Locating Oil-Water Interfaces in Process Vessels

Description:

Introduction:

            In the first part of the project, the goal was to evaluate the maximum possible pressure resolution attainable using differential detection methods and a receiver utilizing spatial demultiplexing.  A test procedure was outlined in the project proposal and QA narrative to determine the maximum resolution.  In the previous report, difficulties in constructing the required experiments were reported.  An optical spectrum analyzer (OSA) replaced the intended spatial demultiplexer in order to perform a preliminary evaluation of sensor resolution.  Oil and water were successfully differentiated using a float sensor design.  A diaphragm sensor design showed promise, but required an instrument with more resolution than the OSA. 

Since the last report, the spatial demultiplexing receiver was successfully implemented and thoroughly tested.  This report describes the results of these tests and the procedures used.  The receiver is capable of higher wavelength resolution (and therefore pressure resolution) than the OSA. 

Receiver Optimization:

            The receiver configuration under evaluation is depicted in Figure 1.  As mentioned in the last report, we are using a grating with less-than-optimal properties due to difficulties in securing an ideal grating from a manufacturer.  Since last report, an imaging lens has been added.  The lens lightly focuses the incoming light onto the camera.  The location of the focused light depends on the location of the incident light on the lens, which is in turn dependent on the wavelength.  As mentioned in the QA narrative, there exists a design trade-off between the size of the focal spot and the instrument resolution. 

Our efforts focused on the proper placement of the lens with respect to the grating and the camera, and the proper choice of the lens’ focal length.  Three different lenses having focal lengths of 19mm, 38mm, and 50mm were evaluated.  Each lens was positioned at several distances d from the camera, starting with d equal to the focal length and increasing in 10% increments.  The distance was further refined by identifying the two distances that provided the best resolution and positioning the lens at a point halfway between these distances.  The lens position with respect to the grating was determined with multiple sensor signals, as the lens diameter limits the maximum spatial difference between any two wavelength to be detected.

Results Since Last Report:

The first experiments focused on selecting the proper focal length of the lens and positioning with respect to the camera.  For the experiments, a single sensor signal was used, and the signal wavelength varied using a piezo-electric stretcher.  Wavelength changes were calibrated using the OSA at selected values of the voltage applied to the stretcher.  The lens position was adjusted so that the focused light formed a circular spot on a relatively noise-free part of the camera.  A circular spot ensures that the incoming optical rays are parallel to the optical axis of the lens, and therefore changes in location and shape of the recorded light pattern are due only to wavelength changes.  Camera data was converted to a bit-map file and then into spreadsheet format for performing data processing.  The centroid algorithm described in the QA narrative was applied to the data using MS Excel.  Before applying the algorithm, however, some pre-processing was required, as light levels often caused the digital output from the camera to “cycle” one or more times – that is to reach 255 and cycle back to 0 one or more times for high optical powers incident at the pixel location.

Typical results as a function of focal length are shown in Figure 2.  The 50mm focal length lens did not produce significant movement in the spot location on the camera for most distances, and was quickly discounted.  The 38mm lens produced significant movement (c.f. Figure 3), with an optimal location at 1.36 times the focal length, as determined by measurements of displacement and standard deviation at each position.  For a 100V volt change on the piezo-electric controller, corresponding to a 0.26nm change in wavelength, a 4.9 pixel change in location was achieved. Thus the linear dispersion at the camera input was 0.053 nm/pixel.  As before, the minimum resolution of the instrument was set by twice the standard deviation of the measurements. The minimum resolution afforded by the 38mm lens was 0.0387 pixels, corresponding to 2.05 pm.  The 19mm lens produced an even higher resolution of 0.060 nm/pixel, but at a distance of 1.2 times the focal length.  Typical results from the 19mm lens are shown in Figure 4.

A second sensor was added to the test configuration to determine whether the lens was capable of capturing both wavelengths simultaneously at the distance for optimal resolution.  The wavelengths of the two sensors were separated by 46.085 pixels when both sensors were at rest.  A one-inch diameter lens was just able to collect both wavelengths without significant diffraction of light from the edges of the lens, which would increase system noise.  No measurable increase in diffraction was recorded when 17.2 V was applied to the piezo-electric driver on either sensor. 

For the current experimental parameters, a maximum of four sensors per fiber string could be discriminated accurately.  The primary limitation on system performance is the still the grating we are currently using.  An imaging (concave) grating or one with a greater linear dispersion would improve both the resolution and the range of wavelengths that could be collected. 

In the two sensor configuration, an experiment was conducted to determine the improvement in resolution of the receiver using differential measurement.  The experiment was performed using the 38mm lens.  Results of the experiment are shown in Table 1 and Figures 5 and 6.  On average, the resolution was improved by a factor of 44.3%, from 0.0595 to 0.0362 pixels.  Based on these results, we expect that water and oil can be differentiated with the appropriate sensor design.

Current Efforts:

We are scheduled to begin resolution tests within the oil-water column in the next several weeks.  As noted above, the resolutions obtained with the 38mm and 19mm lenses indicates that we will be able to differentiate between oil and water for both sensor designs.  A diaphragm-based design is more desirable at present, since we believe it can be implemented with a minimum of moving parts and without exposure of the fiber optic cable to the external (and often hostile) environment present in a production column.  We are conferring with the project consultants on the sensor designs and practical concerns in implementing the receiver.

FIGURES

						Mirror
				GRATING
	Fiber Output

 

Figure 1:  Extended path experimental setup for spatial de-multiplexing receiver.

Figure 2:  Displacement at camera plane as a function of the lens focal length.  Displacement is measured with respect to the voltage change on the piezoelectric controller.

Figure 3:  Shift in the peak location on the camera as a function of the distance between lens and camera for the 38mm focal length lens.

Figure 4:  Peak location in pixels as a function of the voltage setting of the piezoelectric controller.

Figure 5:  Typical processed output for dual sensor measurement.  The pixel value has been adjusted for noise and for rollover in the quantized digital output.

Figure 6:  Results of differential measurement experiment.  Voltage and Position difference are sensor 2 – sensor 1.  Note that there are two sets of measurements at –8.6V – one for 0V – 8.6V, and one for 8.6V – 17.2V. 

Table 1:  Summary of resolution data for dilfferential measurement.  Measurements are quoted in terms of standard deviation.