May-2018

Countering hydrate formation in an NGL recovery unit

Process modelling can be used to validate data from the plant and identify the most effective approach to process optimisation.

JENN WELSH, Chevron North America Exploration and Production Company
JUSTIN C SLAGLE, Bryan Research & Engineering
STEVE CHEANEY, Dickson Process Systems
GLENN SHIVELER, Sulzer Chemtech USA

Article Summary

In the early 1990s, Chevron installed a new process to recover natural gas liquids (NGL) from recycled CO2 in the Rangely Weber Sand unit, Colorado. The facility was designed based on a patent issued in June 1988 (US Patent 4753666). The process, as claimed in the patent, used a refluxed distillation tower to produce an overhead stream virtually free of n-butane and heavier components, and a bottoms stream containing the majority of the heavier components. Based on the success of the Rangely plant’s operation, a similar facility was designed and fabricated for Chevron’s Mabee field near Midland, Texas.

When the Mabee plant faced some unique challenges in the NGL recovery unit, Chevron contacted Dickson Process Systems, Bryan Research & Engineering (BR&E), and Sulzer Chemtech USA to help troubleshoot the process to operate as designed. The Mabee NGL recovery unit is a distillation column with a partial condenser and a reboiler. The condenser, operating at 4-15ºF (-15 to -9ºC), experienced what were described as ‘reflux events’, where the liquid level would rapidly build up in the reflux accumulator.

While troubleshooting, it was found that hydrates formed in the top section of the column, restricting the liquid from continuing down to the next tray. The resulting build-up of liquid not only prevented reflux from entering the column, but also flooded the trays and quickly overwhelmed the top section of the column. This article describes the process used to identify and address the formation of hydrates in the top section of the distillation column and the use of process simulation modelling combined with plant data to improve plant operability.

NGL recovery unit design
The NGL recovery unit designed and fabricated by Dickson Process Systems was added onto the existing compressor station in Chevron’s Mabee field in 2013. This unit was installed between second and third stage compression following the existing triethylene glycol (TEG) dehydrator. The NGL recovery unit was designed based on a feed gas water content of 3 lb/MMscf in accordance with Chevron’s US Patent.1 TEG dehydration units routinely remove water to levels below 7 lb/MMscf in the dry gas. If stripping gas is utilised, the water content of 3 lb/MMscf can be achieved.2 In this system, NGLs are extracted using a reflux column to produce an overhead stream free of n-butane and heavier components and a bottoms stream containing the heavier components.

The distillation column is a 26-tray tower equipped with a partial condenser and reboiler (see Figure 1). A pressure differential transmitter spans from tray 1 through Tray 26. Temperature transmitters are equipped at the column overhead, column inlet located at Tray 15, Tray 23, and column bottoms. Propane is used as the refrigerant in the partial condenser. The reflux accumulator level is maintained through the use of continuously operating pumps along with a level control valve feeding the reflux back into the column and a flow control valve recycling reflux back to the accumulator.

After deployment, the distillation tower experienced what plant operations described as ‘reflux events’. During these events, the reflux level in the reflux accumulator would suddenly rise at a rapid rate. The level would remain high and overwhelm the tower reflux system. Corrective action involved warming up the chiller outlet temperature to stop liquid generation until the reflux accumulator level fell within the normal operating range. This pattern is shown in Figure 2. These corrective measures disrupted the temperature profile across the column, reduced NGL production, increased H2S content in the NGL product, and required continuous involvement from plant operators.

While this solution maintained operability in the short term, a long-term solution was desired. Therefore, a team consisting of engineers from Chevron, Dickson Process Systems, BR&E, and Sulzer Chemtech USA was assembled to quality check and troubleshoot the process to ensure the facility operated as designed. With Sulzer Chemtech’s assistance, column hydraulics were ruled out as the source of the reflux events.

Identifying the cause of rapid liquid build-up in the reflux accumulator
The first step was to develop a process simulation of the distillation column to model the process conditions and study the possible causes of the reflux events. The NGL extraction model was developed in ProMax3 based on data collected from the field. The composition results from the simulator agreed well with the data (see Table 1).

To thoroughly diagnose the abnormal conditions, the team reviewed operations upstream of the NGL recovery unit. Upstream of the distillation tower, glycol dehydration removes water from the process gas in preparation for NGL extraction (see Figure 3). If water exceeds certain thresholds in the NGL recovery column, it can become trapped in the tower trays, leading to blockages and, if conditions allow, hydrate formation.

Initially, it was thought that the reflux events were not a result of hydrate formation or trapped water as field instrumentation indicated very low water content in the feed gas. When modelled with an inlet feed water content below 3 lb/MMscf, the process simulation model predicted no hydrate formation in the NGL recovery column. Additionally, the pressure differential transmitter across the tower spanned 0-1 psi across the tower trays, offering little insight into tray loading both leading up to and during the events. Figure 4 illustrates data collected in the facility where reflux events were observed despite data indicating moisture content to the tower well below the design requirement of 3 lb/MMscf.

In the interests of creating a model of the entire plant to evaluate upstream effects on this process, a model of the dehydration unit was added with current operating data and integrated with the NGL recovery model. Once the team modelled the upstream TEG dehydration unit, the predicted water content at the inlet of the NGL recovery unit was noticeably higher than instrumentation indicated.

As designed by Dickson Process Systems, the NGL recovery unit required a dried feed gas with a water content of no more than 3 lb/MMscf.

When modelled at the operating conditions, a gas water content of 9 lb/MMscf was predicted. The plant engineers manually measured the moisture content in the inlet feed to the NGL recovery column using Draeger tubes and a hand-held automatic dew point hygrometer, both of which confirmed the higher water content predicted by the model. The in-line water content analyser was found to be miscalibrated, which explained why it was difficult to identify the cause of the reflux events. In the interests of collecting the most accurate data moving forward, the in-line moisture analyser was recalibrated and currently delivers reasonable values.