Oil & Gas - Maximising Recovery Programme (formerly SHARP) IOR Views

Introduction

The activity of sulfate-reducing bacteria (SRB) in oil field brines causes numerous problems associated with reservoir souring [1]. This has large economic consequences and numerous control strategies have been attempted to alleviate the problem. Methods include eliminating sulfate from water prior to injection, use of biocides to suppress microbial activity, or stripping the resulting H2S by caustic washing, or further oxidation of H2S to elemental sulfur by chemical methods.

More recently, biological approaches have been applied to control souring. These approaches are of two types: (1) utilisation of hNRB to out compete SRB for limited carbon sources, or (2) use of NR-SOB to oxidise the H2S products of SRB and thus remove the problem. In the present work we present models for these biological solutions, based on extensive experimental work done in our laboratory [2-5].

Several numerical simulation models at the laboratory and field scale have been developed previously for bacterial enhanced oil recovery methods. Numerical models for bacterial souring and treatment are less common, with the models of Ligthelm et al [6] and Sunde et al [7] being the best examples to date. In this work, we will apply a commercial, fully-featured, thermal-compositional simulator STARS with microbial kinetics capabilities to nitrate- and nitrite-based souring control for the first time.

Description of Experiments

Microbial Metabolic Action - Nitrate or nitrite addition can be used to control sulfide production by SRB according to a variety of mechanisms as summarised in Figure 1. Nitrate and nitrite can stimulate both autotrophic nitrate-reducing sulfide-oxidising bacteria (NR-SOB; Figure 1B) and heterotrophic nitrate-reducing bacteria (hNRB; Figure 1C). NR-SOB are effective souring control agents as their growth depends on the removal of H2S, their electron donor. Complete oxidation of sulfide to sulfate by NR-SOB is indicated in Figure 1B. Because NR-SOB recycle sulfide back to sulfate, SRB and NR-SOB are able to coexist symbiotically as long as electron donors for SRB growth remain. Souring control by NR-SOB thus depends on enough nitrate (or nitrite) to recycle sulfide until the SRB electron donor is exhausted.

Nitrate (or nitrite) achieves souring control very differently in the presence of hNRB. Electron donors for these organisms are the same organic components of oil that SRB oxidise. This creates a direct competition between sulfate reduction (Figure 1A) and the thermodynamically more favourable nitrate reduction (Figure 1C). For sulfide production to be contained by this mechanism the amount of nitrate added must again be sufficient to completely oxidise common electron donors, otherwise sulfate reduction will still be allowed to occur with what remains.

In the absence of NR-SOB or hNRB, nitrate does not have any effect on the growth and activity of SRB. Nitrite, however, can be strongly inhibitory to SRB in the absence of nitrite-reducing bacteria. Nitrite interferes with the terminal enzymatic step of the biochemical pathway by which SRB reduce sulfate to sulfide, as indicated in Figure 1D.

Figure 1: Metabolic Activities of Various Anaerobic Microbes

Upflow Bioreactor Experiments and Results - Understanding mechanisms of souring control using nitrate or nitrite have been initially worked out in batch culture studies using pure cultures or bacterial consortia growing in serum bottles with defined media (e.g. containing lactate and sulfate to promote SRB growth and nitrite or nitrate to promote souring control) [2, 3, 5]. In order to verify batch culture findings in a more realistic scenario, continuous up-flow packed-bed bioreactors were constructed [3, 4] to simulate oil reservoir conditions during secondary oil recovery (Figure 2). Glass columns (4.5 cm x 64 cm) equipped with five sampling ports were packed with sand (average size 225 µm) to provide a matrix for biofilm establishment. Anaerobic liquid medium containing sulfate and lactate was introduced at the bottom of the bioreactor. Biofilms were allowed to establish by inoculating bioreactors with produced water from a Canadian reservoir. SRB from the inocula were enriched in the sulfate/lactate environment during start-up periods of ca. 20 days of batch wise operation followed by ca. 20 days of increasing the rate of inflow from 0 to 9 ml h^-1 using a peristaltic pump. To these bioreactors, nitrate or nitrite was added which allowed enrichment of nitrate-reducing bacteria and hence souring control according to the mechanisms discussed above.

Figure 2: Schematic of Upflow Bioreactor System

This experimental system was used to test four different medium conditions with both nitrate and nitrite. Initially, SRB medium containing 7.8 mM sulfate and 25 mM lactate was used. Nitrate or nitrite was added incrementally to determine the amount of each required for souring control under these conditions. These tests were followed by similar tests using medium containing: 7.8 mM sulfate and 12.5 mM lactate, 4.7 mM sulfate and 12.5 mM lactate, and 3.0 mM sulfate and 6.25 mM lactate. The results of these experiments indicate that the amount of nitrate or nitrite required to contain souring is directly proportional to the amount of organic electron donor (lactate) present, not the amount of sulfate (and hence sulfide). This agrees with and has contributed toward our understanding of souring control mechanisms mediated by NR-SOB or hNRB reducing nitrate or nitrite, respectively. Nitrate is not required in as high an amount as nitrite in these situations due to its greater oxidative power. Souring control in the nitrite-treated bioreactors was due mainly to nitrite reduction mechanisms (Figures 1B and C) and not nitrite inhibition (Figure 1D) presumably because effective nitrite concentrations were arrived at gradually and not delivered all at once.

Bioreactors are currently being used to investigate the effects of nitrate and nitrite souring control mechanisms on corrosion problems associated with souring, the effect of very low sulfate concentrations and the effect of bio-augmentation with isolated pure strains of NR-SOB.

Description of Laboratory Simulation Model

Microbial Growth and Metabolite Production Model - The metabolic activity of sulfate reducing bacteria is modelled employing the approach of Okabe and Characklis [8, 9]. Two stoichiometric reactions are written representing the oxidation of an organic carbon source (energy) and the synthesis (growth) of bacterial cells, with lactate as the electron donor and the energy source. These two equations are then combined into one overall reaction assuming 94% usage of lactate for energy and 6% use for growth:

1 lactate + 0.47 H₂SO₄ + 0.036 NH₃ → .94 acetate + 0.94 CO₂ + 0.47 H₂S + 0.18 SRB + 1.05 H₂O

This latter overall reaction is employed in the simulator model.

A completely analogous approach can be employed for nitrate-reducing bacteria, with the overall reaction for hNRB as:

1 lactate + 0.752 HNO₃ + 0.036 NH₃ → .94 acetate + 0.94 CO₂ + 0.376 N₂ + 0.18 NRB + 1.8 H₂O

again assuming a 94%/6% split between energy/growth usage of lactate.

Finally the NR-SOB oxidation reaction uses CO2 as the carbon source for cell synthesis and H2S as the energy source leading to an overall stoichiometric reaction of the form:

.06 CO2 + 1.5 HNO₃ + 0.012 NH₃ + 0.973 H₂S + 0.036 H₂O → 0.973 H₂SO₄ + 0.752 N₂ + 0.06 NRSOB +.752 H₂O

This reaction also assumes the same split between energy needs and growth. We have decided to keep energy/growth split fixed for all reactions during our matching and sensitivity analysis.

In addition to the above stoichiometric requirements, each of the three overall reactions (for the three bacteria classes) requires kinetic expressions for the overall conversion rates. Following Okabe and Characklis [8, 9], we will assume a Monod reaction form for each reactant, except we assume that NH3 concentrations are always in excess and hence don't participate in the reaction. We use matching from the bioreactor experiments to assess the appropriate parameter values.

Upflow Bioreactor Numerical Model - The grid for the laboratory model has 64 cells (of 1 cm length) in the vertical direction, with 1 cell in cross section (area 16 cm2). The cells were chosen such that cells 4, 18, 32, 46, and 60 correspond to the locations of the sampling ports in the experimental bioreactor. The model is initialised with values of pressure and temperature of 150 kPa and 31°C respectively. An initial uniform microbial concentration was used for each of the three microbial species types (SRB, hNRB, NR-SOB) assumed in the microbial consortium. These conditions were assumed to be representative of the state of the system after many days of growth after the inoculum was injected through each port.

The timing and injection sequence attempts to follow that conducted in the experiment. The experiment is divided into three parts: (a) SRB growth (21 days); (b) NR-SOB growth with varying nitrate concentrations; and (c) system response to varying carbon source with varying nitrate concentrations. For the SRB growth period, the flow rate was increased daily in a stepwise fashion from 12 cm3/day (0.5 cm3/hr) to the final value of 216 cm3/day (9.0 cm3/hr) over a period of three weeks, with injected concentrations of all species matching the initial in-situ values. Although experimental details are lacking, the simulation shows a drop in initial sulfate concentration to zero, and a concurrent rise in H2S production over this time. Also a predicted increase in SRB concentration is seen, such that the majority of SRB growth occurs near the inlet end, where a higher concentration of lactate and sulfate is maintained through injection.

Thereafter the period of NR-SOB growth is simulated by stepwise addition of nitrate concentration (0.0 to 17.5 mM in 2.5 mM steps) with each step on the order of 10 days. The result is a reduction in H2S concentration, a re-increase in sulfate concentration, and the growth of NR-SOB bacteria. This latter growth also shows a non-uniform distribution of bacteria along the bioreactor.

Figure 3: Predicted Metabolic Responses to Nitrate Injection at Maximum Lactate and Sulfate Levels

Figure 3 illustrates the simulated growth mechanism in more detail, for the restricted portion of nitrate addition in steps. The times 0.0 to 23 days represents the establishment of an SRB biofilm grown over lactate/sulfate as seen by 8mM of sulfate reduction being supported by oxidation of 16 mM (out of 15 mM) lactate to acetate and CO2. At day 23, addition of NO3^- starts. This allows growth of NR-SOB, which removes H2S and produces SO4^-2 again. This SO4^-2 is immediately used by the SRB (with further lactate consumption). This is why the H2S concentration remains level. The same process continues to occur at successively increasing NO3^- concentration, until finally this sulfide/sulfate cycling exhausts all of the lactate present (around day 50). Now that the lactate (carbon source and electron donor) is all gone, the SO4^-2 produced by the NR-SOB cannot be reduced to H2S anymore, and we see the remaining nitrate additions remove the H₂S according to the NR-SOB metabolism. SRB activity is still occurring, but we have finally achieved a nitrate concentration that exhausts all lactate and allows net H₂S removal.

The final period consists of monitoring the bacterial response to reduction of sulfate concentration by half (again varying nitrate concentrations in steps) over a 50 day period; reduction of lactate concentrations by half (with varying nitrate concentrations in steps) over a 30 day period; and reduction in both lactate and sulfate by half (with varying nitrate concentrations). These results can be found in the original papers.

As indicated by the nonuniform bacteria population over the various stages of the experiment, it is useful to monitor the complete metabolic changes at different locations in the upflow bioreactor. Figure 4 illustrates a comparison of experimental results and simulation at Port 5. Both simulation and experiment show that the Ports 2, 3, and 4 behaviour is very similar to the Port 5 response, while Port 1 has a significantly different response.

Figure 4: Comparison of Experimental and Simulated Metabolite Profiles at Port 5 (NR-SOB Response to Nitrate)

Sensitivities of the Bioreactor Model - As an aid to further understanding of the model, and as a precursor to bring the model to the field, numerous sensitivity runs were performed. These include changes to the rate of NR-SOB and SRB metabolic activities, and the assessment of hNRB activity with simulations using the "competitive exclusion" mechanism. The latter simulations indicated very similar activity predicted for NR-SOB and hNRB, since both mechanisms have the same net overall removal of oil organics by nitrate. Hubert et al [4] conducted bioreactor experiments with nitrite as well, resulting in similar souring control. Nitrite addition can also be modelled by modifying the metabolic equations to represent nitrite reduction.

Irrespective of the above-stated current limitations of the model, one aspect bears emphasising. The time to conduct one complete experiment in the upflow bioreactor typically runs three months to six months, while the equivalent simulation typically takes one minute. Thus there is an obvious motivation to continue to calibrate and improve the numerical representation of the process in order to reduce the actual number of experiments run to study the sensitivities of the process design. As envisioned, the simulation model will never replace the need for extensive selected experiments.

A Field Model Extension of Reservoir Souring and Treatment

A STARS field prototype reservoir model was constructed to illustrate some of the consequences and issues of applying this new technology in a field context. A layered steeply dipping model was chosen, with permeabilities, well spacing and oil properties characteristic of North Sea operations. Numerous examples of reservoir souring issues have been reported by North Sea operators.

The fluid model consists of the live oil components (dead oil plus solution gas) with a certain fraction of the dead oil component specified as the natural carbon source for bacterial growth (e.g. the volatile fatty oil fraction). Additionally the water species, gas species, and microbial species utilised in the laboratory model are included, with the exception of lactate (whose effects are represented by the VFA component in the field model).

The base case model consists of several years of fluid production up-dip at varying water injection rates down-dip. A slow overall movement of water up-dip and a fast tongue of water moving upwards in the bottom portion of the upper high permeability zone can be expected. During the time of water injected, the amount of H2S generated by the in-situ SRB is predicted to grow continuously. In addition to the increase in H2S, other products (CO2 and acetate) of SRB metabolism also increase according to the metabolic equations.

A period of HNO3 injection starts with nutrients used to stimulate growth of NR-SOB. Decrease of CO2 and H2S are initially observed as these species are utilised in the overall NR-SOB reaction while their subsequent increase corresponds to a re-growth of SRB stimulated by the production of additional H2SO4. Figure 5 shows the predicted spatial distributions of H2S after the treatment period.

Although the generic model is three dimensional, the cross-sectional view was chosen to illustrate the similarities and differences from the laboratory experiment and modelling. In addition to the length scale of the process and the difference in the initial distribution of the carbon source caused by the initial oil saturation distribution, other obvious differences are the tilted effects of gravity and a layered permeability system. These factors result in a non-uniform distribution of metabolites. Nevertheless, the metabolic production response is surprisingly similar to the laboratory results, giving confidence to the laboratory experimental approach to capture relevant mechanisms. Field sensitivity studies of additional possible mechanisms provide impetus for further laboratory design and study of these factors.

Figure 5: Predicted H2S Distribution in Generic Field Model after Nitrate Injection

Conclusions and Future Work

This article has illustrated evolving capabilities to model microbial souring treatments at the laboratory and field scale. A relatively simple but metabolically sound representation of competing bacteria kinetic processes has been used in a reservoir simulator model to match detailed mechanistic studies of the activities of SRB, hNRB, and NR-SOB bacteria in a very detailed laboratory bioreactor, and to predict consequences of such a treatment in a field prototype model. Where exact information is lacking, results appear consistent with acceptable microbial growth characteristics.

A wide range of additional experimental information on this process is available from our University of Calgary laboratory, and could be usefully simulated. In addition to the mechanisms associated with different nitrate- and nitrite-reducers, bioreactor studies on corrosion induced by H2S production should be modelled for a more complete description of souring issues. The field modelling has indicated the necessity of acquiring additional information on the nature of the petroleum carbon source, microbial decay processes, and microbial plugging and propagation.

As with laboratory simulations, it is important to directly compare model field simulations with actual field results. To be most effective, the simulator can be used in a feedback manner to define critical parameters to be measured and matched to ensure a successful pilot treatment design. Future work will involve more detailed comparison with a wide range of varying experimental characteristics and direct analysis of field pilot projects where possible.

References

• Cord-Ruwisch, R., Kleinitz, W., Widdel, F., "Sulfate-Reducing Bacteria and Their Activities in Oil Production", J. Pet. Tech., Vol. 39, p. 97, (1987).
• Nemati, M., Jenneman, G., Voordouw, G., "Mechanistic Study of Microbial Control of Hydrogen Sulfide Production in Oil Reservoirs", Biotechnology and Bioengineering Eng., Vol. 74, p. 424, (2001).
• Nemati, M., Mazutinec, T.J., Jenneman, G., Voordouw, G., "Control of Biogenic H2S Production with Nitrite or Molybdate", J. Industrial Microbiology and Biotechnology, Vol. 26, p. 350, (2001).
• Hubert, C., Nemati, M., Jenneman, G., Voordouw, G., "Containment of Biogenic Sulfide Production in Continuous Up-Flow Packed-Bed Bioreactors with Nitrate or Nitrite", Biotechnology Progress, Vol. 19, #2, p. 328, (2003).
• Greene, E., Hubert, C., Nemati, M., Jenneman, G., Voordouw, G., "Nitrite Reductase Activity of Sulphate-reducing Bacteria Prevents their Inhibition by Nitrate-Reducing, Sulphide-oxidizing Bacteria", Environmental Microbiology, Vol. 5, p. 607, (2003).
• Ligthelm, D.J., de Boer, R.B., Brint, J.F., and Schulte, W.M., "Reservoir Souring: An Analytic Model for H2S Generation and Transportation in an Oil Reservoir owing to Bacterial Activity", paper SPE 23141, presented at the Offshore Europe Conference, Aberdeen, U.K., Sept. 3-6, (1991).
• Sunde, E., Thorstenson, T., Torsvik, T., Vaag, J., Espedal, F, "Field Related Mathematical Model to Predict and Reduce Reservoir Souring", paper SPE 25197, presented at the SPE Int. Symp. on Oilfield Chem., New Orleans, LA, March 2-5, (1993).
• Okabe, S., Characklis, W., "Effects of Temperature and Phosporous Concentration on Microbial Sulfate Reduction by Desulfovibrio Desulfuricans", Biotechnology and Bioengineering, Vol. 39, p. 1031, (1992).
• Okabe, S., Nielsen, P., Characklis, W., "Factors Affecting Microbial Sulfate Reduction by Desulfovibrio Desulfuricans in Continuous Culture: Limiting Nutrients and Sulfide Concentration", Biotechnology and Bioengineering, Vol. 40, p. 725, (1992).