Ian A. Frigaard, UBC
June 8, 2016
Executive Summary
This is a final report for work on the project “Fluid Mechanic Causes of Gas Migration”, which has been conducted at the University of British Columbia, Department of Mechanical Engineering, primarily by Ms Marjan Zare under the supervision of Prof Ian Frigaard. This work forms the main part of the PhD thesis of M. Zare, (due to be examined in Summer 2018), which will then be a publicly available permanent record of the research.
This work has been funded by the Alberta Upstream Petroleum Research Fund Program (AUPRF) through PTAC (project number PTAC-16-WARI-02), from October 1st 2016 to June 30 2018. The work continued a research project of the same name funded by BCOGRIS (project number EI-2016-09), from October 1st 2015 to September 30 2016. Earlier work on this project was initiated under a larger collaborative research and development grant between Schlumberger and NSERC (CRD project number 444985-12: “Topics in Oilfield Cementing Fluid Mechanics”), which addressed wider range of technical challenges. M. Zare commenced her PhD studies on the project in 2013.
The project concerns the in depth study of a topic of critical importance to the upstream oil & gas industry: namely the leakage of oil & gas wells. This is a common occurrence worldwide and leakage rates have been reported in the Canadian context by various authors; see Dusterhoft et al (2002), Dusseault et al (2014), Atherton et al (2017). Broadly speaking, an average of 10-20% of wells in Western Canada leak. There appears to be significant variability depending on geology, operational factors and of course the criteria used to define leakage. The concerns are various, depending on the stakeholder: reduced productivity (due to reservoir pressure reduction), environmental (emissions and near-surface ecology), health (groundwater pollution or emitted gas toxicity), public perception, regulatory effectiveness.
Although there are instances of gas detection distant from a wellbore, by far the most common is near-wellbore leakage: through the cemented surface casing (SCVF). This means that pathways exist through the cemented annular gap that are more permeable to reservoir gas than any other pathway from gas source to surface. Although of topical interest, this is not a new phenomenon and has received industrial attention for decades. The complexity arises from the fact that many different physical phenomena and operational procedures can lead to gas migration.
In this project we have selected 3 problem areas of relevance to gas migration, in each of which the underlying process is governed by fluid mechanics.
The main results achieved in the project concern the first two of these areas and are explained in this report. For the 3rd problem, our results have mainly been in formulating a mathematical model for this complex process, but so far we have not progressed further with results of industrial relevance. This 3rd part will only be documented in the PhD thesis of M. Zare.
Best practices and tangible benefits are summarized in separate sections of this report. Below we explain the main results of problems 1 & 2.
Problem 1: Micro-annulus formation during mud removal
Project 1 gives a concise computational and model based study of the fluid mechanical phenomenon of a wet microannulus. This is a layer of drilling mud that is left behind on the walls of the annulus following primary cementing. By concise we mean that the study reduces the dimensionless parameter space from 12 to 4 parameters, which is still complex. The simplification with respect to the annulus is also geometric, in that we consider a long vertical channel which can be interpreted as a longitudinal section of a narrow annulus. The detailed results of the paper are in Zare et al (2017) and Zare & Frigaard (2018b). More than 1000 two-dimensional (2D) detailed computations have been carried out, coupled to simpler models and analysis.
Density stable displacements are most common, i.e. the heavier fluid pushes the lighter fluid upwards. For these we generally see stable flows. Static mud layers are easily found in our simulation study, when the yield stress is strong enough to overcome the combination of buoyancy and frictional stresses. In post-processing our simulations we can compute the residual layer thickness ℎ at the end of the channel. Many of the parametric effects we have observed are intuitive: increasing displacing fluid viscosity improves the displacement, as does increasing the buoyancy force. Counter-intuitive however, is the decrease in ℎ as the yield stress of the drilling mud increases (an effect most evident at small to moderate buoyancy). As well as the thickness ℎ, a critical feature is whether or not the wall layer is moving. This depends on both ℎ and ℎ??????, which is the maximal static wall layer thickness. The latter is easily computed for density stable displacement. We have developed various techniques to qualitatively predict mud layer behavior – summarized later in the report. We also have derived conditions under which there can be no wet microannulus
Density unstable displacements occur with some lightweight spacer fluids and with washes (although these are often in turbulent flow). The main differences that arise are: (i) a range of hydrodynamic instabilities are found (detailed in the report below); (ii) we find that static mud layers can result even for very low yield stresses. This can happen because buoyancy acts against the mean flow. Thus, with thick enough wall layer and strong enough buoyancy force we are able to have wall shear stresses that are very small. This allows a yield stress fluid to remain static near the wall at smaller yield stresses than would be the case for a density stable displacement. We have observed these layers in our 2D simulations.
We have developed a 1D analysis, similar to that for density stable displacements, which allows us to define a maximal static layer thickness. This is however slightly more complex than the density stable case to compute. We have also been able to develop a 1D analysis that can successfully predict when the more complex 2D flows will be stable or unstable. The hydrodynamically unstable flows we have observed are disappointing from the perspective of mud removal. We had hypothesized that the instabilities would result in thinner residual mud layers, but there is no strong evidence for this in our study. Overall, we feel that density unstable flows do not have any positive effect on wall layers/wet micro-annuli, and probably the reverse. A simple criterion developed for eliminating static wall layers in stable displacements is not valid for density unstable displacements and we must balance the (unproven) potential for destabilization with the demonstrated ability of having partially static residual layers as we increase the buoyancy.
A number of recommended future directions are given later in the report.
Problem 2: Fluids invasion
We have conducted lab-scale experiments targeted at exposing the effects of the yield stress on the (over-)pressure required for one fluid to invade a column of another fluid, through a small hole. This setup was designed to simulate invasion into a well during primary cementing, for relatively low porosity reservoirs where pores may be considered isolated, e.g. in our experiment the hole radius was ~1% of the column radius.
For miscible fluids our results show approximately linear increase in the required over-pressure for invasion, with both the height of fluid column and the yield stress of the invaded fluid column. A number of interesting stages have been identified during the experiments: mixing, invasion, transition, fracture and arrest. The passage from invasion to transition pressure seems to represent elastic-plastic yielding close to the invasion hole. The initial growth of the transition dome and then slowing of growth (in cases where the fracture does not start immediately) suggests a relaxation of the stress field, due to the over-pressure now being applied over a larger area. We have seen that the expanding dome interface can be either relatively smooth or granular. This does not appear to have any bearing on the stability of the transition dome: either may be stable or unstable. Invasion and transition domes are approximately axisymmetric. Fracture initiation and propagation represent a departure from symmetry, probably due to either a local defect or a non-uniformity of the stress-field.
Our tests with glycerin solutions, comparing against the water invasion studies revealed: (i) the glycerin solutions have a slower mixing/invasion stage (longer times before transition); (ii) invasion pressures are consequently higher; (iii) after transition the invading domes grow more rapidly and are significantly larger than those for water at the point when fracturing initiates. These differences are partly attributed to the larger viscosities of the glycerin, which we believe slows diffusive/mixing processes. On the other hand, there are many similarities to the water invasion studies. The invasion stages are similar and for all miscible fluids tested invasion is a strongly localised process. The increase in invasion pressure with column height is also far slower for miscible fluids than the Poiseuille flow limit would suggest, confirming the local nature of the flow.
Our experiments with immiscible fluids involved a density matched silicon oil (R550) and air. It was immediately apparent that interfacial tension had a large effect on invasion. Our experimental protocol was changed to increase the invasion pressure in larger steps than for the miscible fluids. Mixing was eliminated as was the occurrence of an initial micro-invasion dome. Instead, once invasion started we observe a clear interface expanding into the Carbopol column. In the case of the R550 the interface evolved as an approximately hemispherical dome, slowly filling the bottom of the column. The invasion pressure was significantly larger than for the glycerin solutions and was nonlocal in that it was resisted by yielding at the walls of the column, increasing as the Poiseuille flow.
The air also did not mix and had no micro-invasion dome. However, the invasion was localised in the sense that buoyancy dominated after any significant influx, leading to the invading fluid stretching upwards into a long bubble, which eventually detached and propagated to the surface. Thus, resistance of invasion at the walls did not occur and the invasion pressure was approximately constant with height of column. Also the measured invasion pressure was much larger for air than for R550. Further analysis of this increased invasion pressure suggested that about 50\% of the increase is directly attributable to interfacial tension and the remainder to yield stress. However, the yield stress is only able to resist effectively when there is no mixing/diffusion.
A number of interesting avenues suggest themselves for future work in this area, outlined in the recommendations section.
# 16-WARI-02