On-line for business
7 Jun 2004
Speciality chemicals is a tough business, where competition is intense. According to Brian Murphy, managing director of speciality firm Robinson Brothers and leader of the Chemistry Leadership Council's Business Intensification project, 'to make money in speciality organic chemicals in the 21st century, we need an unfair advantage.'
One of the key elements of the 'unfair advantage' is on-line instrumentation and control. SONAR, a partnership between SOCSA (the Specialised Organic Chemicals Sector Association) and CPACT (the Centre for Process Analytics and Control Technologies), is an inter-disciplinary network which promotes the uptake of process analytics and control technology in the specialised organic chemicals sector.
It has recently completed a DTI-funded demonstration project at three industrial speciality chemicals sites where on-line analysers were installed on a caustic scrubber (AH Marks), a caustic abatement system (Lambson Group) and a fluidised bed dryer (Robinson Brothers).
The main aim was to stimulate the increased uptake of on-line technology by showing the cost-benefit of such analysers operating over an extended period and to provide an example to those wishing to replicate such technologies.
The project partners included the three user companies, the analyser suppliers, Clairet Scientific (NIR / MIR technology) and Applikon Analytical (autotitrator technology), and the University of Newcastle under the leadership of SOCSA. They were supported by a team of experts: Anderson Barr Consulting (cost benefit analysis), Morris Partnership (sampling systems engineering) and the Oxis Partnership (project management).
Applikon Analytical became involved in the SONAR Demonstration project with two main objectives: to demonstrate how reliable and robust on-line titration is for process and environmental control and to take advantage of the independent cost/benefit analysis to confirm its own data, showing that control of even relatively small scrubbers using on-line titration pays for itself in a very short time.
Titration is ideal for on-line application. As an absolute technique it requires no calibration and has a very low level of inaccuracy, typically less than 1%. Titration will tolerate high levels of solids and is extremely specific - both features that were shown to be important during the project.
Perhaps titration's greatest benefit is its simplicity. It often replicates the existing laboratory method, so operators find the technique simple to adopt.
Applikon analysers operate on a batch principle. The first stage of the analytical cycle is to acquire a fresh, representative sample. For titration a highly accurate sample volume is required and there are a number of modules available to cover all sample sizes and characteristics.
The second stage is to carry out the analysis. For the demonstration project, AH Marks used the dynamic endpoint titration method to give rapid and extremely accurate analysis. Although for A H Marks the caustic concentration was the only parameter of interest, at Lambson's, the carbonate and bicarbonate levels were also determined, as these also provide valuable scrubbing capacity.
The final stage of analysis is to clean the analytical parts used by rinsing with clean water. This cleaning stage increases the reliability enormously and reduces operator involvement to less than thirty minutes a week.
The demonstration site at A H Marks was a fairly standard approach starting from a high level of caustic concentration and depleting to a set point.
The analyser provided for the project was programmed to perform analysis at hourly frequency until the caustic concentration fell below 6%. Below this concentration, the analysis frequency was increased to every fifteen minutes to give greater amounts of data in the 'interesting' part of the cycle. The analyser can operate at a higher frequency but some pragmatism was involved in the conservation of reagents.
The Lambson's demonstration was a shorter project. Here the scrubber operated on a 'top-up' system controlled by two pH analysers mounted in the scrubber. Additionally, the levels of caustic were far lower but very high levels of carbonate and bicarbonate were present. Although the time on site was short, the analysis showed clearly that there is no correlation between pH and caustic concentration in the scrubber and that even when no caustic was present the carbonates/bicarbonates provided sufficient scrubbing capacity to avoid fugitive emissions.
In summary, Applikon's objectives for the project were met and, in the cost/benefit aspect, exceeded. The company already has many customers in the UK, such as Novartis, C6 Solutions and Johnson Matthey, already saving considerable amounts of money by using Applikon analysers.
Clairet Scientific installed an ABB Bomem FT-NIR analyser at A H Marks, placed in the control room laboratory and linked to the sampling system on the scrubber with 50 metres of fibre-optics. A multivariate (PLS partial least squares) model was constructed using data from multiple batches and multiple products. The model predicts the caustic strength from the NIR signal.
The system remained in-line monitoring caustic strength for over six months, including a routine spectrometer source change. The calibration model required no adjustment after the new source was fitted and predictions were as accurate as with the previous source.
Caustic concentration was reported every 100 seconds, resulting in continuous monitoring of the scrubber performance and the ability to reduce the caustic action level from 4% to 1%. The system can be integrated into a closed loop control system to automatically replenish the caustic reservoir when necessary.
The project was a clear demonstration of the usefulness of on-line monitoring using NIR. However, the real importance of this work is not so much in the science as in the economics of the analysis.
Caustic scrubbers are common to many plants and processes and so this is a highly 'portable' application, both in terms of the physical equipment and the model.
The entire analysis system was transferred from one plant to another, totally different, plant 30 km away at Lambson's. Although these two plants make different products, they both use caustic scrubbers. Much manpower and resource normally has to be expended to develop models in the early stages. By focusing on a generic application a great deal of common learning can be transferred to new locations. The generic caustic strength model greatly reduces the cost of entry into the technology.
Arguments can be made for including such monitoring purely on environmental grounds (reduction of waste and lowering of gaseous emissions), as well as on health and safety grounds (by the reduction of operator exposure to the scrubber stream when taking samples for titration).
However, the economic argument is also a very powerful one. Although caustic is relatively cheap, savings can be made from both reduced consumption and the reduction in costs of disposal of 'partially exhausted' caustic, which arises from reducing the amount of 'exhausted' caustic being generated. FT-NIR is easy to adapt so that one analyser can monitor multiple scrubbers, and in this configuration, pay-back for the equipment can be less than one year.
A second FT-IR system was installed by Clairet Scientific at Robinson Brothers to monitor a fluidised bed dryer by monitoring the water and solvent levels in the dryer exhaust gas. This is a straightforward measurement using mid-infrared (MIR) which can be made with simple univariate models. The system was able to produce drying curves that could provide the process operators with real-time information on the water or solvent content of the product. Savings in time and in energy costs can also make this a very attractive economic proposal.