July 2021


Debottleneck analysis on a coker debutanizer via simulation models

Production requirements for more valuable refined products and/or increased refining flexibility are increasing due to new regulations.

Production requirements for more valuable refined products and/or increased refining flexibility are increasing due to new regulations. Delayed coking units (DCUs) help refineries to achieve these goals and maintain sustainable profitability in shrinking market conditions. Therefore, it is important that DCUs must be problem-free facilities since any struggles in these units will negatively affect refinery margins.

A DCU consists of two primary sections: the coker section and the gas plant section. The coker section is designed to produce coke, light- and heavy-coker gasoils, unstabilized naphtha and overhead vapors. The unstabilized naphtha and overhead vapors are further processed in the gas plant section to produce treated fuel gas, LPG and stabilized naphtha. In the gas plant, stripper and debutanizer sections of the DCU are connected via streams and heating systems—therefore, a problem in any of these sections affects the other sections. The authors’ refinery has struggled with a bottleneck problem in the debutanizer column at turndown conditions. The column’s pressure profile increased rapidly and would trip the pressure safety valve (PSV) if feed decreased by 50%.

To solve the bottleneck in the debutanizer and stripper sections of the DCU, column models were prepared and validated with site data using a commercial simulation program. After this step, debottleneck studies were performed to solve the problem and sustain production without giveaways. These studies demonstrated that looking inside and outside the affected columns simultaneously can be a critical factor when evaluating possible solutions. The simulation study is a more comprehensive review to provide a guideline on anticipating, troubleshooting and overcoming bottleneck problems.

After implementing the simulation results to the field, the refinery’s energy intensity value decreased 0.06 points and its carbon dioxide (CO2) emissions decreased by 1,500 tpy.

Why was this study needed?

After the startup of the DCU, it was observed that the debutanizer column performance was unstable, especially at a low feed rate. Sometimes, these instabilities resulted in a sudden pressure increase at the top of the tower, causing the PSV to trip. The aim of the project was to analyze the behavior of the debutanizer at turndown conditions and to eliminate the environmental effect by preventing a PSV trip.

To discover the bottleneck of the system, debutanizer and stripper column models—with condensers and reboilers—were simulated by using a proper data set. This set included capacity rate, maximum capacity and turndown capacity. Simulation models and data historian analyses were used simultaneously to determine the root causes of the problems.

The study’s first phase focused on trying to discover processing problems via simulation models. The second phase was verifying the simulation results via data historian analysis. The third phase focused on improving solutions in the simulation models and then implementing them onsite. This study enabled plant personnel to detect trouble spots and to put forth solutions. It was essential to properly define the debutanizer feed; therefore, the stripper column was also covered in the simulation. Both columns needed to be simulated simultaneously, since these two systems were connected via outlet/charge streams and reboiler systems. All related operating parameters were revised to find the optimum unit operation according to unit capacity. Simulation model outputs were used to solve bottlenecks in the column’s operation.

Main parameters used in the debottleneck study

Although DCUs are designed to maximize diesel cut, these units also produce a high amount of fuel gas, LPG and naphtha. LPG purity is higher vs. other LPG production units. LPG purity directly depends on the debutanizer column’s performance. In addition, naphtha taken from debutanizer bottoms is sent to the naphtha desulfurizer. The stable flow of naphtha is required for the sake of catalyst activity. Consequently, the debutanizer column performance is important for the DCU. To illustrate this system, a schematic view of the process is shown in FIG. 1.

FIG. 1. Overview of the debutanizer and stripper sections of the DCU.

One of the main performance parameters in the debutanizer column is feed characteristic. To define feed characteristics properly, stripper columns and reboilers were simulated, as well.

Operational parameters that dramatically affect column performance were studied via simulation models. The following are operating parameters and their possible effects:

  • Stripper column bottom temperature. Stripper column bottom temperature is a key parameter. The light ends and water should be separated in the stripper column, and it is not expected to carry these to the debutanizer column. For better operations, the stripper bottom temperature must be at an optimum by adjusting the stripper reboiler duty.
  • Debutanizer column recycle flowrate. Recycle flow is supplied from the debutanizer overhead drum to the debutanizer column. The aim of this recycle is to optimize column separation. Proper recycle flowrate is crucial for debutanizer operation.
  • Debutanizer column feed rate. It was observed that the debutanizer column performance was fluctuating at a low debutanizer feed rate. Therefore, a column tray analysis was performed, and the weeping condition in all trays was obtained.
  • Debutanizer column bottom temperature. The temperature at the bottom of the debutanizer was also one of the parameters affecting column performance. For the debutanizer top section load, the bottom temperature and reboiler duty needed to be optimized.

Simulation of the coker debutanizer and stripper columns

A commercial simulation program was used to achieve accurate simulation models on the stripper and debutanizer sections of the DCU. The configuration of the simulations covered rigorous column, condensers (heat exchangers and air coolers), reboilers, drums, pumps, valves and pipe segments.

The first step in simulation modeling of the unit was to carry out several selections and identifications. To estimate the properties of hydrocarbons, water and steam, appropriate property packages should be used. In this study, the Peng-Robinson equation of state (EOS) and ASME steam fluid packages were used. A unit process flow diagram was the basis for the simulation flowsheet.

The debottlenecking study via simulation models had the following primary equipment: a stripper column, a debutanizer column, condensers (heat exchangers and air coolers), reboilers and overhead receiver drums. The primary goal of the study was to investigate and provide solutions to a bottleneck problem in the debutanizer column section. Product back blend was used in all simulation case studies at the beginning of the detailed analysis. The base model of the stripper and debutanizer section of the unit was completed (at steady-state conditions, according to normal operational conditions) and then used for the bottleneck analysis.

Simulation model validation

Simulation models were validated according to the unit’s design data or field data before they were used in the study. The steps used during this validation are summarized here:

  1. Generation of the model, using field data: Plant data (such as distillation, flowrates and operational data) and equipment process data sheets were used to configure the model. The debutanizer and stripper column models were completed according to steady-state operational conditions via two different data sets. Using different data sets, feed conditions were varied to obtain the most validated simulation models and to witness the effects of operational parameters on process dynamics, such as bottleneck conditions.
  2. Validation of the model results: Based on predefined validation limits, the simulation results were validated according to plant data. Critical parameters for validating the simulation model include:
  • Product flowrates (LPG, light naphtha and heavy naphtha)
  • Product specifications for light naphtha, heavy naphtha (ASTM D86 T5-T95) and LPG (C5 limit)
  • Temperature and pressure profiles of the stripper and debutanizer columns
  • Condensers (heat exchangers and air coolers) and reboiler outlet temperatures.
  1. Acceptance of the deviations from actual data: When the model was compared with plant data,
    the results were acceptable if the difference was
    ±5%. These criteria were based on possible measurement errors and general refinery practice.

Debottleneck analysis

Before evaluating alternative solutions for a bottleneck analysis, the unit base model was created according to the 8-hr average values of a day when product and charge analyses were made, and production was stable. Because bottleneck problems occurred at low flowrates, the debottleneck study was performed via turndown simulation models. From the simulation model results, the authors observed the following:

  • The simulation results showed weeping in all trays of the column at turndown conditions.
  • The lower feed was raised to higher temperatures in the reboilers of the debutanizer column, with vapors remaining nearly unchanged at the top (given the weeping conditions).
  • If the column began to weep, then there was no equilibrium in the trays, and the vapor from the feed (including the vapor produced by the reboiler) bypassed the trays on its way up the column. This typically causes higher pressure and very low tray efficiency.
  • If the reboiler duty was kept the same, then more light components could reach the top of the column, thus increasing pressure and tripping the PSV.

This sequence obtained from simulation models was one of the bottleneck problems in the debutanizer columns section. The second problem included the following:

  • During a troubleshooting review, it was also observed that water was continuously being removed from the debutanizer reflux drum.
  • The reflux drum is equipped with a control valve on the water drain, which normally does not open during normal operating conditions.
  • At lower feed loads, the control valve opening was high, which corresponded to the discharged water flowrate from the reflux drum. This was not an expected issue because dissolved water had to be taken from the stripper column—and the debutanizer feed had not contained water composition.
  • Simulation models showed that stripper reboilers (these exchangers used hot debutanizer bottom products as heating fluid) did not have enough duty at lower feed rates.
  • The lower feed fell to lower temperatures in the reboilers of the stripper column, and then water in the reboiled stream could not be removed.
  • Water was carried via the stripper bottom to the debutanizer column at these conditions, thereby upsetting operations and tripping the PSV in the debutanizer.

Implementation of model results in the field

As a result of the simulation modeling, the stripper bottom temperature began to stay at optimum temperature (it was extracted from simulation models) and not carry light ends and water to the debutanizer column, especially at low feed rates. When the water reached the top of the debutanizer, there was a freezing risk that could result in a PSV trip. To prevent freezing and blocking, electrical tracing was used in pipelines and process instruments.

Following the implementation of these model results, high-pressure steam savings increased by 3.36%. Additionally, the refinery’s energy intensity decreased by 0.06, while CO2 emissions decreased by 1,500 tpy. This was achieved by reducing high-pressure steam consumption in the process via decreasing fuel gas rates at conventional boilers in the refinery. Avoiding PSV trips also helped the refinery decrease losses and mitigate environmental issues. This was an invaluable gain for the refinery’s reputation.


This study analyzed the effects of each operating parameter, both individually and collectively. When a system is viewed in its entirety, some parameters have a snowball effect, while some have a diminishing effect. Therefore, before performing any debottleneck study in a process and/or in a specific section of a process, it is better to check the overall effects. In this case, by understanding and targeting the source of the bottleneck, the first step was obtaining the root causes of the problem based on the symptoms and on the unsteady operation of the stripper and the debutanizer under turndown loads. The second step was optimizing operating conditions according to the simulation results.

In many situations, it is uneconomical to solve the problem—especially for large-capacity units in a refinery—with conventional trial-and-error methods, which can cause off-spec products or upsets in refinery units. However, the authors found alternative solutions via simulation models and corrected the problems on the unit’s data historian. After discovering the root causes of the problem, alternative solutions were applied on the models first, prior to implementing them onsite. This solved the problem in the most economical way. Therefore, to obtain a correct diagnosis of a problem, it is important to first understand the underlying principles before addressing the problem. HP

The Authors

Related Articles

From the Archive



{{ error }}
{{ comment.comment.Name }} • {{ comment.timeAgo }}
{{ comment.comment.Text }}