Effective economic evaluation and exploitation of hydrocarbon-bearing formations requires the accurate determination of some important physical properties of the reservoir fluids. Current formation evaluation techniques require sampling of the reservoir fluids, typically using a wireline tool, retrieval to the surface and transportation to a PVT laboratory for analysis. This stepwise process is time-consuming, costly, and in some cases unreliable. More direct and reliable ways of assessing the fluid in the reservoir are therefore sought in the petroleum exploration industry.

The current project focuses on the identification, development and characterisation of compact and robust sensors for the in situ measurement of reservoir fluid physical properties. In particular, the properties of interest were narrowed down to the density, viscosity and bubble pressure since these important properties are currently difficult to measure in the reservoir environment. In general terms, knowledge of density contributes to an estimate of the commercial value of the produced fluid, while viscosity serves as one indicator of the ease with which the fluid can be extracted from the porous formation. Measurements of the bubble curve, or even of just the bubble pressure at the formation temperature, also provide key information for successful exploitation of the reservoir.

The initial stage of the project was to identify, develop and characterise novel sensing technologies for the determination of the aforementioned physical properties and phase behaviour of petroleum reservoir fluids. The requirements for such sensors were: a measurable density range of 600 - 1000 kg•m^-3 and a viscosity range of 0.3 - 300 mPa•s; compact, robust and reliable design for downhole use; the ability to withstand reservoir conditions; and the ability to handle corrosive fluids. Three sensor types were found to be broadly compatible with these requirements: the vibrating wire viscometer modified for downhole applications, a Micro-Electro-Mechanical-Systems (MEMS) cantilever oscillator, and a shear-mode piezoelectric transducer. The vibrating wire sensor for downhole application uses a tensioned metallic wire fixed at both ends. It is sensitive to viscosity but not sensitive (enough) to density, unlike the laboratory device that uses a freely-suspended tensioning weight and which is sensitive to both density and viscosity. The potential of this device to detect bubble points has not been investigated previously. The wire is driven into transverse oscillation and its motion is detected electromagnetically. The emf induced across the wire due to its motion in the magnetic field is proportional to the velocity and is related to the properties of the surrounding fluid through rigorous working equations. The MEMS devices are sensitive to both density and viscosity. These devices are based on a similar operating principle to the vibrating wire but make use of a cantilever flexural plate instead. Owing to the complex geometry of the plate, only a simplified model exists to relate the oscillation characteristics to the properties of the surrounding fluid. Shear-mode piezoelectric transducers are sensitive to the product of density and viscosity and they have been shown in the past to be excellent detectors of the vapour-liquid phase boundary. They are made up of an electroded quartz crystal plate set into shear-mode oscillations through the piezoelectric effect by application of an electric field across the plate. The electrical impedance measured across the electrodes may be related to the fluid properties through theory developed from first principles.

In order to evaluate these devices, a tubular test facility (loop) has been constructed that accommodates a number of sensing devices at the same time, and is able to handle complex mixtures and to simulate the reservoir environment with temperatures up to 175ºC and pressures up to 1379 bar (20,000 psi). Temperature control was provided by an air-bath and pressure control by means of a computer-controlled precision syringe pump that delivered hydraulic fluid to the system. This syringe pump was placed outside the controlled temperature environment since it was not rated for the maximum test temperature. To enable the handling of two-phase multi-component mixtures a magnetic circulation pump and a variable-volume device were included in the loop. Both were designed and built in–house. The magnetic pump was needed to circulate and homogenise the test mixture. It was based on an innovative all-metallic design that was able to offer contactless and near-continuous pumping with a maximum flow rate of 0.3 l∙min^-1, and a maximum differential pressure head of 2.2 bar. The variable volume device served as the interface between the test fluid and the hydraulic circuit. It comprised a cylindrical pressure vessel fitted with an internal piston allowing for a swept volume of approximately 100 cm³. The test loop was mounted on a plate suspended from a pivot point about which it could be rocked back and forth by means of an air-actuator. This rocking served to activate an agitation device within the variable volume vessel and thereby to aid homogenisation of the test fluid. A high-precision pressure transducer and a number of platinum resistance thermometers were also included in the test loop.

Sensors were tested for accuracy and reliability. Accuracy was assessed by comparison of the experimental data with established literature correlations or independent measurements made using conventional high-precision laboratory instruments for a few key test fluids. These tests were designed to probe as many aspects of the sensors behaviour as possible and included: static viscosity measurements with pure fluids of widely differing viscosities; viscosity measurements at various flow velocities up to 2 m•s^-1; viscosity and bubble pressure measurements with two-phase mixtures; sensor behaviour in a non-Newtonian fluid. The vibrating wire was found to have an overall accuracy of ±2 % in viscosity and was capable of detecting the bubble pressure to within ±0.5 bar. For the later, four indicators were tested: the indicated viscosity, the precision with which the measured resonance curve was fitted by the working equation, the estimated vacuum resonance frequency (an indication of the fluid density), and the induced voltage at a constant frequency. All four were found to change markedly on crossing the phase boundary and to be good indicators. The MEMS device was found to have an accuracy of ±1 % in density but errors of up to ±8 % were found in viscosity.

In the final stage of the project, we plan to consider the three devices together as a composite sensor system, and to explore the optimal way of using this system to indicate the desired physical properties.