Mar-2012

Estimating silicon accumulation in coker naphtha hydrotreaters

Improved sampling and analysis of silicon in the feed enable a significant gain in the cycle life of coker naphtha hydrotreater catalysts

Thienan Tran, Patrick Gripka and Larry Kraus
Criterion Catalysts & Technologies

Article Summary

Silicon poisoning is a major concern in coker naphtha hydrotreaters. The source of silicon in coker naphtha can be traced back to the delayed coking process, which typically uses silicon-containing oils, polydimethylsiloxane (PDMS), to suppress foaming in the coker drums. At the elevated temperature inside the coker drums, these high molecular weight, silicon-containing oils crack to form lighter silicon oil fragments, such as dimers and trimers of the dimethylsiloxane. The majority of these silicon oil fragments boil in naphtha range and therefore are routed to the downstream naphtha hydrotreaters together with the coker naphtha.

Under the operating conditions of the naphtha hydrotreaters, the silicon oil fragments present in the feed transform to modified silica gels and absorb onto the catalyst surface.¹ As silicon accumulates on the catalyst surface, it covers active sites and restricts catalyst pores; the latter process eventually blocks access to the active sites. Once silicon is bound to the catalyst surface it cannot be removed and results in an irreversible loss of catalyst activity. Without silicon in the feed, the typical cycle length of a naphtha hydrotreater is three or more years. When processing coker naphtha, the cycle length is typically 12 months. In extreme cases, the cycle length can be six months or less.

The cycle life of a coker naphtha hydrotreater is dictated by the silicon capacity of the selected catalyst system and the silicon accumulating rate. The silicon capacity of a selected catalyst system is known. However, the silicon accumulating rate often cannot be determined due to a lack of accurate feed characterisation data.

Refiners normally collect feed samples for silicon analysis on a frequent basis. However, because of the transient nature of the delayed coking process, the frequency of the feed sampling is often not sufficient to determine the actual amount of silicon being fed to the unit. In addition to the unrepresentative feed sampling issue, the commonly used inductively coupled plasma (ICP) test method does not accurately measure the silicon species present in the coker naphtha feed. These issues cause the calculated silicon accumulating rate to be unreliable and therefore the cycle life of coker naphtha hydrotreaters often cannot be predicted. This results in refiners changing out coker naphtha hydrotreater catalyst based on fixed cycle length or silicon slippage. This either under-utilises the silicon capacity of the catalyst system or results in an unplanned shutdown.

In this case study, a hot loop feed sampling station was installed on a commercial coker naphtha hydrotreater to obtain composite samples. The weekly composited samples were tested for silicon using a Shell proprietary ICP direct injection nebuliser (ICPDIN) analysis. The results were used to estimate the amount of silicon accumulated on the catalyst as the cycle progressed. At the end of the cycle, spent catalysts from the unit were analysed to determine the amount of silicon accumulated on the catalyst. Results indicated the silicon deposition estimated using hot loop sampling and ICPDIN was within 10% of the Si deposition determined from spent catalyst analysis. Due to the accuracy of the estimate, the cycle life of the unit could have been extended up to 4.5 months beyond the scheduled 12-month cycle length if the unit was not shut down due to furnace fouling.

Case study
The coker naphtha hydrotreater in this case study consists of a guard reactor followed by two main reactors, which are in parallel. The guard reactor contains OptiTrap grading materials and DN-200. The primary function of the guard reactor is to saturate diolefins. The main reactors contain 45% MaxTrap[Si], 20% DN-140 and 35% DN-3531. All of these materials are Criterion catalyst grades. MaxTrap[Si] is a silicon trap catalyst. DN-140 is a dual functional NiMo catalyst, which was used to provide both significant silicon uptake capacity and hydrotreating activity for the load. DN-3531 is a high-activity NiMo hydrotreating catalyst, which provides the majority of the HDS and HDN activity requirements for the unit. The catalyst system had a silicon capacity of 8750 lb and was scheduled for a 12-month run length. The unit process conditions and feed properties are shown in Table 1.

Feed sampling
The majority of coker naphtha hydrotreaters sample feed once per week for silicon analysis. Fluctuations in the amount of coker naphtha in the feed and the amount of foam-suppressing oils used in the coker drums will affect the silicon content in the feed. Recognising the fact that frequent feed sampling is important to capture the changes in the silicon content, a hot loop feed sampling system was installed. This sampling system was designed to collect and composite a defined number of feed samples throughout the week to obtain weekly composite samples. 

Silicon analysis of coker naphtha feed
The silicon present in the coker naphtha is in the form of dimers and trimers of the dimethylsiloxane. These dimethylsiloxane molecules are volatile. Due to their volatility, the standard ICP method, which is used by most refiners, cannot accurately measure the silicon content in the coker naphtha. Depending on the form of the silicon compound and the dynamics of the sample introduction system of the ICP method being used, the silicon concentration determined can be off by a factor of 10 to 20.²

To obtain reliable silicon content data, the weekly composite samples were sent to the Shell Global Solutions Westhollow Technology Center for silicon analysis. A Shell proprietary ICP analysis was used to measure the silicon present in the weekly composite sample. This analysis uses a custom design DIN to introduce the sample into the ICP. Based on results of extensive research, the error margin of the ICPDIN test is typically less than 10%.

Spent catalyst analysis
After 10 months online, the unit was shut down due to furnace fouling issues. The decision was made to change out the catalyst while the unit was down. The spent catalyst was unloaded by vacuuming, which allowed an accurate silicon deposition profile to be constructed. Samples of the spent catalyst were collected to determine the silicon uptake capacity of each type of catalyst and the total amount of silicon accumulated on the catalyst system.