Short overview of the case

F3

In this case study, two aromatic substances are coupled by a lithiation reaction which is a prominent example in pharmaceutical industry. The two aromatic reactants (A1, A2) are mixed with a Lithium-base (Li-Base) in a helical reactor so that the desired product (B) and a salt (Li-Salt) are produced:

A1 + A2 + Li-Base → B + Li-Salt

The salt precipitates which leads to the formation of particles. Unfortunately, a fouling layer is formed on the inner surface of the reactor due to the precipitation. After some time, this grows and leads to clogging of the reactor hence the production has to be interrupted for cleaning the reactor. Moreover, the reactivity and temperature of the feed streams are subject to variation.

 The plant that is used in the CONSENS project has been built in the F³ Factory project (see photo).

Which technologies will be validated

  • Novel inline fouling sensor that measures non-invasive and space-resolved the thickness and the growth of the fouling layer along the reactor
  • Novel online sensor that measures fast and accurately the composition of the product stream (product, residues of reactants, side products) with minimum calibration effort
  • Combined use of the new online composition sensor and NIR
  • Novel self-adapting control method
  • Tool for sensor failure detection
  • Tool for an integrated approach for data mining and performance monitoring
  • Design evaluation of the overall system before realization

What is the added value to switch from batch to the continuous process

In contrast to the traditional batch process which is run at -78°C for safety and selectivity reasons, we use a continuous process at room temperature. As the reaction is exothermic, the continuous process saves a lot of cooling energy compared to the batch process.

Expected achievements of the case

In comparison with the existing experiences on the continuous process from the F³ Factory project we want to achieve the following:

  • Understand the fouling mechanism
  • Reduce fouling tendency
  • Identify the right moment for cleaning just before clogging
  • Reduce waste generated by cleaning of the reactor
  • Reduce the required amount of solvent
  • Reduce the required amount of Lithium-base
  • Increase capacity
  • Increase the robustness of process control to keep always the desired quality level

First year achievements

1.1Figure1 Modular production plant

Case Study 1 is focused on the continuous synthesis of organic compounds. The main goal is to guarantee high and constant quality levels of the product while also reducing fouling and the consumption of raw materials and energy. This will be achieved by the use of novel measuring, control and monitoring tools. A modular production plant at pilot scale as shown in figure 1 will be used for the validation of the technologies.

1.2Figure2 ATEX enclosure for NMR sensor

Significant progress has been achieved within the first year of the project on the basis of an active collaboration between the involved project partners. Right from the project beginning the legal and technical requirements for a safety-compliant implementation of the novel technologies within an ATEX area have been considered. In terms of the online NMR (Nuclear Magnetic Resonance) sensor it is envisioned to place the spectrometer as well as its associated electronics and peripheral equipment into an explosion proof enclosure as it is shown in figure 2.

With regards to the ultrasonic detection of fouling, a preliminary lab-scale sensor has been developed. A first indication of the feasibility of the method has been achieved using a model fouling layer.

Based on the data of the measuring equipment, closed-loop control and data monitoring will be used to operate the plant in a more efficient and more reliable manner. Therefore the integration of the technologies into the control system of the chemical plant has been coordinated. In order to evaluate the performance of the integrated control systems, a performance monitoring algorithm has been prepared. It is based on the comparison of the actual plant performance with a best demonstrated practice in the plant which will demonstrate the potential for future optimization steps in the plant.

The Sensor Validation System works well. In a first test in the Hydrogenium Plant “Heracles” in Rozenburg/NL (Air Liquide) the system was installed successfully. With 33 segments, 10 models and 320 sensors the system shows an average error of 2% (figure 3). The critical point is the slow training speed of the model. The actual challenge is to speed up the training of the model by converting the source code into a massive parallel architecture in order to use multi-cores and shaders of graphics adapters for parallel computing.

In order to gain a deeper understanding of the chemistry various lab scale experiments were carried out. Recent work using NMR technology proofed that online acquisition of quasi-simultaneous 1H and 19F spectra combined with automatic data pretreatment and evaluation methods is feasible. The spectra in the proton domain along the reaction path are shown in figure 4.

1.3Figure3 Relative error of the sensor validation

1.4Figure4 Online proton spectra during the progression of the reaction

1.5Figure5 Result of sensitivity analysis

A 2-dimensional model for the process in Case Study 1 was set up. A sensitivity analysis is carried out to understand the effects of changes in inputs to the outputs. To develop a PAT based control scheme the effects of inputs are analyzed. Additionally the optimal positions of the sensors to gain better parameter estimates can be determined. The model based analysis acts as the first step to develop a PAT based adaptive control and online parameter and state estimation.

All Case Study members are looking forward to validate their technologies at the INVITE facility in 2017.