For decades, efforts have been made to automate the HAZOP process. The motivation has mainly been to displace expensive manual HAZOP approaches, that are furthermore known to suffer from systemic quality issues related to system complexity, uncertainty, vagueness and level of knowledge completeness. With offset in a review of the main historic arguments for automating the HAZOP analysis, and an outline of the particular benefits of employing Multilevel Flow Modelling (MFM) theory in this context, this paper emphasises the opportunity to redeploy the insights achieved by the HAZOP team to assist an operator facing an abnormal event years later. By means of a detailed analysis of an actual catastrophic failure of a FPSO compression module, the paper demonstrates how MFM enabled HAZOP captures explicitly tacit expert knowledge about the complex interdependencies between process design, equipment design, safety barriers and instrumentation. The paper further describes a methodology to interpret measurements online by means of the MFM analysis, thereby establishing real-time cause and consequence analysis in sufficient time to interrupt the escalation from a benign sensor malfunction to a topside explosion.
Objective/scope – This paper is an extension of the work presented in URTeC paper #2856750, 2018, where a decline curve reproducing the standard transition from linear flow to pseudo steady state flow regimes was proposed for fractured horizontal wells in unconventional reservoirs. In this new work, the formulation was generalized to a succession of flow regimes. This enables the new decline curve to reproduce a succession of periods with power-law behavior, as predicted by recent work on anomalous diffusion models. By construction, this decline curve is fully consistent with RTA concepts and the predictions of physical models.
Method/Procedures/Process – The decline curve is obtained by numerically integrating, in the material balance time domain, a “base function” defined as a succession of straight lines reproducing the successive flow regimes, linked with continuous transition periods. By construction, the base function follows the characteristic evolution of the rate-normalized pressure derivative on a loglog plot, which ensures the physical consistency of the obtained decline curve. Once the curve is matched, its parameters can then be used to infer combinations of physical parameters. Two approaches are possible: (1) In the absence of bottomhole pressures, a classical match and forecast of the instantaneous rate or of the cumulative can be performed in the real time domain directly, with only 3 parameters. (2) If a bottomhole pressure history is available, the evolution of the rate-normalized pressure and its derivative can first be displayed on a loglog plot, where the different regimes emerging can be rigorously identified, and the corresponding lines traced. Once the different straight lines have been positioned, they are automatically linked with continuous transition periods, and seamlessly integrated into the final decline curve. The forecast can be made either by extending the last observed flow regime, or by using a more conservative regime.
Results/Observations/Conclusions – The calculation steps leading to the decline curve are detailed. Comparisons of the results from the decline curve and from various physical models (including analytical anomalous diffusion models with different flow regimes paths) show excellent agreement. Several application examples are shown and the estimation of physical parameters from the parameters of the decline curve is demonstrated.
Applications/Significance/Novelty - The proposed decline curve reproduces the successive emergence of expected flow regimes and is fully consistent with the predictions of currently available physical models – including recent anomalous diffusion models. The parameters of the curve can be used to estimate combinations of the corresponding physical model parameters. As a consequence, the proposed approach bridges the gap between empirical decline curve techniques and the physical concepts used for rate transient analysis.
The intent of this work is to outline the workflow that engineers can apply in the selection and modelling of inflow control devices (ICDs) for use in completion design. It provides a step-by-step guide and examples of how to address the key challenges in the field of modelling of ICDs, namely: the choice of ICD technology that is most suitable for the particular well; the process of building a custom ICD performance model and import into the reservoir simulator; considerations in setting up the simulation model; and, verification of the results. The paper highlights the risk of using empirical ICD models, available in reservoir simulators, in the early stages of evaluating inflow control technology. It then demonstrates that the decision on whether or not an inflow control is required, and what type, can be made without resorting to these built-in empirical models in reservoir simulators. The paper introduces a new approach to evaluation, comparison and modelling of inflow control devices of different types and geometry. Focused on simplicity and accuracy, the method of ICD characterisation is described, as is how to use this characterisation to generate input to the reservoir simulator. A method for verification of the output from the reservoir simulator is also presented. The application of this modelling approach is illustrated through a case study on evaluating inflow control technology in a mature oil reservoir.
Significant effort around bacteria monitoring has been a focus since July 2016. Water testing program covers all operating areas including southwestern Pennsylvania and Marshall county, West Virginia.
The program began with most probable number (MPN); otherwise known as serial dilution testing, for sulfate reducing bacteria (SRB) and acid producing bacteria (APB). Based on the water test results, biocide treatments were performed for selected wells from October 2016. More than 300 horizontal wells from Marcellus formation were tested for SRB and APB by third quarter 2017 and biocide treatments were optimized by the end of 2017.
Bacteria DNA testing in the form of next generation sequencing (NGS) and quantitative polymerase chain reaction (qPCR) was performed on selected wells to characterize the microbial consortium present in these samples. Due to the limitations of serial dilution testing and understanding of the bacteria makeup with DNA testing, the program has evolved to utilize fieldwide adenosine triphosphate (ATP) testing and a locked-down biocide treatment procedure. Biocide treatments are prioritized during extended midstream maintenance to minimize downtime and maximize contact time and have shown to be effective for extended durations.
To investigate the root causes of bacteria, data science tools are utilized to look at the effect of the completions design and other variables on bacteria during operation. Variables under consideration include biocide type and concentration, fracturing fluid chemical additives, produced water reuse during completions, produced water chemistry, and geologic variation. Initial findings will be discussed and shared with lessons learned from production operations.
Roostaei, M. (University of Alberta) | Nouri, A. (University of Alberta) | Fattahpour, V. (University of Alberta) | Mahmoudi, M. (University of Alberta) | Izadi, M. (Louisiana State University) | Ghalambor, A. (Oil Center Research International) | Fermaniuk, B. (RGL Reservoir Management)
Standalone screen (SAS) design conventionally relies on particle size distribution (PSD) of the reservoir sands. The sand control systems generally use D-values, which are certain points on the PSD curve. The D-values are usually determined by a linear interpretation between adjacent measured points on the PSD curve. However, the linear interpretation could result in a significant error in the D-value estimation, particularly when measured PSD points are limited and the uniformity coefficient is high. Using the mathematical representation of the PSD is an efficient method to mitigate these errors. The aim of this paper is to assess the performance of different mathematical models to find the most suitable equation that can describe a given PSD.
The study collected a large databank of PSDs from published SPE papers and historical drilling reports. These data indicate significant variations in the PSD for different reservoirs and geographical areas. The literature review identified more than 30 mathematical equations that have been developed and used to represent the PSD curves. Different statistical comparators, namely, adjusted R-squared, Akaike's Information Criterion (AIC), Geometric Mean Error Ratio, and Adjusted Root Mean Square Error were used to evaluate the match between the measured PSD data with the calculated PSD from the formulae. The curve fit performance of the equations for the overall data set as well as PSD measurement techniques were studied. A particular attention was paid towards investigating the effect of fines content on the match quality for the calculated versus measured curves.
It was found that certain equations are better suited for the PSD database used in this investigation. In particular, Modified Logestic Growth, Fredlund, Sigmoid and Weibull models show the best performance for a larger number of cases (highest adjusted R-squared, lowest Sum of Squared of Errors predictions (SSE), and very low AIC). Some of the models show superior performance for limited number of PSDs. Additionally, the performance of PSD parameterized models is affected by soil texture: For higher fines content, the performance of equations tends to deteriorate. Moreover, it appears the PSD measurement techenique can influence the performance of the equations. Since the majority of the PSD resources used here did not mention their method of measurement, the effect of measurement technique could only be tested for a limited data, which indicates the measurement technique may impact the match quality.
Fitting of parameterized models to measured PSD curves, although well known in sedimentology and soil sciences, is a relatively unexplored area in petroleum applications. Mathematical representation of the PSD curve improves the accuracy of D-values determination, hence, the sand control design. This mathematical representation could result in a more scientific classification of the PSDs for sand control design and sand control testing purposes.
Golenkin, Mikhail Yurievich (LUKOIL-Nizhnevolzhskneft, LLC) | Latypov, Artur Spartakovich (Schlumberger) | Shestov, Sergei Alexandrovich (Schlumberger) | Bulygin, Igor Alexandrovich (Schlumberger) | Khakmedov, Azat Meredovich (Schlumberger)
First Intelligent multilateral TAML5 wells on Filanovskogo Field is the great example of how new technologies help to optimize CAPEX, and, thanks to higher productivity index, achieve higher production rate. Multilateral well geometry combined with ability to monitor and control each leg separately helps to optimize flow patterns, prolongs well life and contributes to higher cumulative production. The paper focuses on well design, project execution and production results.
In order to achieve results, work was done in several phases: Choose well design which would optimize CAPEX and allow to reach production and recovery targets. Perform two trial jobs on existing mature field to learn technology and prove the concept. Use experience gained on trial jobs to optimize requirements, well design and procedures. Execute the job, control and manage execution to ensure compliance to the plan. Review first production results and estimate benefits obtained from project execution.
Choose well design which would optimize CAPEX and allow to reach production and recovery targets.
Perform two trial jobs on existing mature field to learn technology and prove the concept.
Use experience gained on trial jobs to optimize requirements, well design and procedures.
Execute the job, control and manage execution to ensure compliance to the plan.
Review first production results and estimate benefits obtained from project execution.
This paper describes all the steps focusing mainly on installation procedure, execution and production results review.
As a result of the work done, LUKOIL successfully installed two first intelligent TAML5 completions on Filanovskogo field and achieved
The paper describes introduction of complex intelligent multilateral well design on the field. This practical example can be used for future reference by drilling and production focused petroleum industry professionals to better understand benefits and limitations of existing technologies. Actual production result can also be used as a benchmark for field development planning.
New technologies that contribute to enhanced production in ultralong tiebacks have recently been developed. These new developments include higher differential pressure in multiphase pumps and compressors, mechanical designs for high pressures and temperatures, and power systems suited for ultralong tiebacks.
When developing new, cost-efficient boosting technology for long subsea tiebacks and deep water, a system approach is important. This includes power systems, installation methods, maintenance, reliability, and condition monitoring. The new technologies described have been developed based on operational experience and physical theory combined with practical experiments and validation, both scaled and full size. The importance of developing simple and reliable solutions in facilities that enables comprehensive experimenting and testing is also explained. Today’s oil and gas price level also requires cost-efficient solutions, and the paper explains how this can be obtained through standardization and modularization.
The first pump systems that are able to provide a more-than 200-bar differential pressure are already developed, qualified, and put in operation. A game-changing multiphase gas compressor technology that provides differential pressure up to 55 bar has also been built, tested, and verified. In parallel with these developments, subsea power systems have been further developed so that they can be used for step-outs longer than 200 km. Recently, a multiphase pumping station designed for 2,500-m water depth and 15,000-psi design pressure was installed and set in operation in the Gulf of Mexico. All of this contributes to enhanced production and lower field developments costs in subsea environments and provides a platform for further technology developments that can potentially make extremely remote subsea field developments economically attractive.
This paper presents new technology related to multiphase pumping and compression and a system approach that can make production from remote and deepwater subsea fields more capital efficient.
Because of the substantial cost involved, subsea environments demand technologies which improve efficiency. This statement is particularly true in marginal developments. In such scenarios, multilateral completions have played an integral part in improving field economics through improved well performance and reducing both field and operating costs to a point where the field development is financially appealing. In the northwest shelf of Australia, there have been four major oil field developments to date which have been based either entirely or partially on multilateral completions. This paper describes the evolution of multilateral completions throughout the past 10 years in Australia for two such developments, outlining the efficiency and operational gains during that time period.
Sand face completion techniques in Australia's northwest shelf vary from conventional standalone screens (SAS) to more sophisticated sand control methods, such as gravel packing. For multilateral wells in this region, even though gravel packs have been installed in other regions, developments to date have been limited to oil producers with SAS with inflow control devices (ICDs) and swell packers for compartmentalization. Nonetheless, the process of constructing and completing a subsea multilateral completion can be complex, and often trip intensive. Early multilateral field developments borrowed technology from the Norwegian northwest shelf; however, as such, the completions were not optimized for the specific Australian operating environment. Following the first successful subsea multilateral field developments in Australia, customized technology began to be developed to address and improve completion efficiencies. Specifically, several global and regional first installations were implemented between 2014 and 2016 that reduced dedicated installation time by 42% and helped reduce or eliminate operational risks.
In addition to discrete technologies, well architectures have also been revisited to further the economic advantages of multilateral completions. Trilateral wells, where three laterals penetrate the reservoir, are now commonplace in subsea developments in Australia. The custom solutions developed throughout the past 10 years demonstrate not only industry maturity, but also region specific advancements in multilateral completion technology.
Australia's history of multilateral completions is a success story, highlighting an industry's adoption of new technology and appetite for improvement on a broader scale. In ten years, the step change in multilateral completion has been appreciable. The result of a decade worth of improvement and customization is a fit-for-purpose, reliable, robust and efficient completion practice.
ABSTRACT: A numerical-model-based approach was recently developed by researchers at the National Institute for Occupational Safety and Health (NIOSH) for estimating the changes in both the horizontal and vertical loading conditions induced by an approaching longwall face. In this approach, a systematic procedure is used to estimate the model’s inputs. Shearing along the bedding planes is modeled with ubiquitous joint elements and interface elements. Coal is modeled with a newly developed coal mass model at NIOSH’s Pittsburgh Mining Research Division (PMRD). The response of the gob is calibrated with back analysis of subsidence data and the results of previously published laboratory tests on rock fragments. The model results were verified with the subsidence and stress data recently collected by PMRD from a longwall mine in the eastern United States, and with published cases studies from both eastern and western U.S. mines.
In 2015, there were 40 longwall mines operating in the United States, each producing an average of 4.5 million tons of coal per year. The largest concentrations of longwall faces are found in West Virginia with 13 longwalls, followed by Illinois (7), Pennsylvania (5) and Alabama (5) (Fiskor, 2016).
In 2015, these forty longwall mines supplied 60% of the U.S. underground coal production. This represents a substantial increase from 50% over the previous three years (Sears et al., 2017). During this period, reportable roof fall rates in U.S. longwall mines also increased. Large roof falls that can block the gateroads are not only a ground fall hazard; they disrupt the ventilation system, can block the escapeways, and can increase the potential for elevated methane levels in the gob. To address the recent increase in reportable roof falls, the National Institute for Occupational Safety and Health (NIOSH) Pittsburgh Mining Research Division (PMRD) started a research project to improve ground control in longwall gateroads.
ABSTRACT: Rib falls are serious safety hazards in underground coal mines. Studies of injuries and fatalities caused by rib falls have shown that the significant mining conditions that contribute to rib falls are mining height, depth of cover, and stress resulting from mining of multiple seams. In longwall mining, rib stability in gate entries is also influenced by the induced abutment pressures. To reduce the risk of rib injuries and fatalities in longwall mining, it is important to understand the mechanisms of rib failure under different geologic and mining conditions, and the requirements for rib support in the gate entries. Rib scoping was performed by NIOSH researchers to detect fractures in the ribs of the gate entries in the Pittsburgh coal seam. Rib fractures and failures were observed for various loading conditions throughout different mining cycles. Numerical modeling was performed to calibrate a coal mass model to the field data and to further predict rib fractures under different abutment stress levels. Based on field monitoring and numerical modeling, a conceptual model is proposed for considerations of rib support in longwall gate entries.
In the past several decades, the installation of roof support in underground coal mines has been a priority in preventing potential roof falls as entries and crosscuts are developed. Conversely, rib support has not been taken as seriously as roof support in the past, especially under the conditions where rib sloughage is not so pronounced. Although the scale of rib falls is usually smaller compared to the scale of roof falls, a chunk of coal or rock falling from rib could be a potential risk of injury. The current practice in coal mines in the United States is to choose rib bolts of specific diameter and length as a reactive action; that is, the rib bolting is designed in response to the scale of rib falls and injuries that have occurred (Mohamed et al., 2015). In some cases, whether or not to implement rib support can be easily determined through observations of rib sloughage, while in other cases, the rib may be over-supported or undersupported based on what is perceived through observations. Rib stability is primarily influenced by entry height, overburden depth, and stress resulting from mining of multiple seams. In longwall gate entries, rib stability is also affected by the induced abutment pressures created by longwall retreating.