Copyright 2001, Society of Petroleum Engineers Inc. This paper was prepared for presentation at the 2001 SPE Annual Technical Conference and Exhibition held in New Orleans, Louisiana, 30 September–3 October 2001. This paper was selected for presentation by an SPE Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper, as presented, have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material, as presented, does not necessarily reflect any position of the Society of Petroleum Engineers, its officers, or members. Papers presented at SPE meetings are subject to publication review by Editorial Committees of the Society of Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract The toxicity of hydrogen sulphide in hydrocarbon streams is well known in the industry and considerable expense and efforts are expended annually to remove hydrogen sulphide to a safe level. In large production facilities, it is generally economical to install a regenerative system for treating sour gas streams. However, during the development stages of relatively small low-sour gas fields at remote and normally unmanned locations where regenerative systems are not practical nor economical, it is necessary to treat the sour gas production with non-regenerable scavenging processes. In the development of its low-sour Zechstein gas reserves in the Coevorden field in the North-East The Netherlands, the Nederlandse Aardolie Maatschappij (a Shell operating unit, hereafter referred to as NAM) decided to adopt continuous direct-injection of liquid scavenging agents as the lowest overall cost process having the least environmental impact and the highest energy efficiency. At the inception of the project, the operating parameters controlling the scavenging efficiency using direct injection of liquid scavengers in this system were largely unknown. Consequently, numerous field trials using different chemistries and different injection mechanics had to be carried out. In this paper we present the results of these field trials, which ultimately led to a very successful and profitable field development strategy. A variety of very challenging operational problems were encountered, and solved. Reference is made to injection nozzle blockages, fouling of glycol gas dehydration systems, severe scaling problems in production and downstream water treatment/injection facilities,

inadequate hydrogen sulphide removal efficiencies and HS&E related issues. A better understanding of the fundamental relationships between operating parameters governing direct-injection processes and associated chemical development and application methods has been gained. Communication and integration of experience and knowledge between the operating unit and its chemical supplier were key success factors in this achievement, as was the endurance and continuing support of field operations staff in facilitating the resolution of difficult problems.

Introduction NAM’s Ten Arlo system (Fig. 1) produces gas from 32 satellite locations and four gas treatment / compression plants. The majority of the fields in the system produce sweet gas from the Limburg reservoir. However, the system’s heavily compartmentalized Coevorden field also contains gas accumulations in the Zechstein reservoir with H2S concentrations up to 300 ppm(v). Developing these accumulations would require extensive modifications to existing (sweet) gas production / treatment facilities to ensure safe operations and the prevention of H2S emissions. The producible reserves were small yet economically attractive. However, due to their geographical scatter, a pipeline grid connecting these low-sour accumulations to an existing nearby plant utilizing a regenerative solvent process for gas sweetening was found to be economically unfeasible. In-field desulpherisation with, for example, a solid-bed adsorption process, was found to be technically feasible, but the required capital investment and perceived life-cycle costs could not be justified. In order to allow production of the low-sour gas without violating the Ten Arlo sales gas specification for maximum allowable H2S concentration (5 mg H2S/Nm3 or 3 ppm(v)), in-field desulpherisation utilizing a commercially attractive alternative technology was looked for.

Direct-Injection H2S Scavenging The injection of chemicals into produced gas streams to remove H2S is a fairly old industry practice. Formaldehyde

SPE 71541

The Development of Low-Sour Gas Reserves Utilizing Direct-Injection Liquid Hydrogen Sulphide Scavengers J.G.R. Eylander, SPE, H.A. Holtman, Nederlandse Aardolie Maatschappij, T. Salma, SPE, M. Yuan, SPE, M. Callaway, and J.R. Johnstone, SPE, Baker Petrolite

2 J.G.R. Eylander, H.A. Holtman, T. Salma, M. Yuan. M. Callaway, J.R. Johnstone SPE 71541

has been one of the most frequently used materials, but its use is strongly discouraged owing to the reported carcinogenic properties. Chemistries such as sodium chlorite, caustic soda, glyoxal and others have been tested as well. These chemistries often have severe disadvantages associated with them, ranging from handling and operational problems as a result of high reactivity to slow reaction rates. The development of a non-regenerable class of chemistry commonly referred to as triazines was disclosed in 1990. The term ‘triazines’ is used for a group of compounds, which in reality are substituted hexa-hydro triazines. In this paper, the terms ‘triazine’ and ‘hexa-hydrotriazine’ will be used interchangeably. A general representation of a substituted hexa-hydro triazine is given in Fig. 2, in which R1 to R6 may be arbitrary hydrocarbon groups. Two well-known triazine forms are disclosed in patents1,2, in which R1, R2 and R3 are ethanol or methyl groups, respectively, and R4, R5 and R6 are hydrogen. The products are produced on a technical scale by reacting either monoethanolamine (MEA) or a trimethylamine (TMA) with formaldehyde. The actual reaction mechanisms of triazine compounds with H2S are not well understood. The reaction is complex and produces multiple reaction products. However, repeated laboratory analysis of spent MEA triazine using NMR C13 spectroscopy has established that H2S reacts irreversibly with MEA triazine with the S-atom being built into the ring structure forming primarily two sulfur containing products – thiazine and dithiazine. In the reaction of MEA triazine with H2S, the sulfur atoms are in the S2- oxidation state so that solid elemental sulfur does not form and the reaction products are liquid. Depending upon treatment rates, the reaction also releases either one or two molecules of water-soluble alkanolamine molecules: R-NH2. Because at least one R group is retained in each product, the primary products of the reaction of MEA triazine with H2S are low volatility liquids that are soluble in water (see Fig. 2). First field experiences A first field trial was carried out in 1995 at the Coevorden-24 location utilizing a 1,3,5-trimethyl-hexahydro-1,3,5 triazine (product A from supplier A), with injection upstream of the glycol dehydration unit (Fig. 3). During the trial gas production originated primarily from two wells, with the gas containing 24 and 45 ppm(v) H2S respectively. Injection took place without the use of an atomizer and at a dosage rate of approximately 6 L/kg H2S. As a result of the injection, the H2S concentration in the location’s export gas decreased to well below the 5 mg/Nm3 limit. However, it was found that 40% by weight of the chemical did not react and ended up in the produced water phase. Not only did this result in a potential water disposal problem (owing to prevailing legislation), also an unusual accumulation of solids was encountered in the glycol system. The subsequently required change-out of the wet glycol filters resulted in unacceptable amine-type emissions spreading well beyond the location. Furthermore, the functioning of the in-line H2S analyzer (of the type utilizing a lead acetate tape) was disrupted.

Given the disappointing product yield (i.e. percent reduction in H2S) and the encountered operational problems, the decision was taken to repeat the trial with a more concentrated version of the same triazine. Initially this product was injected downstream of the glycol dehydration unit directly into the location’s export gas flowline with the aid of an atomization nozzle at a dosage rate of approximately 4 L/kg H2S. Repeated blockages of the nozzle quickly resulted in its removal and the trial was continued without it. Emphasis was placed on determining the product yield and substantiation of the manufacturer’s claim that reaction products would preferentially dissolve in the co-produced condensate. It was found that injection of the more concentrated product in the pipeline-wet export gas stream rapidly reduced the H2S level to well below the contractual limit, with an H2S reduction of approximately 97%. This represented a major improvement with respect to the first trial. As illustrated in Table 1, both unreacted and reaction products were preferentially soluble in the produced water phase. The presence of unspent and reacted scavenger in the water necessitated permit application for the installation and use of permanent injection facilities on the location, which were installed in 1996. The chemical cost associated with the subsequent continuous H2S scavenging treatment of the export gas stream amounted to approximately US$ 30 per kg H2S. Field development progression Several problem areas developed over time with the chemical scavenging application as described above. Ten Arlo’s central gas treatment plant was confronted with increasing pH levels in the glycol dehydration system and amine emissions from vessels and drain pits, phenomena which were traced to the scavenger application on the Coevorden-24 location. Development of low sour gas accumulations at certain satellite locations lacking gas drying facilities was not feasible with this scavenger application, not only because transport lines did not meet the NACE specification for sour service but also chemical costs would quickly become prohibitive. By 1997, the requirement for H2S scavenging at the wellhead, driven by economics of the operation combined with increasing operator complaints about amine emissions, triggered field testing of an alternative H2S scavenger consisting of 1,3,5-tri(2-hydroxyethyl)-hexahydro-1,3,5-triazine (product B from supplier B). To distinguish between the tests, three principal injection regimes were tested as depicted in Fig. 4, without the use of atomization nozzles. Results are given in Table 2. Test A: wet gas injection regime, moderate residence time

Scavenging of H2S from 40 ppm(v) to the required <3 ppm(v) required a scavenger dosage of approximately 10 L/kg H2S. In order to establish whether the H2S is indeed removed via the produced water phase, a sulfur balance was made on the water, condensate and gas phases. During the test period the sulphide concentration in the produced water phase is seen to increase and the sulfur balance (gas to liquid) climbs from an initial 44% to an 85% fit at the end of the test. The

SPE 71541 The Development of Low-Sour Gas Reserves Utilizing Direct-Injection Liquid Hydrogen Sulphide Scavengers 3

slow establishment of equilibrium is explained by the inability to fully drain produced water from the test separator prior to test execution.

During the test, the pH of the produced water increased from an initial 5.6 to 7.4 as the scavenger dosage increased. Massive overdosing of the scavenger raised the pH to 7.6. Throughout the test period and even for a prolonged period thereafter, no solids or scale formation was observed in the sampled fluids. However, produced water samples developed a yellowish color over time. Test B: wet gas injection regime, short residence time Decreasing the contact time between the scavenger and gas had a negative impact on the consumption of the product. Given the short duration of the test, no other conclusion from this observation was drawn other than that it could possibly support an observation from other work3 that the chemical injection rate has an effect on the mass transfer of the chemical. Test C: dry gas injection regime, long residence time In this regime, we expected the scavenger to scavenge H2S from the dry gas, albeit inefficiently, with the reaction products falling out as solids in the absence of water. In this test, injection of the scavenger in the dry gas stream downstream of the glycol dehydration system had no noticeable effect on the H2S concentration in the gas. This confirmed the earlier observation that the scavenging reaction takes places in the produced water phase. The product’s manufacturer claimed that a minimum required water content of 2 L/million Nm3 of gas was needed. One of the objectives of the above tests was to prove that neither the chemical nor its reaction products would cause any adverse effects on water-condensate separation/quality and on the performance of the glycol dehydration system. No such effects were observed, but the relatively short test period could not provide conclusive evidence. From the test results it was concluded that the scavenger was very effective at removing small quantities of H2S from wet/saturated gas streams. If sufficient contact time was provided for, the required dosage of 10L/kg H2S, while a factor 2.5 higher than that required for the export gas scavenger, still provided a reduction in chemical costs to approximately US$ 10 per kg H2S. Given the positive trial outcome, permanent injection facilities for product C were installed on the location’s main header while retaining the scavenger (product B) injection into the export gas stream as back-up facility. In order to gain confidence that product C could also be used on satellite locations by direct-injection at the wellhead, a more rigorous field trial was designed that was also aimed at optimizing the treatment to achieve further cost reductions.

Extended field testing Well Coevorden-26 on the Coevorden-24 location was selected for this field trial. This well is completed on both the (sweet gas producing) Limburg reservoir and the (low sour gas producing) Zechstein reservoir. The pressure balance quickly favors (fractured carbonate) Zechstein production after the well is opened up to the extent that steady state gas production with an H2S concentration increasing to approximately 130 ppm(v) can be maintained for a period of about 10 days (see Fig. 5). After this time, the on-set of liquid loading necessitates well shut-in for several weeks to achieve pressure build-up. Early 1998 the well was connected to the test header (see Fig. 4), with the injection point (without atomization nozzle) as per Test A of the previous field trial. Injection of product C commenced simultaneously with gas production, with the scavenger injection rate set at the maximum expected H2S level at the recommended dosage of 10 L/kg H2S. The implied over-injection of the chemical was intended. During the trial period, the H2S scavenging performance matched that of the first trial, yielding H2S reduction in excess of 99%. However, a gradual increase in the pressure drop across the flowline connecting the well to the test header from an initial (normal) 2 bar to 8 bar was also observed. Subsequent internal inspection of the flowline revealed that massive calcium carbonate scaling had occurred at and several meters downstream of the injection point. Following flowline clean-up through acidization, a repeat trial was carried out, this time utilizing an atomization nozzle and preventing over-injection by tailoring the scavenger injection rate to the actually produced H2S concentration. Although this reduced scale deposition to a certain extent, the net results were far from satisfactory. Given the potential threat to the Coevorden field development plans, a major effort was undertaken in conjunction with the chemical supplier to solve the scaling problem. An experimental scavenger was developed (product D from supplier B), in which product C was blended with a phosphonate-type scale dispersant. Late 1998 through 1999, a prolonged period of often-problematic field-testing followed utilizing this experimental product. Without going into details, the following observations were made during this period:

The experimental product appeared to be effective in mitigating the deposition of calcium carbonate scale. However, the chemistry employed in formulating the product resulted in excessive formation of a very viscous foam, disrupting flow and liquid level control systems and carrying the risk of spillover into the glycol dehydration facilities.

To counteract foam formation, a silicone polymer emulsion was mixed into the scavenger/scale dispersant blend. The resulting mixture appeared to be inherently unstable, requiring continuous mixing on site to avoid phase separation. Notwithstanding, it proved to be impossible to achieve a homogeneous mixture and foam suppression was only partially achieved. In addition, a

4 J.G.R. Eylander, H.A. Holtman, T. Salma, M. Yuan. M. Callaway, J.R. Johnstone SPE 71541

50% reduction in the product performance in terms of liters of product required per unit mass of hydrogen sulphide removed was experienced.

Sludge formation occurred in the flowline (see Fig. 6), in the test separator and in downstream water/condensate storage facilities to the extent that it became unacceptable.

Irrespective of the encountered problems, it was observed that the scavenger injection rate had an effect on the product yield, but contrary to published results from earlier work3, no clear pattern could be established. Factors complicating result interpretation are that (a) lowering of the flowing tubing head pressure results in higher linear gas velocities in the flowline and (b) the concomitant increase in the hydrostatic head in the well results in higher water-gas ratio’s. The dilution resulting from the latter could alter the stoichiometry of the reaction between the triazine and H2S such that the triazine could possibly react with more H2S (on a molar basis) than in the (relative) absence of dilution effects. During the various trials, in going from higher to lower linear gas velocities the net reduction in H2S was seen to decrease at first, but after passing through a minimum a higher reduction in H2S was observed again. This would seem to indicate higher scavenger injection rate requirements at lower linear gas velocities, when there is less water production. The significance of these observations could only be ascertained after a solution was found for the encountered problems. It was conceded at the time by the incumbent chemical supplier that such a solution could not be made available in the time remaining for the required firming up of field development plans. Further product developments The field testing of hydrogen sulphide scavengers carried out thus far had generated a number of very clear technical performance criteria for wet gas application. Claiming the successful development of a combined scavenger/scale inhibitor formulation, which could meet these criteria, supplier C was invited to participate in field testing of this product (product E). The claim was founded as follows:- Product E was developed specifically to minimize calcium carbonate scale formation that would normally arise from using a highly alkaline triazine scavenger in calcium and / or high bicarbonate produced water. Based on extensive laboratory evaluation of various scale inhibitors, a proprietary phosphonate scale inhibitor was identified as the most effective at combating calcium carbonate scaling resulting from injection of triazine into a produced brine. The laboratory experiments indicated that, (a) addition of the chosen scale inhibitor significantly reduced calcium carbonate scaling tendency, (b) it had no adverse effect on H2S scavenging efficiency of the triazine, and (c) it did not cause fluid foaming. The result was the development of product E that contained an optimum percentage of the chosen scale inhibitor. The scale inhibitor is completely soluble in high pH brines and it exhibits a high efficiency at inhibiting high pH induced scale precipitation. It is known that this is a nucleation

inhibitor, that is, it inhibits the nucleation of calcium carbonate scale crystals, thus keeping the scaling ions in solution. Next generation field-testing Following flowline clean up through acidization, field testing of product E was initiated on well Coevorden-26 in the same manner as the preceding field test, i.e. injection at the wellhead through an atomization nozzle. Results are shown in Fig. 7. Injection at a dosage of 10 L/kg H2S (slightly higher than the supplier recommendation of 8 L/kg H2S) yielded an average H2S reduction of 70%. By increasing the dosage to approximately 15 L/kg H2S, the reduction in H2S rose to approximately 93% (against a stipulated performance criterion of >99%). Reducing this dosage to lower levels was immediately coupled with increasing residual H2S levels. The influence of the linear gas velocity on the product yield is shown in Fig. 8, whereby the observed data have been fitted to a calculated trend utilizing Holt’s two-parameter linear exponential smoothing method4. The trend appears to indicate the existence of a threshold limiting value above which the product yield is independent of the linear gas velocity; the injection rate seems to have little influence. During the test no increase in the pressure drop across the flowline was observed. However, a camera-run revealed the presence of a thick layer of calcium carbonate scale blocking the top half of the flowline (Fig. 9). From this observation was concluded that the flowline acidization preceding the test had only been partially effective. It was therefore decided to repeat the test after cleaning the flowline through sequential high-pressure water jetting and acidization, followed by a camera-run to confirm that the required clean-up had been achieved. With the scavenger injection rate set at the maximum expected H2S level (and hence an initial massive overdosing), the results shown in Fig. 10 were obtained. Throughout the test, as the amount of overdosing decreased with increasing H2S level in the produced gas, an average H2S reduction (product yield) of 97% was achieved at the expense of an overall average dosage of 20 L/kg H2S. Again, no increase in pressure drop across the flowline was observed, but a camera-run revealed the presence of thick sludge layer on the bottom of the flowline and randomly deposited cauliflower-like deposits (Fig. 11). Clearly, overdosing to achieve the targeted H2S reduction was not the way forward. The disappointing product yield at higher than expected dosage requirements seen in the first test also indicated that an alternative approach was required to reach our objectives. Challenge of chemical delivery method Following a period of consultation with the chemical supplier, the suitability of the hitherto used chemical delivery method was challenged. In the preceding field tests, use had been made of an atomization nozzle specifically designed to avoid the blockage problems experienced with the conventionally designed nozzles some years earlier. Design review raised the suspicion that an inefficient atomization process might be the key contributing factor to the problems

SPE 71541 The Development of Low-Sour Gas Reserves Utilizing Direct-Injection Liquid Hydrogen Sulphide Scavengers 5

experienced. It was proposed to utilize Bete® PJ series direct-pressure atomization nozzles, adapted to allow the use of high-pressure gas to boost the velocity in the atomizer nozzle. Both laboratory and field evaluation of such gas-assisted (2-phase) atomization nozzles had shown approximately 30% efficiency improvement over the normal single phase atomization. Again, utilizing sequential high-pressure water jetting and acidization, the flowline was cleaned (confirmed through camera-run) and a 2-phase atomization nozzle installed, the high-pressure gas being delivered via nitrogen batteries. Injection was subsequently initiated, initially at the supplier recommended product dosage and gas-assist pressure of 20 – 25 bars above the flowing tubing head pressure. Results of this next test are provided in Fig. 12. The percentage H2S removal initially obtained with the 2-phase atomizer was disappointing. However, scavenging performance could be improved upon by changing product injection rates and nozzle sizes. Ultimately, the desired reduction in H2S was achieved, but only at a high product dosage of some 20 L/kg H2S. By reducing the gas-assist pressure to only 2 – 3 bar above the flowing tubing head pressure, an H2S reduction (product yield) >99% could be consistently achieved at a product dosage of approximately 15 L/kg H2S. This result was maintained at the tail end of the test, when the gas-assist needed to be shut-in due to depletion of the nitrogen batteries. A subsequent camera-run through the flowline showed the complete absence of any scale deposits or sludge. Mitigation of scale build-up in the flowline was only one of the two objectives that were successfully met by the use of the gas-assisted atomizer. The other objective was to obtain higher scavenging efficiencies as was observed in previous laboratory and field trials using the gas-assisted atomizer. However, further evaluation indicated that such increased scavenging efficiencies were primarily observed in low-pressure systems with operating pressures less than 5 to 10 barg. In the Coevorden field, the operating pressures were significantly higher, ranging from 35 to 75 barg. This is thought to be the key contributing factor in not achieving further improvement in scavenging efficiency using gas-assisted atomization. The matter was not further pursued, principally because very few of the locations targeted for development had gas available at sufficiently high pressure to achieve 2-phase atomization (implying a compression requirement) and commercially available nitrogen gas generators could also not deliver the required pressures. Given the observation at the tail end of the last test that product yield could be maintained without gas-assist, the decision was made to carry out an additional trial. Prior to discussing this additional trial, it is of interest to revisit the trial discussed above. The influence of the linear gas velocity on the product yield is shown in Fig. 13, whereby as before the observed data have been fitted to a calculated trend utilizing Holt’s two-parameter linear exponential smoothing method4. Comparing with previously obtained results, recalling that then also different atomization nozzles are compared, it is again seen that the product dosage (and hence injection rate) seems to have little influence. In this test,

the linear gas velocity did not enter the region lower than 4 m/s where the earlier test showed evidence of the existence of a threshold limiting velocity value. A more rigorous examination of the data obtained from the current test revealed that they could be fitted to the equation: Product yield (%) = m * linear velocity (ms-1) + b (r2 = 0.71)

through linear regression analysis with the same mean absolute percentage error as the fit obtained with Holt’s method. Dual injection In gas systems where short contact times and high H2S loading are specifically challenging towards efficient reduction in H2S, injection of chemical at multiple injection points has often been used to improve the scavenging efficiency. After the 2-phase atomizer trial, a dual injection approach was utilized in the Coevorden system by injecting the chemical both at the wellhead and at the test header. Use of dual injection resulted in >99% removal of H2S and thus yielded the desired results in scavenging efficiency. Improved scavenging efficiency in a multiple injection system (in this case two-point injection) can be explained as follows. Firstly, the volume of scavenger injected at the first injection point (wellhead) is typically less than the stoichiometric amount needed to react with the total H2S in the gas. Presence of scavenger in limiting quantities allows for optimum utilization of the scavenger. Therefore, scavenger injection at the first point removed the bulk of the H2S (75 – 85% reductions in H2S). Secondly, injection of the scavenger at the second point (test header) provides a polishing or finishing effect. The neat high strength scavenger contacts the lean gas (low in H2S) to react with the residual H2S molecules that were not removed after the first scavenger injection. The limiting reactant at the second point is the H2S and can be removed by the high concentration of the scavenger. The dual injection system using product E, applied through single phase Bete® nozzles, yielded removal efficiencies in the range of 99%, reduced scavenger consumption to, on average 9 L/kg H2S and provided effective control of calcium carbonate scaling. Current development status Prior to the high gas nomination period of the Winter 2000/2001, two satellite locations of the Coevorden-17 gas treatment plant were converted to facilitate the production of low-sour gas wells with H2S levels of up to 220 ppm(v) through dual injection of product E. An extended optimization program, through which overall product dosages of as low as 7 L/kg H2S could be achieved as well as consistent product yields of >99%, is coupled with further investigation of the operating parameters affecting the scavenging reaction. Scientifically intriguing but operationally of little consequence are the following observations. Well Coevorden-31 produces gas at an initial rate of 420,000 Nm3/day, gradually declining to 310,000 Nm3/day at a constant flowing tubing head

6 J.G.R. Eylander, H.A. Holtman, T. Salma, M. Yuan. M. Callaway, J.R. Johnstone SPE 71541

pressure of approximately 90 bar and with a constant H2S concentration of approximately 220 ppm(v). The linear gas velocity is in the range 9 – 12 m/s. Injection of product E at the wellhead alone does not result in the desired H2S reduction (product yield), and it needs to be augmented with a second point injection at the main production header. At another satellite location, well Coevorden 36 produces gas at a fairly constant rate of 400,000 Nm3/day at a flowing tubing head pressure of 70 bar and with a constant H2S concentration of 220 ppm(v). The linear gas velocity is approximately 6 m/s. Injection of product E at the wellhead alone yields the desired H2S reduction, and at this location no second point injection is required. As other wells are being brought on stream, more data can be gathered to hopefully unravel the apparently complex role of interlinked process parameters on the physical chemistry of direct-injection hydrogen sulphide scavenging.

No evidence of scale formation has been encountered to date at any of the locations where product E is being applied. Chemical costs are approximately US$ 11 per kg H2S (slightly lower or slightly higher in individual cases). Back-up facilities utilizing product B (in pipeline-dry export gas treament) are required during well start-up, but only when fast production ramp-ups are employed. The benefits5 of automation of the hydrogen sulphide chemical scavenger injection systems, combining currently available technology into a multi-faceted cost saving and reliable tool, are recognized and its economic implementation is being pursued.

Conclusions 1. Selection and application of direct-injection liquid

hydrogen sulphide scavengers requires an understanding of the fundamental relationships between operating parameters and scavenging (physico-)dynamics.

2. More fundamental research, preferably executed ‘in the field’ is required to unequivocally identify these relationships and to allow ‘best in class’ design of treatment applications.

3. In applying direct-injection liquid hydrogen sulphide scavenging technology, a systematic as well as a systemic approach, with continuing challenge of chemical supplier claims, is required to achieve the desired results.

4. Mutually beneficial solutions to problems can be found between operating companies and their preferred chemical suppliers by not concentrating on internally focused commercial ambitions but by concentrating on: we shall solve this problem together, regardless. And reaping the benefits of such solutions.

Acknowledgements We thank NAM and Baker Petrolite Corporation for their permission to publish this paper. In particular, very special thanks goes out to NAM’s operations staff, not only for providing gas treatment plants and field personnel to support this project while maintaining active production operations, but more importantly also for their continuing

support and dedication in the face of adversity. We have exceptional pride in their and our refusal to yield.

References 1. Dillon, E.T., Composition and method for sweetening

hydrocarbons. U.S. Patent no. 4,987,512, December 1990. 2. Bhatia, K., Thomas, A.R. & Sullivan, D.S., Method of

treating sour gas and liquid hydrocarbon streams. European Patent Application 94305225.8, February 1995.

3. Fisher, K., Initial results from GRI’s 30 MMscf/day direct-injection H2S scavenging test facility. Paper presented at the Eight Gas Research Institute Sulfur Recovery Conference, Austin, Texas, October 1997.

4. Kvanli, A.H. et al, Introduction to Business Statistics. West Publishing Co., 1992, p. 745 f(f).

5. Roth, D. et al, Automated chemical control of H2S content of natural gas. Paper SPE 67247, presented at the SPE Production and Operations Symposium, Oklahoma City, Oklahoma, 24-27 March 2001.

Table 1: Fate of unreacted triazine (product A) and reaction products

Stream Concentration unreacted triazine

(mg/kg)

Concentration reaction products

(mg/kg)

Produced water ± 200 ± 250

Condensate None detected < 1

Table 2: Test results with product B

Qgas

(E3 m3)

Pgas

(bar)

Tgas

(oC)

Contact

Time

(s)

Linear

gas velocity

(m/s)

Qinj.pump

(L/d)

H2S

(ppm(v))

H2S

removed

(%)

Scavenger

Dosage

(L/kg)

S2- in water

(mg/kg)

Sulfur

balance

(%)

Produced

water pH

Test A 145 61 36 - 5.4 0 40 0 0 4 - 5.6

145 61 36 19.3 5.4 41 21 48 9.9 2100 43.9 7.0

245 45 45 8.2 12.7 123 4 90 9.6 6800 75.1 7.4

245 45 45 8.2 12.7 159 0.4 99 10.9 8500 85.3 7.4

245 45 45 8.2 12.7 327 0 100 - 7.6

Test B 234 41 49 - 13.6 0 29 0 0 - - -

234 41 49 1.8 13.6 95 10 65 14.2 - - -

234 41 49 1.8 13.6 136 3 90 14.7 - - -

234 41 49 1.8 13.6 218 1.2 96 22.1 - - -

8 J.G.R. Eylander, H.A. Holtman, T. Salma, M. Yuan. M. Callaway, J.R. Johnstone SPE 71541

Figure 1: the Ten Arlo gas gathering system

Fig. 2. Reaction pathway of triazine and H2S

Ten Arlogathering station

Wannerperveen De Wijksatellites satellites

Coevorden-24 gathering Coevorden-17 gatheringstation station

Coevorden-33 satellite Coevorden-31 satellite

Coevorden-20 satellite Coevorden-5 satellite

Coevorden-21 satellite Coevorden-10 satellite

Den Velde satellite Hardenberg-2 satellite

Hoogenweg satellite Hardenberg-4 satellite

H2S(g) H2S absorbs into the aqueous triazine solution where, because of the latter's alkaline pH, it can be expected to fully dissociate into the HS- ion. It is thisspecies which subsequently reacts with the triazine molecule. The rate ofthe chemical reaction is dependent on the mass transfer rate, which in turnis dependent both on physical (diffusion) and chemical processes in theaqueous phase.

R1 R1 R1

N N N

H2C CH2 H2S H+ + HS- H2C CH2 H2C CH2

N NR3 R2 N S S S

C R3 C C H2 H2 H2

1,3,5-hexahydro-triazine thiazine dithiazine

SPE 71541 The Development of Low-Sour Gas Reserves Utilizing Direct-Injection Liquid Hydrogen Sulphide Scavengers 9

Fig. 3. Simplified set-up field trial with product A

Fig. 4. Simplified set-up field trial with product B

Injection point with product A

Gas from satellites

Main headerCOV-2

GlycolCOV-21 contactor

tower Injection point with product B

COV-33Export gas to Ten Arlo

Gas samples Injection point Test B

Gas to slugcatcher

Injection point Test A

Testheader

Testseparator (approx. 100m, 4" line)

Liquid samples COV-24 well, 40 ppm(v) H2S (approx. 25 m, 4" line)

Water/condensate to tank

Gas from satellites

Main headerCOV-2

GlycolCOV-21 contactor

tower Injection point Test C

COV-33 Export gas to Ten Arlo

(approx. 70m, 10" line)

Gas samples

10 J.G.R. Eylander, H.A. Holtman, T. Salma, M. Yuan. M. Callaway, J.R. Johnstone SPE 71541

Fig. 5. H2S production profile of well Coevorden-26 at Q = 100,000 Nm3/day

Fig. 6. Sludge formation in flowline

0

1

10

100

1000

0.00

4.00

14.3

3

33.5

0

50.5

0

61.0

0

74.0

0

86.0

0

96.6

7

114.

50

117.

58

120.

00

122.

00

127.

67

141.

17

145.

50

156.

27

218.

27

236.

52

Time (hours)

H2S

con

cent

ratio

nppm(v) kg/day

Sludge layer

SPE 71541 The Development of Low-Sour Gas Reserves Utilizing Direct-Injection Liquid Hydrogen Sulphide Scavengers 11

Fig. 7. Results 1st field trial with product E

Fig. 8. Influence of linear gas velocity

0.00

2.00

4.00

6.00

8.00

10.00

12.00

14.00

16.00

18.00

20.00

0.00

1.00

4.00

9.00

14.33

29.00

33.50

45.25

50.50

55.00

61.00

68.00

74.00

80.00

86.00

91.50

96.67

98.00

114.5

011

5.50

117.5

811

8.50

120.0

012

1.00

122.0

012

3.42

139.8

414

1.17

142.5

014

5.50

146.4

215

6.27

212.7

721

8.27

234.5

223

6.52

Normalized time (hours)

kg H

2S/d

ay

0.0

2.0

4.0

6.0

8.0

10.0

12.0

14.0

16.0

18.0

L/k

g H

2S

H2S at wellhead H2S at manifold Scavenger injection

0.0

20.0

40.0

60.0

80.0

100.0

3.83 3.87 3.91 3.95 3.99 4.03 4.04 4.21 4.29 4.33 4.35 4.38 5.31 5.40

Linear gas velocity (m/s)

Prod

uct y

eild

(%)

0

100

200

300

400

500

Scav

enge

r ing

. rat

e (L

/d)

Measured Calculated Inj. rate

12 J.G.R. Eylander, H.A. Holtman, T. Salma, M. Yuan. M. Callaway, J.R. Johnstone SPE 71541

Fig. 9. Calcium carbonate scale in flowline

CaCO3 scale deposit top of flowline

Bottom of line

Top of line

Picture Orientation

SPE 71541 The Development of Low-Sour Gas Reserves Utilizing Direct-Injection Liquid Hydrogen Sulphide Scavengers 13

Fig. 10. Results 2nd field trial with product E (overdosing conditions)

Fig. 11. Sludge layer and cauliflower-like deposits

0.00

5.00

10.00

15.00

20.00

25.00

0.00

4.75

26.00

29.50

48.00

75.75

119.5

0

142.0

0

148.8

3

162.8

3

186.3

3

254.8

3

Normalized time (hours)

kg H

2S/d

ay

0.05.010.015.020.025.030.035.040.0

L/kg

H2S

H2S wellhead H2S manifold Scavenger inj.

14 J.G.R. Eylander, H.A. Holtman, T. Salma, M. Yuan. M. Callaway, J.R. Johnstone SPE 71541

Fig. 12. Results of 2-phase atomization trial with product E

Fig. 13. Influence of linear gas velocity

0.00

5.00

10.00

15.00

20.00

25.00

0.00

7.08

22.7

5

40.2

5

43.2

5

46.2

5

64.7

5

93.2

5

133.

25

162.

25

168.

25

189.

92

210.

00

225.

25

Normalised time (hrs)

kg H

2S/d

ay

0.0

5.0

10.0

15.0

20.0

25.0

30.0

L/kg

H2S

H2S wellhead H2S manifold Scavenger inj.

0.00

20.00

40.00

60.00

80.00

100.00

120.00

4.15

4.46

4.68

5.02

5.14

5.27

5.53

5.94

6.63

6.70

6.86

7.02

7.02

Linear velocity (m/s)

Prod

uct y

ield

(%)

0

5

10

15

20

25

30Pr

oduc

t dos

age

(L/k

g H

2S)

Measured Calculated Dosage