8 petrophysical techniques

7/29/2019 8 Petrophysical Techniques

1/13

8Petrophysical Techniques

GENERAL

Petrophysical techniques provide most of the sub-surface data available to an exploration geologist.Besides their importance in completion decisions, they

are also invaluable methods for mapping and identify-ing lithologies. Presented here are nine techniques thatcan assist geologists with lithologic determination andmapping. They are:

Neutron-density lithology plot

Neutron-sonic lithology plot

Density-sonic lithology plot

M-Nlithology plot

Matrix identification plot (maa vs. tmaa)

Matrix identification plot (maa vs. Umaa)

Alpha mapping from the SP log

Clean sand or carbonate from gamma ray log Rock typing and facies mapping.

These techniques are especially important to a ge-ologist when lithologic data from core or samples areunavailable.

Figures 8.1 and 8.2 are logs through the SilurianFusselman Formation in West Texas. The log suite forthis well consists of a dual induction-SFL, SP, gammaray, and neutron-lithodensity log (Figure 8.1), and asonic log with gamma ray (Figure 8.2). Using datafrom these logs, we will illustrate how lithologies aredetermined using the first six methods listed above. It

is important to mention that the porosity zone (9088 to9126 ft) in the Fusselman has a combination of vuggyand intercrystalline porosity. How the presence ofvuggy porosity affects the different lithology plots willbe discussed in the text below. The software used togenerate the results from the original log data is calledCBA (Carbonate Advisor; The Logic Group, 1994).The various lithology plots and crossplots were ren-dered in PetroWorks, from Landmark Graphics, a Hal-liburton company.

NEUTRON-DENSITY LITHOLOGY PLOT

The gamma ray log (Chapter 3) measures the natu-ral radiation of a formation and primarily functions asa lithology log. It helps differentiate shales (high

radioactivity) from sands, carbonates, and anhydrites(low radioactivity). The neutron log is a porositydevice that is used to measure the amount of hydrogenin a formation, which is assumed to be related toporosity (Chapter 4). The density log is a porositydevice that measures electron density, and from that,formation bulk density (Chapter 4). When these threelogs are used together, lithologies can be determined.Figure 4.8 is a schematic illustration of how responsesfrom the gamma ray, neutron, and density logs arerelated to lithology.

Figure 8.3 is a neutron-density lithology plot of thedata from the Fusselman Formation (Figures 8.1 and

8.2). On the lithology plot (Figure 8.3), the Fusselmandata cluster between the lithology lines of dolomiteand limestone. The amounts of each lithology areillustrated in the column next to the lithology plot. Asa quick-look method, where there are a limited num-ber of lithologies, this plot works well for basic litho-logic and facies mapping. The problem is how thegeologist determines if the Fusselman is a mixture ofdolomite and limestone or a mixture of dolomite andsandstone.

NEUTRON-SONIC LITHOLOGY PLOT

The sonic log is a porosity device (Chapter 3) thatmeasures the speed of a compressional sound wavethrough the formation. Figure 8.4 is neutron-sonic plotfor the Fusselman Formation. The Fusselman data fromthe porous zone (9088 to 9126 ft) cluster along thedolomite lithology line, and the data from the non-porous zone above 9088 ft cluster along the limestonelithology line (Figure 8.4). The amounts of each lithol-ogy are illustrated in the column next to the lithology

137

Asquith, G., and D. Krygowski, 2004, Petrophysical

Techniques, inG. Asquith and D. Krygowski, Basic

Well Log Analysis: AAPG Methods in Exploration

16, p. 137149.


2/13

plot. A comparison of Figures 8.3 and 8.4 reveals thatthe neutron-sonic plot indicates a much more dolomiticlithology compared to the neutron-density plot. Thisdifference is due to the presence of vuggy porosityfrom 9088 to 9126 ft. Remember that the sonic logmeasures only matrix porosity (intergranular and inter-

crystalline) and the nuclear logs (neutron and density)measure total porosity. Therefore, when vuggy porosi-ty is present sonic porosity is less than total (neutron-density) porosity and the data cluster lower on the neu-tron-sonic plot (i.e., more dolomitic; Figure 8.4).

DENSITY-SONIC LITHOLOGY PLOT

Figure 8.5 is a density-sonic plot of the Fusselmaninterval illustrated in Figures 8.1 and 8.2. The calcu-lated lithologies are displayed in the column next tothe lithology plot. The density-sonic plot is the leastdesirable of the lithology plots because the lines rep-

resenting the different lithologies (i.e., sandstone, lime-stone, and dolomite) are much closer together. There-fore, it is harder to differentiate the different litholo-gies. An examination of the lithology for the vuggyzone (9088 to 9126 ft) reveals that the lithology is nowlimestone (Figure 8.5). The presence of the vuggyporosity causes the sonic log to underestimate porosi-ty, which shifts the data to the left of the plot so thatthe lithology appears to be limestone. It is important toremember that all lithology plots that require a soniclog may be unreliable if there is a large amount ofvuggy porosity. However, as can be observed in Fig-ures 8.4 and 8.5, this confusion about lithology in the

interval from 9088 to 9126 ft informs the log analystthat the reservoir has vuggy porosity. Or to state itanother way, if all the porosity were intergranular orintercrystalline, all the lithology plots would indicatesimilar lithologies.

M-NLITHOLOGY PLOT

TheM-Nplot requires a sonic log, neutron log, anddensity log, of which are all necessary to calculate thelithology-dependent variablesMandN.MandNval-ues are largely independent of matrix porosity (inter-granular and intercrystalline). A crossplot of these

two variables makes litholgy more apparent. MandN values are calculated by the following equations(Schlumberger, 1972):

8.1

8.2

where:

t= interval-transit time in the formation (from thelog)

tfl = interval-transit time in the fluid in the forma-tion

b = formation bulk density (from the log)

fl = fluid density

N= neutron porosity (in limestone units, from thelog)

Nfl = neutron porosity of the fluid of the formation(usually = 1.0)

When the matrix parameters (tma, ma, Nma;Table 8.1) are used in theMandNequations instead offormation parameters,MandNvalues can be obtained

for the various minerals (Table 8.2). Note that thequantities M and N above are related to lithology andhave no relation to the cementation exponent (m) andthe saturation exponent (n) in Archies equation.

Figure 8.6 is anM-Nplot of data from the SilurianFusselman Formation illustrated in Figures 8.1 and8.2. The data above a depth of 9088 ft plot in the sand-calcite-dolomite triangle. The calculated lithologiesare illustrated in the column next to theM-Nplot. Thedata from the porous vuggy interval (9088 to 9126 ft)plot above the tie line between calcite and dolomite,indicating the presence of vuggy porosity. If onlymatrix intercrystalline or intergranular porosity were

present, the data would have plotted on or below thecalcite-dolomite tie line. Remember, that the presenceof vuggy porosity results in an underestimation ofsonic interval transit time and therefore an overestima-tion ofM. Because the data from the interval from9088 to 9126 ft plot out of the lithology triangle, thecalculated lithologies are not reliable.

MATRIX IDENTIFICATION PLOT (maaVS. tmaa)

The matrix identification plot, like theM-Nplot, isa crossplot technique that helps identify lithology andsecondary porosity. As with the M-Nplot, the matrix-identification plot requires data from neutron, density,and sonic logs.

The first step in constructing a matrix identificationplot is to determine values for the apparent matrixparameters, apparent matrix density (maa) and appar-ent matrix traveltime (tmaa). These values are calcu-lated from neutron (N), density (b) and sonic (t)data using the following equations (Western Atlas,1995):

138 ASQUITH AND KRYGOWSKI


3/13

Petrophysical Techniques 139

Table 8.1. Matrix Coefficients of Several Minerals and Types of Porosity (Liquid-filled Boreholes).

Minerals tma ma SNPma* CNLma* Pe

sec/ft g/cm3 (decimal) (decimal) (b/e)

Sandstone (1): Vma = 18,000 ft/sec; > 0.10 55.5 2.65 -0.035a -0.05a 1.81

Sandstone (2): Vma = 19,500 ft/sec; < 0.10 51.2 2.65 -0.035a

-0.005 1.81

Limestone 47.5 2.71 0.00 0.00 5.08

Dolomite (1): = 0.055 to 0.30 43.5 2.87 0.035a 0.085a 3.14

Dolomite (2): = 0.015 to 0.055 and > 0.30 43.5 2.87 0.02a 0.065a 3.14

Dolomite (3): = 0.0 to 0.015 43.5 2.87 0.005a 0.04a 3.14

Anhydrite 50.0 2.98 -0.005 -0.002 5.06

Gypsum 52.0 2.35 0.49b 0.6 3.99

Salt 67.0 2.03 0.04 -0.01 4.65

a Average values

b Based on hydrogen-index computation

*SNPma and CNLma values are specific to Schlumberger neutron tools.

From Schlumberger (1972). Courtesy Schlumberger Well Services.

Table 8.2. Values of M and Nlithology parameters, calculated for common minerals.

MineralsFresh mud Salt mud

= 1.0 g/cm3 = 1.1 g/cm3

M N M N

Sandstone (1): Vma = 18,000 ft/sec; > 0.10 0.810 0.628 0.835 0.669

Sandstone (2): Vma = 19,500 ft/sec; < 0.10 0.835 0.628 0.862 0.669

Limestone 0.827 0.585 0.854 0.621

Dolomite (1): = 0.055 to 0.30 0.778 0.516 0.800 0.544

Dolomite (2): = 0.015 to 0.055 and > 0.30 0.778 0.524 0.800 0.554

Dolomite (3): = 0.0 to 0.015 0.778 0.532 0.800 0.561

Anhydrite 0.702 0.505 0.718 0.532

Gypsum 1.015 0.378 1.064 0.408

Salt 1.269 1.032From Schlumberger Log Interpretation Principles, 1972 Schlumberger. Courtesy Schlumberger Well Services.


4/13

8.3

8.4

where:

maa = apparent grain density in g/cm3 or Kg/m3

tmaa = apparent matrix interval transit time insec/ft orsec/m

b = bulk density from the log

fl = density of fluid

t= interval transit time from the log

tfl = interval transit time of fluid

ND = neutron-density crossplot porosity

SN= sonic-neutron crossplot porosity

When matrix parameters (tma, ma, and Nma;Table 8.1) are used in the matrix identification equa-tions instead of formation parameters, tmaa and maavalues can be obtained for the various minerals (Table8.3).

Figure 8.7 is a matrix identification plot of datafrom the Silurian Fusselman Formation illustrated inFigures 8.1 and 8.2. The data above a depth of 9088 ftplot in the quartz-calcite-dolomite triangle with thecalculated lithologies presented in the column to the

right of the matrix identification plot. The porousinterval with vuggy porosity (9088 to 9126 ft) plotsabove the quartz-calcite-dolomite triangle as it did onthe M-Nplot (Figure 8.6). The calculated lithologiesare illustrated in the column to the right of the litholo-gy plot (Figure 8.7). Therefore, as with theM-Nplot(Figure 8.6), the lithologies determined for the porosi-ty zone (9088 to 9126 ft) from the matrix identificationplot (Figure 8.7) are unreliable.

MATRIX-IDENTIFICATION PLOT (UmaaVS. maa)

The third-generation density logs provide not onlya bulk density (b) measurement but also a photoelec-tric absorption (Pe) measurement. The photoelectricmeasurement reflects the average atomic number of

the formation and is therefore a good indicator of theformations lithology (Dewan, 1983). The measure-ments ofb and Pe are made from the same densitytool at the long-spaced detector using energy selection(high energy for b and low energy forPe) of theincoming gamma rays. On the log (Figure 8.1), thePecurve is displayed with the neutron and density curvesin track 3. The units for the Pe curve are barns/elec-tron.Pe values for some of the common minerals arelisted in Table 8.1.

The matrix-identification (Umaa vs. maa) plot usesapparent matrix density (maa) and apparent volumet-ric cross section (Umaa) in barns/cm

3 (Table 8.3). Umaa

is calculated from the Pe, N, and b values using thefollowing equation (Western Atlas, 1995):

8.5

where:

Umaa = apparent volumetric cross section in barns/electron

Pe = photoelectric absorption in barns/electron

b = bulk density

ND = neutron-density porosity

Ufl = photoelectric absorption of fluidWhen matrix parameters (ma,Pe and Nma; Table

8.1) are used in the matrix identification equationsinstead of formation parameters, Umaa and maa valuescan be obtained for the various minerals (Table 8.3).

Figure 8.8 is a matrix identification plot (Umaa vs.maa) of data from the Silurian Fusselman Formationillustrated in Figures 8.1 and 8.2. The data above adepth of 9088 ft and the data from the vuggy porosityzone (9088 to 9126 ft) all plot in the quartz-calcite-dolomite triangle with the calculated lithologies pre-sented in the column to the right of the matrix identi-fication plot. Because the Pe measurement is insensi-

tive to the presence of vuggy porosity, the lithologiesare all reliable.

Therefore, if only two lithologies are present in avuggy reservoir the preferred lithology plots are neu-tron-density and matrix identification (Umaa vs. maa);however, if three lithologies are present in a vuggyreservoir the preferred lithology plot is the matrixidentification plot (Umaa vs. maa). If the porosity ismatrix intergranular or intercrystalline the neutron-


Table 8.3. Apparent matrix values for calculated for common minerals

Mineral maa tmaa Umaa(g/cm3) (sec/ft) (barns/cm3)

Sandstone 2.65 55.5 4.78

Limestone 2.71 47.5 13.80

Dolomite 2.87 43.5 8.98

Anhydrite 2.98 50.0 14.90

Gypsum 2.35 52.0 13.97

Salt 2.03 67.0 9.68


5/13

density, neutron-sonic, and density-sonic plots workfor two lithologies and theM-Nand both matrix iden-tification plots work for three lithologies.

Figure 8.9 is a plot of the lithologies from thematrix identification plot (Umaa vs. maa) (Figure 8.8)and neutron-density porosities for the Silurian Fussel-

man Formation. The digital lithology and porosity datatogether with depths can be used in a mapping pro-gram for further analysis.

ALPHA MAPPING FROM THE SP LOG

The spontaneous potential (SP) log (Chapter 2) canbe used to map clean (shale-free) sands vs. shalysands. The technique is called alpha () mapping(Dresser Atlas, 1974) and is based on the observationthat the presence of shale in a formation decreases theSP response.

The alpha method can be extremely valuable in

mapping because it can more narrowly define desir-able zones. Alpha values from nearby wells can beused to construct clean-sand (high-energy) maps (ineffect, mapping isoalpha values).

To construct an alpha map, the static spontaneouspotential (SSP) that a sand would have, if it were com-pletely shale-free and unaffected by bed thickness, iscalculated. The equation for SSP is:

8.6

where:

SSP = static spontaneous potentialK= 60 + [0.133 formation temperature (F)]

Rmf = resistivity of mud filtrate at formation tem-perature

Rw = resistivity of formation water at formationtemperature

The SSP must be calculated for the formation ineach well, so that variations in Rmf and Rw can beaccounted for. Next, the alpha values are determinedby the method shown in Figure 8.10. The alpha cutoff(0.75, 0.50, or whatever value chosen) is arbitrary,but should be based on production histories in the area.

The resulting alpha () map delineates clean sandenvironments. In the above example (Figure 8.10), thegreater alpha thickness for a given alpha cutoff (i.e.,0.75 or 0.50) indicates a greater thickness of high-er energy, low-shale sandstones. Also, because thepresence of shale in a sandstone can decrease its per-meability, an alpha map indicates better reservoir con-ditions.

The problem with alpha mapping from an SP log isthat SP response is decreased, not only by shale, butalso by thin beds (less than 10 ft) and the presence ofhydrocarbons (Chapter 2). Bed thickness problems areminimized by making an SP log bed thickness correc-tion (Chapter 2). But the SP log cant be corrected for

hydrocarbons.

CLEAN SAND OR CARBONATE MAPS

FROM GAMMA RAY LOG

The gamma ray log can be used to map clean(shale-free) sandstones or carbonates vs. shaly sand-stones and carbonates. Because shales are more radio-active than clean sandstones or carbonates (Chapter 3),when the percentage of shale increases in these rocktypes, the gamma ray reading also increases.

Figure 8.11 is a gamma ray density neutron logthrough the Mississippian, upper Mission Canyon For-mation in Roosevelt County, Montana. In this interval,crinoid and fenestrate-bryozoan bioherms are com-monly developed. Because the bioherm facies is com-posed of clean carbonate relative to the facies that isnot bioherm, the gamma ray log can be used to mapthe bioherm facies. The procedure for obtaining aclean carbonate cutoff from a gamma ray log isdescribed in Figure 8.10.

A gamma ray value of 20 API units on the gammaray log (Figure 8.11) represents clean carbonate with avolume of shale (Vshale) equal to or less than 0.05(5%). By drawing a vertical line on the gamma ray log

equal to 20 API units (Figure 8.10), the geologist canidentify and map the clean carbonate (or sand).Figure 8.12 is an isopach map of clean carbonate

for the upper Mission Canyon Formation in RooseveltCounty, Montana. Because the relationship betweenclean carbonate and the crinoid fenestrate-bryozoanbioherm facies is already established, the map (Figure8.12) delineates the distribution of the bioherm facies.Clean carbonate maps have also been used to map thePennsylvanian banks (bioherms) of north centralTexas (Wermund, 1975).

ROCK TYPING AND FACIES MAPPING

An important contribution to subsurface analysis ofcarbonate rocks has been the attempt to establish rela-tionships between log responses and carbonate facies.Pickett (1977), Asquith (1979), and Watney (1979,1980) used crossplots to identify the relationship oflog response to rock type. Table 8.4 is a list of thecrossplots applied by these authors.



6/13


To date, crossplots have been used to establish logvs. lithology relationships only when petrographicdata are available from cores or cuttings in selectedwells. Petrographic analysis from selected wells is

essential to firmly establish rock type.When establishing log/lithology relationships, log

responses from control wells (i.e., wells with petro-graphic analysis) are crossplotted. Next, areas thatdelineate rock-type clusters are outlined (see Figure8.13) on the crossplot. Finally, log responses fromwells without cores or cuttings are added to the cross-plot. The carbonate rock type and depositional envi-ronment of wells without petrographic analysis canthen be determined by the cluster in which each occurson the crossplot chart (see Figure 8.13).

In Figure 8.13, the solid black circles and squaresrepresent data from wells where petrographic analysis

was used to determine carbonate rock type and depo-sitional environment. The open circles represent datafrom wells without petrographic analysis. The carbon-ate rock types and depositional environments weredetermined by the cluster in which the open circleswere plotted.

Figure 8.14 is a crossplot of deep resistivity (Rt) vs.sonic porosity (S) for the Lower Permian, Council

Grove B-zone in Ochiltree County, Texas. Clusters forthe three carbonate rock types (oolite grainstone,oolitic wackestone, and argillaceous bioclastic wacke-stone) were established by petrographic analysis ofcores and cuttings (open circles). The solid circles rep-resent data from wells with only log control. Figure

8.15 is a facies map of the Council Grove B-zonebased on the percentage distribution of the three car-bonate rock types established by the resistivity/sonic-porosity crossplot (Figure 8.14).

The advantage of log crossplot techniques is thatthey maximize use of available information. Cores andcuttings are required from only a few control wellsrather than all wells. This is very important in subsur-face facies mapping because of the difficulty in obtain-ing cores and cuttings from every well in an area. Also,because petrographic analysis of every well is unnec-essary, a great deal of time can be saved.

However, it should be emphasized that petrograph-

ic analysis of cores or cuttings from control wells is anessential first step to firmly establish the rock-typecluster used in the crossplots.

REVIEW

1. Neutron and density logs can be used in a cross-plot to determine lithology when a limited number ofrock types are present.

2. Where lithology is more complex, a sonic log ora Pe curve used in conjunction with the neutron anddensity curves is necessary to constructM-Nlithology

or matrix identification crossplots.3. Spontaneous potential (SP) and gamma ray logs

can be used to map shaly vs. nonshaly carbonates orsandstones.

4. Crossplotting of multiple log responses can beused to establish relationships between log responsesand rock types, provided some petrographic data fromcores or cuttings are available.

Table 8.4. Types of Carbonate Rock-identification Crossplots (after Pickett, 1977;Asquith, 1979; and Watney, 1979 and 1980).

t(interval transit time) vs. N(neutron porosity)

b (bulk density) vs. N(neutron porosity)

b

(bulk density) vs. t(interval transit time)

Rt(deep resistivity) vs. N(neutron porosity)

GR (gamma ray) vs. N(neutron porosity)*

Rt(deep resistivity) vs. S (sonic porosity)

*Watney (1979 and 1980) also uses neutron log readings

measured in counts/second.


7/13


Figure 8.1. Induction, neutron, and density log through the Silurian Fusselman Formation, West Texas.

Figure 8.2. Sonic log through the Silurian Fusselman Formation, West Texas.


8/13


Figure 8.3. Neutron-density crossplot and resulting lithology estimation, Fusselman Formation, West Texas. Data are from Figures 8.1 and 8.2. A lithology key is shown in Figure 8.6.

Figure 8.4. Neutron-sonic crossplot and resulting lithology estimation, Fusselman Formation, West Texas. Data are from Figures 8.1 and 8.2. A lithology key is shown in Figure 8.6.


9/13


Figure 8.5. Density-sonic crossplot and resulting lithology estimation, Fusselman Formation, West Texas. Data are from Figures 8.1 and 8.2. A lithology key is shown in Figure 8.6.

Figure 8.6. M-Nlithology crossplot and resulting lithology estimation, Fusselman Formation, West Texas. Data are from Figures 8.1 and 8.2.


10/13


Figure 8.7. Matrix identification plot (apparent matrix density vs. apparent matrix traveltime) and resulting lithology estimation, Fusselman Formation, West Texas. Data are from Figures8.1 and 8.2. A lithology key is shown in Figure 8.6.

Figure 8.8. Matrix identification plot (apparent matrix density vs. apparent matrix capture cross section) and resulting lithologic estimation. Fusselman Formation, West Texas. Data arefrom Figures 8.1 and 8.2. A lithology key is shown in Figure 8.6.


11/13


Figure 8.9. Lithologiesfrom the matrix identificationplot and porosities from theneutron-density crossplot,Fusselman Formation, WestTexas.

Figure 8.10. Determining alpha () from an SP log.

Two different cutoffs are demonstrated: 50% alpha ( < 0.50) and 75% alpha(


12/13


Figure 8.11. Neutron-density-gammaray log in the Mission Canyon Formation,Montana

Given: The volume of shale (Vshale) cutoffis arbitrarily set at 5% (Vshale= 0.05).

Next, determine the gamma ray index

from the chart in Figure 3.2, Chapter 3

Gamma ray index, IGR= 0.10 when Vshale= 0.05

Determine gamma ray cutoff (see log andFigure 3.2).

Remember:

3.1

where:

GRlog= gamma ray logGRmax= gamma ray maximum(shale)

GRmin= gamma ray minimum (shale-free sandstone or carbonate)

Figure 8.12. Isopachmap of clean carbonatesfrom the MississippianMission Canyon Formation,Roosevelt County, Montana,described in the text and inFigure 8.11.

From the log:

GRmax= 90 API units (from shale zone on log)

GRmin= 12 API units (from clean carbonate zone on log)

IGR= 0.10 (IGRfor Vshale= 0.05; given)

Then, rearranging the equation above to solve for GRlog:

GRlog= 19.8 (round off to 20 API units)

20 API units represents clean carbonate where the volume of shale is equal to (or less than)5%.

Draw a vertical line from the scale value of 20 API units and determine the thickness andlimits of the clean carbonate formation (bioherm) much as you determined alpha values in

Figure 8.10.


13/13


Figure 8.13. Example crossplot of formation resistivity, Rt (in this case from a deeplaterolog) vs. neutron porosity.

This comparison of log response to facies helps the geologist develop rock-type clusters.This example is from the Ordovician Red River C and D zones and Richland and RooseveltCounties, Montana (after Asquith, 1979).

Solid squares and circles represent wells with core or cuttings available, in addition to logresponse. Open circles represent wells with log control only. Facies classifications are firstconfirmed by core or cuttings analysis, but once clusters are established only log control isnecessary for facies prediction.

Figure 8.14. Example crossplot of formation resistivity (Rt) (in this case from deepinduction) with sonic porosity (S).

As with Figure 8.13, the rock-type clusters are developed by core or cuttings analysis, butwell logs are all thats necessary once the relationship is defined. This example comes fromthe Lower Permian Council Grove B-zone, Ochiltree County, Texas. After Asquith (1979).

Open circles represent wells with both core/cuttings analysis and log control. Solid blackcircles represent wells with only log control.

Figure 8.15. Example facies map of the LowerPermian Council Grove B-zone, Ochiltree County, Texas.

The map was prepared from the facies clustersestablished by crossplotted log data in Figure 8.13.After Asquith (1979). The legend defines the positionof core, cuttings, and log-only control.

8 petrophysical techniques

Documents