1. Introduction
Delayed coking is one of the important processes in refineries. It is used for the thermal cracking of high molecular weight feedstocks (usually vacuum tar in a vacuum oil producing unit), which are converted into acid gas, naphtha, light oil, heavy oil, and coke.
Figure 1 illustrates the typical operation process of a Delayed Coker Unit (DCU). The feedstocks are first heated and then transported to the bottom of the fractionator to cool the superheated steam, and the preheated feedstocks from the bottom of the fractionator, together with the steam with higher water content, are pressurized by the pump into the radiant section of the furnace for rapid heating. After partial evaporation in the heater tube, the feedstocks are introduced into one of the coke drums for coking reaction, and then, high-pressure water is injected into the furnace tube to minimize the tar deposition and delay the coking reaction in the furnace tube. The superheated steam from the top of the coke drums is pumped back to the bottom of the fractionator and further separated into various products such as liquid natural gas, naphtha, light oil, and heavy oil, according to their boiling points. The gas at the top of the fractionator is cooled in an air cooler at the top of the fractionator [
1].
Depew et al. [
2] proposed a process model for simulating DCU operation. The authors used the simulation results to evaluate different control systems. Vatsal Kedia et al. [
3] developed a numerical method to estimate the standard deviation reduction rate of the main control variables of DCU operation. The authors combined the method with MATLAB and applied it to the actual DCU process of refineries subject to periodic disturbances to demonstrate the effectiveness of the method. Zhang and Yu [
4] used a radial basis function (RBF) neural network to establish a multivariate model for DCU liquid products. The model can provide the production ratio of gasoline, diesel, petroleum tar oil, etc., as well as the overall production proportion of liquid products. Chen and Wang [
5] used the n-d-M, E-d-M, and hydrocarbon analysis methods to analyze the composition and properties of delayed coking feedstocks (vacuum oil residue, fluidized catalytic cracking slurry) and ethylene tar. Their results showed that mixing ethylene tar into delayed coking feedstocks would lead to a decrease in the saturation of the feedstocks and an increase in the aromatic hydrocarbon content in the feedstocks. Lei et al. [
6] established a comprehensive optimization model for the hierarchical structure of the heat exchanger network, with heat removal in the complex fractionator as the main coupling variable. They compared the results of three optimization studies. The authors pointed out that it is better to consider steam generation in the comprehensive optimization. Ge Xin [
7] successfully analyzed and solved the problem of waste oil blending through a technical optimization program and the results showed that the program could increase diesel production by 2.59% after blending waste oil. Deng et al. [
8] solved the problem of the use of catalytic cracking slurry by testing the high proportion of catalytic cracking slurry. The results showed that when the mixing ratio of catalytic cracking slurry increased from 25% to 29%, the production of petroleum coke decreased significantly, while the production of kerosene, light oil, and the total production of liquid increased. Fan et al. [
9] applied a thermal load automatic adjustment simulation method to explore the three-point steam injection of DCU, and analyzed the effect of three-point steam injection rate on the coking degree and heat consumption. The results showed that the steam injection rate affected the heat consumption and coking degree. The coking degree could be reduced by improving the steam injection rate. Paladino et al. [
10] developed a CFD model that explores the DCU scrubbing zone (including steam and scrubbing liquid) and considers the heat and mass transfer among different phases to predict the region where the vapor reaches the desired temperature and to avoid the formation of coke in this region. The model can reproduce the complex phenomenon of interfacial heat and mass transfer in multi-component multiphase flow. Díaz et al. [
11] applied CFD to simulate the DCU of a pilot plant. They simulated the cooling process of three different vacuum residue coking beds and compared the results with experimental data. Salma Abdalla Mohamed Ibrahim et al. [
12] conducted computer simulations of two crude oils to obtain the ideal blending ratio of heavy crude oil as a crude oil substitute in DCU. The results of experimental tests and computer simulations showed that mixing 50% DAR blend with 50% Fula blend in DCU can significantly improve the quality and quantity of the product. Albers [
13] developed models for predicting product quantity and quality. The authors tried to use three different methods, including kinetics, Monte Carlo, and empirical methods, to improve the control and optimization of the delayed coking process. Pedro Amorim Valenca et al. [
14] used a two-dimensional axisymmetric numerical simulation to investigate an experimental coke furnace receiving nitrogen. Based on safety and process considerations, the device was preheated at a specific temperature. The results predicted a linear temperature profile, which was the same as the experimental trend. Samy Nabil Mohamed [
15] used Aspen HYSYS modeling to optimize the DCU process parameters and compared them with the original design conditions to achieve maximum diesel production with process safety in consideration.
In terms of pipeline research, Dai et al. [
16] discussed the types and causes of pipeline vibration, and proposed relevant anti-vibration methods to improve the safety of the pipelines. Chen et al. [
17] discussed the deformation and overall stress of a pipeline under different wind speeds. They analyzed the stress distribution of a pipeline under the action of a magnitude 17 wind to confirm its safety under a strong wind. Wu et al. [
18] discussed the influence of seismic action direction on the pipeline displacement direction. The authors used dynamic stress analysis to study the pipelines located in a seismic zone. The results showed that any seismic action direction would cause a significant effect on the lateral displacement of the pipelines. Zhou et al. [
19] discussed the stress distribution of the pipelines. The authors introduced the pipeline stress classification and the content of the pipeline analysis, which are helpful in identifying the methods of the pipeline analysis more quickly. Ren et al. [
20] used ANSYS to establish an elbow model. The simulation results showed that the maximum stress occurred at the elbow and changed with different lengths. Zhou [
21] studied the problem of pipeline damage and used the finite element method to explore the dynamic characteristics of the pipelines under different conditions. Zhao et al. [
22] discussed the vibration and the fatigue damage caused by the fluid flow in pipelines. They proposed vibration reduction methods in terms of the structure and fluid flow to improve the safety and service life of the pipelines. Yin et al. [
23] discussed the vibration generated by the outlet valve of the oil pump. The authors used a spectrometer and analysis software to measure and optimize the pipeline system. Xu et al. [
24] discussed the force distribution in pipes with different diameters and wall thicknesses under earthquakes. They found that the pipe wall thickness is the dominant factor of pipe damage. Terán et al. [
25] applied the nonlinear finite element method (FEM) and 3-D pipeline models to investigate the pipeline failure pressure and mechanical behavior. The failure pressure was predicted by FEM and compared with the results of traditional methods to determine their applicability in considering the interaction of pitting corrosion defects.
From the review of the above references, it can be seen that most of the DCU-relevant studies focus on optimizing DCU process parameters to improve yields. Very few studies discuss the abnormalities of DCU pipeline systems. In view of the great impact of the DCU pipeline system on production and operation safety, we applied CFD in a previous study [
26] to investigate the flow in a DCU fractionator overhead vapor line connected to an air cooler, as shown in
Figure 2. Because of the complex geometry and flow development in the pipelines, damage to the pipes has been found. For example, leakage near a T-junction at the east part of the pipeline system was found. Further, the pipe wall thickness at the east part of the pipeline system became thinner. Ref. [
26] discussed the causes of the damage and the strategies to alleviate the occurrence of the damage. It is found that when two sections of 24″ pipes are connected and five sections of 18″ pipes are also connected, but the 30″ main pipe is not raised, better force uniformity and lower total force can be obtained. In this paper, we further apply the finite element method to analyze the structure of the DCU pipeline system to confirm the strength of the original and the improved pipeline systems. The purpose of this series study is to explore the flow development and the structural strength of the DCU pipeline system to improve its operational safety. The material of the pipeline system is A106 Gr. B carbon steel with density 7850 kg/m
3, thermal expansion coefficient 1.2, Young’s modulus 200 GPa, Poisson ratio 0.3, yield strength 240 MPa, and UTS 415 MPa. The outside diameter (Do) and wall thickness (t) for the pipeline system are listed in
Table 1.
2. Numerical Methods
In this study, the numerical model of the pipeline system is established by using SOLIDWORKS(R) Premium 2020 SP0.1 [
27], and the ANSYS Workbench 2020 R1 [
28] is employed for the stress analysis. SOLIDWORKS 2020 SP0.1 is a computer-aided design software developed by SOLIDWORKS, a subsidiary of DASSAULT SYSTÈMES in Vélizy-Villacoublay, France. It is mainly used in engineering design, product design, and modeling, and is widely used in many industries, including machinery, electronics, construction, automotive, aerospace, etc. ANSYS Workbench 2020 R1 is an engineering simulation software developed by Swanson Analysis Systems in the United States. It has extremely realistic simulation analysis capabilities and is widely used by engineers and designers in various fields, greatly reducing the time required to solve engineering problems.
Structure analysis has been widely used in industrial design. According to the nature of mechanics, structure analysis can be classified as static analysis and dynamic analysis. The latter includes modal analysis and harmonic response analysis. In structure design, static analysis is usually used to evaluate the displacement, strain, and stress of the structure due to inertial forces, e.g., gravity, centrifugal force, etc. Modal analysis is one of the methods of structure dynamic analysis. Mode is an intrinsic vibration characteristic of a linear system. Each mode has its own specific intrinsic vibration frequency, form, and damping ratio. Harmonic response analysis can be used to analyze the response of the structure to the time-variant harmonic wave. Through the solution of frequency domain, the frequency response of the structure to harmonic wave loadings can be obtained to examine if the system can alleviate the resonance, fatigue, or other harmful effects. The solutions of the equations for static and dynamic analyses of a structure can be found in books on structure engineering and mechanics, e.g., [
29,
30,
31].