Scale under Turbulent Flow

Tekna $23^{\text {rd }}$ International Oil Field Chemistry Symposium, 18-21 March 2012, Geilo, Norway

Scale under Turbulent Flow

Leif O. Jøsang ${ }^{1}$, Maria N. Psarrou ${ }^{1}$, Terje $\varnothing$ stvold ${ }^{1}$, Tore Tjomsland ${ }^{2}$, Kari Ramstad ${ }^{2}$ and Britt Marie Hustad ${ }^{2}$
${ }^{1}$ Department of Materials Science and Engineering, NTNU, Trondheim, Norway
${ }^{2}$ Statoil ASA, Norway

Abstract

Field observations presented indicate that scale has a preference to deposit at positions in production wells where irregularities in the flow pattern occur. This could be at bottle necks where the shear forces and turbulence are particularly high, or after a tubing expansion where back currents and eddies occur. One negative consequence of this is that critical well equipment such as Inflow Control Valves (ICV), Inflow Control Devices (ICD) and Down Hole Safety Valves (DHSV) will be particularly exposed to scale deposition, which may cause equipment failure and increased operational risk.

To mitigate the local scaling problems, it is necessary to understand the mechanisms that result in elevated scaling potential at turbulent conditions. Laboratory equipment has been developed to a) Confirm that the scaling potential is higher at turbulent conditions b) Investigate the mechanisms leading to this increased scale potential.

Experiments show that calcium carbonate scale deposits faster after a tubing expansion than in a pipe with constant diameter. Several mechanisms are discussed in the paper, but it was found that the most relevant mechanisms resulting in increased scaling potential after an expansion are i) higher residence time in back currents ii) bulk particles trapped and settling, and iii) less erosion to remove loose particles in back currents.

The effect of scale inhibitors present in the system has also been studied.
This new knowledge can be used to reduce the risk of scale deposition in ICVs, ICDs and DHSVs. Further work to exploit these new opportunities has been suggested.

Introduction

Field observations

Mineral scale, i.e. the precipitation and deposition of minerals on surfaces, occurs in many processes where water is involved, from household percolators, via heat exchangers, water boilers, power plant condensers, desalination plants, to oil wells. Common scale minerals include $\mathrm{CaCO}_{3}, \mathrm{CaSO}_{4}, \mathrm{SrSO}_{4}, \mathrm{BaSO}_{4}, \mathrm{Mg}(\mathrm{OH})_{2}$. Industrially, scale control is important both from an economic point of view and for safety reasons. Therefore scale has been a subject of study for decades. The thermodynamics of scale formation is fairly well understood by now, and several commercial scale modelling packages are available at the market (Pitzer 1995; Kaasa 1998). The effects of precipitation kinetics, flow rates, turbulence, particle deposition and removal have not been studied as thoroughly.

In oil production, scale is sometimes found selectively or in higher amounts in areas where turbulence or change in flow patterns are expected, such as near expansions and constrictions. This has also been observed in wells with an inhibitor concentration above the industry standard Minimum Inhibitor Concentration (MIC).

Tekna $23^{\text {rd }}$ International Oil Field Chemistry Symposium, 18-21 March 2012, Geilo, Norway
Gustavsen et al. (2010) reported observations of barium sulphate scale located just downstream each of the nozzles in a lower completion with nozzle based inflow control devices (ICD) in a Heidrun field oil production well (Norwegian Sea). Since sulphate scale saturation is less dependent on pressure, it is considered likely that the high deposition rate downstream the nozzles were due to turbulence and back currents. Tailby et al. (1999) reports similar observations in a North Sea well with gas lift mandrels. Calliper and gamma ray logs showed enhanced scale deposition in the area with the gas lift mandrels, representing an irregularity in the completion. Payne (1987) reported observations of calcium carbonate scale in the safety valves at the North Sea Murchison oil field. The reduced bore of a downhole safety valve (DHSV), will according to Payne both give a pressure drop that increases the calcium carbonate saturation ratio, and also create turbulence and shear forces that enhance the precipitation reaction kinetics.

Statoil’s internal DHSV failure frequency statistics also show a correlation between oil fields with scaling potential and failure frequency, see Fig. 1. This indicates that in fact scale may be a frequent reason for DHSV failure.

Scale induced DHSV failures represent both a significant safety issue and a large economic losses since wells that do not pass the regular safety valve tests have to be shut in, waiting for workover. This is not a new or Statoil specific issue according to the Norwegian Petroleum Safety Authority. This can be observed in data published in their latest reporting of risk levels in the Norwegian petroleum industry (Petroleumstilsynet 2011), which is shown in Fig. 2.

The DHSV failure frequency is well above Statoil’s target of $2 \%$ for many of the fields in operation, hence measures should be taken to reduce the safety risk and production losses related to scale induced well integrity issues. To do this, it is necessary to increase the understanding of the mechanisms leading to enhanced scale deposition in parts of the production wells and systems where increased turbulence as well as back currents and eddies occur.

Fig. 1 Scaling tendency plotted versus DHSV failure frequency for Statoil operated oil and gas fields at the Norwegian continental shelf. The scaling tendency index has been derived from the number of scale control related well interventions performed per field.

Tekna $23^{\text {rd }}$ International Oil Field Chemistry Symposium, 18-21 March 2012, Geilo, Norway

Figur 92 Andel feil for DHSV

Fig. 2 Field average DHSV failure for oil and gas fields in operation at the Norwegian continental shelf. Average failure frequency for the period 2002-2010 as well as 2010 failure frequency is shown. Figure from Petroleum Safety Authority Norway (Petroleumstilsynet 2011)

From field to laboratory

The field observations made the background for establishing the current study. Wells that were assumed to be protected by scale inhibitors flowing back from the formation at concentrations higher than the Minimum Inhibitor Concentration (MIC), turned out not to be protected. The MIC of a field or well is normally determined through lab experiments where the inhibitor performance is established with appropriate Saturation Ratio (SR) values expected to be found in the well during production. However, the lab test procedures generally do not include the effect of turbulence.

SR of calcium carbonate is defined as
$S R=\frac{a_{C a^{2+}} \cdot a_{C O_{3}^{2-}}}{K_{S P}^{C a C O_{3}}}$
where $a$ represents the activities of calcium and carbonate and $K_{S P}$ represents the thermodynamic solubility product of calcite.

It is not always easy to reproduce well data in the laboratory. Representative flow rates, particle content and organic components extracted from the oil cannot be accurately reproduced in the lab. Even the P, T gradient from reservoir to topside may not be well known and the flow in the well will not be uniform due to restrictions of different sorts, e.g. safety valves. Such restrictions may create turbulence or eddies where $\mathrm{CO}_{2}(\mathrm{~g})$ evolves more easily. Dead-water areas may also occur, where particles are more easily trapped and particles only loosely attached to the pipe walls are not removed by the fluid flow. At such places scale may be formed more easily at lower SRs than otherwise.

Tekna $23^{\text {rd }}$ International Oil Field Chemistry Symposium, 18-21 March 2012, Geilo, Norway
The two most commonly used test methods today to understand scale formation problems and inhibitor efficiency are the static bottle and capillary tube blocking tests. The main advantages with these tests are that they are quick and simple to perform. None of the tests involve the effect of turbulence or shear forces, and the initial goals in the present project were to establish if tube expansions/restrictions had an influence on MIC, and if so, find a new method for determining MIC. It became obvious fairly early in the project that a fundamental understanding of scaling processes during oil field operations were needed in order to develop a new method to establish MIC.

In this study, $\mathrm{CaCO}_{3}$ was chosen as a model scaling mineral for several reasons: The chemicals involved are harmless even in large quantities, and the scale formed can easily be dissolved in acid, enabling simple cleaning and re-use of all experimental devices exposed to $\mathrm{CaCO}_{3}$ precipitation.

Why scale forms necks on constrictions and expansions

Scale deposition happens according to one or more of the following mechanisms:

• Bulk precipitation followed by particle deposition and removal
• Heterogeneous nucleation
• Secondary nucleation
• Crystal growth

In the early stages of scale deposition in cooling water systems bulk precipitation and particle deposition is the most important mechanism. The particle size distribution of scale and silt deposited has been found to be the same, and the addition of scale inhibitors reduces both the deposition of scale and silt (Hawthorn 2009). The growth rate will then be governed by the ratio of particle deposition and particle removal. According to Cleaver and Yates (1976) particle removal is related to wall shear stress. Below a critical shear stress, particle removal does not occur. The removal is assumed to occur in turbulent bursts, the frequency of which is directly related to flow rate. This is a strong indication that initial scale deposition could occur first in areas where the shear forces are low, such as in dead water areas in front of tube constrictions or behind expansions. On the other hand, Quddus and Allam (2000) after studying barium sulphate deposition on a rotating steel cylinder found that deposition rates increased with increasing Reynolds numbers (Re) and that the growth was diffusion controlled, indicating that once a scale layer is formed, and the growth is more rapid when the shear rate is high. Garcia et al. (2001) show the same tendency with $\mathrm{CaCO}_{3}$, at least under laminar flow.

The most important method to reduce scaling in oil production is the use of scale inhibitors. The main effect of these is to adhere to and block the growth sites of crystals. This can have several secondary effects:

• Small, suspended crystals are more easily transported out of the well
• Inhibitor molecules adsorbed on the surface of both tube walls and crystals hinder direct contact between steel and crystals, thus blocking secondary nucleation.
• Inhibitor adsorbed on the well surfaces hinder heterogeneous nucleation
• The inhibitor weakens the attractive forces between the walls and the scale particles, making it:
• Less likely that particles stick to the surface on collision.
• More likely that particles are removed by shear forces
• The inhibitor promotes precipitation of polymorphs that are more soluble and/or more easily removed by shear forces than the most stable polymorphs, e.g. vaterite rather than calcite.

Scale inhibition by a well-known phosphonic acid: HEDP was evaluated by Garcia et al. (2001) The scaling rate increased dramatically when the flow rated was reduced from Re 3200 to 1600. At the low flow rate, a uniform layer of small particles covered the whole surface of the equipment. At Re 3200 scale was only found in areas protected from the shear forces of the flow. The protected particles were large and irregular. This indicates that

Tekna $23^{\text {rd }}$ International Oil Field Chemistry Symposium, 18-21 March 2012, Geilo, Norway
this inhibitor lowers the adhesion of bulk particles to the wall, and thus scale may form preferentially in areas where the wall shear stress is particularly low.

Scale is sometimes formed in systems where inhibitor levels are above MIC. During milling operations, necks are typically found in constrictions and expansions, or other areas where eddies are expected. Scale starts to form on walls either by direct heterogeneous nucleation, secondary nucleation after wall/particle collisions or by particles from bulk colliding sticking to the wall. The following possible mechanisms for preferential growth in turbulent areas are suggested:

Hypotheses

1. Scale forms in bulk regardless of inhibitor concentrations exceeding the traditionally determined MIC. This is supported by available literature. (Hawthorn 2009) (Morizot, Neville et al. 1999)

a. Particles settle more easily in dead-water and eddies in the flow, allowing crystal growth and recrystallization
b. Particles loosely attached to the wall are not removed by shear forces in dead-water areas.
c. Particle-wall interactions are increased around constrictions and expansions
2. Scaling rate increases under turbulent conditions.

The only reference found on this indicates the opposite for scale inhibitors. Increased flow rates increases HEPD (Garcia, Courbin et al. 2001) inhibitor efficiency. This is probably due to crystal modification lowering wall- to crystallite attachment and/or secondary nucleation as the inhibitor stops direct interaction between wall and crystal.
a. Carbonate scaling rates increases under turbulent multiphase flow due to increased $\mathrm{CO}_{2}$ transfer from the water phase to the gas and oil phases
3. Inhibitor efficiency is lowered by turbulence
a. Protective layers on surfaces are disrupted by high shear forces or bursts (local violent collapse of the main flow regime in multi-phase flow)
The HEPD study mentioned above does not indicate this, but the turbulent energy was not very high (Re 3600) during this experiment.
b. Protective layers on particles or nuclei are disrupted by bursts or shear forces
c. The inhibitor is destroyed under turbulent conditions

The only indirect literature reference touching on this is Hasson et al. (2003) who found that a $\mathrm{CaSO}_{4}$ inhibitor was not adversely affected by turbulence from pump impellers.
d. Residence time for parts of the flow exceeds the inhibitor retardation time in wakes and eddies

Sulphate scaling under turbulent flow

Ramstad et al. (1994) studied sulphate scaling using a high flow rate “once-through” dynamic tube set-up similar to the turbulence rig described in the present work (see below). The experiments were performed at $40^{\circ} \mathrm{C}$ mixing full synthetic Brent type of formation water and sea water 50/50 by volume. According to Multiscale simulations this gives a $\mathrm{BaSO}, \mathrm{SR}$ of $\sim 560$. Concentrated brines were pumped into a main flow of tap water providing the correct salinity solutions and flown together into a T-junction (ID 16 mm , length 0.21 m ) followed by a vertical tube (ID 9 mm , length $6-9 \mathrm{~m}, \mathrm{Re}=33600$, flow velocity $2.6 \mathrm{~m} / \mathrm{s}$ ). Precipitation started immediately at mixing of the two brines.

The scale deposition was highest near the mixing point, and the deposition rate increased with decreasing tube diameter when the flow rate was kept constant ( $10 \mathrm{~L} / \mathrm{min}$ ). The reduced ID from 16 to 9 mm increased the surface area per volume of solution and the amount of scale deposition increased from $8 \mathrm{~g} / \mathrm{m}$ to $13 \mathrm{~g} / \mathrm{m}$ within

Tekna $23^{\text {rd }}$ International Oil Field Chemistry Symposium, 18-21 March 2012, Geilo, Norway
the first meter. Along the vertical tube the scale thickness determined by cutting and weighing declined as the supersaturation of the solution decreased.

Slightly more than twice the amount of scale was formed when the duration of the experiment was doubled. This is in agreement with the present observations of $\mathrm{CaCO}_{3}$ scale and of sulphate scale experiments performed by Rollheim (1992); once a layer of scale is formed (pre-scale) the deposition accelerates and crystal growth is dominating the precipitation process. When the inner diameter of the tubing decreases, more scale is deposited. In the non-linear parts of the tube more deposition was found, probably as an effect of the turbulent flow, forming eddies along the line.

The crystal layer formed on the tubing wall at turbulent flow appeared to be soft and like powder and was easy to remove from the steel surface. Scanning Electron Microscope (SEM) analysis performed showed heavy growth out from the wall by a pseudo-rosette crystal shape. There was no significant difference in average size of the crystals along the tube length. Large crystals ( $5-10 \mu \mathrm{~m}$ ) were surrounded by smaller ones ( $1-3 \mu \mathrm{~m}$ ), Fig. 3. Each crystal seemed to be rounded at the edges and not showing the stringent rosette like shape that typically forms under laminar flow. This may be the effect of erosion by the fluid flow. The mineral composition of the scale was determined by X-ray Diffraction (XRD) to approximately $85 \% \mathrm{BaSO}_{4}, 9 \% \mathrm{SrSO}_{4}$ and $6 \% \mathrm{CaCO}_{3}$ with the highest $\mathrm{SrSO}_{4}$ content at the T-junction (ID 16 mm ). This could support the theory of precipitation reaction occurring more slowly (slower kinetics for $\mathrm{SrSO}_{4}$ ) at the high diameter part of the tube.

A time delaying 316 SS tank ( 486 l, ID 80 cm , with horizontal layers inside) with residence time of total 48 minutes was implemented between the first 6 m and the last 3 m of the test tube. The flow velocity decreased significantly inside the tank to 3-6 min time volume at constant flow rate of $10 \mathrm{~L} / \mathrm{min}$.

The dynamic tests under turbulent flow showed that DETPMP (penta-phosphonate scale inhibitor) prevented $(\mathrm{Ba} / \mathrm{Sr}) \mathrm{SO}_{4}$ scale to form on the tubing walls at a dosage of 10 ppm active, however, particles would still form in the bulk solution and settle at the horizontal layers inside the time delaying tank. The threshold concentration was found to be between 7.5 and 10 ppm active in turbulent flow. No scale was found growing on the walls of the tank when inhibitor was present above this concentration. DETPMP scale inhibitor present will change the crystal morphology of the $(\mathrm{Ba} / \mathrm{Sr}) \mathrm{SO}_{4}$ crystals settled at the bottom inlet of the tank. An example of a system with 10 ppm DETPMP present is shown in the SEM micrograph in Fig. 4. The particles could accumulate at dead ends or be transported out with the flow depending on the flow regime of the system.

The efficiency of the DETPMP scale inhibitor found at turbulent flow was compared to standard capillary dynamic tube blocking tests (DTB) at laminar flow (flow rate $480 \mathrm{ml} / \mathrm{h}$, ID 0.75 mm , length 1.5 m ). At the same pH (7.8) MIC was lower at laminar flow (MIC 5-7 ppm active) than at turbulent flow (MIC 10 ppm active). The results could indicate that more inhibitor might be needed in the field where the flow is turbulent. However, the laboratory experiments do not consider presence of particles, oil, gas etc. that might influence on the inhibitor performance. The behaviour of the well or the process system in question should always be carefully monitored to identify the scaling potential and to guide the chemical management.

Fig. 3 SEM micrograph of (Ba/Sr) $\mathrm{SO}_{4}$ scale deposition in the vertical tube at turbulent flow, $\mathrm{Re}=33600$.

Fig. 4 SEM micrograph of ( $\mathrm{Ba} / \mathrm{Sr}) \mathrm{SO}_{4}$ scale deposition settled in the bottom inlet of the tank at turbulent flow with 10 ppm DETPMP present.

Hydrodynamic simulations

As an aid to designing experimental work as well as evaluating which mechanisms might be involved in the increased scaling observed around cross section changes, Larsen and Krampa ( 2009) performed CFD (Computational Flow Dynamics) simulations of water flow over forward and backward facing steps (FFS and BFS, respectively) in a 5 inch pipe and flow over backward facing steps in 2.5 and 0.5 inch pipes.

The objective was to predict flow parameters and patterns around a safety valve like geometry in real wells as well as in a laboratory setup. The following conclusions were drawn (Larsen and Krampa 2009):

• Significant recirculation occurs in the flow behind and in the region of the backward facing step. Recirculation also occurs upstream of a forward facing step, but in a smaller region. These recirculation zones may trap particles (including scale particles) formed upstream of this point that can settle and act as nucleation points for precipitation of more scale. Swirling flow can also require a higher MIC than a flow which is not interrupted.
• The reattachment length reflects the size of the recirculation zone and shows reported dependence on both expansion ratio (ER) and Reynolds number.
• While several expansion ratios were not investigated, the dependence of the reattachment length on the expansion ratio appears to be continuous (also reported in the literature).

• The dependence on Reynolds number is evident through geometries with the same expansion ratios as well as those with different expansion ratios.
• Reattachment lengths can be estimated for laboratory set-ups without performing new simulations.
• $\quad$ Significant turbulent energies were only observed in the “realistic cases” where the Reynolds numbers were in the order 125000-500000. In order to reproduce the Reynolds numbers for the “realistic cases” it is necessary to pump $35-140 \mathrm{l} / \mathrm{min}$, even for the smallest diameter studied ( 0.5 inches). With respect to pump rate this can be achieved in the laboratory, but not with respect to the amount of solution that has to be prepared if such an experiment was to be performed over more than a few minutes. In that case one would need to consider a recirculation set-up and not mixing of fresh solutions to obtain the needed water composition to be studied.
• The simulated local pressure changes observed over the steps (i.e. $\mathrm{D}_{\mathrm{b}}-\mathrm{D}_{\mathrm{c}}$ of 5.0-4.5; 2.5-2.0 and 0.5-3/8 inches change in pipe diameter, respectively), should not itself contribute significantly to induce scale formation.
• For fully-developed flow configurations, the effect of the contraction pipe length (i.e. the pipe with the smaller diameter) may not be important. However, sufficient length is required to avoid having a developing flow exiting into the expanded pipe (backward-facing step) region.

Based on the flow simulation work it was decided to build a 0.5 inch flow rig which would enable flow rates up to $2 \mathrm{l} / \mathrm{min}$. This would not create field representative flow velocities and shear forces, see Table 1, but the test rig could still provide very useful input regarding scale deposition mechanisms under turbulent conditions. See Fig. 1.

Table 1 Computational Flow Dynamics cases simulated (Larsen and Krampa 2009).

Fig. 5 Velocity vector close to the backward-facing step for Case 09 and 10. ( 0.5 in . base pipe, $\mathrm{Re} \sim 2170$ and 3617, respectively) (Larsen and Krampa 2009).

Apparatus

Dynamic tube blocking (DTB)

In the capillary dynamic tube blocking (DTB) tests two brines were heated separately under flow through $1 / 8$ inch OD stainless steel tubing coils ( $\sim 3 \mathrm{~m}$ long) coated with poly-tetra-fluoro-ethylene (PTFE). The brines were mixed in a Swagelok T connector directly in front of a 1 m long 316 stainless steel coil ( 1 mm ID, $1 / 16$ inch OD). pH of the mixed solutions was monitored at the outlet outside the oven along with temperature and pressure drop across the coil. Pharmacia P500 HPLC pumps were used. Pressure build-up in the capillary determines the scaling potential and the efficiency of scale inhibitors may be studied as described e.g. by Montgomerie et al. (2008).

Turbulence rig

All the experiments with turbulent flow were performed in a “once-through” dynamic flow rig as shown in Fig. 6. Total flow rate was $0.6 \mathrm{~L} / \mathrm{min}$ or $1.1 \mathrm{~L} / \mathrm{min}$ and the temperature was at $80^{\circ} \mathrm{C}$. The main flow was provided by tap water. The carrier stream was split and calcium chloride brine and sodium bicarbonate brine was fed and mixed separately into the two carrier streams. The two streams are then merged again and mixed before entering the test tubing. All connecting tubing was made from PTFE.

Test tubes of 316L stainless steel were used. The tube parts were threaded to allow splitting and reassembly of the tubes. The assembled tube was mounted vertically with the 50 mm long 17 mm section facing upwards. The 140 mm long 12.7 mm inner diameter section was connected to a 20 cm polyether ether ketone (PEEK) to ensure that flow was laminar when it reached the test tube. The 12.7 mm section was split into shorter sections, enabling separate weighing of pre-scaled and blank tube sections, as well as inserting new sections during an experiment. The lower weight of the smaller individual pieces also enabled use of a more accurate balance. The tubing design is shown in Fig. 7

Fig. 6 Schematic of the experimental setup and the water flow for tube scaling with once-through flow.

Fig. 7 The turbulence rig test tube design. The tube is mounted vertically with the 17 mm part up.
A) 316L (stainless steel) tubing split and threaded to allow easier inspection and separate weighing.
B) Test tube with expansion.
C) 12.7 mm test tube threaded in both ends. One un-scaled tube section would be placed on each side of a pre-scaled section.
D) 12.7 mm section with only inner threading used to connect the steel pipe to the PEEK tube.
E) Assembled test tube. Direction of flow left to right.

Tekna $23^{\text {rd }}$ International Oil Field Chemistry Symposium, 18-21 March 2012, Geilo, Norway

Experimental procedures

The saturation ratio (SR) of the brines studies in this paper were calculated using Multiscale (Kaasa 1998). In tap water, calcium and bicarbonate levels were assumed to be fairly constant. These were measured by titration. It was further assumed that bicarbonate was the only source of alkalinity in the tap water.

Dynamic tube blocking

Capillary dynamic tube blocking (DTB) experiments were performed according to standard method, as described by for example Montgomorie et al. (2008), with the following exceptions:

• Pump rates were varied from $2 \times 60 \mathrm{~mL} /$ hour to $2 \times 499 \mathrm{~mL} /$ hour
• No back pressure was applied in order to keep conditions as similar as possible to the conditions used in the turbulence rig described below.
• Some of the experiments were performed using tap water. Then the separate brines were supersaturated at room temperature, and supersaturation rose as the brines were heated to $80^{\circ} \mathrm{C}$. This may have caused precipitation in the tubing before the two brines were mixed.

Blocking times were found as the average time to achieve a pressure drop increase of 70 mbar.
In order to interpret the effect of flow rate on scaling potential, the effect of flow rate on uninhibited scale was studied. Further, tests were performed with and without polymer and penta-phosphonate scale inhibitors present to determine the MIC of the inhibitors.

Turbulence rig

Experiments with tap water carrier stream were performed with one hour pre-scaling step at $\mathrm{CaCO}_{3}$ SR 50 followed by 5 hours at SR 28. With ion free tap water as carrier stream, 2 hours at SR 58 was followed by 5 hours at SR 30. The tubes were then weighed and most of the scale was dissolved in 0.5 M HCl , leaving specifically chosen parts of the tubing covered with scale. The calcium content of the acid solutions was measured by EDTA titration in some of the experiments. To avoid capillary suction of acid into the scale not to be removed, and to stop foaming and acid exposure due to rapid $\mathrm{CO}_{2}$ development, the scale was soaked thoroughly in water before acid removal.

The flow was then resumed for $\sim 50 \mathrm{~h}$ at SR 16. The experiment was stopped every evening, the tube sections rinsed tap water, dried at $105^{\circ} \mathrm{C}$ overnight and weighed. The tube sections were visually inspected for $\mathrm{CaCO}_{3}$ scale on the “initially” bare steel surfaces, and flow was resumed. Accumulated weight per $\mathrm{cm}^{2}$ was plotted against time, and a least means square (LMS) trend line was added using Excel. The slope of these trend lines represent a linear scale growth rate, and the growth rates in the 12.7 mm section in each experiment were compared to the rates in the 17 mm section.

Three experiments, A, A-2 and D were performed with pre-scale in the areas indicated with orange colour in Fig. 8. Actual scale length and placement was measured after pre scale acid removal and at termination of all experiments. The scale lengths and placements in Fig. 8 and Fig. 9 are typical values, and were decided based upon the CFD simulations (Larsen and Krampa 2009).

To test whether any change in scaling rate was only due to different linear flow rates in different tube diameters, three experiments, B, C and E were also performed with pre-scale as indicated in Fig. 9.

Tekna $23^{\text {rd }}$ International Oil Field Chemistry Symposium, 18-21 March 2012, Geilo, Norway

Fig. 8 Illustration of pre-scale placement in experiments A, A-2, and D. The pre-scaled areas are indicated in orange. The thickness of the scale is greatly exaggerated. Actual thickness was less than 0.1 mm

Fig. 9 Illustration of pre-scale placement in experiments B, C and E. The pre-scaled areas are indicated in orange. The thickness of the scale is greatly exaggerated. Actual thickness was less than 0.1 mm .

Flow rate was measured several times during the experiment, and the pH electrode was calibrated daily at $\sim 80$ ${ }^{\circ} \mathrm{C}$ using Radiometer IUPAC standard buffer solutions.

Experiments with inhibitor were run at slightly higher SR (60). Tap water carrier stream was used for all experiments, and inhibitor was added to both brine solutions. The experiment was stopped every two hours, the tube disassembled, rinsed gently in tap water and dried at $105^{\circ} \mathrm{C}$. The tube sections were then weighed separately. One set of tubing was scaled without inhibitor present and then inhibitor concentration was gradually increased. An experiment with a second set of tubing was started at the inhibitor concentration determined to be MIC using DTB.

Scale free tube sections were inserted at higher inhibitor concentrations to check if blank steel was better protected from scale than already scaled sections.

Results and Discussion

The results from the following experiments were interpreted:

• Six experiments without inhibitor using tap water carrier stream; A, A-2 and D with pre-scale directly after the expansion, illustrated in Fig. 8, and B, C and E with pre-scale 2 cm after the expansion, as indicated in Fig. 9.
• Seven experiments with ion free carrier stream; also with partial pre-scale.
• Finally, experiments with inhibitor present are presented and compared to MIC determination by modified DTB tests.
• Experiments were run at both $0.6 \mathrm{~L} / \mathrm{min}$ and $1.1 \mathrm{~L} / \mathrm{min}$ to examine the effect of increased shear forces on inhibitor performance.

Tekna $23^{\text {rd }}$ International Oil Field Chemistry Symposium, 18-21 March 2012, Geilo, Norway

General observations turbulence rig

The scale formed in all experiments without inhibitor present was hard and stuck very well to the wall. During flow at SR 16 scale mainly formed on pre-scaled areas. Two images of scale formed in the 17 mm tube sections during the experiments are shown in Fig. 10. These images are representative for the scale formed with tap water carrier stream in all the 17 mm tubes. The scale in tap water carrier stream experiments was reddish to brown, either by iron from tap water or dissolved from the heat exchanger, or organic particles from the tap water. Ion analysis of the tap water did not show iron, however.

When using ion free carrier stream, the scale formed was whiter, but still slightly discoloured. The hardness and adhesion to the tube walls seemed similar, but the amount precipitated was smaller.

Introducing penta-phosphate scale inhibitors dramatically changed the scale morphology, exemplified in Fig. 11 .

Fig. 10 Scale formed in the 17 mm tube sections after pre-scale and 50 hours at SR 16 flow rate $0.6 \mathrm{~L} / \mathrm{min}$ and $80^{\circ} \mathrm{C}$. Left: Experiment A. Pre-scale on and directly behind the expansion Right: Experiment B. Pre-scale starting 2 cm after the expansion.

Fig. 11 Tube sections scaled at $80^{\circ} \mathrm{C}$, SR $60,0.6 \mathrm{~L} / \mathrm{min}$ flow rate with inhibitor present.
A) $\quad 17 \mathrm{~mm}$ outlet, no pre-scale exposed to 1 ppm penta-phosphonate
B) $\quad 17 \mathrm{~mm}$ outlet, pre-scaled for $4 \times 2$ hours, then exposed to 1 ppm penta-phosphonate

Tekna $23^{\text {rd }}$ International Oil Field Chemistry Symposium, 18-21 March 2012, Geilo, Norway

Effect of expansion on uninhibited scale growth

Three experiments were performed with pre-scale directly after the expansion as indicated in Fig. 8, and three with pre-scale starting 2 cm after the expansion, see Fig. 9. Accumulated weight divided by the area of pre-scale was plotted as shown in Fig. 12, and the growth rate calculated by linear regression. The scale growth rate per area was greater directly after the expansion than in the 12.7 mm sections. Scale growth was more rapid at this point both during pre-scale at SR $\sim 50$, where precipitation occurs in the whole tube and during the SR 16 flow, where only already formed scale grew. The accumulated amounts at $\mathrm{t}=0$ represents what has precipitated during pre-scale. With pre-scale placed after the re-attachment length, where flow again was laminar, the scaling rate was now lower in the 17 mm section. According to Garcia et al. (2001) it is indicated that the $\mathrm{CaCO}_{3}$ scaling rate in laminar flow increases with increasing flow rates. This is probably because the scale growth is diffusion controlled, and the diffusion layer is thinner at higher flow rates.

Fig. 12 A: Accumulated scale in experiment A with partial pre-scale directly behind the expansion. B: Accumulated scale in experiment $B$ with partial pre-scale starting 2 cm after the expansion. For both: SR 16, $\mathrm{T}=80^{\circ} \mathrm{C}$. The filled symbols represent scale in the 12.7 mm sections, and the open symbols represent scale immediately after the expansion. The lines represent a least mean square (LMS) trend line.

The slopes from the scale accumulation experiments can be found in Fig. 13. As mentioned above, the scaling rate increases directly after the expansion, and this is observed in all three experiments with pre-scale placed according to Fig. 8. When moving the pre-scaled area outside of the region with eddies, a drop in scaling rate is observed in all three experiments.

If we compare the growth rates in the 12.7 mm tube sections in all six experiments, the average rate with standard deviation is $0.124 \pm 0.016 \mathrm{mg} / \mathrm{cm}^{2} \mathrm{~h}$. The relative standard deviation in experiments under “similar conditions” is fairly high. However, since all experiments show the same trend, it looks safe to conclude that the expansion does indeed cause increased growth rates.

Seven experiments with ion free carrier stream were performed. The rates were generally much lower than in tap water, even if care was taken to achieve approximately the same SRs as in tap water experiments. The same trends in scaling rates directly behind an expansion were still observed in ion free carrier streams.

Tekna $23^{\text {rd }}$ International Oil Field Chemistry Symposium, 18-21 March 2012, Geilo, Norway

Fig. 13 Slopes from LMS scale growth data ( $\mathrm{mg} / \mathrm{cm} 2 * \mathrm{~h}$ ) in 12.7 and 17 mm tube sections, with pre-scale placed as indicated in the figure.

Dynamic tube blocking

In order to interpret the effect of flow ate on scaling potential at laminar flow, the effect of flow rate on uninhibited scale was studied. Blocking time in an uninhibited solution without tap water present was determined at SR $\sim 40$ at $80^{\circ} \mathrm{C}$. Flow rates from $2 \times 60 \mathrm{~mL} / \mathrm{h}$ to $2 \times 499 \mathrm{~mL} / \mathrm{h}$ were tested. The results indicate that increased flow rate increases the scaling rate, see Fig. 14. This is related to three factors: Higher flow rate means the diffusion boundary layer near the pipe walls is thinner and the kinetics of crystal growth therefore higher. Also, the shorter transition time of the solution from the entering point of the capillary tube to the end means that the concentration of the solution leaving the capillary tube will be higher than at low flow rates, resulting in higher growth at the coil outlet. Finally, if the flow rate is doubled twice as many calcium and carbonate ions, and thus potential $\mathrm{CaCO}_{3}$ scale passes through the tubing per time unit. On the other hand, the shear forces will be higher, thus removing loose particles more easily.

Polymer scale inhibitor

Three experiments with 8 ppm polymer scale inhibitor and SR 60 in tap water at $2 \times 499 \mathrm{~mL} /$ hour are compared with runs at $2 \times 300 \mathrm{~mL} /$ hour in Fig. 14. The vertical lines represent average uninhibited blocking time under similar conditions. At $2 \times 499 \mathrm{~mL} /$ hour 8 ppm polymer scale inhibitor average blocking time was 4.8 times higher than uninhibited blocking time. At $2 \times 300 \mathrm{~mL} / \mathrm{h}$ blocking time was 3.6 times higher. At 9 ppm blocking was not observed even at 10 times the blank blocking time. The MIC of the polymer scale inhibitor to inhibit $\mathrm{CaCO}_{3}$ at SR 60 at $80^{\circ} \mathrm{C}$ and laminar flow was interpreted to be 8 ppm .

Tekna $23^{\text {rd }}$ International Oil Field Chemistry Symposium, 18-21 March 2012, Geilo, Norway

Fig. 14 The effect of pumping rate on pressure drop vs. time at $80^{\circ} \mathrm{C}$, SR 60, tap water present. 8 ppm polymer scale inhibitor. Red: $2 \times 300 \mathrm{~mL} / \mathrm{h}$. Blue: $2 \times 499 \mathrm{~mL} / \mathrm{h}$. The vertical dotted line represents average blocking time without inhibitor at $2 \times 499 \mathrm{~mL} / \mathrm{h}$. The solid vertical line represents average blocking time at $2 \times 300 \mathrm{~mL} / \mathrm{h}$.

Penta-phosphonate

All dynamic tube blocking tests (DTB) with penta-phosphonate were run without the presence of tap water. At 1 ppm no blocking occurred at $2 \times 60 \mathrm{~mL} /$ hour. At $2 \times 499 \mathrm{~mL} /$ hour blocking occurred at 1.6 times blank blocking time. At 2 ppm at $2 \times 499 \mathrm{~mL} /$ hour blocking time increased to 12.4 times blank. The MIC of the polymer scale inhibitor to inhibit $\mathrm{CaCO}_{3}$ at SR 60 at $80^{\circ} \mathrm{C}$ and laminar flow was interpreted to be 2 ppm .

Inhibited scale growth in turbulence rig

Two inhibitors were tested; the polymer inhibitor and the penta-phosphonate studied in the DTB tests. A comparison of the appearance of the scale formed in the turbulence rig experiments is shown in Fig. 15.

Both inhibitors change the morphology significantly, and the phosphonate most dramatically. This is in agreement with the founding’s of Ramstad et al. (1994), Fig. 4. Adherence to the walls was also lowest with 1 ppm penta-phosphonate, where forceful knocking loosened a lot of particles. In X-ray diffraction (XRD) these loose particles were identified as mainly vaterite with small amounts of calcite. The particles still on the wall contained significantly more calcite, but were still predominantly vaterite. All samples analysed from scale formed with the polymer scale inhibitor were identified as calcite, even the needle-like crystals seen in Fig. 17, which suggests aragonite. It is possible that aragonite forms initially but rapidly transforms to calcite while marinating the macroscopic shape of aragonite needles.

If scale formed is as loosely attached as that shown in Fig. 15F, it would be expected that it would be washed out under more turbulent flow with higher shear forces. In dead-waters, this might not happen, but any neck forming under such conditions would be washed away as soon as it grew large enough to be exposed to shear forces. The polymer inhibitor, which does seem very efficient at lowering growth rate, but has less impact on morphology might form necks that can withstand these shear forces. This could indicate that problematic neck forming in dead-waters could be more likely with inhibitors that do not affect scale morphology significantly.

Tekna $23^{\text {rd }}$ International Oil Field Chemistry Symposium, 18-21 March 2012, Geilo, Norway

Fig. 15 Tube sections scaled at $80^{\circ} \mathrm{C}$, SR 60, $0.6 \mathrm{~L} / \mathrm{min}$ flow rate with inhibitor present.
A) $\quad 12.7 \mathrm{~mm}$ section exposed to $3 \times 1$ hour pre-scale without inhibitor present, then with increasing concentrations of the polymer scale inhibitor
B) $\quad 12.7 \mathrm{~mm}$ section exposed to increasing concentrations of polymer scale inhibitor without pre-scale
C) Expansion from 12.7 to 17 mm exposed to polymer scale inhibitor without pre-scale
D) $\quad 17 \mathrm{~mm}$ outlet, no pre-scale exposed to 1 ppm penta-phosphonate
E) $\quad 17 \mathrm{~mm}$ outlet, pre-scaled for $4 \times 2$ hours, then exposed to 1 ppm penta-phosphonate
F) $\quad 12.7$ to 17 mm expansion seen from 12.7 mm end, exposed to 1 ppm penta-phosphonate without pre-scale
G) $\quad 12.7$ to 17 mm expansion seen from 12.7 mm end, exposed to 1 ppm penta-phosphonate after $2 \times 4$ hours of pre-scale

Polymer scale inhibitor

Two experiments with turbulent flow were run in parallel: The first with a $3 \times 1$ hour pre-scaling phase without any inhibitor present, and the second experiment with inhibitor concentration of starting at 8 ppm (MIC found from DTB) and increasing to 80 ppm . Even at $\mathbf{8 0} \mathbf{~ p p m}$, already formed scale grows. The growth rate was, however reduced by more than $90 \%$ at this high concentration. Also it should be noted that the laboratory chosen SR 60 for scale growth is very high for calcium carbonate scale, and as such may not be representative for well conditions. It also seems that scale does not form readily on blank steel at $\mathbf{8 0} \mathbf{~ p p m}$. The results from the growth rate experiments are illustrated in Fig. 16 .

Penta-phosphonate

The scale growth rates were determined at SR 60 and $80^{\circ} \mathrm{C}$ using penta-phosphonate at 0,1 and 2 ppm respectively. The experiments at 1 and 2 ppm dosage showed almost the same growth rate independently whether the tube had been pre-scaled or not. The growth rate was higher during the pre-scaling.

Photographs of the scale formed in presence of the various inhibitors are shown in Fig. 17.

Fig. 16 Comparison of growth rates in $\mathrm{mg} / \mathrm{cm}^{2}$-hour at given polymer scale inhibitor concentrations, SR $\mathbf{6 0}, \mathbf{8 0}{ }^{\circ} \mathbf{C} . \mathbf{1 2 . 7} \mathbf{~ m m}$ values are calculated as the average of three tube sections, and the error bars represent the maximum deviation from the average value. Standard deviation is indicated on each bar.

Effect of increased flow rate in turbulence rig

The effect of increased flow rate was studied on the scale growth rate with the polymer inhibitor present (SR 60 and $80^{\circ} \mathrm{C}$ ). The results indicate that the efficiency of the polymer inhibitor is reduced significantly when rate is increased from $0.6 \mathrm{~L} / \mathrm{min}$ to $1.1 \mathrm{~L} / \mathrm{min}$ when applying the MIC of 8 ppm determined by the DTB. It should be noted that the uninhibited growth rate would also be increased with increased flow rate. The results indicate that there is a threshold inhibitor concentration (between 40 and 80 ppm ) for the polymer inhibitor where increased flow rate does not affect growth rate, Fig. 18.

The effect of increased flow on the scale growth rate using the phosphonate inhibitor is shown in Fig. 18. At 1 ppm the growth rates start out quite similar regardless of flow. After six hours the growth rate in the $1.1 \mathrm{~L} / \mathrm{min}$ experiments starts to increase. After ten hour, the growth rate at $1.1 \mathrm{~L} / \mathrm{min}$ was more than double the rate at 0.6 $\mathrm{L} / \mathrm{min}$. When the inhibitor concentration was increased to 2 ppm , however, growth completely stopped, and loose particles were lost during the experiment. In the polymer inhibitor, the same low growth rate was achieved at both flow rates. It is believed this tendency will increase with even higher shear forces.

For both inhibitors, the effect of flow rate is lower in the 17 mm tube section than in the 12.7 mm tube section.

Tekna $23^{\text {rd }}$ International Oil Field Chemistry Symposium, 18-21 March 2012, Geilo, Norway

Fig. 17 Scale growth in turbulence rig at $80^{\circ} \mathrm{C}$ and SR 60. Flow rate $1.1 \mathrm{~L} / \mathrm{min}$. A) 14 hours with 8 ppm Polymer scale inhibitor. B) 26 hours, $40-80 \mathrm{ppm}$ Polymer scale inhibitor. C) 10 hours with 8 ppm Polymer scale inhibitor. D) 22 hours 1-2 ppm penta-phosphonate.

Fig. 18 Comparison of growth rates in $\mathrm{mg} / \mathrm{cm}^{2}$-hour at given polymer scale inhibitor concentrations, SR $60,80^{\circ} \mathrm{C}$, at 0.6 and $1.1 \mathrm{~L} / \mathrm{min}$ flow rate. 12.7 mm values are calculated as the average of three tube sections, and the error bars represent the maximum deviation from the average value.

Fig. 19 Comparison of growth rates in $\mathrm{mg} / \mathrm{cm}^{2}$-hour at given penta-phosphonate inhibitor concentrations, SR $60,80^{\circ} \mathrm{C}$, at 0.6 and $1.1 \mathrm{~L} / \mathrm{min}$ flow rate. 12.7 mm values are calculated as the average of three tube sections, and the error bars represent the maximum deviation from the average value.

Tekna $23^{\text {rd }}$ International Oil Field Chemistry Symposium, 18-21 March 2012, Geilo, Norway

What have we learned so far?

Which of the mechanisms suggested look most reasonable where increased precipitation is observed near expansions and constrictions?

Hypotheses revisited

1. Scale forms in bulk regardless of inhibitor concentrations exceeding MIC

a. Particles settle more easily in dead-water and eddies area in the flow, allowing crystal growth and re-crystallization.
This seems a very likely contribution to the effects we see in the laboratory experiments, especially in the pre-scale stage, where the SR is high enough to cause precipitation in the whole tube section.
b. Particles loosely attached to the wall are not removed by shear forces in wakes. This would enhance the effect of hypothesis 1a. Settled particles remain settled and are allowed to grow. This is considered likely based on the experiments performed.
c. Particle-wall interactions are increased around constrictions and expansions This cannot be ruled out or confirmed by these experiments.
2. Scaling rate increases under turbulent conditions
a. Scaling rates increase under turbulent multiphase flow

The scaling rate increases with increasing flow in the current experiments. The highest scaling rates are found directly behind an expansion, where the back currents are most developed. CFD simulations (Larsen and Krampa 2009) also indicate that with the flow rates achievable in the laboratory, the turbulent energy should be too low to influence crystallization rates.
3. Inhibitor efficiency is lowered by turbulence

This was not confirmed by the experiments performed, since the shear forces applied were not representative to shear forces that may apply in the field. At the shear forces in the turbulence test rig it was not observed that the scale inhibitor efficiency was reduced at increasing degree of turbulence.
a. Protective layers on surfaces are disrupted by high shear forces or bursts (local violent collapse of the main flow regime in multi-phase flow)
Not assessed.
b. Protective layers on particles or nuclei are disrupted by bursts or shear forces Not assessed.
c. The inhibitor is destroyed by turbulent conditions Not assessed
d. Residence time for parts of the flow exceeds the inhibitor retardation time in wakes and eddies
This cannot be ruled out, but it is difficult to separate this effect from sedimentation and lower shear forces to remove loosely attached particles.

Conclusion

• Scale has a preference to deposit at positions in production wells where irregularities in the flow pattern occur, such as inflow control devices, gas lift mandrels and downhole safety valves. This results in increased operational risk and economical losses.
• A laboratory once-through rig has been developed. It has been experimentally proven that an expansion in the tubing causing back currents increases the $\mathrm{CaCO}_{3}$ scale deposition rate…
• A likely mechanism is increased residence time causing settling of nuclei for scale growth and decreased erosion of scale deposits.
• The performance of two generic scale inhibitors, a polymer inhibitor and apenta-phosphonate, have been tested under turbulent conditions in the flow rig. Neither of the inhibitors completely stopped the scale deposition at $0.6 \mathrm{~L} / \mathrm{min}$ flow rate, even at concentrations in the range $40-80 \mathrm{ppm}$.
• At $1.1 \mathrm{~L} / \mathrm{min}$, scale growth stopped completely at 2 ppm penta-phosphonate inhibitor. With the polymer inhibitor similar low growth rates were achieved at both flow rates. It should be noted that the fluid systems tested were highly supersaturated with regards to $\mathrm{CaCO}_{3}(\mathrm{SR}=60)$.
• Both inhibitors had an impact on the morphology of the precipitated $\mathrm{CaCO}_{3}$. The penta-phosphonate promoted vaterite, whereas the polymer inhibitor promoted large, needle-like calcite crystals.
• The penta-phosphonate had a much larger impact on the morphology than the polymeric inhibitor. This is important since calcite is much stronger attached to a steel surface than vaterite. Consequently, the penta-phosphonate could have a dual effect: Amount of scale formed is reduced, and in addition to this it makes the scale easier to remove by shear forces set up by the fluid flow.
• Based on the above, it is recommended to test scale inhibitors systematically with regards to performance under turbulent conditions during qualification for field application.
• Some of the questions regarding deposition mechanisms in turbulent parts of the production systems still remain unanswered. It is recommended to continue the work by performing additional tests in e.g. a rotating cylinder setup, which would also allow for systematic screening of the available scale inhibitor chemistries and their performance under turbulent conditions.

References

Cleaver, J. W. and B. Yates (1976). “The effect of re-entrainment on particle deposition.” Chemical Engineering Science 31(2): 147-151.
Garcia, C., G. Courbin, et al. (2001). “Study of the scale inhibition by HEDP in a channel flow cell using a quartz crystal microbalance.” Electrochimica Acta 46(7): 973-985.
Gustavsen, Ø., Selle. O.M., et al. (2010). “Inflow-Control-Device Completion in a Scaling Environment: Findings and Experiences Obtained During Production Logging in the Heidrun Field”, SPE 134994, SPE Annual Technical Conference and Exibition, Florence, Italy.
Hasson, D., A. Drak, et al. (2003). “Induction times induced in an RO system by antiscalants delaying CaSO4 precipitation.” Desalination 157(1-3): 193-207.
Hawthorn, D. (2009). “Heat transfer: Solving scaling problems at the design stage.” Chemical Engineering Research and Design 87(2): 193-199.
Jøsang, L. O. and T. Østvold (2010). Scale under Turbulent Flow: Status report June 2010. Trondheim, NTNU: 5 .
Kaasa, B. (1998). Prediction of pH, Mineral Precipitations and multiphase equilibria during oil recovery, Fakultet for naturvitenskap og teknologi.

Tekna $23^{\text {rd }}$ International Oil Field Chemistry Symposium, 18-21 March 2012, Geilo, Norway
Larsen, T. and F. Krampa (2009). Influence of turbulence on scaling - A preliminary study. Influence of turbulence on scaling. Trondheim, SINTEF Petroleum Research: 40.
Montgomorie, H., Chen, P. Hagen, T., Vikane, O., Matheson, R., Leirvik, V, Frøytlog, C and Sæten, J.O. (2008) “Development of a New Polymer Inhibitor Chemistry for Downhole Squeeze Applications”. Paper SPE 113926 presented at the SPE International Oilfield Scale Conference, 28-29 May 2008, Aberdeen, UK.
Morizot, A., A. Neville, et al. (1999). “Studies of the deposition of $\mathrm{CaCO}_{3}$ on a stainless steel surface by a novel electrochemical technique.” Journal of Crystal Growth 198-199(PART I): 738-743.
Payne, G.E. (1987). “A History of Downhole Scale Inhibition by Squeeze Treatments on the Murchison Platform”, SPE 16539, Offshore Europe, Aberdeen, United Kingdom.
Petroleumstilsynet (2011). “Risikonivå i petroleumsvirksomheten, Hovedrapport, utviklingstrekk 2010, norsk sokkel”, report, Norway, 215 p.
Pitzer, K. S. (1995). Thermodynamics. New York, McGraw-Hill.
Quddus, A. and I. M. Allam (2000). “BaSO4 scale deposition on stainless steel.” Desalination 127(3): 219-224.
Ramstad, K., Thorsen, H. & Tydal, T. (1994) “Sulphate Scale Formation Under Turbulent Flow”, internal report, Norsk Hydro a.s, Bergen.
Rollheim, M. (1992) «Scale formation During Oil recovery - A Kinetic Method», Ph. D Thesis, NTH.
Tailby, R.J., Ben Amor, C., et al. (1999). “Scale Removal From the Recesses of Side-Pocket Mandrels”, SPE 54477, SPE/ICoTA Coiled Tubing Roundtable, Houston, Texas.

Acknowledgement

The authors want to acknowledge Statoil ASA for permission to publish this work.