A normal curve has kurtosis of 0.

This paper presents a novel approach of continuously measuring drilling fluid rheology and density by use of sound signals. A unique apparatus is built with a series of pipe sections designed to exact pre-calculated dimensions to achieve equivalent standard shear rates as stipulated in the American Petroleum Institute (API) Recommended Practice 13D for measuring the rheology of oil-well drilling fluids (from 3 to 600 RPM). Acoustics waves are passed through the fluids of interest and their interaction is recorded and analyzed to deduce the density and rheological properties of the fluids.

The concept of resonance as demonstrated by the Barton's pendulums are the basis of the methodology. Sound signals are known to exhibit a damping effect when passing through various media. Pairs of sensors are employed in this set-up and their signal response are first characterized and calibrated with fluids of known properties. Electric current is converted into acoustic signals by piezoelectric sensors mounted of the flowline which are then emitted through the fluids desired to be measured without interrupting the flow. A matching sensor receives these damped signed and reconverts them back to electromotive potentials for recording by a data acquisition unit. The signals are then analyzed by applying statistical techniques to interpret and obtain the fluids physical properties.

Owing to the nature of the task, the goal of accurately achieving simultaneous measurement of density and viscosity is attained by applying an ensemble machine learning algorithm, known as Multivariate Random Forest. Pure chemicals and fluids of known properties form the training group on which the predictive model is built for subsequent testing on new mud samples flowing through each section. The pipe sections generate shear rates covering the standard range adopted in oilfield reports. Results from each pair of sensors are analyzed and compared with dial readings from rotational viscometers; these have shown to be within a narrow band of error.

As a result of this work, the voltage outputs are sent continuously and in real-time to a processing computer that converts the values to dial readings at standard shear rates, while not disrupting the flow. This can aid in the better monitoring and surveillance of the entire fluid system of the well, which is highly beneficial to well control. The system can also be arranged to acquire gel strengths or how the fluid behaves after a fixed period of rest. Improvements can be made on the current procedures for fluid characterization which have remained relatively static for many years. This work engages the disciplines of rheology, acoustics and machine learning, creating a mechanism for continuous and real-time drilling fluid surveillance critical to the enhancement of safe development of petroleum resources.

ABSTRACT This paper provides a practical framework for efficient assessment structural fatigue damage in the non-Gaussian seas. In the novel framework, a generalized state-space based model is developed to depict the dynamic characteristics of the floating system, and the effect of non-Gaussian stochastic wave generated based on the translation process theory is incorporated into the model. The key feature of the framework is that the convolution terms in the dynamic analysis model are represented by the state-space model with constant coefficients matrices, and the computational effort of the dynamic analysis can be reduced significantly. With the framework, a comprehensive investigation on the fatigue damage of mooring system to non-Gaussian stochastic wave is conducted with a deep-water semi-submersible platform. Numerical results show that the non-Gaussian stochastic wave can accelerate structural fatigue damage significantly, and more attention should be paid to the non-Gaussian stochastic wave in the design procedure of mooring system. INTRODUCTION The stochastic wave loads acting on offshore structure can cause the cumulative fatigue damage of structure and result in fatigue failure of structural components and systems. The stochastic wave is usually modeled as a Gaussian process in the design procedure of offshore structure. However, the field measurement data indicate that the stochastic wave in shallow water or complex terrain often exhibits non-Gaussian characteristics with the tail of the probability density function (PDF) deviated from Gaussian distribution (Wang, 2014). In contrast with the Gaussian stochastic wave, the probability of wave elevation over three times standard deviation of the softening non-Gaussian stochastic wave becomes larger, and this may accelerate structural fatigue damage significantly (Ding and Chen, 2014; Gong and Chen, 2014). Therefore, understanding the fatigue performance of offshore structure to non- Gaussian stochastic wave is of vital important for its safety and reliability. To estimate the fatigue damage caused by the non-Gaussian stochastic wave, there are two approaches available, namely the frequency domain method and the time domain method (Ding and Chen, 2015). For the frequency domain assessment method, the non-Gaussian stochastic wave is related to the Gaussian stochastic wave through a monotonic translation function, and the distribution of the stress range associated with the non-Gaussian stochastic wave can be determined if the one for its underlying Gaussian process is specified (Ding and Chen, 2015). The analytical formula for structural fatigue damage under the non-Gaussian stochastic wave can be derived under the Rayleigh distribution assumption based on the nonlinear translation theory (Benasciutti and Tovo, 2005). However, due to the nonlinearities inherent in system, the dynamic response of the floating system is usually multimodal broadband process, and the stress range of the response does not follow the Rayleigh distribution exactly (Chang et al., 2017). Therefore, the frequency domain assessment method cannot estimate the fatigue damage of floating system accurately, and the frequency domain assessment method for the multimodal broadband non-Gaussian process warrants further investigation.

In the Black Ridge region of the Atlantic Ocean, Guerin et al This study aims to detect the attenuation effects of gas (1999) have shown that hydrate-bearing lithologies have hydrates by applying statistical analyses to waveforms in the ability to attenuate seismic signal. In the Alaskan regions that indicate Bottom Simulating Reflector (BSR) in permafrost, laboratory studies of the Malik well have also seismic data. Statistical methods of kurtosis and skewness show that hydrates are likely to cause attenuation (Wood et will be utilized as seismic attributes applied to the seismic al (2000). Additionally, seismic studies, such as those by waveforms to determine if the presence of gas hydrates Sérgio et al (2010) were able to identify spectral anomalies within the pore spaces result in observable frequency indicative of hydrate concentration within the GHSZ and attenuation. Since not all gas hydrate accumulations result free gas accumulations below both the GHSZ and gas in a clearly imaged BSR, the ultimate goal of this project is chimneys.

ABSTRACT: This paper utilizes Itasca's PFC3D to simulate linear rock cutting using disc cutters. The objective is to determine the correlation between excavated surface roughness and cutting forces experienced during excavation. The flat-joint BPM model was used to create a synthetic sandstone block in this paper. A rigid wall was modeled as a disc cutter. Cutting was achieved by applying angular and translational velocities to the cutter to achieve both rotational and translational motion. Several cuts were made on the block, completely planing the surface while recording the contact forces on the cutter. The 3D positions of the top-most balls on the excavated surface were determined and used to create a digital elevation model (DEM) of the excavated surface. Roughness parameters were calculated from the DEMs. Analyses of the forces and roughness data show that the surface roughness is positively correlated with the cutting forces. Cutting forces data obtained from laboratory tests that were conducted on rocks similar to the rock modeled in this paper were compared with the numerical results. Though the numerical results were consistently higher than the lab results, they had similar trends.

1. INTRODUCTION

Mechanical rock excavation is a means of breaking rock masses for the purpose of production of ore or for the purpose of creating space for storage of material or for civil engineering projects using machines mounted with different types of cutting tools (Bilgin et al., 2014). It is an alternative to the traditional drill-and-blast method of excavation. It is usually chosen based on economic and environmental reasons and sometimes because of the rock type (rock properties). Mechanical excavation involves the use of mechanical tools to cause shear or tensile fracturing in rock masses resulting in fragmentation. When the fractures ‘daylight’ or coalesce with others, they form chips that break off the surface of the rock. The fracturing mechanism, either tensile or shear fracturing, depends on the mechanical properties of the rock, the tool type and other excavation machine parameters (Rostami, 2013). There is the need to optimize excavation machine and cutting parameters for efficient mechanical excavation for various rock types and conditions. The major factors considered in performance prediction of mechanical excavation that can be adjusted are the spacing between cutters and the depth of cut (DOC) of the cutters (Han et al., 2017).

ABSTRACT

Inhomogeneous media can change the nonlinear properties of waves propagating on them. In the ocean, this phenomenon can be observed when waves travel on a surface current. In the case of negative horizontal velocity gradients (i.e. an accelerating opposing current or a decelerating following current), waves shorten and heighten, enhancing wave steepness. As a result, a nonlinear mechanism known as modulational instability develops, leading to the formation of large-amplitude waves (the so-called rogue waves), even if they would otherwise be unexpected. Laboratory experiments and numerical simulations with a current-modified version of the Euler equations are presented to assess the role of an opposing current in changing the statistical properties of unidirectional random wave fields. Results demonstrate in a consistent and robust manner that an opposing current induces a sharp and rapid transition from weakly to strongly non-Gaussian properties with a consequent increase of the probability of occurrence of rogue waves. Agreement with numerical simulations confirms that this transformation can be attributed to quasi-resonant nonlinear interactions triggered by the background current.

INTRODUCTION

Extreme waves larger than two times the significant wave height (also known as rogue waves) represent a serious threat for marine structures and operations (e.g. Clauss, 2002). Therefore, an accurate description of the statistical properties of the surface elevation and wave height can contribute to improving the design process and warning criteria for marine operations (Toffoli et al., 2005).

There are many mechanisms that cause large amplitude waves to occur (see Kharif and Pelinovsky, 2002; Onorato et al., 2013, for a complete review). Among them, nonlinear energy focusing due to the modulational instability of uniform wave trains to side band perturbations remains the most accredited (e.g. Janssen, 2004; Onorato et al., 2006; Toffoli et al., 2008; Onorato et al., 2009; Waseda et al., 2009; Babanin et al., 2011, Toffoli et al. 2013, among others). It has been verified theoretically and experimentally, however, that such mechanisms occur if waves are sufficiently steep and narrow banded both in the frequency and directional domain. Under these circumstances, rogue waves may occur within a fairly short scale of tens of wavelengths leading to substantial deviations from Gaussian and second-order-based statistics (e.g. Mori and Yasuda, 2002; Socquet-Juglard et al., 2005; Onorato et al., 2006; Onorato et al., 2009; Janssen, 2004; Toffoli et al. 2008; Waseda et al., 2009,; Toffoli et al., 2010; Toffoli et al. 2017 among others).

We present an automatic arrival picking workflow as a preprocessing for the real-time microseismic event location, based on a data-adaptive spectral subtraction filtering and a kurtosis based arrival picking approach. We modified the characteristic functions using kurtosis to highlight the statistical property changes from the microseismic event onsets and adopt the eigenvalue decomposition on three component seismograph data to determine the polarization in order to distinguish P and S waves. The proposed arrival picking system is fully data driven and parameter free, which makes it a perfect option for an autonomous real-time microseismic event location solution. We have tested our arrival identification system on field data, and obtained encouraging results.

Presentation Date: Tuesday, September 26, 2017

Start Time: 11:25 AM

Location: Exhibit Hall C/D

Presentation Type: POSTER

ABSTRACT

Numerical simulations performed by nonlinear Schrodinger equation and high-order spectral method are compared with the laboratory experiments in this paper. A detailed assessment of the capability of the two wave models to descript irregular waves propagation is carried out. It is found that the HOS method shows a better agreement with the experiments in terms of simulating the wave parameters, while HOS method shows a little overestimation in predicting extreme waves.

INTRODUCTION

Extreme waves, which occurs far more frequently than we have imagined, cause severe damages to offshore structures and vessels in recent years. It is critical to precisely predict the distribution of wave heights for the designing of offshore platforms. Besides accurate description of long term propagation of nonlinear waves in intermediate and deep water depth is also a key issue in ocean engineering. Usually, numerical models were used to implement this target.

One simple but powerful approach for the study of nonlinear slow modulated waves is the nonlinear Schrödinger equation which was first put forward by Zakharov (1968) adopting a spectral method. Hasimoto and Ono (1972) also derived the equation employing a multiple scale technique. The nonlinear Schrödinger equation is capable of describing slowly modulated waves which implies constraints in bandwidth and steepness. The NLS equation successfully predicts the phenomena such as envelope solitons (Zakharov and Shabat, 1972) and recurrence (Yuen and Ferguson, 1978). Onorato et al. (2001) studied the dynamics of freak waves using this equation. While this equation also has shortcomings that require the assumption of narrow-bandwidth and slow modulation. Dysthe (1979) later extended the NLS to fourth order and the resulting equation is the so called modified Schrodinger equation (MNLS). Trulsen and Dysthe (1996) went a step further by relaxing the constraint on the bandwidth. Many studies try to explain the formation of extreme waves adopting this method and this modified method is capable of describing phenomenon that is overlooked by NLS equation. For instance, Lo and Mei (1985) found the unequally growth of sideband perturbations and downshift of carrier wave using fourth-order nonlinear Schrödinger equation in a numerical study. Other features such as the split of wave packet is not observed in NLS equation either. Subsequent studies (Melville, 1982; Su, 1982) also confirmed the phenomenon. Zhang (2016) compared the MNLS model with the experiments in terms of wave parameters and exceedance distribution. Cousins and Sapsis (2016) found a way to forecast extreme waves employing the MNLS, and the results show high accuracy in random waves.

Abstract In this paper wavelet phase is extracted using a combination of different moments (CDM), which is introduced as an indicator of localized seismic signal phase. The analysis demonstrates sharpness of the probability distribution of a discrete time series that is more robust than applying kurtosis. The improvement occurs because zero-phase wavelets with reverse polarities have equal maximum kurtosis value hence the correct wavelet polarity cannot be identified. Zero-phase wavelets with reverse polarities have minimum and maximum CDM values. Those enable the technique to handle a finite data segment based on the combination of different moments and it can also choose the correct wavelet polarity. High order statistics are sensitive to outliers so they will be instable in region with high level of noise. Combination of different moments can solve or decrease this limitation because of existing of low degree behind high degree moments. The most important parameter for correctly estimating the phase pertains to the best length of time window used for phase correction. Incorrect window length creates non-zero phase wavelets. To decompose signal to wavelets, an envelope of each trace is applied. Based on the time distance between two minima of envelope value, it is possible to choose the wavelet lengths to do phase correction. Synthetic and real data examples are used to demonstrate the effectiveness of this method in both wavelet extraction and for local correction of signal phase. Results of the seismic phase correction using this method demonstrate superiority than the local kurtosis method, because of high stability and dynamical range.

Abstract As the drilling industry enters the era of big data, it has become necessary to find ways to organize and understand the vast amounts of real-time high frequency data recorded. Shor et al. (2014) show that real-time accelerations recorded down hole actually differ from those predicted by various models, either in the high frequency range or at peak values. These differences in vibration levels can be attributed to noise from various sources, such as the motor or surface equipment, design of measurement devices, uneven shape of the borehole, and/or dynamic effects, such as dampening/cushioning of fluids, bit rock interactions, drillstring-borehole wall interactions, etc. These differences can cause real-time drill log analysts to miss instances of actual failure or predict false failures. The industry currently focuses on the peak and average values of the accelerations recorded down hole, and there is no established method to examine the trend of these accelerations. The methodology proposed in this paper first focuses on obtaining the actual unmodulated acceleration values from the recorded downhole gauge data and then processes these values to understand the trend of acceleration as the drilling proceeds. This demodulation (or deconvolution) of the recorded data can reduce the likliehood of false predictions and, at the same time, increase the credibility of the mode of acceleration predicted (and hence more accurately predict failure), which currently relies only on the experience of the engineer. Thus, this technique can make real-time data monitoring more reliable and simple. Further, if combined with a gamma ray log to know the lithology of the formation being drilled, this data monitoring technique can reveal a significant amount of information. In addition, this paper suggests calculating instantaneous jerk intensity and decomposing it to its monotonic intrinsic mode function. This reveals that these monotonic functions follow a certain trend, which might be useful in future endeavors to understand the underlying physics of drillstring failure.