Dingzhou Cao, Donald G. Hender, Sam Ariabod, Chris James, Y. Ben, Micheal Lee
This paper provides the technical details to develop a real-time deep learning model to detect and estimate the duration of downlinking sequences of Rotary Steerable Systems (RSS) based on a single measurement (standpipe pressure, SPP). Further analytics are derived based on the downlink recognition results together with other real-time log data (ROP, RPM, Torque, etc.) to drive directional drilling efficiency. Real-time RSS downlink recognition is treated as an image segmentation problem. The Deep Learning (DL) models were created using the dynamic U-Net concept and materialized with a pre-trained ResNet-34 as the underlying architecture. Transfer learning was used due to the limited number of training samples (≪ 100 downlinks per onshore well) to help with speed and accuracy. The SPP time series data was segmented based on stand of pipe drilled (one image per stand). This "image" was then fed into the model for downlink recognition. To further increase the accuracy, a second opinion mechanism was applied when the models were tested and deployed into the Real-Time Drilling (RTD) system. Using a dual model approach greatly reduced the number of false positives due to non-downlink pressure fluctuations causing "noise". The patterns of SPP and its rate of change (delta SPP) are quite different. They both have pros and cons for identifying the downlink, thus two independent models were built based on these two signals. The DL model A is trained based on the original SPP signal and the DL model B is trained based on delta SPP. A downlink is confirmed only when both models show positive results. Data of 10 onshore wells (2 rigs) drilled with RSS were segmented (8165 images in total) and labeled. There were 671 images with 795 downlinks and 7980 images without downlink. The five-fold cross-validation technique was used to identify the best model(s). The F1 score of blind test result was .991 (accuracy was ~99.82%, see Table 2). The relative error of duration estimation is 2.49%. The current rig fleet within the RTD system has a mix of drilling tool configurations - RSS and mud motors. To further validate the models’ robustness regarding drilling tools, additional tests were conducted using mud motor wells’ datasets from 21 rigs (25431 images without downlink). There were 3 false negatives from this extended test set, which resulted in a ~99.93% accuracy for the aggregated 31 wells dataset. These results suggest that the models are accurate, reliable and robust. The real-time DL solution presented in this paper enables operators to analyze RSS performance during and between downlinking events. This would allow drilling engineers and rig supervisors to make faster, more reliable data-driven decisions to optimize performance and directional control of the well path.
{"title":"The Development and Application of Real-Time Deep Learning Models to Drive Directional Drilling Efficiency","authors":"Dingzhou Cao, Donald G. Hender, Sam Ariabod, Chris James, Y. Ben, Micheal Lee","doi":"10.2118/199584-ms","DOIUrl":"https://doi.org/10.2118/199584-ms","url":null,"abstract":"\u0000 This paper provides the technical details to develop a real-time deep learning model to detect and estimate the duration of downlinking sequences of Rotary Steerable Systems (RSS) based on a single measurement (standpipe pressure, SPP). Further analytics are derived based on the downlink recognition results together with other real-time log data (ROP, RPM, Torque, etc.) to drive directional drilling efficiency.\u0000 Real-time RSS downlink recognition is treated as an image segmentation problem. The Deep Learning (DL) models were created using the dynamic U-Net concept and materialized with a pre-trained ResNet-34 as the underlying architecture. Transfer learning was used due to the limited number of training samples (≪ 100 downlinks per onshore well) to help with speed and accuracy. The SPP time series data was segmented based on stand of pipe drilled (one image per stand). This \"image\" was then fed into the model for downlink recognition. To further increase the accuracy, a second opinion mechanism was applied when the models were tested and deployed into the Real-Time Drilling (RTD) system. Using a dual model approach greatly reduced the number of false positives due to non-downlink pressure fluctuations causing \"noise\". The patterns of SPP and its rate of change (delta SPP) are quite different. They both have pros and cons for identifying the downlink, thus two independent models were built based on these two signals. The DL model A is trained based on the original SPP signal and the DL model B is trained based on delta SPP. A downlink is confirmed only when both models show positive results.\u0000 Data of 10 onshore wells (2 rigs) drilled with RSS were segmented (8165 images in total) and labeled. There were 671 images with 795 downlinks and 7980 images without downlink. The five-fold cross-validation technique was used to identify the best model(s). The F1 score of blind test result was .991 (accuracy was ~99.82%, see Table 2). The relative error of duration estimation is 2.49%. The current rig fleet within the RTD system has a mix of drilling tool configurations - RSS and mud motors. To further validate the models’ robustness regarding drilling tools, additional tests were conducted using mud motor wells’ datasets from 21 rigs (25431 images without downlink). There were 3 false negatives from this extended test set, which resulted in a ~99.93% accuracy for the aggregated 31 wells dataset. These results suggest that the models are accurate, reliable and robust.\u0000 The real-time DL solution presented in this paper enables operators to analyze RSS performance during and between downlinking events. This would allow drilling engineers and rig supervisors to make faster, more reliable data-driven decisions to optimize performance and directional control of the well path.","PeriodicalId":429641,"journal":{"name":"Day 1 Tue, March 03, 2020","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-02-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128604830","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
E. Oort, M. Juenger, M. Aldin, A. Thombare, M. McDonald, Alex Lucas, F. Ditlevsen
Well abandonment is one of the biggest challenges in the oil and gas industry, both in terms of cost and effort as well as the technical hurdles associated with wellbore isolation for an indefinite term. A mechanism that may be exploited to simplify well abandonments is using natural shale formations for the creation of annular barriers. Currently, uncemented annuli often require casing milling and pulling before abandonment plugs can be set, which necessitates the use of a drilling rig. This is an expensive, time- and labor-intensive process, particularly offshore. However, shale creep may naturally form a barrier behind uncemented casing sections. With a qualified annular shale barrier in place, the well may only require the setting of abandonment plugs within the existing casing string(s), a task that can often be done rigless and with significantly less effort. The work described in this paper presents the results of a rock mechanical investigation into the creep behavior of North Sea shales and their ability to form effective annular barriers. Field core from the Lark-Horda shale was used to conduct dedicated, customized experiments that simulated the behavior of shale confined under downhole effective stress, pressure and temperature conditions to fill in an annular space behind a simulated casing string. Full scale tri-axial rock mechanics equipment was used for testing cylindrical shale samples obtained from well-preserved field core in a set-up that mimicked an uncemented casing section of a well. The deformation behavior of the shale was monitored for days to weeks, and the formation of the annular barrier was characterized using dedicated strain measurements and pressure pulse decay probing of the annular space. The large-scale lab results clearly show that the Lark-Horda shales will form competent low permeability annular barriers when left uncemented, as confirmed using pressure-pulse decay measurements. They also show that experimental conditions influence the rate of barrier formation: higher effective stress, higher temperature and beneficial manipulation of the annular fluid chemistry all have a significant effect. This then opens up the possibility of activating shale formations that do not naturally create barriers by themselves into forming them, e.g. by exposing them to low annular pressure, elevated temperature, different annular fluid chemistry, or a combination. The results are in very good agreement with field observations reported earlier by several North Sea operators.
{"title":"Simplifying Well Abandonments Using Shale as a Barrier","authors":"E. Oort, M. Juenger, M. Aldin, A. Thombare, M. McDonald, Alex Lucas, F. Ditlevsen","doi":"10.2118/199654-ms","DOIUrl":"https://doi.org/10.2118/199654-ms","url":null,"abstract":"\u0000 Well abandonment is one of the biggest challenges in the oil and gas industry, both in terms of cost and effort as well as the technical hurdles associated with wellbore isolation for an indefinite term. A mechanism that may be exploited to simplify well abandonments is using natural shale formations for the creation of annular barriers. Currently, uncemented annuli often require casing milling and pulling before abandonment plugs can be set, which necessitates the use of a drilling rig. This is an expensive, time- and labor-intensive process, particularly offshore. However, shale creep may naturally form a barrier behind uncemented casing sections. With a qualified annular shale barrier in place, the well may only require the setting of abandonment plugs within the existing casing string(s), a task that can often be done rigless and with significantly less effort.\u0000 The work described in this paper presents the results of a rock mechanical investigation into the creep behavior of North Sea shales and their ability to form effective annular barriers. Field core from the Lark-Horda shale was used to conduct dedicated, customized experiments that simulated the behavior of shale confined under downhole effective stress, pressure and temperature conditions to fill in an annular space behind a simulated casing string. Full scale tri-axial rock mechanics equipment was used for testing cylindrical shale samples obtained from well-preserved field core in a set-up that mimicked an uncemented casing section of a well. The deformation behavior of the shale was monitored for days to weeks, and the formation of the annular barrier was characterized using dedicated strain measurements and pressure pulse decay probing of the annular space.\u0000 The large-scale lab results clearly show that the Lark-Horda shales will form competent low permeability annular barriers when left uncemented, as confirmed using pressure-pulse decay measurements. They also show that experimental conditions influence the rate of barrier formation: higher effective stress, higher temperature and beneficial manipulation of the annular fluid chemistry all have a significant effect. This then opens up the possibility of activating shale formations that do not naturally create barriers by themselves into forming them, e.g. by exposing them to low annular pressure, elevated temperature, different annular fluid chemistry, or a combination. The results are in very good agreement with field observations reported earlier by several North Sea operators.","PeriodicalId":429641,"journal":{"name":"Day 1 Tue, March 03, 2020","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-02-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115375110","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The Company Man or Drilling Supervisor is one of the key personnel who can mitigate and minimize risk in drilling operations and ensure Non-Productive Time is kept to a minimum. Their technical competency is crucial to running safe and efficient drilling operations and industry generally ensures these key personnel have the knowledge and skills to do this. However equally or even more importantly, are those skills in the human factors sphere since these account for some 80% of the total causal factors for accidents and incidents offshore. This is especially pertinent for well control incidents but similarly true for all other events. The application of non-technical skills to safety and efficiency should be seen as crucial in all high-risk operations. This is well understood in the aviation and other industries where CRM (Crew Resource Management) has been practiced for some time. The oil and gas industry is starting to recognise the ever-increasing importance that psychological factors play in safe and efficient operations. Furthermore to make the step-change improvement needed in operational safety and efficiency requires all members of the well operations team to have effective development in the application of non-technical skills. Non-technical skills and human factors encompass many elements all of which influence our behaviour. In a work environment these include different areas; for example environmental and organisational components, the use of a variety of equipment, various processes and procedures and the characteristics of many different personnel along with their skills and competencies. It is widely acknowledged many of the world's worst oil field incidents are directly attributable to human error and decision-making. It is notable that some incidents may initially appear straightforward but are frequently exacerbated by human error. Risk and costs increase and in the worst cases results in blowouts or other major disasters. A new programme has been developed to address this and assist in embedding these non-technical skills. It is recognised this is a new area where traditionally industry has only trained in a technical manner and this will require commitment from all.
{"title":"The Company Man Programme","authors":"T. Morton","doi":"10.2118/199659-ms","DOIUrl":"https://doi.org/10.2118/199659-ms","url":null,"abstract":"\u0000 The Company Man or Drilling Supervisor is one of the key personnel who can mitigate and minimize risk in drilling operations and ensure Non-Productive Time is kept to a minimum. Their technical competency is crucial to running safe and efficient drilling operations and industry generally ensures these key personnel have the knowledge and skills to do this. However equally or even more importantly, are those skills in the human factors sphere since these account for some 80% of the total causal factors for accidents and incidents offshore. This is especially pertinent for well control incidents but similarly true for all other events.\u0000 The application of non-technical skills to safety and efficiency should be seen as crucial in all high-risk operations. This is well understood in the aviation and other industries where CRM (Crew Resource Management) has been practiced for some time.\u0000 The oil and gas industry is starting to recognise the ever-increasing importance that psychological factors play in safe and efficient operations. Furthermore to make the step-change improvement needed in operational safety and efficiency requires all members of the well operations team to have effective development in the application of non-technical skills.\u0000 Non-technical skills and human factors encompass many elements all of which influence our behaviour. In a work environment these include different areas; for example environmental and organisational components, the use of a variety of equipment, various processes and procedures and the characteristics of many different personnel along with their skills and competencies.\u0000 It is widely acknowledged many of the world's worst oil field incidents are directly attributable to human error and decision-making. It is notable that some incidents may initially appear straightforward but are frequently exacerbated by human error. Risk and costs increase and in the worst cases results in blowouts or other major disasters.\u0000 A new programme has been developed to address this and assist in embedding these non-technical skills. It is recognised this is a new area where traditionally industry has only trained in a technical manner and this will require commitment from all.","PeriodicalId":429641,"journal":{"name":"Day 1 Tue, March 03, 2020","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-02-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127796051","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：