Abstract

The development of coiled tubing technology and its applications have been the hot issues in the oil and gas industry for over two decades. Since improper data analysis may lead to quite a different result in the estimation of friction pressure losses and flow regime, the development of precise data processing methodology is an essential tool to provide reliable experimental analysis results. Coiled tubing experimental data are greatly influenced by mechanical system vibrations in coiled tubing reel, delayed response time, and harmonized response of sensors due to long coiled tubing flow loops. However, no reliable standardized experimental data analysis methods are currently available that take into account quantitatively the parameters affecting experimental data.

A new highly reliable statistical engineering method for analyzing coiled tubing experimental data has been developed. It is based on an enormous database gathered over a many years of coiled tubing fluid tests using 1,000 and 2,000 ft long coiled tubing with field-scale equipment. This method provides engineers and researchers statistical tools to differentiate the highly suspected outliers from the experimental data.

This paper discusses the risk of the industry method (simple average), which is widely accepted by the oil and gas industry, in removing the outliers from the experimental data. The differences in the eventual data analysis results obtained from the industry method and from the statistical engineering approach are discussed. A specific process of the total uncertainty analysis of experimental measurements with a 95 percent confidence estimation in precision limit and the bias limit is also presented. This statistical engineering method, developed by the Well Construction Technology Center at the University of Oklahoma, is disclosed to the public in order to enhance coiled tubing technologies and to broaden its applications.

Introduction

Though technology advances have allowed researchers to obtain highly reliable and precise experimental devices, the accuracy of measuring and analyzing Coiled Tubing (CT) experimental data has been of great concern. To direct application and minimize the misleading of scale up study, it is a general trend to use field scale equipment in experiments, which bring more accurate result than micro-scale lab test. Since field scale CT experimental apparatus include considerably long curvature CT and long straight CT incorporated with the different signal transmitters, experimental data are greatly influenced by mechanical system vibrations in coiled tubing reel, delayed response time, and harmonized response of sensors due to long coiled tubing flow loops.

Since improper data analysis may lead to quite a different result in the estimation of friction pressure losses and flow regime, the development of precise data processing method is essential to provide reliable experimental analysis results. One of the data analysis methods is the averaged data, which is widely accepted by the industry. This gets rid of outliers in collected experimental data set and averages the data set in order to produce the representative data point in data analysis process. The problem laid on the averaged method is that low confidence can be assessed on the elimination process of outliers from experimentally collected data set with a researcher's engineering practice. No reliable standardized experimental data analysis methods are currently available that take into account elimination of experimental outliers.

This paper presents a highly reliable statistical engineering method for analyzing coiled tubing experimental data. This was developed based on an enormous database gathered over many years of coiled tubing fluid tests using 1,000 and 2,000 ft long coiled tubing with field-scale equipment. Methodologies to reduce the data inaccuracy from experimental apparatus are also discussed. It is still problematic how confident the representing data obtained from the data analysis may be in view of uncertainty and the precision limit. This paper presents a specific process of the total uncertainty analysis of experimental measurements with a 95 percent confidence estimation in precision limit and the bias limit.

This content is only available via PDF.
You can access this article if you purchase or spend a download.