PTQ Q3 2025 Issue

The decision to employ dynamic modelling is normally made on a case-by-case basis where the additional time and effort are warranted by potential reductions in scope. This is common in revamp and debottleneck projects where a change in configuration or increase in capacity has the potential to increase relief loads. Dynamic simula- tion provides the opportunity to avoid costly relief system modifications.³ Case study 1: depropaniser loss of condenser The initial relief load analysis for a depropaniser column was performed using the UBH approach. Refer to Figure 1 for the system sketch. The governing relief scenario was found to be a total loss of cooling duty in the air-cooled overhead condenser due to the airside louvres malfunc- tioning closed. The basis for the UBH calculation can be summarised as follows:  Overhead condenser duty is set to zero. v Reboiler duty during relief is assumed equal to normal duty. As a measure of conservatism, no credit is taken for reduced LMTD as the column bottoms temperature rises. Per API Standard 5212 guidelines, no credit is taken for any favourable instrument response that would reduce the relief load. In this system, the depropaniser sump level con- troller sends a remote setpoint to the LP steam condensate flow controller, adjusting reboiler duty to keep the sump level constant. For the UBH calculation, it was assumed that the level in the sump would tend to fall due to loss of reflux, and no credit was taken for the level controller’s response, which would reduce the flow of steam to the reboiler. The relief load by UBH was calculated to be 497,000 kg/ hr for a PDH plant producing 500,000 t/y propylene. In a later stage of the project, a dynamic simulation was performed to further refine this relief load estimate. The dynamic study was driven by project constraints in the flare system downstream of the relief valve, which would have benefited from a reduced relief load. The basis for the dynamic simulation can be summarised as follows: u Overhead condenser duty is set to zero, initiating the relief scenario. Reflux to the column is lost once the reflux drum runs dry. v Reboiler is modelled as a UA exchanger, taking credit for reduced LMTD as column bottoms temperature rises. The flow rate of steam condensate exiting the reboiler is adjusted by the column sump liquid level controller. The dynamic relief load was found to be 562,650 kg/hr, about 13% higher than the UBH relief load. This increase can be attributed to the complex instrument response of the depropaniser sump level controller. During the early stages of the upset, a rapid condensation of C₃ components occurs due to the rise in pressure, causing the liquid level in the sump to surge. This causes the level controller output to raise the setpoint of the steam condensate flow controller. While there is an initial dip in reboiler duty due to reduced LMTD, this is quickly eclipsed by the action of the steam condensate flow controller, leading to a peak reboiler duty that is 160% of normal duty. Results of the dynamic simulation of this scenario are shown in Figure 2 . Note that since it was identified that the control

Relief

PDIC

PIC

TIC

Overhead condenser

FIC

Overhead drum

Depropaniser

Overhead pumps

Feed

LIC

FIC

LP steam

Condensate

Reboiler

C+

Bottoms pumps

Figure 1 Depropaniser system

that would not be included in a steady-state model. A dynamic model must include far more details than a steady- state simulation, including the following inputs: • Equipment geometry and elevation. • Piping volume based on isometric drawings. • System pressure profile calibrated with hydraulic calculations. • Performance curves for pumps and compressors. • Valve characteristics. • Plant operation control parameters for key control loops. • Column sump and drum liquid levels. • Exchanger UA values based on a detailed HTRI model. Dynamic simulations present the most accurate relief results and frequently estimate lower relief loads than the UBH or steady-state methods. This is mainly attributed to the rigorous estimation of vapour and liquid composition within the column over time, particularly when dealing with material with a wide boiling range. As the lighter compo- nents tend to vaporise earlier while pressure is building in the column, the latent heat of the remaining liquid is ele- vated as it becomes heavier. The following time-dependent variables also impact the relief profile during a relief event: • Changes to stream temperatures and compositions. • Accumulation and depletion of material within system volumes. • Propagation of upset conditions from an upstream sys- tem to a downstream system. • Thermal limitations due to available heat transfer area (UA) and changes to LMTD. • Hydraulic limitations, including pump curves, frictional pressure drop, and static head. • Instrumentation control response.

PTQ Q3 2025