Open-hole logging: why, how and when

A well is the only direct way to reach a formation that lies several kilometres deep. Core gives us the rock itself, but taking it is expensive and slow, and it is done in only a few intervals. Logging works differently. It measures only a few properties, but it measures them continuously, over the whole section and in almost every well.

A tool cannot see porosity or oil. It records physical quantities: how many gamma rays (a penetrating radiation) the rock gives off, how well it conducts electric current, how strongly it scatters the gamma rays of a source, how quickly it slows down neutrons, and how fast sound travels through it. Passive methods record what the rock gives off or produces by itself: natural radioactivity and its own electrical potentials (voltages). Active methods send current, sound, gamma rays or neutrons into the rock and measure the response.

To get from physical quantities to the properties of the formation, we use petrophysical models: relationships that are fitted to core data and checked against core. We convert density into porosity (Porosity), resistivity together with porosity into water saturation (Archie's equation and saturation of clean reservoirs), and radioactivity into shale volume (Determining shale volume). So the porosity and saturation curves on the final interpretation plots are the result of a calculation that rests on assumptions. They are not a direct measurement.

In wireline logging, a service that is separate from the drilling rig makes the measurements. On land it uses a logging unit: a truck with a winch (a drum that winds the cable), the cable and a recording computer. On an offshore platform it uses a permanent module of the same kind. The tools go down the well on a logging cable. The cable has steel armour on the outside and insulated conductors inside. Power goes to the tool along the conductors, and data come back from the tool. The cable both holds the tool and carries the signals.

A measuring wheel, which the cable passes over, measures the depth. The depth is checked against magnetic marks that were put on the cable at equal distances under a set tension. The cable stretches under the weight of the tool and its own weight. So in the well the marks are spaced wider than when they were put on, and the depth read from the marks is corrected for the stretch. The wheel already allows for this stretch, but it goes wrong when the cable tension changes after the wheel: when the tool sticks, when it moves in jerks, and when running in changes to pulling out (depth errors on wireline). We count depth along the hole from a zero mark, usually the rotary table (the turning table on the rig floor). The heading of the log states this mark.

The sensors sit metres apart along the tool string, so they pass the same point of the hole at different times. While recording, the system shifts the readings of each sensor by its depth offset, so that all the curves on the log refer to the same depth. The same offset brings a practical limit. A sensor that sits high in the string cannot reach the bottom of the hole. The deepest point it records is as far above the bottom as the sensor is above the bottom of the string.

In logging while drilling (LWD) the sensors are built into the drill collars (heavy, thick-walled pipes) above the bit. They measure the rock minutes or hours after the bit has cut it. The delay equals the distance from the bit to the sensor, divided by the rate of penetration (the drilling speed). The related system MWD (measurement while drilling) measures the trajectory of the well: the inclination and the azimuth, from accelerometers and magnetometers. It also measures drilling parameters: weight on bit, torque and vibration. Both systems usually work in the same assembly (Measurement while drilling).

Data reach the surface through mud-pulse telemetry. A valve in the assembly makes pressure pulses in the flow of mud, and a sensor on the standpipe (the vertical pipe that feeds mud to the drill string) decodes them. Such a link carries a few bits per second, and up to about twenty in the fastest systems. This is several orders of magnitude lower than over a logging cable (each order of magnitude is a factor of ten). So only the main curves go up in real time, and only at a coarse sampling interval. The full data are written into the tool memory and read out after the assembly has been pulled out. There are other channels too: an electromagnetic one, which does not need the mud to circulate, and wired drill pipe, which has a built-in cable and sends data thousands of times faster. But they are used less often.

The main advantages of LWD:

• The formation before invasion. The filtrate has not yet gone deep, the hole is not yet washed out (widened by erosion), and the measured resistivity is closer to the true resistivity of the formation (Borehole environment: invaded zone and mudcake).
• Steering the well (geosteering). We use the GR and the resistivity measured near the bit to keep a horizontal hole inside the bed.
• Difficult wells. LWD gives data in deviated and horizontal holes, where a wireline tool will not go down by itself.
• Insurance. If the hole runs into trouble and wireline logging becomes impossible, the data have already been recorded.

The limits come from the same conditions. LWD tools must withstand shock, vibration and rotation. So for a long time LWD offered fewer methods than wireline. Today it includes, besides the GR, the resistivity, density and neutron logs, and also shear-wave sonic logging, nuclear magnetic resonance logging, imagers (tools that make pictures of the borehole wall) and formation pressure measurement. But wireline still does the taking of sidewall cores (small rock samples cut from the borehole wall) and many special surveys. We measure LWD depth from the length of the drill string, not by the logging cable, so the two depth systems have to be matched.

A tool reading is not the value at a single point. It is an average over some volume of rock. Two characteristics describe the size of this volume. Depth of investigation is the distance from the borehole wall from which most of the signal comes. Vertical resolution is the smallest bed thickness that the tool shows with its true value.

The design of the tool sets both characteristics: the spacing between the source and the receiver (or between electrodes or coils), the size of the detectors, and how the current is focused. The larger the spacing, the deeper the tool sees, but the coarser its vertical resolution. That is why a suite includes sondes of different lengths. Short ones are for the zone near the borehole and for thin beds. Long ones are for the uninvaded formation (the rock that the filtrate has not reached).

If a bed is thinner than the resolution, the tool sees the bed and the surrounding rocks at the same time, so the reading never reaches the true value. This is the shoulder-bed effect. A thin sand bed in shales looks more shaly on the log than it really is. A thin tight bed (a bed with very little pore space) in a porous sandstone looks more porous. The simplest model is that the tool averages the property evenly over a window of length $L$, about equal to its resolution.

The sampling interval is usually 15 cm, and down to a few millimetres for high-resolution tools. It must be several times smaller than the resolution. Otherwise a thin bed can slip between the samples. But more frequent sampling does not improve the resolution by itself. The physics of the tool sets the resolution, and frequent samples only describe the already smoothed curve in more detail.

We read a log from top to bottom, like a section of the well. The vertical axis is the depth along the hole. Across the page there are several tracks (columns), each with its own scale. In the classic format the depth track stands between the first and second tracks. On the left are the lithology curves (GR, SP, caliper). On the right are the resistivities on a logarithmic scale and the porosity methods.

We choose the depth scale to suit the task: 1:500 for an overview and for correlation (1 cm of the log is 5 m of hole), and 1:200 for a detailed study of beds (1 cm is 2 m). In Russian-language practice, general surveys over the whole hole use 1:500, and detailed surveys in the prospective (possibly productive) intervals use 1:200. Resistivity is shown on a logarithmic scale (each step is ten times the last), because in one section it varies from tenths to thousands of ohm-metres.

Depth on a log is measured along the hole. In a deviated well, a bed looks thicker along the hole than it does vertically. So for maps and volume calculations we use the directional survey to convert the depths to true vertical depths and the thicknesses to vertical or true thicknesses. We match the records of different runs with each other by the GR curve, which every tool string includes. We shift both the LWD data and the core to fit its characteristic peaks.

Today we get logs in digital form (for example in the text format LAS, Log ASCII Standard) and plot them at any scale and with any set of tracks. But the logic of the classic format has survived. Recording, formats and the heading are covered in the topic Log acquisition, processing and presentation. The order of interpretation is covered in the topic Log interpretation workflow.

Wireline logging is usually done at the end of drilling each interval, before the next casing string (the next length of casing) is run, while the hole is still open. One recording of an interval by one tool is called a log run. Runs are numbered separately for each method, starting from its first recording in the well. The run number is given in the heading of the log. So the run numbers of different methods in one interval need not be the same. A difference only shows that one of the methods was not recorded in the upper intervals, which were drilled earlier.

The objectives of the well and the cost of its time decide which surveys to run. In pay intervals, and where drilling showed signs of oil and gas, we record more methods than in a part of the section that is clearly of no interest. In an exploration well we record the full suite even if the result is negative: every such well is expensive, and it is the only chance to study the section. In development wells (wells drilled to produce from a known field) of a well-studied field a reduced suite is usually enough, and some of the methods move to LWD.