Search

Summary Methane hydrates and paraffin plugs on flexible lines are concerns in offshore production. They may stop wells for months, causing high financial losses. Sometimes, operators use depressurization techniques for hydrate removal. Another strategy is using coiled tubing or a similar unit to perform local heating or solvent injection. However, frequently these strategies are not successful. In those cases, a rig may perform the operation, or the line may be lost. We developed a robotic system to perform controlled local heating and remove obstructions. The system developed can access the line from the production platform. It uses a self-locking system to exert high traction forces. An umbilical with neutral buoyancy and low friction coefficient allows significant friction reduction. It allows moving upward and in pipes with a large number of curves. Coiled tubing and similar units cannot do that. Carbon fiber vessels and compact circuits give the flexibility to move inside 4-in. flexible pipes. In addition, a novel theoretical model allows the cable traction calculation using an evolution of the Euler-Eytelwein equation. Experimental tests validated this model using curved pipes, both empty and filled with fluid, and using different loads. Experimental tests also confirmed the external layer traction resistance. Furthermore, the carbon fiber vessels were pressure tested, indicating a collapse resistance of 57 MPa (8,300 psi). Besides, exhaustive tests of the onboard electronics and the surface control system guarantee the communication reliability. In addition, a theoretical model allowed the design of the 25 kN (5.6 kip) traction system considering the self-locking system, the contact with the wall, and a diameter range. Four prototypes allowed us to compare hydraulic and electric drive systems, validate the self-locking mechanism up to its limit, analyze the hydraulic system for leg opening and translation, and prove the traction capacity. Finally, a theoretical model allowed the local heating system and the temperature to increase. The experimental validation of the system on a cooled environment demonstrated its ability to increase temperature. Further, it allowed the obstruction removal in a controlled manner, avoiding damage to the polymeric layer of the flexible line.

To describe the performance of machine learning (ML) applied to predict future metabolic syndrome (MS), and to estimate lifestyle changes effects in MS predictions.We analyzed data from 17,182 adults attending a checkup program sequentially (37,999 visit pairs) over 17 years. Variables on sociodemographic attributes, clinical, laboratory, and lifestyle characteristics were used to develop ML models to predict MS [logistic regression, linear discriminant analysis, k-nearest neighbors, decision trees, Light Gradient Boosting Machine (LGBM), Extreme Gradient Boosting]. We have tested the effects of changes in lifestyle in MS prediction at individual levels.All models showed adequate calibration and good discrimination, but the LGBM showed better performance (Sensitivity = 87.8 %, Specificity = 70.2 %, AUC-ROC = 0.86). Causal inference analysis showed that increasing physical activity level and reducing BMI by at least 2 % had an effect of reducing the predicted probability of MS by 3.8 % (95 % CI = -4.8 %; -2.7 %).ML models based on data from a checkup program showed good performance to predict MS and allowed testing for effects of lifestyle changes in this prediction. External validation is recommended to verify models' ability to identify at-risk individuals, and potentially increase their engagement in preventive measures.

Orientadora: Profª. Dr.ª Carolina Aguiar Moreira Coorientadora: Profª. Dr.ª Victoria Zeghbi Cochenski Borba Dissertação (mestrado) - Universidade Federal do Paraná, Setor de Ciências da Saúde, Programa de Pós-Graduação em Medicina Interna e Ciências da Saúde. Defesa : Curitiba, 24/01/2022 Inclui referências Resumo: Introdução: As fraturas osteoporóticas são comuns e sua incidência está aumentando em todo o mundo. A primeira fratura dobra o risco de novas fraturas. Apesar disso, até 80% dos pacientes com fratura por fragilidade não são avaliados ou tratados para reduzir o risco de novas fraturas. Objetivos: Avaliar os resultados da instalação de um Fracture Liaison Service (FLS) na cidade de Curitiba, analisar as características clínicas dos pacientes atendidos no serviço nos seus primeiros 2 anos de funcionamento e estimar a redução do risco de fraturas nesta população. Métodos: O FLS acompanhou pacientes com mais de 50 anos de idade admitidos com fratura traumática de baixa energia entre janeiro de 2017 e abril de 2018. Esse acompanhamento a consistiu em avaliação médica completa, orientação nutricional e de atividade física e tratamento específico para osteoporose, se necessário. Resultados: Monitoramos e tratamos 135 pacientes, 123 mulheres e 12 homens, com média de idade de 74,8 anos no momento da fratura, por 14 meses. Quarenta por cento deles tinham fratura prévia e apenas 20,3% deles receberam tratamento para prevenir novas fraturas. Na análise de Kaplan-Meier, a incidência estimada de novas fraturas em 24 meses foi de 12,1% (IC 95% 7,2–20,8%), indicando que a porcentagem de pacientes sem novas fraturas devido à fragilidade óssea durante o tratamento foi estimada em 87,9% (95 % CI 79,2–92,8%). Conclusões: A avaliação clínica e o tratamento de pacientes com fratura por fragilidade para a prevenção de fraturas secundárias é eficaz em reduzir o risco de novas fraturas em pacientes de alto risco. Abstract: Background: Osteoporotic fractures are common, and its incidence are increasing worldwide. The first fracture doubles the risk of new fractures. Despite that, up to 80% of patients with a fragility fracture are not evaluated nor treated to reduce the risk of new fractures. Aims: To evaluate the results of the installation of a Fracture Liaison Service (FLS) in Curitiba - Brazil, to analyze the clinical characteristics of the patients attending the service in its first 2 years of operation and to estimate the fracture risk reduction ratio. Methods: The FLS managed patients older than 50 years old which were admitted with a low-energy trauma fracture between January 2017 to April 2018. This management consists in a full medical evaluation, nutritional and physical activity guidance and specific osteoporosis treatment, if needed. Results: We monitored and treated 135 patients, 123 woman and 12 men, with a mean age at the fracture of 74,8 years old, for 14 months. Forty per cent of them had a previous fracture and only 20,3% of them received treatment to prevent new fractures. On Kaplan-Meier analysis, the estimated incidence of new fractures over 24 months was 12.1% (95% CI 7.2–20.8%), indicating that the percentage of patients without new fractures due to bone fragility during treatment was estimated at 87.9% (95% CI 79.2–92.8%). Conclusions: The evaluation and treatment of patients who sustained a fragility fracture to prevent a secondary fracture is effective in reducing the risk of new fractures in high-risk patients.

Two major concerns in offshore production are Methane Hydrates and Paraffin Plugs. They may stop wells for months, causing high financial losses. Sometimes, depressurization techniques allow hydrate removal. Another strategy is using coiled tubing or a similar unit to perform local heating or solvent injection. However, frequently these strategies are not successful. In those cases, a rig may be a suitable but expensive solution, or the line may be lost. The present project aimed to develop a robotic system capable of performing controlled local heating for removing Paraffin and Methane Hydrates. The robotic system accesses the line from the production platform. It uses a peristaltic self-locking traction system to exert high traction forces. An umbilical with quasi-neutral buoyancy and low friction coefficient reduces the cable traction. It also allows moving upwards and in pipes with a large number of curves, something that coiled tubing and similar units cannot. Carbon fiber vessels and compact circuits allowed downsizing it to move inside 4-inch flexible pipes. Initially, a theoretical model for the local heating system allowed the evaluation of this strategy. A prototype allowed testing the system in a cooled environment. This heating system removes the obstruction in a controlled manner, avoiding damages to the polymeric layer of the flexible line. Simultaneously, a modified Euler-Eytelwein equation allowed the development of a theoretical model for cable traction. Experimental tests validated this model. Those tests used straight and curved pipes, both empty and filled with fluid and using different loads. Also, the 20 kN (4.3 kip) traction system was modeled theoretically considering the self-locking system, the contact with the wall, and a diameter range. Prototypes allowed the comparison between electric and hydraulic systems. Those prototypes also validated the traction capacity. Besides, force transmission from the traction system to the umbilical occurs through an external aramid layer. A Universal Testing Machine validated the traction resistance of the external layer. Furthermore, carbon fiber vessels protect the electronic circuits from oil and external pressure. The power electronics designed can provide up to 4kW for the motors to operate the hydraulic system. The onboard computer runs with a real-time operational system and, together with a sensors network, is responsible for monitoring the pressures, temperatures, currents and tensions throughout the entire robot. The fail-safe design allows the robot to operate without risks of catastrophic accidents and guarantees that it can be pulled out at any time. A pressure vessel validated the collapse resistance, reaching more than 700 bar (10.000 psi). In addition, exhaustive integration tests validated the onboard electronics and the surface control system. Finally, factory tests validated the umbilical design.

Methane Hydrates and Paraffin Plugs in flexible lines are concerns in offshore production. They may stop wells for months, causing high financial losses. Sometimes, operators use depressurization techniques for hydrate removal. Other strategy is using coiled tubing or a similar unit in order to perform local heating or solvent injection. However, frequently these strategies are not successful. In those cases, a rig may perform the operation or the line may be lost. This project developed a robotic system in order to perform a controlled local heating and remove obstructions. The robotic system developed is able to access the line from the production platform. It uses a self-locking system in order to exert high traction forces. An umbilical with neutral buoyancy and low friction coefficient allows significant drag reduction. It allows moving upwards and in pipes with a large number of curves. Coiled tubing and similar units cannot do that. Carbon fiber vessels and compact circuits give flexibility to move inside 4-inch flexible pipes. A novel theoretical model allows the cable traction calculation using an evolution of the Euler-Eytelwein equation. Experimental tests validated this model using curved pipes, both empty and filled with a fluid and using different loads. Experimental tests also validated the external layer traction resistance. Furthermore, the carbon fiber vessels were pressure tested, indicating a collapse resistance of more than 550 bar (8.000 psi). In addition, exhaustive tests of the onboard electronics and of the surface control system guarantee the communication reliability. Additionally, the 25 kN (5.6 kip) traction system was modeled theoretically considering the self-locking system, the contact with the wall and a diameter range. Four prototypes allowed to: a) compare hydraulic and electric drive systems, b) validate the self-locking mechanism up to its limit, c) analyze the hydraulic system for leg opening and translation and d) prove the traction capacity. Finally, a theoretical model for the local heating system was developed. The system experimental validation on a cooled environment demonstrated its capacity of increasing temperature. Furthermore, it allows the obstruction removal in a controlled manner, avoiding damage to the polymeric layer of the flexible line.

International audience; Restricted Boltzmann machines (RBMs) are energy-based neural networks which are commonly used as the building blocks for deep-architecture neural architectures. In this work, we derive a deterministic framework for the training, evaluation, and use of RBMs based upon the Thouless-Anderson-Palmer (TAP) mean-field approximation of widely connected systems with weak interactions coming from spin-glass theory. While the TAP approach has been extensively studied for fully visible binary spin systems, our construction is generalized to latent-variable models, as well as to arbitrarily distributed real-valued spin systems with bounded support. In our numerical experiments, we demonstrate the effective deterministic training of our proposed models and are able to show interesting features of unsupervised learning which could not be directly observed with sampling. Additionally, we demonstrate how to utilize our TAP-based framework for leveraging trained RBMs as joint priors in denoising problems.

In statistical learning for real-world large-scale data problems, one must often resort to "streaming" algorithms which operate sequentially on small batches of data. In this work, we present an analysis of the information-theoretic limits of mini-batch inference in the context of generalized linear models and low-rank matrix factorization. In a controlled Bayes-optimal setting, we characterize the optimal performance and phase transitions as a function of mini-batch size. We base part of our results on a detailed analysis of a mini-batch version of the approximate message-passing algorithm (Mini-AMP), which we introduce. Additionally, we show that this theoretical optimality carries over into real-data problems by illustrating that Mini-AMP is competitive with standard streaming algorithms for clustering., Comment: 19 pages, 4 figures

We consider the problem of reconstructing a signal from multi-layered (possibly) non-linear measurements. Using non-rigorous but standard methods from statistical physics we present the Multi-Layer Approximate Message Passing (ML-AMP) algorithm for computing marginal probabilities of the corresponding estimation problem and derive the associated state evolution equations to analyze its performance. We also give the expression of the asymptotic free energy and the minimal information-theoretically achievable reconstruction error. Finally, we present some applications of this measurement model for compressed sensing and perceptron learning with structured matrices/patterns, and for a simple model of estimation of latent variables in an auto-encoder., Comment: 5 pages, 1 figure

Summary When crude oil is stored in large tanks, invariably high-molecular-weight organic sediments are deposited. If left to accumulate, these deposits build up to form a sludge, which causes a reduction in the storage capacity. Routine industrial maintenance of storage tanks in vessels, terminals, and refineries unavoidably means the equipment's temporary inoperability. Furthermore, when conventional treatments are used to remove crude oil sludge, there is a potential for high environmental impact. Petrobras has developed a thermochemical process to remove organic deposits in submersed oil pipelines, wax damage in production reservoirs, and petroleum sludge removal from storage tanks. Based on a sludge removal case study, this paper describes the laboratory methodology to characterize the organic deposit physical/chemical properties, calculate thermochemical reaction kinetics and treatment dimensioning, and stage a physical simulation. It then reports the treatment's operational application to remove 800 m3 of organic deposit from an oil tanker storage tank. The process relies on the strong exothermic reaction between two nitrogen salts estimated at 90°C that also produces large volumes of nitrogen generating turbulence. This reaction heats the sludge, which melts, and irreversibly disperses in the organic solvent. This condition was predicted in a previously studied phases diagram. The method proved to be efficient, safe, and low cost compared to existing methods. The process' financial balance showed that the cost of the chemical reagents and operational facilities was covered by the value of the oil recovered from the organic deposit.

We consider the variational free energy approach for compressed sensing. We first show that the na\"ive mean field approach performs remarkably well when coupled with a noise learning procedure. We also notice that it leads to the same equations as those used for iterative thresholding. We then discuss the Bethe free energy and how it corresponds to the fixed points of the approximate message passing algorithm. In both cases, we test numerically the direct optimization of the free energies as a converging sparse-estimationalgorithm., Comment: 5 pages, 3 figures

Reputation systems seek to infer which members of a community can be trusted based on ratings they issue about each other. We construct a Bayesian inference model and simulate approximate estimates using belief propagation (BP). The model is then mapped onto computing equilibrium properties of a spin glass in a random field and analyzed by employing the replica symmetric cavity approach. Having the fraction of trustful nodes and environment noise level as control parameters, we evaluate the theoretical performance in terms of estimation error and the robustness of the BP approximation in different scenarios. Regions of degraded performance are then explained by the convergence properties of the BP algorithm and by the emergence of a glassy phase., Comment: 20 pages, 14 figures

Summary Well interventions, especially in offshore wells, are costly and usually very time-consuming to conduct. This is related mainly to the need for a rig or a similar unit to perform these operations. In this paper we propose the development of an autonomous robot for rigless well interventions. The robot will be inserted through the tree using a special tool. This proposal will result in significant cost reduction by eliminating the rig when performing light–workover operations in offshore locations. This reduces intervention time because a simpler unit might be used to deploy the robot. Examples of operations that could be performed by the robot are production logging, gas–lift valve replacement, sliding–sleeve opening and closing, and plug insertion or removal. To evaluate this proposal, a conceptual robot was designed. The feasibility of this conceptual robot was evaluated theoretically. After that, a simplified test robot was constructed to test critical aspects experimentally, in laboratories, and in a test well. The traction system, energy storage, and a positioning algorithm were evaluated. In this paper we will outline the intervention strategy proposed, perform a theoretical evaluation of this proposal on the basis of the conceptual robot, and present experimental results obtained with the simplified test robot.

We examine a class of stochastic deep learning models with a tractable method to compute information-theoretic quantities. Our contributions are three-fold: (i) we show how entropies and mutual informations can be derived from heuristic statistical physics methods, under the assumption that weight matrices are independent and orthogonally-invariant. (ii) We extend particular cases in which this result is known to be rigorously exact by providing a proof for two-layers networks with Gaussian random weights, using the recently introduced adaptive interpolation method. (iii) We propose an experiment framework with generative models of synthetic datasets, on which we train deep neural networks with a weight constraint designed so that the assumption in (i) is verified during learning. We study the behavior of entropies and mutual informations throughout learning and conclude that, in the proposed setting, the relationship between compression and generalization remains elusive.