About

Costas Kravaris is a Professor in the Department of Chemical Engineering at Texas A&M University. He holds a Ph.D. and an M.S. in Chemical Engineering from the California Institute of Technology, earned in 1984 and 1981 respectively, and a Diploma in Chemical Engineering from the National Technical University of Athens, Greece, obtained in 1979. His research interests include process control, state observers, and fault detection and isolation. Kravaris has contributed to the understanding of process dynamics and control, and has published extensively on topics related to nonlinear systems, functional observers, and diagnostic approaches in process engineering.

Research topics

• Computer Science
• Artificial Intelligence
• Machine Learning
• Data Mining
• Process engineering
• Engineering
• Chemistry
• Biological system
• Mathematics
• Control engineering

Selected publications

• <scp>Model</scp> ‐based fault diagnosis and fault tolerant control in safety‐critical chemical reactors: An experimental study

  AIChE Journal · 2026-04-09

  articleOpen accessSenior authorCorresponding

  Abstract This study investigates a fault‐tolerant control (FTC) approach for continuous stirred‐tank reactors (CSTR), emphasizing the importance of timely interventions to ensure operational safety under fault conditions. A systematic methodology combining residual‐based fault estimation and Dynamic Safety Margin (DSM) monitoring is developed to guide the activation of a backup solvent injection strategy (“Plan B”). We introduce the concepts of critical fault magnitude and critical time, which provide explicit criteria for assessing when the nominal controller alone is insufficient. Integrating these concepts, we propose a unified FTC decision‐making framework to activate Plan B promptly. Experimental results confirm that this combined strategy robustly maintains the reactor within its safety constraints, accurately tracks setpoints, and effectively manages faults of varying magnitudes. The approach presented herein provides clear, practical guidelines for enhancing safety and reliability in chemical reactor operations.

  PublisherDOI

• Experimental validation of the <scp>RESPONSE</scp> framework against cyberattacks on cyber‐physical process systems

  AIChE Journal · 2026-01-30

  articleOpen access

  Abstract This work experimentally validates the RESPONSE (Resilient Process cONtrol SystEm) framework as a solution for maintaining safe, continuous operation of cyber‐physical process systems under cyberattacks. RESPONSE implements a dual‐loop architecture that runs a networked online controller in parallel with a hard‐isolated offline controller, augmented by a control‐action monitor and controller state reconfiguration for bumpless transfer. Using a laboratory continuous stirred‐tank heater (CSTH), we emulate integrity (false‐data injection) and availability (DDoS) attacks, and benchmark RESPONSE against a parallel unconnected redundancy without controller reconfiguration across different controller tunings. Experiments show that RESPONSE (i) sustains stable operation without enforced shutdown when detection is delayed or absent, (ii) suppresses switching transients via integral/state manipulation, and (iii) accelerates restoration because the plant remains near set conditions. These results demonstrate a practical pathway to retrofit legacy systems with minimal modification, operationalizing the “respond/recover” functions of modern cybersecurity guidance directly within the control architecture.

  PublisherOA PDFDOI

• RESPONSE- a resilient framework to manage cyber-attacks on cyber-physical process systems

  Digital Chemical Engineering · 2025-04-26 · 4 citations

  articleOpen access

  • Presented a novel framework to address the challenge of cyberattacks on cyber-physical process systems (CPS). • The framework enhances process system continuity and recovery during and after cyber incidents. • The framework CPS security, reliability, and economic performance. • Case studies are presented to demonstrate the framework’s capabilities. This paper presents a novel framework – Res ilient P rocess c ON trol S yst E m (RESPONSE) - to address the critical challenge of cyberattacks on cyber-physical process systems (CPS). The RESPONSE emphasizes adaptability to existing systems, operational stability independent of detection mechanism reliability, and enhanced system continuity and recovery during and after cyber incidents. RESPONSE is built on the National Institute of Standards and Technology (NIST) cybersecurity recommendation by leveraging redundant control architecture, secure detection mechanism, and integral error manipulation to maintain safe operations under attack conditions. It has transformative potential for enhancing CPS security, reliability, and economic performance. A comparative analysis and case studies are presented to demonstrate the framework’s ability to mitigate cyber threats, minimize downtime, and ensure rapid recovery.

  PublisherDOI

• Fault Diagnosis in Chemical Reactors with Data-Driven Methods

  Industrial & Engineering Chemistry Research · 2025-03-08 · 6 citations

  articleOpen accessSenior authorCorresponding

  This study investigates fault diagnosis, encompassing fault detection, isolation, and estimation, with experimental data in a continuous stirred-tank reactor (CSTR) for the liquid-phase catalytic oxidation of 3-picoline with hydrogen peroxide. Two key faults were examined: coolant inlet temperature spikes (fault 1) and 3-picoline feed concentration decreases (fault 2). Data-driven methods, including random forest (RF) and k-nearest neighbors (KNN), successfully detected, isolated, and estimated faults under nominal conditions. However, both data-driven and model-based residual generators were disrupted by a shift in the heat transfer coefficient (U). An isolation forest (IF) algorithm was used for anomaly detection and model recalibration, restoring model-based performance. Updated data sets enabled RF and KNN to adapt effectively, demonstrating their scalability and adaptability. Experimental results highlight the strengths of both methods, advocating for a combined framework for robust fault diagnosis.

  PublisherOA PDFDOI

• Design of Linear Residual Generators for Combined Fault Detection and Estimation in Nonlinear Systems

  arXiv (Cornell University) · 2025-12-30

  preprintOpen accessSenior author

  A systematic method for the design of linear residual generators for combined fault detection and estimation in nonlinear systems is developed. The proposed residual generator is a linear functional observer built for an extended system that incorporates the fault dynamics from a linear exo-system, and in addition possesses disturbance-decoupling properties. Necessary and sufficient conditions for the existence of such residual generators for nonlinear systems are derived. As long as these conditions are satisfied, we obtain explicit design formulas for the residual generator. The results are illustrated through a chemical reactor case study, which demonstrates the effectiveness of the proposed methodology.

  PublisherDOI

• A numerical algorithm for maximal admissible set calculation and its application to fault-tolerant control of chemical reactors

  Computers & Chemical Engineering · 2025-05-02 · 3 citations

  articleSenior authorCorresponding
  PublisherDOI

• Observer-based diagnosis and predictive framework for sensor-fault tolerant control of process systems

  SSRN Electronic Journal · 2025-01-01

  preprintOpen accessSenior author
  PublisherDOI

• Model‐based fault diagnosis in closed‐loop safety‐critical chemical reactors: An experimental study

  AIChE Journal · 2025-05-22 · 2 citations

  articleOpen accessSenior authorCorresponding

  Abstract This experimental study investigates fault detection and estimation in a continuous stirred‐tank reactor (CSTR) system under closed‐loop feedback control, including an analysis of different manipulative inputs for temperature regulation. A novel fault diagnosis approach is proposed, combining residual signal analysis and T 2 statistic for real‐time fault detection and size estimation. The closed‐loop system demonstrated robust setpoint tracking and fault tolerance across a range of fault magnitudes. Residual signals serve as direct estimators of fault size, critical for adaptive control, while the T 2 statistic enhances reliability by identifying deviations from normal behavior with fault‐confidence thresholds. As a step towards fault‐tolerant control, the proposed methodology lays the groundwork for advanced control strategies that can ensure safe and efficient operation of chemical reactor systems.

  PublisherOA PDFDOI

• Observer-based diagnosis and predictive framework for sensor-fault tolerant control of process systems

  Journal of Process Control · 2025-12-19 · 2 citations

  articleOpen accessSenior authorCorresponding

  Sensors are ubiquitous in modern industrial systems, and are prone to faults due to harsh condition in which they are placed. Sensor faults can impact the product quality and operational safety as control loops heavily depends on the accuracy of sensor measurement feedback. In this paper, we propose active sensor-fault tolerant control (FTC) strategies that can take proactive measures during faults involving timely correction of faulty measurements, ensuring the system remains within predefined safety and quality constraints. The concept of maximal output admissible set is leveraged to determine acceptable operating set (AOS) which is the set of initial process states that meet safety and quality constraints at all times. To support decision-making, we introduce a novel critical fault function (CFF) that quantifies the fault size and time available before the system exits the AOS if no corrective action is taken. While the AOS and CFF are computed offline, the CFF is implemented online with real-time fault estimates to trigger measurement correction in time. A linear functional observer and a nonlinear state observer, combined with a predictive scheme is proposed to estimate fault size and enhance robustness during transient phase of observers. Alternatively, a bank of state observers is used for fault detection and isolation and subsequently the state observer estimator based on healthy sensors are utilized for state feedback in control loops. The proposed sensor FTC strategy is tested on an exothermic Continuous Stirred Tank Reactor (CSTR) as a case study. The results demonstrate the strategy's effectiveness in handling sensor faults, ensuring both quality and safety constraints are met. Thus, this paper contributes to the advancement of practical active sensor FTC ensuring the resilience of industrial systems. • A FTC algorithm is developed to maintain quality and safety constraints in the presence of sensor faults. • Proposes a predictive fault estimation scheme to estimate fault magnitude, enabling proactive control actions. • Proposes an alternative observer-bank approach for fault detection, isolation, and signal reconstruction.

  PublisherDOI

• On Functional Observability of Nonlinear Systems and the Design of Functional Observers with Assignable Error Dynamics

  IFAC-PapersOnLine · 2025-01-01 · 1 citations

  articleOpen access1st authorCorresponding

  This paper proposes a novel approach for designing functional observers for nonlinear systems, with linear error dynamics and assignable poles. Sufficient conditions for functional observability are first derived, leading to functional relationships between the Lie derivatives of the output to be estimated and the ones of the measured output. These are directly used in the proposed design of the functional observer. The functional observer is defined in differential input-output form, satisfying an appropriate invariance condition that emerges from the state-space invariance conditions of the literature. A concept of functional observer index is also proposed, to characterize the lowest feasible order of functional observer with pole assignment.

  PublisherDOI

Recent grants

• Process diagnostics and event-driven control for safety-critical chemical processes and plants

  NSF · $404k · 2021–2026

• Multi-rate Nonlinear Observers for Process Monitoring, with Application to Polymerization Reactors

  NSF · $350k · 2017–2022

• Digital Control of Nonlinear Processes

  NSF · $231k · 1994–1998

Frequent coauthors

• Raymond A. Wright

  53 shared

• Nikolaos Kazantzis

  Beck Institute for Cognitive Behavior Therapy

  39 shared

• Iasson Karafyllis

  National Technical University of Athens

  25 shared

• Masoud Soroush

  Drexel University

  24 shared

• Sunjeev Venkateswaran

  Texas A&M University

  16 shared

• M. Niemiec

  University of Bremen

  14 shared

• Pródromos Daoutidis

  University of Minnesota

  12 shared

• M. Ziyan Sheriff

  Texas A&M University at Qatar

  12 shared