Dynamic safety analysis of managed pressure drilling operations

The exploration and development of oil and gas reserves located in harsh offshore environments are characterized with high risk. Some of these reserves would be uneconomical if produced using conventional drilling technology due to increased drilling problems and prolonged non-productive time. Seeking new ways to reduce drilling cost and minimize risks has led to the development of Managed Pressure Drilling techniques. Managed pressure drilling methods address the drawbacks of conventional overbalanced and underbalanced drilling techniques. As managed pressure drilling techniques are evolving, there are many unanswered questions related to safety and operating pressure regimes. Quantitative risk assessment techniques are often used to answer these questions. Quantitative risk assessment is conducted for the various stages of drilling operations – drilling ahead, tripping operation, casing and cementing. A diagnostic model for analyzing the rotating control device, the main component of managed pressure drilling techniques, is also studied. The logic concept of Noisy-OR is explored to capture the unique relationship between casing and cementing operations in leading to well integrity failure as well as its usage to model the critical components of constant bottom-hole pressure drilling technique of managed pressure drilling during tripping operation. Relevant safety functions and inherent safety principles are utilized to improve well integrity operations. Loss function modelling approach to enable dynamic consequence analysis is adopted to study blowout risk for real-time decision making. The aggregation of the blowout loss categories, comprising: production, asset, human health, environmental response and reputation losses leads to risk estimation using dynamically determined probability of occurrence. Lastly, various sub-models developed for the stages/sub-operations of drilling operations and the consequence modelling approach are integrated for a holistic risk analysis of drilling operations.

Bayesian Inference in Large-scale Problems

Many modern applications fall into the category of "large-scale" statistical problems, in which both the number of observations n and the number of features or parameters p may be large. Many existing methods focus on point estimation, despite the continued relevance of uncertainty quantification in the sciences, where the number of parameters to estimate often exceeds the sample size, despite huge increases in the value of n typically seen in many fields. Thus, the tendency in some areas of industry to dispense with traditional statistical analysis on the basis that "n=all" is of little relevance outside of certain narrow applications. The main result of the Big Data revolution in most fields has instead been to make computation much harder without reducing the importance of uncertainty quantification. Bayesian methods excel at uncertainty quantification, but often scale poorly relative to alternatives. This conflict between the statistical advantages of Bayesian procedures and their substantial computational disadvantages is perhaps the greatest challenge facing modern Bayesian statistics, and is the primary motivation for the work presented here.

Two general strategies for scaling Bayesian inference are considered. The first is the development of methods that lend themselves to faster computation, and the second is design and characterization of computational algorithms that scale better in n or p. In the first instance, the focus is on joint inference outside of the standard problem of multivariate continuous data that has been a major focus of previous theoretical work in this area. In the second area, we pursue strategies for improving the speed of Markov chain Monte Carlo algorithms, and characterizing their performance in large-scale settings. Throughout, the focus is on rigorous theoretical evaluation combined with empirical demonstrations of performance and concordance with the theory.

One topic we consider is modeling the joint distribution of multivariate categorical data, often summarized in a contingency table. Contingency table analysis routinely relies on log-linear models, with latent structure analysis providing a common alternative. Latent structure models lead to a reduced rank tensor factorization of the probability mass function for multivariate categorical data, while log-linear models achieve dimensionality reduction through sparsity. Little is known about the relationship between these notions of dimensionality reduction in the two paradigms. In Chapter 2, we derive several results relating the support of a log-linear model to nonnegative ranks of the associated probability tensor. Motivated by these findings, we propose a new collapsed Tucker class of tensor decompositions, which bridge existing PARAFAC and Tucker decompositions, providing a more flexible framework for parsimoniously characterizing multivariate categorical data. Taking a Bayesian approach to inference, we illustrate empirical advantages of the new decompositions.

Latent class models for the joint distribution of multivariate categorical, such as the PARAFAC decomposition, data play an important role in the analysis of population structure. In this context, the number of latent classes is interpreted as the number of genetically distinct subpopulations of an organism, an important factor in the analysis of evolutionary processes and conservation status. Existing methods focus on point estimates of the number of subpopulations, and lack robust uncertainty quantification. Moreover, whether the number of latent classes in these models is even an identified parameter is an open question. In Chapter 3, we show that when the model is properly specified, the correct number of subpopulations can be recovered almost surely. We then propose an alternative method for estimating the number of latent subpopulations that provides good quantification of uncertainty, and provide a simple procedure for verifying that the proposed method is consistent for the number of subpopulations. The performance of the model in estimating the number of subpopulations and other common population structure inference problems is assessed in simulations and a real data application.

In contingency table analysis, sparse data is frequently encountered for even modest numbers of variables, resulting in non-existence of maximum likelihood estimates. A common solution is to obtain regularized estimates of the parameters of a log-linear model. Bayesian methods provide a coherent approach to regularization, but are often computationally intensive. Conjugate priors ease computational demands, but the conjugate Diaconis--Ylvisaker priors for the parameters of log-linear models do not give rise to closed form credible regions, complicating posterior inference. In Chapter 4 we derive the optimal Gaussian approximation to the posterior for log-linear models with Diaconis--Ylvisaker priors, and provide convergence rate and finite-sample bounds for the Kullback-Leibler divergence between the exact posterior and the optimal Gaussian approximation. We demonstrate empirically in simulations and a real data application that the approximation is highly accurate, even in relatively small samples. The proposed approximation provides a computationally scalable and principled approach to regularized estimation and approximate Bayesian inference for log-linear models.

Another challenging and somewhat non-standard joint modeling problem is inference on tail dependence in stochastic processes. In applications where extreme dependence is of interest, data are almost always time-indexed. Existing methods for inference and modeling in this setting often cluster extreme events or choose window sizes with the goal of preserving temporal information. In Chapter 5, we propose an alternative paradigm for inference on tail dependence in stochastic processes with arbitrary temporal dependence structure in the extremes, based on the idea that the information on strength of tail dependence and the temporal structure in this dependence are both encoded in waiting times between exceedances of high thresholds. We construct a class of time-indexed stochastic processes with tail dependence obtained by endowing the support points in de Haan's spectral representation of max-stable processes with velocities and lifetimes. We extend Smith's model to these max-stable velocity processes and obtain the distribution of waiting times between extreme events at multiple locations. Motivated by this result, a new definition of tail dependence is proposed that is a function of the distribution of waiting times between threshold exceedances, and an inferential framework is constructed for estimating the strength of extremal dependence and quantifying uncertainty in this paradigm. The method is applied to climatological, financial, and electrophysiology data.

The remainder of this thesis focuses on posterior computation by Markov chain Monte Carlo. The Markov Chain Monte Carlo method is the dominant paradigm for posterior computation in Bayesian analysis. It has long been common to control computation time by making approximations to the Markov transition kernel. Comparatively little attention has been paid to convergence and estimation error in these approximating Markov Chains. In Chapter 6, we propose a framework for assessing when to use approximations in MCMC algorithms, and how much error in the transition kernel should be tolerated to obtain optimal estimation performance with respect to a specified loss function and computational budget. The results require only ergodicity of the exact kernel and control of the kernel approximation accuracy. The theoretical framework is applied to approximations based on random subsets of data, low-rank approximations of Gaussian processes, and a novel approximating Markov chain for discrete mixture models.

Data augmentation Gibbs samplers are arguably the most popular class of algorithm for approximately sampling from the posterior distribution for the parameters of generalized linear models. The truncated Normal and Polya-Gamma data augmentation samplers are standard examples for probit and logit links, respectively. Motivated by an important problem in quantitative advertising, in Chapter 7 we consider the application of these algorithms to modeling rare events. We show that when the sample size is large but the observed number of successes is small, these data augmentation samplers mix very slowly, with a spectral gap that converges to zero at a rate at least proportional to the reciprocal of the square root of the sample size up to a log factor. In simulation studies, moderate sample sizes result in high autocorrelations and small effective sample sizes. Similar empirical results are observed for related data augmentation samplers for multinomial logit and probit models. When applied to a real quantitative advertising dataset, the data augmentation samplers mix very poorly. Conversely, Hamiltonian Monte Carlo and a type of independence chain Metropolis algorithm show good mixing on the same dataset.

Bayesian Emulation for Sequential Modeling, Inference and Decision Analysis

The advances in three related areas of state-space modeling, sequential Bayesian learning, and decision analysis are addressed, with the statistical challenges of scalability and associated dynamic sparsity. The key theme that ties the three areas is Bayesian model emulation: solving challenging analysis/computational problems using creative model emulators. This idea defines theoretical and applied advances in non-linear, non-Gaussian state-space modeling, dynamic sparsity, decision analysis and statistical computation, across linked contexts of multivariate time series and dynamic networks studies. Examples and applications in financial time series and portfolio analysis, macroeconomics and internet studies from computational advertising demonstrate the utility of the core methodological innovations.

Chapter 1 summarizes the three areas/problems and the key idea of emulating in those areas. Chapter 2 discusses the sequential analysis of latent threshold models with use of emulating models that allows for analytical filtering to enhance the efficiency of posterior sampling. Chapter 3 examines the emulator model in decision analysis, or the synthetic model, that is equivalent to the loss function in the original minimization problem, and shows its performance in the context of sequential portfolio optimization. Chapter 4 describes the method for modeling the steaming data of counts observed on a large network that relies on emulating the whole, dependent network model by independent, conjugate sub-models customized to each set of flow. Chapter 5 reviews those advances and makes the concluding remarks.

Likelihood-Based Inference for Partially Observed Multi-Type Markov Branching Processes

Thesis (Ph.D.)--University of Washington, 2016-08

Synthetic portfolio for event studies: Estimating the effects of volatility call auctions

We propose a method denoted as synthetic portfolio for event studies in market microstructure that is particularly interesting to use with high frequency data and thinly traded markets. The method is based on Synthetic Control Method and provides a robust data driven method to build a counterfactual for evaluating the effects of the volatility call auctions. We find that SMC could be used if the loss function is defined as the difference between the returns of the asset and the returns of a synthetic portfolio. We apply SCM to test the performance of the volatility call auction as a circuit breaker in the context of an event study. We find that for Colombian Stock Market securities, the asynchronicity of intraday data reduces the analysis to a selected group of stocks, however it is possible to build a tracking portfolio. The realized volatility increases after the auction, indicating that the mechanism is not enhancing the price discovery process.

A deep learning model for controlling the galvanizing process in a production line

In the industry of steelmaking, the process of galvanizing is a treatment which is applied to protect the steel from corrosion. The air knife effect (AKE) occurs when nozzles emit a steam of air on the surfaces of a steel strip to remove excess zinc from it. In our work we formalized the problem to control the AKE and we implemented, with the R&D dept.of MarcegagliaSPA, a DL model able to drive the AKE. We call it controller. It takes as input the tuple : a tuple of the physical conditions of the process line (t,h,s) with the target value of the zinc coating (c); and generates the expected tuple of (pres and dist) to drive the mechanical nozzles towards the (c). According to the requirements we designed the structure of the network. We collected and explored the data set of the historical data of the smart factory. Finally, we designed the loss function as sum of three components: the minimization between the coating addressed by the network and the target value we want to reach; and two weighted minimization components for both pressure and distance. In our solution we construct a second module, named coating net, to predict the coating of zinc resulting from the AKE when the conditions are applied to the prod. line. Its structure is made by a linear and a deep nonlinear “residual” component learned by empirical observations. The predictions made by the coating nets are used as ground truth in the loss function of the controller. By tuning the weights of the different components of the loss function, it is possible to train models with slightly different optimization purposes. In the tests we compared the regularization of different strategies with the standard one in condition of optimal estimation for both; the overall accuracy is ± 3 g/m^2 dal target for all of them. Lastly, we analyze how the controller modeled the current solutions with the new logic: the sub-optimal values of pres and dist can be optimize of 50% and 20%.

Loss-of-function mutations in the genes encoding prokineticin-2 or prokineticin receptor-2 cause autosomal recessive Kallmann syndrome

Context: Physiological activation of the prokineticin pathway has a critical role in olfactory bulb morphogenesis and GnRH secretion in mice. Objective: To investigate PROK2 and PROKR2 mutations in patients with hypogonadotropic hypogonadism (HH) associated or not with olfactory abnormalities. Design: We studied 107 Brazilian patients with HH (63 with Kallmann syndrome and 44 with normosmic HH) and 100 control individuals. The coding regions of PROK2 and PROKR2 were amplified by PCR followed by direct automatic sequencing. Results: In PROK2, two known frameshift mutations were identified. Two brothers with Kallmann syndrome harbored the homozygous p. G100fsX121 mutation, whereas one male with normosmic HH harbored the heterozygous p. I55fsX56 mutation. In PROKR2, four distinct mutations (p. R80C, p. Y140X, p. L173R, and p. R268C) were identified in five patients with Kallmann syndrome and in one patient with normosmic HH. These mutations were not found in the control group. The p. R80C, p. L173R, and p. R268C missense mutations were identified in the heterozygous state in the HH patients and in their asymptomatic first-degree relatives. In addition, nomutations of FGFR1, KAL1, GnRHR, KiSS-1, or GPR54 were identified in these patients. Notably, the new nonsense mutation (p. Y140X) was identified in the homozygous state in an anosmic boy with micropenis, bilateral cryptorchidism, and high-arched palate. His asymptomatic parents were heterozygous for this severe defect. Conclusion: We expanded the repertoire of PROK2 and PROKR2 mutations in patients with HH. In addition, we show that PROKR2 haploinsufficiency is not sufficient to cause Kallmann syndrome or normosmic HH, whereas homozygous loss-of-function mutations either in PROKR2 or PROK2 are sufficient to cause disease phenotype, in accordance with the Prokr2 and Prok2 knockout mouse models.

Identification of genes differentially expressed in a strain of the mold Aspergillus nidulans carrying a loss-of-function mutation in the palA gene

To identify genes differentially expressed in a strain of the mold Aspergillus nidulans carrying a loss-of-function mutation in palA, a gene in the pH-responsive signal transduction pathway, suppression subtractive hybridization was performed between RNA isolated from the biA1 and biA1 palA1 strains grown under limiting inorganic phosphate at pH 5.0. We have identified several genes upregulated in the biA1 palA1 mutant strain that play important roles in mitotic fidelity, stress responses, enzyme secretion, signal transduction mechanisms, development, genome stability, phosphate sensing, and transcriptional regulation among others. The upregulation of eight of these transcripts was also validated by Northern blot. Moreover, we show that a loss of function mutation in the palA gene drastically reduced the neutral sugar content of the acid phosphatase PacA secreted by the fungus A. nidulans grown at pH 5.0 compared with a control strain.

Improvement in coronary circulatory function in morbidly obese individuals after gastric bypass-induced weight loss: relation to alterations in endocannabinoids and adipocytokines.

AIMS: To investigate the effect of surgical gastric bypass-induced weight loss and related alterations in endocannabinoids (ECs) and adipocytokine plasma levels on coronary circulatory dysfunction in morbidly obese (MOB) individuals. METHODS AND RESULTS: Myocardial blood flow (MBF) responses to cold pressor test (CPT) from rest (ΔMBF) and during pharmacologically induced hyperaemia were measured with &supl;³N-ammonia PET/CT in 18 MOB individuals with a body mass index (BMI) > 40 kg/m² at baseline and after a median follow-up period of 22 months. Gastric bypass intervention decreased BMI from a median of 44.8 (inter-quartile range: 43.3, 48.2) to 30.8 (27.3, 34.7) kg/m² (P < 0.0001). This decrease in BMI was accompanied by a marked improvement in endothelium-related ΔMBF to CPT and hyperaemic MBFs, respectively [0.34 (0.18, 0.41) from 0.03 (-0.08, 0.15) mL/g/min, P = 0.002; and 2.51 (2.17, 2.64) from 1.53 (1.39, 2.18) mL/g/min, P < 0.001]. There was an inverse correlation between decreases in plasma concentrations of the EC anandamide and improvement in ΔMBF to CPT (r = -0.59, P = 0.009), while increases in adiponectin plasma levels correlated positively with hyperaemic MBFs (r = 0.60, P = 0.050). Conversely, decreases in leptin plasma concentrations were not observed to correlate with the improvement in coronary circulatory function (r = 0.22, P = 0.400, and r = -0.31, P = 0.250). CONCLUSIONS: Gastric bypass-related reduction of BMI in MOB individuals beneficially affects coronary circulatory dysfunction. The dysbalance between ECs and adipocytokines appears to be an important determinant of coronary circulatory function in obesity.

Natural Arabidopsis brx loss-of-function alleles confer root adaptation to acidic soil.

Soil acidification is a major agricultural problem that negatively affects crop yield. Root systems counteract detrimental passive proton influx from acidic soil through increased proton pumping into the apoplast, which is presumably also required for cell elongation and stimulated by auxin. Here, we found an unexpected impact of extracellular pH on auxin activity and cell proliferation rate in the root meristem of two Arabidopsis mutants with impaired auxin perception, axr3 and brx. Surprisingly, neutral to slightly alkaline media rescued their severely reduced root (meristem) growth by stimulating auxin signaling, independent of auxin uptake. The finding that proton pumps are hyperactive in brx roots could explain this phenomenon and is consistent with more robust growth and increased fitness of brx mutants on overly acidic media or soil. Interestingly, the original brx allele was isolated from a natural stock center accession collected from acidic soil. Our discovery of a novel brx allele in accessions recently collected from another acidic sampling site demonstrates the existence of independently maintained brx loss-of-function alleles in nature and supports the notion that they are advantageous in acidic soil pH conditions, a finding that might be exploited for crop breeding.

ALDH1A3 loss of function causes bilateral anophthalmia/microphthalmia and hypoplasia of the optic nerve and optic chiasm.

The major active retinoid, all-trans retinoic acid, has long been recognized as critical for the development of several organs, including the eye. Mutations in STRA6, the gene encoding the cellular receptor for vitamin A, in patients with Matthew-Wood syndrome and anophthalmia/microphthalmia (A/M), have previously demonstrated the importance of retinol metabolism in human eye disease. We used homozygosity mapping combined with next-generation sequencing to interrogate patients with anophthalmia and microphthalmia for new causative genes. We used whole-exome and whole-genome sequencing to study a family with two affected brothers with bilateral A/M and a simplex case with bilateral anophthalmia and hypoplasia of the optic nerve and optic chiasm. Analysis of novel sequence variants revealed homozygosity for two nonsense mutations in ALDH1A3, c.568A>G, predicting p.Lys190*, in the familial cases, and c.1165A>T, predicting p.Lys389*, in the simplex case. Both mutations predict nonsense-mediated decay and complete loss of function. We performed antisense morpholino (MO) studies in Danio rerio to characterize the developmental effects of loss of Aldh1a3 function. MO-injected larvae showed a significant reduction in eye size, and aberrant axonal projections to the tectum were noted. We conclude that ALDH1A3 loss of function causes anophthalmia and aberrant eye development in humans and in animal model systems.

Loss-of-Function Mutations in PTPN11 Cause Metachondromatosis, but Not Ollier Disease or Maffucci Syndrome.

Metachondromatosis (MC) is a rare, autosomal dominant, incompletely penetrant combined exostosis and enchondromatosis tumor syndrome. MC is clinically distinct from other multiple exostosis or multiple enchondromatosis syndromes and is unlinked to EXT1 and EXT2, the genes responsible for autosomal dominant multiple osteochondromas (MO). To identify a gene for MC, we performed linkage analysis with high-density SNP arrays in a single family, used a targeted array to capture exons and promoter sequences from the linked interval in 16 participants from 11 MC families, and sequenced the captured DNA using high-throughput parallel sequencing technologies. DNA capture and parallel sequencing identified heterozygous putative loss-of-function mutations in PTPN11 in 4 of the 11 families. Sanger sequence analysis of PTPN11 coding regions in a total of 17 MC families identified mutations in 10 of them (5 frameshift, 2 nonsense, and 3 splice-site mutations). Copy number analysis of sequencing reads from a second targeted capture that included the entire PTPN11 gene identified an additional family with a 15 kb deletion spanning exon 7 of PTPN11. Microdissected MC lesions from two patients with PTPN11 mutations demonstrated loss-of-heterozygosity for the wild-type allele. We next sequenced PTPN11 in DNA samples from 54 patients with the multiple enchondromatosis disorders Ollier disease or Maffucci syndrome, but found no coding sequence PTPN11 mutations. We conclude that heterozygous loss-of-function mutations in PTPN11 are a frequent cause of MC, that lesions in patients with MC appear to arise following a "second hit," that MC may be locus heterogeneous since 1 familial and 5 sporadically occurring cases lacked obvious disease-causing PTPN11 mutations, and that PTPN11 mutations are not a common cause of Ollier disease or Maffucci syndrome.

Loss of ATF2 function leads to cranial motoneuron degeneration during embryonic mouse development.

The AP-1 family transcription factor ATF2 is essential for development and tissue maintenance in mammals. In particular, ATF2 is highly expressed and activated in the brain and previous studies using mouse knockouts have confirmed its requirement in the cerebellum as well as in vestibular sense organs. Here we present the analysis of the requirement for ATF2 in CNS development in mouse embryos, specifically in the brainstem. We discovered that neuron-specific inactivation of ATF2 leads to significant loss of motoneurons of the hypoglossal, abducens and facial nuclei. While the generation of ATF2 mutant motoneurons appears normal during early development, they undergo caspase-dependent and independent cell death during later embryonic and foetal stages. The loss of these motoneurons correlates with increased levels of stress activated MAP kinases, JNK and p38, as well as aberrant accumulation of phosphorylated neurofilament proteins, NF-H and NF-M, known substrates for these kinases. This, together with other neuropathological phenotypes, including aberrant vacuolisation and lipid accumulation, indicates that deficiency in ATF2 leads to neurodegeneration of subsets of somatic and visceral motoneurons of the brainstem. It also confirms that ATF2 has a critical role in limiting the activities of stress kinases JNK and p38 which are potent inducers of cell death in the CNS.

Autosomal Recessive Posterior Microphthalmos/Nanophthalmos is Caused by Loss-of-function Mutations in LOC646960, a Novel Gene Encoding a Trypsin-like Serine Protease

Purpose: Posterior microphthalmos (MCOP)/nanophthalmos (NNO) is a developmental anomaly characterized by extreme hyperopia due to short axial length. The population of the Faroe Islands shows a high prevalence of an autosomal recessive form (arMCOP). The gene mutated in arMCOP is not yet known.Methods: Genetic mapping by linkage analysis using microsatellite and single nucleotide polymorphisms, mutation analysis by PCR and sequencing, molecular modellingResults: Having refined the position of the disease locus (MCOP6) in an interval of 250 kb in chromosome 2q37.1 in Faroese families, we detected 3 mutations in a novel gene, LOC646960: Patients of 10 different Faroese families were either homozygous (n=22) for c.926G>C (p.Trp309Ser) or compound heterozygous (n=6) for c.926G>C and c.526C>G (p.Arg176Gly), whereas a homozygous 1 bp duplication (c.1066dupC) was identified in patients with arNNO from a Tunisian family. In two unrelated patients with MCOP, no LOC646960 mutation was found. LOC646960 is expressed in the human adult retina and RPE. The expression of the mouse homologue in the eye can be first detected at E17 and is highest in adults. The predicted protein is a 603 amino acid long secreted trypsin-like serine peptidase. c.1066dupC should result in a functional null allele. Molecular modelling of the p.Trp309Ser mutant suggests that both affinity and reactivity of the enzyme towards in vivo substrates are substantially reduced.Conclusions: Postnatal growth of the eye is important for proper development of the refractive components (emmetropization), and is mainly due to elongation of the posterior segment from 10-11 mm at birth to 15-16 mm at the age of 13 years. Optical defocus leads to changes in axial length by moving the retina towards the image plane. arMCOP may theoretically be explained, in line with the expression pattern of LOC646960, by a postnatal growth retardation of the posterior segment.

Loss-of-function mutations of an inhibitory upstream ORF in the human hairless transcript cause Marie Unna hereditary hypotrichosis.

Marie Unna hereditary hypotrichosis (MUHH) is an autosomal dominant form of genetic hair loss. In a large Chinese family carrying MUHH, we identified a pathogenic initiation codon mutation in U2HR, an inhibitory upstream ORF in the 5' UTR of the gene encoding the human hairless homolog (HR). U2HR is predicted to encode a 34-amino acid peptide that is highly conserved among mammals. In 18 more families from different ancestral groups, we identified a range of defects in U2HR, including loss of initiation, delayed termination codon and nonsense and missense mutations. Functional analysis showed that these classes of mutations all resulted in increased translation of the main HR physiological ORF. Our results establish the link between MUHH and U2HR, show that fine-tuning of HR protein levels is important in control of hair growth, and identify a potential mechanism for preventing hair loss or promoting hair removal.