The history
When the production volume in a pharmaceutical multi-purpose plant expanded significantly, the existing regenerative waste gas combustion was beyond its capacity. The air volume had grown beyond what could be treated. Therefore the unit was to be retrofitted with a concentrator to extend its use.
The engineering company commissioned with the retrofit had one major problem, though – there were no definitive waste gas concentrations to be had. Every measurement taken was only a snapshot in time, the product variety was so large and the possible product combinations so numerous that a continuous online VOC measurement of several months would have been necessary to give an idea of the waste gas composition.
To get a starting idea of what we were dealing with, I calculated what a „model waste air“ would look like, based on yearly solvent and chemical consumption and several production scenarios. The concentration filter (in this case a three-bed zeolite unit) was designed accordingly and tested with some bench-scale experiments in a university lab.
In this case – since time was pressing – management decided to forego on-site pilot-scale tests and to accept a certain degree of overengineering instead. While tests that lead to the correct dimensioning of a waste gas purification unit usually pay for themselves rather quickly, here time was at a premium, due to difficulties that had arisen with the local EPA.
How the model waste gas was calculated
Yearly lists of solvents and other volatile organic compounds used in production were obtained. It was not quite clear at the point how much of it was used for lab work or how much of it was regenerated. We assumed that eventually everything would end up as waste gas.
To get a model for the composition of the average waste gas, we calculated vapour pressure curves for all compounds and gauged the mass portion of each VOC by its vapour pressure and its portion within yearly consumption. This composition was adjusted to an average CFID-concentration, the only measurement that could be obtained.
Based on these figures, several scenarios were calculated, including more or less pronounced peaks and unfavourable combinations of VOC with strong coadsorption effects.
Bench scale tests
To design the thermal desorption process, bench scale tests on thermal conductivity of packed beds were performed by an independent lab. It helped to design the temperature profile for desorption which would be at once effective and energy efficient.
Adsorption tests were not performed, because the potential combination of two, three or more adsorptives was not even remotely comprehensible. We chose the most unfavorable combination of solvents which would create the strongest coadsorption effect in the system (in that case, xylene and dichloromethane) and based the dimensions of the concentrator on that. For any situation but that coadsorption case the filter was larger than it needed to be, but the customer preferred a certain degree of overengineering to even a small risk of temporary failure.
Design
To lower the size of the individual adsorbent beds and to reduce pressure drop, the concentrator contained three adsorbent beds instead of the two which are the normal case. Bed III was mostly a standby, to be used only when breakthrough in bed II occured before the regeneration of bed I was finished.
Up to now, the concentrator has been running for more than 10 years without issues.