Solution of the network equations

Equations (3.3) and (3.4) have become known respectively as the valence sum rule and the loop, or equal valence, rule, and are known collectively as the network equations. Equation (3.4) represents the condition that each atom distributes its valence equally among its bonds subject to the constraints of eqn (3.3) as shown in the appendix to Brown (1992a). The two network equations provide sufficient constraints to determine all the bond valences, given a knowledge of the bond graph and the valences of the atoms. The solutions of the network equations are called the theoretical bond valences and are designated by the lower case letter 5. Methods for solving the network equations are described in Appendix 3. ... 

An example of the solution of the network equations for the bond graph of CaCrFj is shown in Figure lO.lOd. 

We now consider elemental potentials of general form, for which the evaluation of the elemental deformation functions is essential to the solution of the network equations. 

Probably the least flexible of all methods with respect to the time-step and distance relationship is the method of characteristics (MOC). It requires the pipe lengths in a network to be adjusted to satisfy the condition of a common time interval, but provides an accurate solution of the differential equations. MOC has been successfully implemented by Goacher (G4), Streeter and associates (S6), and Masliyah and Shook (M5). More recently,... 

However, if the atoms are not related by symmetry, the normal rules break down. The homoionic N-N bond in the hydrazinium ion is an electron pair bond, but one in which N1 contributes 1.25 and N2 0.75 electrons. How can we apply the bond valence model in such cases where no solution to the network equations is possible One approach is to isolate the non-bipartite portion of the graph into a complex pseudo-atom. Thus in the hydrazinium ion the homoionic bond and its two terminating N atoms are treated as a single pseudo-anion which forms six bonds with a valence sum equal to the formal charge of —4. 

It can thus be seen that training radial basis function networks uses both supervised and unsupervised learning determining basis function parameters is unsupervised and solution of the linear equations is supervised. 

Software for direct, "brute-force" solution of the rate equations is available [4-9] and can be used if the network consists of only a few elementary steps. In practice, however, effective fundamental modeling usually calls for a reduction in the number of simultaneous rate equations and their coefficients. As Chapter 6 has shown, a systematic application of the Bodenstein approximation to all trace-level intermediates can achieve this, at least unless the network is largely non-simple. 

Voluminous literature exists on the calculation of reaction equilibria in complex networks. The following two procedures are particularly useful simultaneous solution of the equilibrium equations and minimization of the free energy. 

The model representing diffusion/reaction involved solution of the transport equations for each single pore simultaneously to give concentration profile in the pore network. The calculations related to capillary condensation were performed in the same way as for the Fickian model, described in Section lll.C. 

The simplest method of solution of the Kirchhoff equations, corresponding to the random network of conductance elements, is the single-bond effective medium approximation (SB-EMA), wherein a single effective bond between two pores is considered in an effective medium of surrounding bonds. The conductivity ab of the effective bond is obtained as the self-consistent solution of the equation... 

Attack by alkali solution, hydrofluoric acid and phosphoric acid A common feature of these corrosive agents is their ability to disrupt the network. Equation 18.1 shows the nature of the attack in alkaline solution where unlimited numbers of OH ions are available. This process is not encumbered by the formation of porous layers and the amount of leached matter is linearly dependent on time. Consequently the extent of attack by strong alkali is usually far greater than either acid or water attack. 

Kunii and Levenspiel(1991, pp. 294-298) extend the bubbling-bed model to networks of first-order reactions and generate rather complex algebraic relations for the net reaction rates along various pathways. As an alternative, we focus on the development of the basic design equations, which can also be adapted for nonlinear kinetics, and numerical solution of the resulting system of algebraic and ordinary differential equations (with the E-Z Solve software). This is illustrated in Example 23-4 below. 

For other reaction networks, a similar set of equations may be developed, with the kinetics terms adapted to account for each reaction occurring. To determine the conversion and selectivity for a given bed depth, Ljh equations 23.4-11 and -14 are numerically integrated from x = 0 to x = Lfl, with simultaneous solution of the algebraic expressions in 23.4-12, -13, -15, and -16. The following example illustrates the approach for a series network. 

The obvious advantage is that the steady-state solution of an S-system model is accessible analytically. However, while the drastic reduction of complexity can be formally justified by a (logarithmic) expansion of the rate equation, it forsakes the interpretability of the involved parameters. The utilization of basic biochemical interrelations, such as an interpretation of fluxes in terms of a nullspace matrix is no longer possible. Rather, an incorporation of flux-balance constraints would result in complicated and unintuitive dependencies among the kinetic parameters. Furthermore, it must be emphasized that an S-system model does not necessarily result in a reduced number of reactions. Quite on the contrary, the number of reactions r = 2m usually exceeds the value found in typical metabolic networks. 

Commercial impedance analyzers offer equivalent circuit interpretation software that greatly simplifies the interpretation of results. In this Appendix we show two simple steps that were encountered in Chapters 3 and 4 and that illustrate the approach to the solution of equivalent electrical circuits. First is the conversion of parallel to series resistor/capacitor combination (Fig. D.l). This is a very useful procedure that can be used to simplify complex RC networks. Second is the step for separation of real and imaginary parts of the complex equations. 

The integration of these mass balances, often with the help of the computer because the nonlinearity of the equation does not allow for an analytical solution, allows for a calculation of the changes in the concentrations of the molecules over time, given all concentration at time zero, values of all the kinetic parameters, and a description of the interaction of the network (the system ) with its environment. 

Consider a single-phase homogeneous stirred-tank reactor with a time-invariant velocity field U(x, y, z ) a single reaction of the form /) —> B. (This approach can be extended to the case of time-dependent velocity fields. If the flow in the tank is turbulent, then the velocity field is the solution of the Reynolds averaged Navier-Stokes equations). The tank is divided into a three-dimensional network of n spatially fixed volumetric elements, or n-interacting... 

Another application of the analysis of the stoichiometric matrix is flux balance analysis (Edwards et al. 2002). Often the number of fluxes in the system exceeds the number variable metabolites making equation (3) an underdetermined set of linear equations, that is, many different combinations of fluxes are consistent with system steady state. One approach is to measure the fluxes that enter and exit the cell. Because intracellularly there are many redundant pathways, this does not enable one to determine all fluxes. Isotope labelling may help then (Wiechert 2002). Another approach to then find a smaller number of solutions is to postulate that the solution should satisfy an additional objective. This objective is taken to be associated with optimal functioning of the network, for instance maximization of some flux or combination... 

Although the TAP reactor equations can be solved analytically for some simple systems, more complicated reation networks require numerical solutions. Cleaves et al (35) show that useful relations can be found between quantities such as conversion and residence time and the moments of the response of the TAP reactor, analytically obtained from the solutions of the linear system described previously. 

Equation (7-54) allows calculation of the residence time required to achieve a given conversion or effluent composition. In the case of a network of reactions, knowing the reaction rates as a function of volumetric concentrations allows solution of the set of often nonlinear algebraic material balance equations using an implicit solver such as the multi variable Newton-Raphson method to determine the CSTR effluent concentration as a function of the residence time. As for batch reactors, for a single reaction all compositions can be expressed in terms of a component conversion or volumetric concentration, and Eq. (7-54) then becomes a single nonlinear algebraic equation solved by the Newton-Raphson method (for more details on this method see the relevant section this handbook). 

---