IODP Proceedings    Volume contents     Search


Expedition 327 tracer experiments

Tracer injection overview

Tracer injection during Expedition 327 was designed to occur as part of a 24 h pumping experiment in Hole U1362B, used to assess formation properties at a lateral scale greater than that permitted through more conventional 1–2 h packer tests (see the “Site U1362” chapter). A drill string packer has been used on the JOIDES Resolution most commonly for shorter term hydrologic testing, but because that tool is subject to deflation in heavy seas and this test was to last much longer than earlier tests, we used a casing running tool to seal the top of 10¾ inch casing in the throat of the reentry cone (Fig. F2D). A stinger made of standard drill pipe was extended from the base of the casing running tool to the end of 10¾ inch casing and was used to direct the tracer injectate into upper basement. Additional downhole seals were installed as part of earlier casing operations between the top of 10¾ and 16 inch casing (swellable packer and a seal sub with an O-ring), 10¾ inch casing and basement (cement), and 16 inch casing and across the sediment/basement interface (cement) (Fisher, Wheat, et al.). The integrity of these seals was verified during the 24 h pumping experiment by visual inspection of the reentry cone; no plume was seen being emitted during the experiment. Initial inspection of downhole pressure records during injection are also consistent with the hole being fully sealed. Additional confirmation will come from a careful evaluation of pressure records from nearby boreholes (to be recovered in summer 2011).

For several reasons, the tracer mixing and injection system for this experiment was designed to make use of standard shipboard equipment to the greatest extent possible. The pumping rate was planned to be 5–10 L/s (enough to generate a measurable pressure signal at distant boreholes), which required injection of ~500 m3 of fluid in 24 h. It would have been prohibitively expensive and complex to build a completely separate tracer mixing and injection system that was compatible with shipboard equipment and capable of handling this pumping rate and total volume. A custom mixing and injection system could have been made cleaner than the shipboard plumbing, but we would still be pumping the fluid through rig hoses and fittings and nearly 3 km of drill pipe coated with rust, pipe grease, and other contamination. Also, we needed to be able to set and control pumping and injection rates from the rig floor and to hold these rates constant. We planned to use a gas tracer during most of the 24 h injection period. It is better for gas to be injected under pressure than mixed in an open tank where there could be loss to the atmosphere. We planned to inject solute and particle tracers in short pulses to maximize injectate concentrations, and this required mainly that we introduce the tracer into the drill pipe quickly. To the extent that there is contamination during tracer injection, this can be assessed through analysis of pre- and postinjection concentrations and comparison of shipboard and downhole samples. As described later, injectate samples were collected on the ship and downhole throughout the injection period, which will allow postcruise analysis of the masses and concentrations of all tracers.

Fluid and tracers were injected using two pumping systems: mud pumps and cement pumps (Fig. F3). Mud pumps are used routinely during coring, hole cleaning, and other downhole operations on the JOIDES Resolution and are located on the lower tween deck, four levels below and slightly aft of the rig floor. Surface seawater is the primary drilling fluid during most operations, but freshwater can be delivered from a holding tank (either filled on shore and transported to sea or generated using the shipboard desalination system). Salt- and freshwater muds are commonly used on the JOIDES Resolution to increase the density and viscosity of the drilling fluid, which can help to circulate cuttings from the hole. Most conventional drilling rigs have mud returns to the platform, which allows filtering, collection of cuttings, and reuse of drilling mud through repeated circulation. There is no fluid return system on the JOIDES Resolution, so any drilling mud pumped into the hole flows either into the formation or up and out of the hole at the seafloor. No attempt was made to clean the mud pumps, intake, hoses, or other components prior to the injection test, which followed an extended period of seawater pumping.

The driller operates the mud pumps using a series of rotating switches in the driller’s shack adjacent to the rig floor, with separate gauges reporting the total amount of fluid pumped and the current pumping rate in units of strokes and strokes per minute (SPM), respectively. Pumping data are also recorded with the Rig Instrumentation System (RIS), which was configured to collect data at 1 s intervals during Expedition 327. The JOIDES Resolution’s pumps are Oilwell A1700-PT triplex units, each of which has three pistons attached to a crank shaft. The displacement of the pump is equal to liner bore area × plunger stroke × number of pistons × crank shaft revolutions × pump efficiency. As currently configured, the mud pumps use three 16.5 cm diameter pistons and liners and have an efficiency of ~95%, such that 1 SPM = 5.2 gallon/min = 0.33 L/s. The pumps generally require operation at a minimum rate of 10–12 SPM (3.3–4.0 L/s), and with both pumps online simultaneously (as during hole cleaning operations during Expedition 327), the typical pumping rate is 80–120 SPM (26.7–40.1 L/s). Fluid displaced with the mud pumps rises up a standpipe on the aft starboard side of the rig floor and is routed either up the rig and back down through the top drive or through a circulation head and directly into the drill pipe. The former path was used for all tracer injection operations during Expedition 327. A pumping rate of 20 SPM (6.7 L/s) was used through most of the 24 h pumping and tracer injection test. This pumping rate was considered to be great enough to generate a significant single- and cross-hole response in this area (Becker and Fisher, 2000, 2008; Fisher et al., 2008), and faster pumping rates would have contributed to greater tracer dilution prior to introduction of injectate into the borehole.

The JOIDES Resolution’s cement pump system (Fig. F3) is used to mix cement that is subsequently injected into the drill pipe at the rig floor using pipes and hoses that are independent of the mud pump system. The cement mixing and pumping system is located on the upper tween deck, one deck up and directly above the mud pumps, three levels below the rig floor. During a normal cementing job, dry cement is added to water, and the mixture is circulated locally until desired properties (density, viscosity, and composition) are achieved, after which the cement is pumped up to the rig floor and injected into the drill pipe. The cement mixing tank is relatively small, so when completing a typical cementing job (20–40 bbl, 3250–6500 L), there is usually a steady inflow of water and dry cement and an outflow of wet cement until the total desired quantity has been pumped. The two cement additive tanks, each having a volume of 10 bbl (420 gallons, 1630 L), can be used to dissolve or suspend additional compounds that are added to the cement slurry during mixing and injection. We used these additive tanks to pre-mix soluble salts and fluorescent particles prior to injection as part of the Expedition 327 tracer experiment.

The two cement pumps are controlled from the cement room using a rotating switch, and a gauge in the driller’s shack indicates total volume pumped and pumping rate of two separate pumps, also in strokes and SPM. Like the mud pumps, the cement pumps are positive displacement pumps, so fluids injected using this system are added to whatever is being displaced with the mud pumps. The cement pumps have smaller pistons and liners, and the volumetric pump rate conversion is 1 SPM = 1.6 gallon/min = 0.104 L/s.

Prior to the start of the tracer injection experiment, the cement mixing system was circulated and cleaned with detergent to remove rust and residual cement and mud from the tanks, hoses, and pipes. This system is inherently dirty, and complete cleaning is impossible. Instead, we worked with the rig crew to identify the dirtiest and most accessible parts of the mixing and injection system and to clean these repeatedly to remove the largest potential sources of contamination. Cement mixing tanks were flushed repeatedly between brief periods of solute and particle tracer injection. We collected injectate samples throughout the 24 h pumping period. During most of this time, no dissolved or particle tracers were added to the injected fluid, which will allow a postcruise assessment of contamination for the tracers used and an assessment of system retention time following the end of injection of each tracer.

Neither the mud pumps nor the cement pumps can be set in advance to pump at a particular rate. Rather, the driller or cement pump operator adjusts a rotating switch to get the pumps started and set an initial rate. The rate is subsequently adjusted by hand until the rig floor gauges show that the desired rate has been achieved. Adjustment typically takes 1–2 min. Both pumps will stall out if the pumping rate is set too low. If a low pumping rate is desired, one must generally set a relatively high initial pumping rate to get the system started and then reduce the rate and verify that the system continues to run. In addition, separate valves must be shifted to pump freshwater rather than seawater and to add fluid from the cement pump system to the flow of fluid from the mud pumps. All of these activities require careful telephone communication between the rig floor and the mud pump and cement pump rooms, several decks below the rig floor, and this can be complicated by the noise generated when either of these pumping systems is in operation (strong hearing protection is required).

The tracer injection sequence was designed to maximize the initial concentrations of tracers in injectate during the experiment, even though this meant that solute and particulate tracers were injected during periods of only a few minutes. It can be helpful to extend the period of injection to learn more about tracer transport properties, but we opted for a shorter injection period for most tracers because of the cost and difficulty of injecting large quantities of solute and particles for 20–24 h and the extent of anticipated dilution and limits on detection (Table T1). It was much easier to inject the gas tracer for an extended period, as we did, and comparison of results based on longer and shorter periods of tracer injection should be helpful for planning future experiments.

Sulfur hexafluoride (SF6) tracer


Sulfur hexafluoride (SF6) was introduced into fluid being delivered by the mud pumps using a manifold installed in the mud pump room (Figs. F3, F4). The manifold design was based on a system developed for SF6 tracer testing in an aquifer on land (Clark et al., 2005) and was designed to introduce the gas tracer into a backpressured standpipe at a constant rate throughout most of a 24 h injection period. We knew that the pressure in the standpipe, positioned upstream from the mud pumps that deliver water to the drill pipe, would be relatively constant because this pipe is precharged by a positive displacement pump.

The manifold used two eight-port switching valves, each set to actuate at 30 s intervals, to deliver SF6 from compressed gas bottles. Each of the switching valves was configured with two coils of stainless steel, ¼ inch outside diameter (OD) tubing having a nominal coil volume of 20 mL. The switching valves operate by rotating 45° during each actuation. This opens one tubing coil to the incoming (compressed) gas while the other coil is opened to the outflow line (Fig. F4B). The volume of the coils and the difference between inlet and outlet pressures determines the quantity of gas delivered with each valve actuation. One bottle of SF6 was used at a time, and two switching valves were configured in parallel, with their 30 s activation times staggered. Check valves on the manifold prevented drilling fluid from flowing back into the gas regulators. Analog and digital pressure gauges installed on the regulators and at different points on the manifold allowed system performance to be checked regularly during injection.

The volume of the tubing coils and associated fittings attached to each eight-port valve was determined by laboratory calibration, as follows. Once the manifold was assembled and tested for leaks, high-pressure air was used to determine tubing coil volumes. A plastic bottle was filled with 1.000 L of water (measured using a volumetric flask) and then inverted and placed on a ring stand below water level in a bucket. A tube was run from the outlet of the tracer injection manifold into the inverted, submerged bottle, and one of the switching valves was operated while inflowing gas pressure was monitored and the number of actuations was counted. The valve was allowed to operate until most of the water was displaced from the inverted plastic bottle. The bottle was then recovered, and the quantity of water remaining was determined using a graduated cylinder. The difference between the initial and final water volumes was equal to the volume of gas (at standard temperature and pressure [STP] conditions) released by the manifold during the test. This test was repeated three or four times for each valve at four values of differential pressure, allowing generation of calibration curves that relate differential pressure to volume of gas introduced by each valve at STP (Figs. F4, F5).

Just prior to the start of the injection test, the SF6 manifold was installed on a rail in the mud pump room adjacent to the standpipe where the gas would be injected (Fig. F6). Gas was added to the standpipe using a ¼ inch NPT fitting that has been used in the past for injection of perfluorocarbon tracer (PFT) during coring and other operations (House et al., 2003). An analog gauge is attached to the fitting, and a digital pressure gauge was added early during Expedition 327, with data recorded automatically by the shipboard RIS. We verified earlier during the expedition that pressure in the standpipe rarely deviated from a precharge pressure of 41–43 psi during normal operation of the mud pumps with surface seawater across a range of pumping rates from 20 to 120 SPM.

Tracer injection

Injection of the SF6 tracer began at pumping test hour 0020 and continued until test hour 2032, when both bottles of SF6 were depleted. Pressures indicated by gauges on the regulators and manifold were monitored continuously and recorded every 15 min, or when a change in conditions occurred, by personnel stationed in the mud pump room. The pressure at the inlet to the gas injection manifold averaged 120.3 psi (±4.4 psi, 1σ) during the period of SF6 gas injection, the mean backpressure in the standpipe during this time was 40.9 psi (±1.8 psi), and the mean differential pressure was 79.5 psi (±6.1 psi). Calibration of the SF6 injection manifold (Fig. F5), with two valves operating every 30 s and a mean coil volume for both valves of 21.8 mL, indicates a mean SF6 injection rate of 0.0192 mol/min.

Pressure differences between the SF6 manifold and the standpipe varied somewhat when the water source was switched from seawater to freshwater and when additional tracers were added with the cement pump, but these deviations lasted just a few minutes and had little influence on the mass or molar rate of SF6 injection. Given the mean pumping rate during SF6 injection of 6.68 L/s, the typical injectate concentration during this time was 47.6 µM, and 23.3 mol of SF6 was injected into the formation in total. The high concentration and rapid rate of SF6 injection meant that much of the gas tracer was likely to be in the form of small bubbles initially, but these bubbles occupied <0.1% of the flow generated by the mud pumps. SF6 gas liquefies at relatively low pressures. Any bubbles transported below ~200 m water depth (>2 MPa) should have condensed into small droplets and dissolved during transport (Watson et al., 1988). The drill pipe is a highly efficient heat exchanger, and the high pressure and low temperature of bottom water (26 MPa and 2°C, respectively), combined with the turbulent flow within the pipe, assured that SF6 was dissolved in solution before it was introduced to the open hole at depth.

Fluorescent particles


Two kinds of fluorescent particle (colloid) materials were prepared for use during the Expedition 327 tracer experiment: microspheres and stained bacteria. Microspheres having distinct characteristics were acquired from Polysciences, Inc., and mixed in the cement additive tanks prior to injection (Tables T1, T2; Figs. F3, F7). Microspheres were composed of polystyrene tagged with BB Coumarin or YG Fluorescein dyes, compounds that fluoresce at specific wavelengths, and had diameters of 0.5–1.1 µm (Table T2). Microspheres and fluorescent compounds were selected to avoid misidentification of microspheres commonly employed on the JOIDES Resolution for contamination studies during coring operations (0.5 µm, YG Fluorescein dye). The total stock volume of microspheres injected during the experiment was 700 mL, comprising ~6 × 1013 microspheres in suspension. Microspheres have a characteristic density of 990–1010 kg/m3, slightly lower than seawater, so they should remain suspended during mixing and transport.

Fluorescent-stained microorganisms were injected separately. A natural population of surface microorganisms (along with other colloids) was concentrated from 100 L of 63 µm screened surface seawater into a 2 L solution using a standard 0.16 µm nominal pore-size Tangential Flow Filtration (TFF) system (Pall, Inc.). The concentrated microbes and colloids were fixed in 0.5% (final) formaldehyde and stained with the DNA-specific fluorochrome 4′,6-diamidino-2-phenylindole (DAPI) at 5 µM (final concentration) (following procedures described by Harvey et al., 1989). On the basis of a separate determination of the concentration of microorganisms in surface seawater in this part of the northeastern Pacific Ocean (Sherr et al., 2001), we estimate that ~1011 stained cells were injected during this experiment.


Microspheres were diluted initially in deionized water and then circulated within the cement mixing system in freshwater prior to injection. Injection was completed during hours 2013–2019 of the pumping test, near the start of a 1 h period of freshwater injection, with fluid from the cement mixing system being added at ~6.3 L/s to the flow from the mud pumps. Microspheres were mixed and injected in freshwater to minimize clumping in suspension. The resulting concentration of microspheres in fluid pumped down the drill pipe and into the formation was ~3 × 107 microspheres/mL of injectate.

Stained bacteria were also mixed and circulated in the cement additive tanks prior to injection, but with surface seawater rather than freshwater. The microbes and colloids were injected using the cement pumps at a mean rate of 3.4 L/s (added to flow generated with the mud pumps) during hours 2115–2122 of the pumping experiment. Based on the total quantity of fluid injected during this time, the mean concentration of DAPI-stained cells was ~104 cells/mL, well below DAPI’s threshold staining concentration. A more reliable estimate of stained microbe concentrations will be based on shipboard sampling and shore-based analysis of injectate, as described later.

Metal salt tracers

Metal salt tracers cesium chloride (CsCl), erbium chloride hexahydrate (ErCl3·6H2O), and holmium chloride hexahydrate (HoCl3·6H2O) were injected in two batches. Both batches included an alkali metal (Cs) and one of two lanthanide rare earth metals (Er and Ho). These salts were selected because of their relatively low concentrations in seawater and low detection limits using standard analytical techniques. We expect that these metals may behave nonconservatively during injection and transport and plan to assess this behavior through comparison to the behavior of SF6.

CsCl was prepared in two injection batches of 15 kg each (2 × 89.1 mol of Cs), the first mixed in solution with 23 kg of ErCl3·6H2O (60.3 mol of Er) and the second mixed with 23 kg of HoCl3·6H2O (60.6 mol of Ho). These metal salts were dissolved with seawater initially in 5 gallon buckets (Fig. F7). This salt brine was added to salt water in the cement additive tanks (~10 bbl, 1590 L) and then circulated internally until this fluid was introduced with the cement pumps into the flow coming from the mud pumps. The first batch of trace metal tracers was introduced during pumping hours 0301–0308 at 4.3 L/s, and the second batch was introduced during pumping hours 1900–1908 at 5.3 L/s. The quantity of metal salts added and the overall pumping rate indicate that concentrations of Cs, Er, and Ho were ~10–18 µM during injection (Table T1). A more accurate assessment of salt tracer concentrations in injectate will be completed on the basis of shipboard and downhole sampling.


Freshwater was pumped into the borehole at 6.7 L/s during two 1 h periods (pumping hours 0058–0158 and 1959–2058), resulting in a total volume of freshwater injected of 48,000 L. This is a small fraction of the total amount of surface seawater that was pumped and that flowed independently into the formation during drilling and other Expedition 327 operations. Freshwater could be a useful tracer through measurements of chlorinity, which is relatively easy to measure and has a precision of 0.2% for seawater values. The freshwater pumped during Expedition 327 was not pure deionized water, and we have assumed for initial calculations that the total salt content of this water was 200 mg/L. We pumped freshwater during two intervals primarily to generate large changes in fluid chemistry that occurred at known times and could serve as time stamps for calibration for the pumping rate of downhole samplers deployed during tracer injection. This sampling system is discussed in the next section of this paper.