristics to maintain in a quality wine is its colour. The invention claims a method that makes it possible to predict the colour resulting from the blending of various wines. Considering the spectral transmission factor or the colour co-ordinates desired for a wine blend, the invention also offers methods that make it possible to formulate what will be the composition of the blend, i.e. the percentage or concentration of each one of the component wines in the blend. Also included is prediction of the composition of the concentration of component wines by the isoline graph method. The methods described in detail for ternary blends may be extended to blending with more than three wines.

Description

Methods of Controlling and Formulating the Color of Wine Blends by Spegtrogolorimetric Prediction description of the invention

The invention describes methods that are based, in principle, on the verification that Beer's law for the spectral molecular absorbance of the wines considered is approximately valid.

The verification for wine i to be considered in the mixture is made based on the expression:

Τ (λχ = ε ' ^{α <λ)} ·' ^δ '(1) where α (λ) ί = absorption coefficient per unit thickness of the wine sample i. dj = thickness of the wine sample i.

T (À) i = transmittance factor of the wine sample i for the wavelength, λ.

When considering the mixture of n wines, considering Beer's law, we will have:

T (Ã) _mixture = _ε - ^δΣ - ^{α (λ)} · ° 'δί = δ (2) where: δ = thickness of the mixture sample, cj = wine concentration i.

Or equivalent:

- Ai 10 T Aj _λ Cj

T (Z) mixture = C (3) where:

Ajx = absorbance of the wine sample i for a thick sample, δ, that is, Αί _; χ = logioe α (λ) ί δ, for the wavelength, λ.

Ci = proportion of wine i in the sample composition

2-13

Being η

Σ -. = Ι i = l ^Α Μ, λ = Σω ^Α · Α ^c i where: Α _Μλ = absorbance of the sample of the wine mixture with thickness, δ.

(4) (5)

The sample of the thick mixture, δ, can be imagined, in the domain of the validity of Beer's law, as the juxtaposition of samples of the component wines, according to the scheme:

| <- δ; = δ. Ç;

Deviations from Beer's law can be observed at the level of each component wine, by:

[Α (λ \ cJ * 0 (6) or when the mixture occurs due to interfering effects of physical-chemical origin.

In these cases, it will be necessary to know the values of Α; χ as a function of the value of c; for applying corrections, described below.

The invention concerns the direct method, being the method of predicting the color of a mixture, considering that n components i = 1,2, ... n, are combined in the relative concentrations Cj Cj = 1, 0 <c, <1 vj , which corresponds to the absorbance values α (λ) „where Αί (λ) = logioe α (λ) ί δ.

Admitting:

^Α Μ.λ = Σ · =, ^A .A ^C , (7)

3-13 and that the three-dimensional color coordinates are given by:

X = KEjS _x x, TA) (8) (9)

Z ^ & zJA) (10)

The normalization factor K being:

κ = ιοο / (Σ ^ Ϋ, Δΰ (11) where: Sx = spectrum of the illuminant used χ _λ , γ _λ , ζ _χ = the color matching functions, for the standard observer considered

Δχ = the sampling interval

Τχ = sample transmittance for the wavelength, λ. the relationship between transmittance and absorbance given by

Τ _λ = 10 ' ^Αλ or Ax = -logTx (12)

The three-coordinate coordinates of the mixture will be given by:

10- ^to -) (13) and similarly for Ym and Zm.

Considering the transmittance spectrum for the δ thickness of wine i, a component of the mixture given by:

Tía (14) we will have to:

Τ _ΜΛ = Κ · -¾. (15) which indicates the method for obtaining the three-coordinate coordinates of the mixture, from the Tma transmittance, equivalent for the wine mixture.

If Beer's law is not verified, it is a better approximation to replace the lfi values with those obtained from the measurement of a sample of wine i, with a thickness δ-Cj. This feature may not correct the interfering effects between component wines.

4-13

The expression (15) also allows to determine the domain of colors that can be obtained by mixing, from a set of component wines, determining the domains of Xm, Ym and by variation of the c. It is even possible to create abacuses in which one of the concentrations is fixed, varying the others, creating families of concentration isolines. This will be particularly useful in ternary mixtures, which will allow to obtain the solution of the inverse problem, that is, the determination of the concentrations of the three wines that allow, by mixing, to make a specified color, solving it graphically, using the abacus. with the isoconcentration lines. One possibility is outlined in the following diagrams for the formulation of the Mixture by Graphical Route.

*! ▼

Based on the abacuses it is possible to compose the graph:

O

5-13

From this representation, and using additional criteria, the composition of the preferred mixture can be formulated.

The invention also relates to an inverse color formulation method, which consists of determining the concentrations of the component wine blend, which allow a specified color to be made for the blend.

The case of the mixture of three wines is described, and later the generalization to more than three wines is dealt with.

The tri-stimulus values X _p , Y _p , Zp, for a mixture of three wines is a function of the relative concentrations of the component wines, that is:

X _p = fi (ci, c ₂ , c ₃ )

Yp = f2 (Cl, C ₂ , C ₃ ) (16)

Zp f ₃ (ci, c ₂ , c ₃ )

Small variations in the concentrations of the components Acj, Ac ₂ , Ac ₃ , produce small variations ΔΧ _Ρ , ΔΥ _Ρ and ΔΖρ, in the values X _p , Yp, Zp, which in a first order approximation, can be translated as follows:

5X „ 5X. 5X ΔΧ: = —-Ac, + —-Ac, + —-Ac · P 5c, 5c ₂ ôc ₃ 5Y _n 5Y _d 5Y _o ΔΥ = = —-Ac, + —- Ac, + —-Ac, P 5c, õc ₂ 5c ₃ 5Z _n 5Z „5Z Δζ. = ----B.C, + —-Ac ₂ + —-Ac, P 5c, 5c ₂ õc ₃

Obtaining from the previous relationships (8), (9), (10), the values of the elements of the influence matrix:

5X - ^l = ka, Σ _λ ôc, ^{λ λ χ} λ 5th. ρ, λ ÕC, 5Y „ Γ ^9Τ ρλ1 ÕC, -Κ · Δ _λ Σ _λ Ôc, ^{λ λ s} >. Α 5Ζ _η ôt _λ Ί - ^Ρ - = Κ · Δ, Σ _λ ρ, λ õc, ^λ 5c, _

(18) ^J ogioT _iA = Τ _ρλ Λι10 log _I0 T _u (19)

6-13 of which:

5T.

T.

ρ, λ ρ, λ dc; log ₁₀ and

We can now, by inversion of the influence matrix (17), obtain the correction matrix, whose elements,,, with i = 1,2 and 3, allow to calculate the corrections Aci, Δς ₂ ^J δΧ δΥ δΖ and Δθ3, to be introduced in the current values Cj, c ₂ and C3 to determine a mixture with a color closer to the color specified as the one intended for the mixture, Xm, Ym, ZmPortanto:

look, 8X? hello, w hello, δζ z \ 'Ac, P P P í hello, Hey, hello. Ac ₂ = 2 2 J AY ^δχ ρ δ ^γ ρ ^δΖ Ρ P ΔΖ _Ρ OC ₃ OC ₃ hello ₃ \ P / 1 K ^ 7 J

(20)

The values of:

ΔΧ _Ρ - Xm - -X _p are determined from X _p , the initial concentrations, c, * 'current

ΔΥ _Ρ - Ym - Y _p ΔΖ _Ρ - Zm —Z _p

Yr> Zp, which resulted from the initial hypothesis formulated for initial.

The values to be taken for the new interaction will be as follows:

c, = Cj + Ac, i = 1, 2 and 3 'new link' eonent ¹ where the Ac, were determined from (20).

(21)

If oy ¹ Ac, = Δ presents Δ 0, then Cj are corrected according to: i = l

Cj = ^c i ⁱ new iteration 'conciante i = 1,2,3 + Δϋ; ~ (22)

7-13

8-13

The iterative process continues until the values (X _p , Y _p and Z _p ) differ from the desired values (X _M , Y _M , Z _M ) for the mixing of wines of the permitted discrepancy (in general, less than a CIELAB unit of difference in color AE * _b ).

The values obtained from the concentrations, satisfying the color matching condition, can now be used for an experimental test. In the presence of discrepancies, resulting from non-linearities, it will be necessary to correct the transmittance values of the component wines, according to the concentrations.

It may happen that there is no convergence to a solution. In this case, you should start by checking if the desired color for the mixture is possible with the component wines. If not, you will have to change component wines or add more component wines. As the color of the mixture is possible, it will be necessary to resume the iterative process, starting with a different initial mixture composition, since if the distance between the initial mixture and the one that performs the desired mixture color is large, the iterative process may not converge for a solution, that is, to present negative values of the concentrations, or a sum exceeding the unit.

Note that:

ci + c ₂ + C3 = 1 (23) and

Ac, + Ac ₂ + Ac3 = 0 (24) so these conditions can be introduced, for example, by doing:

c ₃ = 1 - ci - c ₂ (25)

Ac ₃ = - Ac, - Ac ₂ (26)

If the spectral transmittance of the desired mixture is known, as is the case of intending to reproduce a reference wine, then it is possible to find an initial combination of concentrations of the component wines, more appropriate as an initial combination in the iterative process.

The non-linear relations that determine the three-coordinate coordinates:

X = / l (Cl, C2, C ₃ )

Y = / 2 (Cl, c ₂ , c ₃ )

Z = / ₃ (Ci, c ₂ , c ₃ ) (27)

9-13 can be written in matrix form, assuming:

X ' ^λ 38Ο · ·· ^X 780 í Ci v = Y T = y 380 · • · Υ 780 c = ^C 2 k Z J k ^Z 380 · • ^Z 780 k c ₃

'S _38O 0 . . . 0> 0 ί τ Ί ^A M, 380 β τ Ί ^A p, 38O If , θ ... θ S ₇₈₀ , M = T ¹ M, 78O) P = t < ^A P, 78O) (28)

where: x, y, z = are the values of the color matching functions for the illuminant and standard observer

M = the spectral transmittance values for the blend of reference wines P = the transmittance values for the combination of component wines determined during the iterative process C = values of the relative concentrations of the wines in the blend sub-indices 380, etc. represent the length of nm wave.

In the conditions of perfect spectral agreement it would be verified:

TSP = TSM

Therefore:

TS (M-P) = 0

The condition to be observed:

M = P which translates the expression (7) in the ideal case, being for the case of ternary mixture ^ Μ, λ ⁼ Σί = 1 ^C i (29) (30) (31) (32)

In practice, the best that can be done is to look for the best fit, in the sense of least squares, which can be expressed in matrix form.

10-13

Considering the matrices:

(ya ² «= Σ _λ Α _1Λ

Σχ Α, _χ Α _{3 χ}

Σχ Α _1λ Α ₂ _ _χ Σχ Α ^, _λ Σχ Α _{2 λ} Α _{3 λ}

Σ _λ Α _1λ Α, _λ

Σχ Α _{2> λ} Α ₃ , _λ 2

Σ _λ Α:

Σχ Α _Μλ Α _1χ Σχ Α ^

Ç Σχ Α _Μλ A _{: j)} y (33)

The initial concentrations are those resulting from the solution of the linear system of equations:

(34) where (MC ¹ is the inverse matrix of M.

The method can obviously be generalized for more than three wines that make up the blend. The method presents the limitation of the need to know the spectral transmittance of the desired wine mixture. This does not become a problem when it is precisely intended to reproduce the color of a reference wine. Also, when working with wines of a given type, it is possible, statistically, to establish methods of spectral reconstruction based on parameters, which may be the color coordinates. A transmittance spectrum will then be reconstructed, which within the statistical type of wines, represents a probable transmittance spectrum for the mixture, which can be used in determining the initial composition of the concentrations of the mixture, to serve the iterative procedure for determining the composition of the wines. concentrations to be adopted.

As mentioned, ci + c ₂ + c ₃ = 1, so it may be necessary to normalize the solution found, adopting the values C ':

C '= - (35)

Cj + + ^ 3 which is equivalent to reducing the physical thickness of the wine sample.

The methods presented are likely to be generalized to mixtures with more than three wines, and mixtures other than quaternary are generally not used.

11-13

The solutions obtained in the iterative process for concentrations should always be positive. Negative solutions imply that there may not be a solution, because the color sought is out of the possible colors for the possible mixtures with the selected component wines, or that the initial concentration combination adopted for the iterative process is very far from the solution combination, at least that another initial combination should be tested. Significant color discrepancies in the order of 3 CIELAB units are considered.

The practical test to perform the mixture may lead to a different color from that predicted, due to interfering effects. Particular importance will be given to the interfering effects, the combination of wines that differ in appreciable value in pH and in the content of sugars. It is therefore recommended that the combination of the mixture use wines with close pH values and sugar content, whenever possible.

The iterative calculation processes can be performed based on the chromatic coordinates L * a * b *, with an advantage in relation to the effects of non-linearity.

The importance of non-linearities in the relationship between absorbance and concentrations implies that non-linear modeling is used. This will be the case for the use of the CIE 1976 L * a * b * color space. In this domain, the relations to be written will be:

L _m = g! (CpCjjCj) ^a * M = g2 ( ^C P ^C 2, ^C 3) (36) h ' _M = g ₃ (C „c ₂ , c ₃ )

Also, small variations in the concentrations of the component wines, Ac3, will give rise to variations in color coordinates:

AL, (Tm dc, da dc,

Ac, + ^5L m, ^ÔL M dc,

-Ac, + dc,

B.C,

Aa _M = ^ - Ac, + ^ - Ac ₂ + ^ - Ac ₃ dc ₂ dc ₃ dc.

dc, (37)

12-13

The values of the influence matrix are now obtained by the relationships:

sl ' _m <5L ' _m <5X _m 3Ym . sl ' _m sz _M dCj A.D,. dCj SZ _M A.D, SXm 5Y _m ! ^ ^a M az _M 5c, 5X _m A.D; tfY _M d _Ci dCj sx _M , ÕbM dY _M 3Z _m dx _M dCj A.D, Hey

(38)

The values of e are given by the relations (18) and (19). The values of dc, ctej dc, derived from L *, a * and b * are obtained from the relationships (36):

with: Y / Y _n > 0.008856

L '= 116 (Y / Y _n ) ^1/3 -16

Y / Y „<0.008856 L * = 903.3 (Y / Y _n ) a * = 500 | / (X / X _n ) ^- / (Y / Yn)] b * = 200 [/ '(Y / Y _n ) - / (Z / Zn)] / (X / X _n ) = (X / X _n ) Z (Y / Y „) = (Y / Y„) / (Z / Z „) = (Z / Z _n )

1/3

1/3

1/3 / (X / X „) = 7.787 (X / X„) +16/116 / (Y / Yn) = 7.787 (Y / Y „) +10/116 / (Z / Z„) = 7.787 ( Z / Zn) + 16/116 (39) where:

with: Y / Y „> 0.008856 gL ' _M Π6 f Y dY .. 3 Y„ Y

Y / Y „<0.008856

5Y ,,

903.3

Y.

ôa ‘500 ί X δΧ.

^5a M δΥ ,,

3XJX

500 ί Y 3ΫΓ Ϋ δΧ.

gives*

X „

7,787 δΥ _Μ Y „ ^gb M dY„,

200 3 Y ..

(ν ') δ1> Μ 7,787 δΥ „jK.

_ki az

200 f Z 3Z „Z„ ^ab M δΖ „

Y „

7,787 (40)

13-13

Similarly, the correction matrix with elements of the shape with

ÔL * ôa ‘ôb * i = 1,2,3, can be obtained by matrix inversion. We have thus:

(R-. C- λ <ÚC, Ac,

V AC ₃ J

Hey Hey ÔC | X X OC ₂ hello ₂ hello ₂ δζΓ ^g to ' _M 8b ' _M hello ₃ hello ₃ dc ₃ X 6a ' _M ôbT

(AT *

Áa _M

X / (41) where:

ΔΤ _Μ —Tr * T _m

Aa _M —a _R -a _M

Ab ^ = b ^ -b ^ (42)

The flow diagram for the iterative process assumes an identical format, replacing only the color coordinates to be considered.

This invention is of interest to everyone when dealing with wine blends and they intend to control the color of the blend from the component wines, or to formulate mix compositions for the component wines, which make it possible to make a specific color for the blend.

Any modification or generalization without an inventive aspect, such as an increase in the number of component wines, variations in detail, combination of elements or inoperative operations in terms of operating principles of the method described is included in the protection framework of the present invention.

Lisbon,

Ho de

ARTUR FURTADO J

MANOATÂRIO IN PROP. INOOSTRM {». da Madalena, 214 - 1 I00 LiStiC-) Tal. 887 0 / -, 57 - Rax 887 97 M {

Claims (3)

Methods of controlling and formulating the color of wine mixtures by spectrocolorimetric prediction, which are based, in principle, on the verification that the Beer law for the molecular absorbance of the wines considered is approximately valid, characterized by presenting a direct method of color prediction a mixture of wines from the transmission spectra of the wines that make up the mixture; Methods for controlling and formulating the color of wine blends by spectrocolorimetric prediction, as claimed in claim 1, characterized by a graphical method of predicting the composition of concentrations of the wines that make up a blend, to obtain a specified color for blending the component wines, essentially applicable to ternary mixtures; Methods of controlling and formulating the color of wine mixtures by spectrocolorimetric prediction, according to claim 1, characterized by the application of an inverse method, of color formulation, in which the composition of concentrations will be determined for the component wines of the mixture, so that the wine mixture has predetermined color coordinates; Methods for controlling and formulating the color of wine mixtures by spectrocolorimetric prediction, as claimed in claim 1, characterized by the application of methods for determining the combination of concentrations of 2-2 components of wine mixtures, to be taken as an initial, for determination by iterative process, of the combination of concentrations of the component-solution wines, with the transmission spectrum of the reference wine mixture being known, or it can be reconstructed from characteristic parameters of the wine family regarding a type of wine; 5- Methods of controlling and formulating the color of wine mixtures by spectrocolorimetric prediction, according to previous claims, characterized by the general application of the methods mentioned above for mixing wines in addition to ternary mixtures.