Raman Technology for Faster Methane Detection in NGL Production

According to a 2019 US Energy Information Administration (EIA) study, 510 natural gas processing plants operate in the United States. This amounts to around 80.8 billion cubic feet of processing power in the lower 48 states.

A modern processing plant produces two principal commodities: natural gas liquids (Y-grade) and residual gas.

Raw gas from a collection network passes through a cryogenic cooling tower, which cools it to extremely low temperatures. Most gaseous hydrocarbon components, such as C₂-C₆+ (ethane, propane, butane, etc.), are condensed into the liquid Y-grade product.

Methane, the residual hydrocarbon component, boils beyond the cooling tower's working temperature and is largely in the gas phase. This methane, known as residual gas, fuels power plants and residential and commercial structures.

A small amount of methane remains in the liquid Y-grade product. Accurately measuring and controlling that quantity has significant ramifications for the process owner.

Methane in Y-grade is a pollutant that impacts product quality and causes issues with process efficiency, custody transfer, and safety.

Some common methane criteria include concentrations of no more than 0.5% (5000 ppm) or 1.5 times the Ethane concentration (ppm).

NGL fractionators (Y-grade clients) typically do not use a methane scrubber (a demethanizer column) as their first unit operation.

As a result, exceeding the methane specification can impact the processing speed, efficiency, and quality of fractionator columns. High methane Y-grade can be accommodated through blending; however, this reduces process efficiency and profitability.

Excess methane can cause outgassing problems during the subsequent pipeline transfer of NGLs to customers. Beyond the practical issues, off-spec Y-grade can result in financial penalties and contract violations.

Monitoring the methane concentration is crucial for ensuring the Y-grade producer's cryogenic tower is operational and on schedule.

Traditional online analysis methods, such as gas chromatography (GC), report cooling tower operational status as a C1/C2% ratio number, a popular indicator for optimizing tower performance. Individual measurement cycles for online GC equipment may take 5 to 20 minutes to report.

Raman spectroscopy has recently shown its utility and value in the online measurement of hydrocarbon products in midstream and refinery environments.

It has established itself as the leading optical method for determining composition and physical properties in NGL, natural gas, and purity products because of its selectivity, rapidity, stability, and sensitivity.

Many industrial processes, such as cryogenic cooling towers, can change faster than a conventional GC measurement cycle can complete.

While attempting to measure a rapidly changing process, a slow cycle time gives operators no relevant information during the GC cycle durations, resulting in critical events and transitions being missed or discovered late.

In comparison, a Raman measurement takes 3 to 5 seconds, which is substantially faster than a comparable GC measurement and can be utilized to enable continuous monitoring and early identification of process changes.

This speed gives operators more visibility into processes as they occur and the capacity to optimize process outcomes in relation to specifications, for example, ensuring ethane concentrations in cryogenic cooling towers are within permissible limits.

One argument against using Raman spectroscopy is its apparent limited sensitivity compared to other techniques such as gas chromatography. However, this study shows that Raman can detect low amounts of methane in NGLs (down to 130 ppm) with high precision in real time.

Experimental

The spectra were acquired using a Thermo Scientific^™ MarqMetrix^™ All-In-One Process Raman Analyzer. The acquisition parameters were tuned so that a new dark-subtracted spectra was acquired for each sample tested, resulting in optimal acquisition durations and averages (Figure 1).

Spectra were acquired with a Thermo Scientific^™ MarqMetrix^™ FlowCell Sampling Optic (Figure 2). After collection, the data was processed and modeled with Solo 9.2.1 (Eigenvector Research, Inc., Manson, WA). Minor, non-chemical fluctuations in the spectra were mitigated using extended multiplicative signal correction (EMSC).

Figure 1. Y-grade spectra collected on a MarqMetrix All-In-One Process Raman Analyzer and FlowCell Sampling Optic. The acquisition parameters were set to collect a new dark-subtracted spectrum for each sample analyzed. Image Credit: Thermo Fisher Scientific – Portable and Handheld Process Raman Spectroscopy

Figure 2. Integrated high-pressure low-temperature FlowCell, rated to 2,500 psi and a temperature range from cryogenic temperatures to 350 K. Image Credit: Thermo Fisher Scientific – Portable and Handheld Process Raman Spectroscopy

In this investigation, liquid samples were introduced to the FlowCell in a stopped-flow manner from compressed piston cylinders with backpressure maintained at more than 1000 psi.

Back pressure was applied to the FlowCell's exit leg to allow stop-flow manipulation and pressure measurement while keeping volatile substances in solution.

The identical samples were analyzed with GPA 2177 to provide laboratory reference values. This experiment investigated nine samples as part of the Proof-of-Concept (POC) investigation.

Results and Discussion

This study shows that it is possible to measure low-level methane in a Y-grade product at a level equivalent to standard GC measurements. Based on this POC model, which only includes nine samples, the methane model shows that an RMSEC of 100 ppm is possible (Figure 3).

Figure 3. The methane model suggests that a prediction RMSE of 100 ppm is achievable based on this POC model, which contains only nine samples. Image Credit: Thermo Fisher Scientific – Portable and Handheld Process Raman Spectroscopy

The capability of a multivariate model is always determined by the quality and characteristics of its calibration dataset.

This dataset contributed to the methane response's independence from other hydrocarbon species. A suitable calibration dataset compels modeling to rely entirely on methane Raman spectral characteristics, as shown by the Regression Vector (Figure 4).

Figure 4. The Regression Vector shows that methane has a single strong Raman band at approximately 2910 cm-1, demonstrating that Raman can detect methane in low concentrations. Image Credit: Thermo Fisher Scientific – Portable and Handheld Process Raman Spectroscopy

Methane has a single strong Raman band at around 2910 cm^-1, making it ideal for measuring and modeling methane at low levels. The methane model's PLS regression vector highlights the potential for specificity in Raman spectroscopy models, using the methane band as the essential feature.

While this band almost always heavily overlaps with other hydrocarbons, a rigorous calibration set design combined with multivariate modeling enables the use of the modest but strongly overlapped methane Raman signal.

It is also worth investigating the concentrations of other hydrocarbons in Y-grade products. Raman spectroscopy can easily quantify and anticipate these additional bulk-level components.

Table 1 displays the RMSEC values and component ranges from this modeling study, demonstrating the ability to forecast all Y-grade species using Raman spectroscopy.

Table 1. Y-Grade PLS Model RMSE Metrics. Source: Thermo Fisher Scientific – Portable and Handheld Process Raman Spectroscopy

Conclusion

This study shows that Raman spectroscopy is a viable method for detecting low quantities of methane in Y-grade products.

Raman's rapid measurement frequency of 3 to 5 seconds makes it ideal for process monitoring and control. Solid-state Raman analyzers, such as the MarqMetrix All-In-One Process Raman Analyzer, come factory-calibrated and require no consumables.

This means that the customer may complete their investigation without incurring costly disruptions. This efficiency, combined with the use of Raman spectroscopy and chemometric approaches, has already made Raman more widely used in various industrial applications.

This information has been sourced, reviewed and adapted from materials provided by Thermo Fisher Scientific – Portable and Handheld Process Raman Spectroscopy.

For more information on this source, please visit Thermo Fisher Scientific – Portable and Handheld Process Raman Spectroscopy.