Dolphin Energy: Sorting sulphur shutdowns

As part of our series of energy insights from the region’s national oil companies, Dolphin Energy shares its insights from analysing the failure of its sulphur recovery units at its massive gas processing and compression facility in Ras Laffan, Qatar, last year.

Background
Dolphin Energy’s Ras Laffan gas processing plant comprises two process streams, both of which include sulphur recovery units (SRUs) which aim for a recovery rate of 99%. The combined plant is designed to produce 1,170 tons/day of sulphur as a by-product.

Each SRU is a conventional two-stage Claus catalyst bed design with a Scot Tail Gas Treatment Unit.

During May and June 2011, there were two unplanned shutdowns of the SRUs, which affected the overall plant throughput at that time. Before May 2011, these units were operating normally for over 3 years without any significant operating issues.

The first unplanned shutdown of SRU-2 lasted from 11 May to 4 June 2011. This occurred due to tube failures on the first stage condenser heat exchanger with subsequent damage during the cool down procedure to the associated first Claus Reactor re-heater heat exchanger.

Erosion was found on the bottom three tube rows of the first and second sulphur condensers. These tubes which account for approximately 3% of the total tubes were plugged.

During the shutdown, all acid gas was processed by SRU-1, requiring a reduction in gas throughput to 1,850 MMSCFD to avoid flaring acid gas.

The second unplanned shutdown was of SRU-1 unit from 15 June until 28 June 2011. This shutdown was again due to a condenser tube failure but this time it was on the second sulphur condenser.

Inspection revealed erosion signs similar to those seen at SRU-2. The bottom four rows of these exchangers, representing approximately 5% of the total tubes, were plugged.

ANALYSIS
A failure analysis of the SR-1 and SR-2 condenser tube leaks was performed by both the licensor and an independent third party.

The analyses agreed that the root cause of the tube erosion in SRU-1 and SRU-2 condensers was due to improper flow distribution at the entry nozzles to the exchangers.

Both the parties reviewed the information related to the design, operation and failures of SRU-1 and SRU-2. Damage in the form of metal loss, including holing through of tubes, as well as some metal loss to the inlet tube sheet at the top of the tube bundle were observable after about three years of operation.

The root cause of the failure was found to be improper design of the system returning HP steam condensate from the Claus Reactor Re-heaters to the subject condensers.

The HP steam condensate outlet from the re-heater connects to a steam trap and then to the condenser inlet. This hot condensate flashes, both in the pipe, and at the condenser inlet nozzle. The flow of large amounts of steam into the underside of the bundle causes high local velocities and turbulence. The damage to the tubes and tube sheet was caused by ‘Flow Induced Corrosion’ (FIC).

In boiler water systems, corrosion prevention depends on the formation of a protective magnetite scale on the steel surface of tubes. De-aerated water with a standard water treatment programme will promote the formation of the protective magnetite scale.

However, under very turbulent two-phase flow conditions, like the one that exists near the HP steam condensate nozzle inlet to the condenser on the shell side, the scale can be compromised and accelerated corrosion can occur.

Article continues on next page …

Findings
Metal loss on the three tubes that leaked was most likely the result of FIC. The three tubes that leaked were in the immediate vicinity of the A3 nozzle which carried mostly saturated steam vapor at a significant velocity into the condenser.

The observed metal loss on the inlet tube sheet and tube outer diameter (OD) surfaces near the tube sheet was most likely caused by a ‘liquid bubble flow’ pattern in the condenser shell as a result of introducing steam vapor through nozzle A3.

The flow pattern results in a lower heat transfer, possibly leading to higher metal temperatures. It also causes an alternating “wet and dry” condition on the metal surface, especially in the hottest areas of the condenser, which compromises the protection provided by the magnetite scale and results in a potential for corrosion.

All of the observed metal loss was caused by conditions in the condenser shell created by the steam vapor and hot condensate introduced in the bottom of the shell through nozzle A3.

The potential for high temperature sulfidation was also reviewed from the process side of this condenser because the process stream from the combustor to the final condenser contains large amounts of sulphur and sulphur compounds.

The elemental sulphur, and in particular, the H2S in the vapor or liquid phase of this stream, will corrode carbon steel at elevated temperatures by way of sulfidation.

The predicted sulfidation rate of carbon steel exposed to this stream increases very rapidly from less than 0.50 mm/yr at 205C to 1mm/yr just below 345C. The first condenser, which is the hottest, operates with an inlet temperature of 322C so even if the tubes were partially vapor blanketed, it is not anticipated that the temperatures on the process side will be high enough to result in high internal tube sulfidation corrosion.

There is also concern at potential corrosion associated with condensate, as the line from the steam trap to the condenser contains a two phase mixture of steam and condensate. Within the line at nozzle A3, the stream contains 99% steam vapor by volume with a calculated mass flow rate of between 14 and 26 m/s, depending upon operating conditions.

It is difficult to evaluate all of the literature to develop a clear corrosion rate prediction. Flow-induced corrosion may be possible in areas where the two-phase flow is turbulent. This includes piping just downstream of the steam trap where the condensate flashes and fittings such as elbows, tees and reducers.

As a result, it would appear that the best mitigation for the observed corrosion is to eliminate steam vapor from the bottom of the condenser shell.

As long as the water continues to be treated as indicated by the analysis results and only water enters the condenser shell, corrosion should not occur on the shell side surfaces.

Action
The key to fixing the problem of deviating conditions is correcting the design of the condensate return from the HP steam traps to the condensers.

Option 1 would involve the addition of a conventional disengaging drum – a single drum can be provided to receive the condensate from the two steam traps, at the elevation of the condensers. The disengaging drum would send flashed steam to the low pressure steam system, upstream of the control valve holding pressure on the condensers. This requires additional equipment and execution difficulty, so was not pursued.

Option 2 was to relocate nozzles A3 into the vapor space of the condensers. To avoid excessive velocities entering the bundle (to comply with TEMA limits), the nozzle diameter would need to be increased to NPS 8 if the steam/condensate rate is 2,304 kg/hr. With an NPS 8 nozzle and that mass flow rate, an impingement plate is not required.

This option was not judged to be feasible, as it requires extensive modification to the condenser shell and complexity in executing the modification.

Option 3 was to divert or reroute the HP condensate to an alternate location instead routing it to the condensers, and supplement the water requirement for the condensers. It was found that there is an existing steam condensate flash drum in the plant that can receive the condensate and send it to the condensate recovery system.

This was the quickest feasible solution as the drum was checked and found to be adequately sized for the additional condensate load too.

Option 3 was implemented during the recent planned shutdowns in both SRUs, both of which are now performing soundly.