Brandon Corbett

Data-Based Modeling and Control of Nylon-6, 6 Batch Polymerization

This work addresses the problem of modeling the complex nonlinear behavior of the nylon-6, 6 batch polymerization process and then subsequently tracking trajectories of important process variables, namely the reaction medium temperature and reactor pressure, using model predictive control. To this end, a data-based multi-model approach is proposed in which multiple local linear models are identified from previous batch data using latent variable regression and then combined using an appropriate (continuous) weighting function that arises from fuzzy c-means clustering. The proposed approach unifies the concepts of auto-regressive exogenous (ARX) modeling, latent variable regression techniques, fuzzy c-means clustering, and multiple local linear models in an integrated framework capable of capturing the nonlinearities and multivariate nature of batch data. The resulting data-based model is then used to formulate a trajectory tracking predictive controller. Through simulation studies, the modeling approach is shown to capture the major nonlinearities in the nylon-6, 6 polymerization process and closed-loop simulation results demonstrate the efficacy of the proposed predictive controller and illustrate its advantages over existing trajectory tracking approaches such as conventional proportional-integral control and latent variable model predictive control.

Injectable Anisotropic Nanocomposite Hydrogels Direct in Situ Growth and Alignment of Myotubes

While injectable in situ cross-linking hydrogels have attracted increasing attention as minimally invasive tissue scaffolds and controlled delivery systems, their inherently disorganized and isotropic network structure limits their utility in engineering oriented biological tissues. Traditional methods to prepare anisotropic hydrogels are not easily translatable to injectable systems given the need for external equipment to direct anisotropic gel fabrication and/or the required use of temperatures or solvents incompatible with biological systems. Herein, we report a new class of injectable nanocomposite hydrogels based on hydrazone cross-linked poly(oligoethylene glycol methacrylate) and magnetically aligned cellulose nanocrystals (CNCs) capable of encapsulating skeletal muscle myoblasts and promoting their differentiation into highly oriented myotubes in situ. CNC alignment occurs on the same time scale as network gelation and remains fixed after the removal of the magnetic field, enabling concurrent CNC orientation and hydrogel injection. The aligned hydrogels show mechanical and swelling profiles that can be rationally modulated by the degree of CNC alignment and can direct myotube alignment both in two- and three-dimensions following coinjection of the myoblasts with the gel precursor components. As such, these hydrogels represent a critical advancement in anisotropic biomimetic scaffolds that can be generated noninvasively in vivo following simple injection.

Model Predictive Quality Control of Polymethyl Methacrylate

This work considers the production of Polymethyl methacrylate (PMMA) to achieve target quality variables such as number and weight average molecular weights. A dynamic multiple-model based approach is first used to capture the process dynamics using data generated from a detailed first principles model. Subsequently, the multiple-model is integrated with a quality model to enable predicting the end quality based on initial conditions and candidate control input (jacket temperature) moves. A data-driven model predictive controller is then designed to achieve the desired product quality while satisfying input and a lower bound on the conversion, as well as additional constraints that enforce the validity of data-driven models for the range of chosen input moves. Simulation results demonstrate the superior performance (10.4% and 6.5% relative error in number average and weight average molecular weight compared to 19.8% and 18.5%) of the controller over traditional trajectory tracking approaches.

Model predictive quality control of batch processes

This work addresses the problem of driving a batch process to a specified product quality using model predictive control (MPC) with data-driven models. To address the problem of unavailability of online quality measurements, an inferential quality model, which relates the process conditions over the entire batch duration to the final quality, is required for this problem. At a given sampling instant, the accuracy of this type of quality model, however, is sensitive to the prediction of the future (unknown) batch behavior. That is, errors in the predicted future data are propagated to the quality prediction, adding uncertainty to any control action based on the predicted quality. To address this “missing data” problem, we integrate a previously developed data-driven modeling methodology, which combines multiple local linear models with an appropriate weighting function to describe nonlinearities, with the inferential model in a predictive control framework. The key benefit of this approach is that the causality and nonlinear relationship between the future inputs and outputs are accounted for in predicting the final quality, resulting in more effective control action. The efficacy of the proposed predictive control design is illustrated via closed-loop simulations of an industrially relevant nylon-6,6 batch polymerization process.

Subspace-Based Model Identification of a Hydrogen Plant Startup Dynamics

Abstract This work addresses the problem of determining a data-driven model for the startup of a hydrogen production unit, and demonstrates the approach both on a detailed first principles simulation model and by application to real data. To this end, first a detailed first principles model of the hydrogen plant is developed in Honeywells UniSim design by adapting the plant standard operating procedure (SOP). Illustrative simulations are next presented to establish the meaningfulness of approximating process nonlinearity with a (higher order) linear time invariant (LTI) model. Then an LTI data-driven model of the hydrogen unit startup process using subspace identification based methods is identified. The framework is then implemented and successfully validated data on simulated data and on data from an industrial hydrogen unit.

Model predictive quality control of Polymethyl methacrylate

This work considers the production of Polymethyl methacrylate (PMMA) to achieve target quality variables such as number and weight average molecular weights. A dynamic multiple-model based approach is first used to capture the process dynamics using data generated from a detailed first principles model. Subsequently, the multiple-model is integrated with a quality model to enable predicting the end quality based on initial conditions and candidate control input (jacket temperature) moves. A data-driven model predictive controller is then designed to achieve the desired product quality while satisfying input and a lower bound on the conversion, as well as additional constraints that enforce the validity of data-driven models for the range of chosen input moves. Simulation results demonstrate the superior performance (10.4% and 6.5% relative error in number average and weight average molecular weight compared to 19.8% and 18.5%) of the controller over traditional trajectory tracking approaches.

Data-based modeling and control of nylon-6,6 batch polymerization

This work addresses the problem of modeling the complex nonlinear behavior of a nylon-6,6 batch polymerization process and subsequently tracking trajectories of the important process variables, namely the reaction medium temperature and reactor pressure, using model predictive control (MPC). To this end, a data-based multi-model approach is proposed in which local linear models are identified from previous batch data using latent variable regression and then combined using a continuous weighting function that arises from fuzzy c-means clustering. The resulting data-based model is used to formulate a trajectory tracking predictive controller. Through simulation studies, the modeling approach is shown to capture the major nonlinearities of the process, and closed-loop simulation results demonstrate the efficacy of the proposed predictive controller and its advantages over conventional proportional-integral (PI) trajectory tracking.

Development of a high fidelity and subspace identification model of a hydrogen plant startup dynamics

In this work, the problem of determining a data-driven model of a hydrogen production unit is addressed. The framework is applied to a high fidelity simulation model developed in this work. To this end, first a high fidelity model of the entire plant is developed in Honeywells UniSim Design, capable of simulating the startup and shutdown phase, with appropriate adaptation of the plant standard operating procedure (SOP). Several startups are simulated to generate training data for identification of a data-driven model. Then an LTI data-driven model of the process using subspace identification based methods is determined and validated against new simulated startup. Simulation results demonstrate the prediction capabilities of the identified model.

Quality control of variable duration batch processes via subspace identification

Batch process reactors are often used for products where quality is of paramount importance. To this end, this work addresses the problem of direct, data-driven, quality control for batch processes. Specifically, previous results using subspace identification for modeling dynamic evolution and making quality predictions are extended with two key novel contributions: first, a method is proposed to account for mid-batch ingredient additions in both the modeling and control stages. Second, a novel model predictive control scheme is proposed that includes batch duration as a decision variable. The efficacy of the proposed modeling and control approaches are demonstrated using a simulation study of a polymethyl methacrylate (PMMA) reactor. Closed loop simulation results show that the proposed controller is able to reject disturbances in feed stock and drive the number average molecular weight, weight average molecular weight, and conversion to their respective set-points. Specifically, mean absolute percentage errors (MAPE) in these variables are reduced from 8.66%, 7.87%, and 6.13% under traditional PI control to 1.61%, 1.90%, and 1.67% respectively.

Data-driven quality control of batch processes via subspace identification

In this work we present a novel, data-driven, quality modeling and control approach for batch processes. Specifically, we adapt subspace identification methods for use with batch data to identify a state-space model from available process measurements and input moves. We demonstrate that the resulting LTI, dynamic, state-space model is able to describe the transient behavior of finite duration batch processes. Next, we relate the terminal quality to the terminal value of the identified states. Finally, we apply the resulting model in a shrinking-horizon, model predictive control scheme to directly control terminal product quality. The theoretical properties of the proposed approach are studied and compared to state-of-the-art latent variable control approaches. The efficacy of the proposed approach is demonstrated through a simulation study of a batch polymethyl methacrylate (PMMA) polymerization reactor. Results for both disturbance rejection and set-point changes (that is, new quality grades) are demonstrated.