Layer | Fill | Outline |
---|
Map layers
Theme | Visible | Selectable | Appearance | Zoom Range (now: 0) |
---|
Fill | Stroke |
---|---|
Collaborating Authors
Abstract Understanding and quantifying the uncertainty in petrophysical interpretation inputs and results are increasingly important in subsurface settings as drilling targets and reservoirs are getting harder and more complex to evaluate. To make the right key decision in exploration or field development it is important to quantify the uncertainty of the interpreted property outputs. In our approach we assume the source of uncertainty originates from three sources, measured logs, the parameter settings used for our equation sets and finally also the imperfectness of the applied interpretation model. In-depth studies of the probability distribution of each parameter in every zone in every well are normally too time consuming to be part of standard well interpretation procedures. It would also open for too much subjectivity and inconsistency in how the uncertainty is deemed from case to case, and petrophysicist to petrophysicist. It is also known that the meta data describing the relevant conditions at the downhole sensor at every depth and in every zone is not always fully and perfectly measured or recorded and available to the analyst. One can ask what the exact properties of the mud are (density, temperature, salinity, pad contact, pad tilt, etc.) depth increment for depth increment? Also, what is the sensor really measuring in heterogenous formation, is it at the lamina with m and n = 2 or was it also measuring some of the rock at the other side of the borehole with higher m and n? There will always be some error associated with both the measurements, and the parameters used as input to the petrophysical equation sets. There will even be some error in the models used, as the chosen interpretation model will fit better to some zones and depth intervals than others. To help quantify the resulting uncertainty of the petrophysical interpretations we therefore have built a context-based approach where the interpreter specifies the context for the various key parameters, the input logs, and the applied model. Every context defined will call a predefined probability density functions fitting to the described scenario/situation as chosen by the interpreter. The predefined probability density functions are a product of consensus reached by a board of expert petrophysicists.
- Europe > Norway (0.67)
- North America > United States > Texas (0.46)
- Asia > Middle East > Israel > Mediterranean Sea (0.34)
Abstract Casing shoe setting depths selection, and the consequent determination of geometrical and mechanical characteristics of casings, is a crucial activity that heavily influences well planning and affects both safety and economical aspects of the operation. Accurate well and casing design can significantly reduce drilling costs and risks; for that purpose CASCADE (Computer Aided Support for Casing Design), an expert system to support the choice of shoe depths and the selection of casing characteristics, has been developed. According to well data provided by the user (lithological information, gradients development forecasts, drilling problems in related wells, targets), the system hypothesizes the casing shoe setting depths and corresponding mechanical characteristics. Different hypotheses can be generated by the system, feeding design criteria with new user defined constraints, but the final choice is up to the user. The system runs on a UNIT platform and is based on X-windows and KEE, a LISP-based object-oriented tool. It exploits an interactive, graphical interface; dialog boxes and menus guide the user. The first version of the system is now undergoing field testing. Introduction CASCADE is a decisional support system for shoe depth selection and casing design in exploration and production wells. This expert system is part of the ADIS project (Advanced Drilling Information System), jointly developed by three ENI companies: AGIP, ENIDATA, SAIPEM. The purpose of ADIS is to provide drilling engineering software tools for well planning and drilling operations support both at the office and rig site, to monitor real time operations and to diagnose drilling problems. Its ultimate aim is to provide better operations support and improved accuracy in well planning, at the same time reducing risks and costs. CASCADE is part of the well planning tools developed for ADIS project: other software tools are related to directional well planning, hydraulics modelling, temperature prediction and gradients forecast. Casing shoe depth selection, and the consequent determination of geometrical and mechanical characteristics of casings, is a crucial activity. It heavily influences well planning and affects both safety and economical aspects of drilling operations. Accurate well and casing design can significantly reduce drilling costs and risks. Currently, casing design is carried out according to several requirements: company standards, common practice procedures, drilling engineers' experience and local regulations. Therefore, the choice of the best casing parameters which meet all the requirements and minimize cost is not a trivial task. We chose to address the problem with expert systems technology focusing on the following issues: - an expert system can combine several different skills; - it guarantees easy and modular expandibility with the acquisition of new experience; - moreover, new experience must be taken into account inorder to continuously increase the quality of the results; - available data come from many sources, and all have to be considered when choosing the shoe depths' and casing'sgeometrical and mechanical characteristics; - an expert system can explicitly represent both domain and strategical knowledge that can easily be modified, added or expunged. It can also give the user a feedback on its own behavior, and it can also be used as a training tool. The expert systems approach has already been exploited to solve both the casing design and other drilling engineering problems. P. 127^
- Information Technology > Knowledge Management (0.40)
- Information Technology > Communications > Collaboration (0.40)
Abstract Because of the diversity of problems which can and do arise, diagnosis of formation damage is a complex task requiring considerable practicalexperience and technical knowhow. To qualify as an expert in this field an individual must read, understand and remember findings reported in amyriad of technical publications and be able to apply this knowledge in a logical and systematic way. Recently programs, called Expert or Artificial Intelligence (AI) Systems, have been developed which perform the activities of a qualified expert in a specialized area. To use these programs the"knowledge base" must first be codified and introduced into the computer's memory, after which the Expert System can (approximately) simulate the expert's analyses. This paper presents a progress report on an Expert System developed to analyze formation damage problems. The goal of the development is to demonstrate the practicality of making available to an average engineer an automated formation damage consultant. We conclude that diagnosis of formation damage is a fruitful application of AI. Introduction Formation damage is any barrier in the vicinity of the wellbore within the well completion interval that reduces the natural capacity to produce or inject fluid or gases. If a well'sperformance is below expectations, the cause must be determined in order to select the appropriate corrective measure. This diagnosis uses practicalexperience combined with technical knowhow. We have developed a prototype expert system to perform such a diagnosis. The presentation of the results of this investigation is divided into four segments. First, we briefly survey types of formation damage. Second, we examine how the reasoning process used by humans to diagnose the causes of formation damage can be translated into a computerized expert system which simulates human analyses. Third, we sketch the structure of the prototype's knowledge base and outline how the prototype uses this store of information to reach conclusions. Finally, we support our conclusion that this is a fruitful area in which to apply AI and present recommendations and plans for future work. FORMATION DAMAGE Formation damage, which is broadly and extensively documented, reduces efficiency of production operations. To maximize the productivity/injectivity of wells the cause of any significant damage must be identified and appropriate treatment carried out to reduce the deleterious effects. Given the variety of interactions which can reduce productivity, correct diagnosis is essential to assure proper design of the treatment. This study is concerned only with the diagnosis. P. 559^
Of course, if our list does not contain the right solution, we can always invent it! Flow assurance is a discipline in which innovations are often relatively easy to prototype and transfer to the field and can provide a competitive advantage. So, how do we do all the above in practice? We first have to make sure we deeply understand the behavior of the several relevant substances. The optimum places for dissecting and understanding fluid behavior are laboratories and flow loops.