Abstract: Characterizing, Simulating, and Eliminating Vibration Induced Counts in Measurement While Drilling Gamma Ray Detectors authors K. Kimmich and C. Grodsinsky 1 A Measurement While Drilling (MWD) gamma ray tool will report a higher than actual count rate if its scintillation crystal package produces vibration induced light pulses. The scintillation crystal is used as a transducer that measures the count rate of gamma rays coming from naturally occurring radioisotopes in down hole shales. Each gamma ray causes the crystal to emit a number of visible light photons that are converted into an electrical pulse by a photomultiplier tube. The pulse is shaped and digitized by tool electronics and registers as a count if it is above some threshold energy. The harsh shock and vibration conditions MWD gamma detectors experience can cause a crystal package to emit spurious light pulses which typical tool electronics can not distinguish from genuine gamma counts, so the detector will report a higher than actual count rate. For Geosteering applications a higher count rate typically will indicate the drill string trajectory must be changed to remain within a reservoir because such a response is indicative of a shale reservoir boundary, therefore vibration induced counts will provide a false control input making it difficult to steer correctly. This could cause the trajectory of the drill string to exit the reservoir. Measurements of vibration induced counts from MWD scintillation packages were made using an electro-dynamic vibration table, gamma ray spectroscopy electronics, and a digitizing oscilloscope. It was found that a scintillation package will produce counts as a function of the input g-level, if the resonant frequency of the crystal in its support structure is within the bandwidth of the operational vibration environment. These measurements provided the basis for computer simulations of MWD gamma ray logging that showed the detrimental effect of vibration induced counts on log accuracy. It was concluded that stiffening the scintillation crystal mechanical support and therefore, moving the rigid body resonant frequencies of the system outside the operational vibration bandwidth ensures the scintillation package will not produce vibration induced counts in a specified bandwidth. Historical Background The Measurement While Drilling (MWD) well logging technique is growing in its importance for formation evaluation in the petroleum industry. In fact, each year there are an increasing number of boreholes drilled and evaluated using MWD tools and techniques. This technique involves the incorporation of a variety of logging tools into the drill string immediately above the drill bit. The data are either stored locally until the tool is returned to the surface or transmitted to the surface during the drilling. In addition, the data can be used to make in situ decisions about the drilling operation, i.e. GeoSteering [1]. 1 Bicron, a Division of Saint-Gobain Industrial Ceramics, 1 Inc. 12345 Kinsman Road, Newbury, OH 44065 USA.
The harsh vibration, shock and thermal environments MWD tools experience has increased survival and operational specifications on the instruments that have been used in wireline logging applications. The first generation of gamma ray scintillation detectors that were designed specifically for MWD applications were constructed to survive repeated exposure to the environment. Thallium doped sodium iodide scintillation crystals were cushioned against vibration and shock in a hermetically sealed container, and were tested to ensure that performance remained consistent after a number of shock, vibration, and thermal cycles. However, survival specifications were insufficient to guarantee that the detector would operate correctly in the field, and it was discovered that the scintillation detectors would produce vibration induced counts during drilling. Introduction A natural gamma ray counting scintillation detector will register a gamma ray event once it has been absorbed by the scintillation crystal and converted into light which in turn is converted into an electrical signal. Radiation measurement scintillation devices can produce light as a function of their dynamic environment which is not caused by the incident radiation. The nuclear gamma ray log will be a function of both sources of light if the detector is not designed with the dynamic environment taken into consideration. The nonscintillated light is produced once the motion of the crystal relative to its surrounding elastomer and reflective materials is significant enough to cause a build up of static charge. Once the potential difference is great enough the voltage will discharge and cause luminous sparks and subsequently be detected by the interfaced photomultiplier tube. In order to provide an estimate of nonscintillated light produced as a function of inertial dynamic response of the scintillation crystal material, a dynamic model was developed to give the peak dynamic response of an assembled hermetic/crystal scintillator package. This model was then validated by measuring the inertial response of a plastic scintillator package which was instrumented with three linear accelerometers giving the transverse, longitudinal and rotational accelerations of the experimental configuration. (See Figure 1). In addition, the vibration induced light was measured to provide a relationship between the amount of nonscintillated light produced and the relative response of the scintillator. A simulation, based on these results, was produced to demonstrate the effects of such spurious signals on typical logging data. MWD Dynamic Equations of Motion In general there will be six rigid body degrees of freedom for each scintillator assembly, three rotational and three translational. At present, it is not known if static discharge photo-electric events are the only source of non-scintillation counts caused by mechanical excitations; however, any mechanically induced photo-electric event must be deterministic in time. Therefore, there will be at a minimum, six distinguishable event frequencies in the photo-tube signal, provided each mode is excited. If any or all of these modes cause a relative deflection by which the scintillator loses interface or contact with its surrounding 2
structure, and discharges its static electrical potential, there will be nonscintillated light produced. Each MWD detector response can be represented by a dynamic set of equations all having the following matrix form: [ M]{ u && } + [ C]{ u & } + [ K]{ u} = { f ( t ) } where: [ M ] = mass matrix, [ C ] = damping matrix, [ K ] = stiffness matrix,{ f ( t) } = time varying load vector, and{ u&& }, { u& }, { u } are the relative acceleration, velocity, and displacement vectors, respectively. By solving this generic set of equations for each MWD design, the mechanical dynamic response per design can be evaluated. These dynamic models predict the response of MWD detector designs to vibration and shock inputs, and so provide a guide to define the appropriate radial and axial compression to guarantee the required dynamic response. Model Definitions and Assumptions Each MWD style detector which has dynamic performance specifications is supported radially and axially in the hermetic housing by an elastomeric material. A wave spring loads the crystal in compression axially and maintains the optical coupling between the scintillator and the glass interface. Dynamic Response of Scintillator Assembly A finite element representation of an MWD style detector was developed in order to solve the natural frequencies or eigenvalues of the equations of motion for the scintillation package assembly. The MWD package components, boot, elastomeric pads, and scintillator, were represented by solid linear elements while the radial, and axial wave spring supports were modeled by discrete stiffness elements. Once these modes were calculated, the model was run to solve the inertial response of the system to a base dynamic excitation equivalent to the experimental validation tests performed using the vibration test stand. Figure 2 illustrates the solid and discrete elements which modeled the instrumented 1.75x4.0 scintillator package assembly. A base excitation was defined as an x axis 30 g peak sine sweep. The output given in Figure 3 gives the acceleration response curves of two scintillator nodal points from the front and rear ends of the crystal. Experimental Validation of Inertial Scintillator Response A 1.75x4.0 gamma ray scintillation detector was fabricated with three linear accelerometers placed along its long axis and aligned in the transverse direction while the 3
third accelerometer was aligned in the long axis, at the rear of the detector. This allowed for a measurement of the translational, rotational, and long axis motion of the scintillator mass. Figure 3 gives the experimental values measured of the input acceleration disturbance and the two transverse accelerometers placed at the front and rear of the scintillator. As seen in Figure 3, the experimental results support the modeled accelertion estimates. However, at the higher frequencies the experimental detector assembly deviates from the modeled results substantially. It is believed that this is caused by the scintillator slipping along the scintillator/reflector interface. In the model, the stick slip motion of the actual system was not modeled and therefore, will not provide a good estimate of the experimentally measured inertial response of the detector at the higher frequency modes. Vibration Induced Counts As discussed earlier, once the scintillator reaches some relative displacement magnitude, a discharge of the voltage potential build up between the scintillator and the surrounding materials will cause light and thus, produce vibration induced counts. This was experimentally demonstrated by exciting the scintillation package with a swept sine function in the bandwidth of the scintillators first few translational rigid body modes. This caused large relative displacements at the natural frequencies shown in Figure 3. Light output data was collected with a multichannel analyzer (MCA) in the multiscaling mode, and time correlated to the input and output acceleration data. A 3.234 Hz/sec. sine sweep was conducted from 30 to 2000 Hz providing the acceleration response shown in Figure 4. This response is the same as the experimental data shown in Figure 3. The start and stop times were noted for the data collected from the PMT and a counts per second value was calculated for the vibration time history providing the CPS values for the 30 to 2000 Hz sine sweep. Figure 4 illustrates the CPS response as a function of frequency in comparison to the scintillators acceleration response. This shows that vibration induced counts occur at some maximum displacement. While this data provides design information for MWD detectors, it does not directly predict the vibration induced count rate per random excitation. While the down hole environment will be described by some random excitation bandwidth, the inertial response data gives insight as to the amount of vibration induced counts per acceleration input one may experience. Table 1 gives tabulated results from random vibration data for a number of random acceleration disturbances with their corresponding average vibration induced CPS, calculated from the MCA experimental data time histories. As seen in Table 1, the number of average counts per second per g rms input is not linear. Therefore, the amount of light produced is not just a function of dynamics but also the physics of the discharge mechanisms. These values were used in order to simulate the effect of vibration induced light on measurement while drilling logging opperations. Effects of Vibration Induced Light on MWD Data Figure 5 shows simulated natural gamma ray MWD data from a detector that produces vibration induced counts at a rate that is a function of the g-rms input. The functional 4
relationship between vibration induced counts and input g-rms varies from detector to detector and depends strongly on how the scintillation crystal is packaged. The curves were generated by adding two values selected at random from a gaussian distribution that was centered on periodically recurring values of 10 and 40 counts per second (Radiation Counts Only), to one selected from another gaussian distribution that was centered on the values in Table 1. The gaussian distributions had a standard deviation that was equal to the square root of the mean. In the field, measured radiation count rates depend on the background count rate, which is generally a function of the shale content of the rock [2], the detector and borehole geometry, and the stopping power of the materials that are between the crystal and the borehole wall. Detectors with more efficient geometry s, i.e. larger scintillation crystals, and low density, or thin walled crystal package materials will measure a higher count rate in a given borehole than less efficient detectors. The periodic function simulates how the background count rate changes as the drill string passes through hypothetical alternating strata of shale (high background) and sand (low background). The vibration induced noise count rate increases as the input g-rms increases and causes the measured count rate to depart from the actual count rate. Input g rms Vibration Induced Noise Count Rate in cps 15 2.4 25 9.1 30 28.5 Table 1: Vibration Induced Noise as a Function of Input g-rms. The effect of vibration induced counts on the relative accuracy of the log depends on combinations of the measured background count rate and the vibration input. For example, consider a scenario where the drill string vibrates violently when the drill string bores through shaley strata, and then with less excitation when passing through sand. The vibration induced counts will not greatly effect the relative accuracy of the measurement for this particular scenario. However, if the drill string bores through rock which cause it to shake violently at 30 g-rms, but contains little shale, and then does not vibrate as it bores through a highly radioactive stratum, the data will be highly inaccurate and useless. (See Figure 6). Gamma ray logs are generally given in API units to normalize differences in detector geometry. One API unit is 1/200th of the difference between, the count rate a particular detector measures in a hot region and the count rate measured in a cold region for a standard, calibrated source [2]. The percent error due to vibration induced counts will be preserved whether a log is given in API units or in counts per second, regardless of crystal size or calibration [2]. 5
Conclusions If a well is logged while drilling with a gamma ray detector that exhibits vibration induced counts, accurate formation evaluation and correlation between logs could be impossible. MWD detectors produce counts as a function of g-level and disturbance frequency. The levels and energy bandwidth at which the detectors produce these spurious counts is largely governed by the dynamic response of the detector assemblies. These mechanically induced photo-electric events are distinguishable in time and are functions of the modal rigid body degrees of freedom for each crystal assembly. To minimize the effect of spurious counts on log data, the expected operating dynamic environment and the detector response must be understood, predicted and experimentally validated. These dynamic response predictions can then be used in designing an appropriately performing MWD detector which will not produce mechanically induced counts in its operating bandwidth. References 1. Nuclear Logging Techniques for Hydrocarbon, Mineral and Geological Applications, IEEE Vol. 35, No. 1 Feb. 1988. 2. Introduction to Wireline Log Analysis, Atlas Wireline Services, Western Atlas International Inc., 1992. 3. Cased Hole and Production Log Evaluation, James J. Smolen, PennWell, Tulsa Oklahoma, 1996. 6
Crystal f(t) table transverse accelerometers axial accelerometer hermetic package Interface Pad elastomeric support Figure 1: Experimental configuration of scintillator package. Rear Interface Pad Solid Linear Elements Discrete Spring Elements Scintillator Solid Linear Elements Front Interface Pad Linear Solid Elements Figure 2: Finite element representation of scintillator assembly. 7
1.00E+03 Transverse Acceleration (g's) 1.00E+02 1.00E+01 Front Nodal Accel. Back Nodal Accel. Input Accel. Front Trans. Accel. Rear Trans. Accel. 1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.00E+04 Frequency (Hz) Figure 3: Modeled and experimental results of transverse acceleration response. Acceleration Response (g's) and Counts Per Sec. (CPS) 1.00E+03 1.00E+02 1.00E+01 1.00E+00 1.00E+00 1.00E+01 1.00E+02 1.00E+03 1.00E+04 Frequency (Hz) Input Accel. Front Trans. Accel. Rear Trans. Accel. CPS Figure 4: Experimental results of transverse acceleration response and vibration induced light versus frequency. 8
Simulation of Measurement While Drilling Radiation Measurement Data: Count Rate as a Function of Background and Random Energy Input 90 Count Rate and Input g-rms 80 70 60 50 40 30 20 10 Radiation Counts Only Radiation Counts + Noise Random Input in g- rms 0 0 256 512 768 1024 Depth in Arbitrary Units Figure 5: Effects of nonscintillated light on MWD data. Simulation of Measurement While Drilling Data: Count Rate as a Function of Input Vibration and Depth Counts per Second 90 80 70 60 50 40 30 20 10 Radiation+Spurious Counts Radiation Counts 0 0 200 400 600 800 1000 Depth in Arbitrary Units Figure 6: Effect of vibration induced counts on MWD data as function of depth. 9