August 2019

Process Control and Instrumentation

Estimation of relief load and realistic relieving temperature for heavy-end fractionating columns

Relief calculation is one of the most discussed aspects of chemical engineering design.

Saha, J., Fluor Daniel India Pvt. Ltd.; Chaudhuri, S., Bechtel India Pvt Ltd.; Groenendijk, S., Fluor

Relief calculation is one of the most discussed aspects of chemical engineering design. Licensors, contractors, industry literature and the American Petroleum Institute (API) specify the broad boundaries of “dos” and “don’ts” for relief system analysis and sizing. Still, much is left for engineering judgement to define the optimum safe design. This article examines the purview of relief load estimation and a realistic relieving temperature calculation method for distillation columns handling heavier cuts in a refinery.

FIG. 1. Schematic representation of hydrotreated gasoil product fractionation.
FIG. 1. Schematic representation of hydrotreated gasoil product fractionation.

The conventional approach of tower relief load calculation, especially for grassroots units, is to balance the unbalanced heat across the tower during an upset scenario. Although the unbalanced heat method has its limitations, it is one of the most trusted methods for relief load estimation for a distillation column. One of the basic assumptions for the method is an unlimited supply of liquid to the top tray, and the liquid is considered to vaporize from the top tray during a relieving scenario. This results in a conservative (high) relief load owing to the low latent heat of vaporization of top-tray liquid.

However, it is important to recognize that top-tray liquid is lighter and demonstrates a lower relieving temperature, risking incorrect material selection and design of the column overhead, relief valve and downstream system, in many cases. The effect on relieving temperature is more pronounced in a column that has a wide range of boiling temperatures between the top and bottom trays. The total inventory of the system, including the diameter of the column and the number of side draws and side strippers, is also critical in the scenario in question. If the reboiling/stripping is continued for a relief scenario, then the likelihood of column overheads being exposed to higher boiling fluids during that relief scenario is more realistic for a small-diameter column with no or a limited number of side draws, rendering the design overhead system vulnerable to high temperature exposure.

FIG. 2. Fractionator normal operating temperature profile.
FIG. 2. Fractionator normal operating temperature profile.

The case study presented here demonstrates the difference of steady-state relieving temperatures calculated by the unbalanced heat method (UBH) and a simulation method, and its impact on design temperatures of column overhead system and relief valve laterals for a hydrotreated heavy gasoil fractionating column.

Overpressure protection of hydro-treated gasoil fractionation system

A heavy gasoil fractionating column (FIG. 1) is downstream of a heavy gasoil hydrotreating reactor and its high- and medium-pressure separators. The hydrotreated gasoil is first stripped to remove hydrogen sulfide (H2S) and light gases in a stripper. The stripped gasoil is further fractionated in the fractionator column to produce the hydrofinished gasoil from the bottom of the tower. The feed from the stripper bottom to the fractionator is first heated in the feed/effluent exchanger, followed by the fired tube feed heater. The hot feed is fractionated in the tower with the assistance of live steam injection. A part of the overhead stream is taken out as naphtha product after condensation, and the other part is fed as reflux to the top tray of the tower.

A diesel side draw from the tower is further steam stripped in a side stripper, and the product diesel is taken out from the bottom of the side stripper. The vapor from the side stripper is returned to the fractionator. The column heat is removed by overhead condensation and reflux cooling.

Relief scenarios analysis

Upon careful analysis of all probable causes of overpressure, the following credible overpressure scenarios are identified as significant for the system:

  • Total power failure
  • Partial power failure
  • Reflux failure
  • Abnormal heat input from fired heater
  • External fire
  • Column overfilling
  • Steam valve failure in open position.

As per steady-state relief calculation, reflux failure (due to partial power loss to the reflux pump or reflux valve failure) is the governing case for relief valve sizing—i.e., results in the highest relief load. Additionally, this case also defines the governing relieving temperature for the unit flare header design.

Governing scenario: Tower reflux failure

Boundary conditions and probable impacts on the system on loss of reflux are listed in TABLE 1.

To further explain the impact, the liquid draw from the reflux drum is partially blocked, which eventually floods the condenser and ultimately results in a blocked outlet scenario for the tower.

For impact (A), the peak (initial) relief load is bigger, but the initial temperature is much lower as the lighter liquid vaporizes from the top trays. However, for both cases the final relief temperature, when liquid on the trays has vaporized and after the condenser is flooded, is the same.

Relief load calculation methods

Various calculation methods are proposed in the following subsections.

FIG. 3. Steady-state simulation results.
FIG. 3. Steady-state simulation results.

Unbalanced heat (UBH) method. Considering an unlimited supply of liquid at the top tray, latent heat was calculated for the top tray liquid composition, and the relieving temperature was the bubble point temperature of the top-tray liquid. Since the top tray contained the minimum boiling liquid, this resulted in a conservative high relief load but a lower relieving temperature. The relief load calculated for the governing scenario of this case study in the UBH method was 23,700 kg/hr, with a relief temperature of 165°C (329°F).

The limitation of this method was the prediction of an accurate relieving temperature. The liquid in the trays below the top tray are progressively heavier, and the bottom tray liquid has the heaviest liquid in the column, as shown in the temperature profile in FIG. 2. In case of reflux failure, the top trays became dry after a certain time and the liquid began to vaporize from the feed tray, resulting in a higher relief temperature than that of the top-tray liquid. This effect was much more pronounced in small- to medium-sized towers where the top trays dried out quickly as the reflux failed, as observed for this case study.

Steady-state simulation method. The case study results demonstrate that the concern of inaccurate relieving temperature can be adequately addressed by using a steady-state simulation method, which was used to simulate loss of reflux.

For columns having only reflux cooling and no side pumparound heat removal, the failure of top reflux leads to a total cooling loss to the column, and the overhead section of the column is exposed to the feed vapor temperature. To simulate this, a flash drum was modeled, and the feed tray was considered as the top tray. In case of loss of reflux, all trays above the feed tray became dry after some time, which is the basis for the assumption here. In this example, there was also no diesel side draw, as there was no liquid in the draw-off tray (FIG. 2) and no vapor return from the diesel side stripper. However, the stripping steam supply to the diesel side stripper continued, returned to the column and was relieved through relief valve, along with the feed vapor. The flash drum was modeled at relief pressure (FIG. 3) with the normal fractionator feed and the stripping steam supply to the fractionator as another feed. The stripping effect below the feed tray was ignored. The net vapor from the flash drum was mixed with diesel side stripper stripping steam, and this total flow was to be relieved through the relief valve.

FIG. 4. Comparison of UBH and steady-state simulation method results.
FIG. 4. Comparison of UBH and steady-state simulation method results.

The calculated relief load was 21,000 kg/hr with a relief temperature of 349°C (660°F). The relief load was 12% lower than the UBH method, but the variation in relieving temperature is significant, as shown in TABLE 2.

Following the UBH method results, the column overhead system, relief valve and its downstream piping design temperature and/or piping stress calculation cut-off was 165°C (329°F) if no other design consideration prevailed, which is the case; whereas per the steady-state simulation approach, the relief and/or design temperature should be 349°C (660°F). The impact on the system design temperature profile specific to the case study is demonstrated in FIG. 4.

The logical next step: Dynamic simulation method

Dynamic simulation is a tool that has been proven to provide realistic results with respect to the time-dependent variations in relief load. As a natural progression from a simple steady-state simulation, the problem was simulated with a dynamic module.a Using a simple dynamic model to simulate the relief behavior during reflux failure, a similar trend of a high relief rate with lower relieving temperature at the initial point followed by a lower relief rate with a higher relieving temperature were demonstrated.

FIG. 5. Simple dynamic model.<sup>a</sup>
FIG. 5. Simple dynamic model.a

For simplifying the dynamic model (FIG. 5), the following approaches were taken: the gasoil fractionator and diesel side stripper were simulated as per front-end design specification, pump curves were not modeled, control loops were simplified, piping holdups were not considered and static head effects were ignored. To simulate reflux failure, a dummy valve upstream of the condenser was introduced to model a flooded condenser. A valve in the side draw from the main column was introduced to close it after reflux failure. Once the dynamic model could sustain in steady state, reflux and side draw were stopped by placing the controllers into manual mode and the setting to zero opening. Next, the condenser inlet valve was set to zero, giving rise to overpressure under the reflux failure scenario.

The dynamic model data logger (FIG. 6) showed an initial high relieving point (A) of 26,100 kg/hr when the PSV pops open, but an associated temperature of 291°C (559°F) was significantly higher than the bubble point of the steady-state top-tray boiling liquid at relieving condition. While this point was not the same as the UBH method point, the initial expectation of high relief load was established. A relieving point (B) was observed, and the results were extremely close to the steady-state simulation model—i.e., a relief rate of 21,300 kg/hr with a relieving temperature of 345°C (653°F). With an infinite supply of liquid from upstream, the dynamic relief rate becomes steady at point (C): 17,200 kg/hr with a relieving temperature of 356°C (673°F).

Dynamic simulation is a useful tool to validate steady-state simulation results and trends, but the system specification and volume definition must be accurate for the time frame to be realistic for a relieving scenario, which is often unrealistic in front-end design.

Takeaway

A conventional method like UBH provides a conservative relief load for columns, but it does not account the compositional changes on the trays toward the bottom of the tower and, therefore, affects the computation of correct relief temperature.

FIG. 6. Dynamic simulation output, reflux failure.
FIG. 6. Dynamic simulation output, reflux failure.

For columns handling heavy hydrocarbon components with a wide range of boiling components, the column operating temperature profile is quite varied from overhead vapor to bottom product, due to the compositional changes from the top to the bottom of the tower. This results in a gradual increase in relief temperature of the components descending from the top to the bottom of the tower, the bottom-tray liquid being the heaviest in the column. The case study demonstrated such an example. Here, the difference in relief temperature calculated using UBH and steady-state simulation methods was approximately 180°C.

For columns handling lighter petroleum cuts, this effect might not be very pronounced, since the difference in bubble point temperature for light and heavy cuts is not so significant.

Therefore, for columns handling heavier petroleum cuts, especially small- to medium-sized towers, the demonstrated simulation method can be recommended for correct prediction of relief and/or design temperature on a case-to-case basis with stakeholders buy-in. The temperature excursion limit of the affected piping and equipment material can also be exploited in view of the economic impact weighed against the low likelihood of pressure-relieving valve (PSV) relieving during plant design life.

An operator’s intervention time is also one of the deciding parameters. For very-large-diameter towers, time is particularly important as the time required to dry out all trays above the feed section is significant (e.g., more than 10 min–30 min, which is the globally acceptable time range for operator intervention). In that case, an intermediate tray vapor temperature can be selected as the relieving factor, as well as column overhead design temperature.

For laterals and flare header sizing in columns with heavier components (heavy gasoil, etc.), the UBH method gives a conservative relief load, provided that reboiling or stripping media is continued for the governing relieving scenario. For flare header hydraulics, the UBH provides a conservative approach; for fixing the design temperature of the flare header, the simulation-based method is better suited. Lateral and flare header design temperature must also consider the ambient temperature loss to avoid overdesign.

The observations were validated in a dynamic simulation environment establishing the variation of relieving temperature to be a real scenario. For dynamic results to replace the steady-state results for engineering design, several parameters must be accurately defined, including a real-time model with realistic system and piping volumes, realistic control loops and control valves, realistic rotary equipment behaviors at overpressure scenario and the depletion of upstream fluid supply. However, as this article has shown, the UBH method combined with the steady-state simulation method requires fewer details and less effort, and still leads to a comprehensive and sturdy heavy-ends column relief system design. HP

NOTE

  a Aspen Hysys

ACKNOWLEDGEMENT

The conclusions presented in this article are solely those of the authors and cannot be ascribed to Fluor Corp. and/or any of its subsidiaries.

The Authors

Related Articles

From the Archive

Comments

Comments

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