A-Study of Bulk Cement Handling and Testing Procedures

Abstract

Summary The variation between thickening-time test results from tests on pilotblends and cement-blend sample tests has been a subject of frequent pilotblends and cement-blend sample tests has been a subject of frequentinvestigation. Disagreement between these test results is often blamed onimproper additive proportioning, inadequate blending, or incorrect samplingtechniques. A comprehensive study of blending, sampling, and testing variableswas conducted. This paper describes the results of this research testing with afull-scale laboratory bulk plant. It also includes the results of a south Texasfield study in which thickening time test results from more than 300cement-blend samples were examined. Results of these studies suggest thatsignificant statistical variations may be involved in blending, sampling, andtesting procedures under deep-well conditions. The variabilities are used asguidelines for determining thickening-time acceptance windows. Variabilities oruncertainties can be minimized by optimization and standardization of theseprocedures. Introduction A recurring concern about oilwell cementing involves the effectiveness ofbulk cement blending and handling procedures. This concern is manifested byvariations (sometimes large) between the results of thickening-time testsperformed with pilot blends prepared in the laboratory and results from samplescollected from prepared in the laboratory and results from samples collectedfrom actual blends prepared at field bulk plants. This disagreement can createmany problems. Cement jobs may be delayed for remedial procedures, such asretesting and reblending, adding considerable procedures, such as retesting andreblending, adding considerable expense for operators in lost rig time and forservice companies in extra laboratory time and materials. Previous blending and sampling studies provided a basis for what is nowcurrent practice at pneumatic bulk blending stations. These studies showed thatthe current procedures are basically accurate and well founded. Recentexperience in cementing deeper, hotter wells, however, prompted new concernsabout the precision of blending, sampling, and testing procedures. Thus, manybasic questions about bulk blending, sampling, and testing have beenrestated.What constitutes a "good blend" of a bulk cement composition? Howmuch variation in cement-additive quantities could and should be expected?How many pneumatic transfers are required to blend a bulk cementcomposition properly? Will different compositions require different blendingprocedures?When and how should a cement blend be sampled? How much variation insample composition should be expected as a result of different samplingprocedures?What amount of variation should be expected between thickening-time testsperformed on pilot blends and bulk field-blend samples? How much variationshould be allowed? This paper studies these questions by drawing on results from three areas ofresearch: a frill-scale laboratory bulk plant, laboratory thickening-timerepeatability tests under deep-well conditions, and a south Texas field studyof pilot-/field-blend-sample thickening-time agreement. The results providemeasurements of variations in blending, sampling, and testing procedures inactual field operations. While these variations are significant, the basicaccuracy of cement testing and pneumatic blending is supported by thousands ofsuccessful cementing jobs annually. The success of cementing jobs examined inthe field study further suggests that substantial variations may be tolerable, even where the effects of these variations were compounded by the hightemperatures, pressures, and cement-slurry densities native to south Texas. Background and Theory In deep-well cementing, considerable time and expense are expended in thelaboratory to design a cement formulation with adequate performance propertiesas measured by thickening-time, fluid-loss, performance properties as measuredby thickening-time, fluid-loss, compressive-strength, and other standard tests. As greater casing depths and higher bottomhole temperatures are encountered, cement-slurry design and testing become more difficult. Careful attention mustbe given, however, to cement composition design from initial pilot testingthrough all phases of the cementing process-bulk pilot testing through allphases of the cementing process-bulk cement blending, sampling, mixing, anddisplacement-to achieve a full, competent cement sheath in the desiredinterval. In south Texas, pilot testing may begin weeks before total depth (TD) isreached. Once a slurry that meets the requirements is identified, full-scaleblending may begin. Where well conditions are very demanding, small-scalebatches of approximately 50 sacks may be blended initially to verify agreementbetween the pilot-blend and field-blend thickening times. If the agreement isreasonably good, full scale blending of 50- to 250-sack batch truck blendsbegins. Samples from some or all of the truck blends are commonly tested as aquality-control measure. At all these points in the cement formulating and blending process, thickening time is the best test of whether the cement design process, thickening time is the best test of whether the cement design will meet itsgoal. This standard test has important limitations, how-ever, that should beunderstood. First, the test simulates continuos pumping and does not reflectany effects of planned or unplanned pumping and does not reflect any effects ofplanned or unplanned shutdowns during the cementing job - Second, temperaturearid pressure applied during the test are estimates of actual downhole pressureapplied during the test are estimates of actual downhole conditions. Third, thethickening-time test results have an inherent variability, a limitation that isnot widely acknowledged. Statistical Considerations. Previous examiners recognized significantvariations between repeated thickening-time tests. Calculated standarddeviations, s (expressed as a percent of the means), and variationcoefficients, K, have been used to reflect test repeat-ability or experimentaluncertainty. It is important to realize that these calculated values assumethat the test results are normally distributed and are actually estimates ofthe true values that would b given by a large number of repeat tests. The mean result and s, or similarly mean and K, describe a normaldistribution of test results. In discussions of test repeatability limits ortest variability, 2K, is used because the interval, mean +/ 2K, includes 95.4% of the distribution of results, whereas 1K, or one standard deviation, includes only 68.3% of the results. Field-Study Background. This study was intended to examine the effects onthickening-time agreement of cement composition, the volume of cement blendedin a batch, and the number of pneumatic transfers before ft batch is sampled. Because the variations among these test results would be largely statistical, the intention of the study was to collect a body of data sufficiently largethat meaningful conclusions could be drawn. Cement-blend samples from 20 cementing jobs on nine wells in south Texaswere tested. A variety of cementing procedures and depths were included in thestudy. SPEPE P. 425