High-value propylene glycol from low-value

biodiesel glycerol: A techno-economic and

environmental assessment under uncertainty

Andres Gonzalez-Garay, † Maria Gonzalez-Miquel, ‡ Gonzalo Guillen-Gosalbez,*, †

†Department of Chemical Engineering, Centre for Process Systems Engineering, Imperial

College, South Kensington Campus, London, SW7 2AZ, UK

‡School of Chemical Engineering and Analytical Science, University of Manchester, Manchester,

M13 9PL, UK

KEYWORDS

Biodiesel glycerol, propylene glycol, LCA, economic assessment

1

ABSTRACT

Recent governmental policies that promote a bio-based economy have led to an increasing

production of biodiesel, resulting in large amounts of waste glycerol being generated as low-cost

and readily available feedstock. Here, the production of high-value bio-based propylene glycol as

an alternative chemical route to valorize biodiesel glycerol was studied and assessed considering

economic and life cycle environmental criteria. To this end, the conventional industrial process

for propylene glycol production, which uses petroleum-based propylene oxide as feedstock, was

compared against three different hydrogenolysis routes based on biodiesel glycerol using process

modeling and optimization tools. The environmental impact of each alternative was evaluated

following Life Cycle Assessment principles, while the main uncertainties were explicitly

accounted for via stochastic modelling. Comparison among the various cases reveals that there

are process alternatives based on biodiesel glycerol that outperform the current propylene glycol

production scheme simultaneously in profit and environmental impact (i.e. 90 % increment in

profit and 74 % reduction in environmental impact under optimum process conditions). Overall,

this work demonstrates the viability to develop sustainable biorefinery schemes that convert

waste glycerol into high-value commodity chemicals, like propylene glycol, thereby promoting

holistic bioeconomy frameworks.

2

INTRODUCTION

The need to develop a more sustainable chemical industry has spurred substantial research for

replacing petroleum-based feedstocks by renewable ones1. Several studies have already

demonstrated that bio-based chemicals can meet the quality standards required within the

industry while at the same time bringing significant environmental benefits when compared with

their corresponding fossil-based counterparts2,3. Consequently, different economies around the

world are implementing policies and legislations to promote the bioeconomy, which focuses on

the use of renewable resources across the industry4,5.

Biofuels production has been one of areas with large potential to contribute towards the

development of a more sustainable chemical industry. This has led to many regulations seeking

to promote their industrialization and commercialization. As a result, large amounts of biofuels

have been manufactured over the last years, opening up new opportunities for using their by-

products in other chemical routes (e.g., bio-ethanol derivatives, bio-diesel glycerol, etc.). The use

of these molecules as platform molecules in the production of chemicals enhances the so called

bioeconomy, while at the same time reduces the use of petroleum-based compounds.

Among these by-products, glycerol has received much attention, since it is a highly active

molecule with a wide range of applications6. Biomass-based glycerol is generated as by-product

in the transesterification of vegetable oils during the production of biodiesel (10 wt. % of total

biodiesel production7). Before the biodiesel market took off, glycerol was an expensive chemical

seldom used as feedstock. The large amounts of biodiesel glycerol produced in the last decade

caused a drastic price drop, stimulating its use as platform chemical (i.e. price for crude glycerol

declined from 380 $/ton in 2002 to less than 100 $/ton in 20127). In fact, the fast growth of

biodiesel production has resulted in 88% of the global glycerol demand being supplied by this

3

process in 20137. Considering that biodiesel production is expected to grow at an estimated

annual rate of 10%8, there will be soon a surplus of glycerol supply that the current market

cannot accommodate. As a result, exploitation of glycerol as inexpensive, abundant feedstock is

receiving increasing attention as an strategy to develop more sustainable processes and products,

including valuable bio-based chemicals and novel bio-renewable solvents9–11.

The conversion of biodiesel glycerol to propylene glycol (PG) emerges as an appealing

alternative, since the market demand of PG can absorb large consumptions of glycerol12. PG is a

major commodity across the world having an annual production over 2.18 million tons in 2014

and annual growth of 8%13. The main application for its industrial grade arises in the production

of polymers, while the human-safe grade (USP grade) has a wide application as solvent in the

food and pharmaceutical industry14

PG is traditionally produced from propylene oxide (PO), which reacts with water to produce

PG along with di- and tripropylene glycols. Propylene oxide is a petroleum-based chemical

derived by the chlorohydrin or the hydroperoxide processes15. Over the last decade, different

studies have analyzed the production of PG from renewable sources such as glycerol, sorbitol or

biomass13,16,17. Among these options, catalytic hydrogenolysis of glycerol to PG has been put

forward as a sustainable production route and studied under several operating conditions. Some

of the alternatives evaluated include systems at high or atmospheric pressure12,18,19, isothermal or

non-isothermal conditions12,18–21, external or in situ generated hydrogen20–25 and liquid or vapor

phase reactions26–28. However, little focus has been placed on the design and evaluation of the

process at an industrial level, which plays a crucial role in the development of a feasible

bioeconomy in terms of economic, environmental and social impact.

4

Computer-aided process engineering tools enable this type of assessment by estimating the

performance of a chemical process via techno-economic analysis, which should ideally account

for both economic and environmental criteria at the design stage29–34. In this context, Life Cycle

Assessment (LCA) has emerged recently as the preferred tool to quantify the environmental

impact of a chemical product, mainly because of its holistic scope that embraces all the material

and energy flows taking place across the product’s supply chain35. Unfortunately, chemical

processes are subject to different sources of uncertainty that introduce variability into the

decision-making problem. Variations in technical, market and supply chain parameters certainly

affect the performance of the processes, and the proper understanding of their impact become

essential for the success of a sustainable design. The incorporation of uncertainty analysis in the

assessment and optimization of more sustainable processes has been addressed in areas such as

supply chain management36–38, process synthesis36,39, energy systems40 and water management41,

among others. Despite these advances, many techno-economic studies still neglect uncertainties

and report nominal values for the economic and environmental performance rather than

stochastic ones.

Some authors have studied the production of PG from biodiesel glycerol. Posada et al.42

analyzed the economic performance of chemical and biochemical processes that convert glycerol

into six different valuable products, concluding that the production of PG represented the best

economic option, with a sale price / total cost of production ratio of 1.57. The authors considered

the use of external hydrogen at atmospheric pressure, focusing on a standard design with no heat

integration and which was not subject to process optimization. Focusing on environmental

issues, Adom et al17 found that savings of around 60% in energy consumption and greenhouse

gas emissions could be attained in the production of PG by replacing the conventional

5

petroleum-based process by hydrogenolysis of bio-based glycerol. The authors, however,

provided no details on the specific hydrogenolysis route used in the assessment. Following a

different approach and considering another biomass-source of glycerol, Gong and You16

developed a superstructure to assess the use of microalgae as raw material in the production of

biodiesel, hydrogen, PG, glycerol-tert-butyl ether and poly-3-hydroxy-ybutyrate. During the

assessment, Gong and You considered the use of different routes to produce hydrogen from

glycerol as well as external hydrogen. Their analysis, however, is restricted to one single route

concerning the hydrogenolysis process. The authors found that when PG is the only bio-product

generated, 1.82 kg of CO2 equivalent per kg of PG produced are generated, which represents a

reduction of 51.5% compared to the propylene oxide technology. The economic results for this

case are nevertheless not reported. It is worthy to mention that none of the previous studies

handled uncertainties in their assessments.

Focusing on the enhancement of the bioeconomy and the replacement of petroleum-based

compounds by renewable feedstocks, we here address the economic and environmental

assessment under uncertainty of different routes for the production of PG from biodiesel

glycerol. More precisely, three different routes are compared against a benchmark industrial PG

technology based on the use of petroleum-derived propylene oxide.

The paper is organized as follows. First, we describe the four different routes considered in the

assessment. We then introduce the methodology followed and present and discuss the results of

the analysis. Finally, the conclusions of the assessment are drawn and the most sustainable route

for the production of PG is further discussed.

PROCESS DESCRIPTION

6

We consider crude glycerol produced from the transesterification of vegetable oils in biodiesel

plants. This glycerol usually contains water, methanol, salts and other organic material. In this

study, we assume that crude glycerol is purified before it enters any other process by removing

methanol, salts and organics, obtaining a feed stream 90 wt. % glycerol and 10 wt. % water43.

The liquid feed stream of propylene oxide/glycerol for the proposed cases has a flow rate of 75

kgmole/h. Propylene glycol is produced with 99.5 wt. % purity in all of the cases. A description

of the alternatives proposed along with the mechanisms considered for each route is presented

next, while further details can be found in Appendix A of the supplementary information.

Route business as usual (BAU): Propylene oxide conversion. Figure 1 shows a flow

diagram of the standard BAU process, where PG is produced from liquid-phase hydrolysis of

propylene oxide (PO) under a non-catalytic reaction14. Propylene oxide and water are mixed

according to the ratio 1:1514, since an excess of water is required in the process to limit the

generation of by-products dipropylene glycol (DPG) and tripropylene glycol (TPG). The reaction

takes place at 18.25 bars and 190 °C, achieving full conversion of PO with a yield of 85 % to

PG, 10 % to DPG and 5% to TPG. Distillation columns operate under vacuum at 0.1 bars to

avoid decomposition of PG. By-products are recovered with 99.5 wt. % purity in both cases.

Figure 1. Production of PG from propylene oxide conversion (BAU).

7

Route glycerol-based 1 (GB-1): Isothermal hydrogenolysis at high pressure and external

hydrogen. Figure 2 shows the flowsheet of the second route. This alternative follows the two-

step mechanism introduced by Pudi et al.44. The reaction is carried out at 205 °C, 20 bars and

with a glycerol concentration of 75 wt. %45. The conversion of glycerol reported at these

conditions is 88.7 % with a selectivity to propylene glycol of 94.3%. A molar ratio

hydrogen/glycerol 5:1 is used in the simulation18. As in the previous alternative, distillation

columns operate under vacuum to avoid decomposition of PG. Methanol and ethylene glycol

(EG) are generated as by-products and are recovered with 99.5 wt. % purity.

Figure 2. Hydrogenolysis of glycerol at high pressure and isothermal conditions with external

hydrogen (GB-1).

Route glycerol-based 2 (GB-2): Non-isothermal hydrogenolysis at ambient pressure and

external hydrogen. Figure 3 shows the flowsheet of the process. This alternative is based on the

work by Akiyama et al.21, following the two-step mechanism presented in alternative GB-1. This

option requires a gradient temperature reactor which operates at 200 and 120 °C at the top and

bottoms, respectively. Fresh glycerol is not diluted, since Akiyama et al. showed that the

concentration of glycerol has no impact on the conversion. The molar ratio hydrogen/glycerol is

8

5:118. Distillation columns operate at atmospheric pressure and by-products methanol and EG are

recovered with 99.5 wt. % purity.

Figure 3. Hydrogenolysis of glycerol at ambient pressure and non-isothermal conditions with

external hydrogen (GB-2).

Route glycerol-based 3 (GB-3): Isothermal hydrogenolysis at high pressure and in situ

generated hydrogen. The use of external hydrogen may lead to high operation costs as well as

higher environmental impact, since most of it is produced from fossil fuel in refineries. In order

to circumvent this limitation, we consider as third alternative hydrogenolysis using in situ

generated hydrogen. Figure 4 shows the flowsheet of the process. In the simulation, we

implement the reaction mechanism presented by Maglinao et al.20,25, where methanol, ethanol

and propanol are generated as by-products. The reaction takes place at 240 °C and 20 bars with a

glycerol solution 50 wt. %. Glycerol conversion reported at these conditions is 96% with a yield

towards PG of 33%. The liquid products of the reactor are primarily separated into light alcohols

(methanol, ethanol and propanol) and heavy alcohols (PG and glycerol). The separation of heavy

alcohols is performed under vacuum to avoid degradation of PG. As for the light alcohols,

purification is carried out at atmospheric pressure requiring an extractive distillation column,

9

since an azeotrope ethanol/propanol/water is formed. Methanol and propanol are recovered with

99.5 wt. % purity while ethanol achieves 99.3 wt. %.

Figure 4. Isothermal hydrogenolysis of glycerol at high pressure with in situ generated hydrogen

(GB-3).

MATERIALS AND METHODS

The full description of the methodology applied to assess each alternative is presented in

Appendix B of the supporting information. In essence, a simulation model of each process was

first developed using traditional equipment models and then optimized through standard

heuristics46, sensitivity analysis and heat integration47. The economic performance and life cycle

impact were both assessed afterward considering different uncertainties modeled via Monte

Carlo sampling.

RESULTS AND DISCUSSION

The four processes described above were simulated with Aspen-Hysys v8.8 using UNIQUAC

activity coefficients to model the liquid-vapor equilibrium of the system48. Conversion reactors

were defined using stoichiometric data retrieved from different sources14,20,21,25,44,45. Process

10

integration is essential to attain sustainable designs and can include heat, mass and property

integration. In our assessment, only heat integration was addressed since we considered that the

potential savings of mass integration are marginal while property integration had no application

to the processes described. We first present the results for the optimized flowsheet of each

alternative to later on discuss the uncertainty attached to the models. Energy and mass balances

per kg of PG produced are summarized in Table 1.

Table 1. Overall mass and energy balances for the production of 1 kg of propylene glycol.

Concept BAU GB-1 GB-2 GB-3

Raw materials

Propylene oxide (kg) 0.9034 - - -

Glycerol solution 90 wt. % (kg)

- 1.4238 1.3707 3.7300

Hydrogen (kg) - 0.0297 0.0321 -

Water (kg) 0.2165 0.0093 - 0.5687

Waste streams

Gas Purge (kg) - 0.0052 0.0071 2.7926

Wastewater (kg) - 0.4305 0.3798 0.3205

Products

By-products (kg) DPG: 0.1326

TPG: 0.0087

Me: 0.0111

EG: 0.0178

Me: 0.0080

EG: 0.0146

Me: 0.0325

Et: 0.1316

Pr: 0.0165

Energy consumption

Electricity (kW) 0.1229 0.0578 0.0582 0.1214

Heating demand (MJ) 11.231 4.635 4.819 16.707

Cooling demand (MJ) 12.640 5.970 6.157 12.288

DPG: Dipropylene glycol; TPG: Tripropylene glycol; EG: Ethylene glycol; Me: Methanol; Et:

11

Ethanol; Pr: Propanol.

Economic Assessment. The economic performance is quantified using the economic potential

(EP/kg of PG) and total annualized cost per kg of PG produced (TAC/kg of PG)49,50. EP is

defined as the net profit after taxes while the TAC is the summation of the fixed and variable

costs of operation plus an annual capital charge. To express the capital costs on an annual basis,

330 days of operation are considered and the annual capital charge is calculated following a 10-

year straight line depreciation and considering an interest rate of 15%46. Glycerol price is taken

from the current market (0.25 $/kg), while no further subsidies or incentives are considered in

the assessment. Further detail of the methodology applied is presented in Appendix B of the

supporting information. Raw materials and equipment costs are shown in Tables S12 –S14.

Figure 5 displays the contribution to the TAC and the economic potential per kg of PG

produced for the alternatives proposed. We identify that all the alternatives generate profit, being

alternatives GB-1 and GB-2 the routes with the best economic performance. In terms of the TAC

per kg of PG produced, both options present significant reductions when compared to the BAU

case (0.679 $/kg of PG for GB-1 and 0.636 $/kg of PG for GB-2 versus 1.781 $/kg of PG in the

BAU case). The main reason behind such savings is the difference in price between propylene

oxide and glycerol. In the BAU case, the cost of propylene oxide (1.53 $/kg of PG) is already

higher than the total cost reported for either process GB-1 or process GB-2. The profit obtained

for alternative GB-2 is 1.326 $/kg of PG, which represents an increase of 90% compared to the

BAU case. Alternative GB-1 is the second best option with 1.300 $/kg of PG, representing an

increase of 86% compared to the BAU case. In contrast, the low yield to PG attained in

alternative GB-3 increases the cost per kg of PG by 19% compared to the BAU case (i.e. 2.058

$/kg of PG). Consequently, GB-3 shows a significant decrease in economic potential generating

12

only 0.251 $/kg of PG, which corresponds to 36% of the EP/kg of PG obtained in the BAU case

(0.700 $/kg of PG).

In our assessment, no subsidies nor incentives where considered. However, the significant

improvements attained in the economic performance of alternatives GB-1 or GB-2 certainly

promote the shift from the current process to the glycerol-based options on a long term basis. In

addition, the aim of the industry to boost the bioeconomy and the increasing demand of PG

(resulting in the generation of more plants) can further favor the incorporation of the glycerol-

based options to the current market.

Figure 5. Contribution to the total annualized cost and economic potential per kg of PG

generated.

It is worthy to mention that the economic results presented in this assessment differ from those

reported by Posada et al.42. More precisely, the ratio commercial sales price/ total cost of

production per kg of PG is in our case 4.2 for the best alternative (GB-2), versus 1.57 in their

case. The price of PG used in the assessment plays a crucial role in the final value of the

13

economic indicator evaluated. As far as we are aware, Posada et al. did not report the PG price

used in their assessment, making it hard to carry out a direct comparison of TAC values between

their work and ours. We anyway understand that these discrepancies might be due to the use of a

non-isothermal reactor (as opposed to the isothermal reactor used in their case), as well as the

application of heat integration and sensitivity-based optimization. As an example, the use of a

non-isothermal reactor increases the yield of PG from 80% to 98%, yielding savings of as much

as 20% in the TAC. Furthermore, high contribution of utilities to the TAC reported by Posada et

al. suggests that there was indeed room for improvement in their process via optimization and

heat integration.

Environmental assessment. To quantify the environmental performance of each alternative,

we follow the LCA methodology CML 2001. Glycerol is assumed to be produced from the

transesterification of soybean oil in the US. An analysis to validate the impact loads attached to

glycerol in the Ecoinvent database is presented in the Appendix C of the supplementary

information. Environmental impacts are evaluated per kilogram of PG produced while economic

allocation is applied to distribute the total impact of the processes among products and by-

products. Ecoinvent51 v.3.2 database is used to obtain the impact data of streams located beyond

the plant boundaries. We assume that purge gas streams are burned, while the liquid waste is

immediately sent to a wastewater treatment plant. Entries taken from Ecoinvent database are

displayed in Table S2.

Figure 6 shows the environmental impact of the different alternatives proposed. As in the

economic analysis, alternative GB-2 has the best performance in all the categories, followed by

alternative GB-1. When comparing any of these two options against the propylene oxide case,

we observe a significant reduction in the environmental impact. The greatest improvement is

14

achieved in the category ozone layer depletion with reductions of 89% (1.67·10-6 kg CFC-11-

eq/kg of PG in the BAU case versus 1.82·10-7 and 1.78·10-7 kg CFC-11-eq/kg of PG for

alternatives GB-1 and GB-2, respectively). The lowest improvement is shown in photochemical

oxidation, with a drop of 59% in GB-1 and 60% in GB-2 (2.38·10-3 kg CFC-11-eq/kg of PG in

BAU versus 9.78·10-4 kg CFC-11-eq/kg of PG in GB-1 and 9.43·10-4 kg CFC-11-eq/kg of PG for

GB-2). All the categories are improved on average in as much as 73% in GB-1, and 74% for GB-

2. Alternative GB-3 shows the highest environmental impact among the glycerol-based options.

This is due to i) the low yield of PG (35% versus 98% in GB-2 and 96 in GB-1), ii) the large

amount of emissions generated, and iii) the utilities and equipment necessary to carry out the

reaction and separation steps.

The potential benefits of these savings on the planet are hard to ascertain since they depend on

the extent to which the new technologies might be deployed, their location, transportation routes,

logistics, etc. In an attempt to achieve a better appreciation of these savings, we focus next on the

category global warming potential, for which specific targets have been pledged to reduce the

greenhouse gas emissions by 2020. If we take the production of PG in the US during 2014, the

total contribution of PG to the CO2 emissions account for 3.10 million tons of CO2-eq if

produced with the BAU case. If we consider the production of PG from one of the glycerol-

based options GB-1 or GB-2, the CO2 emissions would account for 1.21 million tons of CO2-eq.

Hence, the production of PG from any of the glycerol-based options GB-1 or GB-2 could reduce

the CO2 emissions of the chemical sector in the US by 2.79% according to the emissions reported

in 2005. The reduction of the total greenhouse gas emission in the US would represent 0.08%.

Despite the global benefits are related to global warming mitigation, we would certainly benefit

in turn from lower levels of atmospheric pollution causing local impacts.

15

With regards to the main contributors of impact, for the BAU case, raw materials entail 94% of

the total environmental impact, from which propylene oxide contributes with 99% and water

accounts for the remaining 1%. Utilities are responsible for the remaining 6% of the impact, with

the heating demand representing 80% of the utilities impact. In options GB-1 and GB-2, based

on external hydrogen, raw materials account for 89% of the total impact. In alternatives GB-1

and GB-2, glycerol represents 98% of the raw materials impact, while hydrogen is responsible

for the remaining 2%. As for alternative GB-3, based on the in situ generation of hydrogen, the

contribution of raw materials is 75%, while utilities represent 21%, steel 1% and waste 3%.

Figure 6. Environmental life cycle assessment results for the proposed alternatives. [Impacts

expressed per kg of PG. AP: Acidification potential; GWP: Global warming potential; DAR:

Depletion of abiotic resources; FAET: Fresh aquatic ecotoxicity; MAET: Marine aquatic

ecotoxicity; TE: Terrestrial ecotoxicity; EP: Eutrophication potential; HT: Human toxicity; OLD:

Ozone layer depletion; PO: Photochemical oxidation].

16

Note that the reductions in total GHG emissions in GB-1 and GB-2 compared to the propylene

oxide case are above those described by Adom et al.17, who reported 8 and 4.5 kg CO2-eq/kg of

PG for the propylene oxide case and glycerol option, respectively, versus 4.8 and 1.85 kg CO2-

eq/kg of PG in our case. Hence, we achieve an impact reduction of 61% (compared to the BAU

case) versus 56% in their work. Note, however, that the work of Adom et al. was based on a

cradle to grave analysis, while ours is cradle to gate and thus omits end-of-life stages. This

difference in scope might explain the discrepancy between the two assessments. Conversely, the

value of 1.85 kgCO2-eq for option GB-2 is close to the one reported by Gong and You 16 (1.82

kgCO2-eq), who applied a similar cradle to gate LCA study, although in their assessment

microalgae are considered as raw material (as opposed to the biodiesel glycerol used in ours).

It is worthy to note the use of intensified processes may lead to additional improvements in

both economic and environmental criteria. Intensified processes are known to lead to more

compact, energy-efficient and environmentally friendly processes. However, they were left out

of the analysis because we lack the necessary data for the realistic modeling of such equipment

units.

Despite obtaining results of the same order of magnitude as other studies, caution must be

placed concerning the source of biomass used during the assessment. The use of different types

of biomass and/or logistics (e.g. locations of the facilities for the production of soybean oil) may

lead to different results. In Table S6 of the supporting information we further assess alternative

GB-1 considering five different sources of biomass, showing how different conclusions might be

reached depending on the assumptions made.

Uncertainty analysis. The process is affected by several technical and environmental

uncertainties. To handle them, we first performed a sensitivity analysis over the technical

17

parameters (Tables S7 and S8) in order to identify those with the highest impact on the economic

and environmental performance. The most critical parameters were found to be the prices of

products and raw materials, process conversions and raw materials flowrates. These parameters

were then modelled via normal distributions using the mean values and standard deviations

shown in Table S8. Furthermore, environmental uncertainties associated with the data retrieved

from Ecoinvent51 were modelled following a simplified version of the approach proposed by

Weidema and Wesnaes52, which makes use of the Pedigree matrix. More precisely, the life cycle

impacts embodied in the inputs to the processes were modelled as lognormal distributions, using

the mean values obtained from Ecoinvent and standard deviations (SD) obtained from the

Pedigree matrix (and considering the main emission causing the corresponding impact, see Table

S10).

After modelling all the uncertain parameters, Monte Carlo sampling was applied to generate a

set of samples, each entailing a specific set of values of the uncertain parameters and for which

the calculations were repeated iteratively. A total of 3,000 samples were generated, ensuring that

the mean relative error of the economic and environmental performance indicators would fall

below 5% for a confidence level of 95% following the statistical test developed by Law53 (see

details in appendix B of the supplementary information).

In terms of the mass and energy flows, the uncertainty analysis reveals fluctuations of up to

10% in variables such as PG production and utilities consumption, while by-products and waste

streams are more sensitive (from 5% up to 70%). Figure 7 presents the EP/kg of PG and TAC/kg

of PG evaluated considering the different uncertainties. Results are displayed using box plots,

where the central mark indicates the mean and the bottom and top edges indicate the 25th and

75th percentiles, respectively. The whiskers extend to the most extreme data points within ±2.7

18

standard deviations. The results show that the TAC/kg of PG falls in the interval 1.792±15% for

the BAU case, 0.695±9% in GB-1, 0.669±12% in GB-2 and 1.888±21% in GB-3. The EP/kg of

PG falls in the interval 0.688±35% for the BAU case, 1.285±23% in GB-1, 1.336±22% in GB-2

and 0.393±106% in GB-3. In GB-3, the high fluctuation of EP/kg of PG makes the process

economically unappealing in some scenarios. As in the deterministic evaluation, the mean values

of the economic indicators present alternative GB-2 as the option with the best performance,

followed by cases GB-1, BAU and GB-3. The overall mass and energy balances for all the

alternatives under uncertainty are presented in Table S11.

Figure 7. Total annualized cost and economic potential per kg of PG generated under

uncertainty.

The full LCA results are presented in Figure S6 of the supplementary material. In Figure 8, we

display the results of the categories with the highest uncertainty. As in the economic indicators,

the central mark indicates the mean value and the bottom and top edges indicate the 25th and

75th percentiles, respectively. The whiskers extend to the most extreme data points within ±2.7

standard deviations. Glycerol-based options GB-1 and GB-2 remained as the best alternatives

19

and appear more robust against the environmental uncertainty under the assumptions of the

study. In the deterministic evaluation of the environmental performance, alternative GB-2 shows

the lowest impact in the 10 categories evaluated. However, when uncertainty in the processes is

considered, the mean value of alternative GB-1 slightly outperforms alternative GB-2 in 5 of the

10 categories evaluated. The BAU case and alternative GB-3 show the highest environmental

impact and wider distribution among the processes evaluated. In the BAU case, the variation of

the results is attributed mainly to the uncertainty in the Ecoinvent data. As for alternative GB-3,

the high variation in the environmental categories is attributed mainly to the impact of the

conversion in the process. Among the categories, global warming potential shows the lowest

variation from its corresponding mean (17 % in BAU, 6 % in GB-1, 6% in GB-2 and 9% in GB-

3). The largest variation is identified in the category of marine aquatic ecotoxicity (186 % in

BAU, 171 % in GB-1, 176% in GB-2 and 169 % in GB-3).

20

Figure 8. Environmental life cycle assessment results under uncertainty for the proposed

alternatives. Impacts expressed per kg of PG.

CONCLUSIONS

Herein, four different processes for propylene glycol production have been economically and

environmentally evaluated through process simulation and optimization tools along with life

cycle assessment. The results presented for the deterministic evaluation of the alternatives show

that the use of an external source of hydrogen at atmospheric pressure and gradient of

temperatures (GB-2) offers the best glycerol route with potential to increase profitability and

reduce the environmental impact (compared to the BAU process) in all the categories evaluated.

An additional benefit is the use of atmospheric pressure through all the process. The use of high

pressure at isothermal conditions with an external hydrogen source (GB-1) is presented as the

21

second best option, leading to a win-win scenario compared to the BAU case but with slightly

lower economic potential and environmental impact reduction than route GB-2. The assessment

shows that hydrogen has a low contribution towards both the economic and environmental

performance. Therefore, the use of in situ generated hydrogen at high pressure (GB-3) presents

the worst performance given its low yield towards PG. The recovery of the by-products

generated has no significant impact either in economic or environmental terms, while it requires

an expensive and complex process configuration. Since all the routes evaluated have shown a

high dependence on raw materials, we can conclude that the use of biodiesel glycerol represents

a more sustainable route for the production of PG as far as the source of biomass has low

environmental impact embodied. Hence, the production of PG from biodiesel glycerol can

represent not only a more sustainable option compared to the conventional process, but also an

important route to overcome the surplus of glycerol.

The uncertainty analysis shows that the most critical parameters are the prices of products and

raw materials, conversions of the process and feed flowrates. The economic indicators can vary

in as much as 106% from the corresponding mean. Nevertheless, the ranking according to the

mean value remains the same as the deterministic evaluation. As for the environmental

indicators, variations in as much as 186% from the corresponding mean are observed. Among the

alternatives, the results obtained in GB-1 and GB-2 appear more robust against uncertainty than

routes BAU and GB-3. Overall, the uncertainty analysis presents alternatives GB-1 and GB-2 as

the most appealing routes to be further considered for industrial development. As for alternative

GB-3, we advise improvement of the catalytic reaction prior further analysis.

Given that the alternatives are based on experimental studies and computer simulations, the

results presented still carry a relatively high degree of uncertainty. While consistent with other

22

reference studies, a more detailed assessment via pilot plants should be carried out before the

scale up of the process. Ultimately, alternative processes like the ones discussed in this article

can contribute to boost the bioeconomy, ensure a more sustainable industrial development and

address major social, environmental and economic challenges faced nowadays.

ASSOCIATED CONTENT

Supporting Information.

Full description of the alternatives proposed and methodology applied in the assessment, table of

prices for the commodities used, environmental entries taken from Ecoinvent database, LCA

model for the transesterification process, evaluation of PG production using different biomass

sources, technical parameters considered in the sensitivity and uncertainty analysis, data entries

and uncertainty basic factors for the evaluation of the Pedigree matrix, equipment characteristics

and prices, summary of the mass and energy balances for all the alternatives under uncertainty,

economic evaluation under uncertainty and uncertainty graphs for all the environmental

categories are included. This material is available free of charge via the Internet at

http://pubs.acs.org.

AUTHOR INFORMATION

Corresponding Author

*G. Guillén-Gosálbez. E-mail: g.guillen05@imperial.ac.uk.

Notes

The authors declare no competing financial interest.

ACKNOWLEDGMENTS

23

Gonzalo Guillén-Gosálbez would like to acknowledge the financial support received from the

Spanish "Ministerio de Ciencia y Competitividad" through the project CTQ2016-77968-C3-1-P.

Andrés González-Garay acknowledge the financial support granted by the Mexican “Consejo

Nacional de Ciencia y Tecnologia (CONACyT)”.

REFERENCES

(1) Clark, J. H.; Farmer, T. J.; Hunt, A. J.; Sherwood, J. Opportunities for Bio-Based Solvents

Created as Petrochemical and Fuel Products Transition towards Renewable Resources. Int.

J. Mol. Sci. 2015, 16 (8), 17101–17159.

(2) Farrán, A.; Cai, C.; Sandoval, M.; Xu, Y.; Liu, J.; Hernáiz, M. J.; Linhardt, R. J. Green

Solvents in Carbohydrate Chemistry: From Raw Materials to Fine Chemicals. Chem. Rev.

2015, 115 (14), 6811–6853.

(3) Isikgor, F. H.; Becer, C. R. Lignocellulosic Biomass: A Sustainable Platform for the

Production of Bio-Based Chemicals and Polymers. Polym. Chem. 2015, 6 (25), 4497–

4559.

(4) European Commission. Innovating for Sustainable Growth: A Bioeconomy for Europe;

2012.

(5) The White House. National Bioeconomy Blueprint; Washington, 2012.

(6) Werpy, T.; Petersen, G. Top Value Added Chemicals from Biomass. U.S. Dep. energy

2004, 1, 76.

(7) Quispe, C. A. G.; Coronado, C. J. R.; Carvalho, J. A. Glycerol: Production, Consumption,

Prices, Characterization and New Trends in Combustion. Renew. Sustain. Energy Rev.

2013, 27, 475–493.

(8) BP. BP Statistical Review of World Energy; UK, 2015.

(9) Esteban, J.; Ladero, M.; García-Ochoa, F. Kinetic Modelling of the Solventless Synthesis

24

of Solketal with a Sulphonic Ion Exchange Resin. Chem. Eng. J. 2015, 269, 194–202.

(10) Esteban, J.; Fuente, E.; Gonzalez-Miquel, M.; Blanco, A.; Ladero, M.; Garcia-Ochoa, F.

Sustainable Joint Solventless Coproduction of Glycerol Carbonate and Ethylene Glycol

via Thermal Transesterification of Glycerol. RSC Adv. 2014, 4, 53206–53215.

(11) López-Porfiri, P.; Brennecke, J. F.; Gonzalez-Miquel, M. Excess Molar Enthalpies of

Deep Eutectic Solvents (DESs) Composed of Quaternary Ammonium Salts and Glycerol

or Ethylene Glycol. J. Chem. Eng. Data 2016, acs.jced.6b00608.

(12) Dasari, M. A.; Kiatsimkul, P. P.; Sutterlin, W. R.; Suppes, G. J. Low-Pressure

Hydrogenolysis of Glycerol to Propylene Glycol. Appl. Catal. A Gen. 2005, 281 (1–2),

225–231.

(13) Merchant Research & Consulting ltd. World Propylene Glycol Market to Reach Supply-

Demand Balance in 2015 https://mcgroup.co.uk/news/20140418/propylene-glycol-market-

reach-supplydemand-balance-2015.html (accessed May 20, 2016).

(14) McKetta, J. J.; Cunningham, W. A. Propylene Glycol. Encyclopedia of Chemical

Processing and Design: Volume 45; CRC Press, 1993; Vol. 45, p 568.

(15) McKetta, J. J.; Cunningham, W. A. Propylene Oxide. Encyclopedia of Chemical

Processing and Design: Volume 45; CRC Press, 1993; p 589.

(16) Gong, J.; You, F. Value-Added Chemicals from Microalgae: Greener, More Economical,

or Both? ACS Sustain. Chem. Eng. 2015, 3 (1), 82–96.

(17) Adom, F.; Dunn, J. B.; Han, J.; Sather, N. Life-Cycle Fossil Energy Consumption and

Greenhouse Gas Emissions of Bioderived Chemicals and Their Conventional

Counterparts. Environ. Sci. Technol. 2014, 48 (24), 14624–14631.

(18) Zhou, Z.; Li, X.; Zeng, T.; Hong, W.; Cheng, Z.; Yuan, W. Kinetics of Hydrogenolysis of

Glycerol to Propylene Glycol over Cu-ZnO-Al2O3 Catalysts. Chinese J. Chem. Eng.

2010, 18 (3), 384–390.

25

(19) Wang, Y.; Zhou, J.; Guo, X. Catalytic Hydrogenolysis of Glycerol to Propanediols: A

Review. RSC Adv. 2015, 5 (91), 74611–74628.

(20) Maglinao, R.; He, B. Verification of Propylene Glycol Preparation from Glycerol via the

Acetol Pathway by in Situ Hydrogenolysis. Biofuels 2012, 3, 675–682.

(21) Akiyama, M.; Sato, S.; Takahashi, R.; Inui, K.; Yokota, M. Dehydration-Hydrogenation

of Glycerol into 1,2-Propanediol at Ambient Hydrogen Pressure. Appl. Catal. A Gen.

2009, 371 (1–2), 60–66.

(22) Chen, H.; Wang, Q.; Zhang, X.; Wang, L. Applied Catalysis B : Environmental

Quantitative Conversion of Triglycerides to Hydrocarbons over Hierarchical ZSM-5

Catalyst. "Applied Catal. B, Environ. 2015, 166–167, 327–334.

(23) Martin, A.; Armbruster, U.; Gandarias, I.; Arias, P. L. Glycerol Hydrogenolysis into

Propanediols Using in Situ Generated Hydrogen - A Critical Review. Eur. J. Lipid Sci.

Technol. 2013, 115 (1), 9–27.

(24) Seretis, A.; Tsiakaras, P. Hydrogenolysis of Glycerol to Propylene Glycol by in Situ

Produced Hydrogen from Aqueous Phase Reforming of Glycerol over SiO2-Al2O3

Supported Nickel Catalyst. Fuel Process. Technol. 2016, 142, 135–146.

(25) Maglinao, R. L.; He, B. B. Catalytic Thermochemical Conversion of Glycerol to Simple

and Polyhydric Alcohols Using Raney Nickel Catalyst. Ind. Eng. Chem. Res. 2011, 50

(10), 6028–6033.

(26) Dieuzeide, M. L.; Jobbagy, M.; Amadeo, N. Vapor-Phase Hydrogenolysis of Glycerol to

1,2-Propanediol over Cu/Al2O3 Catalyst at Ambient Hydrogen Pressure. Ind. Eng. Chem.

Res. 2016, 55 (9), 2527–2533.

(27) Vasiliadou, E. S.; Lemonidou, A. A. Kinetic Study of Liquid-Phase Glycerol

Hydrogenolysis over Cu/SiO2 Catalyst. Chem. Eng. J. 2013, 231, 103–112.

(28) Harisekhar, M.; Pavan Kumar, V.; Shanthi Priya, S.; Chary, K. V. R. Vapour Phase

Hydrogenolysis of Glycerol to Propanediols over Cu/SBA-15 Catalysts. J. Chem. Technol.

26

Biotechnol. 2015, 90 (10), 1906–1917.

(29) Julián-Durán, L. M.; Ortiz-Espinoza, A. P.; El-Halwagi, M. M.; Jiménez-Gutiérrez, A.

Techno-Economic Assessment and Environmental Impact of Shale Gas Alternatives to

Methanol. ACS Sustain. Chem. Eng. 2014, 2 (10), 2338–2344.

(30) Gao, J.; You, F. Shale Gas Supply Chain Design and Operations toward Better Economic

and Life Cycle Environmental Performance: MINLP Model and Global Optimization

Algorithm. ACS Sustain. Chem. Eng. 2015, 3 (7), 1282–1291.

(31) Orfield, N. D.; Fang, A. J.; Valdez, P. J.; Nelson, M. C.; Savage, P. E.; Lin, X. N.;

Keoleian, G. A. Life Cycle Design of an Algal Biorefinery Featuring Hydrothermal

Liquefaction: Effect of Reaction Conditions and an Alternative Pathway Including

Microbial Regrowth. ACS Sustain. Chem. Eng. 2014, 2 (4), 867–874.

(32) Kumar, S.; Lange, J.-P.; Rossum, G. V.; Kersten, S. R. A. Liquefaction of Lignocellulose

in Fractionated Light Bio-Oil: Proof of Concept and Techno-Economic Assessment. ACS

Sustain. Chem. Eng. 2015, 3 (9), 2271–2280.

(33) Biddy, M. J.; Davis, R.; Humbird, D.; Tao, L.; Dowe, N.; Guarnieri, M. T.; Linger, J. G.;

Karp, E. M.; Salvachúa, D.; Vardon, D. R.; Beckham, G. T. The Techno-Economic Basis

for Coproduct Manufacturing To Enable Hydrocarbon Fuel Production from

Lignocellulosic Biomass. ACS Sustain. Chem. Eng. 2016, 4 (6), 3196–3211.

(34) Ehlinger, V. M.; Gabriel, K. J.; Noureldin, M. M. B.; El-Halwagi, M. M. Process Design

and Integration of Shale Gas to Methanol. ACS Sustain. Chem. Eng. 2014, 2 (1), 30–37.

(35) Azapagic, A. Life Cycle Assessment and Its Application to Process Selection, Design and

Optimisation. Chem. Eng. J. 1999, 73 (1), 1–21.

(36) Sabio, N.; Pozo, C.; Guillén-Gosálbez, G.; Jiménez, L.; Karuppiah, R.; Vasudevan, V.;

Sawaya, N.; Farrell, J. T. Multiobjective Optimization under Uncertainty of the Economic

and Life-Cycle Environmental Performance of Industrial Processes. AIChE J. 2014, 60

(6), 2098–2121.

27

(37) Kostin, A.; Guillén-Gosálbez, G.; Mele, F. D.; Jiménez, L. Identifying Key Life Cycle

Assessment Metrics in the Multiobjective Design of Bioethanol Supply Chains Using a

Rigorous Mixed-Integer Linear Programming Approach. Ind. Eng. Chem. Res. 2012, 51

(14), 5282–5291.

(38) Guillén-Gosálbez, G.; Grossmann, I. E. Optimal Design and Planning of Sustainable

Chemical Supply Chains Under Uncertainty. AIChE J. 2008, 55 (1), 99–121.

(39) Douglas, T.; Denny, N.; Tan, R. Robust Optimization Approach for Synthesis of

Integrated Biorefineries with Supply and Demand Uncertainties. Environ. Prog. Sustain.

Energy 2013, 32 (2), 384–389.

(40) Kasivisvanathan, H.; Ubando, A. T.; Ng, D. K. S.; Tan, R. R. Robust Optimization for

Process Synthesis and Design of Multifunctional Energy Systems with Uncertainties. Ind.

Eng. Chem. Res. 2014, 53, 3196–3209.

(41) Cai, Y.; Yue, W.; Xu, L.; Yang, Z.; Rong, Q. Sustainable Urban Water Resources

Management Considering Life-Cycle Environmental Impacts of Water Utilization under

Uncertainty. Resour. Conserv. Recycl. 2016, 108, 21–40.

(42) Posada, J. A.; Rincón, L. E.; Cardona, C. A. Design and Analysis of Biorefineries Based

on Raw Glycerol: Addressing the Glycerol Problem. Bioresour. Technol. 2012, 111, 282–

293.

(43) Xiao, Y.; Xiao, G.; Varma, A. A Universal Procedure for Crude Glycerol Purification

from Different Feedstocks in Biodiesel Production: Experimental and Simulation Study.

Ind. Eng. Chem. Res. 2013, 52, 14291--14296.

(44) Pudi, S. M.; Biswas, P.; Kumar, S. Selective Hydrogenolysis of Glycerol to 1,2-

Propanediol over Highly Active Copper-Magnesia Catalysts: Reaction Parameter, Catalyst

Stability and Mechanism Study. J. Chem. Technol. Biotechnol. 2015, 91 (August), 2063–

2075.

(45) Wołosiak-Hnat, A.; Milchert, E.; Lewandowski, G. Optimization of Hydrogenolysis of

Glycerol to 1,2-Propanediol. Org. Process Res. Dev. 2013, 17 (4), 701–713.

28

(46) Seider, W. D.; Seader, J. D.; Lewin, D. R.; Widagdo, S. Product and Process Design

Principles : Synthesis, Analysis, and Evaluation, 3rd ed.; John Wiley & Sons: US, 2009.

(47) Yee, T. F.; Grossmann, I. E. Simultanueous Optimization Models for Heat Integration - II.

Heat Exchanger Network Syntheis. Comput. Chem. Eng. 1990, 14 (10), 1165–1184.

(48) Carlson, E. C. Don’t Gamble with Physical Properties for Simulations. Chem. Eng. Prog.

1996, 92 (10), 35–46.

(49) Guthrie, K. M. Capital Cost Estimating, Second.; Elsevier Ltd., 1969.

(50) Towler, G. P.; Sinnott, R. K. Chemical Engineering Design : Principles, Practice, and

Economics of Plant and Process Design; Butterworth-Heinemann, 2013.

(51) Wernet, G.; Bauer, C.; Steubing, B.; Reinhard, J.; Moreno-Ruiz, E.; Weidema, B. The

ecoinvent database version 3 (part I): overview and methodology.

http://link.springer.com/10.1007/s11367-016-1087-8 (accessed May 31, 2016).

(52) Weidema, B. P.; Wesnæs, M. S. Data Quality Management for Life Cycle Inventories—an

Example of Using Data Quality Indicators. J. Clean. Prod. 1996, 4 (3), 167–174.

(53) Law, A. M. Simulation Modeling and Analysis, 5th ed.; McGraw Hill: New York, 2015.

We design and evaluate the economic and environmental performance of three different routes

for the production of propylene glycol from biodiesel glycerol.

29

30

Figure 1. Production of PG from propylene oxide conversion (BAU).

31

Figure 2. Hydrogenolysis of glycerol at high pressure and isothermal conditions with external

hydrogen (GB-1).

32

Figure 3. Hydrogenolysis of glycerol at ambient pressure and non-isothermal conditions with

external hydrogen (GB-2).

33

Figure 4. Isothermal hydrogenolysis of glycerol at high pressure with in situ generated hydrogen

(GB-3).

34

Figure 5. Contribution to the total annualized cost and economic potential per kg of PG

generated.

35

Figure 6. Environmental life cycle assessment results for the proposed alternatives. [Impacts

expressed per kg of PG. AP: Acidification potential; GWP: Global warming potential; DAR:

Depletion of abiotic resources; FAET: Fresh aquatic ecotoxicity; MAET: Marine aquatic

ecotoxicity; TE: Terrestrial ecotoxicity; EP: Eutrophication potential; HT: Human toxicity; OLD:

Ozone layer depletion; PO: Photochemical oxidation].

36

Figure 7. Total annualized cost and economic potential per kg of PG generated under

uncertainty.

37

Figure 8. Environmental life cycle assessment results under uncertainty for the proposed

alternatives. Impacts expressed per kg of PG.

38

TOC

39