Jan-2019

Dynamic simulation to estimate tower relief

When applied in the correct circumstances, dynamic analysis can help avoid unnecessary scope and project delays.

HARRY Z HA and BEN LEVITON
Fluor Canada

Article Summary

Accurately predicting the relief loads for a distillation column is a challenge in engineering design. Most often, the unbalanced heat (UBH) method1 is used in the grassroots design of a project. The relief loads predicted by the UBH method are typically conservative due to the assumptions intrinsic to the method of calculation. This level of conservatism is normally acceptable for a grassroots project as it incorporates a safety margin into the relief system design. For a revamp or debottleneck project, however, the acceptable column relief load is limited by either the installed pressure safety valve (PSV) size or by the capacity of the existing flare header. The relief load calculated by the UBH method frequently exceeds these physical limitations and can lead to the installation of additional PSVs or extensive flare header modifications, which are typically costly and difficult to implement in an operating plant.

Application of a steady state relief model is often the next step to more accurately predict the relief load by better characterisation of product streams at relief conditions. Within the limitations of steady state simulation, the transitional changes of an upset case are still ignored or substituted by conservative stream conditions. As a result, the column relief loads predicted by steady state simulation also tend to be conservative. Additionally, steady state models are not easily applied to upset scenarios where the column trays run dry due to loss of liquid loading (for instance, reflux failure or loss of feed) as the column generally fails to converge.

When facing these challenges on a revamp project, dynamic simulation can be employed to provide the most reliable relief estimates. Dynamic modelling accounts for the impact of stream compositional changes, system volume contributions, and the available inventory within the system during an upset scenario. It also better characterises the relief stream by estimating depletion of light ends, temperature variations, and latent heat changes over time. The column relief loads predicted by dynamic simulation are often promising in eliminating potential modifications to the existing PSV or flare system.

This article outlines the situations in which dynamic simulation is most helpful in reducing calculated relief loads. An example relief system for a product fractionator in a hydrotreating unit is presented in order to compare the relief loads predicted by UBH, steady state, and dynamic analysis for the same system.

Hydrotreating unit product fractionator example
In order to compare the relief loads predicted by UBH, steady state, and dynamic models, this article presents a case study which employs all three calculation methods. In this study, the total power failure (TPF) relief load associated with the product fractionator column of a typical hydrotreating unit was investigated using all three methods. A typical distillate hydrotreating process flow scheme is shown in Figure 1.
The hydrotreating unit investigated is divided into reaction, stripping, and fractionation sections, as shown in Figure 1. Key features of the system configuration are as follows:
•    The reaction section includes a recycle gas compressor loop along with high and low pressure separator drums. The reactor effluent is cooled by heat exchange with the reactor feed and low pressure separator liquid.
•    The stripping section includes two columns in series. Flow from one column to the next is driven by system pressure.
•    Feed to the fractionator section is pumped from the stripping section through a set of feed/bottoms exchangers and a fired heater.
The relief scenario investigated in this example is TPF which is a global scenario with simultaneous relief loads from all units in the plant. The detailed scenario is defined by the following considerations, driven by guidelines in API Standard 521:2
1.    All pumps and compressors driven by electric motors are assumed offline.
2.    Where both electric motor and steam turbine drivers are available for a given service, the turbine driver is assumed to be in service only if it favours higher relief loads.
3.    Electric fans on air-cooled heat exchangers are assumed offline. Credit is taken for natural draft cooling duty based on a detailed HTRI model.
4.    No credit is taken for any favourable instrument response from automatic control valves during the relieving period to mitigate relief.
5.    The upset conditions in the upstream (reaction) section of the unit affect the downstream (stripping and fractionation) sections and must be accounted for.

Unbalanced heat (UBH) approach
The UBH method1 is a conventional practice for calculating column relief loads and widely used in industry. It is preferred in grassroots design where conservative PSV and flare header sizing is advantageous.

The method employs a heat and material balance around the column envelope (see Figure 2) at relieving conditions in order to estimate excess heat input. The relief rate (WA) is calculated from the excess heat divided by the latent heat of the relieving material.

The excess heat calculation considers the enthalpy of each stream at relieving pressure assuming all product stream compositions remain constant. An endless supply of relieving material is assumed available (typically represented by the top tray liquid of the column). Normally, no credit is taken for the following mitigating factors:
• Compositional changes including depletion of light components
• Accumulation of mass within the system volume as pressure increases
• Hydraulic limitations
• Overhead cooling before the reflux drum is flooded.

The UBH method is simple, effective and conservative. Aside from typical over-estimation of relief loads, it is also subject to the following key limitations:
1.    The UBH method is unreliable for scenarios where the upset leads to a significant compositional change within the column envelope. A typical example is a blow-through of vapour from an upstream high pressure section to the low pressure column via a failed open control valve.
2.    UBH can over- or underestimate relief loads for systems where the column energy balance is sensitive to minor compositional changes. In these systems, the enthalpy of bottoms stream is typically affected by the presence of light components which skews the calculated relief load. Some examples are:
• Stripping columns which remove absorbed components from a solvent (for instance, a sour water stripper)
• Columns with low overhead to bottoms flow ratio (for instance, a stabiliser).
3.    UBH is not normally suitable for complex systems such as:
• Reactive distillation columns
• Columns with relief occurring near the critical region
• Scenarios with significant transient effects such as a major upset upstream affecting feed conditions.

UBH results
The UBH method was applied to estimate the TPF relief load for a hydrotreating unit product fractionator. While the UBH method typically takes the feed at normal conditions, in this case the feed enthalpy was estimated based on simulation of the upset conditions upstream, since significant upsets do occur in the reaction and stripping sections during TPF. The calculation resulted in a required relief area of 44.8 in2. This is 172% of the available relief area of 26.0 in2.