A O PUED H A n t H V n f A l /KCDHU ISC
ELSEVIER A p p lied M a th em atic al M odelling 22 (1998) 4 8 5 -4 9 4
Unsteady state fugacity model by a dynamic control system
R a f a e l B ru a’*, J o s é M a r ia C a r ra s c o b, L o u r iv a l C o s ta P a r a íb a a,c
Departament de M atemática Aplicada, E T SIA , Universität Politécnica de Valência, Cami de Vera, sin, 46022 Valência, Spain
b Departament de Biotecnologia, E TSE A , Universität Politécnica de Valência, Valência, Spain
Centro Nacional de Pesquisa de Monitoramento e Avaliação de Impacto Ambiental, C N PM A, E M B R A P A , Jaguariúna, SP, Brazil
R eceived 11 N o v em b er 1997; received in revised form 16 A p ril 1998; accep ted 2 Ju n e 1998
A bstract
A co n tin u o u s tim e d y n am ic system o f an u n steady state fugacity m o d el is p resen ted . P ro p erties o f this m odel as stability are studied. In o rd e r to evaluate num erical results a discretization preserving th e stab ility an d yielding the positivity p ro p e rty o f th e m o d el is used. Finally, algorithm s to determ ine th e values o f th e fugacities, th e c o n cen tratio n s an d th e d issip atio n tim e a re given. T h e above stu d y is illu strated w ith num erical resu lts in a th ree co m p a rtm e n ta l en v iro n m en tal system . © 1998 Elsevier Science Inc. All rights reserved.
Keywords: F u g acity m odel; M u ltip h ase environm ental m odel; U n ste a d y state m odel; D y n am ic co n tro l systems; N o n n eg ativ e c o n tro l system s; F e n itro th io n
1. Introduction
The therm odynam ic concept o f fugacity was introduced in 1910 fo r Lewis [1] in o rd er to ex plain the behaviour o f the real gases with respect to th at o f ideal gases, in the study o f free energy corresponding to an expansion process, isotherm , reversible a n d infinitesim al.
The fugacity is a therm odynam ic m agnitude related to the chemical p o ten tia l and characterized by the leak trend o f a substance in a com partm ent [2,3]. T he fugacity express the chem istry ac tivity o f a substance and has been applied m ainly in th erm odynam ic problem s im plicating equilibrium am ong phases, especially in com putations encountered in chem ical separation p ro cesses such as liquid extraction, distillation and adsorption.
M athem atical m odels based on the therm odynam ic theory o f the fugacity are outlined fre quently by a linear system o f equations describing the bu lk balance o f a chem ical substance in an ecosystem constituted by com partm ents. Thus, when all fugacities are equal an d co n stan t in all com partm ents the concentrations are evaluated directly, this case corresponds to the well-known ‘Level I Fugacity M odel’ or ‘Level II Fugacity M odel’ if in ad d itio n there are reactions and advections (see [4]). ‘Level III Fugacity M odel’ supposes th a t the d istrib u tio n o f the substance is not in equilibrium and th a t each fugacity can have different values, w hich are determ ined by a
C o rre sp o n d in g a u th o r. Tel.: + 3 4 96 387 7660; fax: +34 96 387 7669: e-m ail: rb ru @ m a t.u p v .e s. 0307-904X/98/S 19.00 © 1998 E lsevier Science Inc. All rig h ts reserved.
486 R. Bru el al. / Appl. Math. Modelling 22 (1998) 485-494
linear system o f equations when there are reactions, advections, em issions and transfers o f the substance am ong com partm ents in stationary state [5,6]. In ad d ition, there is a n o th e r m odel describing the unsteady state behaviour o f a substance in the environm ent, which perm its to observe substances w hose emissions vary w ith the tim e an d to determ ine the tim e in w hich the system reaches the steady state. This last m odel, know n as ‘Level IV F ugacity M odel’, usually is described by a system o f differential equations (3) (see [7]).
In this w ork, we will present a proposal o f the m ultiphase fugacity environm ental ‘Level IV ’ m odel fo r a ecosystem constituted by n com partm ents where the fugacities change w ith the tim e in response to m emissions and they are determ ined by a continuous tim e dynam ic control system describing the to tal bu lk balance o f the substance.
Som e n o n -statio n ary m odels for the study o f the kinetic o f a substance in the environm ent have been pro p o sed for the analysis o f the concentrations changing w ith the tim e [8-10]. But, it seems th a t no n e o f them presents a m athem atical form alism th ro u g h c o n tro l theory.
T he m athem atical unsteady state fugacity m odel o f a substance in an ecosystem for continuous tim e c o n tro l system perm its to analyse the asym ptotic stability and the non-negativity o f the corresponding discrete system. M ore precisely, we will see th a t the fugacity m odel can be solved as a positive discrete tim e control system obtained by discretizing the continuous tim e control system. F u rth e r, we will prove th a t the continuous an d the discrete tim e c o n tro l systems are b o th asym ptotically stable a n d th a t the discrete system is non-negative. T he num erical algorithm s presented indicate th a t this control m odel turns o u t to be easy to com pute. Finally, we illustrate o u r m odel by a num erical exam ple studying the insecticide fen itro th io n in the fugacity m ultiphase m odel constituted by air, w ater and b o tto m sedim ent.
2. Notations and definitions
The distribution o f a small qu an tity o f a substance betw een tw o com p artm en ts denoted by the indices i an d j , respectively, u n d er co n stan t tem perature an d pressure, yields constant concen tra tio n ratio s between these two com partm ents. The p a rtitio n coefficient ky o f the substance betw een tw o com partm ents is then defined as the quotient CJCj o f the concentrations o f the substance in each one o f the com partm ents.
T he relationship betw een the fugacity an d the co n cen tratio n is given by C = Z f where C is the co n cen tratio n in m ol m “3, / i s the fugacity given in Pascal (Pa) an d the c o n sta n t o f p ro p o rtionality Z is the capacity o f fugacity in m ol m -3 P a -1. The estim ate o f the capacity o f fugacity Z, o f a substance in a com partm ent i depends on the n a tu re o f the c o m p a rtm e n t and o f the partitio n coefficient o f the substance in this com partm ent.
F o r exam ple, in the atm osphere the fugacity o f a substance is equal to the p artial pressure, the one th a t can be expressed in term s o f the concentration in the air Ca a n d by the equation o f the ideal gases given by
/a = CaR T °. (1)
where ^ = 8.314 x Pa m 3 m ol-1 is the gas constant and 7° is the absolute tem perature. By ex pression (1), the air capacity o f fugacity is Z a = 1 ART0.
T he fugacity o f a substance dissolved in w ater is ap p ro x im ated by its p a rtial v ap o u r pressure, the one th a t is p ro p o rtio n al to its concentration in w ater, th a t is,
R Bru et al. / Appl. Math. Modelling 22 (1998) 485-494 487
where H is the H en ry ’s c o n stan t in Pa m 3 m ol 1. C onsequently, the capacity o f fugacity o f w ater is Z w = \/H .
F o r the oth er com partm ents, the capacity o f fugacity is expressed as the p ro d u ct o f the ca pacity o f fugacity in w ater or in air, by using the density o f the co m p artm en t and the ad sorption coefficient [6,11]. F o r exam ple, fo r the b o tto m sedim ent the capacity o f fugacity is evaluated by expression Z s = k dp sZ w, where k d is the ad so rp tio n coefficient and ps in kg m “3 is the density o f the dry soil, consequently, the fugacity or concentration in this co m p artm en t can be calculated by equation
Cs = Ps
Zv„/s-In an unsteady state, the fugacities are functions o f tim e and they can be determ ined as a system o f o rdinary differential equations, describing the to tal bulk balance during an infinitesim al tim e interval d t. In this balance, the storage capacity o f a substance in a co m p artm en t i is given by the p ro d u ct o f its volum e Vt and by its capacity o f fugacity Z,.
The biological and chem istry reactions are supposed to be a first o rder process where the rate o f reaction is rt in m in -1 and the reaction com ponent for each co m p artm en t is r ^ Z , .
W hen there is a gradient o f fugacity betw een two co m partm ents i a n d j it results a flow o f the substance in the direction o f this gradient. The flow o f the substance is the p ro d u ct o f the dif ference (ft - f j ) by the transfer dy betw een these two com partm ents. T he coefficients o f tra n s ference dy a n d djt are equal an d positive. The differences (/j- - f ) are positive or negative depending on the direction o f the transfer determ ined by the relative values o f f and f .
T he advection in a co m partm ent i can be introduced in the m odel as a first o rder process. In fact, the advection can be considered as a constant speed defined as the algebraic sum betw een the entry flow G,CBi and the exit flow G,Cj o r in term s o f fugacity as G i Z f , w here G, is the m atte r flow in m 3 m h r 1 entering into co m partm ent i with concentration CBi and leaving this com partm ent w ith concentration Q .
The emissions in com partm ent i, as a function o f the tim e in m ol m in -1, are denoted by
Ej = Ei(t). In this position we are supposing th at there is n o t any effect o f dilution, th a t is, the
volum e Vj o f each one o f the com partm ents is constant.
3. Fugacity model of a continuous and discrete time system
G iven a co m p artm en t i and a co m p artm en t j there exist the follow ing excluding possibilities for transferring substance betw een these tw o com partm ents: (a) there is a co n tact area between co m partm ent i and com p artm en t j , in this case, there is a positive g rad ien t while the equilibrium o f the fugacities will n o t be reached am ong these two com partm ents; (b) there is no contact area betw een co m p artm en t i and co m p artm en t j , consequently, there is n o t any possibility o f a direct transfer o f the substance am ong these two com partm ents. These tw o possibilities are fundam ental in the position and in the analysis o f the fugacity m odel for a con tin u o u s tim e control system.
Let / , — { / € N: the set o f indices i ^ j o f com partm ents in w hich there exists a contact area w ith co m p artm en t
/}-T he v ariatio n rate o f the fugacity in the tim e, for each one o f the com p artm en t, is calculated using the to ta l bulk balance, w hich is described by the follow ing system o f ordinary differential equations:
VtZt ~ = E, + GiCm + Y d f i f j ~
(
G‘Z‘+
r‘ V‘z >+
Y f l jJ
f" ^j€J, \ jeJ: J
488 R. Bru et al. I Appl. Math. Modelling 22 (1998) 485-494
T he system expressed by Eq. (3) can be w ritten as the follow ing continuous tim e co n tro l system:
f = A f + Bu,
w here f = f(i) is the fugacity derivative vector o f com ponents [/}] = d f / d t .
GiZj
+
riVjZi+
Y2jeJi dij(4)
A — [a,y] —
&ki '
Vlzl
if k = i,&ki
v,z,
dkiin e J l'
. a/a = 0 if k £ J i
is the n x n state m atrix, f = f(f) is the state vector o f fugacity o f com ponents [/]} = f ( t ) , B = I„ is the identity m atrix o f o rder n and u = u(7) is the co n tro l vector o f com ponents
[«,] =
(Ei+
GiCxyViZi.3.1. E xam ple
F o r a simple ecosystem constituted by air ( / = 1), w ater (i = 2) and b o tto m sedim ent (z' = 3), system (3) is determ ined by the system o f ordinary differential equations:
V\Z\ = E\ + G\CB\ + d2\ f i — (d\2 + G\Z\ + r\ V \Z \)f\ ,
VlZl = Eo G2CB2 + d \lf\ + <^3 2 /3 {d2\ + ^23 + G1Z2 + t 2 V2Z2) f 2,
P3Z3 = £3 + G^Cb3 + d jifl —(^ 32 + G3Z3 + ^3^ Z ^ fs
w ith the initial conditions /i( 0 ) = / f \ f 2{0) = / 2° a n d f 3(0) = / 3°, or by the continuous time co n tro l system (4) w here the m atrices A and I3, are respectively,
A -— (G\Z\ + d\i + r\ V\Z\) V\Z\ d\2 W 2 0 a 21 V & — {G2Z2 + Ai -I- d2i + r2 V2Z2) V2Z2 ^23 W i 0 i/32 V&2 — ( G3Z3 + d^2 + r3V3Z 3) '1 0 O' I 3 = 0 1 0 _0 0 1
The vectors f = f(i), f = f(i), u = u(i), fo = f(0) are, respectively,
f = d /i d/ 2 d/ 3
d t di d t f « = [ / H 0 / 2 ( 0 / 3 « ] T,
u(0 = [{El + G\CB\)/V \Z \ (E2 + G2CB2)/V 2Z2 ( £ 3 + G2CBi)/V 3Z3]T, and f(0) = [f? . In this
R Bru et al. I Appl. Math. Modelling 22 (1998) 485-494 489
N ow we retu rn to Eq. (4) w hich adm its the follow ing integral solution in [0; t\ (see [12]):
t
f(f) = eA'f(0 ) + J e A^ \ u ( t ) d r. (5) o
Let m be the n u m b er o f emissions a n d let T be the tim e interval betw een each emission. Discretizing Eq. (5), see [12] for details, for each t = k T , w here k = 0 , 1 , 2 , . . . , (m — 1) we have
f((* + l ) r ) = W( T ) f { k T ) + H (T )u (k T ), (6)
where
T
w (T) = eAT, H( T ) = J e Ml n dx, x = T - t .
o
W hen the tim e interval T betw een each em ission is fixed, the m atrices W(7~) an d H ( I ) are constants. W e are supposing th a t the emissions E l = E ^ t) an d consequently the vectors u = u(/) are co n stan t in a given interval betw een any tw o consecutive in stants o f em issions, th a t is, u = u(£7), for k T ^ t < k T + T.
Thus, if the emissions E ^ k T ) fo r each com partm ent i are know n, for the corresponding times
k = 0 , 1 , 2 , . . . , (m — 1), and consequently the control vectors u(kT ), then, it is possible to com pute
the fugacities an d the respective concentrations in these times for each one o f the com partm ents
i = 1 , 2 , . . . ,n, using Eq. (6). N o te th a t u(i) = 0, for all t ^ m T , an d in this case system (4) is
determ ined by
f((Jfc + 1)T) = W ( T ) f ( k T ) fo r k ^ m . (7)
We will see below th a t the discretization (6) has the p ro p erty o f m aintaining the stability o f Eq. (5).
In the m odel given by Eq. (4) and the corresponding solution (5), o r alternatively the dis cretized m odel (6) we suppose that: (i) the volum e o f each co m p artm en t is co n stan t, (ii) the emissions are know n fo r each tim e k T in the tim e interval [0; (m - 1)7] an d (iii) for the times outside o f th a t interval the em issions are null.
4. Analysis of the model
In this section we deal w ith som e properties o f o u r m odel. F irst we study the stability o f the continuous tim e control system described by the m atrix eq u atio n (4). L ater, we will prove th a t the discrete m odel (6) is asym ptotically stable. In ad d itio n , since we are m odelling fugacities, it is im p o rta n t to know the non-negativity o f the discrete tim e c o n tro l system given by m atrix equation (6) (i.e. all m atrices o f (6) has only non-negative entries). In the follow ing theorem s we will use the term inology, corresponding to non-negative m atrices used by Berm an and Plem m ons [13],
Theorem 1. The discrete control system defined by the pair (W( 7), H(7")) is a positive system, that is,
the components o f the matrices W(7~) and H(T) are non-negatives.
Proof. T he m atrix A is essentially non-negative since there exists a c o n stan t <5 > 0 sufficiently
large, for exam ple 5 > m axi=12 „[a,-,-], such th a t (A+<5I) > 0. T hen, by T heorem (3.12) o f [13],
490 R. Bru et al. / Appl. Math. Modelling 22 (1998) 485—494
As a practical consequence o f this theorem one can deduce th at if we apply non-negative controls and the initial conditions are non-negative then we always ob tain non-negative state vectors th a t in our problem are fugacities. In fact, this is the definition o f positive discrete tim e control systems, and this is equivalent to the m atrices W (T) an d H (7) are non-negatives.
W ith the following theorem we can assure th at system (4) is stable because the m atrix A has eigenvalues w ith negative real p arts (see [14]).
Theorem 2. L e t A be the m a trix o f system (4), - A is positively stable, that is, its eigenvalues have
positive real part.
Proof. F irst, by the definition o f the p roblem all param eters dtj are positive. F u rth erm o re, it is
observed th a t
then - A is strictly diagonal dom inant. Since - A has positive diagonal elem ents then the Ge- rschgorin’s disks are in the positive sem iplane. F rom b o th properties we deduce th a t the eigen values o f - A have positive real p a rt an d consequently - A is positively stable. Then A has eigenvalues w ith negative real p a rt. In ad d itio n A is n o t singular. □
The discretization (6) m aintains the stability o f the original continuous time co n tro l system, then the po in t where all fugacities vanish is asym ptotically stable for system (6). By com pleteness, we give the short p ro o f o f this result fo r o u r particular m atrices.
Theorem 3. The discrete positive control system (W (7), H (7 )) is asym ptotically stable, that is, the
point where all fugacities are null is an equilibrium stable point.
Proof. I f a, are the eigenvalues o f A then, e'-,r are the eigenvalues o f W (7). By T heorem 2 the
eigenvalues A,- have negative real p a rt, then we have |e/-, r | < 1 and therefore, | | W ( —» 0, when
p —> +oo. H ence the system (W (7), H (7 )) is asym ptotically stable. □
5. Algorithm to compute the settling time
To determ ine the needed tim e to get low concentrations in all com partm ents, we have to com pute the time for w hich the fugacities vanish, after the emissions finish. This tim e is called fugacity settling time. T he fu g a city settling time is defined as the tim e required for the fugacity values stay w ithin a range o f the final value. The settling tim e can be conceived as the time re quired so th a t the system arrives and stays within a range o f the equilibrium point. It can be com puted by the follow ing algorithm whose convergence is guaran teed by T heorem 3.
The com p u tatio n o f the m atrices W (7) and H(T) w hen the m atrix A is know n can be ac com plished by the m athem atical scientific package M A T L A B [15], w hich calculates the expo nential o f a m atrix th ro u g h P a d e ’s approxim ations with the function expm using the com m and
[W,H] = c2d (A,I„,7).
Algorithm 1 (A pseudo-code algorithm to com pute the settling time).
Begin
C om pute m atrices W(T) and H(T) by: [W,H] = c2d (A ,I„,7) F o r k = 0 until (m - 1) calculate
R. Bru et al. / Appl. Math. Modelling 22 (1998) 485-494 491
f((k + l)T ) = W( T ) f ( k T ) + U (T )u(kT )
fo r i = 1 until n com pute
Q (k T ) = Z jfi(kT )
L et tol > 0 be a tolerance o f the origin. F o r k > m com pute
i((k + l)T ) = W (T )f(k T ) + H (7)u (J)
until [|f(k + 1)71| ^ tol fo r i = 1 until n com pute
Ci{ { k + \) T = Z if i{(k+ l)T )
End;
T he first value o f k s = k , fo r w hich the inequality o f A lgorithm 1 is satisfied, determ ines the settling tim e ts = k sT for the tolerance tol. The value o f tolerance can be established as a function o f the lim its o f c h ro m ato g rap h y detection o f the concentrations o f the substance in the com partm ents.
6. Algorithm to compute the dissipation time
By Eq. (7), for each co m partm ent, it is possible to com pute the in sta n t tim e for w hich the value o f the concen tratio n re d u c e d p % o f the first value o f the co n cen tratio n after the em issions finish. This tim e is called the dissipation time in the co m p artm en t i, d enoted by fp a n d calculated by the follow ing procedure:
Algorithm 2 (A pseudo-code alg o rith m to com pute the d issipation time).
Begin
C om pute m atrices W (T) a n d H (7 ) by: [W,H] = c l d (A,I„,7) F o r k ^ m com pute
f((&+1) 7) = W ( T )f(kT )
Ci(kT) = Z f ( k T )
until C,(kT) {(100 -/? )/1 0 0 ■ Q ((m - 1 )T j End;
The first value o f k p — k , fo r w hich the inequality is satisfied, determ ines the dissipation tim e at level p % , fv — kpT fo r the co m p artm en t i.
7. Numerical simulation
W e have selected the insecticide fenitrothion to verify o u r m odel. F e n itro th io n is an insecticide frequently used in crops an d forest protection. F o r th a t, we have considered a hypothetical three com partm ents environm ental system consisting o f air, w ater an d b o tto m sedim ent w ith volum es o f 1.0 x 106, 1.0 x 104 a n d 1.5 x HP m 3, respectively. T he sedim ent density and co n te n t volu m etric o f carb o n are, respectively, 1.5 x 103 kg m ~3 an d 4.0%. In this system , we apply in the air one m ole o f the insecticide fen itro th io n during 60 m in, th a t is, E \(k ) — 1/60 and E2(k) = E^ik) = 0 fo r k = 0 , 1 , 2 , . . . , 59. A ccording to Z itko and M cLeese [16], the c o n sta n t rate o f disappearance in m inutes o f the fen itro th io n in air, w ater and b o tto m sedim ent are respectively 4.76 x 10-4, 3.80 x 10“4 and 2.85 x 10-5. H e n ry ’s co n stan t for the fen itro th io n (see [17]) is 6.65 x 10-2 P a m 3 m o l-1. T he coefficient o f a d so rp tio n k A for this substance in this b o tto m sedim ent is 27 x 10-3 m 3
492 R. Bru et al. I Appl. Math. Modelling 22 (1998) 485-494 1.0E-HH l,0E+00 1.0E-02 l,0E-a3 1.0E-M 1.0E-05 M>E-06 l,0E -07 1,0EMB 1.0E-09 M>E-10 1,0E-11 0 30 60 7200 14400 21600 28800 36000 43200 50400 57600 64800 72000 79200 86400 93600100800 Time (m inutes)
-•-A ir + W ater - k Sedim ent
Fig. 1. F ugacities o f fe n itro th io n in a ir, w a te r an d b o tto m sedim ent b efo re a n d a fte r th e em issions, o b ta in e d by A l g o rith m 1.
k g -1. T he value o f T an d tol are 1 m in and 1.0 x 10_u Pa, respectively. T he exit advection flow in air, w ater and b o tto m sedim ent are, respectively 6.87 x 102, 5.49 and 6.20 x 10-2 m 3 m in -1.
W ith these d a ta we wish to evaluate, during and after the em issions, the values o f the con cen tratio n a n d the fugacity o f fen itro th io n in the air, w ater an d b o tto m sedim ent an d then to determ ine the needed tim e to stabilize the fugacity, th a t is the fugacity settling time. In this ex am ple the air is the only one co m p artm en t receiving em issions a n d we can consider th a t the advections are null fo r all com partm ents.
1,OE+Ol 1,OE+00 1,0E-01 l,0E-02 i L,0E-fl3 e l,0Er04 c g l,0E -05 a o 1,0E~06 o o vj l,0E -07 I,0E-08 l,0E-09 1,0E-10 1,0E-U Time (m inutes) ■♦•Air “h W ater rh Sedim ent
Fig. 2. C o n c e n tra tio n s o f fe n itro th io n in air, w ater a n d b o tto m sed im en t b efore a n d a fte r th e em issions, o b ta in e d by A lg o rith m 1.
494 R. Bru et al. I Appl. Math. Modelling 22 (1998) 485—494
References
[1] G .N . Lewis, T he law o f physico-chem ical change, Proc. A m . A cad. Sci. 34 (1901) 49. [2] D . M ack ay , S. P aterso n , C alcu latin g fugacity, E nviron. Sci. T echnol. 15 (1981) 1006-1014. [3] D . M ackay, S. P a te rso n , F u g acity revisited, E nviron. Sci. T echnol. 16 (1982) 654A -660A .
[4] J. K o p riv n jak , L. P o issan t, E v a lu a tio n an d ap p licatio n o f a fugacity m odel to explain the p a rtitio n in g o f c o n ta m in a n ts in the St. L aw rence river valley, W ater, A ir a n d Soil P o llu tio n 97 (1997) 379-395.
[5] S. P aterso n , D . M ack ay , T he F u g acity C oncept in E n v iro n m en tal M odelling, in: O. H u tzin g er (E d.), T he H a n d b o o k o f E n v iro n m en tal C hem istry, vol. 2, P a rt C, Springer, B erlin, 1985, pp. 121-145.
[6] J. C am pfens, D . M ack ay , F u g acity -b ase d m odel o f PCB b io acc u m u latio n in com plex a q u a tic food webs, E nviron. Sci. T echnol. 31 (1997) 577-583.
[7] D . M ack ay , M u ltim ed ia E n v iro n m en tal M odels: T he F u g acity A p p ro a c h , Lewis P ublishers, A n n A rb o r, M I, 1991, p. 257.
[8] Y. C o h en , A .P. R y a n , M u ltim ed ia m odelling o f en v iro n m en tal tra n sp o rt: T rich lo ro eth y len e test case, E nviron. Sci. T echnol. 19 (1985) 412-417.
[9] Y. M a to b a , J. O hnishi, M . M atsu o , In d o o r sim ulation o f insecticides in b ro a d c a s t spraying, C hem osphere 30 (1995) 345-356.
[10] W . Stiver, D. M ack ay , L in ear su p erp o sitio n in m odelling c o n ta m in a n t b eh av io u r in a q u a tic system s, W ater R esearch 29 (1995) 329-335.
[11] E. Bacci, E cotoxicology o f O rganic C o n tam in an ts, Lewis P ublishers, B oca R a to n , F lo rid a , 1994, p. 165. [12] K. O g ata, D iscrete-T im e C o n tro l System s, P rentice-H all, E nglew ood Cliffs, N J, 1995, p. 745.
[13] A. B erm an, R .J. P lem m ons, N o n n eg ativ e M atrices in the M a th e m a tic a l Sciences, Classics in A pplied M ath em atics, S IA M , P hiladelphia, P A , 1994, p. 340.
[14] D .G . L uenberger, In tro d u c tio n to D y n am ic Systems, W iley, N ew Y o rk , 1979, p. 446.
[15] M a th W o rk s Inc., T h e S tu d en t E d itio n o f M A T L A B , P rentice-H all, E nglew ood Cliffs, N J, 1992, p. 820.
[16] V. Z itk o , D .W . M cL eese, E v a lu a tio n o f hazard s o f pesticides used in fo rest sp ray in g to the a q u a tic environm ent, C a n a d ia n T echnical R e p o rt o f F isheries A q u atic Sciences, G o v e rn m e n t o f C a n a d a , D ecem ber, 1980. p. 21. [17] O. Y enigiin, D . S o h to rin k , C alcu latio n s w ith the level II fugacity m odel fo r selected o rg a n o p h o ru s insecticides,
W ate r, A ir an d Soil P o llu tio n 84 (1985) 175-185.
[18] K. K aw ata, A. Y a su h a ra , D e te rm in a tio n o f fen itro th io n a n d d iazin o n in air. Bull. E n v iro n . C o n tam . T oxicol. 52 (1994) 419-424.
[19] N a tio n a l R esearch C ouncil o f C a n a d a , F en itro th io n : T he effects o f its use on en v iro n m en tal q u ality a n d its chem istry, A ssociate C o m m ittee on Scientific C riteria fo r E n v iro n m en tal Q uality, N R C C N o. 14104, 1975, p. 162, O tta w a , C an ad a.