Analyzing Hydrocarbons in the Borehole

A rapid evaluation of hydrocarbon uid composition is now available with a new uid-sampling tool. The quality of samples taken for later analysis can be determined before lling the sampling cylinder. The tool is sufciently sensitive to determine compositional gradients within a formation.

Soraya Betancourt Go Fujisawa Oliver C. Mullins Ridgeeld, Connecticut, USA Andrew Carnegie Abu Dhabi, United Arab Emirates (UAE) Chengli Dong Andrew Kurkjian Sugar Land, Texas, USA Kre Otto Eriksen Statoil Stavanger, Norway Mostafa Haggag Antonio R. Jaramillo Abu Dhabi Company for Onshore Oil Operations Abu Dhabi, UAE Harry Terabayashi Fuchinobe, Kanagawa, Japan
CFA (Composition Fluid Analyzer), LFA (Live Fluid Analyzer for MDT tool), MDT (Modular Formation Dynamics Tester) and PVT Express are marks of Schlumberger. For help in preparation of this article, thanks to Sylvain Jayawardane and Jiasen Tan, Edmonton, Alberta, Canada; Sudhir Pai, Rosharon, Texas, USA; Ibrahim Shawky, Abu Dhabi, UAE; and Tsutomu Yamate, Fuchinobe, Kanagawa, Japan.

Near-infrared radiation

Absorption and excitation

Understanding crude oil composition early in the development process helps optimize resource exploitation. Such information is now available from a wireline tool, giving results in real time that allow optimization of uid sampling based on the measured in-situ composition.

An early determination of gas composition and gas/oil ratio (GOR) may be necessary to decide whether to complete a well, or even whether to develop a eld at all. For example, the economics of developing elds that contain a rich hydrocarbon gas and those that contain a

54

Oileld Review

high percentage of carbon dioxide [CO2] in the gas are markedly different. CO2 is highly corrosive, so its presence can change requirements for owlines and surface equipment. In addition, commingling prospects with incompatible compositions may need to be avoided. Flow assurance also is impacted by problems with asphaltene, wax, hydrate and organic scale buildup in owlines.1 The uid composition can restrict allowable drawdown pressures and ow rates to prevent uid dropout. This article update presents recent developments in uid analysis available with the MDT Modular Formation Dynamics Tester.2 A new module, the CFA Composition Fluid Analyzer, provides a measurement of uid composition from samples drawn directly from the formation. It discriminates the fractions of methane, light hydrocarbons, heavy hydrocarbons, carbon dioxide and water present in a sample. The tool makes this determination based on light absorption and uorescence of the uids; results are transmitted to surface in real time. Case studies from the Middle East and the North Sea demonstrate the effectiveness of this new module. Analyzing Oil and Gas The terms gas and oil describe the state of a hydrocarbon as vapor or liquid, but do not specify the chemical composition. A detailed measurement of the constituents of a hydrocarbon, as determined in a surface laboratory, can be used to predict the constituents of the gas and oil phasesas well as other physical properties like density and viscosityat various temperatures and pressures. These detailed laboratory measurements can take a long time to obtain. The new CFA tool, in conjunction with other modules on an MDT assembly, provides a quick determination of some of the components and indicates the degree of drilling-mud contamination before samples are taken for further analysis. Hydrocarbon uids comprise a multitude of constituents ranging from single-carbon methane to very long-chain carbon compounds, as well as cyclic, aromatic and other complex molecules such as asphaltenes and waxes. These constituents determine the phase behavior of a given reservoir uid, which is often indicated using a pressure-volume-temperature (PVT) phase diagram (above right).3 A hydrocarbon uid is in single phase if the pressure and temperature are outside the phase envelope. At conditions within this envelope, two phases coexist. However, the phase composition changes within this two-phase region. Near the bubblepoint curve, the gas phase

Reservoir depletion Dewpoint curve Critical point

Pressure

Bubblepoint curve

Cricondentherm

Temperature

> A hydrocarbon phase envelope for a retrograde condensate. Between the bubblepoint and dewpoint curves, hydrocarbon uids are in two phases. The lines of constant liquid mole fraction (dashed) meet at the critical point. Fluids that enter the two-phase region to the right of the critical point are termed retrograde condensates. Fluids at temperatures greater than the cricondentherm remain single phase at all pressures. If the initial reservoir condition of temperature and pressure is above the phase envelope and between the critical temperature and the cricondentherm, the uid goes through a dewpoint and liquid drops out of the gas phase as the reservoir pressure declines. This condition (vertical line) starts at initial reservoir condition, shown here at an arbitrarily chosen temperature and pressure.

is predominantly methane, but further into the two-phase region, more light and intermediate components enter the gas phase. Similarly, the rst liquid components to drop out after passing the dewpoint are the heavier components; lighter components go into the liquid phase at conditions further from the dewpoint curve. This phenomenon is important when sampling gas-condensate fluids: once a fluid enters the two-phase region, heavy components are lost into the liquid phase. This behavior is used in the CFA design to determine when a uid crosses the dewpoint. The pressure and temperature condition at which the bubblepoint and dewpoint curves meet is called the critical point. At that point, the density and composition of the liquid and gas phases are identical. The maximum temperature at which two phases can coexist is termed the cricondentherm. Reservoir temperature is usually almost constantunless cold or hot uids are injected into the reservoirso most depleting reservoirs follow a downward vertical path on a pressuretemperature phase diagram. If reservoir temperature is between the critical temperature and the cricondentherm, liquid can drop out of the gas phase within the reservoir. These are termed

gas-condensate, or retrograde-condensate, reservoirs. Gas in a reservoir with a temperature greater than the cricondentherm is termed a wet gas if liquid drops out because of pressure and temperature decreases in the production system, or a dry gas if no liquid falls out in either the reservoir or production system. Economic decisions early in an exploration project often hinge upon characterizing the type of hydrocarbon in a reservoir. This determination is particularly true offshore, where an expensive platform infrastructure or subsea tiebacks may need to be designed to handle reservoir uids. Early hydrocarbon typing is also needed in remote areas where satellite elds may not be
1. Wasden FK: Flow Assurance in Deepwater Flowlines/Pipelines, Deepwater Technology, World Oil Magazine Supplement (October 2003): 3538. 2. Andrews RJ, Beck G, Castelijns K, Chen A, Cribbs ME, Fadnes FH, Irvine-Fortescue J, Williams S, Hashem M, Jamaluddin A, Kurkjian A, Sass B, Mullins OC, Rylander E and Van Dusen A: Quantifying Contamination Using Color of Crude and Condensate, Oileld Review 13, no. 3 (Autumn 2001): 2443. 3. For more on phase diagrams and pseudouids: Composing Pseudouids in: Alaka JO, Bahamaish J, Bowen G, Bratvedt K, Holmes JA, Miller T, Fjerstad P, Grinestaff G, Jalali Y, Lucas C, Jimenez Z, Lolomari T, May E and Randall E: Improving the Virtual Reservoir, Oileld Review 13, no. 1 (Spring 2001): 4445.

Autumn 2003

55

2.0 Methane Ethane N-heptane Carbon dioxide

1.5 Optical density

Miscible ood programs, such as reinjection of separator gases, can change the composition and phase behavior of the mixture of formation and injected uids. Capturing reservoir uid samples may be necessary to understand this process also. Fluid Sampling For many years, the industry has evaluated uids by collecting samples from a formation, bringing them to surface, and analyzing them in a laboratory that may be far from the wellsite. This process can be time-consuming and is subject to errors in collection, handling or sample degradation during transport. The PVT Express onsite well-uid analysis service is a recent advance in uid-property determination. This system can deliver detailed uid-analysis data a few hours after samples reach the surface. A unique minicell for PVT property determination allows onsite measurement of dewpoint pressure on gas-condensate samples. The compact, modular, mobile laboratory can be transported to any geographic location. Delays associated with sample shipment are eliminated. Fluid quality and uid properties can be determined while the opportunity to obtain additional samples is still available. Decisions relating to additional wireline formation testing or drillstem testing operations can be made more quickly with the PVT Express service. Taking the next step, Schlumberger makes some uid properties evaluations downhole. The LFA Live Fluid Analyzer for the MDT tool provides a means to analyze in-situ fluids to determine when contamination from drilling mud has decreased sufciently to obtain a uid sample with acceptable quality.4 This minimizes the time required to collect fluid samples, decreasing both rig costs and the risk of the tool becoming stuck because it was on the formation for too long. The LFA module includes a channel specically tuned to record the presence of methane, providing a means to obtain GOR.5 Downhole GOR measurements help identify whether different formations are compartmentalized. A sampling program can be directed to reveal compositional variation within a given compartment, helping to optimize completion programs. Agreement between downhole, wellsite and laboratory crudeoil property measurements engenders condence in the derived uid properties. The LFA channels also measure the oils color, which usually changes as drilling mud is ushed out of the formation. A sophisticated algorithm indicates the cleanup time required to

1.0

0.5

0.0 1600 1700 1800 Wavelength, nm 1900 2000 2100

Medium-weight oil

Optical density

Water 2

Hydrocarbons Condensate

0 500 1000 Wavelength, nm 1500 2000

> Visible and near-infrared absorption spectrum. As the wavelength increases, hydrocarbon optical densityor light absorptionis due to successively heavier and more complex molecules (bottom). Gas condensates and oils have different responses in the visible region. Hydrocarbon molecularexcitation bands appear at about 1700 nanometers (nm), where light interacting with hydrocarbon bonds induces molecular vibrations (top). Methane has a peak at the CH4 vibrational mode, and ethane peaks at the CH3 mode. Longer-chain hydrocarbons, such as n-heptane, have many CH2 bonds, but also have CH3 bonds at the ends of chains. The carbon dioxide excitation wavelength is longer than the hydrocarbon mode wavelengths. Water has two strong, broad absorption peaks, which can interfere with detection of the hydrocarbon excitation peaks (bottom).

economic to produce unless a tieback conguration or additional facilities are built to market the gas. Sound production practices also require knowledge of uid-phase behavior. If reservoir pressure drops below the dewpoint, liquid condensate drops out in the formation. At low saturation, liquid in the pore spaces is not mobile

and decreases the gas relative permeability. Two negative economic impacts result: productivity declines and valuable condensate liquids are left behind in the reservoir. Pressure support through gas or water injection often is required to keep reservoir pressure above the dewpoint. Similar practices can be followed to keep an oil reservoir above the bubblepoint to avoid gas breakout.

56

Oileld Review

obtain a representative formation-uid sample in the MDT sample modules.6 This evaluation prequalifies fluid samples for more extensive analysis at the surface, provides basic uidproperty data such as GOR and helps dene uid variability at different depths. These measurements are critical for adjusting a sampling and analysis plan while the MDT tool is in the borehole, which helps an operator realize maximum benet of a logging run. Gas-condensate reservoirs present special challenges for uid-sample collection. A sampling tool must apply a pressure differential to pull uid out of the formation into its sampling chambers. If this drawdown is too large, the pressure can drop below the dewpoint and the separated liquid phase may be trapped in the reservoir. As a result, the sample collected will not be representative. Even if the phase transition occurs outside the formation, that is, the uid becomes multiphase within the tools probe and pumpout modules or in the owlines leading to the sampling chamber, differences in uid density and viscosity and phase segregation within the tool can lead to an unrepresentative sample composition. This problem of a reservoir uid breaking into two phases is even more severe when the samples are obtained at surface during a drillstem test, which uses a larger pressure drawdown than a sampling tool on wireline. The new CFA modulea joint development of Schlumberger-Doll Research Center, Ridgeeld, Connecticut, USA; Schlumberger Kabushiki Kaisha Technology Center, Fuchinobe, Kanagawa, Japan; and Schlumberger Sugar Land Product Center, Sugar Land, Texas, USAwas designed specically to detect dew formation as a second hydrocarbon phase using a fluorescence detector. This is the rst downhole tool with dew-detection capabilities. With this capability, the module can differentiate between single- and multiple-phase ow and can show when pressure in the tool drops below the dewpoint pressure. Used in combination with an LFA module, a CFA tool indicates the proper time and conditions for obtaining a uid sample, even in the difcult environment of gas-condensate reservoirs. In addition to the uorescence detector, the CFA tool incorporates absorption spectrometers that measure the opacity, or optical density, of a uid at several wavelengths. These measurements distinguish several components of hydrocarbon uids, not only enhancing dew detection, but also providing a compositional analysis. This compositional analysis capability will be discussed first, with fluorescence detection described later in the article.

Evaluating Gas Composition Hydrocarbon molecules interact with light in the visible and near-infrared wavelength band that is sampled by the CFA spectrometers. Interaction with electronic energy bands gives oils their color, with complex molecules absorbing more light than simple ones (previous page).7 Oils with a signicant quantity of resins and asphaltenes are darker than oils containing primarily parafns.8 Gas condensates tend to be relatively clear, with little electronic absorption. A different type of interaction occurs in the near-infrared region, where light absorption excites molecular vibration. The type of molecular bond between carbon [C] and hydrogen [H] atoms determines the frequency of absorbed light. The dominant vibrational absorption interactions occur in three types of molecular congurations:9 a carbon atom surrounded by four hydrogen atoms, that is, CH4 a carbon atom with three hydrogen atoms, CH3 a carbon atom with two hydrogen atoms, CH2. Methane is the unique example of the rst mode. Ethane is an example of the second case, because it contains two carbon atoms that are each connected to three hydrogen atoms. However, longer-chain hydrocarbons are predominantly CH2 but also have the CH3 group at each end of a chain. The CH2 group dominates light absorption by such long-chain compounds, but there also is some CH3 absorption. For example, 77% of the carbon-hydrogen bonds of n-dodecane, a common parafn with 12 carbon atoms in a linear chain, are in CH2 groups. There is a complication in the analysis of hydrocarbon spectra: the absorption spectra overlap. Spectral interpretation requires proper accounting for this overlap. These complexities are overcome in the CFA analysis by employing a technique called principal component regression. This mathematical procedure extracts maximal information content in any dataset, in this case the vibrational spectra. The CFA interpretation algorithm incorporates ve detectors to determine four components:10 methane, which is termed C1 in the CFA analysis 11 other hydrocarbon gases, termed C2-C5 hydrocarbon liquids, termed C6+ carbon dioxide, CO2. Distinct spectral signatures can distinguish both methane and carbon dioxide. The other hydrocarbon gases are dominated by CH3, and CH2 groups dominate hydrocarbon liquids. Thus,

the principal component regression results are interpretable in terms of spectral characteristics. In the same part of the infrared spectrum, water has a broad and strong absorption peak. The presence of water can swamp the other signals, particularly the CO2 signal. The CFA module has a detector tuned to the water vibrational mode, indicating when the responses from the other detectors are inuenced by water. The CFA tool is recommended for uids with a GOR exceeding 1000 scf/bbl [180 m3/m3], because uids with a lower GOR have a color signal that is strong enough to interfere with the vibrational-mode absorption peaks. This recommended range includes gases, gas condensates, volatile oils and some black oils.12 The following sections show how these compositional measurements were used to detect injected gas in a monitoring well and to discover a compositional gradient within the oil leg of a reservoir with a gas cap.
4. Andrews et al, reference 2. 5. Mullins OC, Beck GF, Cribbs ME, Terabayashi T and Kegasawa K: Downhole Determination of GOR on Single-Phase Fluids by Optical Spectroscopy, Transactions of the SPWLA 42nd Annual Logging Symposium, Houston, Texas, USA, June 1720, 2001, paper M. Dong C, Hegeman PS, Elshahawi H, Mullins OC, Fujisawa G and Kurkjian A: Advances in Downhole Contamination Monitoring and GOR Measurement of Formation Fluid Samples, Transactions of the SPWLA 44th Annual Logging Symposium, Galveston, Texas, USA, June 2225, 2003, paper FF. 6. Mullins OC, Schroer J and Beck GF: Real-Time Quantication of OBM Filtrate Contamination During Openhole Wireline Sampling by Optical Spectroscopy, Transactions of the SPWLA 41st Annual Logging Symposium, Dallas, Texas, June 47, 2000, paper SS. 7. For more on visible and near-infrared light interactions with crude oil: Andrews et al, reference 2. 8. Nonhydrocarbon compounds found in oil, such as those containing nitrogen, oxygen and sulfur, also contribute to color. Dark oils can contain large amounts of these components. 9. The dashes indicate a connection to other carbon atoms: CH3 connects to one carbon atom, and CH2 connects to a carbon atom on each side of the indicated carbon atom. 10. Van Agthoven MA, Fujisawa G, Rabbito P and Mullins OC: Near-Infrared Spectral Analysis of Gas Mixtures, Applied Spectroscopy 56, no. 5 (2002): 593598. Fujisawa G, van Agthoven MA, Jenet F, Rabbito PA and Mullins OC: Near-Infrared Compositional Analysis of Gas and Condensate Reservoir Fluids at Elevated Pressures and Temperatures, Applied Spectroscopy 56, no. 12 (2002): 16151620. 11. In this terminology, the number following the letter C indicates the number of carbon atoms in the compound. Thus, C1 is methane, with molecular formula CH4. 12. A standard correlation of oil type to GOR is that black oil GOR is less than 2000 scf/bbl [360 m3/m3]; volatile oils range from that value to 3300 scf/bbl [594 m3/m3]; then gas condensates extend to 50,000 scf/bbl [9006 m3/m3]; and gases have GOR greater than 50,000 scf/bbl.

Autumn 2003

57

Composition percent Water Flag

100

CO2
Quality Water Volume Fraction

C6+ C2C5 C1 Highly Scattering Fluid 0

Gas/Oil Ratio Tool GOR Laboratory GOR scf/bbl 7500

Low

Medium 0 percent 100 Elapsed Time, s 10,260 10,215 10,170 10,125 10,080 10,035 9990 9945 9900 9855 Lab result 8145 8100 8055 8010 7965 7920 7875 7830 7785 7740 Lab result High

8145 8100 8055 8010 7965 7920 7875 7830 7785 7740 Lab result 8145 8100 8055 8010 7965 7920 7875 7830 7785 7740 Lab result

Detecting Injection Gas The CFA module was used in a gas-injection pilot project that had been in operation for several years in an onshore carbonate reservoir in the United Arab Emirates (UAE). As part of an ongoing evaluation program, the operator, Abu Dhabi Company for Onshore Oil Operations (ADCO) drilled a new monitoring well to determine the progress of the injected gas.13 An MDT sampling string containing the CFA tool was equipped with a dual-packer module, a pumpout module, an LFA module and 18 single-phase multisample chambers. Single-phase samples of sufficient quality for later, detailed laboratory analyses were obtained from six different stations. The CFA tool provided uid compositional information prior to sample collection at the rst four stations. The drilling uid was a water-base mud, so some water was detected during the analysis. The characteristics of the pumpout moduledescribed in Detecting a Multiphase Condition, page 60caused the water to appear as slugs passing the CFA window. The tool monitored the cleanup of the drilling uid prior to sampling. The upper zone sample was virtually all oil (left). The second sample station clearly showed a high concentration of gas coming from the formation. The two lowest zones produced some gas. This indicated that, at the monitor well, the top zone was unswept, and the second zone had been swept the most by the injected gas. The results demonstrated that these two upper zones were not in communication. The CFA results were obtained after about two hours of pumping to clean out drilling mud. After an additional two to three hours at each station, changes in the LFA color channels indicated that the uid had cleaned up enough to take a sample in a single-phase sampling cylinder. These samples were analyzed in a laboratory, and the results match reasonably well with the real-time CFA data. These results helped ADCO understand the ow characteristics and gasinjection efciency of their eld. Discovering a Compositional Gradient Statoil, the operator of a Norwegian Sea appraisal well, wanted to establish the gas/oil contact (GOC) and oil/water contact (OWC) and obtain uid samples for laboratory analysis. A drillstem test in a discovery well had not provided conclusive uid-phase property data. This was the only appraisal well drilled before developing facilities to process a complex, nearcritical reservoir-uid system. Statoil felt it was

> CFA uid composition in a UAE carbonate reservoir. The CFA result indicated that the upper zone, A, was unswept. The second station, B, had the greatest concentration of the gas components, C1 and C2-C5, and the highest gas/oil ratio (GOR), indicating that the injected gas had swept this zone. The two lower stations also had been partly swept by injected gas. Results from samples collected during this logging run were analyzed in a laboratory, conrming the composition and GOR values measured by the CFA module.

58

Oileld Review

C1 Formation Resistivity 0.01 ohm-m 10,000 C2C5 C6+ Water

0 Optical Density 3

Bulk Density 0.6 g/cm3

Invaded-Zone Resistivity 0 0.1 ohm-m 10,000 MDT Pressure Gas bar

Gamma Ray XX60 20

Neutron Porosity p.u.

Invaded-Zone Diameter

Oil Water

Laboratory GOR scf/bbl 8000 Tool GOR scf/bbl 8000

API 140 1.7

2.7 0

in.

60 375

385 0

2 Gas/oil contact 5

XX80
3

XY00
Oil/water contact

1 4

XY20

XY40

XY60

XY80

> Compositional gradient in a North Sea well. Gamma ray (Track 1), bulk density and neutron porosity (Track 2), and formation resistivity (Track 3) logs indicate a relatively featureless zone of about 100 m [328 ft]. A thin, possibly impermeable zone exists at about XY30. The invaded-zone resistivity (Track 3) implies a water zone up to XY10, with a transition zone up to about XX95 and perhaps a third zone above XX75. The pressure measurements (Track 4) conrm three gradients, with a gas/oil contact at XX75 and an oil/water contact at XY10. Both optical density from the CFA color channel and gas/oil ratio (GOR) (Track 5) show a gradient in composition, which is also seen in the CFA compositional analysis (Track 4). The numbers to the left of the CFA compositions indicate the sampling order in the wellbore. The thin bars below each CFA result are later laboratory results, which were scaled to exclude the water fraction measured by the CFA tool, allowing direct comparison of the hydrocarbon components. Laboratory GOR measurements (Track 5) also conrm the compositional gradient, although the magnitude is somewhat different from the CFA result.

important to obtain a good description of the uid properties within the reservoir. A wireline triple combo log indicated 100 m [328 ft] of relatively featureless reservoir, except for a possibly impermeable streak at about XY30 m (above). The density and neutron porosity logs did not indicate a crossover. Crossover is normally a sign of a gas zone, but the formation and uid properties in this reservoir were such that no separation was seen, probably because of high gas density, low oil density and invasion by water-base mud ltrate. The formation resistivity was uniform, but the resistivity due to the waterbase mud in the invaded region indicated a prob-

able change in the ushed-zone water saturation, with a likely OWC at XY10. Statoil next obtained pressure gradients in the formation using an MDT tool to nd the different uid sections. The MDT logging system comprised a probe module, a pumpout module, a CFA module, an LFA module and multiple sample-chamber modules. The probe module had a high-grade quartz pressure gauge. Within the 100-m span identied by the wireline logs, the operator obtained 25 pressure measurements. These data identied three different pressure gradients corresponding to gas, oil and water, all in hydraulic communication. However,

the pressure gradient alone was inadequate to resolve a compositional gradient in the hydrocarbon zone. Statoil had investigated compositional grading at other locations around the world and wanted to study this oil leg in greater detail.14 Once the pressures were obtained, the operator repositioned the tool string to analyze the formation uids using the CFA module and to collect samples in high-pressure sample bottles from an area in the lower part of the oil leg. The CFA module obtained a quick reading of sample composition before filling each sampling cylinder. The results, transmitted to surface in real time, indicated that the uid composition stabilized within 1000 to 2000 seconds, or 17 to 33 minutes. The CFA compositional measurements typically were taken after an hour of cleanup time. However, slugs of water-base mud continued to pass through the apparatus, and it typically took more than two additional hours for the mud contamination to become acceptably low to obtain each sample for surface analysis. Next, the probe was positioned in the gas cap. The GOR was high, with large concentrations of C1 and C2C5 components. The optical density at the color channel was almost zero in the gas cap, consistent with the presence of a very light hydrocarbon system. After obtaining a uid sample in the gas cap, the tool was repositioned to obtain a second sample 14 m [46 ft] above the rst oil sample. The operator suspected there might be a compositional gradient in the oil zone. The CFA results from the two oil stations indicated a signicantly higher GOR at the upper position. The color optical density also was less at the upper station, indicating a greater content of gas components at the higher position. With this information, the operator was able to change the logging plan immediately. The probe module was placed as close to the oil/water contact as possible, then as close to the gas/oil contact as possible. The CFA readings over the full extent of the oil column conrmed the existence of a uid compositional gradient and more than a 60% increase in GOR over about
13. Fujisawa G, Mullins OC, Dong C, Carnegie A, Betancourt SS, Terabayashi T, Yoshida S, Jaramillo AR and Haggag M: Analyzing Reservoir Fluid Composition In-Situ in Real Time: Case Study in a Carbonate Reservoir, paper SPE 84092, presented at the SPE Annual Technical Conference and Exhibition, Denver, Colorado, USA, October 58, 2003. 14. Hier L and Whitson CH: Compositional Grading Theory and Practice, paper SPE 63085, presented at the SPE Annual Technical Conference and Exhibition, Dallas, Texas, USA, October 14, 2000. For another compositional-gradient example: Metcalfe RS, Vogel JL and Morris RW: Compositional Gradients in the Anschutz Ranch East Field, paper SPE 14412, SPE Reservoir Engineering 3, no. 3 (August 1988): 10251032.

Autumn 2003

59

Excitation Fluorescence channel 1

Fluorescence intensity

Gas condensate

Subtle changes in composition can be easily obscured if the uid passing through the detectors has separated into two phases. This can occur if the drawdown pressure is too great. The use of uorescence is key to detecting when a uid passes through its dewpoint. An Aromatic Afterglow Aromatic hydrocarbons uoresce. The distinguishing characteristic of uorescence is that there is a brief time delay between light absorption and its reemission, and that the reemission occurs at a lower energythat is, a longer wavelengththan the absorbed light (left).15 The CFA module incorporates a uorescence detection unit (FDU) along the owline, about 7 cm [3 in.] from the absorption spectrometer. Since they are close together, the two types of detector sample essentially the same uid. This allows the two measurements to be used simultaneously to evaluate uids. The FDU shines blue light onto a window in the ow tube. One detector tuned at the source wavelength is placed at the reection angle. This provides a measure of direct reection of light, reducing the possibility of false-positive uorescence detection. Two other detectors in the FDU record the intensity and spectrum of the uorescence. The FDU is particularly sensitive to uorescence from uid on the surface of the ow-tube window. Dew formation often causes a liquid coating on the ow-tube surfaces. When the uid is in a single phase, the detector measures the properties of the uid owing near the window. Once the pressure drops below the dewpoint, liquid drops out of solution and condenses. The condensed liquid phase wets the detector window, so the uorescence detector is most sensitive to the properties of the liquid phase. Since the heavy ends are enriched in the liquid phase, the FDU is sensitive to the presence of a liquid phase dropping out from a gas condensate. This makes it an excellent tool for detecting when a uid drops below its dewpoint. Detecting a Multiphase Condition The rst use of an FDU in the eld showed that the drawdown pressure being used at that sampling station was too large, generating a twophase condition. The operator moved to another location a few centimeters away and resampled, this time obtaining a good sample. This section describes how the fluids separated in the pumpout module and how the FDU detected this two-phase condition caused by excessive drawdown at the rst location.

Light oil

Fluorescence channel 2

Wavelength

Fluorescence Fluorescence channel 2 channel 1 Blue light Reflection source detector Lamp

Fluid flow

Water

Fluorescence detection unit

Spectrometer

> Detecting uorescence. The uorescence detector and spectrometer are about 7 cm [3 in.] apart along the CFA owline (bottom). In the uorescence detection unit, a blue light source reects from a sampling window into a detection channel tuned to the same wavelength. The excitation beam is reemitted at longer wavelengths. Two other channels are tuned to detect this hydrocarbon uorescence over two broad ranges of wavelengths. Most of the signal from condensates occurs in the rst of these two detectors (top).

a 32-m [105-ft] interval within the oil column. Fluid samples taken at each of these points and later analyzed in a laboratory confirmed a compositional gradient. Finding this gradient without the real-time CFA analysis would be unlikely. First, small differences between samples in a laboratory often are interpreted as reecting sampling difculties rather than uid-property variations. Second, a company would not be likely to position a tool at four separate heights in this oil leg without some prior evidence of the presence of a gradient. For Statoil, the important point was that the existence of a compositional gradient within the oil leg was noticed and conrmed in real time. This allowed the operator to adjust the MDT sampling program to identify appropriate uid sampling depths and collect sufcient quantities of samples for complete reservoir uid description.

Knowing that there was a compositional gradient in the formation and knowing the expected range of GOR helped the company develop a wellbore drawdown strategy to optimize production. Because the operator planned to develop the eld using horizontal wells, placement of the wells in relation to the gas/oil and water/oil contacts was crucial. An MDT permeability test performed during this testing sequence provided additional information for well placement. The in-situ properties determination provided by the combination of the CFA and LFA modules assured that quality samples were obtained in the sampling cylinders. Since the produced uids from this eld will be tied in to other elds, a quality uid compositional analysis and determination of compatibility with the other uids were important for ow assurance.

60

Oileld Review

During an MDT logging run, a probe seals against the formation, then the pump pulls uid from the formation into the tool. Ideally, the drawdown will be sufcient to remove the uid from the formation, but not so much as to drop the uid below its dewpoint pressure. However, both the formation permeability and the differential between dewpoint pressure and formation pressure are unknown or poorly known before most MDT jobs in exploration wells begin. In fact, determining the dewpoint is one of the main reasons for obtaining uid samples. Without knowing permeability and dewpoint pressure, establishing a proper drawdown pressure to keep a gas in single phase is difcult. The FDU on the CFA tool provides a check of this condition in situ. When water may be present in the owline, the CFA unit should be placed downstream of the pumpout module to avoid continuous swamping of the longer wavelength absorption spectrometers by the strong water-absorption peak. The water is still present in the owlines in this conguration, but the residence time in the pump is sufcient for phase segregation to occur. Thus, the water, oil and gas phases will ow through the spectrometers separately (right). On the downstroke, the lower part of the pump lls from the formation, and the upper part discharges to the owline. The discharge point is at the bottom of the upper pump chamber, so the rst uid pumped out is water, followed by oil, then gas. On the upstroke, the chambers reverse function. Now, the lower chamber discharges to the owline, but this time does so from the upper part of the chamber. The rst uid expelled is gas, followed by oil, then water. The CFA tool clearly distinguishes owing phases using the water and hydrocarbon vibrational-energy absorption peaks and the main FDU fluorescence channel. The hydrocarbon vibrational channels provide an indicator of the amount of gas passing through the owline when the ratio of C1 to C6+ is high. The FDU is sensitive to the presence of a liquid hydrocarbon phase. Thus, a plot of these three quantities water absorption, C1/C6+ ratio and the main uorescence channelshows the three phases passing through the owline. This procedure indicated three-phase ow during sampling of a gas cap in a North Sea well. In this case, the operator suspected that the gas
15. One way for a molecule to decay from an excited state is by emitting a photon. If some of the excitation energy has dissipated, for example through collision, the reemitted light is at a lower energy than the absorbed light.

Downstroke

Upstroke

To flowline

From reservoir

Gas Oil Water Hydraulic oil

From reservoir

To flowline

Downstroke

Upstroke

Fluorescence intensity, volt Water apparent density, g/cm3 Ratio of C1 to C6+, dimensionless

1.2 1.0 0.8 0.6 0.4 0.2 0 0 20 40 60 Elapsed time, s 80 100 120

> Detecting multiple phases downstream of the pumpout module in a North Sea well. The pumpout module is a reciprocating pump with two separate chambers sharing one piston. When the pumpout module strokes down (top left), a multiplex valve directs uid from the formation into the lower chamber and from the upper chamber into the owline. On the upstroke, the multiplex valve switches the inlet and outlet sources (top right). The CFA module, which is downstream of the pump, detects three phases. The chart (middle) compares signals from the water vibrational channel, from the main uorescence channel that indicates liquid oil, and from the C1/C6+ ratio that indicates gas. On the downstroke, water is expelled rst, followed by oil, then gas (assuming all are present, as they are here). On the upstroke, the order is reversed. The color bar indicates the primary ow contributor (bottom). This uid sample came from a gas-condensate zone, and the separation of the uid into gas and liquid phases indicated the drawdown was too large.

cap might contain a retrograde condensate. The gas was near a saturated condition, and the pumpout unit drew the pressure down to about 25 bar [370 psi] below formation pressure. Because a large drawdown was used to pump the uids out of the formation, two hydrocarbon phases were observed; a sample taken from this depth would be invalid because of a high likelihood of formation-uid damage at this point. After moving the MDT string a few centimeters to obtain an unaffected uid sample, the operator found this new sample contained more liquid than the rst. Detecting the inappropriate sampling condition and moving to a new location was possible because of the real-time CFA results.

Real-Time Advantages Capabilities provided by the FDU are being incorporated into real-time CFA services, increasing the sensitivity for detecting phase transitions and providing additional information about in-situ uid compositions. The ability to distinguish methane and light hydrocarbons from heavier hydrocarbons greatly increases the amount of information available in real time from gas-condensate reservoirs. This determination allows an operator to quickly make important economic decisions about a reservoir. The operator can then follow up with more extensive measurements in a surface laboratory, using samples whose quality has been assuredbefore collectionusing these innovative downhole sampling tools. MAA

Autumn 2003

61