Richard B. Vilim

An identification scheme combining first principle knowledge, neural networks, and the likelihood function

An identification scheme is described for modeling uncertain systems. The method combines a physics-based model with a nonlinear mapping for capturing unmodeled physics and a statistical estimation procedure for quantifying any remaining process uncertainty. The technique has been used in predictive maintenance applications to detect operational changes of mechanical equipment by comparing the model output with the actual process output. Tests conducted on a peristaltic pump to detect incipient failure are described. The inclusion of unmodeled physics and a statistical representation of uncertainties results in lower false alarm and missed detection rates than other methods.

Nuclear Hybrid Energy Systems Regional Studies: West Texas & Northeastern Arizona

The primary objective of this study is to conduct a preliminary dynamic analysis of two realistic hybrid energy systems (HES) including a nuclear reactor as the main baseload heat generator (denoted as nuclear HES or nuclear hybrid energy systems [NHES]) and to assess the local (e.g., HES owners) and system (e.g., the electric grid) benefits attainable by the application of NHES in scenarios with multiple commodity production and high penetration of renewable energy. It is performed for regional cases - not generic examples - based on available resources, existing infrastructure, and markets within the selected regions. This study also briefly addresses the computational capabilities developed to conduct such analyses, reviews technical gaps, and suggests some research paths forward.

Tracking of weak radioactive sources in crowded venues

Monitoring for the clandestine transport of nuclear and radiological materials at large public gatherings or special events, such as the presidential inauguration, is one element in a strategy for preventing their subsequent dispersal. Being able to track these materials as they approach or move through such a venue can provide law-enforcement personnel with important information. The focus of this paper is on methods developed by the authors to meet the challenges inherent in the tracking problem. Results of laboratory experiments performed using the RadTrac prototype system are described.

Hierarchical Control of Reactor Inlet Temperature in Pool-Type Plants —II: Implementation and Results

The implementation of a computer-based controller for regulating reactor inlet temperature (RIT) in a pool-type power plant is described. The mathematical description of the controller is given in a companion paper. The elements of the control system are organized in a master-follower hierarchical architecture that takes advantage of existing in-plant hardware and software to minimize the need for plant modifications. Low-level control algorithms are executed on existing local digital controllers (followers) with the high-level algorithms executed on a new plant supervisory computer (master). A distributed computing strategy provides integration of the existing and additional computer platforms. The control system operated by having the master controller first estimate the secondary sodium flow needed to achieve a given RIT. The estimated flow is then used as a setpoint by the follower controller to regulate sodium flow using a motor-generator pump set. The control system has been implemented in a hardware-in-the-loop (HIL) setup and qualified for operation in the Experimental Breeder Reactor II at Argonne National Laboratory. The HIL results are provided.

Profitability Evaluation of Load-Following Nuclear Units with Physics-Induced Operational Constraints

AbstractThis work explores the technical challenges associated with flexible operation for nuclear power plants (NPPs) and evaluates whether a flexible operational mode could improve the profitability of nuclear units by allowing nuclear plant owners/operators to reduce output when prices are low and instead shift capacity to the ancillary services markets. As compared to conventional power plants, NPP flexible operation capabilities are affected by additional physics-induced constraints. Among the most limiting constraints is the negative reactivity insertion following every reactor power drop due to the increased concentration of xenon, a strong neutron poison. In this work, a previously available power system operation model based on mixed-integer linear programming optimization was improved by implementing a dedicated representation of these physics-induced constraints for pressurized water reactors (PWRs). Because the xenon-related constraint involves nonlinear governing dynamics, a dedicated paramet...

Treatment of Shielding in Real-Time Source Tracking Software

Abstract Within the homeland security and emergency response communities, there is a need for a low-profile system to detect, locate, and identify radioactive sources in real time. Such a system could be deployed for area monitoring around venues for special events. A system was developed at Argonne National Laboratory, called RADTRAC, which is based on a network of radiation detectors and advanced signal-processing algorithms. The initial implementation of RADTRAC did not account for dynamically changing shielding due to crowd movements. An algorithm was developed that utilizes the gamma-ray energy spectrum from each detector to estimate the amount of attenuation and scattering that is present between the source location (a priori unknown) and the detector location in real time. The attenuation and scattering estimations are then included in the maximum likelihood model to significantly improve the source localization solution. Results are presented for several test cases showing the improvement in the real-time source localization solution. This algorithm has been implemented into the current version of RADTRAC such that it now accounts for the effects of dynamically changing shielding and scattering due to crowd movements in real time in order to accurately determine the source location in crowded venues.

Impact of Active Control on Passive Safety Response Characteristics of Sodium-Cooled Fast Reactors: I—Theoretical Background

Abstract The interaction of the active control system with passive safety behavior is investigated for sodium-cooled fast reactors. A claim often made of advanced reactors is that they are passively safe against unprotected upset events. In practice, such upset events are not analyzed in the context of the plant control system, but rather the analyses are performed without considering the normally programmed response of the control system (open-loop approach). This represents an oversimplification of the safety case. The issue of passive safety override arises since the control system commands actuators whose motions have safety consequences. Depending on the upset involving the control system (operator error, active control system failure, or inadvertent control system override), an actuator does not necessarily go in the same direction as needed for safety. So neglecting to account for control system action during an unprotected upset is nonconservative from a safety standpoint. It is important then, during the design of the plant, to consider the potential for the control system to work against the inherent and safe regulating effects of purposefully engineered temperature feedbacks.

Integrating Physical Modeling, Neural Computing, and Statistical Analysis for On-Line Process Monitoring

Abstract Two basic approaches can be mentioned to model physical systems. One approach derives a model structure from the known physical laws. However, obtaining a model with the required fidelity may be difficult if the system is not well understood. A second approach is to employ a black-box structure to learn the implicit input-output relationships from measurements in which no particular attention is paid to modeling the underlying processes. A method that draws on the respective strengths of each of these two approaches is described. The technique integrates known first-principles knowledge derived from physical modeling with measured input-output mappings derived from neural processing to produce a computer model of a dynamical process. The technique is used to detect operational changes of mechanical equipment by statistically comparing, using a likelihood test, the predicted model output for the given measured input with the actual process output. Experimental results with a peristaltic pump are presented.

A method for measurement of delayed neutron parameters for liquid-metal-cooled power reactors

The trend toward increased reliance on passive features for power reactor safety makes it important to obtain the characteristics of the reactor system from measurements on the system. A method is described for solving for the delayed neutron parameters in a liquid-metal power reactor by fitting an analytic solution of the point-kinetics equations to the flux die-away from a dropped rod in an initially critical core. The method includes treatment of those conditions found in a power reactor that depart from those in a critical assembly experiment. These include a comparatively long rod drop time and a detector signal that instead of providing an integrated count rate is a sampled data signal proportional to the instantaneous fission power. The delayed neutron parameter values calculated from a rod drop experiment in the Experimental Breeder Reactor II are in agreement with values calculated using first principles and knowledge of core material composition and nuclear cross sections.

Innovative Control Strategy for Fast Runback Operational Transient Applied to SMRs

Abstract The recent interest in the Small Modular Reactor (SMR) for its potential increased economic competitiveness has focused attention in part on reducing operational costs to offset those plant costs that do not benefit from the economies of scale of large traditional units. Plant operation and maintenance economics are significantly driven by plant availability, which can be enhanced by means of innovative control strategies by avoiding unnecessary plant or unit trips. In this context, an effective strategy for achieving fast runback of a sodium-cooled SMR has been developed. In this work, after having defined and modeled a suitable control strategy by adopting the Petri nets formalism, a Model-based Predictive Control regulator has been developed in order to reduce as promptly as possible the power level, without scramming the reactor (fast runback) and possibly limiting the control rod contribution. Such flexibility could lead to significant savings in the operational costs of the reactor while also improving the system availability. The proposed procedure has been characterized by simulating the operational transients on both an oxide-fueled reactor and on a metal-fueled reactor, comparing the responses of the two different configurations and the respectively needed control rod contribution.