cantera/include/cantera/equil/MultiPhaseEquil.h
Ray Speth 002c158761 Cleanup include statements
Move includes from header to implementation files where possible, and remove
unnecessary includes.
2014-08-28 16:54:13 +00:00

191 lines
6.4 KiB
C++

//! @file MultiPhaseEquil.h
#ifndef CT_MULTIPHASE_EQUIL
#define CT_MULTIPHASE_EQUIL
#include "MultiPhase.h"
namespace Cantera
{
/*!
* Multiphase chemical equilibrium solver. Class MultiPhaseEquil is designed
* to be used to set a mixture containing one or more phases to a state of
* chemical equilibrium. It implements the VCS algorithm, described in Smith
* and Missen, "Chemical Reaction Equilibrium."
*
* This class only handles chemical equilibrium at a specified temperature and
* pressure. To compute equilibrium holding other properties fixed, it is
* necessary to iterate on T and P in an "outer" loop, until the specified
* properties have the desired values. This is done, for example, in method
* equilibrate of class MultiPhase.
*
* This class is primarily meant to be used internally by the equilibrate
* method of class MultiPhase, although there is no reason it cannot be used
* directly in application programs if desired.
*
* @ingroup equil
*/
class MultiPhaseEquil
{
public:
//! Construct a multiphase equilibrium manager for a multiphase mixture.
//! @param mix Pointer to a multiphase mixture object.
//! @param start If true, the initial composition will be determined by a
//! linear Gibbs minimization, otherwise the initial mixture
//! composition will be used.
MultiPhaseEquil(MultiPhase* mix, bool start=true, int loglevel = 0);
virtual ~MultiPhaseEquil() {}
size_t constituent(size_t m) {
if (m < m_nel) {
return m_order[m];
} else {
return npos;
}
}
void getStoichVector(size_t rxn, vector_fp& nu) {
size_t k;
nu.resize(m_nsp, 0.0);
if (rxn > nFree()) {
return;
}
for (k = 0; k < m_nsp; k++) {
nu[m_order[k]] = m_N(k, rxn);
}
}
int iterations() {
return m_iter;
}
doublereal equilibrate(int XY, doublereal err = 1.0e-9,
int maxsteps = 1000, int loglevel=-99);
doublereal error();
std::string reactionString(size_t j) {
return std::string("");
}
void printInfo(int loglevel) {}
void setInitialMixMoles(int loglevel = 0) {
setInitialMoles(loglevel);
finish();
}
size_t componentIndex(size_t n) {
return m_species[m_order[n]];
}
void reportCSV(const std::string& reportFile);
double phaseMoles(size_t iph) const;
protected:
//! This method finds a set of component species and a complete set of
//! formation reactions for the non-components in terms of the components.
//! In most cases, many different component sets are possible, and
//! therefore neither the components returned by this method nor the
//! formation reactions are unique. The algorithm used here is described
//! in Smith and Missen, Chemical Reaction Equilibrium Analysis.
//!
//! The component species are taken to be the first M species in array
//! 'species' that have linearly-independent compositions.
//!
//! @param order On entry, vector \a order should contain species index
//! numbers in the order of decreasing desirability as a component.
//! For example, if it is desired to choose the components from among
//! the major species, this array might list species index numbers in
//! decreasing order of mole fraction. If array 'species' does not
//! have length = nSpecies(), then the species will be considered as
//! candidates to be components in declaration order, beginning with
//! the first phase added.
void getComponents(const std::vector<size_t>& order);
//! Estimate the initial mole numbers. This is done by running each
//! reaction as far forward or backward as possible, subject to the
//! constraint that all mole numbers remain non-negative. Reactions for
//! which \f$ \Delta \mu^0 \f$ are positive are run in reverse, and ones
//! for which it is negative are run in the forward direction. The end
//! result is equivalent to solving the linear programming problem of
//! minimizing the linear Gibbs function subject to the element and non-
//! negativity constraints.
int setInitialMoles(int loglevel = 0);
void computeN();
//! Take one step in composition, given the gradient of G at the starting
//! point, and a vector of reaction steps dxi.
doublereal stepComposition(int loglevel = 0);
//! Re-arrange a vector of species properties in sorted form
//! (components first) into unsorted, sequential form.
void unsort(vector_fp& x);
void step(doublereal omega, vector_fp& deltaN, int loglevel = 0);
//! Compute the change in extent of reaction for each reaction.
doublereal computeReactionSteps(vector_fp& dxi);
void updateMixMoles();
//! Clean up the composition. The solution algorithm can leave some
//! species in stoichiometric condensed phases with very small negative
//! mole numbers. This method simply sets these to zero.
void finish();
// moles of the species with sorted index ns
double moles(size_t ns) const {
return m_moles[m_order[ns]];
}
double& moles(size_t ns) {
return m_moles[m_order[ns]];
}
int solutionSpecies(size_t n) const {
return m_dsoln[m_order[n]];
}
bool isStoichPhase(size_t n) const {
return (m_dsoln[m_order[n]] == 0);
}
doublereal mu(size_t n) const {
return m_mu[m_species[m_order[n]]];
}
std::string speciesName(size_t n) const {
return
m_mix->speciesName(m_species[m_order[n]]);
}
//! Number of degrees of freedom
size_t nFree() const {
return (m_nsp > m_nel) ? m_nsp - m_nel : 0;
}
size_t m_nel_mix, m_nsp_mix, m_np;
size_t m_nel, m_nsp;
size_t m_eloc;
int m_iter;
MultiPhase* m_mix;
doublereal m_press, m_temp;
std::vector<size_t> m_order;
DenseMatrix m_N, m_A;
vector_fp m_work, m_work2, m_work3;
vector_fp m_moles, m_lastmoles, m_dxi;
vector_fp m_deltaG_RT, m_mu;
std::vector<bool> m_majorsp;
std::vector<size_t> m_sortindex;
vector_int m_lastsort;
vector_int m_dsoln;
vector_int m_incl_element, m_incl_species;
// Vector of indices for species that are included in the calculation.
// This is used to exclude pure-phase species with invalid thermo data
std::vector<size_t> m_species;
std::vector<size_t> m_element;
std::vector<bool> m_solnrxn;
bool m_force;
};
}
#endif