SEARCH · Search NASA

Results for “Model Predictive Control”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,279 records · Page 71

All sources NASA NTRS DOE OSTI

NASA NTRS1977

Theory for passive microwave remote sensing of near-surface soil moisture

The theory of microwave thermal emission from a nonscattering half-space medium is developed for application to regions with nonuniform subsurface soil-moisture and temperature variations. A coherent stratified model is presented which is valid for nonuniform temperature profiles and rapidly varying moisture profiles, under which conditions the commonly used emissivity and radiative-transfer approaches become inaccurate. For naturally occurring profiles the stratified model gives more accurate results than the other approaches at frequencies below about 4 GHz. Experimental results from ground-based radiometric observations of a controlled target area compare systematically with brightness temperatures predicted from the theoretical model to within approximately 10 K. Results of dielectric-constant measurements of the sand are given at seven frequencies in the microwave range and for moisture contents in the range from 0% to 30% by volume. By using this model, the thermal microwave emission spectrum is computed for a number of representative moisture and temperature profiles in the frequency range from 0.25 to 25 GHz.

Njoku, E. G.↗

NASA NTRS1998

Predicting lettuce canopy photosynthesis with statistical and neural network models

An artificial neural network (NN) and a statistical regression model were developed to predict canopy photosynthetic rates (Pn) for 'Waldman's Green' leaf lettuce (Latuca sativa L.). All data used to develop and test the models were collected for crop stands grown hydroponically and under controlled-environment conditions. In the NN and regression models, canopy Pn was predicted as a function of three independent variables: shootzone CO2 concentration (600 to 1500 micromoles mol-1), photosynthetic photon flux (PPF) (600 to 1100 micromoles m-2 s-1), and canopy age (10 to 20 days after planting). The models were used to determine the combinations of CO2 and PPF setpoints required each day to maintain maximum canopy Pn. The statistical model (a third-order polynomial) predicted Pn more accurately than the simple NN (a three-layer, fully connected net). Over an 11-day validation period, average percent difference between predicted and actual Pn was 12.3% and 24.6% for the statistical and NN models, respectively. Both models lost considerable accuracy when used to determine relatively long-range Pn predictions (> or = 6 days into the future).

Non-NASA Center↗

DOE OSTI2023

Stream Temperature Prediction in a Shifting Environment: Explaining the Influence of Deep Learning Architecture

Stream temperature is a fundamental control on ecosystem health. Recent efforts incorporating process guidance into deep learning models for predicting stream temperature have been shown to outperform existing statistical and physical models. This performance is in part because deep learning architectures can actively learn spatiotemporal relationships that govern how water and energy propagate through a river network. However, exploration of how spatiotemporal awareness and process guidance influence a model's generalizability under shifting environmental conditions such as climate change is limited. Here, we use Explainable Artificial Intelligence (XAI) to interrogate how differing deep learning architectures affect a model's learned spatial and temporal dependencies, and how those learned dependencies affect a model's ability to maintain high accuracy when applied to unseen environmental conditions. Using the Delaware River Basin in the northeastern United States as a test case, we compare two spatiotemporally aware process–guided deep learning models for predicting stream temperature (a recurrent graph convolution network—RGCN, and a temporal convolution graph model—Graph WaveNet). Both models achieve equally high predictive performance when testing data are well represented in the training data (test root mean squared errors of 1.64°C and 1.65°C); however, Graph WaveNet significantly outperforms RGCN in 4 out of 5 experiments where test partitions represent different types of unseen environmental conditions. XAI results show that the architecture of Graph WaveNet leads to learned spatial relationships with greater fidelity to physical processes, and that this fidelity improves the generalizability of the model when applied to shifting and/or unseen environmental conditions.

54 ENVIRONMENTAL SCIENCES↗

NASA NTRS1981

Experimental validation of a numerical model predicting the charging characteristics of Teflon and Kapton under electron beam irradiation

The effect differential charging of spacecraft thermal control surfaces is assessed by studying the dynamics of the charging process. A program to experimentally validate a computer model of the charging process was established. Time resolved measurements of the surface potential were obtained for samples of Kapton and Teflon irradiated with a monoenergetic electron beam. Results indicate that the computer model and experimental measurements agree well and that for Teflon, secondary emission is the governing factor. Experimental data indicate that bulk conductivities play a significant role in the charging of Kapton.

Hazelton, R. C.↗

DOE OSTI2026

Comparison of structurally diverse simulation models for prediction of epidemic outcomes caused by a long-distance dispersed pathogen

Long-distance dispersal (LDD) pathogens pose substantial challenges for epidemic control due to their ability to generate new infection foci at great distances. While various modeling approaches have been developed to understand and manage such outbreaks, little work has compared how models of different structures behave under shared conditions. Here, in this study, we compare four structurally distinct epidemiological models — EPIMUL, GEMF, PoPS, and Warwick — each adapted to simulate the spread of wheat stripe rust (WSR), a wind-dispersed LDD pathogen, under identical epidemiological parameters and dispersal kernel. Using data from a controlled field experiment, we evaluate the ability of each model to replicate disease prevalence under nine intervention scenarios that vary in timing and culling area. While the models differ substantially in design — ranging from spatial grid-based to network-based and raster-based frameworks — the shared dispersal kernel allowed for close alignment in their predictions. All models accurately captured general epidemic trends, particularly the strong effect of early intervention on disease suppression. We qualitatively compared their behavioral responses across scenarios and also evaluated an ensemble prediction by averaging across model outputs. Our findings highlight how integrating shared epidemiological components into distinct modeling frameworks can improve consistency and accuracy, while reinforcing the importance of early culling in managing LDD pathogen outbreaks.

Dispersal kernel↗

DOE OSTI2017

Feedback control design for non-inductively sustained scenarios in NSTX-U using TRANSP

This paper examines a method for real-time control of non-inductively sustained scenarios in NSTX-U by using TRANSP, a time-dependent integrated modeling code for prediction and interpretive analysis of tokamak experimental data, as a simulator. The actuators considered for control in this work are the six neutral beam sources and the plasma boundary shape. To understand the response of the plasma current, stored energy, and central safety factor to these actuators and to enable systematic design of control algorithms, simulations were run in which the actuators were modulated and a linearized dynamic response model was generated. A multi-variable model-based control scheme that accounts for the coupling and slow dynamics of the system while mitigating the effect of actuator limitations was designed and simulated. Simulations show that modest changes in the outer gap and heating power can improve the response time of the system, reject perturbations, and track target values of the controlled values.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

NASA NTRS2000

Soil Moisture Memory in Climate Models

Water balance considerations at the soil surface lead to an equation that relates the autocorrelation of soil moisture in climate models to (1) seasonality in the statistics of the atmospheric forcing, (2) the variation of evaporation with soil moisture, (3) the variation of runoff with soil moisture, and (4) persistence in the atmospheric forcing, as perhaps induced by land atmosphere feedback. Geographical variations in the relative strengths of these factors, which can be established through analysis of model diagnostics and which can be validated to a certain extent against observations, lead to geographical variations in simulated soil moisture memory and thus, in effect, to geographical variations in seasonal precipitation predictability associated with soil moisture. The use of the equation to characterize controls on soil moisture memory is demonstrated with data from the modeling system of the NASA Seasonal-to-Interannual Prediction Project.

Koster, Randal D.↗

DOE OSTI2015

High Penetration of Photovoltaic Generation Study – Flagstaff Community Power (Final Technical Report, Results of Phases 2-5)

APS’s renewable energy portfolio, driven in part by Arizona’s Renewable Energy Standard (RES) currently includes more than 1100 MW of installed capacity, equating to roughly 3000 GWh of annual production. Overall renewable production is expected to grow to 6000 GWh by 2025. It is expected that distributed photovoltaics, driven primarily by lower cost, will contribute to much of this growth and that by 2025, distributed installations will account for half of all renewable production (3000GHW). As solar penetration increases, additional analysis may be required for routine utility processes to ensure continued safe and reliable operation of the electric distribution network. Such processes include residential or commercial interconnection requests and load shifting during normal feeder operations. Circuits with existing high solar penetration will also have to be studied and results will need to be evaluated for adherence to utility practices or strategy. Increased distributed PV penetration may offer benefits such as load offsetting, but it also has the potential to adversely impact distribution system operation. These effects may be exacerbated by the rapid variability of PV production. Detailed effects of these phenomena in distributed PV applications continue to be studied. Comprehensive, high-resolution electrical models of the distribution system were developed to analyze the impacts of PV on distribution circuit protection systems (including coordination and anti-islanding), predict voltage regulation and phase balance issues, and develop volt/VAr control schemes. Modeling methods were refined by validating against field measurements. To augment the field measurements, methods were developed to synthesize high resolution load and PV generation data to facilitate quasi-static time series simulations. The models were then extended to explore boundary conditions for PV hosting capability of the feeder and to simulate common utility practices such as feeder reconfiguration. The modeling and analysis methodology was implemented using open source tools and a process was developed to aid utility engineers in future interconnection requests. Methods to increase PV hosting capacity were also explored during the course of the study. A 700kVA grid-supportive inverter was deployed on the feeder and each grid support mode was demonstrated. Energy storage was explored through simulation and models were developed to calculate the optimum size and placement needed to increase PV hosting capacity. A tool was developed to aid planners in assigning relative costs and benefits to various strategies for increasing PV hosting capacity beyond current levels. Following the completion of the project, APS intends to use the tools and methods to improve the framework of future PV integration on its system. The tools and methods are also expected to aid other utilities to accelerate distributed PV deployment.

14 SOLAR ENERGY↗

DOE OSTI2023

A dynamic model of wind turbine yaw for active farm control

This paper presents a graph-based dynamic yaw model to predict the dynamic response of the hub-height velocities and the power of a wind farm to a change in yaw. The model builds on previous work where the turbines define the nodes of the graph and the edges represent the interactions between turbines. Advances associated with the dynamic yaw model include a novel analytical description of the deformation of wind turbine wakes under yaw to represent the velocity deficits and a more accurate representation of the interturbine travel time of wakes. The accuracy of the model is improved by coupling it with time- and space-dependent estimates of the wind farm inflow based on real-time data from the wind farm. The model is validated both statically and dynamically using large-eddy simulations. An application of the model is presented that incorporates the model into an optimal control loop to control the farm power output.

17 WIND ENERGY↗

DOE OSTI2015

Mathematical Modeling of Electrochemical Flow Capacitors

Electrochemical flow capacitors (EFCs) for grid-scale energy storage are a new technology that is beginning to receive interest. Prediction of the expected performance of such systems is important as modeling can be a useful avenue in the search for design improvements. Models based off of circuit analogues exist to predict EFC performance, but these suffer from deficiencies (e.g. a multitude of fitting constants that are required and the ability to analyze only one spatial direction at a time). In this paper mathematical models based off of three-dimensional macroscopic balances (similar to models for porous electrodes) are reported. Unlike existing three-dimensional porous electrode-based approaches for modeling slurry electrodes, advection (i.e., transport associated with bulk fluid motion) of the overpotential is included in order to account for the surface charge at the interface between flowing particles and the electrolyte. Doing so leads to the presence of overpotential boundary layers that control the performance of EFCs. These models were used to predict the charging behavior of an EFC under both flowing and non-flowing conditions. Agreement with experimental data was good, including proper prediction of the steady-state current that is achieved during charging of a flowing EFC. (C) The Author(s) 2015. Published by ECS. This is an open access article distributed under the terms of the Creative Commons Attribution Non-Commercial No Derivatives 4.0 License (CC BY-NC-ND, http://creativecommons.org/licenses/by-nc-nd/4.0/), which permits non-commercial reuse, distribution, and reproduction in any medium, provided the original work is not changed in any way and is properly cited. For permission for commercial reuse, please email: oa@electrochem.org. All rights reserved.

Hoyt, NC↗

DOE OSTI2021

A Physics-Constrained Deep Learning Model for Simulating Multiphase Flow in 3D Heterogeneous Porous Media

Physics-based simulators for multiphase flow in porous media emulate nonlinear processes with coupled physics, and usually require extensive computational resources for software development, maintenance and simulation execution. As a result, a huge demand exists for fast modeling of coupled processes in a wide range of subsurface applications including geological sequestration, hydrocarbon recovery and geothermal energy extraction. In this work, an efficient physics-constrained deep learning model is developed for solving multiphase flow in 3-Dimensional (3D) heterogeneous porous media. The model fully leverages the spatial topology predictive capability of convolutional neural networks, specifically U-Net with successive contracting and expansive steps, and is coupled with an efficient continuity-based smoother to predict flow responses that need spatial continuity. Furthermore, the transient regions are penalized to steer the training process such that the model can accurately capture flow in these regions. The model takes inputs including properties of porous media, fluid properties and well controls, and predicts the temporal-spatial evolution of the state variables (pressure and saturation). While maintaining the continuity of fluid flow, the 3D spatial domain is decomposed into 2D images for reducing training cost, and the decomposition results in an increased number of training data samples and better training efficiency. Additionally, a surrogate model is separately constructed as a postprocessor to calculate well flow rate based on the predictions of state variables from the deep learning model. We use the example of CO 2 injection into saline aquifers, and apply the physics-constrained deep learning model that is trained from physics-based simulation data and emulates the physics process. The model performs prediction with a speedup of ~ 1400 times compared to physics-based simulations, and the average temporal errors of predicted pressure and saturation plumes are 0.27% and 0.099% respectively. Furthermore, water production rate is efficiently predicted by a surrogate model for well flow rate, with a mean error less than 5%. Therefore, with its unique scheme to cope with the fidelity in fluid flow in porous media, the physics-constrained deep learning model can become an efficient predictive model for computationally demanding inverse problems or other coupled processes.

58 GEOSCIENCES↗

DOE OSTI2015

Bayesian network ensemble as a multivariate strategy to predict radiation pneumonitis risk

Purpose: Prediction of radiation pneumonitis (RP) has been shown to be challenging due to the involvement of a variety of factors including dose–volume metrics and radiosensitivity biomarkers. Some of these factors are highly correlated and might affect prediction results when combined. Bayesian network (BN) provides a probabilistic framework to represent variable dependencies in a directed acyclic graph. The aim of this study is to integrate the BN framework and a systems’ biology approach to detect possible interactions among RP risk factors and exploit these relationships to enhance both the understanding and prediction of RP. Methods: The authors studied 54 nonsmall-cell lung cancer patients who received curative 3D-conformal radiotherapy. Nineteen RP events were observed (common toxicity criteria for adverse events grade 2 or higher). Serum concentration of the following four candidate biomarkers were measured at baseline and midtreatment: alpha-2-macroglobulin, angiotensin converting enzyme (ACE), transforming growth factor, interleukin-6. Dose-volumetric and clinical parameters were also included as covariates. Feature selection was performed using a Markov blanket approach based on the Koller–Sahami filter. The Markov chain Monte Carlo technique estimated the posterior distribution of BN graphs built from the observed data of the selected variables and causality constraints. RP probability was estimated using a limited number of high posterior graphs (ensemble) and was averaged for the final RP estimate using Bayes’ rule. A resampling method based on bootstrapping was applied to model training and validation in order to control under- and overfit pitfalls. Results: RP prediction power of the BN ensemble approach reached its optimum at a size of 200. The optimized performance of the BN model recorded an area under the receiver operating characteristic curve (AUC) of 0.83, which was significantly higher than multivariate logistic regression (0.77), mean heart dose (0.69), and a pre-to-midtreatment change in ACE (0.66). When RP prediction was made only with pretreatment information, the AUC ranged from 0.76 to 0.81 depending on the ensemble size. Bootstrap validation of graph features in the ensemble quantified confidence of association between variables in the graphs where ten interactions were statistically significant. Conclusions: The presented BN methodology provides the flexibility to model hierarchical interactions between RP covariates, which is applied to probabilistic inference on RP. The authors’ preliminary results demonstrate that such framework combined with an ensemble method can possibly improve prediction of RP under real-life clinical circumstances such as missing data or treatment plan adaptation.

60 APPLIED LIFE SCIENCES↗

DOE OSTI2023

Aligning theoretical and empirical representations of soil carbon-to-nitrogen stoichiometry with process-based terrestrial biogeochemistry models

Soil carbon-nitrogen (C:N) stoichiometry acts as a control over decomposition and soil organic matter formation and loss, making it a key soil property for understanding ecosystem dynamics and projected ecosystems responses to global environmental change. However, the controls of soil C:N and how they respond to increasing pressures from global change agents are not fully understood. The “foundational” controls on soil C:N, namely plant and microbial C:N, have been used to predict soil C:N, but fail to accurately simulate all ecosystems and may be insufficient for predictions under global environmental change. Here we present an “emerging” representation of controls of soil C:N that includes plant-microbe-mineral feedbacks that have been shown to regulate soil C:N. We argue that including representation of these emerging drivers in process-based terrestrial biogeochemistry models, which include biological N fixation, mycorrhizae, priming, root exudation of organic acids, and mineralogy (including soil texture, mineral composition, and aggregation), will improve mechanistic representation of soil C:N and associated processes. Such improvements will produce models that will better simulate a variety of ecological states and predict soil C:N when global changes modify plant-microbe-mineral interactions. Here, we align our empirical understanding of controls of soil C:N with those controls represented in models, identifying contexts where emerging drivers might be particularly important to represent (e.g., priming and root exudation in nutrient-limited conditions) and areas of future work. Additionally, we show that implementing emerging drivers of soil C:N results in different simulated outcomes at steady state and in response to elevated atmospheric CO2. Our review and preliminary simulations support the need to incorporate emerging drivers of soil C:N into process-based terrestrial biogeochemistry models, allowing for both theoretical exploration of mechanisms and potentially more accurate predictions of land biogeochemical responses to global change.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2020

Sensor Recommendations for Long Term Monitoring of the F-Area Seepage Basins

In mid-2018, a new paradigm for long-term monitoring was developed after of decade of applied research projects funded by the Department of Energy’s office of Environmental Management Technology Development program. The program at SRNL was focused on transitioning complex environmental waste sites from active to passive remediations strategies. A key result of these studies was that the use of enhanced attenuation approaches at radiologically contaminated sites will result in the creation of secondary source areas in the subsurface that will require monitoring for decades. Alternative monitoring approaches are being developed and tested at the Savannah River Site’s F-Area Hazardous Waste Management Facility, the new paradigm provides innovative solutions that will significantly lower costs of monitoring through the coupling of data collection, machine learning and deterministic groundwater modeling. The foundation of this approach is a well-optimized network of sensors for measuring hydrogeochemical master variables that control, and therefore act as indicators of groundwater contaminant transport. By monitoring changes in the controlling master variables over time arising from geological and environmental shifts, predictive modelling can assist with identifying new strategies for ensuring regulatory requirements are met if trends toward conditions for potential remobilization of attenuated contaminants are detected. In this report, we evaluated commercially available single parameter sensor platforms (e.g., temperature/depth) and configurable multi-parameter sensor platforms (e.g., pH, oxidation-reduction potential, temperature, depth, dissolved oxygen, and conductivity). Each was scored using an optimization function based on how well the system supports the proposed long-term monitoring paradigm, in general, and the site-specific conditions at F-Area, in particular. Several viable sensor systems were identified. Of these, a combined platform including the In-Situ Aqua TROLL 500 multi-parameter sensor platform and the In-Situ temperature/depth sensor had the highest rating and was identified as the most suitable candidate for installation and monitoring of the master variables and potentiometric surface that control groundwater contaminant plumes emanating from the F-Area Seepage Basins. The discussion of recommended potential deployment locations builds upon recommendations made by Denham et al (2019).

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

NASA NTRS2020

Autonomous Real-Time Global Aerodynamic Modeling in the Frequency Domain

A method for autonomous real-time global aerodynamic modeling is developed and explained. Local real-time frequency-domain estimates of stability and control derivatives are assembled over the flight envelope as a function of nominal flight condition, accounting for uncertainty in the individual local model parameter estimates. Combining this with real-time estimates of low-frequency aerodynamics in the time domain produces a real-time global aerodynamic model with good prediction capability, simple evaluation of local stability and control derivatives, insight into the aerodynamic dependencies, and the ability to capture local variations in the aerodynamic dependencies throughout the flight envelope. The technique is demonstrated using data from an F-16 nonlinear simulation and flight test data from a subscale aircraft.

Eugene A Morelli↗

DOE OSTI2026

Quantifying Uncertainties in Earth's Energy Budget by Cloud Feedback and Ocean Heat Uptake Using E3SM-Slab Ocean Configurations

In order to improve predictive skills of Earth System Model, we need to better understand processes that control Earth's energy budget via ocean, atmosphere, and cryosphere interactions. Simulated energy budget in comprehensive Earth System Models shows a wide range, leading to large uncertainties in predicting Earth system dynamic and thermodynamic variations and associated social-economic impacts. Uncertainties in cloud feedbacks have been identified as the main cause of the large inter-model spread, but oceanic adjustments, especially those associated with ocean heat uptake (OHU) and the Atlantic Meridional Overturning Circulation (AMOC), also play an important role. In this proposed work, we focus on understanding the individual and combined roles of cloud feedbacks and ocean adjustments on modulating Earth's energy balance. This research is motivated by our overarching hypothesis that oceanic adjustment is a key source of uncertainty, in addition to those associated with the cloud feedbacks; further, the ocean adjustment and associated OHU work through the cloud feedbacks to modulate Earth's energy budget and temperature variations. We test this hypothesis using numerical experiments where we systematically enable and disable cloud feedbacks in conjunction with perturbations to OHU.

58 GEOSCIENCES↗

DOE OSTI2023

Implementation of AI/DEEP learning disruption predictor into a plasma control system

Abstract This paper reports on advances in the state‐of‐the‐art deep learning disruption prediction models based on the Fusion Recurrent Neural Network (FRNN) originally introduced in a 2019 NATURE publication [ https://doi.org/10.1038/s41586‐019‐1116‐4 ]. In particular, the predictor now features not only the “disruption score,” as an indicator of the probability of an imminent disruption, but also a “sensitivity score” in real time to indicate the underlying reasons for the imminent disruption. This adds valuable physics interpretability for the deep learning model and can provide helpful guidance for control actuators now implemented into a modern plasma control system (PCS). The advance is a significant step forward in moving from modern deep learning disruption prediction to real‐time control and brings novel AI‐enabled capabilities relevant for application to the future burning plasma ITER system. Our analyses use large amounts of data from JET and DIII‐D vetted in the earlier NATURE publication. In addition to “when” a shot is predicted to disrupt, this paper addresses reasons “why” by carrying out sensitivity studies. FRNN is accordingly extended to use more channels of information, including measured DIII‐D signals such as (i) the “n1rms” signal that is correlated with the n = 1 modes with finite frequency, including neoclassical tearing mode and sawtooth dynamics; (ii) the bolometer data indicative of plasma impurity control; and (iii) “q‐min”—the minimum value of the safety factor relevant to the key physics of kink modes. The additional channels and interpretability features expand the ability of the deep learning FRNN software to provide information about disruption subcategories as well as more precise and direct guidance for the actuators in a PCS.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2016

Predicting the 5-Year Risk of Biochemical Relapse After Postprostatectomy Radiation Therapy in ≥PT2, pN0 Patients With a Comprehensive Tumor Control Probability Model

Purpose: To fit the individual biochemical recurrence-free survival (bRFS) data from patients treated with postprostatectomy radiation therapy (RT) with a comprehensive tumor control probability (TCP) model. Methods and Materials: Considering pre-RT prostate-specific antigen (PSA) as a surrogate of the number of clonogens, bRFS may be expressed as a function of dose-per-fraction–dependent radiosensitivity (α{sub eff}), the number of clonogens for pre-RT PSA = 1 ng/mL (C), and the fraction of patients who relapse because of clonogens outside the treated volume (K), assumed to depend (linearly or exponentially) on pre-RT PSA and Gleason score (GS). Data from 894 node-negative, ≥pT2, pN0 hormone-naive patients treated with adjuvant (n=331) or salvage (n=563) intent were available: 5-year bRFS data were fitted grouping patients according to GS (<7:392, =7:383, >7:119). Results: The median follow-up time, pre-RT PSA, and dose were 72 months, 0.25 ng/mL, and 66.6 Gy (range 59.4-77.4 Gy), respectively. The best-fit values were 0.23 to 0.26 Gy{sup −1} and 10{sup 7} for α{sub eff} and C for the model considering a linear dependence between K and PSA. Calibration plots showed good agreement between expected and observed incidences (slope: 0.90-0.93) and moderately high discriminative power (area under the curve [AUC]: 0.68-0.69). Cross-validation showed satisfactory results (average AUCs in the training/validation groups: 0.66-0.70). The resulting dose-effect curves strongly depend on pre-RT PSA and GS. bRFS rapidly decreases with PSA: the maximum obtainable bRFS (defined as 95% of the maximum) declined by about 2.7% and 4.5% for each increment of 0.1 ng/mL for GS <7 and ≥7, respectively. Conclusions: Individual data were fitted by a TCP model, and the resulting best-fit parameters were radiobiologically consistent. The model suggests that relapses frequently result from clonogens outside the irradiated volume, supporting the choice of lymph-node irradiation, systemic therapy, or both for specific subgroups (GS <7: PSA >0.8-1.0 ng/mL; GS ≥7: PSA >0.3 ng/mL). Early RT should be preferred over delayed RT; the detrimental effect of PSA increase can never be fully compensated by increasing the dose, especially for patients with GS ≥7.

62 RADIOLOGY AND NUCLEAR MEDICINE↗