Stochastic optimization algorithms have been developed for decades to provide approaches to solve process synthesis problems, such as Tabu Search (TS) and Simulated Annealing (SA). However, complicated high-throughput synthesis applications, e. g. complex reactor network design, consisting of many components, procedures and parameters, may encounter slow convergence with current optimization algorithms. The development of "grids" technology enables integrated applications and the design of distributed experiments in industrial environments. One objective of this work is the design of efficient high-throughput infrastructures for enabling faster optimization convergence with stochastic optimization algorithms like TS and distributed optimization algorithms like SA Cascade. This will in turn enable the undertaking of computing challenges. Results show that optimization with the assist of grids can achieve faster convergence than those without grids. However the intrinsic parameters of a reactor network, such as number of reactors, reactor types, reactor sequence, reactor volume, feed flow and split fraction, have not been analyzed in previous optimization algorithms. Another objective of the work is to design a knowledge-based optimization model and integrate ontology with the knowledge-based model for conceptualization of relationships between intrinsic parameters and objective value that is equivalent to the concentration of desired product and further guiding the optimization search at run-time. The results show that the computational performance becomes more efficient with the assistance of newly discovered knowledge during the optimization process, which proves the feasibility of the knowledge-based optimization with ontology. Future work involves the full and complete integration of the knowledge-based model with ontology and development of a more intelligent ontology-supported optimization model to achieve completely automatic applications for all steps of the ontology-supported model.