Effect of High Silica Content on Scale Deposition and Pipe-Wall Loss in Oilfield Steam Generators

Khatib, Z.I. (Shell Development Co.) | Olson, E.E. (Shell Development Co.) | Place, M.C. (Shell Development Co.)



Studies were conducted on site in the Coalinga, Belridge, and Midway Sunsetfields in California to research the cause of metal losses detected in theradiant section return bends and immediate piping downstream from the steamgenerators. Surveillance of silica content in the influent and effluent streamsof the selected steam generators and the results of X-ray inspection of bends,elbows, welds, and pipings indicated that (1) a correlation is likely to existbetween the silica and bicarbonate concentration in the feedwater and thesilicate scale buildup, and incident rate of wall loss and (2) the cause ofwall loss/pipe failures is a combination of corrosion and erosion mechanismsaccelerated at higher steam qualities.


Large-scale steamdrives began in the west Coalinga field in 1979. Injectionrates increased from 3,000 B/D of steam (BSPD) (feedwater equivalent) in 1979to 39,000 BSPD currently. Silica content of the steam generator feedwater hasincreased concurrently from 70 to 275 mg/L. The silica content is expected toincrease as more reservoirs are steamflooded and the produced water is recycledcontinuously. Most heavy-oil operators in southern California have experienceda similar increase in silica levels in their generator feedwaters. With littledilution by fresh water and maximum use of recycled produced water, someoperators tended to reduce the steam quality to between 50% and 60% or to lowerthe pressure levels to 500 psig to tolerate the high silica content. TheCoalinga Steam Injection System normally operates at 3,000 BSPD (feedwaterequivalent), 75 % to 80 % steam quality, 950 psig, and 540 degrees F. In 1986,pinhole leaks began to appear bimonthly in the steam generator discharge pipingat the Coalinga field. In 1987, following a leak frequency increase and therupture of two pipes, steam injection was shut in for about 3 weeks forcomprehensive inspection and repairs. Some fittings reportedly had beenreplaced only months earlier, demonstrating how quickly wall loss was occuring.When reinjection was begun, the operating steam quality was reduced to 50% forsafer generator operation until the problem could be defined better. Thisreduction in quality resulted in 20% less heat input to the reservoir for thesame boiler feedwater, with a subsequent decline in oil production. This paperdescribes a study conducted in 1987 in the Coalinga. Belridge, and MidwaySunset fields near Bakersfield. The purpose of this study was to search for thecauses of detected wall loss and frequent pipe failures at Coalinga and tocompare results with those found at the other two fields. This included (1)surveillance of silica content in the influent and effluent streams of selectedsteam generators at a range of steam qualities (50% to 85 %), (2) metallurgicalstudy of the failed sections and identification of major deposits, (3) X-rayinspection of the steam generator's piping systems to locate metal loss andscale buildup, and (4) determination of the influence of piping system andsteam quality on erosion of corrosion films and solids transport.


The produced water from a steamflood generally contains silica. This silicais dissolved from the quartz present in the formation by unvaporized andcondensed water that has a pH greater than 9. Dissolved silica tends to formscales in steam generator tubes if either its concentration is reduced or itssolubility is increased. The API2 has proposed several guidelines for thequality of wet-steam generator feedwater. The guidelines emphasize that afeedwater with hardness less than 1 mg/L (as CaCO3) and silica content nothigher than 150 mg/L in the absence of scaling ions (iron, magnesium, etc.) isnecessary to maintain satisfactory operations. The API also reported that theamount of silica and hardness that can be tolerated in the feedwater of steamboilers decreases as the pressure of the steam generator increases. pressure ofthe steam generator increases. Silica can be removed by coagulation withchemicals such as in cold- or hot-lime softening or demineralization by anionexchange resins. A description of these processes is beyond the scope of thispaper. However, it is worth mentioning that two operators, one in south Texasand one in Peace River (Canada), have reported that the dissolved silicaconcentrations were reduced from 400 to less than 50 mg/L by the hot-limesoftening process.

Field Investigations

Field investigations were carried out in three stages: (1) surveillance ofthe silica content upstream and downstream of selected steam generators, (2)investigation of the effect of steam quality on the precipitation or loss ofsilica in the generator, and (3) X-ray inspection of the piping systems.

Status of the Silica Content and Hardness in Feedwater. The steamfloodproduced water from Coalinga, Belridge, and Midway Sunset fields flows throughdiatomaceous earth (DE) filters and then through a series of softeners beforeit is used for steam generation. An oxygen scavenger, such as catalyzed sodiumsulfite, was added periodically to the de-aerated freshwater streams whenmakeup saw was required at Midway Sunset and Belridge. A sampling technique wasdeveloped to collect representative samples from the influent and effluent ofthe steam generator at various steam qualities up to 85%. Fig. 1 is a schematicof the sampling apparatus and associated equipment. Analytical procedures weremodified to account for H2S interference and the presence of colloidal silicaon silica measurements. Table 1 gives the chemical analyses of the feedwater atthe Coalinga, Midway Sunset, and Belridge fields. The survey of silica content(pre- and postfiltration and softening processes) indicated that there was nodetectable colloidal silica in the feedwater and that the DE filters did notcontribute silica to the overall concentration. The following differencesbetween Coalinga feedwater and those of Belridge and Midway Sunset were found.1. The silica content in the feedwater at Coalinga (254 mg/L) was higher thanthat at Midway Sunset (197 mg/L) or Belridge (142 mg/L). 2. The magnesium andcalcium concentrations in the feedwater at Coalinga were also higher than thelevels detected at Belridge and Midway Sunset. These levels (up to 1 mg/L)correspond to a hardness of up to 4 mg/L CaCO3, which is considered fairly highfor feedwater. 3. The dissolved CO2/HCO3- content in Coalinga feedwater was atleast 10 times higher than in the other two feedwaters. 4. The H2S/sulfide andoxygen content in the Coalinga feedwater was higher than in the other twofeedwaters. 5. The salinity of the feedwater was one-half that of Midway Sunsetand one-third that of Belridge.


P. 357