Jan-2009

Control of primary fractionator fouling

Optimal operation and chemical treatment of primary fractionators requires an in-depth knowledge of many parameters. A combination of analytical tools and benchmarking information can help predict fouling severity

Bob Presenti, Dan Frye and Sandra Linares-Samaniego
Nalco Company

Article Summary

The ethylene industry remains extremely competitive. Fouling in primary fractionators costs ethylene producers millions of dollars annually in lost production, product downgrades, reduced energy efficiency and unplanned shutdowns. The primary fractionator in a liquid cracker represents one of the largest potential material and energy cost debits if fouling is not controlled. While the fouling mechanisms in primary fractionators have been studied extensively over the past few decades, they are by no means fully understood.

The overlying problem for most ethylene producers is the ability to effectively monitor the impact of fouling on primary fractionator performance. Often, proper instrumentation is either unreliable or non-existent and plant operators are challenged to answer the questions “Is my tower fouling?” and, if so, “What do I do to stop it?”. As part of Nalco’s Best Practices primary fractionator treatment programme (PrimAct), benchmarking, analytical and monitoring tools have been developed to give ethylene producers an improved tool to evaluate fouling tendencies. This in turn helps us understand and optimise the economic impact of fouling and operational optimisations.

Fouling in a primary fractionator
Primary fractionator fouling can be broken down into two distinct problem areas: top section fouling and bottom section fouling. The fouling mechanisms in the top and bottom sections of a primary fractionator are quite different. Analyses of top sections of fractionators across the industry indicate that the predominant fouling mechanism is the deposition of cross-linked polystyrene-type polymers formed through free radical polymerisation, whereas fouling in the bottom section of the primary fractionator is caused by the agglomeration and deposition of polynuclear aromatic hydrocarbons (PNAs).

Top section fouling is usually identified by increased differential pressure (DP) and increased overhead endpoint. While increased DP can limit the throughput and decrease the unit run length, an elevated pygas endpoint can lead to emulsions in the quench water loop and product downgrades downstream.

Fouling in the bottoms section of a primary fractionator will cause changes in the viscosity control of the circulating quench oil, loss of heat recovery, a decrease in dilution steam production, an increase in pump energy usage and poor filter operation, as well as many other issues.

Stream characterisation and benchmarking
The primary fractionator is usually the first tower after the transfer line exchanger (TLE) or furnace. The primary functions of this unit are heat recovery and crude fractionation of the furnace effluent. Basic plant configurations vary widely from plant to plant, but most fractionators have a top stream, bottom stream and a light fuel oil stream. Due to the variability in design and operating conditions, a comprehensive approach to primary fractionator treatment and monitoring that includes the characterisation and benchmarking of the process streams was undertaken. Figure 1 shows the sampling points and summarises some of the tests undertaken on each of these streams.

A database of analytical results from ethylene plants throughout the world is being compiled and continues to grow, as over 400 samples have been analysed to date. This database allows for continuous expansion of knowledge and serves as a benchmarking tool for other units. Currently, the database is comprised of towers that have experienced fouling problems as we work with each ethylene producer to improve reliability. A portion of the benchmarking data for reflux streams is included on the following pages in Figures 2 to 5.

Reviewing the benchmarking data in Figures 2 to 5 reveals that stream characterisation is plant specific and that further investigation is required before the fouling severity in a tower can be predicted based on analytical data. Still, several observations exist based on Nalco’s initial survey of reflux streams:
— Styrene content alone does not predict fouling severity, as polystyrene is very soluble in styrene
—  Styrene and indene levels are independent of each other (Figure 2). Indene will cross-link with polystyrene to form a polymer that is less soluble in the liquid stream
—  DCPD is present when the reflux contains recycle streams (Plants 1, 2, 3 and 7 in Figure 2). DCPD reacts via a Diels-Alder mechanism to both form polymer and cross-link with other oligomers and polymers like polystyrene
— Iron does not appear to exist in significant quantities (Figure 3).

Specific gravity of the reflux stream (Figure 4) is indirectly controlled as each ethylene operator attempts to maximise pygas flow while maintaining stream endpoint within a targeted range. As specific gravity approaches 1.0, the potential for emulsion and separation problems in the quench tower and downstream equipment increases.

Figure 5 summarises the soluble polymer levels in the reflux using a Nalco proprietary lab method. This data is interesting not only because it demonstrates significant variation in soluble polymer from plant-to-plant, but also because the levels in some of the reflux streams are much higher than previously anticipated. This suggests that in addition to the polymerisation occurring inside the tower, polymer is being formed at varying levels outside of the primary fractionator and transported back to the tower via recycle streams or the quench tower gasoline reflux.

Field monitoring techniques development
Many ethylene producers have become frustrated with conventional monitoring techniques like delta pressure (DP) as a measure of fouling in the upper section of a primary fractionator. Two goals to address this problem have been identified. The first involves the application of statistical modelling to improve the ability to detect fouling in DP trends. The second is to develop other monitoring techniques to support the DP analysis. Figure 6 outlines Nalco’s initial statistical modelling of one plant’s primary fractionator tower DP.

Using statistical modelling software, three separate models were created and compared. The models helped smooth the DP trend and remove hydraulic impacts. This analysis offers an improvement over traditional flow-corrected multivariate regressions of tower pressure drop by addressing limitations due to non-linearity and autocorrelation of “independent” variables typical of actual plant operations.