Hydrogen Sulfide Gas Detection via Multivariate Optical Computing

Bin Dai, Christopher Jones, Megan Pearl, Mickey Pelletier, Michael L. MyrickSensors 2018 · DOI

TL;DR

This paper introduces a novel sensor system for detecting hydrogen sulfide (H₂S) gas in challenging high-pressure, high-temperature (HPHT) environments, common in the petroleum industry. By leveraging Multivariate Optical Computing (MOC) and a custom-designed optical element (MOE), the system achieves accurate H₂S measurements even in the presence of interfering gases like mercaptans, offering a significant advancement over current methods.

What Problem Does This Paper Address?

The petroleum industry faces a critical need for real-time, in-situ H₂S detection within downhole oil and gas wells. H₂S is a highly toxic and corrosive gas that can cause significant damage to equipment and infrastructure, leading to costly over-engineering of production systems if its presence is not accurately known. Current methods for H₂S measurement typically involve collecting downhole fluid samples and transporting them to a laboratory for analysis. This process is time-consuming and prone to H₂S loss or reaction with sampling equipment, leading to underestimated concentrations. Furthermore, existing sensors often struggle with the harsh HPHT conditions found in subterranean wells, exhibiting reliability issues or being too large for downhole tools. This paper aims to bridge this gap by developing an optical sensor capable of accurate H₂S measurement directly in these challenging environments.

Methods

The core of this research lies in the application of Multivariate Optical Computing (MOC). MOC is an optical technique that performs spectral analysis and calculation directly within the optical domain, simplifying instrument design and enhancing robustness. The process involves:

  1. Spectral Data Acquisition: A laboratory setup was used to collect UV spectra of H₂S and various interfering gases (including mercaptans, CO₂, H₂O, and hydrocarbons) under both room-temperature/low-pressure and HPHT conditions. This data forms the basis for designing the sensor.
  2. Multivariate Optical Element (MOE) Design: Using the collected spectral data, an optimization algorithm was employed to design a specialized optical filter (the MOE). This MOE is engineered to have a specific transmission pattern that, when illuminated by the sample's spectrum, produces a signal proportional to the H₂S concentration. The design process aims to minimize prediction error and uncertainty.
  3. MOE Fabrication: The designed MOE was fabricated using a thin-film stack structure, alternating layers of Al₂O₃ and SiO₂ with precise thickness control.
  4. MOC Sensor Prototype: The fabricated MOE, along with a neutral density (ND) filter, was integrated into a rotating wheel within a prototype MOC sensor system. A UV light source illuminates the gas sample, and the light passing through either the MOE or the ND filter is detected.
  5. Signal Processing: The ratio of the integrated detector signals from the MOE channel to the ND channel (A/B ratio) is calculated. This ratio is designed to be linearly proportional to the H₂S concentration, allowing for quantitative measurement.

Datasets

The study utilized two primary sets of spectral data:

  1. Reference Spectra: UV absorption cross-sections of H₂S and methyl mercaptan (CH₃SH) were obtained from the MPI-Mainz UV-vis Spectra Atlas.
  2. Experimental Training Data: UV spectra of H₂S, CO₂, and CH₃SH mixtures were collected in the lab under various conditions:
  3. Low-Pressure/Room-Temperature: Spectra were acquired at eight pressure setpoints (96.5 kPa to 6894.7 kPa) with varying concentrations of H₂S, CO₂, and CH₃SH in a N₂ buffer gas.
  4. High-Pressure/High-Temperature (HPHT): Spectra were collected at five pressure setpoints (20.68 to 103.42 MPa) and five temperature setpoints (65.6 to 176.7 °C) for H₂S samples.

This experimental data, particularly the HPHT spectra, represents a novel dataset for training and validating an MOC sensor for H₂S in challenging downhole conditions.

Key Insights and Ideas

  • MOC for Robust H₂S Detection: The paper demonstrates that MOC is a powerful technique for developing robust H₂S sensors that can operate in complex gas mixtures and harsh environments. The optical nature of the computation inherently simplifies the instrument and reduces susceptibility to drift.
  • Targeted MOE Design for Interference Rejection: The key to handling interfering species like mercaptans lies in the MOE design. By carefully tailoring the MOE's transmission spectrum based on the spectral characteristics of both H₂S and interferents, the system can effectively "filter out" unwanted signals and isolate the H₂S signature.
  • HPHT Data is Crucial: The collection of spectral data under HPHT conditions was essential for designing an MOE that performs accurately in the target application environment. This highlights the importance of application-specific data for developing effective optical sensors.
  • Linearity of MOC Response: The study confirms that the A/B signal ratio from the MOC sensor exhibits a linear relationship with H₂S concentration across various pressures, simplifying calibration and prediction.

Key Results

  • MOE Calibration Accuracy: The fabricated MOE demonstrated a standard error of calibration (SEC) of 3.3 nmol/mL and a relative accuracy (relative SEC) of approximately 2.2% for H₂S concentrations ranging from 0 to 150 nmol/mL.
  • Validation Accuracy: In validation tests with H₂S and mercaptan mixtures under various pressures, the MOC system achieved a relative accuracy (relative SEP) of 8.05% for H₂S measurement.
  • Sensitivity: The sensor exhibited a detection sensitivity (relative signal change) of approximately 10% for H₂S concentrations in the tested range.
  • Interference Rejection: The MOC sensor was able to achieve better than 10% accuracy for H₂S measurement in mixtures containing mercaptans, demonstrating effective spectral interference management.

Limitations and Future Work

  • Ideal Gas Assumption: The reference concentrations of H₂S samples were calculated based on the ideal gas assumption. The authors note that using an equation-of-state (EoS) model would provide a more accurate calculation, especially at higher pressures.
  • Pressure Control Uncertainty: An estimated 5% uncertainty in pressure control could contribute to the calculated SEC and SEP.
  • Future Sensor Design: For future commercial MOC sensors, the authors suggest using a short-pass filter and a simple photodiode instead of the spectrometer to create a more compact and robust system.
  • Broader Application: While this study focused on H₂S, the MOC approach is broadly applicable to other analytes and complex mixtures, suggesting potential for wider adoption in various industrial monitoring applications.

Summary updated

Abstract

Hydrogen-sulfide gas is a toxic, colorless gas with a pungent odor that occurs naturally as a decomposition by-product. It is critical to monitor the concentration of hydrogen sulfide. Multivariate optical computing (MOC) is a method that can monitor analytes while minimizing responses to interferences. MOC is a technique by which an analogue calculation is performed entirely in the optical domain, which simplifies instrument design, prevents the drift of a calibration, and increases the strength and durability of spectroscopic instrumentation against physical perturbation when used for chemical detection and identification. This paper discusses the detection of hydrogen-sulfide gas in the ultraviolet (UV) spectral region in the presence of interfering gaseous species. A laboratory spectroscopic measurement system was set up to acquire the UV spectra of H₂S and interference gas mixtures in high-pressure/high-temperature (HPHT) conditions. These spectra were used to guide the design and fabrication of a multivariate optical element (MOE), which has an expected measurement relative accuracy of 3.3% for H₂S, with a concentration in the range of 0⁻150 nmol/mL. An MOC validation system with the MOE was used to test three samples of H₂S and mercaptans mixtures under various pressures, and the relative accuracy of H₂S measurement was determined to be 8.05%.
BibTeX
@inproceedings{dai2018hydrogen,
  author = {Bin Dai and Christopher Jones and Megan Pearl and Mickey Pelletier and Michael L. Myrick},
  title = {{Hydrogen Sulfide Gas Detection via Multivariate Optical Computing}},
  booktitle = {Sensors},
  year = {2018},
  doi = {10.3390/s18072006},
}
Figure