Prospects for Developing Low-Cost Optical Fiber Sensors for Liquid-Liquid Interface Detection

Abstract

This work examines the prospects of a low-cost retroreflection optical sensor system for detection of the interface between two immiscible liquids, for applications in oil separation tanks. The sensor in its simplest configuration comprises a bare fiber optic tip, cleaved at an angle of 90º relative to the longitudinal fiber axis. The device tested experimentally is based on a 2x2 bidirectional coupler, which connects the sensor tip to a laser and two photodetectors, one used to measure the signal and the other as a reference. From experiments conducted in a laboratory wateroil system, the measured retro-reflected signal was very small, and of the same order as the differential signal due to low optical contrast liquids. To overcome this difficulty, the detection system was designed with an amplification stage, followed by filtering of the digitally converted signal. With this strategy, an accuracy of 99.7% was reached. On the light of the performance measured by the bare fiber tip configuration, an analysis on the potential improvement in performance obtained, by applying a thin layer of gold to the tip of the sensor, was made. Simulated results indicated that, for an optimized gold film, 13 nm thick, a 13fold improvement in the sensor detection limit is obtained.

Index Terms
Optical fiber; retro-reflection; separator tank; liquid level detection

I. INTRODUCTION

Liquid level detection is an important task for process monitoring in industry. Typically, capacitive [¹], [²] and electromechanical [³] sensors are widely employed. However, in some applications, the use of these sensors is impractical or, as is the case with explosive environments, require extra care and protection. Alternatively, non-contact sensors, such as microwave [⁴] or ultrasonic sensors [⁵], [⁶], may be employed, but these tend to be highly complex and costly. A special case included in this class of problems is the detection of the interface between two immiscible liquids. This problem is particularly important in separator tanks in the oil industry, in which it is necessary to identify precisely the gas-oil and the oil-water interface, so that proper separation for the distinct substances is achieved.

This work examines to what extent a simple, low-cost technique based on measuring the retroreflectance of light from a fiber optic tip can be used to identify the interface between two liquids having low optical contrast. This approach has been employed in previous work [⁷], but the problem was restricted to detecting a liquid-gas interface, thus with high optical contrast between the contacting media. In the present work the feasibility of this method for low optical contrast is evaluated and improvements in the transducer configuration are proposed. The approach has a potential application of interface detection in crude oil separation tanks, typically used in oil extraction plants. For this specific problem, as illustrated in Fig.1, the separator tank receives a mixture of three fluids with distinct densities, namely: water, oil and gas. These substances can be separated naturally, as they are immiscible with distinct densities. Due to the high pressure, which the tank normally operates, the oil-gas interface is well defined and the gas can be easily extracted by an outlet on top of the tank. However, the separation of water and oil presents a more challenging problem. Due to the same high pressure, the oil and water, normally immiscible, can partially dissolve in each other, forming an emulsion zone, causing uncertainty to the separation process. Thus, a very reliable detection system is required.

Fig. 1
Proposed sensor in the oil separator tank, and a possible embodiment for a multipoint configuration.

A variety of optical principles have been employed for liquid level detection. Although non-contact [⁸] and indirect [⁹] methods can be applied for level detection, due to the mechanical characteristics inside the tank, direct contact sensors are more adequate. In this case, optical fibers directly immersed in the liquid have been employed, such as a bent fiber [¹⁰], [¹¹]. In those reports, the sensor measures the amount of liquid in a vessel and the proposed approach does not detect an interface directly. Long period Bragg gratings have also been employed [¹²], but the approach, besides requiring a reasonably sophisticated measurement system, is mostly suitable for detecting a liquid-gas interface. Techniques such as surface plasmon resonance (SPR) [¹³] are able to discriminating low optical contrast liquids with precision, but, although SPR may seem simple, a thin metal film/prism combination as well as polarized light have to be employed, and this adds cost to the measurement principle. Also, interference measurement is a possible way to achieve liquid-liquid level detection. The interference, measured in these sensors, can be achieved, for example, by adding to the optical path Bragg fibers [¹⁴], non-core fibers [¹⁵], [¹⁶], lateral-shifted junctions [¹⁷], [¹⁸] or combinations of these elements [¹⁹]. Unfortunately, similarly to the SPR sensor, these techniques require the use of more complex equipment, in turn increasing device complexity.

It is possible to employ a very simple sensor configuration, as already investigated by Murshid [⁷], by measuring directly the retro-reflection of a cleaved fiber tip immersed in the mixture under test. However, detection of the interface between low optical contrast liquids, such as is the case in the present study, requires a detailed experimental investigation to determine to what extent the approach is feasible.

There are several sophisticated ways that could be used to improve the sensing capability of the sensor tip, including the use of multilayer stacks [²⁰], grating coupled surface plasmon resonance [²¹], to name a few. However, these add cost and complexity to the sensor head. A simple way to improve the reflectance contrast at the sensor interface is to add a single thin semi-transparent film on the fiber tip. Being semi-transparent, the reflectance on the tip can be increased by a factor that can be controlled by the film thickness. Correspondingly, the differential reflectance in going from oil to water, is expected to increase by a factor equals to that of the baseline reflectance of the coated, relative to the uncoated tip configuration. With well-established deposition techniques, a metal film, thin enough so as to become semi-transparent, can be employed. In addition, thin metal film deposition could be carried out for a very large number of fiber tips, on a single run, in turn highly reducing fabrication costs.

In the present work we characterize the capability of the retro-reflection measurement approach for detecting a water-oil interface. Expanding our own previous work on this problem [²²], we implement strategies for improving the detection sensitivity of the measurement system by employing a high input impedance amplification circuit and a digital filter, so as to avoid adding complexity to the optical circuit and to the sensor head configuration. We also include a detailed analysis of the accuracy of the approach. We then evaluate and discuss the potential benefit obtained in the sensing principle by using a modified sensor head having a single layer of gold on the fiber tip, and the prospects of using this modified configuration for reliable detection.

II. BARE FIBER TIP SENSOR

A. Preliminary Analysis of the Sensor Response

Fig.1 shows a possible configuration of a rod holding several fibers positioned vertically, suitable for multipoint detection in a separator tank. The rod would form a right angle with the water-oil interface. In a static rod configuration, a number of fiber tips, separated by the resolution distance d_res would be employed. The number of fibers and the resolution parameter d_res are defined by the dimension of the tank, width of the emulsion zone and the estimated locations of the gas-oil and oil-water interfaces. A moving rod, controlled externally, could also be employed, and in this case, the number of fibers and resolution parameter d_res could also be defined in the specific project. Because the sensitive portion of the fiber tip is of the order of 10 to 50 µm, the orientation of the fiber tip relative to the horizontal plane might be at any angle, with no practical loss in resolution, should it be necessary to avoid the formation of bubbles due to the surface tension between the liquids and the fiber.

For an individual fiber tip, having an end face cleaved at 90º relative to the fiber longitudinal axis, for normal incidence, as illustrated in Fig.2, the reflectance at the interface between the fiber and the measured fluid can be calculated by use of Fresnel’s relation [²³],

(1) $$R=\frac{{\left(n_2-n_1\right)}^2+{\left({\kappa }_2-{\kappa }_1\right)}^2}{{\left(n_2+n_1\right)}^2+{\left({\kappa }_2+{\kappa }_1\right)}^2},$$

Fig. 2
Fiber-Outer Medium Interface.

where n₁ and κ₁ are, respectively, the refractive index and extinction coefficient of the fiber core, and n₂ and κ₂ are, respectively, the corresponding coefficients for the outer medium.

Prior to carrying out experiments, a simulation study was conducted to analyze the feasibility of the sensor. A Mathcad® script was created to calculate the reflectances involved. The fiber was considered as made of silica, having negligible extinction coefficient, and the refractive index was calculated from [²⁴]. The water refractive index was calculated by use of the fitting function reported by Daimon and Masumura [²⁵]. For the oil, we considered two typical types available in the oil industry, namely, the Petrobaltic light oil and the Romashkino heavy oil. The refractive indices and the extinction coefficients of these oils were calculated by use of a fitting function on the data reported by Otremba [²⁶]. Fig.3 shows the wavelength dependency of the reflectances for water and the two types of oil as external media. Typically the reflectance is 2 × 10⁻³ for water and approximately 10⁻⁴ for both types of oil. Given the small difference in reflectance for the two oils, all the calculations reported in this paper were carried out for one type of oil only, namely the Petrobaltic light oil.

Fig. 3
Spectra of the reflectance functions of silica/water, silica/light oil and silica/heavy oil.

As shown in Fig.3, the fact that the reflectance increases by ten times, when changing the immersion medium from oil to water is an indication that the approach has potential to to be feasible. This yields a reflectance change from oil to water approximately of the same order of the water reflectance. The latter is small, however, thus making the signal susceptible to noise. Because of this, strategies to improve the signal to noise ratio, as addressed in this work, are required.

B. Experimental Apparatus

The experimental setup comprises a pigtail laser [²⁷] operating at the wavelength of 780 nm, a 2 x 2 3dB optical coupler [²⁸] and two Si photodiodes FDS-100 [²⁹], both used in the photoconductor mode. One detector is used for signal acquisition and the other serves as the reference to mitigate noise, as illustrated in Fig.4a. For calculation purposes, connector loss was estimated as approximately 0.5 dB. With this, the difference between the estimated current levels generated in the signal photodiode, due to the distinct outer media, is given by

(2) $$\Delta I=PT\Delta RS(\lambda ),$$

Fig. 4
Schematics of the sensor system setup: (a) optical circuit, (b) electronic circuit of the photodetector.

where P is the power output of the laser, T is the overall transmittance through the optical path between laser and signal detector, taking into account insertion and coupling losses, and S(λ) is the responsivity of the photodiode, at the wavelength λ. For our experimental conditions, the calculated differential current for a change of the outer medium between water and oil, at the wavelength of 780 nm is approximately 1 µA.

In order to convert the small current generated by the photodiode into a voltage signal suited to digitalization, for each photodiode, a single stage amplification circuit based on an LF356 JFET operational amplifier was implemented, according to the scheme of Fig.4b. This component has a 106 dB gain, a 100 dB common-mode rejection ratio, and input bias and offset current, respectively of 30 pA and 3pA, which are much smaller than the signal input current. The LF356 is more suited to the application than more general use transistors, such as the LM393, that has input bias and offset current a thousand times higher. In addition, each circuit has two precision trimpots, named P1 and P2 in Fig.4b, to allow adjustment of the gain on the voltage divider and amplifier separately, what gives the circuit a wider range of gain.

C. Measurements

Given the complexity of performing the experiments in a separator tank, the apparatus illustrated in Fig.5a was assembled. In this setup, differently than in a separator tank, the volume of the liquid remains static. The setup was conceived so that the probe location within the liquid phase could be controlled externally. To synchronize data acquisition and probe motion, an Arduino® microncontroller connected to a stepper motor and a computer was used. This stepper motor is connected, by a pulley, to a car, which is supported by a pivot which restricts its movement to one degree of freedom, as shown in Fig.5a. With this, a translation of 165 µm for each half step of the motor as achieved. Due to the unavailability of a suitable crude oil sample, olive oil, whose optical characteristics [³⁰], [³¹] are similar to those of light oil, was employed. The heterogeneous mixture of water and oil was placed in a graduated beaker. All the experiments were carried out at normal conditions of temperature and pressure.

Fig. 5
Experimental setup: (a) mechanical apparatus for moving the optical probe in the liquids, (b) main routine flowchart, (c) signal acquisition subroutine flowchart.

As shown in Fig.5a, the experiment consists in moving the probe through the fluid, and recording and processing the photodetector signals in each half step. As illustrated in the flowchart of Fig.5b, at each half step, the signals on both detectors are recorded, consecutively, a hundred times, according to the more detailed flowchart of Fig.5c. As shown in this flowchart, the reflection signal is divided by the reference signal to obtain a normalized signal. Then the average normalized signal, processed in the Arduino® microcontroller, is transmitted to a computer for display, storage and analysis. The probe moves a distance defined by the maximum number of half steps, as indicated in the flowchart of Fig.5b.

One problem that had to be solved was the inherent amplification of part of the background noise coming from the photodetector. To handle this problem, a low-pass digital filter was implemented. Although it was written in a MATLAB script and executed separately, the digital filter algorithm was kept as simple as possible, to allow its inclusion directly in the same embedded system, i.e. the microcontroller chip, for a more practical future implementation of the process. A moving average function was used as a basis for the digital filter algorithm.

D. Results

Fig.6 illustrates the correlation between the position of the sensor tip and the signal obtained during a sequence of experiments with several up and down cycles, where each cycle corresponds to a displacement of 29.7 mm. In the plot, the digitized signal is represented by the red points and the processed signal, after the application of the digital filter with windows of 3 and 13 points, represented, respectively, by the green and blue lines. From the plot, it is possible to notice regions with three distinct and well-defined average signal values, of 0.920, 0.750 and 0.780, when the tip was exposed, respectively, to air, oil and water. Also, it is possible to notice a spread of the scattered points of approximately 0.1 – 0.15, which is almost of the same order of the reflectance change obtained in going from air to oil and even larger than the corresponding change in going from oil to water. Thus, without digital filtering, interface identification could not be accomplished.

Fig. 6
Correlation between the average signal and the vertical position of the sensor tip during four consecutive measurement cycles, for the original and filtered signal with windows of two different sizes.

The effectiveness of the digital filter in rejecting the high frequency noise, shown in Fig.6, increases with the window size. Since the filter transfer function is non-causal, the increase in the window size implies, in addition to a greater computational effort, in a possible delay in the signal evaluation. To better evaluate the performance of the sensor, a statistical analysis was performed. To do this, the step change between the mean values of the reflectance function in going from oil to water, ΔR was determined. We defined half of this step, i.e.,

(3) $$x\equiv \frac{\Delta R}{2},$$

as the minimum reflectance change for identifying a probed medium transition, i.e., from oil to water. For each window size, we calculated the standard deviation σ. Assuming a gaussian distribution for the noise spread around the mean value, we defined an accuracy function acc(x), corresponding to the probability of detecting the expected fluid. This can be calculated by use of

(4) $$acc(x)=erf\left(\frac{x}{\sqrt{2}\sigma }\right),$$

where erf is the error function [³²].

Fig.7 shows the correlation between accuracy and window size, where the window size of 1 sample represents the original signal, right after the digitalization. It is possible to verify that the accuracy stabilizes after a window size corresponding to 13 half-steps, with a value of approximately 99.7%. Since this sensing principle may be configured in a multipoint configuration, allowing a more comprehensive analysis of the measured signals, the accuracy found is acceptable, encouraging future experiments in environmental conditions closer to those expected in a separator tank.

Fig. 7
Correlation between the accuracy and the window size of the digital filter applied to the bare fiber sensor.

However reasonable might be the performance of the transducer in the bare fiber configuration, there is room for improvement, without compromising the inherent cost-benefit of the system. One simple modification is the application of a single metal thin film on the cleaved fiber surface. In the following section, we analyze the benefits of using a thin gold film as the coating material.

III. GOLD COATED FIBER TIP SENSOR

As evidenced in Fig.3, a critical disadvantage of bare fiber tip sensors is the small magnitude of the reflectances at the interface between the glass on the fiber tip and the outer liquid medium. Consequently, the low power of the reflected signal to be measured by the detector makes the sensor susceptible to noise, in turn directly affecting the sensor detection capability. To address this issue, we evaluate and discuss the potential benefit of increasing the retro-reflectance level, by use of a single layer of gold on the fiber tip.

Fig.8 shows the modified configuration of the fiber tip, with the inclusion of a metal film having thickness d. The reflectance function, with the presence of the film becomes

(5) $$R={\left|\frac{r_1+r_2\exp (-j2kd)}{1+r_1{r}_2\exp (-j2kd)}\right|}^2,$$

with

(6) $$r_1=\frac{\left(n_m-n_1\right)-j\left({\kappa }_m-{\kappa }_1\right)}{\left(n_m+n_1\right)-j\left({\kappa }_m+{\kappa }_1\right)},$$

(7) $$r_2=\frac{\left(n_2-n_m\right)-j\left({\kappa }_2-{\kappa }_m\right)}{\left(n_2+n_m\right)-j\left({\kappa }_2+{\kappa }_m\right)}$$

and

(8) $$k=\frac{2\pi }{\lambda }\left(n_m-j{\kappa }_m\right),$$

Fig. 8
Fiber-Gold-Outer Medium multilayer interface.

with j representing the purely imaginary number, n_m and κ_m, the refractive index and extinction coefficient of the metal film, respectively, and λ, the wavelength.

The web app SPRinG [³³] is an online simulator that can be used to calculate and optimize multilayer grating structures, including flat interfaces, and was used in the present work for the calculation of (5). The optical parameters n_m and κ_m were extracted from Weaver [³⁴]. The other optical parameters are the same defined in Section II-A. To find the optimum gold film thickness d, for which the differential reflectance of oil relative to water is maximized, the reflectance functions were simulated in the range 0 ≤ d ≤ 19 nm. The plot in Fig.9 shows the calculated reflectance function versus gold film thickness for water and oil as the outer medium, and the corresponding difference function. The difference function has a maximum for a gold film thickness of 13 nm. It is important to point out the improvement in adding the single gold film, as anticipated previously. Whereas the difference is about 2 × 10⁻³ for d = 0, it reaches a value approximately 2.2 × 10⁻² at the optimum thickness, i.e., approximately a 10-fold improvement.

Fig. 9
Gold film thickness dependence of the reflectance functions for water and oil and corresponding difference curve.

In order to compare the degree of improvement in detecting the interface between water and oil after replacing the bare fiber with the gold coated tip, a simulation was carried out according to the following procedure. We used the 13-point average curve, represented in Fig.6 to determine a conversion factor of reflectance units to experimental signal units. From the plots of Figs. 3 and 6 one can infer the conversion factor between differential reflectance units and experimental signal units. Then, by using the approximately 10-fold increase in reflectance change obtained with the coated fiber relative to the bare fiber, we can infer how much change this would correspond in experimental signal units. Given that the detector noise level is independent of the signal, it can be added to the scaled simulated signal. These simulated noisy and average signals corresponding to the metallized fiber, give an idea on how the signal would look like, should the metallized fiber be used in the measurements.

Fig.10 shows the bare fiber measurement with the 13-point average signal plotted together with the simulation corresponding to the signal that would be obtained by use of the metal coated fiber, with the corresponding 13-point average. For the sake of comparison, the simulated signal has an arbitrary offset to allow representation together with the original measured signal. For the simulated case, the base level corresponding to direct contact with air is not shown as the value is too high and off the chart to be represented in the same plot. The plots in Fig.10 show that the metal coated fiber tip would be able to detect the interface water-oil, even without use of the moving average procedure, as the signal change is at least 5 times larger than the noise level.

Fig. 10
Correlation between the average signal and the vertical position of the sensor tip during four consecutive measurement cycles, with and without application of a digital filter, for bare and gold coated configurations.

To better evaluate the capability of the gold coated tip on discriminating distinct immiscible liquids, a statistical analysis, similar to that presented in Section II-D, was performed. To allow determining the accuracy relative to a given change in refractive index, the reflectance change was mapped in terms of the refractive index change relative to a reference value equals to that of water, i.e.,

(9) $$\Delta n\equiv n-n_{water}.$$

Then, the accuracy function given by (4) was calculated as a function of Δn.

Fig.11 shows the correlation between the accuracy and the refractive index difference, before and after the application of the digital filter. Considering an acceptable accuracy of 99,7%, without the digital filter, the proposed gold tip sensor achieved an accurate discrimination between liquids for a change larger than 0.1 refractive index units. Taking into account the refractive indices of water and oil of typically 1.33 and 1.46, respectively, i.e., having a difference of Δn = 0.13, the metal coated configuration without filtering exhibits a similar performance to the bare tip sensor with the digital filter, as already anticipated qualitatively. With the application of the digital filter, the metal coated sensor achieves a detection limit of 0.01 refractive index units, i.e., a 13-fold enhancement in sensor detection capability relative to the the bare tip sensor.

Fig. 11
Dependence of the accuracy parameter on the difference between external refractive indices for the gold coated fiber sensor, with and without application of a digital filter.

IV. CONCLUSION

Detection of the interface between water and oil, an important problem in the oil industry, using a bare fiber tip, is a challenging task. This is so because of the small level of the retro-reflected signal from a glass-liquid interface, and a contrast as small as the signal for the step change in reflectance in going from one medium to the other A detailed experimental study and statistical analysis was carried out to determine to what extent a bare fiber tip, used as a retro-reflectance sensor would be feasible for detecting the liquid-liquid interface. The results obtained showed the feasibility of the proposed sensor by use of a detection system employing a specialized amplification stage based on the JFET LM356 operational amplifier, which has high input impedance and high common noise rejection ratio. In order to minimize the effect of background noise, a low pass filter was employed, followed by digitalization and application of a moving average digital filter. Although the moving average digital filter is characterized by a non-causal function, which can delay the evaluation of the measured signal, the results show that the filter stabilizes with a window of 13 samples, making its execution in the same embedded system used in the digitalization of the signal feasible. With the implementation of the digital filter, the sensor accuracy was improved from 64.5% to 99.7%. Although, for single point detection, an accuracy of 99.7% might be insufficient to meet critical system requirements, as could be the case in oil separator tanks, a multipoint sensor configuration would allow further reducing uncertainties, given that multiple signals are analyzed and thus an accuracy of 99.7% would be acceptable.

In this study we also investigated the benefits of adding a thin gold film to the sensor tip. We determined an optimum thickness of 13 nm for gold, for a wavelength of 780 nm. Considering an accuracy of 99.7% and the same standard deviation of the signal measured at the output of the amplifier, simulation results indicated that the sensor would reach, without the use of a digital filter, a refractive index detection threshold of 0.1, a value smaller than the refractive index change Δn = 0.13 from water to oil. Simulation results also indicated that, with digital filtering corresponding to a 13-point average, the gold coated sensor can reach a detection threshold of 0.01, that is, a 13-fold increase relative to the performance of a bare tip sensor. This detection limit, in addition to providing greater reliability, makes it feasible to use the gold coated sensor as a reliable solution for applications in liquid interface detection in the oil industry or as a probe to discriminate liquids with closer values of refractive indices, such as salt solutions, acetone, alcohol, and other organic solvents, which are quite common in the biochemical industry.

V. ACKNOWLEDGMENTS

This work was supported by the following brazilian funding agencies: CNPq – Conselho Nacional de Desenvolvimento Científico e Tecnológico (Grant: 31162520193), FACEPE – Fundação de Amparo à Ciência e Tecnologia do Estado de Pernambuco (Grant: IBPG-1255-3.04/16) and CAPES – Coordenação de Aperfeiçoamento de Pessoal de Ensino Superior (CAPES-PRINT grants: 88887.311936/2018-00; 88887.568744/2020-00 and 88887.371102/2019-00)

REFERENCES

• ^[1]
  W. Q. Yang, M. R. Brant, and M. S. Beck, “A multi-interface level measurement system using a segmented capacitance sensor for oil separators,” Measurement Science and Technology, vol. 5, no. 9, pp. 1177–1180, 1999.
• ^[2]
  J. Vergouw, C. O. Gomez, and J. A. Finch, “Estimating true level in a thickener using a conductivity probe,” Minerals Engineering, vol. 17, no. 1, pp. 87–88, 2004.
• ^[3]
  K. Khalid, I. V. Grozescu, L. K. Tiong, L. T. Sim, and R. Mohd, “Water detection in fuel tanks using the microwave reflection technique,” Measurement Science and Technology, vol. 14, no. 11, p. 1905, 2003. [Online]. Available: http://stacks.iop.org/0957-0233/14/i=11/a=008
  » http://stacks.iop.org/0957-0233/14/i=11/a=008
• ^[4]
  M. H. Zarifi, M. Rahimi, M. Daneshmand, and T. Thundat, “Microwave ring resonator-based non-contact interface sensor for oil sands applications,” Sensors Actuators, B Chem., vol. 224, pp. 632–639, 2015.
• ^[5]
  P. Hauptmann, R. Lucklum, A. Puttmer, B. Henning, A. Piittmer, and B. Henning, “Ultrasonic sensors for process monitoring and chemical analysis: state-of-the-art and trends,” Sensors Actuators, A Phys., vol. 67, no. 1 -3 pt 1, pp. 32–48, 1998.
• ^[6]
  V. E. Sakharov, S. A. Kuznetsov, B. D. Zaitsev, I. E. Kuznetsova, and S. G. Joshi, “Liquid level sensor using ultrasonic Lamb waves,” Ultrasonics, vol. 41, no. 4, pp. 319–322, 2003.
• ^[7]
  S. Murshid, “Universal liquid level sensor employing fresnel coefficient based discrete fiber optic measurement technique,” Proceedings of SPIE - The International Society for Optical Engineering, vol. 9202, pp. 1–9, 2014.
• ^[8]
  J. Y. Lee and S. K. Tsai, “Measurement of refractive index variation of liquids by surface plasmon resonance and wavelength-modulated heterodyne interferometry,” Opt. Commun., vol. 284, no. 4, pp. 925–929, 2011. [Online]. Available: http://dx.doi.org/10.1016/j.optcom.2010.10.060
  » http://dx.doi.org/10.1016/j.optcom.2010.10.060
• ^[9]
  Samian, A. H. Zaidan, M. Yasin, Pujiyanto, and Supadi, “Liquid level sensor using fiber bundle,” Measurement: Journal of the International Measurement Confederation, vol. 129, no. July, pp. 542–547, 2018. [Online]. Available: https://doi.org/10.1016/j.measurement.2018.07.038https://doi.org/10.1016/j.sna.2018.08.032
  » https://doi.org/10.1016/j.measurement.2018.07.038» https://doi.org/10.1016/j.sna.2018.08.032
• ^[10]
  C. Zhao, L. Ye, J. Ge, J. Zou, and X. Yu, “Novel light-leaking optical fiber liquid-level sensor for aircraft fuel gauging,” Optical Engineering, vol. 52, no. 1, p. 014402, 2013. [Online]. Available: http://opticalengineering.spiedigitallibrary.org/article.aspx?doi=10.1117/1.OE.52.1.014402
  » http://opticalengineering.spiedigitallibrary.org/article.aspx?doi=10.1117/1.OE.52.1.014402
• ^[11]
  Y. Zhang, Y. Hou, Y. Zhang, and Y. Hu, “Continuous liquid level detection based on two parallel plastic optical fibers in a helical structure,” Optical Engineering, vol. 57, no. 02, p. 1, feb 2018. [Online]. Available: https://www.spiedigitallibrary.org/journals/optical-engineering/volume-57/issue-02/026112/Continuous-liquid-level-detection-based-on-two-parallel-plastic-optical/10.1117/1.OE.57.2.026112.full
  » https://www.spiedigitallibrary.org/journals/optical-engineering/volume-57/issue-02/026112/Continuous-liquid-level-detection-based-on-two-parallel-plastic-optical/10.1117/1.OE.57.2.026112.full
• ^[12]
  A. L. Ricchiuti, D. Barrera, K. Nonaka, and S. Sales, “Fiber optic liquid-level sensor using a long fiber Bragg grating,” Fifth European Workshop on Optical Fibre Sensors, no. May 2013, p. 87941J, 2013. [Online]. Available: http://proceedings.spiedigitallibrary.org/proceeding.aspx?articleid=1690545
  » http://proceedings.spiedigitallibrary.org/proceeding.aspx?articleid=1690545
• ^[13]
  M. D. Ooms, H. Fadaei, and D. Sinton, “Surface Plasmon Resonance for Crude Oil Characterization,” Energy & Fuels, vol. 29, no. 5, pp. 3019–3023, may 2015. [Online]. Available: http://pubs.acs.org/doi/abs/10.1021/acs.energyfuels.5b00340
  » http://pubs.acs.org/doi/abs/10.1021/acs.energyfuels.5b00340
• ^[14]
  H. Xue, Z. Xu, H. Chen, Y. Yang, J. You, J. Yan, H. Fu, and D. Zhang, “Continuous liquid level sensor based on a reflective long period fiber grating interferometer,” Measurement Science and Technology, vol. 26, p. 037001, 2015. [Online]. Available: http://stacks.iop.org/0957-0233/26/i=3/a=037001?key=crossref.4df31898f6cfd4a29b7368382d737c5dhttp://dx.doi.org/10.1088/0957-0233/26/3/037001
  » http://stacks.iop.org/0957-0233/26/i=3/a=037001?key=crossref.4df31898f6cfd4a29b7368382d737c5d» http://dx.doi.org/10.1088/0957-0233/26/3/037001
• ^[15]
  C. Li, T. Ning, C. Zhang, J. Li, X. Wen, L. Pei, X. Gao, and H. Lin, “Liquid level measurement based on a no-core fiber with temperature compensation using a fiber Bragg grating,” Sensors and Actuators, A: Physical, vol. 245, pp. 49–53, 2016. [Online]. Available: http://dx.doi.org/10.1016/j.sna.2016.04.046
  » http://dx.doi.org/10.1016/j.sna.2016.04.046
• ^[16]
  A. Saha, A. Datta, and S. Kaman, “Ultrahigh-sensitive multimode interference-based fiber optic liquid-level sensor realized using illuminating zero-order Bessel–Gauss beam,” Optical Engineering, vol. 57, no. 03, p. 1, 2018. [Online]. Available: https://www.spiedigitallibrary.org/journals/optical-engineering/volume-57/issue-03/036118/Ultrahigh-sensitive-multimode-interference-based-fiber-optic-liquid-level-sensor/10.1117/1.OE.57.3.036118.full
  » https://www.spiedigitallibrary.org/journals/optical-engineering/volume-57/issue-03/036118/Ultrahigh-sensitive-multimode-interference-based-fiber-optic-liquid-level-sensor/10.1117/1.OE.57.3.036118.full
• ^[17]
  C. Li, T. Ning, C. Zhang, X. Wen, J. Li, and C. Zhang, “Liquid level and temperature sensor based on an asymmetrical fiber Mach-Zehnder interferometer combined with a fiber Bragg grating,” Optics Communications, vol. 372, pp. 196–200, 2016. [Online]. Available: http://dx.doi.org/10.1016/j.optcom.2016.04.025
  » http://dx.doi.org/10.1016/j.optcom.2016.04.025
• ^[18]
  P. Li, H. Yan, and H. Zhang, “Highly sensitive liquid level sensor based on an optical fiber Michelson interferometer with core-offset structure,” Optik, vol. 171, no. June, pp. 781–785, 2018. [Online]. Available: https://doi.org/10.1016/j.ijleo.2018.06.126
  » https://doi.org/10.1016/j.ijleo.2018.06.126
• ^[19]
  Z. Wang, Z. Tan, R. Xing, L. Liang, Y. Qi, and S. Jian, “Liquid level sensor based on fiber ring laser with single-mode-offset coreless-single-mode fiber structure,” Optics and Laser Technology, vol. 84, pp. 59–63, 2016. [Online]. Available: http://dx.doi.org/10.1016/j.optlastec.2016.05.001
  » http://dx.doi.org/10.1016/j.optlastec.2016.05.001
• ^[20]
  R. F. Góis, G. O. Cavalcanti, E. F. de Melo, and E. Fontana, “Highly sensitive retro-reflectance fiber-optic sensors for liquid-liquid interface detection,” in 2019 SBMO/IEEE MTT-S International Microwave and Optoelectronics Conference (IMOC), pp. 1–3, 2019.
• ^[21]
  E. F. de Melo and E. Fontana, “Design of Surface Plasmon Resonance Sensors Having Maximum Response at Normal Incidence,” in Frontiers in Optics 2012/Laser Science XXVIII, p. FTu3A.18, 2012. [Online]. Available: https://www.osapublishing.org/abstract.cfm?URI=FiO-2012-FTu3A.18
  » https://www.osapublishing.org/abstract.cfm?URI=FiO-2012-FTu3A.18
• ^[22]
  R. F. Gois, “Sistema óptico de detecção de nível na mistura água/óleo,” MSc Dissertation in Systems Engineering, Programa de Pós-Graduação em Engenharia de Sistemas - Universidade de Pernambuco, 2017.
• ^[23]
  M. Born and E. Wolf, Principles of optics : electromagnetic theory of propagation, interference, and diffraction of light Cambridge: Cambridge University Press, 2019.
• ^[24]
  I. H. Malitson, “Interspecimen Comparison of the Refractive Index of Fused Silica,” J. Opt. Soc. Am., vol. 55, no. 10, p. 1205, oct 1965. [Online]. Available: https://www.osapublishing.org/abstract.cfm?URI=josa-55-10-1205
  » https://www.osapublishing.org/abstract.cfm?URI=josa-55-10-1205
• ^[25]
  M. Daimon and A. Masumura, “Measurement of the refractive index of distilled water from the near-infrared region to the ultraviolet region,” Appl. Opt., vol. 46, no. 18, p. 3811, jun 2007. [Online]. Available: https://www.osapublishing.org/abstract.cfm?URI=ao-46-18-3811
  » https://www.osapublishing.org/abstract.cfm?URI=ao-46-18-3811
• ^[26]
  Z. Otremba, “Oil droplets as light absorbents in seawater,” Œ, vol. 15, no. 14, pp. 8592–8597, 2007. [Online]. Available: http://www.opticsexpress.org/abstract.cfm?URI=oe-15-14-8592
  » http://www.opticsexpress.org/abstract.cfm?URI=oe-15-14-8592
• ^[27]
  Thorlabs, “Pigtailed Laser Diode, SMF LPS-785-FC,” 2013. [Online]. Available: https://www.thorlabs.com/
  » https://www.thorlabs.com/
• ^[28]
  Thorlabs, “FC780-50B-FC,” 2011. [Online]. Available: https://www.thorlabs.com/
  » https://www.thorlabs.com/
• ^[29]
  Thorlabs, “Si Photodiode 350-1100 nm FDS100,” 2013. [Online]. Available: https://www.thorlabs.com/
  » https://www.thorlabs.com/
• ^[30]
  J. E. de Albuquerque, B. C. L. Santiago, J. C. C. C. Campos, A. M. Reis, J. S. R. Coimbra, C. L. da Silva, and J. P. Martins, “Photoacoustic Spectroscopy as an Approach to Assess Chemical Modifications in Edible Oils,” vol. 24, no. 3, pp. 369–374, 2013.
• ^[31]
  I. Bodurov, I. Vlaeva, M. Marudova, T. Yovcheva, and K. Nikolova, “Detection of adulteration in olive oils using optical and thermal methods,” Bulgarian Chemical Communications, vol. 45, pp. 81–85, 2013.
• ^[32]
  R. Larson and B. Farber, Elementary Statistics: Picturing the World, Books a la Carte Edition Pearson Education Canada, 2011. [Online]. Available: https://books.google.com.br/books?id=i5ikcQAACAAJ
  » https://books.google.com.br/books?id=i5ikcQAACAAJ
• ^[33]
  E. F. de Melo and E. Fontana, “SPRinG - Surface Plasmon Resonance in Gratings,” Mar 2013. [Online]. Available: http://www.ufpe.br/fontana/spring
  » http://www.ufpe.br/fontana/spring
• ^[34]
  J. H. Weaver, C. Krafka, D. W. Lynch, and E. E. Koch, “Optical properties of metals,” Appl. Opt., vol. 20, no. 7, pp. 1124_1—-1125, 1981. [Online]. Available: http://ao.osa.org/abstract.cfm?URI=ao-20-7-1124{_}1
  » http://ao.osa.org/abstract.cfm?URI=ao-20-7-1124{_}1

Publication Dates

• Publication in this collection
  23 Sept 2022
• Date of issue
  Sept 2022

History

• Received
  30 Mar 2022
• Reviewed
  07 Apr 2022
• Accepted
  02 June 2022