The chemical kinetics of the semi-open hydrous pyrolysis system: Time series analysis of lithostatic pressure and fluid pressure

Highlights

• A microreactor was used to artificially mature the marine source rocks.

• The yields of generated, retained and expelled oils from mudstone were quantified.

• Time series analysis was used to determine kinetic parameters.

• The frequency of expulsions is meaningful for a kinetic model of generation.

• The peak of oil generation is delayed by the episodic-hydrocarbon expulsion.

Abstract

As a problem that has plagued geochemists, the perspective of time has long been recognized as an important factor in organic matter evolution and hydrocarbon generation. This article discusses two time series analysis methods based on econometrics, damping trend and autoregressive integrated moving average (ARIMA), to mine the potential value of data and abstract useful information from pyrolysis experiments. The case studies use marine source rocks (depth of 3119–3230 m, in the LH29–2 well) from the Enping Formation in the Baiyun Sag, the deep-water area (depth of >300 m) of the northern South China Sea, which has a geological background of high temperature, high pressure and rapid burial in the late stage. A tubular plug flow microreactor is used to artificially mature the source rock in the laboratory, and the conditions of hydrous pyrolysis are listed as follows: temperatures of 250°C, 300°C, 350°C, 400°C, 450°C, and 500°C; lithostatic pressure of 33~89 MPa; fluid pressure of 16~38 MPa; and time duration of 72 h. The results revealed that as the degree of maturation increased (R_o from 0.27 to 2.04), the cumulative yield of hydrocarbon gas increased rapidly (max = 311.64 ml/g⋅TOC), the yield of the expelled oil increased first and then decreased (peak = 350.69 mg/g⋅TOC at 350°C), and the yield of residual oil extracted from residue decreased gradually. Although the kinetic parameters of kerogen can be calculated by conventional pyrolysis experiments, the long-term cumulative effect of episodic-hydrocarbon expulsion in real conditions is often neglected. The major contributions of our statistical approach for geological applications are described as follows: (1) the recognition of qualitative changes within hydrocarbon generation and expulsion; (2) the characterization of episodic-hydrocarbon expulsion that describes the evolution of the lithostatic pressure and fluid pressure; (3) the disclosure of new influencing factors abstracted from time series models to describe the frequency distribution of hydrocarbon expulsion; and (4) an evaluation of the dependency of new variables.

Introduction

In recent years, the deep-water exploration block (depth >300 m) has been one of the most important global areas of growth for conventional oil and gas reserves (Zhang et al., 2011; Wang et al., 2018; Yang et al., 2018; Zhang et al., 2019). The global deep-water industry reached a milestone in 2019 – when it surpassed the 10 million boe/d mark according to the reports from Wood Mackenzie Lens®.

The South China Sea (SCS), the Gulf of Mexico and the North Sea Basin (NE-Atlantic) are the three major high-temperature and high-pressure basins in the world. The SCS is located in the convergent hinge of the European-Asia plate, Pacific plate, and India-Australia-plate, where oil and gas exploration and development are facing a series of world-class theoretical and technical challenges, such as high temperature, high pressure and the depth of water (Zhang et al., 2019). A large number of new methods have been developed and applied in the deep-water area of the northern South China Sea (Liao et al., 2016, Liao et al., 2018; Kong et al., 2018; Yang et al., 2018). However, the quantitative study of the mode and mechanism of hydrocarbon generation and expulsion continues to rely on traditional geochemical methods, which are methods of visual observation or instrumental analysis. In this paper, the method of laboratory-simulated semi-open hydrous pyrolysis is designed to reveal the model and mechanism of hydrocarbon generation and expulsion for marine source rocks (Enping Formation) in the background of high temperature, high pressure, and rapid burial in the deep-water area (Kong et al., 2018). Compared with previous studies (Fu et al., 2013; Tang et al., 2014; Li et al., 2016b; Jiang, 2017) of the variations in yield during heating experiments in this area, this study investigated the changes in gas yield (hydrocarbon gas, nonhydrocarbon gas), liquid yield (expelled oil, residual oil), and Rock-Eval parameters of solid residue and episodic-hydrocarbon expulsion in the conditions of a temperature range of 250~550°C and a fluid pressure range of 16~38 MPa.

Natural hydrocarbons are most often formed by thermal decomposition of organic matter (OM) following a kinetic law that is dependent on both time and temperature (Tissot et al., 1987). Accordingly, the petroleum industry has performed laboratory pyrolysis experiments to artificially mature source rocks at relatively high temperatures and short times in an inert atmosphere (Lewan, 1997). The laboratory pyrolysis experiment developed based on the time-temperature relationship and Arrhenius theory to artificially mature source rocks is an important method for exploring the genesis mechanism of hydrocarbons and evaluating the hydrocarbon generation potential (Connan, 1974; Hunt et al., 1991). These artificial maturation techniques enable us to study the effect of several conditions present in the natural system, which influence the generation, expulsion and composition of petroleum (e.g., temperature, time, pressure, presence of water, source rock mineralogy and retention) and determine kinetic models that, when extrapolated to geological conditions (lower temperatures and longer times), reproduce the timing and extent of petroleum generation in a sedimentary basin (Lewan et al., 1979; Spigolon et al., 2015).

During the past 30 years, the methods of pyrolysis experiments have seen great improvement and provide a basis to better determine the petroleum formation process and petroleum compositional variations with increasing maturation. Some of these advanced techniques include Rock-Eval pyrolysis (Liao et al., 2018), pyrolysis–gas chromatography (Watson et al., 2012; Pan et al., 2015), gold tube pyrolysis, microscale sealed vessel pyrolysis (MSSV) (Liao et al., 2016), and a hydrothermal diamond anvil cell (Huang, 1996; Mo et al., 2008). Three types of pyrolysis systems are generally employed for these experiments: an open system, an anhydrous or hydrous closed system and a semi-open system. First, an open pyrolysis system (e.g., Rock–Eval and PY-GC–MS) can be used to determine the quantity of hydrocarbon generation; however, it only addresses the primary reaction and dose not address the experimental conditions of pressure and water (Espitalie et al., 1985; Behar et al., 1997; Liao et al., 2018). Second, the closed pyrolysis system (e.g., autoclave, sealed gold tubes, and MSSV) can simulate the quantity of gaseous hydrocarbon and involves the cracking of liquid hydrocarbons at higher temperatures. The overlap between primary reactions and secondary reactions occurs in the closed system (Dieckmann et al., 2000; Gai et al., 2018; Shao et al., 2019).

Typically, a semi-open system that has an inlet and an outlet is separately controlled by back-pressure regulators to maintain the internal pressure by balancing the generation reaction rates with the escape rates (Burnham, 2017; Ma et al., 2020). In this work, deionized water was employed as the internal pressure medium of our reactor, and the values of lithostatic pressure and fluid pressure variables per minute are recorded during the experiment. Therefore, we define a semi-open system as a type of tubular plug flow microreactor. As the plug of fluid flows through the plug flow reactor, reactants are converted to products. In an ideal plug flow reactor, no mixing of the medium along the long axis (X-axis) of the reactor is assumed, although lateral mixing in may occur in the medium at any point along the long axis (i.e., Y-axis). The residence time of pyrolysis products inside the particle depends on the pressure of the reactor due to both the inhibition of liquid vaporization and the smaller volume of a generated mass of gas, and further reaction is quenched once the products are expelled from the particle (Burnham, 2017; Takahashi and Suzuki, 2017; Ma et al., 2020). Furthermore, a semi-open system can provide increasingly more complex process information in the process of artificial simulation of natural oil production and it is conducive to the study of the dependence of product composition on temperature, pressure, time and water (Burnham and McConaghy, 2014; Sun et al., 2015; Wu et al., 2016; Takahashi and Suzuki, 2017; Ma et al., 2020). Relatively, the semi-open system has provided experimental conditions that resemble natural geology.

Time series analysis is a well-established topic in mathematical statistics. A satisfactory review of the history of this topic is provided in Paolella (2018). Mathematical geoscientists should analyze geoscientific time series and space series. Based on the time series definition, the serialized data recorded by a programmable logic controller (PLC) in a semi-open hydrous pyrolysis experiment are pretreated by pure randomness and a stationarity test. The results showed that the data of lithostatic pressure and fluid pressure are recognized as a nonstationary time series of deterministic trends and random trends, respectively. As one of the exponential smoothing methods, the damped trend method is extended to fit the lithostatic pressure data with a deterministic trend. Since the difference operation is very suitable for addressing the random process data of a nonstationary series, the first-order differential ARIMA (Autoregressive Integrated Moving Average) model is adopted for fluid pressure. As geochemical workers, we only have limited knowledge about the hydrocarbon generation processes that determine the observed data. While models that involve these data are formulated by economic theory and then tested using econometric techniques, the theory is inherently insufficient. For instance, the theory may provide minimal evidence about the processes of adjustment, which variables are exogenous, and which variables are irrelevant or constant for the particular model under investigation. A contrasting approach is based on statistical theory, which involves an attempt to characterize the statistical processes where the data are generated. The frequency distribution of hydrocarbon expulsion is further discussed by joining the yield and time series model.

In this article, we aim to study the interaction between the “generation”process and the “expulsion” process and the law of the timing of hydrocarbon expulsion in the overpressure zone based on the results from the semi-open system pyrolysis experiment and time series analysis for lithostatic and fluid pressure. This work not only re-evaluates the effectiveness of the simulation technology but also provides a feasibility guide to establish a hydrocarbon generation kinetic model for a semi-open hydrous pyrolysis system.