Monitoring the borehole condition during the ongoing drilling process is essential for a successful and safe drilling operation. The drilling industry has developed monitoring systems, which offer the possibility to analyze torque and drag effects in real time and typically calculate a borehole pseudo friction factor. The paper presents a new approach of analyzing recorded drilling data by directly using hook load measurements. The target of the analysis is to minimize the time spent on ream and wash operations. Furthermore the proposed method can be used as an indicator of developing borehole problems at an early stage before they become critical to the operation. The method is based on the comparison of the hook load during pick-up and slack-off movement of the drill string while reaming one stand of drill pipe prior to making a connection. A characteristic signature is generated, which is then interpreted automatically. The result shows whether the well bore conditions allow to resume operations or indicate continued reaming, thus optimizing the number of reaming trips and subsequently reducing the total amount of non-productive time. Based on the comparison of ream and wash sequences of subsequent drill pipe stands, it is possible to generate an automatic well bore condition log, which shows well bore sections with abnormal behavior. The paper shows that it is possible to identify non-problematic well bore conditions and to avoid unnecessary ream and wash operations. A software package offers the possibility to post analyze recorded data as well as the evaluation of real-time data. A case study shows the technical and economic potential of this new approach. Introduction During the drilling process it is necessary to obtain immediate information on current downhole conditions in order to take remedial actions in case of an existing or developing drilling problem. Therefore online techniques have been developed, which continuously monitor drilling conditions and evaluate drilling performance. Considering the trend to record more and more different data volumes at ever-higher frequencies, it is necessary to find methods to reduce and condense the amount of data presented, without loss of important information. Furthermore it must be the target to avoid analysis primarily based on manual (i.e. visual) interpretation of a multitude of log type curves. Previous investigations1 tried to calculate a friction factor log with surface measured data in real-time, which was used for identifying hole cleaning problems, stuck pipe, differential sticking, formation change and mud lubrication problems. The authors defined different type curves of hookload patterns comparing tripping out hookload data against these different type curves. At a second stage a so-called pseudo friction factor analysis was performed, if needed. The current paper is an evolution of this concept, automating the analysis by utilizing rule-based algorithms in combination with statistical analysis in a hybrid system. The preprocessing of the recorded data, with the recognition of different operations patterns with the intention to reduce the amount of data to be stored, is used as a means to condense the amount of information presented to the user to a manageable volume2. The objective of the proposed analysis is to reduce ream and wash time during drilling as well as increase drilling performance by timely detection of onsetting downhole problems. The ream/wash-induced non-productive time, which due to its short duration is normally not listed explicitly in the daily drilling operations report, is typically lumped into the drilling time, such as connection time, and therefore represents hidden non-productive time. The basic concept of this work is to recognize such operations automatically and to obtain information about changes of the drilling environment focusing on the time-based ream and wash sequences after a stand has been drilled down. Typically multiple up and downward movements of the drill string over the entire length of the stand are performed. The hookload patterns during these sequences represent a kind of "borehole log", and they have been subject to further investigations.

No abstract