- K. Harris, -R. Hauer, -D. Potter, -B. Lewis,

PRESENTED AT: ISA 52nd Analysis Division Symposium Houston, TX

Keywords

SCOT, TCD, NDUV, Heated Probe, H₂S Measurement, H₂ Measurement, COS Measurement, Tail Gas Clean Up, H₂ and H₂S Analyzer

Abstract

The paper describes the initial installation and subsequent improvements of a gas analyzer installed on a SCOT (Shell Claus Off-gas Treating) tail gas treating unit (TGTU) for Claus SRU (Sulfur Recovery Unit) tail gas. The process requires the measurement of H₂ (Hydrogen) to control the reduction gas to a CoMo (Cobalt Molybdenum) catalytic reactor and the measurement of H₂S + COS (Hydrogen Sulfide and Carbonyl Sulfide) to monitor the efficiency of the amine contactor and CoMo reduction reactor.

The analyzer combines a non-dispersive ultraviolet (NDUV) optical bench with a thermal conductivity detector (TCD) for the simultaneous measurement of H₂S, COS and H₂ from the sample point at the overhead of the amine absorber. This measurement has previously been performed by a gas chromatograph or using discrete cold/dry analyzers. The innovation comes from combining the detection techniques, electronics and sample system into a single integrated continuous analyzer for a specific application. A significant reduction in cost and maintenance is realized.

The sample system was given special attention in this application. The toxicity of H₂S is well known and therefore the sample system was designed so the sample could be returned to the process at the same point as the sample take off. A heated sample probe was designed which has sample wetted parts fabricated from stainless steel and an aluminum heater for temperature control. The probe can be isolated from the process and backflushed (along with the analyzer) so safety integrity is maintained during service intervention. An integral membrane filter removes entrained liquids, and an aspirator is used to return the sample to the process. The probe, sample line, and analytical components of the sample system are heated above the water dew point temperature to avoid water removal from the sample.

Introduction

DESCRIPTION OF AMINE BASED TAIL GAS TREATING (TGT) PROCESS

The Tail Gas Treater (TGT) is the basic unit operation of removing sulfur compounds from Claus SRU tail gas. There are several types of TGTs, including production of byproduct sulfates, but by far “closed-loop” amine-based TGTs where the H2S rich stream is recycled back to the front of the SRU are the most prevalent. This paper is based on the experience for a SCOT tail gas treater but the analyzer and sample system has been applied to many variants of the amine-based TGT.

The process consists of three basic steps

• Catalytic hydrogenation and hydrolysis of all sulfur components (H2S as well as SO2 (Sulfur Dioxide), COS, CS2 (Carbon Disulfide), sulfur vapor and sulfur liquid) in the Claus tail gas to H2S. The catalyst is Cobalt Molybdenum and this stage is sometimes referred to as the “CoMo”, reduction or SCOT reactor. Reduction of all the unreduced sulfur compounds to H2S requires an excess of H2 at the outlet of the CoMo SCOT reactor, hence the need for a H2 analyzer. (1)
• H2S is selectively absorbed from the tail gas by means of an amine solvent, after cooling. This stage is sometimes referred to as the (SCOT) absorber or contactor.
• Amine is regenerated (H2S is desorbed from the solvent). The high concentration H2S stream from the regenerator is recycled back to the input of the Claus SRU. The off-gas from the absorber is sent to the incinerator (operating permits are usually 250 ppm SO2 with typical operation at 100 ppm SO2 or below).

FIGURE 1 – SIMPLIFIED SCHEMATIC OF THE SCOT PROCESS WHERE AND HOW TO MEASURE H2S + H2 IN TAIL GAS TREATERS

There are two common sampling locations for the H₂S/Hydrogen analysis. One sample location is immediately after the CoMo reactor and Quench tower (before the absorber) so the H₂S measurement represents all of the sulfur compounds in the SRU tail gas. This serves as a fairly precise material balance (recovery efficiency measurement) which can be used as an optimization tool by the process engineer.

The other location is after the absorber (before the incinerator) to monitor the operation of the amine treatment section. Also by comparing the H₂S measurement here to the SO₂ CEM’s (Continuous Emission Monitor) value after the incinerator, the difference can be attributed to trace sulfur compounds including COS and CS₂ which is useful information for the process engineer and operations (or the COS and CS₂ can be measured by the UV analyzer). There are examples of TGTUs having H₂S measurements in both process locations with the H₂ measurement installed with the upstream analyzer.

The sample system considerations are not trivial. The strong preference from a simplicity standpoint is to locate the analyzer in either one of the positions described above because it means the sample system can consist of a custom heated probe, heated sample line and the heated cell. In applications requiring sampling before the quench tower (after the CoMo reactor) then the sample has to be conditioned with the addition of a sample chiller to eliminate potential water dew point issues. Note that the chiller is not required if the sample point is after the absorber.

There are two common reasons for choosing the sample point before the quench tower. First, with existing tail gas treaters the sample point may be located before the quench tower and changing the sample point is more costly than the addition of a sample chiller Second, in the case of Alon USA (who operate a 70,000 bpd refinery located in Big Spring, Texas, approximately 45 miles east of Midland, Texas), during the basic engineering phase the contractor expressed a desire to have the safety of being able to temporarily measure the H₂ before the quench tower / absorber during startup while the quench tower/absorber are being bypassed. It was desired to do this in addition to making the normal measurement of H₂S after the absorber when the SCOT unit is in full operation. This option prevents damage to the amine during the short interval of time the absorber is being bypassed. A simple manual switching system was designed, that can select either sample location, back purges the unused sample leg with N2, and uses a sample chiller to remove the water.

Analyzer Design

Based on previous experience of supplying UV analyzers to plants with SCOT units in their SRU, it seemed that there would be several advantages to a single analyzer that was capable of measuring H₂, H₂S, and possibly COS and SO₂. The primary advantage of combining these measurements is a cost savings in both the cost of the analyzer and also installation costs. Manufacturing costs are lower with elimination of duplicate items like micro-controllers, sample system, enclosures, power supplies etc. Further cost savings can be realized during the installation because of reduced number of sample lines, probes, electrical connections and also space savings.

The photometric portion of the analyzer was designed around a well-proven optical bench which has been used for the measurement of these species in the past. The optical bench has been specifically designed to provide excellent baseline stability, exceptional linearity and high sensitivity. These performance specifications are desirable in applications where a single component is being measured at low concentrations with no interfering compounds present and especially when measuring multiple components with overlapping absorbance ranges. Further details on the design of this spectrometer and discussions on the advantages in comparison to other designs are available in References 2 and 3.

FIGURE 2 – DUAL BEAM MULTI-WAVELENGTH SPECTROMETER

A cell length of 81 cm was used for the first installation measuring H₂S in the amine contactor overhead of the SCOT unit, where a full-scale measuring range of 0-500 ppm was required. Since that original installation, the most common full-scale measuring range for H₂S in the contactor overhead applications is 0-1000 ppm, so a 40 cm cell length is often used.

THERMAL CONDUCTIVITY SENSOR
The thermal conductivity sensor uses a Wheatstone bridge type circuit with matched heated-wire elements on both sides of the bridge. One of these heated elements is used as a reference and is sealed inside a compartment containing air while the other element is exposed to the sample gas. A constant voltage is supplied across the bridge and this controls the maximum temperature of the elements. The measuring element loses more heat than the reference element if the sample gas has a higher thermal conductivity than the reference air. Increased heat loss reduces the temperature and lowers the resistance of the measuring element. The reduction of measuring element resistance unbalances the bridge circuit, resulting in an increase in the bridge output voltage which is read as increasing thermal conductivity of the sample gas. Both heated elements are located in a compartment behind a sintered stainless steel flame arrestor, with the sample gas flowing past the sintered disk and diffusing into the area around the elements as shown in Figure 3.

FIGURE 3 – LAYOUT OF THERMAL CONDUCTIVITY SENSORECTROMETER

Thermal conductivity sensors can be used to measure the composition of binary gas mixtures, if the two components in the mixture have sufficiently different thermal conductivities. However, sample gas in the amine contactor overhead in a SCOT unit is not a binary gas mixture. A typical stream contains approximately 10% water vapor (H₂O), 10% carbon dioxide (CO₂), and 3% hydrogen, with most of the balance being nitrogen (N₂). There are small quantities of hydrogen sulfide, various hydrocarbons, carbonyl sulfide, and carbon monoxide, but these do not have a significant effect on the thermal conductivity of the mixture because their concentrations are low relative to the major constituents. It is still possible to make the hydrogen measurement by considering the sample gas to be a binary mixture, if the concentrations of water vapor and carbon dioxide are stable, or if the impact of their variation can be minimized. Water vapor concentration is a function of the temperature and pressure in the amine contactor since the sample gas is essentially saturated with water in the contactor. Good control of the amine temperature and stream pressure will result in a relatively stable water vapor concentration. Carbon dioxide concentration is affected by changes in the composition of SRU plant feed and swings in SRU air demand. It can also vary with the duty cycle of the reducing gas generator (RGG) that is typically used to add hydrogen to the stream before the SCOT catalyst bed. Carbon dioxide concentration varies more than the water vapor, so it is desirable to operate the TC sensor under conditions that minimize the effect of changes in CO₂.

Experiments were carried out to determine the effects of sample pressure and sensor temperature on the thermal conductivities of gas mixtures containing nitrogen, hydrogen, carbon dioxide and water vapor. It was found that the effect of CO₂ variation on the mixture thermal conductivity could be minimized at TC sensor temperatures between 400°C and 600°C where the thermal conductivity of CO₂ is closer to that of nitrogen and air than it is at lower temperatures. Higher temperatures increase the effect of water vapor concentration changes, which is not an issue in applications with stable water vapor content. In applications with unstable water vapor concentration or if the analyzer is being switched between streams, it is possible to dry the sample gas to eliminate this issue. The output of the TC sensor is shown in Figure 4 at hydrogen concentrations of 0 to 10%. The output is very linear with only a very small non-linearity correction being required.

FIGURE 4 – OUTPUT OF TC SENSOR VS HYDROGEN CONCENTRATION

After reviewing the data from these experiments, it was determined the hydrogen concentration of mixtures representative of SCOT contactor overhead sample gas could be calculated using the following equation:

These tests were also used to select the supply voltage for the TC sensor that would result in the element operating temperature that offers the best combination of sensitivity to hydrogen measurement over a 0 to 10% range and also minimizes the effect of CO₂ concentration changes. Since water and CO₂ are not being measured, static concentration values are used in the software to correct the hydrogen reading for their presence in the process gas.

Sample System Design

DEVELOPMENT OF THE HEATED SAMPLE PROBE
Sample gas from the absorber overhead in the SCOT process contains moderately high levels of H2 and H2S that make it undesirable to vent to atmosphere, even at the low flow rates required by an analyzer. The sample is typically saturated with water at a temperature between 40°C and 50°C, requiring heating of the probe and other sample handling components to prevent condensation. An aspirator was selected for the sample transport method to avoid moving parts and return the sample to the process. The sample return point was the preferred location for the aspirator to avoid the vent backpressure issues associated with long sample lines. A probe was developed that combined the sample and vent into a single sample point connection that included two ball valves for manual isolation of the sample and vent, a 25 mm (1”) diameter membrane filter on the sample side, and an aspirator on the vent side. A heater is mounted on the top of the probe to allow all of the components including the sample and vent line connection fittings to be kept at a temperature higher than the water dew point. A flow schematic of the probe is shown in Figure 5. A Pt100 RTD in the probe is connected to the analyzer and used to provide probe temperature control. An internal temperature switch is used as a secondary over-temperature limit control. The heater is certified for use in Division 1 and Zone 1 hazardous areas.

FIGURE 5 – SCHEMATIC OF HEATED SAMPLE PROBE

Installation at Big Spring, Texas Refinery

The refinery has two SRUs, (one 60 t/d and one 75 t/d capacity), both are followed by a SCOT tail gas treater. ”SRU one” was operating under a grandfather clause without a tail gas treater. The load to SRU two is relatively stable with load variations being handled by SRU one. These high load variations seen in SRU one made it a good location to evaluate the SCOT H₂ + H₂S analyzer. The SCOT unit was constructed as a fast track project in 2004 with Ortloff Engineers and it was brought on-line in April 2005.

INSTALLATION
The analyzer included a heated sample switching system which allowed temporary manual switching between the standard sample point at the amine absorber outlet and another sample point before the absorber and the quench tower. A sample dryer was used to remove the water vapor from the sample to avoid problems with the very high water content when sampling before the quench tower. The sample system schematic is shown in Figure 6.

The analyzer was commissioned and after operating for some time, it was determined that the service life of the membrane filter in the probe could be increased by periodically switching between sample points while backflushing the idle probe with nitrogen. This backflushing seemed to be effective in temporarily removing particulate from the filter element. The probe was redesigned to allow the installation of a larger diameter membrane filter (approximately 55 mm diameter), which should result in higher sample flow rate and longer intervals before service is required. A small diameter flat element fiber filter was also added to reduce particulate loading on the membrane filter.

FIGURE 6 – SCHEMATIC OF ANALYZER SAMPLE SYSTEM

The redesigned heated probe was installed at the Big Spring refinery in September 2006 and operated for over 3 months without back-flushing. The probe seemed to be working properly at the end of this testing, but the flow rate had dropped from over 5 L/min (10 scfh) to about 2.5 L/min (5 scfh). The probe was disassembled and the inspection showed that the fiber filter had a 6 mm (1/4”) thick coating of what appeared to be sulfur as shown in Figure 7. It is thought that this sulfur deposited on the filter during a process upset when SO₂ breaking through the CoMo catalyst bed required that the absorber amine solution be replaced. The membrane filter was still fairly clean, in spite of the sulfur deposit on the fiber filter. A spare fiber filter was not available at the time, so the sulfur coating was scraped off and a new membrane filter was installed. Removal of this sulfur coating restored the flow rate to a normal value of 5 L/min.

FIGURE 7 – ULFUR ON PARTICULATE FILTER AFTER PROCESS UPSET

Data

Figure 8 shows some data from the analyzer recorded as one minute averages. Hydrogen sulfide concentration (ppm) at the absorber outlet and SCOT catalyst bed outlet temperature (°C) are plotted on the primary Y axis. The secondary Y axis shows the H₂ concentration at the absorber outlet (%), the COS concentration (ppm) at the absorber outlet, and the SO₂ concentration in the SRU tail gas which was measured by the SRU Air Demand analyzer. The first section of data shows normal, stable operation, with the H₂ between 2 and 3%, the COS at close to 0 ppm, and the H₂S at 100 ppm. During this same period, the SO₂ concentration in the tail gas is about 0.2% and the SCOT catalyst outlet temperature is just below 300°C. An upset occurs in the SRU at about 19:00 which drives the SO₂ concentration up above the 1% full-scale. The flow is switched to bypass around the SCOT unit in order to protect the amine from SO₂ breaking through the SCOT reactor (sometimes determined by a drop in pH of the water in the quench tower). Once the flow is bypassed, the analyzer is basically offline while the flow is in bypass mode. The temperature of the SCOT reactor drops during the period that the flow is bypassing the SCOT unit. SRU operation begins to stabilize at approximately 04:30 and after the SO₂ concentration drops down close to zero, the flow is switched back to the SCOT catalyst bed. The outlet temperature of the SCOT reactor rises up as it is heated by the hot gas stream from the reducing gas generator (RGG) and the hydrogenation reactions. Readings from the analyzer show that the process quickly returns to normal operating conditions once the reactor temperature stabilizes. Normally there is no COS being read by the analyzer at the absorber outlet. The brief rise in the COS to just over 5 ppm occurred when the sample point was temporarily switched to before the quench tower. This COS reading of 5 ppm is consistent with the equilibrium concentration typically seen from fully active catalyst (in this case about 1.5 years old).

FIGURE 8 – DATA FROM PROCESS UPSET IN SRU/TGT

FIGURE 9 – DATA SHOWING EFFECT OF SO2 EXCURSIONS

Although several hydrogenation and hydrolysis reactions are carried out in the SCOT reactor, the hydrogenation of SO₂ consumes a large portion of the hydrogen because it has a higher concentration than other sulfur species (with the exception of H₂S) and because 3 moles of H₂ are required for each mole of SO₂ as shown in Equation 2.

SO2 + 3H2 → H2S + 2H2O

Sulfur vapor and liquid can also consume significant quantities of hydrogen, but their concentrations do not typically vary as much as SO2. Stable operation of the SRU and good control over the SO2 concentration/SRU trim air are critical for successful operation of a SCOT TGT. This effect is shown in Figure 9, where excursions in SO2 concentration in the reactor feed result in sharp drops in H2 measured in the absorber overhead. The reactor outlet temperature increases with SO2 concentration because the reaction in Equation 2 is exothermic. Excess H2 was on manual control, but was at a sufficient level during stable operation to maintain the excess during upsets.

Conclusion

Good control and measurement of excess H₂ is important to provide stable operation of a SCOT TGT. Measurement of the H₂S at the absorber outlet can be used to monitor the operation of the amine treatment section. A process analyzer has been developed that combines these measurements along with optional measurement of COS. The analyzer was field tested for over a year in the SRU One SCOT unit at a refinery in Big Spring, Texas. The COS reading at the SCOT absorber outlet was close to zero which shows that good hydrolysis is being achieved on the catalyst in the SCOT reactor. Feedback from this testing was used to improve the design of the sample system, resulting in a system that operated reliably in a variety of conditions including process upsets.

While collecting data for this paper and investigating the performance as well as improvements in the heated probe design, it was decided to incorporate the probe modifications into the upgrade of the SCOT unit on “SRU two”.

References

1. Massie, Stephen and Huffmaster, Michael and McGillycuddy, Patrick, Low Temperature SCOT, A New Horizon in Tail Gas Treating, Laurence Reid Gas Conditioning Conference, Norman OK, 2006
2. Adam, Hamish and Harris, Phil, The Design and Practical Application of UV Process Photometers Part I. L’Analyse Industrielle, France, 1996.
3. Adam, Hamish and Harris, Phil, The Design and Practical Application of UV Process Photometers Part II. ISA Analysis Division Symposium, Chicago, 1996.