3.2.2.1. Results of solvent extraction method

Results from the liquid extraction of [Co(CN)6]3- are shown in Fig. 48. The primary amine, Primene JM-T, extracted only minor quantities at pH1. At higher pH values, the extraction was not detectable. The secondary amine, Amberlite LA-2, without H2SO4 pretreatment shows as expected a reasonably high extraction at pH1, but has a strongly falling tendency towards higher pH values while the same resin with H2SO4 pretreatment maintains a high extraction yield until pH5 but decreases at higher pH values. Only the tertiary amine extractant, Alamine 336, extracted the Co complex strongly (near 100%) over the whole pH1-7 range.

Since Alamine 336 contains a basic nitrogen atom in the amine group, it may react with a variety of inorganic and organic acids to form amine salts, which are capable of undergoing ion exchange reactions with a host of other anions. As such, Alamine 336 is a liquid ion exchanger operated in a solvent extraction system. The general reactions, which are shown in Table 9, illustrate the two steps protonation and anion exchange:

After extraction, the organic phases containing Amberlite LA-2 and Alamine 336 were tested for stripping by the agents listed in Table 8. Ammonia could be used for stripping the Co complex from the Amberlite phases, while

TABLE 9. PROTONATION AND ION EXCHANGE: GENERAL REACTIONS

1. [R3N]org + [HA]aq 5 [RgNH+A-lorg

2. [RgN]org + [HCl]aq 5 [RgNH+Cl-]org

3. [RgNH+A-]org + [B-]aq 5 [RgNH+B-]org + [A-]aq

4. 3[R3NH+Cl-]org + [Co(CN)6]3-aq 5 {^NH+yCo^NL]3-^ + 3[Cl-]aq

K2(COO)2 (potassium oxalate) could be used for stripping Co complex from both for the Amberlite and Alamine phases. For Alamine 336, for instance, the type of stripping agent to be recommended depends on the overall recovery process. In general, basic stripping agents, which reverse the protonation reaction, show the best stripping efficiency.

Two alternative mechanisms for stripping [Co(CN)6]3- are:

(1) 2{[R3NH+]3[Co(CN)6]3-}org + 3[K2(COO)2]aq 5 З^КЦ

+ 3[H2(COO)2]aq+org + 2K3[Co(CN)6]aq

(2) 2{[R3NH+]3[Co(CN)6]3-}org + 3[K2(COO)2]aq 5 3{[R3NH+]2[(COO)22-]}org

+ 2{K3[Co(CN)6]}

3.2.2.2. Results of anion exchange method (1) [Co(CN)J3-

Table 10 gives the ion exchange yields for tracer concentration of [Co(CN)6]3-on various ion exchange resins with a feed volume of 100 mL.

It is obvious from the data in Table 10 that the strong base resins are superior to the weak base ones. Although both Dowex 1 and Dowex 2 are strong bases and quaternary amines, there is a difference in the functional groups:

• The functional (or ionogenic) group of Dowex 1 (Fig. 49(a)) is — CH2-N+(CH3)3.

• For Dowex 2 the ionogenic group (Fig. 49(b)) is

-CH2-N+(CH3)2-C2H4OH.

• This latter structure implies that the base strength is somewhat lower than for the first structure. This is not seen in the ion exchange yield but became evident in the succeeding stripping process.

Table 11 shows that elution of [Co(CN)6]3- from the Dowex 2 resin is easier than from the Dowex 1 resin both for strong HNO3 and HCl elution agent solutions reflecting the somewhat weaker basicity of Dowex 2.

(2) SCN-

A method for isolation and upconcentration of SCN- from sea water (produced water) has previously been published and is based on the use of the

anion exchange resin BioRad AG1, which is, in principle, equivalent to Dowex 1. Figure 50 shows the stripping yield (green curve) with 2.8M NaClO4 as the stripping agent and the total chemical yield (red curve) of the SCN — separation and enrichment process, both as a function of the collected volume of the eluate. The elution peak is shown in the inset.

The performance of SCN — on the Dowex 2 x 8 resin with the column dimension described earlier was checked and the separation factor from [Co(CN)6]3- with various elution agents investigated. Results of elution yields and separation factors are given in Table 12.

Figure 51 illustrates the absorption characteristics for [Co(CN)6]3- and SCN — on the Dowex 2 x 8 column and compares these data with the sorption characteristics of SCN — on the BioRad AG1 column.

О

„20 ■ 1 ■ I ■ I_________________ ■ I ■ I

0 200 400 600 800 1000

Raffinate volume (mL)

FIG. 51. Absorption yield of [Co(CN)6]3 and SCN — on the 0.5 mL Dowex 2 x8 (100-200 mesh) column, and for SCN — on the 8.5 mL BioRad AG1 column, as a function of raffinate volume (or original sample size) for tracer-containing seawater samples.

It is obvious from the data that SCN — experiences a rather fast breakthrough on the 0.5 mL Dowex 2 x 8 column, so the separation from [Co(CN)6]3- will be substantial in the absorption process.

Since HTO is the most frequently used water tracer, as an example, the possible dose to the operating personnel in the case of a worst case accident is discussed.

In the case of spillage of HTO, some of the water will evaporate and there will be tritiated vapour in the air. As vapour can be inhaled, a spillage of the total volume of HTO should be considered as the worst case scenario.

In the injection cylinder, the 2000 GBq of HTO will be diluted to approximately 75-100 mL. If this water evaporates and mixes with 500 m3 of air (e. g. a volume of air 10 m x 10 m x 5 m), the tritium concentration in the air will be 4 GBq/m3 or 4 MBq/L. If a person stays in this atmosphere for one minute and inhales 20 L of air and the entire vapour in the inhaled air is retained in the lungs, the intake of tritium will be 80 MBq. The annual limit of intake of tritium (from HTO) by professional workers as recommended by the ICRP is 1000 MBq. Thus, the calculated figure in the example above corresponds to approximately 8% of the annual limit of intake.

The worst case scenario for intake will occur if a beam of HTO is directed towards the faces of the operators handling the injection equipment. However, the operators will wear visors, masks and protective clothing, and they are well trained to deal with such situations. Taking into consideration that the total release of tracer can hardly occur unless there is a serious accident in the vicinity or some serious breakdown of the injection equipment at the time of injection, the worst case scenario described is very unlikely to arise. An accident causing destruction of injection equipment and the release of tracer will probably be more dangerous than the release of the radioactive tracer.

The concentrations of tracer in the sampled water will be very low. For HTO, it will be in the range of 200-300 Bq/L at maximum (top of the tracer recovery curve). For other tracers which have been injected in lower quantities, the concentration will be correspondingly lower. Sampling will not require any special protection equipment such as breathing apparatus or rubber gloves.

Figure 111 shows the pattern of injection and production wells where the tracer test was run.

FIG. 110. Tracer records in well 3.

FIG 111. Wells 20-2 and 21-3.

Up to two years after the tracer injection, only three wells had detected tracer breakthrough: wells 12-3, 13-4 and 13. The simulation for the well 13-4 using the parameters provided by the oil company is shown as a black line in Fig. 112.

To obtain a better fit to the experimental data of well 13-4, the layer thickness was reduced from 55.5 m to 3.2 m, resulting in the curve given in Fig. 113. However, the simulation predicted a very significant production of tracer in well 14-4, which did not occur.

A similar situation appears in the mesh of well 20-2. It is possible to fit the tracer record of well 13-3, but the simulator predicts significant tracer production in well 13-2, which is absent (Figs 114 and 115).

A scenario which assumes strong anisotropy along the direction between wells 21-3 and 13-4 (or wells 20-2 and 12-3) is proposed. However, in this case, in the simulation there is tracer production in wells 14-3 and 13-2 (tracer response was undetected in these wells). In consequence, the only way to avoid tracer production in wells 14-3 and 13-2 is to assume that the injectors located to the south of these wells have higher water injection rates.

For example, if the injection water flow rates of the injectors located to the south are duplicated, it is possible to obtain a good fit for the tracer record of well 13-4. A satisfactory fit of simulated response and experimental data (for wells 12-3 and 20-2) was obtained (Fig. 116).

FIG. 113. Tracer records in the well 3 with the new, reduced, thickness layer.

FIG. 114. Streamlines in well 21-3 mesh.

days after injection FIG. 115. Experimental and simulated tracer records for well 13-4.

a = 13 m h =18 m

FIG. 116. Experimental and simulated tracer records for well 12-3.

Case application 4: Interpretation of tracer experiments in laboratory model of non-boundary conditions

2.3.3.1. Tracer analysis of samples from the geothermal fields

(i) Analysis of HTO

HTO in samples from geothermal fluids is analysed using liquid scintillation counting. HTO analysis is performed by a direct counting method after a sample distillation pretreatment process. In the analysis, 11 mL of

мм… т………………

carrier (127I-), precipitation by silver as Agl followed by dissolution of the precipitate in the scintillation cocktail by addition of thiourea. The sample is then analysed by liquid scintillation counting.

All details of the analytical protocol are given in Appendix III. This protocol also includes the detailed procedure used to prepare 125I — for gamma spectroscopy measurement.

Structural correlation indicates that the main conduit of the tritium and the NDS from 1R8D to the production wells is the Sambaloran Fault. Figure 73 shows the schematic of the likely path of the tritium-bearing fluid. By first order of approximation, it is logical to think that wells 2R3D and 2R4D would yield the tracer first since these are situated nearest the injector well. It should then follow that the tracer recovery would diminish from well 2R3D towards well 202.

Figure 74 shows the cross-section and the structures intersected by wells where tritium and NDS were detected. As shown, the Sambaloran Fault is the only structure intersected by the wells which showed breakthroughs. Thus, it is the likely conduit of the tritiated brine.

Considering the proximity of the injector well, 1R8D, with the nearest monitor production well, 2R3D/2R4D, at ~140 m (well bottom separation), a much higher recovery of tracer was expected than was actually measured.

FIG 74. Cross-section of the wells showing the structures intersected. Highlighted in dark green are the intersects of the Sambaloran Fault. (CFL — Central Fault Line; MEF—Mahiao East Fault; Ura — Urangon Fault; SSU — South Sambaloran Unit).

Similarly, the tracer recoveries in wells 214 and 202 are considered low since there were already chemical and thermal indications as early as 2002 that when 1R8D was commissioned for brine injection (from South Sambaloran production wells in an adjacent sector), brine returns were then observed in wells 214 and 202, amongst others.

Although wells 202, 214 and 2R3D have brine returns based on geochemical monitoring and tracer breakthrough, the low tracer recovery suggest that these wells are still predominantly fed by the upper steam zone. This implies that most of the tritium is still in the deeper portion of the reservoir, along with the liquid zone. This is corroborated by flow measurements showing medium to high enthalpy discharges at the well head (W214 ~2500-2700 kJ/kg enthalpy, W202 ~2000-2200 kJ/kg, 2R3D ~1800-1900 kJ/kg).

Tritium, on the other, has a fractionation factor of 1. Thus, equal concentrations go to the water and vapour phases. The low recovery seen in wells 2R3D, 214 and 202 may suggest that the 10 Ci injection is insufficient.

The two pulses seen in wells 2R3D and 214 imply that the first pulse passed through the structural conduit, the Sambaloran Fault, from 1R8D and the second pulse which occurred at a later period could mean that the tracer travelled with the brine to the deeper part of the reservoir and later flashed and again passed through another conduit to appear in these wells (Fig. 75).

It has been reported that the northern Tongonan wells, 101 and 105D, showed declines in their gas concentrations and geothermometers, and this was attributed to the injection returns from well 1R8D. The discharges from these wells remained dry despite the prognosticated brine returns simply because of their shallow production zones, which are tapping the degassed steam from the Tongonan injection sink.

On the basis of the above-mentioned scenario, the absence of tritium in wells 105D, 101 and 109D could indicate two things. Firstly, there may not have been enough tritium injected into well 1R8D to effect a breakthrough in these northern Tongonan wells. Secondly, the current preferential flow of the tracer and the brine returns is to the east through the Sambaloran Fault, as indicated by the

early tracer breakthroughs in the wells along the structure. Thus, breakthrough of the tracer to the northern Tongonan wells could be observed much later. However, as shown by chemical changes in these wells, with the use of well 1R8D, the connection between these wells cannot be discounted.

Using minimal tracer recovery experimental data, no decline in temperature was observed in TRCOOL simulation (Fig. 76). Thus, no historical match was processed for the wells, in terms of cooling or thermal decline. The ~10°C decline in the fluid temperature (based on TQuartz) of well 202 did not match the cooling prediction based on tritium and NDS recoveries.

These observations and processed data could, therefore, imply one, or a combination, of the following: (i) tritium injected may have been insufficient to be detected in the monitor wells, however short their distances are; (ii) because of the nature of NDS tracer, monitoring in a highly two phase environment will give minimal recovery; (iii) the chemical breakthroughs observed in previous years could mean that well 1R8D is not the sole source of brine/cold waters capable of effecting such changes in thermal and chemical parameters.

Normalization of a liquid scintillation counter is done using unquenched sealed standards of 14C, tritium and background supplied by a commercial company (e. g. Packard Instrument Co.) The samples and standards are usually counted for a preset time of 50 min and the whole batch is cycled 10 times. Thus, each sample/standard is counted for a total time of 500 min (the sample counting time and the number of cycles can be adjusted as per requirement). The samples with higher counts can be counted for shorter times and for fewer repeat cycles). The data are statistically evaluated by applying Chauvenet’s criterion and rejecting outliers. This is followed by the calculation of the mean background count rate, the net mean count rate of the standard and the net count rates of unknown samples.

Uncertainty in the tritium activity is calculated by combining the uncertainties of all the factors involved (net counts per minute of counting standard, activity of counting standard, net counts per minute of sample, weights and decay correction) and using the error propagation law.

The energy production potential, or capacity, of geothermal systems is highly variable. It is primarily determined by the pressure decline caused by mass extraction, but also by heat content. Pressure declines continuously with time in systems that are closed or with limited recharge. The production potential of geothermal systems is, therefore, often limited by lack of water rather than lack of heat. Geothermal resource management involves controlling energy extraction from geothermal systems underground so as to maximize the resulting benefits, without overexploiting the resource.

When geothermal systems are overexploited, production from the systems has to be reduced, often drastically, resulting in an insufficient steam supply to power plants or in loss of wells. Overexploitation mostly occurs for two reasons. Firstly, inadequate monitoring and data collection contribute to poor understanding of the system and lack of reliable modeling and, therefore, the systems respond unexpectedly to long term production. Secondly, overexploitation occurs when many users utilize the same resource/system without common management or control.

The main purpose of tracer testing in geothermal reservoir management is to predict possible cooling of production wells resulting from the long term injection of colder fluid and/or the invasion of natural groundwater. In a geothermal field, the primary resource is water, both as liquid and as steam. A direct measure of its behaviour is thus of obvious importance to field management. Water tracing is the only technique that gives a direct indication of underground flow patterns and velocities.

Information gained from tracer testing of geothermal reservoirs is similar to that obtained from oilfields and includes:

(a) Proper diagnostics of the reservoir comprising evidence of direct connections between the tracer injection point (either within or outside the field) and monitoring wells in the field;

(b) Measurement of direction, speed and mean residence time of water movement;

(c) Determination of the extent to which groundwater downflows intrude into production wells;

(d) Identification of breakthrough (arrival time or first appearance of tracer in the production well);

(e) Quantification of the tracer quantity collected in each production well;

(f) Information needed for calibration or verification of physical models of the geothermal reservoir.

All of this information will aid in gaining an understanding of the nature of a geothermal system, but the measurements, which bear upon injection and groundwater intrusion or cooling potential, have the greatest impact on field management.

Tracer candidates that pass the thermal stability tests mentioned above are subject to static batch sorption experiments. Crushed reservoir or reservoir-like rock is added to the same type of vial as used above. Convenient test materials are sandstone (Berea, Clashack, Bentheimer, Felser, etc.) and chalk, which represent the main reservoir rock types, and kaolinite, which represents clays that are nearly always present to some degree.

The survivability yield, Y, is calculated from Eq. (15). One example of a static sorption curve is given in Fig. 35.

As can be seen in the example given, there is a clear indication of sorption to sandstone (Clashack) whereas sorption to the other substrates is not detectable. The decrease in the Y values for chalk and clays is due to the thermal degradation data given in Fig. 35.

11.3.1. Introduction

Interwell passive tracer testing is a powerful tool for the evaluation of secondary recovery projects in oilfield reservoirs. In these projects, water injected into injector wells push the oil to the producer wells from which it is extracted. The water injected patterns are usually very complicated because of the natural heterogeneity of the reservoirs and the mobility differences between water and oil. Channelling of water between injector and producer wells is a very common problem that conspires against achieving acceptable sweep efficiencies. The interwell tracer tests permit this problem to be detected and also to allow some reservoir parameters to be determined.

Owing to the elevated uncertainty associated with the reservoir knowledge, especially later to the waterflooding, there is no need, in an initial phase of interpretation, to assume a very detailed model. Therefore, a simple moment analysis or an analytical solution from a one dimensional model can provide acceptable results about the average residence times and water volumes in many cases.

11.3.2. Objectives

In consequence, employment of the Anduril 2.3 simulator was proposed for estimating the principal parameters from data recorded from an interwell tracer test performed in an oilfield in Argentina. The Anduril 2.3 program has capabilities for both temporal and volumetric analyses. In this instance, it was used to determine the following parameters between different pairs of wells:

• Breakthrough time;

• Mean residence time;

• Peak maximum time;

• Final time;

• Tracer recovery.

From the above mentioned parameters, the following volumetric parameters can be obtained:

• Breakthrough volume;

• Mean swept volume;

• Injected volume at peak;

• Maximum injected volume;

• Pore volume swept between wells.

These parameters are proportional to the ones in the first list (water flow rate is the proportionality factor).

Two common measures of heterogeneity can be obtained from the (F Ф)- plot: the Lorentz coefficient (Lc) and the Dykstra Parsons coefficient (VDP) [32]:

(45)

•

Lc varies between 0 and 1, with 0 representing a homogeneous flow field.

• F is the derivative of (F, Ф). It is evaluated at the mean (0.5) and at one standard deviation below the mean (0.841).

IV. 2.1.5. Volumetric fluid sweep efficiency

Volumetric fluid sweep efficiency is a measure of efficiency for use of the injected fluid. It expresses that fraction of the injected water which is actively contributing to pushing the original fluids in the reservoir. Using the concept of streamlines and the definition of ‘F above, it is possible to estimate sweep efficiency directly from a tracer test.

The F(t) term can be interpreted as the fraction of streamlines that have ‘broken through’ and have started to produce injected fluid. These streamlines are not contributing further to sweeping of the reservoir. In the opposite sense, at the beginning, all the injected water is active and the sweep efficiency must be at a maximum. Hence, sweep efficiency can be expressed in terms of fractional tracer recovered and injection rates:

Sweep efficiency is typically reported as a function of dimensionless time or time normalized by the total pore volume:

„ qinj m t tD — t —

D V Mnj t *

IV. 2.1.6. Limitations of the method

• Conservative tracer travelling jointly with the water bulk flow is required.

• Steady state flow is necessary.

• The necessity for extrapolation of the tracer records (tracer production curve). Sampling for tracer is frequently terminated long before the tracer concentration falls to zero. Because the first moment is a time weighted average, failure to include late time data leads to the underestimation of
both mean residence time and pore volume. A decline pattern composed of exponential terms is most commonly observed.

• The difficulty of distinguishing between the imposed flow geometric effects (resulting from the injection pattern) and heterogeneity effects.

• The estimations can be erroneous if the tracer does not move with the ‘bulk flow’. This happens, for example, in double porosity rocks with diffusion in the secondary porosity.