Browsing by Subject "Data analysis"
Now showing 1 - 3 of 3
Results Per Page
Sort Options
Item Application of space time concept in GIS for visualizing and analyzing travel survey data(2006-05) Lu, Xiaoyun; Zhang, Ming, 1963 April 22-The classic time geography concept (space-time path) provides a powerful framework to study travel survey data which is an important source for travel behavior studies. Based on the space-time concept, this research will present a visualizing approach to analyze travel survey data. By inputting the data into GIS software such as TransCAD and ArcGIS and editing the needed information, this study will explain how to create 3D images of travel paths for showing the variation of trip distribution in relation to different social-economic factors deemed as the driving forces of such patterns. Also, this report will address the technical challenges involved in this kind of study and will discuss directions of future research.Item Equipment data analysis study : failure time data modeling and analysis(2012-05) Zhu, Chen, master of science in engineering; Popova, Elmira; Bickel, J.EricThis report presents the descriptive data analysis and failure time modeling that can be used to find out the characteristics and pattern of failure time. Descriptive data analysis includes the mean, median, 1st quartile, 3rd quartile, frequency, standard deviation, skewness, kurtosis, minimum, maximum and range. Models like exponential distribution, gamma distribution, normal distribution, lognormal distribution, Weibull distribution and log-logistic distribution have been studied for failure time data. The data in this report comes from the South Texas Project that was collected during the last 40 years. We generated more than 1000 groups for STP failure time data based on Mfg Part Number. In all, the top twelve groups of failure time data have been selected as the study group. For each group, we were able to perform different models and obtain the parameters. The significant level and p-value were gained by Kolmogorov-Smirnov test, which is a method of goodness of fit test that represents how well the distribution fits the data. The In this report, Weibull distribution has been proved as the most appropriate model for STP dataset. Among twelve groups, eight groups come from Weibull distribution. In general, Weibull distribution is powerful in failure time modeling.Item Production analysis of oil production from unconventional reservoirs using bottom hole pressures entirely in the Laplace space(2015-05) La, Natalie-Nguyen; Lake, Larry W.; Mohanty, Kishore KLaplace transforms are a powerful mathematical tool to solve many problems that describe fluid flow in unconventional reservoirs. However, for the solutions to be useful in applications, for instance history matching, they must be converted from the Laplace space into the real-time domain. A common practice is to numerically invert the transformed Laplace solution. However, we find substantial benefits if the data sets are handled entirely in the Laplace domain, and fitted to models presented in Laplace space rather than in the time domain. The data set used in this work is oil production rate and bottom hole pressure (BHP) from a liquid-rich shale play in North America, which we study to understand the decline of production from a tight formation produced by a fractured horizontal well. Since the BHP is relatively constant in the long run, a constant BHP solution is appropriate to analyze inflow performance analysis for most wells. However in some cases, as a result of operational changes to some wells, mainly periodic shut-ins, the production rate experiences isolated pressure build-ups. Both the production rate and BHP are transformed into the Laplace domain and accounted for in the model. Ours is the first analysis that combines rate and BHP entirely in the Laplace domain. There is no need for a Laplace transform inversion. Two models whose Laplace solutions are readily available are studied side-by-side, a single-compartment model versus a dual-compartment model. We fit the transformed production data of hundreds of wells to the Laplace models. The algorithm to transform data is fairly simple and computationally inexpensive. Since Laplace transformation smoothes the data, the fits are consistently good. Both models yield realistic and similar estimates of ultimate recovery. In most cases the effect of the second compartment in the dual-compartment model can be ignored, i.e., neglecting the fracture-well interaction. The single-compartment model seems adequate for modeling unconventional reservoirs performance. The knowledge of the reservoir model parameters provides estimation of the drainage volume and forecast future production. One of the main advantages of this novel history matching method is its ability to eliminating noise from data scatter without losing important information. As a result, we can match data more easily. Moreover, real-time solutions to many fluid flow problems in porous media often cannot be obtained analytically but rather via numerical computation. Our current method eliminates the need of inverting to real-time solutions. Additionally, these solutions often assume simple closed forms in Laplace domain even for very complex geometry (higher number of compartments), facilitating the task of history matching.