Selection of Kill Fluid involved formulation of the required properties and laboratory testing the process of buoyant settling (Demirci et al., 2017). Several hydrophobic fluids were evaluated using criteria of buoyant slippage (Archimedes Number, A_r >>1), immiscibility (Partitioning Coefficient, P>>1), and slippage velocity (large stable droplet size). Table-top slippage testing identified brominated organic liquids as the best KF candidates for the design considering their variable viscosities and high densities. Densities of these Newtonian fluids could be adjusted between 11 and 25 ppg with viscosities range from 0.46 to 3,057 cP at temperatures up to at least 150°C. Furthermore, their partitioning coefficient values vary from 2 to 7, which makes the fluids highly hydrophobic. Specific design of KF shall consider known properties of the AF and general rule that an ideal KF require maximizing the density of the KF, maximizing the interfacial tension between the KF and the AF, and maximizing the partitioning coefficient of the KF.

Initial testing of the fluids’ buoyant settling performance involved pouring the fluid at the top of 4′ diameter and 10-ft long vertical pipe filled with water. No mixing of the KF with water was detected. It was observed that submerged injection of KF is necessary to avoid air entrapment by the settling fluid droplets that reduce slippage velocity and would delay the displacement process.

The actual design of KFs was performed in the bench-top compatibility testing in the custom-made vertical (3-ft tall, 4″x2″ diameter) custom-made glass annular models filled with a simulated (translucent) unweighted drilling mud (Demirci et al., 2017). Multiple tests revealed kill fluid viscosity being an important property that controls dispersion and formation of a mixture zone on top of the settled kill fluid column.

Based on the bench-testing results a 11.85 ppg low-viscosity (8 cP) kill fluid was selected for the field-size test to enable efficient cleaning the well after testing. The annular) fluid (AF) was a slightly-weighted bentonite drilling mud having 8.6 ppg density. Table 1 shows properties of the two fluids.

A total of 32 barrels of kill fluid (KF) was blended at the testing facility using three 550-gallon capacity oilfield mixing tanks. Also, a 20-barrel oilfield trip tank was used to prepare the annular fluid (AF). The mixed drilling mud was stored and hydrated in the tank for a week.

The KF injection was studied by considering injection geometry and the rate of stable settling of KF in the stagnant column of AF. The study involved bench-top testing with the slot model of annulus, described above (Demirci and Wojtanowicz, 2015; Demirci and Wojtanowicz, 2016a) and the pilot-scale testing with the annular 30-ft tall model, also mentioned above (Demirci and Wojtanowicz, 2016b). The slot-model tests utilized KFs and AFs with different viscosities and densities. They also considered two configurations for injecting KF at the top of the AF column: a sideward or downward injection. The results showed that sideward geometry was superior to the downward injection due to the impingement effect that provides stable settling of KF fluid droplets at higher injection rate without their break-up and partial loss in the upward overflow of AF.

The slot-model study also provided empirical formulas for maximum sideward injection rate for field-scale process design. Eq. 1 below gives the rate required for desired impingement flow and Eq. 2 gives the maximum rate for undesired atomization flow regime. q cr − i = 1.21 ∆ ρ μ kf σ 1 3 ρ kf (1) q cr − A = 15.76 μ kf   d n 1.4   σ 0.4 ρ kf 0.6 3 μ kf + μ af 0.8 (2)

Understanding of the injection rate effect was improved with the pilot scale 20-ft high physical model of casing annulus designed and manufactured at the Albemarle Process Development Center. The model featured a 6-in [6 5/8-in. outer diameter)] carbon-steel pipe inside an 8-in. [8.329-in. inner diameter] carbon-steel pipe (Demirci and Wojtanowicz, 2016b). A total of 37 experiments were conducted with water-based AFs and brominated organic KFs with densities from 9 to 13 ppg and from 17 to 23 ppg, respectively. Also, an upward gas migration was allowed in some experiments having KF injected with no pressure at the top of the AF column.

The results showed that gas migration would prevent buoyant settling and fluid displacement at any injection rate. Thus, the buoyant kill process must be performed in the pressurized annulus to eliminate gas migration. Moreover, buoyant settling of KF produced a two-fluid mixture column on top of the settled KF column at the bottom of the annuls. The mixture length would build in time and its final size would increase at higher injection rate and smaller density difference of the two fluids. Density of the mixture column changed upwards exponentially from KF density to AF density. The results also indicated that more than one annular volume of KF would be needed to replace the AF in the annulus.

The experiments provided data for derivation of a pressure replacement model that would predict hydrostatic pressure increase at the bottom of the fluid column resulting from AF replacement with KF. A dimensionless parameter, R_u, is the ultimate displacement and represents the ratio of required kill fluid volume to annular volume for maximum annular pressure increase. The parameter value was empirically correlated using the experimental data (Demirci and Wojtanowicz, 2016b) as, R u = 10 − 0.376   ρ af 0.377   ρ kf 0.035   τ 0 0.019   q 0.1719 (3)

The formula shows increased usage of KF for higher injection rate caused by dispersion and loss of KF in the AF overflow. It also implies that there is a maximum low value of injection rate that would minimize volume of KF needed for maximum annular pressure increase due to efficient buoyant settling of KF without its partial dispersion in AF. Maximum injection rate is determined from the buoyant transport model based on the theory of fluid droplet slippage shown in Supplementary Appendix S1 as, q max = q cr − T = 4.92   A 1.85 σ ∆ ρ − τ 0 C D 0   ρ af (4)

Figure 2 is a photograph of top installation at the wellhead of Well#1. Downhole installation schematics is depicted in Figure 3. The wellhead photo in Figure 1 shows the KF injection pump with the KF feed line and 2-in. outlet pipe for overflow of the displaced AF. Also shown is the pipe for downhole injection of gas into the tubing (through the gas-lift valve) so the gas would migrate upwards in the tubing filled with AF liquid. The 2,750-ft long 2–7/8″ (ID = 2.441″) tubing shown in Figure 2 simulates column of AF having 16 barrels volumetric capacity. The tubing’s bottom represents the cement top conditions in actual well with the gas injection valve and two pressure transducers.

The feed line for KF injection is a 0.5″ (ID = 0.375″) diameter microtubing stretching from the injection pump to the wellhead’s top (Figure 2) and going 60 feet down the tubing with an elbow-shaped outlet to assure KF discharge geometry with impingement (Figure 3). Pressure and rate capacities of the positive—displacement chemically—resistant KF injection pump are 900 psi and 9–90 gal/hr, respectively. The 2-in. overflow outlet pipe is equipped with a sampling port and a conductivity-meter to collect fluid samples and measure composition of the overflow stream. A pressure-temperature transducer at the outlet pipe controls constant pressure at the wellhead. The overflow is discharged to a mud-gas separator.

1) Fill the tubing with 15.5 barrels of the mud by leaving 57 feet space at the top in order to create a gas cap and to assure submerged injection of kill fluid KF;

Shown in Figure 4 is the expected theoretical pressure change during the test: tubing fill-up with AF in Step 1; mud gelation in Step 2; pressure-controlled injection of gas in Steps 3; gas migration ending with simulated SCP of 350 psi in Step 4; and KF injection with step-wise bleeding of wellhead pressure in Step 5.

As learned from the previous tests, the efficiency of the displacement process strongly depends on the injection rate. The rate was designed using data from the bench-top testing with the slot model, the theoretical considerations, above, and additional lab verification.

The slot-model study (Demirci and Wojtanowicz, 2016a) included three runs with fluids in Table 1. All the runs showed high volumetric and pressure replacement efficiencies with the model’s annular capacity 1.05 gal/ft. To obtain similar performance in the field-scale test (tubing capacity 0.24 gal/ft) injection rate conversion is made by assuming the same settling velocity shown in Table 2. The assumption ignores different flow geometry in the slot model and the tubing with potential effects such as the whirling flow of heavy fluid through a lighter fluid discussed elsewhere (Calvert et al., 1995).

Settling velocity criterion is further verified in the flip-flop experiment using a 4-foot long clear PVC pipe having diameter 2.5″ (2.323″ ID) and capacity (0.23 gal/ft) similar to the well’s tubing. In the experiment, the kill fluid in Table 1 was placed at the pipe’s bottom while filling the rest with AF form Table 1. Then, the pipe was flipped to have KF on top of AF and the displacement process was timed as shown in Figure 5.

After pipe flipping, the KF/TF interface broke and a plume of kill fluid started to slip downwards. The fluids’ interface produced KF droplets that settled down much slower than the plume. It took 10 s for the two fluids to be fully replaced so the required KF injection rate is estimated as, q = 60 ∙ 0.3 gal / 10   sec = 1.8 gpm . Shown in Table 3 are estimated values of the KF maximum injection rate based on all criteria.

Another practical criterion for the KF injection rate design was the experiment’s duration time. Using the best-performance (minimum KF volume) criterion (Eq. 3) the time needed for displacing 16 barrels of AF at different injection rate is shown in Figure 6. Considering the importance of atomization criteria and trying to keep the operation time below about 12 h the KF injection would begin at 1.5 gpm with visual inspection of the KF content in the overflow in order to reduce the rate until only AF is present.

The pneumatic positive displacement pump used for KF injection generates pressure pulses that result in fluctuation of flow rate. (The fluctuations were averaged in the flow meter readings.) The fluctuating rate repeatedly exceeded the maximum designed flow rate by going above the atomization flow regime threshold (Table 3) which resulted in dispersion of KF droplets. Instead of traveling to the bottom, the droplets floated in the well’s fluid column plugging it off and disrupting the KF settling transport.

Accidental wellhead’s decompression resulted in gas migration and entrapment in AF. Releasing the well-head pressure would reduce the pressure on top of trapped gas bubbles and allow them to expand. Expanding gas bubbles would; 1) decrease the density of the fluid column, 2) start migrating upwards. The similar phenomenon was actually tested in the pilot scale experiments where KF injection during gas migration resulted in poor displacement (Demirci and Wojtanowicz, 2017). In this experiment, gas migration resulted from temporary failure of the auto choke causing wellhead’s decompression and bottom gas invasion. The migrating gas caused flotation of the emulsified mixture column around the KF exit.