1. Introduction
The pressure at the casing head of a well that rebuilds even after the bleed down is called Sustained Casing Pressure (SCP) [
1]. By definition, SCP is not imposed by the operator or caused by temperature fluctuations. The main cause of SCP is a leaking cement sheath through which the gas flows through the annulus from a high-pressure formation. A column of mud may exist in the annulus on top of the cement sheath. The gas percolates through this column and accumulates at the casing head causing excess pressure at the surface.
In the US, Mineral Management Service—presently the Bureau of Safety and Environmental Enforcement (BSEE)—established regulations to address sustained casing pressure in oil and gas wells (BSEE 30 CFR Part 250, 2011). They developed the rules for managing SCP and established criteria to monitor and test these wells.
The B/B test begins with the pressure bleed-down stage when the casing head pressure is vented out through a half-inch valve. Once the pressure drops to zero or stabilizes at a certain value the needle valve is closed so the pressure rebuilds for 24 h. Most typical plots of B/B tests are depictedin
Figure 1 for the case when the annular gas leak is large (
Figure 1a) or small (
Figure 1b). If the leak is large buildup pressure would stabilize at its initial value in 24 h. In case of small leak, ther would be slow pressure buildup with more than 24 h needed for pressure stabilization.
American Petroleum Institute (API) [
2] published simplified diagnostic criteria for B/B test analysis published in API RP-90 and 10 CFR 250 Subpart E. The criteria only consider if the pressure would bleed down to zero and rebuild to its initial value within 24 h time period. As shown in
Figure 2, when the pressure bleeds down to zero and builds back up slowly to a low valuewithin 24 h the rate of gas leak is small and the environmental risk of breaching the pressure containment barrier is considered acceptable. Moreover, a gas leak is considered large when the pressure drops to zero and rebuilds quickly (within 24 h). In case when pressure does not bleed to zero and quckly rebuilds the leak rate is high and so is the environmental risk of the containment failure.
In addition to pressure, API also recommends measuring and recording the amount of fluid recovered during the B/B test. In all, the API approach is mostly qualitative without offering a specific method for sizing the gas leak value. Moreover, API does not have a specific protocal for the B/B testing and the testing procedure depends on the operator. (API RP-90, 2006).
Quantitative analysis of B/B tests has been developed by modifying the mathematical models proposed by Xu et al. in 2001, 2003, and 2017 [
3,
4,
5,
6]. The model developed by Xu et al. (2001) [
4] assumes a Newtonian fluid in the annulus and gas flow is modeled based on this assumption. As a result of such an assumption, the gas is not trapped in the annular fluid sothe fluid’s compressibility remains constant.The gas migration time through the annular fluid is also ignored. Once the gas reaches the top of the annulus, the gas accumulates and forms a gas chamber (below the casing head and above the fluid level). The model calculates the gas chamber pressure at each time step using an analytical formula. The effect of well and formation parameters including the formation pressure, cement leak conductivity (generalized as cement permeability), mud compressibility, and casing gas chamber on SCP buildup pressure can be evaluated using the model. The 2001 Xu model produced oversimplified predictions.
In 2002, Xu et al. [
3] modified their model by considering the two-phase gas flow in a non-Newtonian (power-law) fluid. A transient gas flow in the leaking cement was assumed and coupled with the flow in the annular fluid. In the modified model, the pressure bleed-down stage of the B/B test is represented separately from the build-up stage. The unknown parameters of the system are the size of the gas chamber, high-pressure formation depth and pressure, annular fluid compressibility, and the permeability of the leaking cement. These parameters are found using a trial and error process. The bleed-down and buildup model matching are separated from each other. The improved model was also used to identify typical patterns of the B-B test pressure change by matching data from SCP field testing [
5]. Matching all parameters together in this analysis method causes ambiguity in the obtained unknown parameters.
The 2001 build-up model was utilized by Huerta et al. [
7] to determine the gas source pressure, cement permeability, and depth for two fields with CO
2 leakage issues. The annular fluid was assumed to be Newtonian and therefore the gas entrapment was ignored. This assumption can significantly alter the pressure buildup stage analysis as the mud compressibility notably affects the gas chamber pressure. From known values of the length of Newtonian annular fluid and stabilized casing head pressure, the gas source pressure was calculated hydrostatically. The computations required making educated guesses about the depth of the gas source. The best match between the field pressure buildup data and the model was obtained by iterating the cement permeability. The best match comes from repeated guessing. The best match determines the values of the gas source depth and cement failure severity (permeability). Another assumption made in Huerta’s et al. [
7] study was that the gas chamber did not exist at the start of the buildup stage (zero volume). It was assumed that the gas chamber forms by compressing the annular fluid. This assumption may not be accurate as the free fluid level in the annulus is usually unknown and must be determined by analyzing the B/B test.
Another similar study to Huerta et al. [
7] is the work of Tao et al. [
8]. They used the same hydrostatical method as Huerta [
7] (discussed above) to determine the source pressure with the assumption of free gas mud column. The annular fluid compressibility was determined using its correlations with density. The cement top depth, pressure buildup history, and mud density were used to determine the depth of the high-pressure formation and permeability of the cement. The permeability determined from Tao’s [
8] study was very small as the pressure buildup happened over 400 days while the B/B test pressure buildup data is for 24 h. The Monte-Carlo simulations were performed to determine the cement permeability for assumed boundary values of depth. Similar to Heurta [
7], the best model fit to pressure buildup history provides the values of gas source depth and effective permeability.
Zhu et al. [
9] confirmed the Xu [
5] model by using the model and applying it to once SCP field B/B test. They did not discuss the fitting method or the number of unknown parameters in their simulations.
Another form of SCP study with a focus on the maximum air emission rates (MER) in the case of a failed casing due to SCP was performed by Kinik et al. [
10]. Since the casing was considered failed, atmospheric pressure was assumed at the casing head. The gas flows linearly through the cement column reaching a stagnant Newtonian annular fluid. A two-phase flow of gas and water occurs in this column of fluid reaching the surface with atmospheric pressure. The two-phase flow in the annular fluid was modeled using Caetao et al.’s [
11] mechanistic model which provides the criteria for flow regime transitions and pressure gradients. As the gas bubbles percolate through the annular fluid, their volume increases especially in the top section of the mud column which results in liquid unloading. This liquid unloading is considered in Kinik’s model by computing the gas and mud total volume at each time step and comparing it with the annulus volume. Kinik’s model is different from other SCP models as it assumes that the cement permeability, initial volume of the gas chamber, and the gas source formation pressure are known from the B/B test analysis. Therefore, the model is suitable for open annulus testing.
In our more recent study (Yao et al. [
12]) we quantified the environmental risk of air pollution due to SCP by introducing two metrics to be determined from the B/B test: maximum downhole rate of gas leakage at the top of the cement sheath column (TOC), and the potential annual rate of atmospheric gas emission from leaking well with failed casing head. The second metric considers a two-phase flow in the fluid column above TOC. We also proposed threshold maximum values for the two metrics—adopted from the oil industry standards for downhole safety valves (maximum 15 scf/min downhole gas rate across the leaking cement), and from the regulatory limits for venting volatile organic compounds (6 tons per year annual gas emission).In our study, we verified the new metrics by analyzing field data from B/B testing of 19 wells. In the analysis, we used an in-house B/B test simulator to match all data from
the entire test for each well, i.e., pressure bleed-down and pressure buildup. The approach quantifies the severity of the gas leaks using the proposed metrics of environmental risk.
The results showed that 15.8% of wells exceeded the maximum downhole gas rate threshold and 26.3% of wells failed the annual gas emissions threshold. Moreover, values of the new metrics for the SCP wells that could be bled down to zero pressure were mostly smaller than the maximum threshold values indicating no need for the pressure buildup stage in such wells. It was concluded that the 24-h pressure build-up stage of the B/B test did not uniquely correlate with the risk metrics and, therefore, would not qualify as a unique measure of cement integrity. The conclusion seems questionable since it would disregard the physical effect of gas migration in the column of liquid above TOC. A more reasonable explanation comes from a hypothesis that globally matching all B/B test data would prioritize the pressure bleed-down stage over the buildup stage thus making the latter stage mostly irrelevant in providing information about the whole gas migration system. Therefore, the B/B test stages should be analyzed comparatively to remove ambiguity resulting from each of them alone.
In this work, we would verify the above hypothesis and evaluate the quality of information discerned from the B/B testing as a solution to the inverse problem, i.e., finding four parameters of the leaking well system including the compressibility and the length of the mud column and the permeability and length of the cement sheath. Using the analysis of variance and multivariate regression methods we identify significant parameters controlling each of three stages of the B/B test: pressure bleed-down, stabilization, and buildup. If the significance varies, then a stage-by-stage analysis of the 3-stage test would be more accurate than the presently—practiced method of matching all data from the two-stage testing. To properly represent the effect of gas migration in the liquid above TOC, we would improve the hydraulics of the B/B test simulator to consider a column of the yield-power-law fluid (Herschel-Bulkley) instead of the power-law (or Newtonian) fluid used in the previous models. A stage-by-stage interpretation approach is used to analyze three stages of B-B tests—pressure bleed-down, constant flow, and pressure build-up stage. The study would also include a sensitivity analysis of the three stages to identify the most important parameters affecting each stage of the test. Finally, a new procedure for B/B testing is proposed.
Finally, we would assess the procedure of B/B testing by considering three operational parameters (bleeding rate and time, and pressure recording time step). We demonstrate a significant effect of the parameters on the well response that may cause misinterpretation of test results—a premature closure of the bleed valve eliminates the pressure stabilization stage and deforms the pressure buildup causing erroneous cement leak size determination.
3. Sensitivity Analysis of System Parameters
In order to perform the sensitivity analyses, we need to identify different annulus configurations. Typical well completion includes cementing the annulus between two consecutive casings and the pressure gauge on all annuli should read zero.
Depending on failure type, different gas flow paths (
Figure 3) may result as follow:
If the casing failure happens to be at the annulus A casing shoe, the gas will flow through the liquid existing in annulus A. This annulus doesn’t have cement and hence, the only medium for gas to flow is liquid contained in this annulus.
- 2.
Flow through the cement only
Typically two possible configurations exist for an annulus (except annulus A); cemented to the top and an annulus containing a liquid column above the cement. In case the annulus cemented to the top, the only path for gas flow would be through leaking cement.
- 3.
Flow through both cement and liquid
In the case where both the cement and annular fluid columns exist, the fluid permeates through the leaking cement and then flows through the annular fluid to reach the casing head.
Figure 3.
potential leak paths resulting in SCP.
Figure 3.
potential leak paths resulting in SCP.
In an annulus containing a cement column and the mud column above it, the important parameters are gas source formation pressure (p
f) and depth (D
f), cement permeability (k), trapped gas concentration in mud (F
gt), the initial volume of the gas chamber (V
gci).
In annulus cemented to the surface, the interpretation of the B-B test includes determining these unknown values which are source formation pressure and depth, and cement sheath leak rate (k).
where P
f is the gas source formation pressure, k is the cement permeability, V
gc is the constant gas chamber volume, and D
f is the formation depth.
In annulus A, where only the mud column exists, the important parameters are the initial volume of the gas chamber, and trapped gas concentration in mud.
The sensitivity analysis is performed for the general annulus configuration (annulus containing both liquid and cement columns) by utilizing the multiple linear regression and a two-level factorial design. The procedure for the sensitivity analysis starts with selecting the lower and upper bound of each factor. The mathematical model is then used to determine the casing head pressure responses for those two levels. The linear relationship between the selected factor and the pressure response is then determined using the F-test. The importance of each factor to change the casing head pressure is eventually determined by performing the t-test.
Table 1 demonstrates the 4 independent variables assumed in this work that affects the casing pressure (dependent variable) during the B/B test. A complete factorial design was used in this study which results in 2
4 = 16 pressure bleed-down and buildup simulation runs.
The multiple regression model is
The regression coefficients (βj, j = 0, 1, 2, 3, and 4) are computed using the least square method that minimizes the sum of the squares of the errors. The linear regression model suffices in determining the important parameters affecting the dependent variable while a higher-order regression model would provide a more accurate description of the system.
Of four parameters that were tested only the initial volume of the gas chamber and trapped gas concentration in the mud had a significant effect on pressure bleed down (
Table 2). The trapped gas concentration is a function of the annular fluid yield point and can be determined independently as will be shown in this paper. Although the effect of formation pressure and cement permeability is observed in longer bleed downs, the only dominating parameter affecting the bleed-down stage is the volume of the gas chamber.
As shown in
Table 3, it is found that the buildup stage is mainly affected by the conductivity of the cement column (k), gas source pressure (p
f), and compressibility of the mud column (c
m). In addition to these, the volume of the gas chamber also plays important role in build-up, however; this parameter will be fixed and known after simulating the bleed down.
Figure 4 shows different stages of the B-B test and different parameters affecting each stage.
Sensitivity analysis demonstrates that a stage-by-stage analysis can be performed since each stage of the B/B test is not a function of all unknown parameters.