Controlling production online is an important issue for chemical companies. Visible and near-infrared (NIR) spectroscopy offers a number of important advantages for process monitoring, and has been used since the 1980s. For complex media such as silica precipitation samples, it is interesting to be able to study independently the scattering and absorption effects. From the scattering coefficient it is possible to extract information on the physical structure of the medium. In this work, the physical changes were monitored during a silica precipitation reaction by simple measurement of collimated transmittance NIR spectra. It is shown that it is possible to differentiate samples before and after the gel point, which is a key parameter for monitoring the process. From these NIR spectra the scattering coefficients were simply extracted, allowing a global vision of the physical changes in the medium. Then principal component analysis of the spectra allowed refinement of the understanding of the scattering effects, in combination with particle size monitoring.

Hydrocracking is a crucial refinery process that transforms heavy molecules (i.e., vacuum gas oil) into lighter and highly valued products such as naphtha, jet fuel and diesel fuel. In order to improve margins, the refiner must tackle the following points: (1) select the best catalyst stacking or combination [hydrotreatment catalyst(s) + hydrocracking catalyst(s)] and (2) optimization process: define the optimal operating conditions. The aim of this paper is to illustrate the global methodology proposed by IFPEN to answer to these points with the focus on experimental points. It is illustrated for the revamping of a fluid catalytic cracking feedstock hydrotreater (CFHDT) into a mild hydrocracker. It is divided into several steps. The first step is a virtual screening for possible catalyst stacking and operating conditions. It is based on hydrodenitrogenation (HDN), hydrodesulfurization (HDS) and hydrocracking (HDC or HCK) kinetic modeling, and product property correlations. The second step is to define the design of experiments in order to check the results provided by the kinetic models. The third step is to check the virtual stacking using pilot plant tests. This can be carried out by using a high throughput testing (HTT) experiment pilot plant (reactor volume around 0.5−1 cm 3 ). The fourth step is to test the prescreened catalyst stacking in a classical pilot plant (reactor volume around 50−200 cm 3 ) because they are more robust to refractory feedstock, and it can produce additional product volume for further analysis. The fifth step is to update the kinetic models using the new experimental points. This paper shows also that: (1) it is possible to process refractory feedstock in the HTT experiment pilot plant; (2) catalyst ranking is similar in HTT and classical pilot plants; and (3) in mild hydrocracking, performances (HDS, HDN, and HDC) are similar between HTT and classical pilot plants. However, the classical pilot plants are still useful for process study (optimize each temperature bed, estimate product properties for specific cuts, etc).

Near infrared spectroscopy offers a number of important advantages for process monitoring. In addition to its numerous practical advantages, an important reason to use near infrared spectroscopy for process monitoring is its ability to supply versatile and multivariate information. However, in heterogeneous samples the interaction of light is complex and includes transmission, absorption, and scattering simultaneously which all affect spectra. The measurement of the signal at one point may be insufficient. A solution is to measure the medium at several points and to use specific multivariate analysis. In our study we propose to associate multipoint measurements with a common components and specific weight analysis. We monitored two media online by angular multipoint near infrared spectroscopy. For the first medium, in which only the scattering varies over time, the precipitation of silica was chosen to illustrate such a medium. For the second medium, both scattering and absorption vary, whereby microemulsions implemented for enhanced oil recovery illustrate this medium. The results showed, by combining multiangle measurements to common components and specific weight analysis, the interest of measuring at different angles. In the first case, two scattering regimes have been identified and it was possible to access the anisotropy coefficient during the silica precipitation reaction. In the second case study, on microemulsions, it was possible to identify the different phases and to separate the phenomena related to absorption and those related to diffusion. These encouraging results validate the interest of coupling multiangle measurements with multivariate multiblock analysis tools.