SPE International Oilfield Scale Conference and Exhibition

This paper describes a number of different evaporative processes which can cause flow assurance issues within oilfield production systems including chemical application via gas lift systems, halite deposition and gunking in injection lines. Similarities and differences are described and laboratory test methods are presented for each case.

While the challenges all involve evaporative processes, each system is different and requires suitable approaches to evaluate and mitigate the risks. These attempt to mimic the field system in the laboratory and allow observation under controlled conditions. Laboratory test methods vary from basic static bottle tests, through glass capillaries in autoclaves to dynamic tests using brine and a partially saturated gas phase, or neat chemical and dry gas lift media. In particular, the challenges when applying a chemical via a gas lift system will be described including field case studies.

Static tests with unlimited volume to evaporate produce a worst case for any evaporative process. However, it is frequently too severe to produce any useful results. Instead a test regime should be designed to mimic the field conditions. For example, evaporation within a pressure vessel can mimic the self-limiting process within a downhole injection line. Application of a chemical via a gas lift system requires a dynamic test where hot pressurised dry gas and neat chemical are co-injected with continual monitoring of gunking as indicated by flow path restrictions. Halites require a similar dynamic test method but with extensive modelling of the in situ saturation ratio to fully understand the system.

This paper will present case studies, summarise our understanding of the different evaporative processes, and give best practice guidelines for laboratory evaluation of the risks and mitigation strategies.

Hydraulic fracturing for shale gas production involves pumping large volumes of water; as a consequence of this, produced water management is an important topic to address in order to sustainably produce shale gas. It has been well documented that only approximately 10-40% of the pumped fluids will be produced back to the surface, and that there will be increased concentrations of various ions in the flowback water during this process. This flowback water, with high total dissolved solids and high concentrations of certain ions, presents a significant risk of mineral scaling (Blauch, 2009).

In general, it can be very challenging to identify the in situ formation water composition in shale reservoirs since samples of the formation water can be difficult to obtain. They may have been contaminated during the drilling process, reactions may have taken place due to fluid mixing between the injected fluid and the formation water, or simply they may not have been preserved appropriately (Pan, 2017). Some calculations of formation water compositions require to be preceded based on the observed compositional data; thereafter, the predicted formation water compositions are validated by comparison with the observed total dissolved solids (TDS) data. A two-phase 3D numerical flow model has been developed that includes a hydraulic fracture and is populated with shale reservoir properties. (This model assumes the hydraulic fracture is already established – i.e. the calculations include coupled flow and component transport, but the geomechanics are not considered). It is used to simulate fluid transport mechanisms within the shale system and to address the question – what causes the significant retention of fracture fluid in shale reservoirs. A series of simulations was performed to achieve a history match with observed flowback water data in a western Canadian basin (the Horn River Basin).

A further two-phase 3D flow model was developed to examine the scaling tendency due to the evolving produced brine composition over the lifetime of the well. It is based on the previously history matched model and includes the fracture fluid and formation water compositions to predict precipitation of minerals. Finally, scale inhibitor injection was simulated to examine the impact of inhibitor retention on well protection.

Deposition of inorganic scale has always been a common problem in oilfield pipes, especially in raising safety risk and producing cost. However, the fundamentals of deposition mechanism and the effect of various surface, temperature, flow rate and inhibitors on deposition rate has not been systematically studied. The objective of this research is to reveal the process of barium sulfate deposition on stainless steel surfaces.

In this work a novel continuous flow apparatus has been set up to enable further investigation of deposition rate, crystal size and morphology and the effect of scale inhibitor. In this apparatus supersaturate barium sulfate solution is mixed and passed through a 3 feet stainless steel tubing with ID = 0.04 inch or 0.21 inch at 70 to 120 degree C. The barium concentration is measured at the effluent to quantify the concentration drop. After 1 to 200 hours the tubing is cut into pieces to measure the barite deposition amount and observe the barite crystal morphology using SEM.

Under the experimental conditions, the deposition rate along the stainless steel tubing can be modelled by second order crystal growth kinetics, the SEM micrograph also shows that most of deposited barite is micrometer sized crystals. The highest deposition rate happens at the beginning of the tubing even before the expected induction time of bariums sulfate. The results indicated that the deposition happens even before the mixed solution is expected to form particles, which suggest that the heterogeneous nucleation might be the dominate mechanism in the initial stage, then crystal growth takes place and governs the deposition.

The mechanism of scale attachment to tubing surface has never been well-understood. The apparatus in this work provides a reliable and reproducible method to investigate barium sulfate deposition. The findings in this research will enhance our knowledge of mineral scale deposition process, and aid the use of inhibitors in mineral scale control.

Carbonate and sulphide scales can form in CO₂ and/or H₂S-rich environments in a process which we refer to as "auto-scaling", i.e. these scales form in the produced brine due to a change in conditions such as pressure and temperature, not due to brine mixing. Particularly in production systems, carbonate and sulphide scales can form due to the evolution of CO₂ and H₂S from the aqueous phase to the gas phase caused by a pressure decrease. Carbonate scale formation in this manner is broadly understood; however, there are details of precisely how this occurs in auto-scaling processes which are not widely appreciated.

Measuring the water composition at surface locations (e.g. at the separator) does not give a full indication per se of the amount of scale that has precipitated upstream of the sampling point. However, the composition of the water before precipitation occurs is required for predicting the scaling potential of the system, and this information is seldom available. In this paper, we propose a model that accounts for this issue, and that accurately calculates the carbonate and sulphide scaling profiles in CO₂ and/or H₂S-rich production systems by knowing only commonly available surface data – i.e. pressure, temperature, and fluid compositions (water, gas, and oil). A rigorous workflow which can do this calculation using any aqueous scale prediction model along with a PVT Model has already been published by the authors (Verri et al, 2017a). The current paper describes a new model to do these calculations which also includes an approach for estimating both the "correct" scaling case within a range of cases up to the "worst case" carbonate scaling scenario.

A scale prediction model has been developed to include a three-phase flash algorithm (using the Peng-Robinson Equation of State) coupled with an aqueous electrolyte model (using the Pitzer equations as the activity model). This model is used to run a demonstration example showing the procedure to calculate accurate auto-scaling profiles in CO₂ and/or H₂S-rich production systems, which is based on building a sensitivity analysis on the ions directly involved in precipitation reactions. We also note that auto-scaling profiles in production systems are commonly obtained by sectioning the production system – either by parameterising depth with pressure and temperature, or by selecting specific locations (e.g. DHSV, wellhead, etc.). Then, established guidelines to treat scale (or not) based on the calculated saturation ratios and precipitated masses of scale can be applied. We show that such an approach is not optimal and that it can lead to under or over-estimation of scale treatments. Furthermore, building on our previous method (Verri et al 2017a) we propose an approach to model the cumulative amount of scale formed under full equilibrium conditions, which is not dependent on how the production system is sectioned. By doing so, the correct amount of scale formed in the production system is always calculated, thus avoiding non-optimum scale treatments.

Our approach focuses on calculating the correct auto-scaling profiles in CO₂ and/or H₂S-rich production systems, and on correctly interpreting the results obtained by thermodynamic modelling and it can be easily integrated with commonly available scale prediction software.

An analytical technique was developed to directly determine the concentration of aqueous sulphide in exotic scale studies. The technique is based on measuring the absorbance of sulphide samples in a copper reagent using an ultra-violet visible (UV-Vis) spectrophotometer. Calibration curves, generated using stock sodium sulphide solutions, were compared and validated against earlier reports in the literature.

The instantaneous reaction of a stirred (1000 rpm) sulphide solution and a copper reagent produces copper sulphide nano-particles, which turn the solution colour from light blue to brown. Increasing concentrations of sulphides give more intense shades of brown, resulting in increased absorbance values when analyzed by UV-VIS at a wavelength of λ = 480 nm. The absorbance readings are proportional to the sulphide concentrations in the samples, thus generating a calibration curve.

The repeatability of these absorbance measurements has an average standard deviation of 1.9% and a correlation coefficient (R²) of 0.9998 over a range of 7-386 mg/L sulphide species in synthetic brines. The determination of sulphide concentrations from absorbance readings using a 2^nd order polynomial equation reveals average variations of ±1.2 and ±9.67 mg/L over the ranges of 7-70 and 105-386 mg/L, respectively. Furthermore, this technique was able to determine excess sulphide concentrations in solutions containing sulphide scale particles, as long as only clear filtered samples were analyzed. An alternative copper chloride reagent was used to prevent the formation of sulphate scale when barium or strontium ions were present in the brine being examined.

Comparing the results from the new detection method against the currently used technique reveals higher accuracy and practicality for the straightforward UV-VIS technique over the ICP analysis of quenched solutions. The new detection method allowed for the determination of any oxidative effects on the sulphide solutions, which is not identifiable using ICP.

One of the biggest challenges in designing squeeze treatments is ensuring appropriate chemical placement along the completion interval. Generally, the chemical slug is bull-headed; therefore, in long horizontal wells and/or crossflow wells, exposing the chemical to all the completion intervals might be difficult. In this paper we introduce a method to evaluate placement efficiency. If placement is inadequate, some sections of the well will be unprotected, resulting in an undesirable situation: the well may appear to be protected because the inhibitor return concentrations measured at surface are above the threshold, but there is a loss of production due to scale deposition in areas of the well not contacted by chemical. In these circumstances inhibitor placement can be accurately determined by production logging, but this can be prohibitively expensive. An alternative is to use tracers to evaluate the layer flow rate distribution, and therefore quantify chemical placement. The objective of this paper is to determine if a tracer package could be deployed as part of a squeeze treatment in challenging wells, in particular in the overflush stage. If there are zones in the wellbore at different pressures, then producing the tracer back in steps at different rates will result in the tracer return concentration profile having characteristic features that can be interpreted to estimate chemical placement.

Two three layer cases with crossflow are considered. In both cases, a tracer package was included in the overflush, and the resulting return profiles showed clearly the desired features. The main advantage of this approach is that there is no significant increase in the operational expense. The only additional expense will be the cost of the specific tracer and the subsequent analysis. It is envisaged that the cost is less than 5% of the total squeeze treatment cost. The results of this novel multi-rate post squeeze production stage following injection of tracer demonstrate the feasibility of including such a tracer package in a squeeze treatment. Data collected may then be used to optimise the design of subsequent treatments, to ensure that appropriate placement is achieved by rate control or by diversion, if necessary.

Predicting the formation of pH- dependent scales such as carbonates and sulphides requires a full calculation of all hydrocarbon and aqueous phases present to determine the distribution and speciation of CO₂ and H₂S in the system. Several commercially available software packages combine PVT calculations with scale predictions, but such packages are more targeted to aqueous systems and have limited hydrocarbon capabilities. Likewise, PVT modelling software focusing on the hydrocarbon phase does not always fully model the aqueous phase or can only predict a limited number of scales/complexes. Moreover, within each software we can select a large number of different Equations of State (EOS), activity models, equilibrium parameters etc., which may ultimately impact the final carbonate and sulphide scale prediction profile. The questions we try to answer in this work are: How important is the software selection and which parameters really affect the final scale prediction profiles? In what scenarios do these values matter and when are they not important?

In previous publications we laid out a clear rigorous procedure (workflow) for the prediction of carbonate and sulphide scales which can be applied using any commercial PVT and scale prediction software. Here we apply this general workflow using different software and EOS models to evaluate their impact on the final carbonate and sulphide scale prediction profiles for some specific carbonate/sulphide field scaling scenarios. The results show that despite the large number of modelling options available, there are two parameters that play a key role in pH-dependant scale predictions: partition coefficients of CO₂ and H₂S between gas, oil and water and the relative mole (and volume) distribution between each phase at selected temperature and pressure. The final scale prediction results can be accurate only when these values are accurate, irrespective of how they are obtained. This work shows the impact of choosing different software and equations on carbonate and sulphide scale predictions, not just as a "black box" software comparison exercise but with a clear connection between the aqueous and hydrocarbon phase thermodynamics, the scaling system and the final results.

Scale inhibitor squeeze treatments are used to prevent scale deposition in production wells. A treatment consists of injecting a scale inhibitor slug at a concentration between 5 and 15%, referred to as the main treatment, followed by an overflush, which will push the chemical slug deeper into the reservoir. During injection, the stages might undergo some degree of mixing in the tubing. This paper addresses the impact such mixing would have on the squeeze lifetime. A consequence of mixing between main treatment and overflush stages in the well tubing would be that although the same overall mass of scale inhibitor was injected, it would be distributed over a larger volume of water and therefore be exposed to the rock formation at a lower concentration than planned in the design. The degree of mixing in the tubing depends on a number of factors, such as tubing length and diameter, and the pumping rate. The phenomenon is described by the longitudinal dispersion coefficient, which may be calculated.

The resulting calculation may be defined as the spreading of a solute along the longitudinal axis, which leads to the spread of an initial high concentration slug with a low spatial variance to a final stage of low concentration with high spatial variance. The main objective of the paper is to study the effect of the degree of mixing of the main and overflush stages on the squeeze treatment lifetime. The net effect of full mixing would be that instead of there being two different stages at very different scale inhibitor concentration, a single stage at a lower concentration might be exposed to the rock formation. Two mixing profiles were considered, a short and long tubing; where the total injected volume is greater than and less than the total tubing volume, respectively. A number of levels of mixing were considered and compared to the base case, where no mixing was allowed. The results showed that squeeze lifetime is not significantly reduced if mixing occurs in a short tubing interval, whereas it can be reduced by up to 20% in a longer tubing interval.

The paper reviews operational issues that arise when MEG is used for hydrate inhibition, especially when it is regenerated and recirculated. Thermodynamic equilibrium software can assess the scaling risk to some extent. Utilisation of data on nucleation and growth of scale formers like calcium carbonate can enhance the accuracy of the predictions.

The presence of MEG and the conditions encountered in MEG systems favour aragonite crystallisation when the MEG solutions become supersaturated in CaCO₃. MEG retards the growth rate of all three CaCO₃ polymorphs, but the reduction is smaller for aragonite than for calcite and vaterite. The growth rate of siderite is also slowed down by MEG. However, MEG does not inhibit the nucleation and growth of carbonates.

Alkalinity in recycled MEG will enhance the scaling risk downstream of the MEG injection point when there is calcium in the produced water. Scale can be mitigated by scale inhibitors, but the selection process must ensure that the chemicals are tested at relevant conditions; i.e. with the expected MEG concentration, alkalinity and pH.

Many MEG recovery units have a pre-treatment system for controlled removal of carbonates and to some extent hydroxides. This reduces the amount of scale that may form in the regeneration system. In the pre-treatment, alkalinity dosed as hydroxide and/or carbonate forces precipitation of calcium, strontium and iron carbonates and magnesium hydroxide. The supersaturation is generally so high that scale inhibitors are not able to prevent precipitation of the solids.

Calcium carbonate (CaCO₃) scale can form through an "auto-scaling" process in production systems with a CO₂-rich environment due to fluid (water/oil/gas) depressurisation. Thermodynamic modelling is used to estimate the amount and severity of CaCO₃ scale precipitation in this context in order to design scale inhibitor or other types of treatments. However, field experience has indicated that thermodynamic calculations often lead to an overestimation of the calcite scale problem. One possible source of this discrepancy may be due to kinetic effects; i.e. that the calcite is somewhat oversaturated (Saturation Ratio, SR >1) but the driving force is not sufficiently large and so the deposition is kinetically "slow". The industry response to this situation has been to come up with some simple heuristics based on field observations, and "rules of thumb" have been developed to account for this apparent overestimation of calcite deposition. The central objective of this paper is to try to address the problem of using such an arbitrary field procedure for calcite scale prediction by introducing the kinetics of calcite deposition in a thermodynamically consistent manner. We view the calcite auto-scaling system as one which moves from SR < 1 (non scaling) incrementally to one that is slightly supersaturated (SR slightly > 1). By making the deposition rate a function of SR, this would give slow rate of deposition initially, but as the system moved into the more scaling regime in the production system (SR > 1) then the deposition rate would increase. However, throughout the system, this kinetic formulation must limit correctly (i.e. it must be consistent with) the underlying equilibrium thermodynamics of the full brine/oil/gas system. Thus, we replace the idea of using heuristic estimates of when calcite scaling occurs by one where an estimate (or measurement) of the kinetics is made; indeed, a range of kinetics rates can easily be run to give an envelope of calcite scaling profiles in the wellbore and throughout the production system.

In this paper, we present a model that incorporates a fully consistent kinetic formulation into a general thermodynamic scale prediction model. This model can then calculates scaling profiles in production systems considering both kinetic and thermodynamic effects. In particular, a rate law for the precipitation of CaCO₃ based on the respective degree of super-saturation is coupled with the Heriot-Watt FAST Scale Prediction model (HW FAST). HW FAST uses the Pitzer equations and the Peng-Robinson Equation of State to model, respectively, the aqueous and hydrocarbon phases (gas and oil), and it has been developed to calculate CaCO₃ scaling profiles caused by a de-pressurisation effect in CO₂-rich production systems.

First, we present an equilibrium thermodynamic example calculation showing that CaCO₃ scale precipitates in CO₂-rich production systems due to a de-pressurisation effect, and that precipitation is more severe topsides where the pressure is low than it is downhole where the pressure is high. It is explained that the amount of scale precipitated in this auto-scaling process must be plotted as a cumulative amount, in order to avoid the calculation of a potentially misleading scaling profile. This calculation is then repeated, but also considering kinetic effects for systems with varying temperatures and water flowrates. In the example presented here, we show that this system with sufficiently "low" water flowrates can be approximated by a thermodynamic calculation, and that systems with "high water" flowrates must take kinetic effects into consideration. This scaling profile can then be used to more accurately design scale inhibitor treatments, thus avoiding under or over-treatments (e.g. opting for continuous scale inhibitor injection instead of the more expensive squeeze treatment).

Our approach focuses on calculating the correct scaling profile in auto-scaling processes, both qualitatively and quantitatively, by coupling a kinetic formulation to a thermodynamic model, and it can be readily extended to other auto-scaling processes. Further, our kinetic model can be easily integrated with commonly available scale prediction software.