Raynal, Jean C., SPE-AIME, Societe Nationals des Petroles d'Aquitaine Gstalder, Serge A., Societe Nationale des Petroles d'Aquitaine Sagot, Andre M., Societe Nationale des Petroles d'Aquitaine Muckleroy, J.A.,* Amoco Production Co. The scatter of data in oilfield drilling makes evaluation of results using different bits highly uncertain. Conventional statistical theory is here applied to samples of field results to clarify and quantify evaluation, and recommendations are presented for organizing drillsite tests. Introduction The original text of this paper, which far exceeded the SPE's allowable length, has been condensed by the translator, using American statistical terms and drilling nomenclature. The complete original English text is on file with the SPE. For space reasons, it was necessary to omit the extensive mathematical and statistical development of the original; instead, references are given to standard textbooks where the basis of this development may be found. Although the statistical approach discussed here has been widely employed in agriculture, medicine, pharmacology, chemical analysis, pilot plant operation, and other fields in which tests are made on samples of a product or on changes produced by a cause, this approach of making statistical tests on pairs of experimental results on bits seems to have been largely neglected in drilling. Statement of the Problem Definition of the Problem When in drilling a well a technique or a tool is changed, it is often very difficult afterwards to answer the essential question: Did the change improve the result, or make it worse? Drilling results may be essentially shown in three ways:as footage drilled per hour;as footage drilled per bit; andas the cost per foot drilled, which incorporates the preceding two and is the most important result. The difficulties in judging these results are many and varied, but they always manifest themselves in a loss of time and so of money. If we wish the drilling tests to be proved absolutely, they must be repeated many times; if, on the other hand, we jump to a hasty conclusion, the result can be a bad decision. Either route is costly. Analysis of the Problem There are many variables in drilling, of which the most important are the following: Weight on bit (W) Diameter of bit (D) Type of bitWear on bit when pulled Revolutions per minute (rpm) Flow rate of drilling fluid (Q) Nature and properties of drilling fluid Pressure difference across rock surface ( p) Depth Rock properties Most of these are poorly measured. For the same drilling rig, drilling the same well, it is relatively easy to maintain constant values of W, rpm, and Q. JPT P. 506
Read full abstract