logo


Jan-2013

Tracking performance degradation in a debutaniser column

A project applying statistical and data mining techniques to historical operating data identified the root causes of a debutaniser’s poor performance

ALI ABDULAL, VINOD RAMASESHAN and STANLEY GUSTAS III, Saudi Aramco
HORIA ORENSTEIN, MOHAMMAD KURDI and ANDREY GASKOV, SAS Middle East

Viewed : 5514


Article Summary

Complex, capital-intensive industries such as refining and petrochemicals face a pressing demand to minimise unplanned downtime and achieve higher asset uptime and improved Health Safety & Environment (HSE) targets. To improve the reliability and availability of facilities and assets, Saudi Aramco has been investigating the application of SAS Predictive Analytics and Data Mining in operations and maintenance. In this regard, a pilot project was conducted on the ability to deliver leading indicators to the problem of performance degradation in the Ras Tanura refinery’s hydrocracking unit debutaniser column. This article deals with this pilot project; understanding of the use of a business analytical tool, resulting in successfully achieving the goals that deliver prediction of performance degradation; the flagging of leading indicators; and the development of a dashboard for operating staff, with recommendations for revision to the multivariable control system (MVC) system to alleviate the problem.

Hydrocracking unit
Saudi Aramco’s Ras Tanura refinery operates a 44 000 b/d Unicracking process unit (hydrocracking unit) designed by UOP and commissioned in 1999. This unit is a two-stage maximum conversion unit (>97%), where the primary feed vacuum gas oil (SC-7 and SC-8) is treated and cracked over three reactors (C-100 and C-101 in the first stage, and C-200 in the second stage) to naphtha and distillate products.

Since the unit was originally conceived to operate in both maximum distillate and naphtha mode, the fractionation section is what in hydrocracking parlance is termed a “debutaniser first” fraction scheme. The reactor effluent first comes to the debutaniser, where lighter ends are removed before being routed to the main fractionation section. Consequently, the debutaniser acts both as a conventional stabiliser and as a hydrogen sulphide (H2S) rejection column. To act as a H2S rejection system, it is imperative that the vapour-to-liquid ratio at the bottom of this column is high (>0.5 on mole basis). If it is lower, H2S rejection is poor, leading to corrosion in the downstream fractionation column and pressure drop in the second-stage reactor (C-200) on account of particulate carry-over with the recycled oil, over and above possible leaks in the piping on account of high temperature and wet H2S corrosion.

Subsequently, since the feed to the debutaniser essentially comes from the reaction section, the vapour-to-liquid ratio is strongly influenced by the product slate. Ras Tanura’s hydrocracking unit, since start-up, has predominantly operated in distillate mode (>95% of the time) and with catalyst systems changed to low-zeolite catalyst in the first stage and amorphous catalysts in the second (as opposed to high-zeolite catalyst in the past). The product slate has predominantly been heavier (a flatter true boiling point, TBP, curve), leading to a higher than design requirement in the reboiler duty (based on a “preferential once-through” reboiler [fired heater] design), with no improvement in the vaporisation rate, leading to H2S slippage and associated corrosion and reliability problems. To mitigate this issue, a multivariable controller (MVC)-based control system for the second-stage conversion levels was developed by changing the C-200 inlet temperature and therefore the weighted average bed temperature (WABT). This was done because the distillate selectivity per pass conversion on the second stage is lower when compared to the first stage for the same catalyst (the effect of ammonia comes into the picture). Consequently, the light end make would increase and thereby maintain the vapour-to-liquid ratio in the debutaniser. However, this procedure leads to a debit in distillate make (the product of choice for the refinery).

A pilot project with SAS Institute, using predictive analytics and data mining techniques, was applied to this problem, to ascertain if other leading indicators could be tracked to alleviate the situation and maintain the distillate make from the unit. While the refinery will conduct other engineering modifications to the unit in the future, data mining was considered to be a good concept to deal with the current situation, alleviate the problem, if possible, and therefore optimise any additional engineering solution. Figure 1 shows a high-level overview of the hydrocracking unit in question.

Technique and analytic procedure
The objectives of the project, called the Predictive Performance Degradation (PPD) project, can be summarised as follows:
•   Perform data mining techniques to understand inter-dependencies among different debutaniser operational parameters that include:
    ν  Hot and cold feed temperatures, flows and compositions
    ν  Internal reflux
    ν  Tray 23 temperature: the seventh tray from the top in the rectification section
    ν  Reboiler duty and bottom temperatures
•   Identify the cause and effect relationships among these operating parameters and the following debutaniser performance-related issues:
    ν  Debutaniser instability
    ν  Light hydrocarbon slippage and Reid vapour pressure deviations
    ν  H2S slippage and corrosion problems at the bottom of the debutaniser and the fractionator column
•   Proactively identify the critical event and root causes that contribute the most to the above performance issues
•   Generate proactive dashboards, reports and other user interface screens that provide information about events, symptoms, causes and effects of the performance degradation of the debutaniser, thereby helping engineers and operators to react in time and 
minimise the debutaniser’s instability.

The performed predictive analysis is based on historical process information data. It applies statistical and data mining techniques to this data to find the factors and root causes that affect the debutaniser’s performance. Analytical capability is the highlighted outcome of the project. It has been determined that there is a huge potential in benefits from the exploitation of historical data. It has also been determined that exploitation of such data in a similar way has not been fully applied in the past. Therefore, this project was designed to utilise the power of analytics to unleash the potential of such historical data and to make operations more stable and profitable.


Add your rating:

Current Rating: 4


Your rate: