443519 Using  Modeling to Compare and Optimize Continuous Chromatographic Processes

Tuesday, April 12, 2016: 9:00 AM
337B (Hilton Americas - Houston)
Jay Yun1,2, Laurent David1 and Roger-Marc Nicoud1, (1)Ypso-Facto, Nancy, France, (2)Ypso-Facto, Cambridge, MA

Sugar purification is one of the largest application of continuous chromatography. For instance, fructose is separated from glucose to produce High Fructose Corn Syrup (HFCS).

A typical application consists in producing HFCS 55 (55 % fructose) from HFCS 42 (containing 42 % fructose). 

These separations, typically performed on ion-exchange resins, are extremely productive: relatively linear adsorption isotherm enables very high feed concentration, up to 800 g/L.

The sugar market is extremely competitive, thus fine-tuning the system to maximize recovery, productivity and minimize water consumption is of utmost importance. 

In order to understand the key drivers for optimization, we have developed advanced modeling tools and simulation software.

In spite of their relative linearity, the glucose and fructose adsorption isotherms are in fact slightly anti-Langmuirian (slope increases with concentration) and synergetic (increasing fructose concentration increases glucose adsorption and vice versa).

Using the so-called equilibrium model (triangle representation) with this uncommon adsorption behaviour, we identify SMB parameters that are critical for the required separation, and we show that fructose can be purified and concentrated at the same time.

Furthermore, an alternative system was considered: instead of setting the parameters to obtain 55 % fructose in one step, we investigated the possibility to produce 90 % fructose and use a bypass to obtain the targeted 55 %.

Both processes are compared in terms of equilibrium model.

Finally, using ChromWorksTM software, we made a comparison between more realistic systems, including SMB, S-SMB and I-SMB and we were even able to design new innovative custom multi-column configurations by predicting their performances.

Extended Abstract: File Uploaded