KCKyle V. Camarda

Kyle V. Camarda

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Kyle V. Camarda has a total of 11 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Multicomponent Measurement and Modeling of the Solid–Liquid and Solid–Solid–Liquid Equilibria of N-Phenylsuccinimide and Succinanilic Acid in DMF/Antisolvent Mixtures

    by Diego Trevisan Melfi, Allen To, Douglas Dougherty, Kyle V. Camarda, Aaron M. Scurto

    N-Phenylsuccinimide (NPS) is used in several industrial products and can be synthesized from succinanilic acid (SA) in dimethylformamide (DMF). However, incomplete conversion generally produces mixtures of NPS and residual SA in the reactor effluent. Antisolvent crystallization is a potential separation strategy but requires reliable multicomponent solubility data for NPS and SA and well-tuned thermodynamic models. In this study, openCOSMO-RS was evaluated for cosolvent/antisolvent screening and water, 2-propanol, and ethyl acetate were chosen for experimental studies. The melting point and enthalpy of fusion for the pure NPS and SA were measured. Solid-liquid equilibrium (SLE) and solid-solid-liquid equilibrium (SSLE) of binary, ternary, and quaternary systems involving NPS, SA, DMF, water, ethyl acetate, and 2-propanol were measured at temperatures between 10°C and 25°C. Water and 2-propanol act as antisolvents for NPS in DMF, while only water showed an antisolvent effect for SA. Ethyl acetate behaves as a cosolvent in the studied DMF mixtures. While openCOSMO-RS was able to predict the general qualitative trends, the NRTL activity coefficient model was used to correlate the binary data and predict higher-order mixtures. The NRTL parameters obtained only from binary SLE data described pure-solvent systems well but did not represent ternary DMF/antisolvent systems quantitatively for all systems. However, including ternary SLE data in the regression improved the correlation and would allow design and simulation of a crystallizer.

  • Open Access

    Network Enhanced Multi-Objective Particle Swarm Optimization: A Graph Theory-based MOPSO for Chemical Engineering Applications

    by Fernando Zea, Kyle V. Camarda

    Multiobjective optimization in chemical engineering requires balancing conflicting objectives within highly nonlinear, simulationdriven design spaces. Traditional evolutionary algorithms can struggle with maintaining convergence, diversity, and computational efficiency when applied to such problems. This work presents NEMOPSO, a graphtheoretic multiobjective particle swarm optimization framework that enhances search performance through structured leader selection, topologyaware information flow, and adaptive exploration. By modeling the swarm as a dynamic graph and selecting leaders using centrality metrics, NEMOPSO prioritizes structurally influential solutions rather than relying solely on geometric diversity. Benchmarking on the ZDT suite demonstrates that closenessbased leader selection with caching achieves superior hypervolume and convergence on smooth and moderately complex landscapes, while scalefree topologies excel in deceptive problems. Four chemicalengineering case studies: a lowdensity polyethylene reactor, a distillation column, a methane trireforming reactor, and a continuousflow SNAr reaction, further validate the approach. NEMOPSO consistently reproduces or improves upon reference Pareto fronts, achieving lower spacing, competitive hypervolume, and strong IGD+ performance, while maintaining robust execution times even when coupled to highfidelity simulators. Results show that the selection method of leaders’ influence, not the size or diversity of the stored solutions, is the primary driver of optimization quality. Overall, NEMOPSO provides a flexible, scalable, and computationally efficient framework for multiobjective optimization in chemicalengineering applications.

  • Open Access

    Neural Networks as Surrogate Models in Multi-objective Optimization for Chemical Reactor Design

    by Tate Bestwick, Kyle V. Camarda

    Peak system sensitivity, a measure of a system’s propensity to become unstable, is a crucial consideration for safety in chemical plants. However, many chemical process units have nonlinear dynamics that present significant computational challenges to incorporating peak sensitivity with other design objectives. This work addresses these difficulties by using an neural network (NN) as a surrogate model to calculate design objective values for multi-objective optimization. In so doing, the multi-objective optimization problem is simplified from a bi-level nonlinear program to a mixed-integer linear program. The NN is employed as a surrogate model that takes design decision variables as inputs and computes objective values as outputs.The described method for optimizing design parameters is demonstrated on a Haber-Bosch reactor model. A NN is trained to accept three reactor design parameters as inputs and compute two competing objective values: peak sensitivity and ammonia production rate. The NN surrogate model is used to generate a Pareto-optimal front, which is subsequently compared with data from the first-principles model to validate its accuracy. The Pareto-front produced from the NN surrogate model closely aligns with the data from the first-principles model. Furthermore, the NN surrogate model shows robustness toward outlier training data points. These findings highlight the effectiveness of NNs as surrogate model for multi-objective optimization in chemical reactor design and provide a method for evaluating system sensitivity as a design objective along with economic and environmental objectives.

  • Open Access

    Gaussian Process-Aided Dynamic Control of a Nonlinear Reactor

    by Tate Bestwick, Kyle V. Camarda

    Proportional-Integral-Derivative (PID) control remains one of the most common low-level control strategies in engineering. However, chemical systems often exhibit nonlinear dynamics that are not easily regulated by statically tuned PID controllers. This work proposes a Gaussian Process (GP) method to enable real-time adaptive PID control for nonlinear chemical systems. The proposed method preserves existing PID infrastructure while dynamically adjusting controller parameters in response to changing operating conditions. Results show that GP-tuned PID controllers offer comparable performance to static PID controllers under disturbances which produce linear changes in outputs. However, under multivariable nonlinear disturbances, GP-tuned control significantly improves setpoint tracking—reducing error by up to 90%. Both Matérn 3/2 and exponential kernels are evaluated, with the latter yielding smoother and more stable control. This framework enables dynamic adaptability for PID controllers, offering a low-risk path toward effective control in practical industrial settings.