Kinetic bounds on attainability in the reactor synthesis proble

For a given chemical reaction network with known kinetics, a specified feed composition, and specific process constraints, the attainable region for the reactor synthesis problem is the set of all composition vectors that are achievable through all constraint-consistent reactor configurations that invoke only reaction and mixing. The boundary of the attainable region gives useful information to a process designer in terms of yield and selectivity of a desired chemical species that might be achieved by any constraint-consistent design. The process designer can use this information to assess the effectiveness of existing designs and consider alternate designs that might produce an effluent composition closer to the boundary of the attainable region.;Though much is known about necessary conditions for compositions to lie on the boundary of the attainable region, a definite method to calculate the true boundary of the attainable region is not known. Most of the existing methods compute "candidate" attainable regions by starting at the feed composition and attempt to enlarge the set of attainable compositions by considering various reactor configurations until no further enlargement is seen. These methods raise the question of whether the "candidate" attainable region has come close to the (currently unknown) true attainable region. These methods cannot preclude with certainty the existence of some novel reactor configuration that can enlarge the current set of achievable compositions.;This thesis describes a new technique called the method of bounding hyperplanes that complements existing methods by "bounding from the outside" those composition vectors that are attainable. In other words, the method seeks to create a set of composition vectors such that all composition vectors lying outside this set are inaccessible from the feed. Thus, the true attainable region will be contained in or will coincide within the bounding set so created.;The theoretical underpinnings for the method of bounding hyperplanes are described followed by algorithms to implement the method computationally. Kinetic bounds are computed by the method in several examples and compared with stoichiometric bounds and with compositions computed for various typical reactor designs. The kinetic bounds do well in precluding a large set of compositions that would otherwise be considered feasible from stoichiometric considerations alone. In many instances, the computed bounds are sharp insofar as they come very close to compositions known to be attainable. The methods described in this thesis do not replace existing methods to find candidate attainable regions but rather complement them.