Search NASA⌕ Search

Engineering topics

Maldonado Puente, Bryan

Publications and source records attributed to Maldonado Puente, Bryan.

At least 19 records

Transient Optimization of the Cryogenic Moderator System Controller at the Spallation Neutron Source for Improved Performance

The high-energy neutron beam generated at the Spallation Neutron Source (SNS) at Oak Ridge National Laboratory is moderated to use cold (slow) neutrons for scientific discoveries. The Cryogenic Moderator System (CMS) removes heat from the neutron beam using cryogenic hydrogen (H 2 ) moderators connected via heat exchangers to a helium (He) refrigeration loop that dissipates heat using a compressor-brake system. However, the CMS is affected by sporadic losses in beam power, referred to as "beam trips," as these events generate significant disturbances in cooling requirements. To accommodate the heat load transients during beam trips, the CMS uses a decentralized control strategy consisting of four flow valves and one electric heater adjusted by independent proportional-integral (PI) controllers. During the CMS’s initial commissioning, the PI gains were calibrated based only on tracking performance, overlooking their effectiveness in disturbance rejection. A data-driven, control-oriented closed-loop model was developed to recalibrate the PI gains and minimize the transient disturbances caused by beam trips. The model consists of three main components: (1) a physics-based model of the He refrigeration loop, (2) a machine-learning model of the cryogenic H 2 cooling trains, and (3) the control logic used for feedback set-point tracking. Experimental results showed that the recalibrated gains obtained in this study improved the CMS’s transient response during beam trips.

Maldonado Puente, Bryan↗

Optimization of prefabricated component installation using a real-time evaluator (RTE) connection locating system

Prefabrication promises to industrialize the construction industry. By constructing elements within a manufacturing environment, producers can better control quality and maximize production efficiency. Since the major adoption of prefabrication, a wide variety of prefabricated components have been produced for varying applications such as new construction and exterior wall retrofits. While the production processes of these prefabricated components have seen much innovation, the installation process has remained relatively unchanged for decades. To innovate the installation process with modern technologies, a real-time evaluator (RTE) has been developed to reduce the installation cost of prefabricated components by reducing installation time, decreasing rework, and improving accuracy. The RTE uses developed software solutions with off-the-shelf hardware to assist erectors in completing an installation by measuring the real-time positions of connections and prefabricated components, providing installation guidance through a graphical user interface, and monitoring the accumulated installation errors. An overview of the RTE and proposed workflow is presented. A connection locating system that guides users in expediting the installation of connections is introduced. Laboratory experiments were conducted to determine the accuracy improvement and time savings of the RTE in installing connections for prefabricated components. RTE enabled a time saving of up to 37% compared to traditional connection installation methods using handheld measurement tools.

Hayes, Nolan↗

Component pose reconstruction using a single robotic total station for panelized building envelopes

The construction industry can benefit greatly from increased automation in construction processes in terms of precision, time, and labor costs. This is especially true for prefabricated construction for either new buildings or envelope retrofits. Here, prefabricated components are manufactured offsite according to design specifications and installed onsite. Accurate information on the component’s position and orientation (pose) is needed to achieve this. A tool named “Real-Time Evaluator” (RTE), designed to autonomously track prefabricated components as they are being installed and provide real-time installation instructions, is currently under development using a single robotic total station. This is challenging since at least three points are needed to determine the pose of an object in space. To achieve this goal with a single robotic total station, two key algorithms were developed: the “resection” algorithm aligns the digital twin with the physical twin regardless of the location of the total station, and the “transformation” algorithm gives instructions of translations and rotations to installers to achieve the desired installation pose. The algorithms were evaluated experimentally on a lab-scale demonstration. Results show that the resection algorithm achieved an average error of < 3.1 mm, while the transformation algorithm predicted the rotation angle along a single axis with an error of < 0.5◦.

Tang, Mengjia↗

Flat and Level Analysis Tool (FLAT) for real-time automated segmentation and analysis of concrete slab point clouds

In the United States, the flatness and levelness of concrete floors during construction is traditionally specified by a maximum allowable gap under a 3 meter straightedge. However, the straightedge method is inexact and rarely representative of the entire floor since the technician is free to choose any location on the floor to perform the measurement. In cases requiring a higher degree of precision and repeatability, concrete floor flatness and levelness can be measured using the standard test method ASTM E1155. With the recent introduction of advanced surveying instruments such as robotic theodolites and terrestrial laser scanners (TLS), the means now exist to modernize and expedite the measurement of floor flatness and levelness. This paper details the development and demonstration of a digital tool, named the Flat and Level Analysis Tool (FLAT), to automate and expedite the segmentation and analysis of flatness and levelness from dense point cloud data of concrete floor slabs. Segmentation algorithms were developed using unsupervised machine learning to extract the set of points belonging to the concrete floor slab from a full 360 scan of a construction site. After segmentation, automated analysis algorithms report the results according to the standard method. The developed algorithms were demonstrated on a dense point cloud captured from a concrete slab-on-grade at a construction site. Results show that the digital tool can quickly provide estimates for floor flatness and levelness with minimal human involvement with comparable accuracy to manual methods.

Hayes, Nolan↗

Cable-Driven Parallel Robot (CDPR) for Panelized Envelope Retrofits: Feasible Workspace Analysis

Recent decades have seen remarkable progress in the field of robotic-assisted construction. Cable-driven parallel robots (CDPRs) emerge as promising tools for automating construction processes, due to their advantageous features such as scalability, reconfigurability, compact design, and high payload-to-weight ratio. This paper uses a simple static model to determine the feasibility of a CDPR for overclad panel installation in building envelope retrofits. Given that the building facade needs to be a subset of the CDPR’s wrench-feasible workspace, we focus on the sensitivity of the workspace concerning various cable arrangements and CDPR frame sizes (e.g., height and width extensions). Our analysis indicates that no cable arrangement satisfies the requirement of complete facade coverage and avoids cable-to-panel collisions. Thus, frame extension is needed to enhance coverage. However, in densely populated areas where width extension is limited by space constraints, height extension alone is insufficient to guarantee full facade coverage. This paper pioneers the investigation of CDPRs for panelized envelope retrofits, showcasing their advantages and limitations and paving the way for further research and development.

Liu, Yifang↗

Data-Driven Modeling of a High Capacity Cryogenic System for Control Optimization

The Cryogenic Moderator System (CMS) is responsible for maintaining a steady flow of cold neutrons for numerous physics experiments at the Spallation Neutron Source (SNS) in Oak Ridge National Laboratory (ORNL). Sudden losses in beam power, known as beam trips, cause a major disturbance to the CMS due to large step changes in cooling demands. Ongoing efforts on upgrading the neutron beam power from 1.4 to 2.0MW are expected to generate larger transients that can further strain the CMS subsystems if they are not properly controlled. To manage such disturbances, four flow valves and one electric heater are adjusted by five decentralized proportional-integral-derivative (PID) controllers. However, the original PID gains were calibrated empirically based only on tracking performance and not based on disturbance rejection. To address this issue without compromising current CMS operations, a control-oriented model was developed to recalibrate the PID controllers offline. The zero-dimensional (0-D) model was based on simple physics-based principles and data-driven system identification techniques. The CMS was broken into several subsystems for analysis, each of which corresponds to a parametric model tied to the thermodynamic states of the working fluid. The model parameters were identified using the nonlinear least squares method where the residuals were calculated from available sensor data. Simulation results show that the proposed model can capture the dynamics of the CMS at steady state and during beam trips.

Maldonado Puente, Bryan↗

Analysis of Real-World Preignition Data Using Neural Networks

Increasing adoption of downsized, boosted, spark-ignition engines has improved vehicle fuel economy, and continued improvement is desirable to reduce carbon emissions in the near-term. However, this strategy is limited by damaging preignition events which can cause hardware failure. Research to date has shed light on various contributing factors related to fuel and lubricant properties as well as calibration strategies, but the causal factors behind an individual preignition cycle remain elusive. If actionable precursors could be identified, mitigation through active control strategies would be possible. This paper uses artificial neural networks to search for identifiable precursors in the cylinder pressure data from a large real-world data set containing many preignition cycles. It is found that while follow-up preignition cycles in clusters can be readily predicted, the initial preignition cycle is not predictable based on features of the cylinder pressure. Further, this indicates that the alternating pattern of preignition cycles within clusters is influenced by the thermodynamic state as reflected in the pressure, but that the trigger for the initial preignition cycle is not thermodynamic in nature, but more likely tied to a critical threshold in the chemistry of the fuel/lubricant mixture in the upper crevice or other factors related to the presence of an ignition source.

33 ADVANCED PROPULSION SYSTEMS↗

Dynamic systems modeling of the spallation neutron source cryogenic moderator system to optimize transient control and prepare for power upgrades

Through support of the US Department of Energy's Office of Basic Energy Sciences, Oak Ridge National Laboratory has begun applying machine learning methods to improve accelerator and target performance of the Spallation Neutron Source (SNS). These methods are being applied to the control optimization and power upgrade of the SNS Cryogenic Moderator System (CMS). A numerical model of the CMS has been developed to study these optimizations and system modifications using EcosimPro. This paper compares steady-state and transient numerical results with experimental data. Control optimization studies focused on dampening mass flow, temperature, and pressure fluctuations during sudden losses of accelerator beam power. This analysis was conducted by adjusting five proportional-integral-derivative controllers connected to four flow control valves and one heater. Future efforts include power uprate studies focused on increasing the CMS cooling capacity. The current accelerator power is 1.4 MW; the first target station is being upgraded to 2.0 MW as part of the Proton Power Upgrade effort. The CMS cooling capacity is sufficient for 2.0 MW operation.

47 OTHER INSTRUMENTATION↗

Knowledge-Informed Uncertainty-Aware Machine Learning for Time Series Forecasting of Dynamical Engineered Systems

The high complexity and multiscale nature of many engineered systems—such as those in nuclear power plants—make representing and forecasting their dynamic behavior challenging. Physics-based models can be overly complex and computationally intractable, whereas machine learning (ML) tools are often data-hungry and prone to unphysical solutions. This study proposes a knowledge-informed ML-aided hybrid residual modeling approach that offers accurate and efficient time series forecasting for the operation of dynamical engineered systems. Hybrid residual modeling entails a baseline solution from domain knowledge and known physics expressions about the system dynamics integrated with an ML model to capture undiscovered information from the mismatch (i.e., residuals) between true states from measurements and baseline-predicted outputs. This study further quantifies the ML model uncertainty to provide trustworthy solutions. Real-time operational data from thermal-hydraulic flow loops of the cryogenic moderator system in Oak Ridge National Laboratory’s Spallation Neutron Source facility were used to demonstrate the potential of knowledge-informed uncertainty-aware ML in real-world applications. The state variables of the cryogenic helium loop were modeled with (1) first principles–based system identification (sysID), (2) long short-term memory (LSTM) neural network, and (3) hybrid sysID (baseline) + LSTM (residual). The superior predictive capability of the sysID+LSTM model versus stand-alone sysID and LSTM is confirmed by average performance metrics and individual data points across different prediction horizons. By creating a robust representation of the underlying physical system, the widely applicable hybrid residual modeling approach will enable the future development of digital twins for performance prediction, prognostics, and operation control.

Zhao, Xingang↗

Automatic point Cloud Building Envelope Segmentation (Auto-CuBES) using Machine Learning

Modern retrofit construction practices use 3D point cloud data of the building envelope to obtain the as-built dimensions. However, manual segmentation by a trained professional is required to identify and measure window openings, door openings, and other architectural features, making the use of 3D point clouds labor-intensive. In this study, the Automatic point Cloud Building Envelope Segmentation (Auto-CuBES) algorithm is described, which can significantly reduce the time spent during point cloud segmentation. The Auto-CuBES algorithm inputs a 3D point cloud generated by commonly available surveying equipment and outputs a wire-frame model of the building envelope. Unsupervised machine learning methods were used to identify facades, windows, and doors while minimizing the number of calibration parameters. Additionally, Auto-CuBES generates a heat map of each facade indicating non-planar characteristics that are crucial for the optimization of connections used in overclad envelope retrofits. With a scan resolution of 3 mm, the resulting window dimensions showed a mean absolute error of 4.2 mm compared to manual laser measurements.

Maldonado Puente, Bryan↗

Automated tracking of prefabricated components for areal-time evaluator to optimize and automateinstallation

Trade associations for prefabricated construction estimate that about 50% of prefabricated wall projects have alignment problems that lead to defects and rework. Additionally, component installation times average between 30 and 60 minutes per component. To address these issues, a real-time evaluator (RTE) system was introduced to decrease cost and automate prefabricated component installation by reducing the installation time, decreasing rework, and enhancing energy performance through higher installation quality. The RTE uses commonly available hardware and software to perform autonomous tracking to measure the real-time location and orientation of components as they are crane-lifted and installed. The hardware, software, and algorithms that allow the autonomous tracking of components are detailed. An algorithm to automate the initial search for a component with three attached retroreflectors is proposed. Algorithms to automate the measurement of component position and orientation are also proposed. Simple lab-scale proof-of-concept experiments were conducted to assess the algorithms for automation of component searching, measurement of real-time movement, and measurement of component orientation. With additional development, the system can be used as a tool to generate the commands for autonomous crane operation or single-task construction robots.

Hayes, Nolan↗

Automatic Segmentation of Building Envelope Point Cloud Data Using Machine Learning

About 50% of buildings in the US were constructed before energy codes were introduced. Modular overclad panel retrofits, in which a new envelope is constructed over the existing building, are a promising solution given that it minimizes occupant disruption and shortens construction time at the jobsite. Current state-of-the-art retrofit panel layout and dimensioning consists of three steps: 1) 3D point cloud data generation of the building envelope using commonly available surveying equipment, 2) manual segmentation of 3D point cloud data by a trained professional to identify and dimension window openings, door openings, and other architectural features, and 3) modular panel layout optimization and dimensioning by an architect or engineer. Among these steps, the second one remains the most difficult and costly because it is very labor-intensive. We propose a methodology to automatically label 3D point cloud data to reduce the time and expense spent in manual segmentation. Machine learning methods were employed to classify the point cloud data into distinct groups, each of which corresponds to different features of the building envelope. After classification, a segmentation algorithm was developed to perform boundary detection and separate the components of the façade. Finally, the algorithm returns the relative positions and dimensions of the features in the building envelope. The measurements obtained with the proposed automated method were compared against the actual dimensions to determine the overall algorithm accuracy. The proposed algorithm can then be used to reduce manual efforts for 3D point cloud labeling before modular panel layout optimization is performed.

Maldonado Puente, Bryan↗

Artificial Neural Networks for In-Cycle Prediction of Knock Events

Downsized turbocharged engines have been increasingly popular in modern light-duty vehicles due to their fuel efficiency benefits. However, high power density in such engines is achieved thanks to high in-cylinder pressure and temperature conditions that increase knock propensity. Next-cycle control has been studied as a method to reduce the damaging effects of knock by operating the engine in a low knock probability condition. This exploratory study looks at the feasibility of in-cycle knock prediction as a tool for advanced knock control algorithms. A methodology is proposed to 1) choose in-cycle features of the pressure trace that highly correlate with knock events and 2) train artificial neural networks to predict in-cycle knock events before knock onset. The methodology was validated at different operating conditions and different levels of generalization. Precision and recall were used as metrics to evaluate the binary classifier. However, the Fowlkes-Mallows (FM) index was used to compare the result of the clustering algorithm at different operating conditions. The results showed a maximum FM index of 0.7 when the prediction was done at knock onset and a minimum FM index of 0.45 when the prediction was done at spark timing.

42 ENGINEERING↗

Next-Cycle Optimal Dilute Combustion Control via Online Learning of Cycle-to-Cycle Variability Using Kernel Density Estimators

Dilute combustion using exhaust gas recirculation (EGR) presents a cost-effective method for increasing the efficiency of spark-ignition (SI) engines. However, the maximum amount of EGR that can be used at a given condition is limited by a rapid increment of cycle-to-cycle variability (CCV). This study describes a methodology to design a model-based stochastic optimal controller to adjust the cycle-to-cycle fuel injection quantity in order to reduce CCV and further extend the dilute limit. Given the complexity and chaotic nature of combustion events, the controller was enhanced with online learning in order to identify the statistical properties of combustion efficiency, which are needed to generate predictions for next-cycle events. This study showed that a kernel density estimator (KDE) can be used to learn the combustion properties in real time and can be incorporated into the feedback policy in order to calculate the optimal control command. Experimental results suggested that the dilute limit can be extended from 18.5% to 21% EGR fraction at an operating condition relevant for highway cruising. Additionally, the proposed controller can achieve a large CCV reduction with less fuel enrichment compared to previous methods, overall contributing to an increase in 0.2% indicated fuel conversion efficiency.

33 ADVANCED PROPULSION SYSTEMS↗

Artificial-intelligence-based prediction and control of combustion instabilities in spark-ignition engines

In recent years, as engine control strategies have grown increasingly sophisticated in a continued drive for increasing efficiency and reducing emissions, engine operation has been pushed into regimes that are limited by combustion instabilities. These instabilities produce undesirable abnormal combustion events, which pose barriers to further improvement in engine efficiency. Many of the phenomena involved are difficult to model or control using traditional, purely physics-based models and reactive control approaches. Artificial intelligence (AI) techniques offer some promise for achieving more effective combustion stability control, especially when appropriately applied within a physics-informed framework. This chapter illustrates the current state-of-the-art in applying AI to combustion stability control and examines three case studies with application to the dilute stability limit in spark-ignition engines to illustrate the utility and limitations of AI in these applications.

Maldonado Puente, Bryan↗

Real-Time Evolution and Deployment of Neuromorphic Computing at The Edge

Extremely low power neuromorphic systems are well-suited for deployment to the edge for many applications. In many use cases of neuromorphic computing for control, a spiking neural network is trained off-line using a simulation and then deployed to a neuromorphic system at the edge, where it will operate without ongoing training or learning. However, it may be desirable to continue training or learning at the edge to refine or adapt to the real-world system. In this work, we propose an approach for performing real-time evolutionary optimization for spiking neural networks for neuromorphic deployment at the edge. In particular, we propose a combination of simulation and real-world evaluations, along with feedback from the real-world environment, to train spiking neural networks for continuous deployment to the edge. We show that the real-time evolution at the edge approach achieves comparable performance to an evolution approach that requires constant evaluation in the realworld environment.

Schuman, Catherine↗

Next-Cycle Optimal Fuel Control for Cycle-to-Cycle Variability Reduction in EGR-Diluted Combustion

In this simulation study, cycle-to-cycle fuel control was used to reduce CCV by injecting additional fuel in operating conditions with sporadic misfires and partial burns. An optimal control policy was proposed that utilizes 1) a physics-based model that tracks in-cylinder gas composition and 2) a one-step-ahead prediction of the combustion efficiency based on a kernel density estimator. The optimal solution, however, presents a tradeoff between the reduction in combustion CCV and the increase in fuel injection quantity required to stabilize the charge. Such a tradeoff can be ad- just by a single parameter embedded in the cost function.

Maldonado, BryanP. [Oak Ridge National Lab. (ORNL)↗

Dilute Combustion Control Using Spiking Neural Networks

Dilute combustion with exhaust gas recirculation (EGR) in spark-ignition engines presents a cost-effective method for achieving higher levels of engine efficiency. At high levels of EGR, however, cycle-to-cycle variability (CCV) of the combustion process is exacerbated by sporadic occurrences of misfires and partial burns. Previous studies have shown that temporal deterministic patterns emerge at such conditions and certain combustion cycles have a significant influence over future events. Due to the complexity of the combustion process and the nature of CCV, harnessing all the deterministic information for control purposes has remained challenging even with physics based 0-D, 1-D, and high-fidelity computational fluid dynamics (CFD) models. In this study, we present a data-driven approach to optimize the combustion process by controlling CCV adjusting the cycle-to-cycle fuel injection quantity. Readily available data from in-cylinder pressure was used to train a spiking neural network (SNN) which learns the optimal way to manage fuel injection in order to reduce CCV while maintaining acceptable levels of fuel consumption. SNNs are particularly well suited for powertrain control applications due to their ability to be deployed on FPGA-based neuromorphic hardware which are small, inexpensive, and have a low power demand. The high-performance computing (HPC) resources of Oak Ridge National Laboratory were used to run an evolutionary-based training approach for choosing the best SNN configuration that minimizes the size of the network while achieving the desired goal. The neuromorphic hardware with the optimized SNN deployed was connected to the rapid prototyping engine control system for real-time control implementation and tested on a single cylinder version of a GM LNF 4-cylinder engine. The results show a significant reduction of CCV with a small percentage of additional fuel used to stabilize the charge.

33 ADVANCED PROPULSION SYSTEMS↗