Scale and Deposit Formation in Steam Assisted Gravity Drainage (SAGD) Facilities
Reprint R-1014
By W. Hugh Goodman, Martin R. Godfrey, and Thomas M. Miller, Nalco Comany, Naperville, IL
ABSTRAST Produced water and brackish brackish well water are are the main boiler feedwater sources for the steam assisted gravity drainage oil recovery process. Contaminants in these two water sources differ. Water separated from the produced oil emulsion (produced water) is high in silica and in soluble organic compounds. Brackish well water can be high in hardness ions (calcium and magnesium). The combination of these waters can be unstable and can produce a variety of mineral scales. The primary location of scaling is dependent on the process scheme scheme employed. For systems with extensive extensive pretreatment, deposition is most likely in preheaters and evaporators. For other systems, deposition can occur in the steam generation equipment. In addition to mineral scales, carbon deposition can occur in steam generation equipment. Analytical data on the water sources is discussed, along with the composition of deposits from preheaters, evaporators and steam generation equipment. This can provide provide some insight insight into the processes of formation and strategies for inhibition of those deposits. Strategies for inhibiting scale formation are discussed.
INTRODUCTION Steam assisted gravity drainage (SAGD) facilities inject steam into geologic formations to facilitate the production of heavy heavy hydrocarbon. Northern Alberta Alberta (Canada) is one area where this process has been extensively deployed to produce bitumen (the term used to describe the heavy hydrocarbon from this region). This process uses pairs of horizontal wells – the upper well is used to inject steam and the lower well is used to extract a complex emulsion of bitumen in water. Cyclic steam stimulation (CSS) is a similar process using only a single well. Periods of production alternate with periods of steam injection. This process also produces an oil-water emulsion. Environmental considerations – limited make-up water availability and lack of discharge options –
dictate that the produced water is recycled to make more steam. Treatment is required to remove contaminants and protect steam generation equipment. The extent of this pretreatment depends on the water quality requirements for the chosen steam generation equipment. Even with complete reuse of produced water, some make-up is required to replace steam losses and incomplete return of down-hole steam as well as to accommodate variations in steam production and blowdown rates. Although many facilities facilities are permitted to use higher quality sources – surface water or well water, some sites use water from higher-salinity aquifers. The two common options for steam generation are once through steam generators (OTSGs) and conventional drum boilers. Table 1 compares the typical feedwater specications for these different types of steam generation equipment. Table 1 – 1 – Comparison Comparison of selected feedwater feedwater specications. specications.
Parameter OTSG Hardness (mg/l as CaCO3) 0.2 – 0.2 Conductivity (µS/cm) 2,000 – 10,000 Silica (mg/l as SiO2) 20 – 50 Non-volatile TOC (mg/l C) 200 – 600
Drum Boiler * 150 2 0.2
*not detectable, > ~0.02
The limits for OTSGs are based on a survey of user practices. The limits for drum drum boilers are taken from the ASME guidelines for watertube, drum-type boilers operating at 1001–1500 psig.1 The generally higher tolerance of OTSGs to feedwater contamination arises from the lower concentration factor as compared to conventional boilers and from ease of mechanical cleaning to remove accumulated deposits. OTSGs typically produce 75-80% steam
Presented at the Int ernational Water Conference in San Antonio, Texas, Texas, October 24-28, 2010.
quality, corresponding to a concentration factor of 4-5. OTSGs are designed with long sections of straight tubing bounded by removable end sections. This allows insertion of mechanical devices (“pigs”) that scrape the deposits off of the tube surface. This process is often referred to as pigging.
Table 2 – Range in analytical data for produced water.
Component Ca (mg/l) Mg (mg/l) K (mg/l) Na (mg/l) SiO2 (mg/l) TOC (mg/l) NH3 (mg/l) Cl (mg/l) pH (s.u.) “M” alkalinity (mg/l as CaCO3)
The OTSG option typically includes warm lime softening (WLS) for partial removal of silica and hardness. WLS is typically followed by filtration to reduce suspended solids and ion exchange using weak acid cation exchange (WAC) resins for additional hardness removal. In some cases, hot lime softening is used in place of WLS. Figure 1 shows the WLS process scheme. A second option is to use conventional drum boilers. As compared to OTSGs, these boilers require lower contaminant concentrations in the boiler feed water. The WLS – WAC approach cannot meet the requirements, specically for conductivity (total dissolved solids, TDS) and non-volatile TOC. Wastewater evaporators are typically used for pretreatment. The condensate produced from these evaporators typically meets the water quality requirements for drum boilers. Figure 2 illustrates this scheme.
Minimum 1 1.6 14 130 11 170 11 48 7.3
Maximum 52 14 240 3000 260 430 64 4800 8.8
140
1400
reduce the solubility of calcium carbonate. Likewise, decreasing temperature or any concentrating process would reduce the solubility of silica. The “M” alkalinity measurement presented in this and subsequent tables was measured by titration. 2 This method suffers from the possibility of interference by other species, notably organic acids. It appears likely that for many of the produced water samples the bicarbonate/carbonate concentration is lower than the reported alkalinity.
WATER QUALITY Table 2 illustrates the wide variation in the concentration of selected contaminants in produced water. Examination of these data suggests that the produced water from some facilities may be close to saturation with respect to several species, particularly silica and calcium carbonate. In fact, some of these samples seem over-saturated with calcium carbonate. For example, the concentration of carbonate in equilibrium with 52 mg/l calcium at 100ºC is ~10 mg/l (“M” alkalinity expressed as CaCO3 equivalents).
Table 3 shows similar data for well water. As compared to produced water, well water is typically much lower in silica and TOC. Calcium concentrations cover roughly the same range, but trend toward the higher end of the range.
TREATMENT TARGETS WLS-OTSG OPTION – WLS removes hardness by precipitation of calcium carbonate, dolomite (calcium magnesium carbonate), and magnesium hydroxide. Typically, lime is added to raise pH and promote precipitation of the carbonate species.
Anything that would reduce the solubility of key species could lead to precipitation or scale formation. For example increasing pH, increasing temperature, or any concentrating process (evaporation or boiling) will
Figure 1 – Typical Drum Boiler Pretreatment Scheme.
2
magnesium). These systems typically produce efuent meeting the hardness target for OTSG feed water. The WAC step has no effect on soluble silica.
Table 3 – Range in analytical data for well water.
Component Ca (mg/l) Mg (mg/l) K (mg/l) Na (mg/l) SiO2 (mg/l) TOC (mg/l) NH3 (mg/l) Cl (mg/l) “M” (mg/l as CaCO 3)
Minimum 2.0 1.5 2.2 700 8 nd nm 480 880
Maximum 45 32 250 3700 10 5 nm 5300 1200
EVAPORATOR-DRUM BOILER OPTION – Evaporators are capable of routinely meeting the feedwater quality requirements for a drum-type boiler. These units produce a condensate essentially free of dissolved solids. A purge stream removes contaminants from the system. The purge is typically 2½-5% of the incoming ow; these systems operate at concentra tion ratios (feed ow rate / purge ow rate) of 20-40.
At such high concentration ratios, the contaminants can become insoluble. Typically, caustic soda (NaOH solution) is added to raise the pH and increase silica solubility. This has the undesirable effect of decreasing the solubility of calcium carbonate and of several metal silicates.
nd = not detected nm = not measured
Generally the overow from the WLS is near equi librium with respect to calcium carbonate. Thus, the residual soluble calcium carbonate concentration depends on the carbonate concentration (as well as on the temperature and ionic strength). For systems decient in carbonate, soda ash is added as a source of carbonate. For any system, increasing the carbonate concentration decreases the soluble calcium concentration.
To reduce the overall energy usage, the evaporator condensate can be used to heat the evaporator feed water. This typically uses a plate-and-frame type heat exchanger (shown in Figure 2 as the Preheater). Depending on the composition of the feed water, the rise in temperature across the preheater increases the potential for calcium carbonate deposition.
Typically, the total soluble hardness (calcium and magnesium) is measured rather than the calcium or magnesium alone. This is not problematic since the residual soluble magnesium is usually lower than the calcium (provided that the system pH is in the target region of 10.0 to 10.2). The target for residual hardness is typically around 10 mg/l as CaCO 3.
Some evaporator systems use two evaporators operated in series. The rst evaporator operates at a concentration ratio of 2 – it evaporates roughly onehalf of the water. The second evaporator uses the concentrated brine from the rst evaporator as feed and operates at a concentration ratio of approximately 20. The two evaporators produce roughly the same amount of condensate, but the recirculating brine in the second evaporator is much more concentrated.
Silica removal in the WLS is accomplished through magnesium hydroxide addition to precipitate magnesium silicate. Increasing magnesium dosage decreases the residual silica concentration. Targets for the residual silica concentration range from 20 to 50 mg/l.
SIMILARITIES – While there are many differences between these two approaches to the recycle of produced water to make steam, both systems contain areas where deposit formation is likely. For the WLSOTSG process, the most common location for deposits
WLS is not capable of reaching the calcium concentration target for OTSG feed. The WAC step removes the remaining soluble hardness ions (calcium and
Figure 2 – Typical OTSG Pretreatment Scheme
3
is in the OTSG tubes. For the Evaporator-Drum Boiler option, the most common deposit locations are in the evaporator and the preheater. Minimizing and managing these deposits is critical to the success of both schemes.
Table 4 – Range in analytical data for well water.
Component Ca (mg/l) Mg (mg/l) Ba (mg/l Na (mg/l) K (mg/l) SiO2 (mg/l) TOC (mg/) NH3 (mg/l) Cl (mg/l) pH (s.u.) “M” alkalinity (mg/l as CaCO3)
Both the evaporator and the OTSG benet from the increase in silica solubility with increasing temperature.3 This applies only to silica itself; the solubili ties of the metal silicates decrease with increasing temperature.
DEPOSIT ANALYSES OTSG – Table 5 shows analytical results for deposits from several OTSG systems. These data suggest that several contaminants contribute to the deposit – silica, hardness ions (calcium and magnesium), carbonate ions, and organic compounds (dissolved compounds or TOC; suspended hydrocarbon, or both). Comparing the results from several sites indicates that, while the relative abundance of these components may vary in the deposit, they are all present for all of the sites represented. One possible contributing factor to deposit formation in OTSGs is variation in the performance of upstream treatment equipment.4
Deoiled Water 3.1 <1.5 0.06 570 24 180 220 56 700 8.1
Well Water 24 32 2.8 3,500 24 8 5.4 nm 5,300 8.2
350
880
nm = not measured
The system design anticipated that the fraction of well water needed for make-up to the system would be relatively small, around 2-5% of the total ow. In practice, more well water has been needed to meet targets for steam production. The ScaleSoftPitzer (SSP) program from Rice University5 was used to calculate the scaling tendency of blends of deoiled water and well water. This calculation estimates the saturation ratios (SR) for several potentially scale-forming species (although it does not include calculations for silica or magnesium silicate). The SR is the ratio of the ion product to the equilibrium constant, corrected for ionic strength.
EVAPORATORS – Table 5 also includes the analytical results from deposits in two evaporator systems. Note that the deposit samples were taken from several different places in the system. System 1 uses two evaporators in series. E1 and E2 samples were from the sumps of evaporator 1 (E1) and evaporator 2 (E2) respectively. System 2 is one stage.
The SSP calculation includes a robust correction for ionic strength which is needed due to the high salt concentration in this system. SR = 1 represents the equilibrium condition; SR < 1 is non-scaling; SR > 1 is potentially scaling. SSP actually reports log 10(SR); values here were converted to SR.
The deposits from the preheater portion of System 1 are mainly calcium carbonate, with some magnesium and silica present. During the time that these deposits formed, this site added NaOH to the tank upstream of the preheater to raise the pH to ~9. Additional NaOH was added to the recirculating ow in the rst stage (E1) of the two-stage system.
Figure 3 shows the SR values calculated for calcite (calcium carbonate) without pH adjustment at 75ºC and 90ºC. This corresponds generally to the temperatures in the evaporator feed tank (where the deoiled water and the well water are mixed) and after the preheater, respectively.
The deposits from E1 of the two-stage system are also primarily calcium carbonate, with lesser amounts of magnesium and silica. Deposits from E2 are predominately silica. Note that the carbon content of the deposits from the evaporators is generally lower than that from the OTSGs.
The gure suggests that the deoiled water itself may be over saturation in calcium carbonate. However since some fraction of the TOC may be organic acids that can titrate as alkalinity, the actual carbonate concentration may be a bit lower than the analytical result obtained by titration. Figure 3 illustrates that the scaling tendency for calcium carbonate increases dramatically as the fraction of well water increases. The increased scaling tendency at 90ºC indicates that scaling is likely in the preheater system.
DETAILS ON EVAPORATOR SYSTEM 1 – FEED/PREHEATER SECTION Key analytical results for the deoiled produced water and the brackish well water make-up for Evaporator System 1 are shown in Table 4.
4
Table 5 – Analytical results for deposits from OTSG and evaporator systems.
Si
Fe
Mg
Ca
P
C
% as SIO2 % as Fe 2O3 % as MgO % as CaO % as P 2O5 % as C
Loss 925 Carbonate %
% as CO 2
OTSG Systems
Site 1 Pigging Sample
4
1
1
5
3
70
89
Site 1 Pigging Sample
5
1
1
2
3
74
87
Site 1 Scale on Tube
15
1
8
3
2
49
65
Site 1 Scale on Tube
13
12
8
4
3
45
56
Site 1 Scale on Tube
6
34
2
4
2
32
36
Site 1 Pigging Sample
14
32
8
6
4
28
30
Site 1 Pigging Sample
11
32
8
6
5
33
32
Site 1 Pigging Sample
12
32
8
5
4
37
34
Site 1 Pigging Sample
12
21
7
5
4
42
47
Site 1 Deposit Sample
4
34
8
13
19
13
16
Site 2 Pigging Sample
47
27
5
2
<1
7.4
8
Site 2 Tube Sample
17
57
13
5
<0.5
22
nm
neg
Site 2 Tube Sample
7
55
5
6
6
8.8
20
nm
Site 2 Pigging Sample
41
9
31
1
<0.5
12
nm
neg
Site 2 Pigging Sample
6
59
4
4
<0.5
10
20
neg
OTSG
8
63
3
4
2
7.2
18
neg
OTSG
8
53
6
6
5
10
21
neg
OTSG
25
9
20
1
<0.5
23
40
neg
neg
Evaporator Systems
System 1 Feed Pump
49
1
20
2
1
9.5
25
neg
System 1 Preheater
10
1
6
40
<0.5
nm
41
33
System 1 Preheater
46
1
18
3
<0.5
7.8
29
2
System 1 Preheater
44
1
18
3
<0.5
8.6
31
1
System 1 E1
18
1
7
36
<1
11
36
26
System 1 E1
42
2
15
10
<0.5
10
27
6
System 1 E1
20
1
8
33
<0.5
11
35
24
System 1 E2
50
<0.5
1
12
<0.5
5.4
21
neg
System 1 E2
58
<0.5
1
11
<0.5
3.4
20
3
System 2
53
0.5
2.4
22
mn
nm
7
pos
nm = not measured neg = negative in qualitative carbonate test pos = positive in qualitative carbonate test
5
tion indicate the presence of amorphous silicates and stevensite – Ca0.2Mg2.9Si4O10(OH)2•4H2O. Figure 5 shows the variation in feedwater hardness over several months. The large swings could result from variation in the amount of well water (higher calcium and magnesium concentrations) as well as from the presence of precipitated material in the samples.
Figure 3 – Calcite SR for Deoiled-Well Water Blends at pH = 8.1.
Figure 4 shows the scaling tendency at pH 8.8. Comparing Figure 4 with Figure 3 illustrates the result of caustic addition to the feed tank. Increasing the pH results in an increase in the calcium carbonate scaling tendency both in the feed tank (~75ºC) and in the preheater (~90ºC).
Figure 5 – Baseline Feedwater Hardness.
Scale formation in the pretreatment system led to the need for frequent cleaning. Scale formed in the inlet piping of the feedwater pumps (upstream of the preheater section) as well as in the preheater. This led to the decision to introduce a calcium carbonate scale inhibitor into the feed tank. The inhibitor chosen was a copolymer with good efcacy in prevent ing calcium carbonate scale at high temperatures. This coincided with the introduction of a silica scale inhibitor to reduce silica scaling in E2. Introduction of the scale inhibitor eliminated scaling in the inlet piping of the feedwater pumps. However, the preheater still required regular cleaning. This suggests that the calcium carbonate inhibitor was capable of overcoming the scaling tendency at 75ºC, but not at 90ºC. As mentioned previously, the pH adjustment in the feed tank increases the scaling tendency of calcium carbonate. This adjustment is targeted to increasing the silica solubility in the evaporator. As there should be no tendency for silica scale to form in the preheater (silica solubility increases with increasing temperature), the pH adjustment point could be moved downstream of the preheater without negative impact. This was done and eliminated fouling in the preheater.
Figure 4 – Calcite SR for Deoiled-Wall Water Blends at pH = 8.8.
A deposit sample from the evaporator feed tank showed 40% CaO, 33% carbonate (as CO2), 10% SiO2, and 6% MgO. This is a near stoichiometric ratio of calcium to carbonate (within the precision of the analytical measurement), suggesting that the deposit is predominately (73%) calcium carbonate. Calcite and vaterite phases were identied by x-ray diffraction. Likewise, the molar ratio of magnesium to silica is stoichiometric within the analytical precision. X-ray diffraction results from deposits in the preheater sec6
DETAILS ON EVAPORATOR SYSTEM 1 – EVAPORATOR 1 The changes in the feed system – adding a scale inhibitor and changing the pH adjustment point – stabilized the feedwater hardness, as shown in Figure 6. Although variation was reduced, the amount of hardness entering the evaporator system increased since all of the hardness was now transported through the feed tank, piping, and preheater. This increased the scaling tendency in E1.
Figure 7 – Transport Feedwater to E1 with Antiscalant.
to form magnesium silicate scale would only lead to a small reduction in the silica concentration. So the near 100% transport of silica does not rule out formation of magnesium silicate. Figure 7 indicates that the calcium carbonate inhibitor was not effective in preventing calcium carbonate scale in E1. A deposit sample from E1 showed 32% CaO, 24% carbonate (as CO2), 19% SiO2, and 7% MgO. This is a near stoichiometric ratio of calcium to carbonate within the precision of the analytical measurement, suggesting that the deposit is predominately (56%) calcium carbonate. The molar ratio of magnesium to silica in the deposit is ~½, suggesting that there may be some slight silica deposition in addition to magnesium silicate. X-ray diffraction detected no crystalline phases containing silica.
Figure 6 – Feedwater Hardness with Antiscalant Addition.
While deposit formation is the ultimate indication of scale formation, analysis of the scale requires system shut-down to obtain deposit samples. It is useful to assess the system during operation. One option is to calculate the mass balance (transport) across the evaporators. This requires knowing the input and output ow rates and concentrations of the target ions.
DETAILS ON EVAPORATOR SYSTEM 1 – EVAPORATOR 2
Unfortunately, variations in the operating conditions (ow rates) make it difcult to estimate these values accurately. An alternative is to compare the values for a potentially scale-forming ion to those for an ion not expected to form scale. For a non-scaling ion, dividing the outlet concentration by the inlet concentration allows calculation of the concentration ratio for the system. This is used as a reference point for potentially scaling ions.
Figure 8 shows the baseline (without antiscalant treatment) transport across E2. This suggests that silica, calcium, and magnesium all deposit in the system. A deposit analysis indicated 50% SiO 2, 12% CaO, and 1% MgO. There was no carbonate detected.
For this system, the most suitable reference analyte is boron (B). So: CR = B(brine)/B(feed) %T(Ca) = [Ca(brine)/Ca(feed)] / CR Figure 7 shows the transport across E1 after introduction of the calcium carbonate inhibitor and before the change in NaOH addition point. Silica transport is 100%, within the accuracy of the measurement. The gure suggests that both calcium and magnesium are depositing in E1 Note that since the magnesium concentration is low, loss of ~50% of the magnesium
Figure 8 – Baseline Transport Across Evaporator 2.
7
Figure 9 shows the transport across E2 after introduction of the calcium carbonate and silica inhibitors. As compared to Figure 8, this shows much improved silica transport, although a bit less than 100%. Deposit analysis from this period showed 58% SiO 2, 11% CaO, 1% MgO, and 3% carbonate (as CO2).
Increasing the transport of calcium through the feed and preheater sections increases the driving force for scaling in E1. The calcium carbonate inhibitor did not completely eliminate scale in E1. The evaluation of additional inhibitors is planned. The introduction of a silica inhibitor signicantly reduced scale formation in E2.
ACKNOWLEDGEMENT Thanks to Dr. Jasbir Gill for helpful discussions and for reviewing this manuscript.
REFERENCES 1. “Consensus on Operating Practices for the Control of Feedwater and Boiler Water Chemistry in Modern Industrial Boilers”, American Society of Mechanical Engineers, New York, NY (1994). Figure 9 – Transport Evaporator 1 to Evaporator 2.
CONCLUSIONS A detailed investigation was undertaken into deposit formation in a two-stage evaporator system treating produced water and well water. The mixture of produced water and well water is unstable with respect to the formation of calcium carbonate. This instability is increased by increasing pH and by increasing temperature, both of which were initially present in the feedwater system. Introducing a calcium carbonate inhibitor and moving the pH adjustment point downstream of the preheater was successful in eliminating scale formation in the feed tank and piping and in the preheater.
2. “Standard Methods for the Examination of Water and Wastewater”, 20th Ed., pp 2-26 to 2-28, American Public Health Association, Washington, DC (2008). 3. “The Chemistry of Silica”, R, K. Iler, pp 47-49, John Wiley & Sons, New York, NY (1979). 4. Godfrey, M. R.; IWC 09-36 (2009).
5. ScaleSoftPitzer from Rice University.
Nalco Company 1601 West Diehl Road • Naperville, Illinois 60563-1198 SUBSIDIARIES AND AFFILIATES IN PRINCIPAL LOCATIONS AROUND THE WORLD
© 2011 Nalco Company. Nalco and the logo, are trademarks of Nalco Company. Printed in U.S.A. 3-11