(111e) Production of Bio-Olefins: Tall-Oils and Waste Greases to Green Chemicals and Polymers

There is an increasing trend of using bio-polyethylene and
bio-polypropylene in Europe for making consumer goods. However there is
currently very limited production capacity available for producing these base
chemicals that are used in the polymerization processes. This contribution will
give an overview of the presently available routes for the production of
bio-ethylene and bio-propylene (bio-ethanol to olefins, methanol to olefins,
hydrodeoxygenation of biomass followed by steam cracking, fast pyrolysis of
biomass) and discuss advantages and disadvantages. The talk will be completed
with results obtained from different pilot plant studies starting from tall-oil
and waste fats and greases. In a first step these feedstocks are catalytically
converted, in a second step  they are cracked towards olefins. The total
light olefin yield (ethylene and propylene) that has been obtained is in all
cases higher than with naphtha and is above 50 wt%
depending on process conditions, pretreatment and the biomass origin. 

Crude tall oil is a
viscous liquid obtained as a by-product of the Kraft process for wood pulp
manufacture when pulping mainly coniferous trees. It can be fractionated into
distilled tall oil (DTO) and tall oil fatty acids (TOFA). These fractions
mainly contain long chain fatty acids. DTO also contains significant amounts of
rosin acids, i.e. a mixture of organic acids such as abietic
acid. Catalytic hydrodeoxygenation (HDO) of both DTO as well as TOFA removes
the oxygen in these acids in the form of H2O, CO and CO2, producing highly
paraffinic hydrocarbon liquids, i.e. HDO-TOFA and HDO-DTO respectively.

Similarly, catalytic hydrodeoxygenation of triglyceride
based biomass (TGB), such as algae oils or low cost waste greases like poultry
fat and yellow grease, also produces paraffinic liquids (HDO-TGB) that are
attractive feedstocks for conventional steam crackers.

The detailed composition of the studied feedstocks was
determined using GC×GC-FID/TOF-MS. In Figure 1 the group-type compositions of
the hydrodexygenated oils are presented as well as
the composition of a typical petroleum-derived naphtha, i.e. currently the main
steam cracker feedstock. Compared to this naphtha, the HDO-TGB contains high
amounts of n-paraffins in a significantly higher
carbon range, i.e. C14-C24 for the HDO-TGB versus C3-C13 for the naphtha. Only
small amounts of naphthenes and no more than traces
of aromatics were detected in the HDO-TGB. Also the HDO-TOFA and HDO-DTO feeds
are highly n-paraffinic mixtures (C14-C24). The rosin acids present in the
untreated TOFA and DTO fractions result in significant amounts of tricyclic naphthenes, such as norabietane
(C19), and aromatics, such as norabietatriene (C19).
In both fractions also some fatty acids methyl esters (FAME) were measured.

Figure 1: Group-type composition
of (a) reference petroleum naphtha (C3-C13), (b) HDO-TGB (C14-C24), (c)
HDO-TOFA (C14-C24) and (d) HDO-DTO (C14-C24)

The combination of detailed
feedstock analyses and pilot plant data allows validation of the mechanistic
model describing the radical reactions of the feedstock molecules. Table 1
compares the measured and simulated yields of some important products.

Table 1:
Comparison of experimental and simulated product yields [d = 0.45 kg/kg;
τ = 0.3 s, COP = 1.7 bar, COT = 820 °C]

For the naphtha, the model
performance is quite exceptional. Not surprisingly, since the model was
optimized based on an extensive set of pilot plant data, mainly comprising
experiments with gaseous and naphtha feedstocks. Nevertheless, the model
performance is more than adequate for both HDO-TGB and HDO-TOFA. This, after
inclusion of retro-ene decompositions for long-chain
olefins which proved to be necessary to accurately model the measured product
distribution of these highly paraffinic feeds.

As discussed above, the HDO-TOFA also contains a small
amount of FAME. Since the SEMK model currently does not contain any esters,
these components were represented by n-paraffins.
This can explain the somewhat larger discrepancies between measured and
simulated yields for this feedstock. However, including the decomposition of
esters to the SEMK model should result in more accurate predictions.