The presence of salt (expressed as chloride) in crude oil depends on the origin and of the wells production. If not removed, they can react to form a corrosive hydrochloric acid during the refining process, leading to operational problems as corrosion, incrustation and also in the deactivation of catalysts. Therefore, many analytical techniques or industry standard methods are routinely used to determine chloride in crude oil and fuels. Regardless of the analytical method, some aspects can be considered, such as: cost, time and analytical frequency. In this context, the colorimetric analysis using PhotoMetrix app can be an approach easily performed with low-cost, portable instrumentation and high analytical frequency. The present work proposes the extraction of chloride from the fuel to an aqueous phase, followed by colorimetric quantification of silver chloride using digital images. For that, parameters for extraction (the sample mass, the time, the temperature, the type and use of solvent, and the use of demulsifier and its concentration,) and for detection (the ratio between silver nitrate and indicator, lighting, the volume inside the vial, the focal distance, the number of pixels and the effect of pH) were evaluated. To obtain the digital images, an USB camera was placed in an open source 3D-printed chamber illuminated by a white light-emitting diode (LED) with light intensity controlled (Fig. 1). Sample and reagents (silver nitrate and the indicator) were added to an Eppendorf vessel, which was placed inside the chamber. Thus, the digital images were captured and converted into red, green, and blue (RGB) histograms, and partial least squares regression (PLS) models were used within the app. The PLS regression results (obtained directly from the app) were evaluated in terms of the number of samples and factors, the coefficient of determination (R2), the root mean squared error of calibration (RMSEC), the root mean squared error of cross validation (RMSECV), and the root mean squared error of prediction (RMSEP). The calibration curve (from 100 to 500 mg L-1 of Cl) was obtained using a PLS model with a R2 higher than 0.99. No significant differences (p < 0.05) between the measured and predicted values were identified. The RMSEC, RMSECV and RMSEP of PLS model were of 9.04, 31.2 and 26.1 mg L-1, respectively.

Figure 1 - Source 3D-printed chamber equipped with USB camera and, in detail, sample introduction.

The presence of salt (expressed as chloride) in crude oil depends on the origin and of the wells production. If not removed, they can react to form a corrosive hydrochloric acid during the refining process, leading to operational problems as corrosion, incrustation and also in the deactivation of catalysts. Therefore, many analytical techniques or industry standard methods are routinely used to determine chloride in crude oil and fuels. Regardless of the analytical method, some aspects can be considered, such as: cost, time and analytical frequency. In this context, the colorimetric analysis using PhotoMetrix app can be an approach easily performed with low-cost, portable instrumentation and high analytical frequency. The present work proposes the extraction of chloride from the fuel to an aqueous phase, followed by colorimetric quantification of silver chloride using digital images. For that, parameters for extraction (the sample mass, the time, the temperature, the type and use of solvent, and the use of demulsifier and its concentration,) and for detection (the ratio between silver nitrate and indicator, lighting, the volume inside the vial, the focal distance, the number of pixels and the effect of pH) were evaluated. To obtain the digital images, an USB camera was placed in an open source 3D-printed chamber illuminated by a white light-emitting diode (LED) with light intensity controlled (Fig. 1). Sample and reagents (silver nitrate and the indicator) were added to an Eppendorf vessel, which was placed inside the chamber. Thus, the digital images were captured and converted into red, green, and blue (RGB) histograms, and partial least squares regression (PLS) models were used within the app. The PLS regression results (obtained directly from the app) were evaluated in terms of the number of samples and factors, the coefficient of determination (R2), the root mean squared error of calibration (RMSEC), the root mean squared error of cross validation (RMSECV), and the root mean squared error of prediction (RMSEP). The calibration curve (from 100 to 500 mg L-1 of Cl) was obtained using a PLS model with a R2 higher than 0.99. No significant differences (p < 0.05) between the measured and predicted values were identified. The RMSEC, RMSECV and RMSEP of PLS model were of 9.04, 31.2 and 26.1 mg L-1, respectively.

Figure 1 - Source 3D-printed chamber equipped with USB camera and, in detail, sample introduction.