Berkeley Lab team develops predictive model to optimize biomass blends for commercial-scale biorefinery processing
The type of biomass available for fuels and chemical production varies with the time of year and geography; the ability to integrate multiple feedstocks into the process of a bio-refinery can substantially de-risk the logistics of supply. However, pioneering bio-refineries have been optimized to process a single feedstock such as corn stover.
Now, researchers from Lawrence Berkeley National Laboratory, with their colleagues at Idaho and Sandia National Laboratories, have developed a model to predict continuous envelopes of biomass blends that are optimal for a given pretreatment condition to achieve a pre-determined sugar yield—or vice versa. A paper on their work is published in the journal Bioresource Technology.
The model identified optimal biomass blends that would allow the blending of low-quality, low-price feedstock with higher quality ones to not only expand the scope of establishing a biorefinery in geographical areas beyond the Corn Belt but also reduce costs.
Petroleum industry has long been utilizing the power of statistical modeling to rapidly adapt process conditions per compositional variability of incoming crude oil feedstocks and manufacture a variety of products. Feedstock composition of each batch of crude oil shipped to a refinery can vary considerably and this variation becomes pronounced because a commercial-scale petroleum refinery processes crude oil in quantities exceeding 100,000 barrels per day. This variability can make any attempts at pre-determining optimal process conditions futile. Petroleum refineries utilize non-linear modeling to tune process conditions and fully convert each batch of crude oil but, in consequence, vary the yields of individual products existing in their pre-established suite. Complete conversion of feedstock has a higher impact on process economics than variability in quantities of products manufactured by a refinery. Furthermore, a refinery’s ability to adapt to feedstock variability mitigates risks associated with dedicated feedstock supply chains.
… Idaho National Laboratory (INL) is currently developing a method to blend high-quality feedstocks with low-quality ones to reduce supply-side risks at commercial-scale biorefineries. High-quality feedstocks are defined as those that deconstruct and convert readily to biofuels. Least Cost Formulation (LCF) is being developed to evaluate low-cost biomass feedstocks across the United States and identify geographical locations for bio-refineries that can integrate diverse biomass feedstocks from multiple sources into their supply chain and lower overall feedstock costs. While such integration is theoretically possible, it is vital to ascertain that blending feedstocks does not negatively impact biomass conversion performance, and thereby product yields.
… Biorefineries, whether applying traditional or novel deconstruction catalysts, will have to be equipped with the ability to tune deconstruction processes per compositions of biomass blends and completely convert them to valuable intermediates and products. Blending feedstocks and tuning deconstruction process per the composition of biomass blends will be vital in attaining the goal of de-risking biorefining.—Narani et al.
Researchers at the Advanced Biofuels Process Demonstration Unit (ABPDU) at the Lawrence Berkeley National Laboratory (LBNL) worked with Idaho and Sandia labs on a multi-year project to improve the economics of deconstruction of biomass blends and their conversion to bioproducts.
The goals of the study were to (i) use the LCF model to select three diverse types of biomass feedstocks available in a geographic region and (ii) build a robust predictive model to maximize sugar yields by identifying optimal deconstruction process conditions for a given biomass blend or vice versa.
Their model predicted that a more than 60% glucose yield can be achieved by treating an equal part blend of energy cane, corn stover, and switchgrass with alkali pretreatment at 120 ˚C for 14.8 h. By using ionic liquid to pretreat an equal part blend of the biomass feedstocks at 160 ˚C for 2.2 h, they achieved 87.6% glucose yield.
The use of such a predictive model could potentially overcome dependence on a single feedstock, they concluded.
Akash Narani, Phil Coffman, James Gardner, Chenlin Li, Allison E. Ray, Damon S. Hartley, Allison Stettler, N.V.S.N. Murthy Konda, Blake Simmons, Todd R. Pray and Deepti Tanjore (2017) “Predictive modeling to de-risk bio-based manufacturing by adapting to variability in lignocellulosic biomass supply” Bioresource Technology 243, 676-685 doi: 10.1016/j.biortech.2017.06.156