(419d) Automated Selection of Products and Chemistry in Biomass Conversion Through Network Generation and Optimization | AIChE

(419d) Automated Selection of Products and Chemistry in Biomass Conversion Through Network Generation and Optimization


Marvin, W. A. - Presenter, University of Minnesota
Daoutidis, P. - Presenter, University of Minnesota
Rangarajan, S., University of Minnesota

Biomass can be upgraded to a multitude of valuable products using a variety of chemistries.  This leads to two related questions – What’s the best product to make from biomass? And, what’s the best route to make it? These questions have to be addressed simultaneously because while state-of-the-art molecule design methods can be used to efficiently identify the ideal compound for a particular application, it may be possible that it cannot be synthesized from biomass using known chemistries. However, identifying the spectrum of biomass-derived compounds is nontrivial because the network of potential reactions that upgrade biomass can be large (>1000) and complex.  We have developed an automated method1 to: (a) generate the network of synthesis steps that convert biomass to all possible products using our automated network generator, RING2, and (b) identify optimal compounds for desired applications and their synthesis routes within the network.

     In this talk, we will focus on two different classes of compounds – fuels and chemicals – that can potentially be obtained from biomass using heterogeneous catalysis. First, we will consider the problem of identifying the optimal mixture of biomass-derived oxygenates and hydrocarbons that can be blended with gasoline. We use three different objectives as descriptors of competing metrics for determining optimal blends: (a) energy loss during the synthesis as a descriptor of capital cost, (b) external utility requirements as a measure of operating cost, and (c) overall catalyst requirements as an indicator of the ‘speed’ of synthesis. We find that in all three cases several alternative mixtures of biomass-derived compounds can be identified such that the corresponding gasoline-mixture blend satisfies current ASTM standards on heating value, octane rating, vapor pressure, boiling points, and water solubility. We also find that these alternative solutions are typically composed of varying proportions of higher alcohols such as propanol and/or furans such as dimethylfuran.  We further find that multiple routes exist that can form a mixture.

     Second, we will consider the problem of synthesizing fatty alcohols from biomass such that nonionic surfactants made from these alcohols have desired physical properties. We will show that in addition to selecting the ideal fatty alcohol and potential synthesis routes, we can automatically obtain process information such as operating windows, thermodynamic bottlenecks, and potential coupling of reactions and/or phases.

[1] W. Alex Marvin, Srinivas Rangarajan, and Prodromos Daoutidis. Automated Generation and Optimal Selection of Biofuel-Gasoline Blends and Their Synthesis Routes, Energy & Fuels, In Press (2013).

[2] Srinivas Rangarajan, Aditya Bhan, and Prodromos Daoutidis. Language-oriented rule-based generation and analysis of complex reaction networks: Description of RING, Computers  & Chemical Engineering, 45, 114-123, 2012