National Academies Press: OpenBook
« Previous: 1 Introduction
Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×

2

The Committee’s Responses to the Charge Questions

This chapter represents the committee’s consensus answers to targeted charge questions provided by the Bureau of Safety and Environmental Enforcement (BSEE) as a frame for the committee’s assessment of the interim report prepared by the Naval Research Laboratory (NRL). Chapter 3 expands on broader comments on the interim report addressing the committee’s statement of task (Box 1-1 in Chapter 1).

Were the objectives of the study clearly defined?

The description of the objectives in the NRL interim report lacks detail and fails to provide a broader context regarding the relevance of the study to the problem at hand. These contextual gaps include laboratory-scale and computational fluid dynamics (CFD) validation, especially with respect to scaling up to relevant physical scales in which buoyancy and radiation depend on the physical dimensions of the wellhead. There is no explicit characterization of the relevant nondimensional numbers of the real-world problem when the bench- and intermediate-scale experiments are compared. In particular, radiation will be increasingly important at larger scales, whereas conduction and convection may dominate at the bench scale.

Relatedly, explanations for why the set of fuels was selected for study are limited, and the relevance of these fuels to the expected multicomponent crude oil is unclear. Important processes—e.g., vaporization properties, surface tension, and preferential evaporation missing in a single-component spray—may have been overlooked by this selection. This work appears to have placed emphasis on the lighter end of the spectrum of hydrocarbon components found in crude oils, whereas if what falls to the ground is the priority, the fuels considered need to include heavier hydrocarbons.

One important clarification in the modeling objectives is whether the selection of submodels is intended for engineering calculations or for high-fidelity models that would be used for more scientific analyses. Some of the assumptions provided for the submodels are significant and not necessarily state of the art.

It would also be useful to clarify the manner in which the efficacy of the different submodels would be determined. This clarification would be bolstered by a discussion of whether the choice of submodel is likely to under- or overestimate the burn efficiency, as well as of the sensitivity of burning efficiency to the submodels and tunable parameters within the submodels.

Were the assumptions regarding wellhead conditions and two-phase wellbore flow (including film thickness and instability, liquid entrainment, and droplet diameter and its influence on wellhead ejection behavior) adequately characterized?

Earlier sections of the report include a good discussion of wellhead-spray combustion scenarios. However, the authors do not clearly justify their choices of the submodels and the values of their assumptions, nor do they cite adequate references. Few assumptions on the selection of properties are specified. In the absence of detailed correlations for property estimation, the assumptions may be fine, but the modeling and experimental work do not validate the properties used. The two-phase wellbore flow modeling was focused on film thickness, entrainment, and droplet diameter, but the role of these parameters in combustion efficiency was not established, and it is unclear that these parameters are sufficient to describe combustion efficiency (quantitatively or qualitatively).

Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×

The authors assume annular-mist flow behavior for the sake of brevity and applicability, as these sprays may atomize well. However, the pools or fountains emerging from lower speed flows may not burn well, as evidenced by the authors’ experimental results. Thus, the modeling they performed may not have considered the “worst-case” conditions for combustion efficiency (i.e., conditions in which significant oil droplets drop out of the flow).

Knowledge of the range of nondimensional parameters expected in multiphase wellbore flow and a review of the literature on the regimes of the transition flow in wellbores would help clarify how relevant the authors’ assumptions are for the wellbore flow. The authors provide a reasonable review of the literature on correlations for film thickness, liquid entrainment, and droplet diameters. However, it is not clear whether those correlations are valid for the regimes under consideration. Additionally, while the correlations for Weber numbers may be valid for the bench-scale simulations, it is unclear whether they are applicable for the actual wellhead.

A fundamental and critical concern is the model used to generate the input conditions for the NRL model. The worst-case discharge (WCD) model from Hilcorp is proprietary (per the Hilcorp report, whose Appendix G is not provided). Thus, detailed data, such as the content listed in the Society of Petroleum Engineers Technical Report,1 including flow correlations and uncertainty ranges of the parameters used in the Hilcorp WCD model, are not provided for evaluation. The choice of modeling methods will affect outputs from the Hilcorp WCD model, which were used as input conditions for the NRL model. Perhaps it is possible for BSEE to provide input ranges used for the WCD model for the particular reservoir of interest in the NRL study so there is some control on the input, initial, and boundary conditions used with the NRL model. Moreover, the WCD model emphasizes volume flow rate and does not consider dimension predictions for annular-domain inner radius, spray/mist character in the central core of the pipe flow, or the liquid structures that form in the pipe-exiting cascade process. These aspects will affect atomization or liquid-stream breakup.

Alternatively, the Hilcorp report includes some information on reservoir rock and fluid properties (Section 4) and well design (Section 8), among other data that could be used to build an independent WCD model. An independent model would require additional analog and dynamic data, but would be useful to verify output from the Hilcorp model and to cross-validate the WCD output of different wellbore flow models. Such verification and validation would help characterize the wellhead conditions and provide uncertainty bounds for the NRL study.

A related critical concern is the conditions during actual drilling. During actual drilling, if reservoir or flow conditions changed, the WCD model and the associated results would need to be updated. The uncertainty associated with such a scenario would affect outflow and wellhead conditions, leading in turn to questions about whether wellhead burning will suffice as a response plan in all unpredicted situations.

Key expertise on WCD model building and wellbore hydraulics appears to have been lacking in the NRL study. Important research reports on WCD could help guide and inform future modeling and experimental work on this subject (see Appendix D).

Lastly, the NRL interim report does not address the roles of actual wellhead failure geometry and interference of the flow discharge with superstructure. The condition and type of exit structure can affect the external flow, so the systems used in the modeling and experimental work require justification.

Was the physical model for multiphase flow adequately developed to capture the liquid droplet phase and the gas-phase flow field?

It is unclear whether the collective physical and computational aspects of the NRL modeling are appropriate for predicting the actual wellhead conditions in the field, as opposed to predicting the bench- and intermediate-scale experiments that were conducted. The bench-scale problems could be considered as an initial set of unit problems representing the wellhead conditions and opportunities to investigate the

___________________

1 Society of Petroleum Engineers. 2015. “Calculation of Worst-Case Discharge (WCD).” SPE Technical Report Rev. 1.

Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×

significance of parameters relevant to each unit-scale enquiry. Even the bench-scale problem has significant complexities, and while the authors are reasonable in their preliminary approach, a number of issues remain as relates to the relevance of the selection of models and their validation in terms of the subprocess models, uncertainty quantification, and sensitivity analyses relative to the parameters given. The selection of the CFD models using conventional Reynolds-averaged Navier-Stokes (RANS) approaches and subprocess models is justified based on computational costs and time constraints. Nonetheless, the subprocess models were taken off the shelf with little further development, and the NRL interim report contains sufficient information to show that further development is needed based on the droplet dynamics and combustion data. Given the complexity of the overall problem, subprocess models must work in tandem, and it would have been preferable to provide sensitivity and uncertainty quantification of model assumptions, constants, and boundary and initial conditions as they relate to the ultimate objective of predicting burn efficiency. The report would benefit from a more extensive review of the literature on flow, physical, and chemical properties observed under wellhead conditions to guide the selection of further experiments and CFD models.

Were the soot and radiation models adequately characterized?

The sooting and liquid-phase coke particulate emission characteristics of crude oils are not well represented by n-heptane. Particulate generation and burnout will affect several critical physical and chemical transport mechanisms in the model, including the radiative energy balance. In addition, preferential evaporation of lighter components in crude oil may induce composition and thermal stratification in the mixture not captured by n-heptane, which affect combustion rates. CFD mixing and combustion models need to account for this stratification. While use of a surrogate may be a necessary approach for representing crude oils, the committee had considerable concern regarding how to appropriately develop and validate surrogates for studies of crude oil combustion. Developing a surrogate that could be reproduced by other members of the community for complementary studies would be valuable. Such a surrogate would need to reproduce the relevant properties of crude oil, including viscosities, surface tension, latent heat, boiling point, and heat of combustion. Existing crude oil distillation and chemical properties show the extent to which internal liquid cracking and gasification must vary (as a result of changes in distillation fractions with temperatures exceeding 350 oC) for oils located just tens of miles distant from one another. Much greater effort to characterize and understand the physicochemical property effects of crude oils on atomization and combustion will be needed if the proposed model is to be used for regulatory applications, as proposed by the sponsors. This work would likely require better understanding of current oil property tests and the development of new standardized tests (e.g., by ASTM) specifically suited to crude oils.

In addition to an oil surrogate, the study would be greatly improved by the inclusion of a laminar configuration for parametric studies. The laminar flame configuration would benefit from two types of studies: volatilized combustion and spray flame combustion of the surrogate under different conditions that could test effects of entrainment of cold air. Further suggestions include studying effects of turbulent cross-flow at typical Arctic wind speeds, and potentially leveraging data from large-scale pool fires with wind in the literature and perhaps available from other National Laboratories. However, understanding the cross-flow and discharge ratios of these experiments compared with the NRL system is critical to their utility in model development.

Some specific suggestions for improving the soot and radiation submodel characterization include using a blend of sooting propensity materials (including aromatics and aliphatics) without coking potential to develop a simple soot model that could be validated. The soot submodel needs to be validated using laminar flame configuration and by comparison with the sooting flame data in jet flames from the International Sooting Flame Workshop.2 The submodel could be validated using any of a number of soot

___________________

2 International Sooting Flame Workshop. n.d. Data Sets. The University of Adelaide. https://www.adelaide.edu.au/cet/isfworkshop/data-sets.

Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×

measurements, such as luminosity, extinction, or laser-induced incandescence.3 Another potential approach to validating the soot submodel is searching the literature for information on smoke point for similar fuels and reproducing the data for the surrogate (see Appendix D for suggested resources).

Assessing soot production and radiation with entrainment of air of different temperatures is an example of how to provide valuable information on the sensitivity of the submodels to such input parameters as the colder air temperatures expected in the Arctic. Sensitivity analysis is critical to understanding the effects of model input uncertainties.

Were Lagrangian droplet dynamics and thermophysics adequately incorporated into the model?

Use of a combined Eulerian-Lagrangian approach appears to be the appropriate choice for modeling the droplet dynamics. However, significant physical contributors were missing from the modeling effort. For example, while there was some attempt to validate this aspect of the model, it was not based on droplet dynamics. In fact, the droplet trajectory and velocity were not measured, although it may be possible to analyze the data collected in the experiments to determine this information quantitatively. Additionally, the dynamic model was inadequate; it did not account for gravity/buoyancy, assuming that Stokes drag for large droplets in air is incorrect, and proximity to other droplets was not accounted for. Also not accounted for were other thermophysical phenomena, such as internal droplet heating and circulation, as well as preferential evaporation and swelling, which may induce stratification that affects combustion rates. Some of these phenomena may not be important; however, characteristic length and time scales are necessary to justify omitting or including them in the model and submodels. Fundamentally, the model and submodels are semi-empirical and were not developed or tested for oil well fires, nor were sensitivity and uncertainty due to model assumptions and input parameters analyzed; therefore, high uncertainties are likely.

Other concerns about modeling of the droplet dynamics include the initial conditions and the fuel choice. It would have been useful as well to model the flow in the pipe and use it as initial conditions. Pipe exit geometry effects may also affect the exit flow (e.g., spray characteristics, flow rate). Experiments could explore such features as jagged protrusions and inward or outward tuliping of the pipe to understand how important these issues are. Additionally, representing crude oil using a single-component heptane is unsuitable because the heptane is too light and has specific associated sooting propensity due to its chemical structure; hence the sooting potential and characteristics will be different from those of crude oil.

Does the droplet injection model adequately simulate realistic diameters and velocities of two-phase, high-speed flows that would occur during a wellhead blowout event?

In the absence of data for the two-phase flow in the pipe and injection plane, the authors made a series of clearly stated assumptions. The experiments and simulations were performed at scales that differ significantly from those in actual oil well conditions. Attempting to account for scale effects by performing experiments at different scales is the proper approach. However, it is unclear and hardly discussed whether the laboratory droplet diameters and velocities (which were not measured) are relevant to actual wellhead conditions, which are extremely difficult to achieve. Differences in scales and breakup regimes, including thermal effects, are likely to generate very different droplet statistics and dynamics. Other potential significant effects include primary and secondary breakup, and deformations. The choice (or validation) of the model constants was not evaluated and is not discussed in the interim report. Cross-flow winds could also play a significant role in the droplets’ breakup and transport, and some evidence exists in the literature about how breakup mechanisms change as Reynolds number and Weber number change. The values of these parameters in the domains for the two scales of experiments and in the third domain related to the practical field need to be compared.

___________________

3 Michelsen, H. A. 2017. Probing soot formation, chemical and physical evolution, and oxidation: A review of in situ diagnostic techniques and needs.” Proceedings of the Combustion Institute 36(1):717-735. DOI: https://doi.org/10.1016/j.proci.2016.08.027.

Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×

Does the validation process capture the controlling physical properties to a sufficient level of accuracy including transport and boundary conditions at the bench- and intermediate-scales for both gas-phase and two-phase turbulent spray?

There was general consensus among the committee that the validation process did not capture the controlling physical properties; however, the committee also recognized that validation of such complex fuels, processes, and models is challenging. While the experimental progression is well described in the interim report, discussion of quantitative validation is limited. The experiments provided some valuable information; however, they did not target validation of specific submodels.

Radiation and soot formation submodels were not appropriately validated, and these phenomena are expected to have significant effects on combustion efficiency and dimensionality, among other plume characteristics. In particular, heptane is not a high-sooting fuel, and its sooting propensity is not expected to be consistent with that of wellhead fuels. Other fuels, such as a higher-sooting-propensity single-component fuel (e.g., toluene) or mixtures of such a higher-sooting fuel with heptane, could be used to assess experimentally the effects of sooting propensity on the plume characteristics and observable features. (See the suggestion to create a soot surrogate fuel under the above discussion of soot and radiation models.)

The experimental studies were not specifically directed at validation for specific submodels. Like the radiation submodel, the turbulence combustion closure submodel lacks justification and validation, and there are similar concerns regarding the droplet model. A fundamental concern is the primary assumption about using two-dimensional axisymmetric modeling for what is a highly three-dimensional physical flow. This assumption also has not been validated; horizontal wind speeds in the Arctic are very high, and cross-flow is expected to make wellhead flames highly nonaxisymmetric. The large-scale motion of the macroscopic flows and how they are coupled with the smaller-scale fluid motion have not been validated and may be a significant omission from the modeling and experimental validation efforts. The effects of turbulent mixing and the associated closure models (e.g., progress-variable scalar dissipation rate, mixture fraction dissipation rate, and cross-dissipation rates) also were not modeled and are not discussed in the interim report. Given the significant stratification expected with wellhead flames, these effects will very likely be important in determining the predicted combustion efficiency.

Other concerns relate to (1) transient heating and vaporization of the droplets, including the effect of shear-driven internal circulation within the droplet; (2) multicomponent mass diffusion within the liquid; (3) the importance of group droplet behavior in contrast to the assumption of isolated-droplet heating and vaporization; and (4) the mode of liquid-stream breakup (e.g., lobe-ligament-droplet cascade, lobe-hole-bridge-ligament-droplet cascade, or some other sequence). Understanding the size of the droplets expected would help in assessing the importance of these transport mechanisms.

Were the phase Doppler anemometry imaging diagnostic methods for the droplet behavior measurements appropriately designed, clearly described, and adequate to capture droplet behavior for the Gas Phase and Two-Phase Spray Flame?

There was agreement on phase Doppler anemometry (PDA) being appropriate, with some caveats with respect to the range of its use. The PDA system and the measurement technique are described well in the interim report. The results are valuable for providing insight into the structure of the flow. The PDA system results are not useful near the exit plane because the liquid still has a sheet or ligament-like structure; they are useful only when the spray is composed predominantly of spherical droplets. Additionally, the data plots lack error bars. Regarding the interpretation of the data, on page 61 of the interim report, the authors state that there were essentially no droplets outside r = +/–4 mm; however there were enough droplets to obtain a velocity reading, which appears to represent an inconsistency.

Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×

Were the diffuse back-light illumination imaging diagnostic methods for the droplet behavior measurements appropriately designed, clearly described, and adequate to capture droplet behavior for the Gas Phase and Two-Phase Spray Flame?

The diffuse back-light illumination imaging was appropriately designed and provided meaningful insight and some data on the shape of the droplets and plume configuration, elucidating some of the initial breakup processes. The procedures are described adequately in the interim report. However, the report does not use the data for characterizing the droplet dynamics, other than showing one sample demonstrating a capability to track the droplets. Much more information—e.g., droplet velocity and size distribution—could be obtained by dynamically postprocessing the data. The interim report provides only preliminary results, with detailed analysis left for future investigations. An uncertainty analysis and assessment of accuracy are also absent.

Were the diagnostic methods (Coherent Anti-Stokes Raman Spectrometry-based Thermometry [CARS]) for the temperature measurements appropriately designed, clearly described, and adequate to capture temperature for the Gas Phase and Two-Phase Spray Flame?

The committee reached general agreement that the Coherent Anti-Stokes Raman Spectrometry-based Thermometry (CARS) method was competently applied and appropriate for the experiments and is well described in the interim report, with the caveat of suggesting improvements to the analyses. Specifically, the authors need to do a more thorough uncertainty analysis for their CARS measurements. This is a much more complex task than that for the previously discussed PDA measurement. Figure 45 in the NRL interim report appears to indicate that uncertainty analysis was done for data from the particular flame that was investigated, with 0.1 g/sec of ethane and 0.2 g/sec of heptane. The authors need to explain how they determined these uncertainties and include uncertainty bars on Figures 28 and 30 as well. They also need to explain clearly the differences between the averaged and single-shot measurements shown in Figure 53. For future CARS measurements in this group, it is essential to further develop the computational framework for analysis of single-shot CARS spectra.

Were the diagnostic methods (3-Color High-Speed Pyrometry) for the temperature measurements appropriately designed, clearly described, and adequate to capture temperature for the Gas Phase and Two-Phase Spray Flame?

The assessment of this technique and the results needs to incorporate uncertainties in the measurements, particularly in comparison with CARS gas-phase measurements. The 3CHIP method for temperature analysis of soot raises some issues, as does the interpretation of the results. First, the emissivity of soot changes dramatically with maturity. Lower in the flame, where the soot is less mature, the dispersion exponent ξ will be much larger than 1; that is, the emissivity is proportional to 1/λξ, where ξ >>1 for young soot. Using a value of ξ that is too small will lead to overprediction of the temperature. Soot will age and mature with increasing height in the flame. In normal diffusion flames, soot is also often more mature at the flame front at higher radial distances. At full maturity, the dispersion exponent is less than 1 (ξ <1), and using the value of unity will lead to underprediction of temperature. Thus, the trend for temperature may be the opposite of that suggested in Figure 30 of the interim report for the particulates as a function of height in the flame, considering the change in optical properties with soot maturity. Along the edges of the flame, optical dispersion effects may highly perturb the 3CHIP measurements because of the refractive index changes with temperature. Finally, it is highly unlikely that soot will have a different temperature from that of the gas phase at atmospheric pressure because conductive heat transfer equilibrates temperatures on timescales of nanoseconds at such pressures. In addition, if there are interactions between less volatile crude oil and soot particles, optical properties of the internally mixed particles will deviate from those of mature soot and look more similar to those of young soot, leading to larger deviations from a 1/λ dependence of

Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×

the emissivity and significant errors in inferred temperatures. Furthermore, for high-sooting conditions, luminosity measurements need to be corrected for reabsorption.

Do the results adequately characterize evidence of the droplet characteristics including droplet breakup, the droplet size (diameter), droplet speed, and the duration of droplet in fire (bench- and intermediate-scales)?

There are significant disconnects between the model and the attempts to validate it. The work does not demonstrate key aspects of the models such as interactions between droplets, roles of convection, and radiation or gravity effects (including natural convection and falling droplets). The relevance to wellhead fuel properties, flow conditions (annular vs. bubbly, emulsions) is missing. The authors provide high-quality images that could prove insightful quantitative and qualitative information on the droplet breakup and modeling assumptions, but did not analyze those images showing droplet fragmentation, including interpretation of the behavior of the droplet breakup, the droplet velocity, or a characterization of the relevant physical conditions. Further analysis of the imaging data is needed, as recognized by the authors.

Does the research product accurately expand predictions of droplet diameters beyond current limited validated ranges?

In the introduction to the interim report, the authors acknowledge the importance of scaling and discuss relevant dimensionless parameters, but they did not appropriately address this issue with analysis and experimental results, nor did they attempt to extrapolate the results between their two experimental scales or to full-scale conditions. A wealth of data from experiments performed by the authors could have been used to address scaling trends, but the authors did not perform this work, and did not use the experimental data well to examine the key assumptions about isolated droplet vaporization and heating and the effect of shear force on droplets. Are droplets batched together sufficiently to require the use of group theory for vaporization and burning? If nonspherical droplet shapes appear, shear could be one cause, thereby also being a likely cause of internal droplet circulation that would strongly affect heating and vaporization rates.

Does the research product accurately characterize the impact of two-phase flow regimes (bubble, slug, and churn) on the effluent plume (bench- and intermediate-scales)?

A clear outline of the two-phase flow characterization is critical because it directly impacts burn efficiency, defined as the amount of liquid that falls to the ground. While the experiments were configured around annular flow of liquid coming out of a pipe with some spray in the center, the interim report does not clearly present the evidence for the assumption of this regime. This assumption is critical to the manner in which breakup occurs, so it is difficult to discuss impacts of the flow regime on atomization unless one knows whether there is annular or bubbly flow.

Justification for the assumption of annular flow is necessary, whether experimental limitations or expectations of output from the wellhead. This assumption would have been strengthened by a more robust literature review on two-phase flows through wellbores. Another concern is whether there may be a water phase, which, in addition to making this a three-phase flow (water, oil, gas), would allow for the possible formation of oil-in-water and water-in-oil emulsions. Addressing how this would impact the modeling results would strengthen the model’s applicability to other reservoirs as well, even if water intrusion is not a concern here. See Appendix D for literature on this topic.

More explicitly, in this instance, because the Hilcorp WCD model is proprietary, no details are provided. Information on pipe flow and wellbore model details and types are also missing, as is the application of key expertise in wellbore pipe flow and WCD modeling and experimental research. There is no way to verify whether the WCD volume is valid. Hence, a great deal of uncertainty is associated with the WCD model used for input to the NRL model, and it is unclear how the authors dealt with this

Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×

uncertainty. This is not a small matter; indeed, it likely controls or limits the model predictions. Even if Hilcorp is unwilling or unable to provide specifics, identifying a range of values with confidences with the help of oil experts would enable NRL to build its own WCD model. Valuable references relevant to wellhead modeling, particularly Waltrich et al. (2019) and Society of Petroleum Engineers (2015), are included in Appendix D.

Does the research product adequately address how the wellbore flow would influence the ejected spray plume behavior, which directly influences how the oil and gas burns and how much will either fall back to the surface or remain vapor?

The committee reached agreement that this research product does not adequately address how the wellbore flow would influence the behavior of the ejected spray plume. The initial experiments are foundational, but need to be expanded based on the current limited observations and limited conditions considered. Furthermore, the envelope of conditions needs to include the range of physical properties expected for crude oils, including highly volatile dissolution and water in the fluid. Such experimental pursuits are important to enable ranking and prioritizing of the physical mechanisms that should be included in the model development and consideration of property ranges.

The committee has significant concern that the variability of oil composition dramatically affects many aspects of wellbore fires, including the pipe flow. If the goal is to create a model for a broad range of crude oils, the effects of the different thermophysical properties (e.g., surface tension, volatility, heat capacity) need to be considered. If the goal is to create a model for a more specific type of oil, then the range of oil properties needs to be summarized, and the experiments need to reflect that range of properties. Pipe geometry can likely play a role as well (e.g., shear flow, boundary layer effects). What are the geometric features of the wellbore exit (tapered pipe, flow bends)? Identifying the key pipe attributes required for the boundary conditions is a critical first step in designing the modeling approach and the experimental efforts. Additionally, detailed simulations of different pipe flow conditions are important to improve understanding of the behavior of the ejected spray plume. Emulsified materials may also be a relevant consideration (e.g., what water content is expected in the oil flows?).

While these are significant concerns, the experimental setup could potentially be used to study these effects, such as the role of pipe boundary layers and different crude oils, information that could be used to help understand the magnitude of the effects. Similar to the experiments, the subprocess models (turbulence, combustion, radiation, soot, spray) need to be individually validated against unit test experiments with well-characterized inlet boundary conditions for the envelope of relevant conditions before the combined effects are tested.

Does the research product accurately predict the length of fire plume, location of flame anchoring, height of flame, width/angle, expansion, etc.?

Evaluation of the fire plume dynamics is complex and heavily dependent on the submodels used and developed for liquid- and gas-phase transports. The authors incorporated some of these issues of elemental physics, albeit in the form of model parameters. However, they made many assumptions for the study. In particular, the width/angle of the liquid sheet was used as an input instead of being evaluated with a physical model, which leads to restriction in such parameters as flame dynamics, flame stability, and liftoff height. For flame calculations, tabulated values were used. This method requires that the fuel composition be known as droplet evaporation proceeds, thus necessitating a more detailed multicomponent droplet evaporation model. The authors implemented a simplified droplet evaporation model, which could lead to uncertainties in flame height. Furthermore, the real wellhead conditions with heavy fuels may not be realized with these simplified assumptions.

The authors provide only trends for such parameters as location of flame anchoring and height of the flame, and no detailed results from simulation and experiments, or their comparisons. Additionally, they conducted their investigations within a narrow range of flow conditions, so it is unclear whether the results

Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×

can be related directly to the high-flow-rate condition of the wellhead, where blow-off may occur. Limiting conditions for ignition need further exploration as well.

The authors developed a reasonably robust code that could be used in the future to investigate various parameters. More accurate submodels could also be included in future work.

Does the research product determine the primary mechanism driving burn efficiency?

The authors investigated the breakup dynamics of the liquid phase and the gas-phase combustion of atomized droplets to obtain the “burn efficiency.” Results of the intermediate-scale experiments show that the large droplets fall back to the ground because of their weight. In the modeling approach, some of the underlying physical processes were captured, albeit with empirical correlations. But the experiments did not address the extent to which the droplet size distribution might be skewed by pipe exit geometry and fracture protrusions. The structure of the exit may need greater consideration given that the dropout of large drops appears to be significant.

The flamelet model with a progress variable is a widely used model, but has large uncertainties caused by significant assumptions that are likely wrong. The model relies only on normal rate of strain in the flow and neglects any effect of shear strain and vorticity. At the same time, there is no direct and clear relationship between strain in the Large Eddy Simulation (LES) field and scalar dissipation in the flame; only a vague connection is made in flamelet theory. Unfortunately, a better model is not quite yet available. Still, the NRL report needs to identify the inherent weaknesses of the flamelet model.

The axisymmetric nature of the model may also pose some limitations. Depending on the exit conditions at the wellhead, nonaxisymmetric gas-phase plume structure may evolve in realistic wellhead fire scenarios. Axisymmetric models such as that presented in the NRL report cannot capture the burning efficiencies of such cases. The ambient wind conditions and directions can significantly affect the axisymmetric assumption.

Gas-phase models may not be adequate with inherent evaporation/atomization assumptions. Droplet size distribution is input to the gas-phase combustion model and is a questionable choice. The postulated worst-case scenario, which assumes a slower gas velocity, leads to lower liquid atomization and lower burn efficiency. However, higher gas velocities could lead to more entrainment, and gas-phase combustion could become the controlling mechanism.

The authors concluded that the remnant fuels on the ground are primarily from large liquid fragments, which were not entrained into the flame and settled because of heavier weight. However, the choice of surrogate fuel significantly affects this observation. The mismatch between the distillation range and particulate mass generation potential (and radiation effects) between the simpler surrogate fuels used for the NRL study and the actual crude properties will likely influence the predicted burn efficiency. Particulate mass generation should include all relevant crude oil combustion particles (e.g., liquid-phase coking, ash, sand, rock). The crude oil is expected to have higher particulate mass generation propensity. Large-scale coke particles will also contribute to the solid/liquid accumulation on the ground.

Sensitivity analysis of the submodels is critical to ascertain the controlling physics that determines the burn efficiency. The NRL report does not provide the needed parametric studies from the computational simulations or the experiments. The current model could be used to perform parametric analysis and determine the sensitivities of various physical processes that control the overall burn efficiency. Furthermore, such analysis in conjunction with experiments could be used to develop a more realistic “burn efficiency” definition that could be used in practical situations.

Were the conclusions based on the OSRR 1063 study findings in the report logical and appropriate based on the results?

This is a very difficult research problem, and the authors performed foundational work for modeling and experimental research on some of the physicochemical mechanisms for physically downscaled wellbore processes. They identified some of the important aspects to be considered and developed some

Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×

foundational understanding of physical and chemical processes relevant to wellbore ignition and combustion problems. Details and scaling remain to be done, so the applicability of this work to full scale is still unclear.

At the core, there are concerns about the omission of certain information in the model—e.g., buoyancy, droplet size and velocity distributions, spray falling to the ground, and the potential for a fire whirl—that would have been relevant to the resulting burn efficiency. Additionally, even in the CFD model, a great deal of uncertainty in the submodels was not characterized. Submodel assumptions were not validated, and grid convergence and numerical artifacts also were not well characterized with respect to the submodels. It is therefore difficult to determine whether discrepancies exist and if so, whether they are attributable to numerics and resolution versus inadequacies in the model. In addition to validating the submodels, performing a sensitivity analysis would elucidate the impact of the different submodels on the end result. At this stage, this model is not predictive.

The bench-scale experiments the authors performed were largely experimentally correct, with competently executed measurements (although there were issues with the 3CHIP measurements). However, the authors conducted no diagnostics other than imaging for intermediate scale. The results also do not link clearly to larger scales, from lab scale to meso/room size. Extrapolating these results to the wellhead is problematic, in no small part because of differences in the behavior of the model oil chosen and crude oil. The work would have benefited from a range of nondimensional parameters and some idea about effects of different physical and chemical properties of n-heptane relative to the range of crude oil properties expected in the field.

Overall, the results and comparisons between the experiments and computations are limited, and their relevance to real-world wellhead oil burning is questionable. Many of the conclusions are observational and not in dispute. Conclusions assessing whether this technique will work in the field are lacking. This body of work does not provide a concrete foundation for determining whether wellhead burning is sufficient for mitigation of uncontrolled environmental release of oil in the event of loss of well control.

Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×
Page 5
Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×
Page 6
Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×
Page 7
Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×
Page 8
Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×
Page 9
Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×
Page 10
Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×
Page 11
Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×
Page 12
Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×
Page 13
Suggested Citation:"2 The Committee's Responses to the Charge Questions." National Academies of Sciences, Engineering, and Medicine. 2021. Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales. Washington, DC: The National Academies Press. doi: 10.17226/26211.
×
Page 14
Next: 3 Conclusions »
Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

Peer Review of Interim Report on Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales reviews OSRR 1063: Bureau of Safety and Environmental Enforcement Report: Computational Fluid Dynamics Model for Predicting Wellhead Oil-Burning Efficiency at Bench and Intermediate Scales: Interim Report (July 30, 2020), produced by the U.S. Naval Research Laboratory (NRL) and funded by the Bureau of Safety and Environmental Enforcement (BSEE). Specifically, this report assesses the technical quality and completeness of the NRL report; the assumptions and approach used to develop the computational fluid dynamics model; and the completeness of the modeling results and experimental validation as an evidence base for determining whether wellhead burning is sufficient for mitigation of uncontrolled environmental release of oil in the event of loss of well control.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    Switch between the Original Pages, where you can read the report as it appeared in print, and Text Pages for the web version, where you can highlight and search the text.

    « Back Next »
  6. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  7. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  8. ×

    View our suggested citation for this chapter.

    « Back Next »
  9. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!