Reduced Coated-Tubing Failures Cut Costs at Dickinson Heath Sand Unit

Abstract

Summary In the Dickinson Heath Sand Unit waterflood, corrosion-related failures ofcoated injection tubing began accelerating in 1979 with the increased waterproduction attributed to waterflood response. The trend peaked in 1984 when$164,000 was spent pulling failed tubing (34 workovers). Starting in 1983, theselection of coatings, coating-application-process quality control, andhandling/installation procedures were examined, modified, and refined. Thesechanges and improvements resulted in fewer pulling jobs and less money spent onassociated workovers during 1988. Introduction The Dickinson Heath Sand Unit is near Dickinson, ND, in the southern sectionof the Williston basin. Production is from the Mississippian Heath formation at8,000 ft [2438 m], and bottomhole temperature is about 200F [93C]. Crude isprimarily asphaltic, with a specific gravity of 0.855 g/cm3, and the formationwater was saturated brine before waterflooding (Table 1). The field wasunitized for secondary recovery in 1973. At that time, the produced watervolume was insufficient to sustain the waterflood, so Dakota formation watersupplemented the daily production. Even though the initial injection water wasof lower salinity than the formation water, the entire waterflood system wasdesigned for severe service in anticipation of flood response. All vessels, tanks, injection lines, and tubing strings were protected with an internalcoating or lining. As expected, the produced water volume began increasing(Fig. 1), and subsequently, the total dissolved solids (TDS) of the injectedfluid began to rise. The produced Heath formation water is corrosive because ofits salinity (290,000 mg/L TDS) and high dissolved CO2 content (450 ppm). Evenwhen the Dakota water is added, the resulting fluid remains corrosive. In fact, corrosion rates increase in the waterflood because of oil carryover, solids, and scale deposition from mixing the waters and the occasional oxygencontamination typical in waterfloods. There is also a slight problem withsulfate-reducing bacteria in the injection system, but very little problem isevident in the producing wells. The corrosion-related failure rate in thewaterflood system accelerated concurrently with the increase in injection-watersalinity. This paper focuses on the various causes for the injection-tubingfailure rates and the steps taken to mitigate those failures successfully. Thispaper does not discuss the chemical treating, water quality, or surfaceequipment conditions. Problems This study considers 18 injection wells with at least 240 tubing jointseach. The tubing is primarily J-55 EUE 8-round 2-3/8 or 2-7/8 in. [6.03 or 7.30cm], but several strings of C-75 and N-80 have been run. All the tubing isinternally coated or lined. Injection occurs under a packer with the annulusfilled with packer fluid. Wellhead pressures are 2,900 psig [20 MPa], and fluidtemperatures range from 70 to 130F [21 to 54C]. Relatively few tubing failureswere recorded during 1973–80. It is believed, however, that during this periodinjection wans were pulled only when absolutely necessary, such as foracidizing or cleanout. Now, injection wells are pulled as soon as a leak isnoted through an increase in annular pressure. Starting in 1979, the tubingfailure rate increased dramatically with the volume of produced water (seeTable 2 and Fig. 1). Eventually, it was determined that several mechanicalfactors combined with the corrosive nature of the system to produce prematurefailures in the coatings. The failures were divided into three areas of thetubing joint (in order of cumulative failures): at the tong or makeup marks, atthe pin ends and collars, and in the body of the tube. Economically, it isdifficult to predict the length of service to payout. An 8,000-ft [2438-m]string of J-55 8-round 2-3/8-in. [6.03-cm] internally coated tubing costs about$40,000 (1989 prices). On the basis of average maintenance costs of$4,400/pulling job and replacement cost of $155/tubing joint (average of threejoints replaced per job), the entire string should be replaced the eighth timeit is pulled for leaks. This does not imply a definite time frame for payout, however, because the frequency of failure, resulting from a combination ofmechanical damage and atmosphere-accelerated corrosion, increases each time thewell is pulled. Therefore, this study uses average time to first failureinstead of payout for comparisons. Before 1980, the average time to first recorded failure of a coated tubingstring was 48 months (Table 2), as compiled from well servicing records. Therelated work-over costs during this time were less than $50,000/yr. JPT P. 350⁁