Hydrocarbon Dew Point Prediction in Natural Gas Processing
Modcon Systems Ltd.
Innovative technologies of process analysis and AI-enabled?optimization.
Hydrocarbon Dew Point (HCDP) is a critical property of natural gas, representing the temperature at which the heaviest hydrocarbons start to condense from the gas phase into the liquid phase. ISO-11150:2007 and ISO-14532:2005 standards define the HCDP as the temperature above which no condensation of hydrocarbons occurs at a specified pressure.
The need for precise HCDP measurements arises due to two main reasons. Firstly, it is crucial for safe transportation through pipelines. Secondly, if the hydrocarbon gas temperature drops below its dew point, even a small amount of liquid condensation can lead to a significant loss in hydrocarbon substance, resulting in decreased heating value. This leads to lost revenue for gas producers as they have to supply more gas to compensate for the decreased heating value.
Various methods have been developed to measure the hydrocarbon dew point temperature in the industry. Manual or automatic chilled mirror instruments are the most well-known for direct measurement of the gas dew point value. However, for most general dew point temperature measurements, relying solely on directly measured samples is impractical due to the high dimensionality of the gas composition vector, demanding many examples for precise HCDP value interpolation.
Gas Chromatography (GC) offers an indirect method for HCDP determination. The gas composition obtained through GC, along with the sample pressure, serves as input for an analytic HCDP calculation, providing a practical approach for HCDP calculation in a production environment. ISO 23874:2006 outlines the gas chromatographic requirements for HCDP calculation and specifies the performance specifications for gas chromatographs analyzing natural gas.
In practical experience, marginal cases of non-typical natural gas composition, such as those with a high CO2 proportion, can occasionally lead to numeric convergence problems during HCDP calculations using traditional methods. These convergence issues may result in inaccuracies and reduced precision in HCDP determination, impacting gas producers' operations and revenue.
Modcon.AI Machine Learning (ML) tools offer a solution to address the convergence problem and enhance the precision of HCDP calculations. In a recently implemented project, the effectiveness of Machine Learning in HCDP calculation was demonstrated. The polynomial approximation results for four modelled gases showed improved accuracy and robustness through ML techniques.
领英推荐
By employing a different HCDP calculation paradigm, ML techniques can control the ratio between the data size in the training set, originating from the Equation of State (EoS) equation, and the empirical measurements. This enables the ML predictive model to shift the resulting phase curve closer to the analytical baseline or the directly measured HCDP values, eliminating the need for iterative convergence. Interestingly, increasing the ratio of directly measured HCDP samples added to the ML model's training set does not require taking more measurements. Instead, reusing the empirical measurement samples multiple times yields the same desired effect. The polynomial approximation results for the four modelled gases are shown?below.
In conclusion, HCDP calculation is essential for the gas industry, impacting safe transportation and heating value. Gas chromatography, combined with Equations of State, is commonly used, but it may encounter convergence problems in non-typical gas compositions.
Modcon provides complete integrated analyzer systems and all related services from initial engineering through manufacturing, testing and field start-up. Modcon.AI ML techniques offer a practical and efficient solution to enhance HCDP prediction based on gas composition obtained through GC, along with the sample pressure, without the need for iterative convergence. Implementing ML tools can lead to more reliable and accurate HCDP measurements, contributing to improved operational efficiency and revenue optimization in the gas industry.
Head of Analytical Instrumentation dept.
1 年Any HC DP analyzer (with direct measurement) is much more accurate and chipper than any GC equipped with any EOS...?
Bringing clarity to liquid carryover events
1 年We install process cameras onto gas pipelines to detect contamination and frequently detect the onset of mist flow or stratified liquid flow in sales gas pipelines while the HCDP continues to report dry gas. The last thing you want to do is allow liquids into a GC so complying with API 14.1 ensures that HC liquids are rejected and therefore not included in HCDP or BTU calculations. Consequently we believe that many operators are misled, beliving the gas is dry when a camera clearly shows that the gas is wet.
Freelance Trainer Process Analyzers & Sampling Systems and Custody Transfer / Fiscal Metering,
1 年Gregory Shahnovsky. Is 1+1= zero? How can there be still errors of more than 30 degrees be observed in practice by many companies who are using GC models? A claim of theoretical accuracy of almost zero remains just fiction not reality! As Mr.Sohrab correctly wonders. What is your reference accuracy which should then be unrealistic? Combined errors determine the uncertainty so when only the preparation of a certified primary reference gas is already higher than 0.2% it is simply impossible to claim a smaller overall uncertainty as also the GC analysis systemic errors should be included. One plus one will in practice never become zero! Practice also still proves that models have significant errors compared to direct primary measurements that are inherently more accurate than indirect measurement. This despite Modcon's efforts to prove that 1+1 is zero.
Director, Process Analytical Middle East WLL
1 年Not easy. Putting aside the debate of GC or direct measurement and potential errors the only thing that matters is that your HCDP calculations agree with +- ?hcdp of the customers process simulator values for an exact stream composition and further tested at different compositions so they meet their custody transfer agreement. So how well did you do? In 2007 I designed and supplied custom HCDP calc system for multiple GCs at PDO Oman's Central Gas Gathering Facility. The approved thermodynamics software had a comprehensive BIPS library and had to meet Shell's /Aspentechs HYSIS simulations. It did so to within +-0.2DegC HCDP. It was the most complicated OPC data system I have ever done and was in service for some 15yrs.
Or you could actually use an HCDP analyzer to measure it directly at a lower cost and much higher accuracy, without any guesswork. https://www.zegaz.com/post/a-fallacy-calculating-hydrocarbon-dew-points-from-gc-based-measurements