Dynamic analysis and optimization of flare network system for topside process of offshore plant

https://doi.org/10.1016/j.psep.2019.12.008Get rights and content

Highlights

  • Design and optimization of flare relief system in emergency situation at offshore plant.

  • Designing flare relief system at steady-state has an excessive margin for line size.

  • Reduction of overall capital cost is achieved by using dynamic simulation.

  • Enhancement of flare system safety considering uneven distribution of flare loads.

Abstract

This study introduces a new approach for designing a flare network system that ensures economic feasibility and safety using dynamic simulation analysis with the gPROMS ProcessBuilder software. A case of “separator outlet blocked discharge” was selected as a pressure-relief scenario based on the American Petroleum Institute Standard 521, and design data from a previous offshore project were used. As the main process and flare network system were dynamically simulated, the effects of the pressure, temperature, and liquid-level changes in the vessel on the opening of the pressure safety valve (PSV) were analyzed. Then, dynamic simulation results, including those for the flare load, PSV back pressure, and Mach number, were compared with those obtained from a steady-state model employing the Aspen Flare System Analyzer. Lastly, the sizes of the PSVs, branch lines, and main headers were optimized to minimize the overall capital costs and ensure the safety of the flare network system. This methodology can be applied to all existing and newly designed flare network system to enhance safety and reduce capital costs.

Introduction

In large-scale chemical plants, various types of equipment operate, mainly including rotating machines, separators, reactors, heat exchangers, and vessels. They normally run under the allowable operating conditions (e.g., temperatures, pressures, and flowrates) specified by either the manufactures or the operators. However, in emergency situations, such as fires, utility failures, and outlet blockages, such equipment tends to operate abnormally, leading to overpressure of the equipment (American Petroleum Industry, 2014). This may result in serious fires and explosions, causing severe damage to humans, property, and the environment (Shin et al., 2018). To prevent such accidents, pressure-relief systems are employed in most chemical plants. This pressure-relief systems protect the process equipment and relieve overpressure that arises during normal or abnormal operation.

Furthermore, the emergency shutdown (ESD) system is widely used to protect the equipment and humans. Pengyou Zhu (Zhu et al., 2019) developed an overall framework to identify gaps and challenges of ESD system based on the field study in Norway. The authors mentioned that the workflow of ESD system includes three key parameters (i.e., technical, operational, organizational parameters). The technical parameter is related to equipment and systems, the organizational parameter to personnel with defined roles or functions and specific competence, and the operational parameter to actions and activities that the personnel has to perform. According to identification results using the framework, the authors found 15 challenges during the operation and maintenance of ESD system. Among the 15 challenges, three, seven and five of them were related to technical, organizational, and operational challenges, respectively. As a result, the paper provided effective guideline regarding the operation and maintenance of ESD system.

In addition, the emergency situation can be handled by applying an integrated robust and resilient control scheme to the process. In such a case, the role of robust control is to guarantee the product specifications in emergency situation, and the role of resilient control is to stabilize the process fast during an emergency situation (Jin et al., 2010). However, further research is necessary, and it is important that the flare network system should be integrated with such control system.

Moreover, in terms of risk assessment, the emergency scenario can be predicted based on historical data. Faisal Khan (Khan et al., 2015) presented the overall risk assessment method. In the study, the authors adopted various methods to manage the risk of the process. One of the methods, which used the historical data, was to combine the Fault Tree Analysis (FTA) with the Event Tree Analysis (ETA). FTA graphically illustrated the logical relationship between cause, consequence and failure paths. ETA evaluated the sequence of failure processes and analyzed its consequences. The authors described System Hazard Identification, Prediction and Prevention (SHIPP), which combined FTA and ETA. SHIPP is able to predict the probability of emergency by analyzing the history of emergency. As a results, the frequency of emergency situations can be predicted.

Fig. 1 shows the simplified structure of the pressure-relief system. It contains pressure-relief devices, headers and tail pipes that transfer flare loads, knock-out drums that separate liquid and vapor before combustion, and a flare stack that burns flare gas safely (Bader et al., 2011). Pressure-relief devices typically include a pressure safety valve (PSV), a blowdown valve (BDV), and a rupture disk, and they release fluid from the inside of the equipment to relieve overpressure. The PSV performs depressurization automatically under abnormal conditions that cannot be recognized by the operator. It normally opens via spring or elastic action when the pressure of the equipment reaches a designated set pressure. The BDV also performs depressurization, but it is generally used for manual operation to release fluid quickly from the inside of the vessel for process shutdown or equipment replacement. While the PSV constantly maintains the pressure of the vessel below the set pressure, the BDV reduces the pressure of the vessel to the level set by the operator (Bjeere et al., 2016). A rupture disc is installed in facilities or devices, such as pressure containers, pipes, and reactors, and it is ruptured when the pressure reaches the set point during operation to protect the equipment. Thus, it is only used once and should be replaced with a new one once it is ruptured. In general, it is used as a spare device of the PSV when the valve function is impaired because abnormal materials have accumulated in the PSV and the valve cannot respond quickly enough. The most commonly used relief device in emergency situations is the PSV (Roberts et al., 2004). The design of the PSV depends on the size of the orifice, the flare load, and its back pressure. Therefore, it is important to install PSVs of the correct type and size, as well as the correct number of PSVs, for safe depressurization.

Flare gas released from each pressure-relief device, which is installed in various equipment, flows through a tail pipe and a sub-header and gathers in the main header. Subsequently, the relieved vapor and liquid are collected in a knockout drum (K.O Drum). Then, gas is eventually sent to the flare stack for incineration before its release to the atmosphere (Lee et al., 2012). During normal operation of the flare system, when there is no released fluid from the PSV, small amounts of light hydrocarbon vapor flow through the header to maintain the operation of the flare network (Chopinet and Marcec-Rahelic, 2015). However, in the case of an emergency or abnormal operation, large amounts of released fluids flow through the header with a high velocity at a high pressure (Song et al., 2014). In general, the following aspects should be considered when engineers design flare pipes (i.e., tail pipes, sub-headers, and header pipes).

  • Limits of the PSV back pressure

  • Fluid Mach number

  • Momentum of the released fluid

Generally, the engineers consider the most governing PSV relieving scenarios when they design the flare system. For example, all feasible abnormal scenarios, such as total power failure, external fires, and equipment failure, are considered for designing the size of the flare pipes, size of the PSV orifices, number of PSVs, etc. Among them, the most governing case is considered for designing these parameters of the flare system. Furthermore, an additional margin is added to the developed design parameters to cover unanticipated emergency situations. This leads to increased capital costs of the equipment, limited space for equipment installation, and additional weight of the offshore plant (Goyal and Al-Ansari, 2009).

Over the past few decades, many researchers have investigated the design of the flare system. For example, Muktikana Sahoo (Sahoo, 2013) developed a steady-state model of the flare system in an external fire scenario using industry data and estimated the PSV back pressure and velocity of the relieving fluid. The results indicated that a high flare load in the case of a fire scenario leads to a high back pressure of the PSV, causing severe damage to the PSV. Thus, the study indicated that the diameter of the tail pipe must be increased or the tail pipe should be divided into two separate pipes to reduce the flare load for each pipe. Unfortunately, this study did not consider limit of the Mach number that affects the vibration of the pipeline.

In another study, Sisshartha Mukherjee (Mukherjee, 2008) numerically optimized flare networks in the case of cooling-water failure and external fire scenarios by solving governing equations. As one of the constraints for optimization, the maximum values of the PSV back pressure and Mach number in the header were considered. For estimation of these parameters, the American Petroleum Institute (API) Standard 521 was employed. During optimization, design variables, including the number of PSVs, configuration of the pipelines, and diameters of the pipes, were adjusted, satisfying the proposed constraints. However, only a steady-state model was developed; dynamic behavior, such as changes in the relieving loads over time, was not considered. Thus, the design margin was too large for designing pipes with the optimum size for flare networks.

Recently, Prashanth Siddhamshetty (Siddhamshetty et al., 2019) performed a simplified model of ignition phenomenon at a wellhead blowout situation. In this study, they reviewed previous papers comprehensively to analyze the relationship between a set of parameters and burning efficiency. Moreover, they developed a simplified model to calculate the burning efficiency, and the results showed high accuracy. After PSV opening, the liquid is generally formed in the pipe due to the depressurization. This liquid formation affects the flame length at the flare stack and flame temperature. In terms of the safety, prediction of flame length and the flame temperature is considered to be quite important.

Most flare systems of existing chemical processes were designed according to the steady-state condition. This method uses the sum of flare loads, which are collected in header pipes from different sources of PSVs, to design the diameters of the headers, as described in Fig. 2(a). However, the flare load that is released from each PSV has a peak point with regard to the flowrate immediately after its release, and the flowrate decreases, as shown in Fig. 2(b). Thus, if the diameters of the headers are designed according to the estimated maximum flowrates, the design margin becomes too large. Additionally, all the main equipment is installed with long distances for safety, and it takes different times for the released fluid from each PSV that is installed in the equipment to reach the main header. Consequently, the actual cumulative flare load that passes through the main header is smaller than algebraic sum of the flare loads from the PSVs. Therefore, it is necessary to estimate the actual flare loads that pass through the main headers over time from the point of their release until they reach the flare tips. However, few studies involving the analysis of flare systems with dynamic simulation have been published.

Gilardi and Tonello (Gilardi and Tonello, 2011) estimated the total flare load along the main flare header over time in a power-failure scenario by using a dynamic simulation model. The results indicated that the maximum accumulated flare load that was collected from various sources of PSVs was significantly smaller than the algebraic sum of the flare loads from the individual PSV sources. Alban Sirven et al. (Sirven et al., 2011) estimated the total flare load in the case of reflux failure of a crude distillation unit using a DynaSim simulation. The size and number of PSVs were optimized to minimize the total capital cost of the flare system. However, the study focused on the total flare load but did not consider the Mach number of the fluid in the pipes. Ali Shafaghat (Shafaghat, 2016) performed a dynamic simulation of the flare system in an external-fire scenario and analyzed the changes in the fluid properties over time with regard to the pressure, temperature, and mass flowrate during the depressurization period. Then, the estimated Mach numbers from steady-state and dynamic simulation models were compared. The estimated Mach number from the steady-state model exceeded the design criteria, whereas the Mach number from the dynamic model was significantly smaller than the design criteria during the whole relieving period, except for the initial 1 min. Thus, the size of the header pipes could be reduced. However, only the main header was considered in the study.

Most previous studies on the design of the flare network system involving dynamic simulation have mainly focused on comparing the flare loads, Mach numbers, back pressures, and sizes of the main header pipes in particular emergency scenarios. They seldom revealed the optimum size of the whole flare system, including the tail pipes that are connected to PSVs. Furthermore, onshore chemical plants were mainly considered by researchers, while its impact on capital cost saving is far larger for offshore plants.

Section snippets

General PSV relieving scenarios

In general, a PSV opens owing to overpressure of the equipment, which has various causes. Therefore, it is necessary to consider each scenario that causes overpressure of the equipment and apply the results to the design of the flare network system to ensure its safe operation. Relieving scenarios can be divided into two main categories. The first is the continuous-emission scenario, which includes purging, control valve leakage, and process venting, and the second is the emergency scenario, in

Separator outlet blocked discharge

In this study, “high-pressure separator (HP separator) outlet blocked discharge” was selected as a governing scenario. The feed stream to an HP separator is directly connected to a riser from the subsea region, and operating pressure of the fluid is very high. Therefore, if the outlet valve of the separator is blocked, the risk of explosion becomes high owing to the high-pressure stream inflow to the equipment. Fig. 3 presents a schematic of the “HP separator outlet blocked discharge” scenario.

Results and discussion

The simulation results of the dynamic model were analyzed. First, the dynamic behavior of the HP separator was analyzed. Then, the dynamic behavior of the flare system, such as the opening of the PSVs and the relieving loads from the PSVs, was analyzed. Lastly, the results of the dynamic simulation were compared with those of the steady-state model, and the optimum sizes of the PSV orifices and pipes were estimated according to the sizing criteria of API Standard 521. In this study, the

Conclusion

A governing scenario of “HP separator outlet blocked discharge” in an offshore plant was adopted to compare the design results of a flare network system between steady-state and dynamic models. For the steady-state model, the widely used software of the AFSA was used, whereas gPROMS ProcessBuilder was used to develop the dynamic simulation model. Compared with the steady-state model, the dynamic model estimated the behavior of the relieved fluids over time more accurately, including the

Funding

This research was supported by a Korea Institute for Advancement of Technology (KIAT) grant funded by the Korean Government (MOTIE) (P0008475, Development Program for Smart Digital Engineering Specialist). It was also supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (2019R1F1A1058979).

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

References (25)

  • F. Chopinet et al.

    Upgrade of a flare network for a new refinery unit

    Goriva I Maziva

    (2015)
  • C. Gilardi et al.

    Dynamic Simulation, HIPS & Combined Probability Analysis Reduce Flare Loads

    (2011)
  • Cited by (9)

    • Analysis of pressure behavior during runaway reaction with case studies of various depressurization designs

      2022, Process Safety and Environmental Protection
      Citation Excerpt :

      Safety valves are installed to prevent equipment rupture due to undesired pressure rise from reactions and failures during pressurization or transport operations. Typical abnormal events that initiate pressure rise are malfunction of control valves, cooling system failure of reflux or internal coils of the reactor, and tube failure in heat exchangers and fire case etc (Bjerre et al., 2017; Li et al., 2021; Jo et al., 2020; Yandrapu and Kanidarapu, 2021). Though there are many considerations for these cases, pressure rise caused by runaway reaction is difficult to predict.

    • Development of dynamic simulation model of LNG tank and its operational strategy

      2021, Energy
      Citation Excerpt :

      In both cases, the total simulation was set to proceed until the tank pressure reached the maximum allowable working pressure (MAWP) so that we could analyze the change in the operating condition during the entire operation period. In practice, when MAWP is reached, safety equipment such as the pressure relief valve is expected to prevent critical accidents [26]. Based on the calculated heat duty of the natural heat inflow into the storage tank, the pressure and temperature in the three sections (i.e., vapor, upper, and lower zones) were estimated during the entire operating period.

    • Revealing hidden debottlenecking potential in flare systems on offshore facilities using dynamic simulations – A preliminary investigation

      2020, Journal of Loss Prevention in the Process Industries
      Citation Excerpt :

      Dynamic simulations showed that the maximum backpressure calculated during a full plant depressurization was reduced by approx. 17% compared to steady-state simulation results. Jo et al. (2020) studied a separator blocked-outlet scenario with multiple PSVs discharging into the flare system of an offshore process facility using dynamic simulations. They found that substantial line sizing optimization could be obtained for most lines compared to a traditional steady-state approach, thereby reducing CAPEX.

    View all citing articles on Scopus
    View full text