Laboratory Quality Optical Analysis in Harsh Environments

Abstract

Significant advances have been made in formation testing since the introduction of wireline pumpout testers (WLPT), particularly with respect to downhole fluid compositional measurements. Optical sensors and the use of spectroscopic methods have been developed to improve sample quality and minimize sampling time in downhole environments. As a laboratory technique, spectroscopy is a ubiquitous and powerful technology that has been used worldwide for decades to measure the physical and chemical properties of many materials, including petroleum, geological, and hydrological samples. However, laboratory-grade, high-resolution spectrometers are incompatible with the hostile environments encountered downhole, at wellheads, and on pipelines. Only limited resolution techniques are available for the rugged conditions of the oil field. This paper introduces a new optical technology that can provide high-resolution, laboratory-quality analyses in harsh oilfield environments.A new technology for optical sensing, multivariate optical computing (MOC), has been developed and is a nonspectroscopic technique. This new sensing method uses an integrated computation element (ICE) to combine the power and accuracy of high-resolution, laboratory-quality spectrometers with the ruggedness and simplicity of photometers. Many modern sensors typically merge the sensor with the electronics on an integrated computing chip to perform complex computations, resulting in an elegant yet simplistic design. Now, optical sensing using ICE features an analogue optical computation device to provide a direct, simple, and powerful mathematical computation on the optical information, completely within the optical domain. Because the entire optical range of interest is used without dispersing the light spectrum, the measurements are obtained instantly and rival laboratory-quality results.A proof of concept MOC with ICE has been demonstrated, logging more than 7,000 hours, in nearly continuous use for 14 months. Oils with gravities ranging from 14 to 65°API have been measured in downhole environments that range from 3,000 to 20,000 psi, and from 150 to 350°F. Hydrocarbon composition measurements, including saturates, aromatics, resins, asphaltenes, methane, and ethane, have been demonstrated using the MOC configuration. As compositional calculations therein, GOR and density are validated to within 14 scf/bbl and 1%, respectively. The paper discusses the details of the new ICE-based sensor and describes its adaptations to downhole applications.

Hydrogen Sulfide Gas Detection via Multivariate Optical Computing

“…The experiment setup for UV spectral data collection is illustrated in Figure 3 a. The setup had been previously described with slight modifications [ 20 , 27 ] and consisted of a UV light source (deuterium lamp (D2), low pressure 5 Torrs (0.1 psi), Hamamatsu Inc.) , two convex lenses (Edmund Optics, Barrington, NJ, USA), a high-pressure/high-temperature (HPHT) optical cell with CaF 2 window (thickness of 1 cm) on each sides (custom built by Halliburton company, Houston, TX, USA) , an optical fiber (Ocean Optics), and a UV spectrometer (Ocean Optics HR200, Dunedin, FL, USA). The deuterium lamp was used as a broadband UV source.…”

“…As light traverses the MOE and strikes a detector, an analog dot product of the regression vector is computed with respect to the spectral pattern. MOC has been shown to provide similar analysis accuracy as the laboratory spectrometer from which an MOE is designed [ 24 , 25 , 26 , 27 ]. Proof-of-concept systems have been developed for visible, near-infrared, and mid-infrared systems [ 28 , 29 ].…”

Characterization of ion-assisted induced absorption in A-Si thin-films used for multivariate optical computing

“…The transmission or reflection function of the integrated computational element mimics the regression vector 6 . Unlike typical interference filters or notch filters, which only measure the net change in intensity, integrated computational elements extract high-resolution information about the analyte from the spectrum.…”

“…Because the integrated computational element is designed for a real multivariate optical computer, the regression vector includes other optical throughput effects in the optical train of the system, which is a slight deviation from the Beer-Lambert law mentioned in Equation 3. The predictability of the integrated computational element is described using two merit values: the standard error of calibration (SEC) and A/B detector response (or sensitivity) 6,8 . The SEC is the regression fit of the measured concentration values of the analyte versus the predicted (Figure 10a) using a linear fit to the data.…”

“…Various analytes and interferents are mixed with varying concentrations with the absorption spectrum taken each time. The regression vector is then derived computationally using methods such as partial least squares (PLS) 5,6 . Figure 2a shows the spectral database collection for a mixture of oil and gas in varying concentrations collected using the optical PVT system at Halliburton 6 .…”