<<

Assessment of Selected Aspects of Biodiesel

Production: and Waste Conservation

A thesis submitted to the

School of Energy, Environmental, Biological and Medical Engineering

Division of Graduate Studies

University of Cincinnati

In partial fulfillment of the requirement for the degree of

MASTER OF SCIENCE

2012

By

Qingshi Tu

Bachelor of Engineering, Environmental Engineering

University of Shanghai for Science and Technology, China, 2008

Committee:

Mingming Lu, PhD (Chair)

Drew Mcavoy, PhD

Joo-Youp Lee, PhD

Y. Jeffery Yang, PhD

i

Abstract

Biodiesel has been in commercial use for more than a decade with several known benefits: reducing the nation’s reliance on petroleum import, significant reduction in the emission of air pollutants and green house gases (GHGs), and comparable fuel properties to the petroleum diesel.

However, expansion of the biodiesel industry has also resulted in some concerns. As an example, the “food vs. fuel” debate reflects the competition of biodiesel with food supply when edible seed-oils, such as soybean oil, are used for more profitable production. In order for the biodiesel industry to continue thriving in the future, various governments and organizations have imposed and suggested an array of sustainability factors for biodiesel supply chain, such as land use, water consumption, , cost and availability of feedstocks, etc.

This study addressed two major sustainability aspects regarding biodiesel production: (1) water consumption, and (2) utilization of waste materials. In detail, the following topics were investigated: (1) characterization of water consumption by soybean-derived biodiesel in plant growth and fuel production; (2) parametric study of reducing FFA (free fatty ) in waste cooking oil; and (3) a preliminary evaluation on the utilization of waste coffee ground as biodiesel feedstock and purification material.

Water consumption from biodiesel process was characterized as three stages: plant growth, soybean processing and biodiesel production. Result showed that the nationwide average irrigation accounted for 61.78 gallons of water per gallon of soybean biodiesel while soybean processing (0.17 gal/gal) and biodiesel production (0.36 gal/gal) stages consumed much less. A

ii state-by-state analysis for irrigation water indicated that the water consumption was highly dependent on the location and climate. Overall, on a nationwide basis, the total water consumption for making biodiesel from soybean was approximately 808.7 million gallons water per year.

In general, feedstock can account for up to 80% of the total cost for biodiesel production. This offers potentials for low cost and even waste materials, such as animal fats, waste cooking oil

(WCO), and trap grease (brown grease). However, the high FFA content (>1 wt%) in these waste materials requires pretreatment prior to transesterification . Therefore, a parametric study on FFA reduction in WCO was performed to study the optimum conditions for FFA pretreatment.

WCO with FFA level of 5±0.5 wt% was treated by acid catalytic esterificiation using (H2SO4). The influence of temperature, methanol-to-FFA molar ratio, and catalyst concentration on the conversion rate was investigated. Results indicated that the optimal condition was 60±5°C, 40:1 methanol-to-FFA molar ratio, and 12.5 wt% H2SO4.

Thirdly, a preliminary study was performed to investigate the feasibility of using waste coffee grounds (WCG) as both an oil source and purification material for biodiesel production. Results showed that the oil content of WCG was around 10 wt%. In addition, the post-extraction WCGs were found to be effective in removing impurities from crude biodiesel, such as free glycerin, methanol and metal . Results suggested that WCG may be comparable in purification capability to commercial materials. The use of waste as feedstock and purification material can greatly promote the sustainability of biodiesel production by lowering overall production cost,

iii reducing waste generation (less/no additional purification material needed) and minimizing life cycle environmental impact (/reusing wastes in each stage of the production life cycle).

iv

v

Acknowledgement

Sincere gratitude is extended to my adviser, Dr. Mingming Lu, for her guidance throughout my entire period of graduate study. Meanwhile, thanks are given to the other committee members,

Dr. Y. Jeffery Yang, Dr. Drew C. McAvoy and Dr. Joo-Youp Lee, for their advices and suggestion for the thesis.

Many thanks also go to US EPA for providing financial support for part of the study. Special thanks are given to Mr. Ming Chai for his instruction and help with the experimental work.

Cincinnati Zoo & Botanical Garden, Central Utility Plant (UC) and Bluegrass Biodiesel® are acknowledged for their in-kind support to this research work.

I would also like to say thank you to my colleagues, Lei Cheng, Jarod Gregory, Jiangchuan Hu,

Tongyan Li, Shuang Liang, Mark Schutte, Jingjing Wang, and Thiansathit Worrarat.

Last but not least, I would like to thank my family for their love and support for me to accomplish the degree.

vi

Disclaimer

The names of the brand and company mentioned in this thesis are only provided as the examples of the relevant industries and thereof do not reflect any preference or recommendation.

The U.S. Environmental Protection Agency, through its Office of Research and Development, funded and managed, or partially funded and collaborated in, the research described herein. It has been subjected to the Agency administrative review and has been approved for external publication. Any opinions expressed in this paper are those of the author(s) and do not necessarily reflect the views of the Agency, therefore, no official endorsement should be inferred.

Any mention of trade names or commercial products does not constitute endorsement or recommendation for use.

vii

Table of Contents

Abstract ii Acknowledgement vi Disclaimer vii List of Figures ix List of Tables x Chapter 1 Problem Statement and Motivation 1 1.1 Sustainability Criteria for Biodiesel 1 1.2 Focus of the Study 6 1.3 Structure of Thesis 9 Chapter 2 Literature Review 11 2.1 Current Status of Biodiesel Industry in the US 11 2.2 Water Consumption for Biodiesel Production Stages 13 2.3 Effect of Free Fatty Acid (FFA) in Biodiesel Manufacturing and Its Control 28 2.4 Utilization of Waste Coffee Grounds (WCGs) for Biodiesel Production 31 Chapter 3 Influence of Biodiesel Production on Water Resources 44 3.1 Goal and Scope 44 3.2 Methodology 46 3.3 Results and Discussion 51 Chapter 4 Parameteric Study on Esterification of FFA in Waste Cooking Oil by 82 H2SO4 4.1 Goal and Scope 82 4.2 Methodology 82 4.3 Results and Discussion 88 Chapter 5 Utilization of WCGs for Biodiesel Production 94 5.1 Goal and Scope 94 5.2 Methodology 94 5.3 Results and Discussion 96 Chapter 6 Conclusions and Future Work 107 Conclusions 107 Future Work 110 References 113 Appendix 132

viii

List of Figures

Figure 2.1 Annual biodiesel production in the US (1999~2011) 11 Figure 2.2 Biodiesel production by feedstock type (2007-2011) 12 Figure 2.3 Transesterification reaction 23 Figure 2.4 Schematic of biodiesel production 23 Figure 2.5 Hydrolysis of fat, oil and grease 28 Figure 3.1 Schematic of water consumption stages of biodiesel production from 45 soybean Figure 3.2 Normalized irrigation water intensity (Wni) for 35 states 53 Figure 3.3 Annual total irrigation water consumption (W1) for 35 states 53 Figure 3.4 Ratio between irrigation methods in top 5 states with highest irrigation 55 areas Figure 3.5 Profile of Irrigation Methods in Top 5 States (Market Share) 55 Figure 3.6 Profile of Irrigation Water Sources in Top 5 States (Capacity) 56 Figure 3.7 Profile of Irrigation Water Sources in Top 5 States (Market Share) 56 Figure 3.8 Annual soybean processing water consumption for 38 states (W2) 58 Figure 3.9 Annual water consumption during biodiesel manufacturing stage (W3) for 63 46 states Figure 3.10 Total annual water consumption in 49 states (Wtot) 66 Figure 3.11 Normalized total water consumption for 46 biodiesel production states 66 (Wnt) Figure 4.1 Experiment setup 82 Figure 4.2 Effect of temperature on the FFA reduction (methanol-to-FFA molar ratio: 89 50:1, H2SO4 concentration: 10 wt%) Figure 4.3 Effect of temperature on the FFA reduction (methanol-to-FFA molar ratio: 89 40:1, H2SO4 concentration: 10 wt%) Figure 4.4 Effect of temperature on the FFA reduction (methanol-to-FFA molar ratio: 90 30:1, H2SO4 concentration: 10 wt%) Figure 4.5 Effect of temperature on the FFA reduction (methanol-to-FFA molar ratio: 90 20:1, H2SO4 concentration: 10 wt%) Figure 4.6 Effect of acid catalyst concentration on 2-hour FFA conversion rate 91 (methanol–to-FFA molar ratio 40:1) Figure 4.7 Effect of methanol-to-FFA molar ratio on 2-hour FFA conversion rate (10 92 wt% sulfuric acid) Figure 5.1 Oil concentration in the solvent vs. extraction time (solvent: 97 hexane/isopropanol 1:1 v/v mixture)

ix

List of Tables

Table 2.1 Literature review on WF papers 16 Table 2.2 Literature review on other approaches 20 Table 2.3 Summary on activation energies of esterification reaction from different 31 studies Table 2.4 Summary of costs for selected dry wash materials 34 Table 2.5 Summary of oil level and composition from literature for coffee beans and 38 grounds Table 3.1 Major data sources for current study (Ohio) 49 Table 3.2 Water consumption during soybean processing and refining stage 57 Table 3.3 Washing water consumption data from literature review and personal 59 communications Table 3.4 Washing water consumption from industry survey in current study 60 Table 3.5 Cooling tower makeup water consumption 61 Table 3.6 Different W3 estimation scenarios 62 Table 3.7 Top 10 states with highest total annual water consumption (Wtot) 64 Table 3.8 Summary of water-stressed states from literature 67 Table 3.9 Total annual water consumption (Wtot) for the states in the water-stressed 70 areas Table 3.10 Comparison of waste water COD, BOD5 data from different refining 77 processes (Biodiesel, Bio-ethanol, and petroleum diesel) Table 3.11 COD, BOD5 data of different biodiesel products (Peterson and Moller, 77 2010) Table 4.1 Major composition of waste cooking oil samples from Cincinnati Zoo 88 Table 4.2 Activation energy (Ea) and pre-exponential factor (A) of the experiments 93 Table 5.1 Results of extraction 97 Table 5.2 Elemental analysis of selected coffee oil sample 98 Table 5.3 Summary for three purification runs 101 Table 5.4 Qualitative summary of purification effects by different materials based on 103 the preliminary results Table 5.5 Removal rates of some impurities by different purification materials 104

x

Chapter 1

Problem Statement and Motivation

1.1 Sustainability Criteria for Biodiesel

The combustion of -based fossil fuels, i.e. coal, petroleum and natural gas, is associated with issues that are adverse to both the environment and human health. Acid precipitation, environmental , climate change, increased respiratory diseases, etc. are directly related to fossil fuel emissions. Besides, as the primary energy source, its over-exploitation has caused the concern of the depletion of fossil fuels in the near future. These two factors combined, therefore, entail the development of alternative fuels. Biodiesel, as an example, has been considered an alternative for petroleum diesel, especially in transportation fuel use. Biodiesel, or referred by some, the first generation of biodiesel, is a mixture of fatty acid alky esters that are derived from renewable feedstocks such as vegetable oils, animal fats and waste oil and greases via transesterification. Biodiesel displays comparable fuel properties and significant lower emission for most of the air pollutants and green house gases (GHGs). In addition, the utilization of biodiesel helps to reduce the nation’s reliance on petroleum import, which consequently contributes to energy security proposed by Energy Independence and Security Act (EISA, 2007).

Also, since there is almost no modification needed for a current diesel engine to run on biodiesel, either in the form of blending with petroleum diesel or pure biodiesel alone, biodiesel is considered as a turn-key solution to achieving a sustainable fuel supply. Therefore, recognized as one of the most promising alternative fuels, biodiesel industry has significantly increased production in recent years.

1

Like many other innovations, expansion of the biodiesel industry has also incurred significant concerns about its sustainability. Active advocation has been proposed for the of alternative fuels so that unintended environmental consequences will be minimized, unlike the development of many other products in history. Perhaps the most well- known example of a sustainability issue regarding is the “food or fuel” debate (Canali and Aragrande, 2010; Casman and Liska, 2007; Tilman et al., 2009).

In an effort to promote the sustainable development of biodiesel, various entities have proposed regulations and suggestion. European Council (EREC, 2010) laid out the sustainability criteria for that secures the development of biofuel would not compromise the environment, society and natural resources. The criteria is the result of an initiative led by

European Commission (EC), volunteer industrial practitioners, governments and NGOs to: (1) propose a sustainable biofuel certificate system; (2) protect the natural resources; and (3) promote the biofuels that achieve significant Greenhouse Gas (GHG) reduction. Specifically, the criteria contains five items: GHG savings, , high carbon stock, cross compliance, and social sustainability. Step-wise increment in GHG reduction is set for the biofuels to be qualified as sustainable with 35% in 2010, 50% in 2017 and 60% in 2018. The calculation procedure also integrates the effect of land use change. Biodiversity prevents the disturbance of forests, grasslands and protection areas from being converted into arable land for feedstock production. Biofuels produced from land with high carbon stocks, such as and undrained peatland, will fail to meet the sustainability criteria due to the potential release of GHGs by destruction of these lands. Cross compliance requires the feedstock to meet the EU Common

Agricultural Policy, if applicable. Finally, social sustainability reflects the concern for welfare issues such as labor standards and food security. As a legislated regulation, the compliance is

2 required if the economic operator (industrial practitioners) wants to be considered as a candidate for federal/governmental support. Also, the non-compliance biofuel products would not be counted towards the fulfillment of EU targets of renewable energy, which is similar to the practice in the US.

Renewable Fuel Standard (RFS) was established in 2005 under the context of requesting the obligated parties, petroleum refiners and importers to take the responsibility to promote the utilization of renewable fuels (McMartin and Noyes, 2010). In RFS, ethanol made from corn starch was the fuel for compliance and the measurement of the obligated volume was calculated on the basis of the gasoline sales. In 2007, revisions (RFS 2) were made to integrate the sustainability factors, including: (1) the type of fuel for compliance was expanded from corn ethanol alone to four different biofuel products of their petroleum counterparts (biomass-based diesel, cellulosic biofuel, non-cellulosic advanced fuel, and ethanol); and (2) life-cycle GHG reduction threshold was set for each type of biofuel. The resulting RFS 2 has become the guideline for achieving sustainable biofuels in the US. Un-qualified products (failing to meet the

GHG reduction threshold) will not be deemed as complying fuels and thus, may lose the competitiveness in the market (Argyopoulos, 2008).

Besides the legislative efforts, initiatives have been taken by NGOs to frame a standard for sustainable biofuels. For instance, National Biodiesel Board organized a symposium (NBB,

2008), where a score-based system was proposed by the Roundtable on Sustainable Biofuels

(RSB) to better evaluate and promote the sustainability of biodiesel. Unlike RFS 2, not only

GHG emissions were targeted in the proposal but natural resources and social concerns were also

3 taken into consideration. The subdivisions of natural resource conservation were biodiversity, health, air quality and water use. For the social concerns, specifically, food security and working conditions were emphasized. This further breakdown of the sustainability principles resembles the EU counterpart mentioned above. Western Organization of Resource Councils

(WORC, 2006) also recommended several practices for sustainable biofuel development with a focus on biomass growth. As was advocated, water resources and soil quality should be maintained during energy biomass cultivation; net energy gain and air quality improvement should be achieved; wildlife and biodiversity should not be disturbed; and, adequate income for the farmers should be guaranteed to insure the sustainability practices.

In addition to the governments/organizations, research and academic institutes are active in exploring sustainability opportunities and challenges as well. The Royal Society (2008) released a report in 2008, pointing out aspects to be considered in a life cycle assessment (LCA) for biofuel sustainability, including GHG emission, land use, soil carbon conservation, water consumption and biodiversity. In 2008, a workshop hosted by research divisions from the United

States Department of Agriculture (USDA) and Department of Energy (US DOE) called for ideas to direct the future research areas in sustainable biofuels (USDA & US DOE,

2008). The workshop addressed three dimensions of biofuel sustainability: environmental, economical and social. Environmental sustainability focused on four aspects: (1) improving the soil fertility and reducing agricultural GHG emissions; (2) understanding, predicting and managing the impact of biofuel production on water resources; (3) identifying and increasing the ecosystem services provided by biofuel feedstocks; and (4) investigating the landscape at regional scales. Economical sustainability stressed: (1) understanding the economic and

4 market impacts; (2) setting up the inventory and model for predicting the land availability and productivity; (3) developing an economic LCA for investigating the benefits of biofuel production; (4) determining decision-making support and future infrastructure requirement estimation. Social dimension included building education tools and understanding the social effects, dynamics, human choices, risk management and incentives. David Fridley (2010) wrote an article in the book of “Post Carbon Readers” and suggested nine potential challenges for the healthy growth of alternative energy. Although the article targeted alternative energy, some of the author’s views on sustainability apply to the biofuels too. To illustrate, the water consumption during the full cycle of biofuel production would be of increasing concern due to tight water resource availability. Additionally, the “energy return on investment” (EROI) is also a very important parameter to examine when assessing the sustainability of a biofuel. The improvement of EROI involves a couple of measures, such as using the alternative/waste-derived feedstocks to minimize the energy input during the feedstock growth stage, streamlining the manufacturing processes and applying energy-saving techniques. Similar to the energy input is the material requirement during the entire production cycle of the biofuel. Minimizing material use would enhance the sustainability. Franca et al. (2006) cominbed the sustainable product development (PSD) methodology with the conception of “sustainability principles” in assessing the sustainability opportunities and challenges of biofuel industry. The resulting criteria embraced four perspectives: (1) fossil fuel input; (2) pesticides and other agrochemical application; (3) biodiversity, and soil degradation; and (4) labor, income and food security.

5

1.2 Focus of the Study

As is shown in the previous section, the conception of biofuel/biodiesel sustainability is an open definition. However, a collective summary would contain at least the following significant aspects: GHG reduction, land use, water resources and quality, ecosystem conservation, economic and social impacts. Another factor that is not embodied explicitly in the major standards/proposals discussed above is the application of alternative/waste-derived feedstocks.

In effect, the choice of feedstock is the result of a comprehensive consideration of economic viability, technical feasibility and regulation compliance, which reflects sustainability concerns to some extent. Although soybean oil is still the dominant feedstock for U.S. Biodiesel industry, its market share has been decreasing due to the increase in other feedstocks, among which waste fats, oils and greases are particularly attractive because of their low/no cost, large amount and low parent environmental burden (Wiltsee, 1998). So, the utilization of alternative/waste-derived feedstock should be regarded as, at least supplementarily, one of the sustainable development criteria for biodiesel.

Due to the limitation of the time, among all the above-mentioned sustainability aspects, the scope of this study would cover two of them: water resources and quality, and utilization of alternative/waste-derived feedstock. Specifically, three topics are covered in this thesis: (1) implication of biodiesel production on water consumption and quality; (2) parametric study on FFA reduction in waste cooking oil; and (3) utilization of waste coffee grounds in biodiesel production. The selection of these topics is based on the rationale below.

Currently, soybean oil is the dominant feedstock for biodiesel manufacturing in the US.

According to the estimation from USDA (Asbridge, 2010), approximately 2 billion pounds of

6 soybean oil was consumed for biodiesel production in 2009. On the other hand, agriculture sector is a significant user of the water resource. For example, it accounted for 85% of total fresh water consumption in 2005 in the US (USGS, 2005). Since current biodiesel is primarily produced from oil-crops whose growth requires large amount of irrigation water input; plus biodiesel plants usually seek for the feedstock in the nearby domain considering the cost of transport and storage, it is rational to concern the potential impact on water resources by the development of biodiesel industry in a specific area; or conversely, water resources may be a limiting factor for the further promotion of biodiesel. Similar situation can be found in the biodiesel manufacturing stage as well, with the water used for biodiesel purification and other processes. Actually, the nexus between energy and water has been proposed already and water resources may be a restricting factor that will affect the development of biodiesel industry in water stressed areas

(EPRI, 2003; Hurd et al. 1999; Scown et al., 2011; Yang, 2010). So measures are inevitably needed to prevent these unintended outcomes. Unfortunately, the existing data regarding water consumption and waste water quality varies hugely among individual studies (to be discussed in later section). A uniform recognition of water consumption steps is absent in biodiesel manufacturing stage at the industry level. Therefore, before the control practices and regulations could be applied, there is a need to obtain a clear picture of water consumption during the entire biodiesel production life cycle. Besides the inventory of water consumption, equally important is the water quality deterioration issue. The biodiesel purification process may generate high

BOD/COD-laden waste water, depending on the technology used and once this waste water is released to the local waste plant (WWTP), the unexpected problems may occur.

Similar to the consumptive water inventory, the characterization of waste water from biodiesel

7 manufacturing stage has not been properly established. So the waste water discharge from biodiesel processing and the regulations are investigated in this study as well.

In addition to the water issue, diversification of feedstock market is also an indispensible piece of the puzzle for achieving sustainability of biodiesel manufacturing. While many companies are able to handle multiple feedstocks, the application of high free fatty acid (FFA) containing alternative/waste-derived feedstocks (>1% wt), such as waste cooking oil, still faces some challenges. FFA is a product of hydrolysis reactions between triglycerides (major component of cooking oil) and the water released from the food (Berrios et al., 2010). The presence of FFA in the triglycerides will halt the alkaline catalytic transesterfication, the most widely used technique in biodiesel production, by means of saponification with the catalyst. The result of this will not only cause the loss of catalyst and thus reduce the yield of production, but also increase the difficulty of separating glycerin from the crude biodiesel. As indicated in the literature, to achieve a desirable yield of transesterication the FFA concentration should be reduced to 1 wt%

(Liu, 1994; Freedman and Pryde, 1982) or even lower (Feuge et al., 1945; Nye and Southwell,

1984; Canakci and Gerpen, 2001). Therefore, in order to better utilize these alternative/waste- derived feedstocks, how to efficiently lower the FFA level before the conversion step is crucial.

In this study, waste cooking oil was chosen as the targeted feedstock due to its leading position in market share of the waste-derived feedstocks. Also, its lower cost enabled it to be an ideal bulk material for synthesizing the high FFA oil to simulate the other alternative/waste-derived feedstocks (e.g. trap grease with over 50 wt% FFA). This parametric study not only generated an optimal condition for the operation but also will serve as the potential input for a regression model used to estimate the yield under a given set of operational conditions (future work).

8

In addition to being used as feedstock, utilization of waste-derived materials in other stages of biodiesel production to minimize the overall environmental foot print of biodiesel production is equally preferable. By doing so, the biodiesel producer can improve the sustainability of its biodiesel product through consuming less materials and/or fuels that are substituted by the existing waste materials; and hence, less environmental impacts will occur. In this thesis, waste coffee grounds (WCG) are of special interest because of its demonstrated potential to realize the closed-loop utilization for biodiesel production since it could not only be used as an oil source but is also suitable as purification material and adjunct fuel. Existing studies indicated that coffee oil in the WCGs could be a considerable source of oil for biodiesel production. So, through optimization of the oil extraction process, it could be commercially feasible to produce biodiesel from WCG. In addition, WCG has been proven effective for removing various pollutants in existing studies (details in Chapter 2). So it’s rational to expect that WCG (after extraction) could also serve as the purification material to clean the crude biodiesel. Lastly, the spent WCGs (after oil extraction and purification) can also be used as an adjunct fuel in the biomass burner to supply energy to the biodiesel production facilities considering its high heating value. The results from current preliminary study on oil extraction and purification effectiveness will serve as the building block for accomplishing the ultimate goal of utilizing WCG in multiple stage of biodiesel production in the future.

1.3 Structure of Thesis

The main body of this thesis consists of five chapters. Chapter 2 provides background information about the selected topics in this thesis through a review of existing literature, based on which the specific areas to be explored are stated. Full details are presented in Chapter 3 for

9 characterizing water consumption in making biodiesel from soybean in the US. The study focused on three major stages of producing soybean-derived biodiesel: soybean growth, soybean processing and biodiesel manufacturing. Along with the quantification of water consumption are the summary of waste water quality from biodiesel manufacturing stage as well as the comparison among current biodiesel purification technologies. Chapter 4 contains the information about the second topic of this thesis, parametric study for FFA reduction in waste cooking oil. Chapter 5 introduces the preliminary results for the third topic, utilization of waste coffee ground for biodiesel production, and a detailed future work plan is attached at the end of this chapter to better connect the current results with future research direction. Chapter 6 serves as the conclusion for this thesis. Future work for the first two topics is also included.

10

Chapter 2

Literature Review

2.1 Current Status of Biodiesel Industry in the US

As is shown in Figure 2.1, the boom of biodiesel industry started from 2004, continued and reached the 1st peak at 2008, when the annual production was around 700 million gallons.

Although the uncertainties within federal policy implementation discouraged the stakeholders in recent years, for example the expiration of blender tax credit in 2009, the implement of

Renewable Fuel Standard 2 (RFS2) and the faith in biodiesel as the advanced green fuel still maintained the industry in a “healthy” status and in 2011, the biodiesel industry welcomed the

2nd peak with over one billion gallons production (NBB, 2012).

Figure 2.1 Annual biodiesel production in U.S. (1999~2011) (source: National Biodiesel Board)

11

Figure 2.2 Biodiesel production by feedstock type (2007-2011) (source: USDA)

The expansion of biodiesel industry has also propelled the diversification of feedstock market in recent years and the market share of soybean oil kept decreasing due to the emerging of other feedstocks (Figure 2.2). The reason behind this trend can be multiple: the desire to reduce the feedstock cost, sustainable concerns (such as food reservation, water resource protection), higher yield and improved fuel properties. Normally, the cost of feedstock takes up a major portion of the total cost for the final biodiesel product and this ratio can be as high as 80% (Haas, 2010).

Under the pressure of feedstock price increase (EIA, 2004), many biodiesel plants, especially the newly-built ones, are now able to handle multiple feedstocks, particularly the waste-derived fat, oil and greases. Also, in order not to compete with the human needs, many biodiesel producers tend to use non-crop based feedstock oils, such as jatropha oil. Although soybean is still the leading feedstock, the oils from non-crop/non-edible plants have already attracted considerable attention and relevant R&D projects are well underway. In addition, despite that current biodiesel manufacturing technology is mature enough to secure a high yield, the exploration has never been stopped for promoting the biodiesel productivity to a notch higher. Consequently,

12 looking for the feedstock that has the potential to yield more biodiesel is one important driving force for the diversification of the feedstock market. Last but not least, efforts are always put in the investigation of feedstocks that are able to address the challenges of the current biodiesel product, especially the fuel properties such as cold flow performance.

2.2 Water Consumption for Biodiesel Production Stages

2.2.1 Estimation of agriculture water consumption based on “water footprint”

Allen (1998) introduced the concept of “crop water requirement” (CWR), which is defined as the total water needed due to the evapotranspiration (ET) effect to maintain the crop in a normal growth manner during the entire life cycle of a particular crop, from planting to harvest, under a specific climate condition. By quantifying the CWR, it is possible to estimate the “water footprint” (m3/ton) based on the crop yield and production, which indicates the total water input needed for a crop product if assuming actual water use is equal to CWR. Based on this idea, the

Food and Agriculture Organization of the United Nations (FAO) developed a model,

CROPWAT, which offers not only the CWR option, but also the “irrigation schedule option” that takes into account the actual water stress condition and estimates the WF in terms of soil water balance. Also realized in the model is the separation between irrigation water consumption

(blue WF) and rainfall (green WF), which provides a better understanding about the profile of water consumption for a crop under a specific climate condition. The details of the principles and methods are given in “Water Footprint Manual” (Hoekstra et al., 2009).

As a well-established approach, water footprint has been adopted by many studies in investigating the water consumption in the agriculture sector. In this study, a literature review was performed on the papers published by “Water Footprint Network” involving soybean

13 biodiesel and related studies, and the result is listed in Table 2.1. The CROPWAT model was used in all these studies. Mekonnen and Hoekstra (2010) applied the “irrigation schedule option” while others used the CWR option. Separation between green WF and blue WF was carried out in the latest two studies (Gerbens-Leenes et al., 2009; Mekonnen and Hoekstra, 2010). From

Table 2.1, it can be found that the results are quite heterogeneous. For example, the highest value is 13, 149 gallons of water per gallon of biodiesel (gal/gal) (Gerbens-Leenes et al., 2008) while the lowest value is only 491 gal/gal (Mekonnen and Hoekstra, 2010). The huge variation can be caused by a couple of factors:

(1) The spatial scope

The spatial scope of the study affects the outcome directly. For example, both looking at the blue

WF, the US average is 491 gal/gal (Mekonnen and Hoekstra, 2010) which is substantially lower than 7,521 gal/gal, the global average (Gerbens-Leenes et al., 2009). Similarly, Gerbens-Leenes et al. (2008) calculated the WF based on data sources of Iowa instead of the nationwide average

(Chapagain and Hoekstra, 2004; Hoekstra and Chapagain, 2007), which may explain, at least partially, the difference in the final results.

(2) Blue WF vs total WF

Blue WF specifically reflects the irrigation water consumption and therefore is more informative to our study. The percentage of blue WF in total WF is determined by the climate condition.

(3) Inconsistency in data period

Since the CROPWAT model relies heavily on the data input, the inconsistency in temporal scope of data sources will influence the outcome too. As is shown in Table 2.1, the data period of the individual study varies significantly between each other. For instance, climate data was averaged over 1961-1990 (Chapagain and Hoekstra, 2004), 1996-2002 (Mekonnen and Hoekstra, 2010)

14 and 2000 (Gerbens-Leenes et al., 2008; Gerbens-Leenes et al., 2009) respectively in different studies. Similarly, the yield data of these studies also involved three different periods as well.

4) Difference in assumptions

Besides the above-mentioned factors, difference in the underlying assumptions also to variation. To illustrate, Mekonnen and Hoekstra (2010) applied “irrigation schedule option”, which involved the water stress factor (Ks) and thus yielded a lower number in blue WF.

Likewise, Gerbens-Leenes et al. (2008, 2009) took into account the conception of harvest index, which is the ratio of crop yield to the total biomass yield (e.g. stem, husk), and thus potentially increased the inconsistency of the final results by introducing more parameters. Other factors that were not clearly indicated in all of the studies, such as planting and harvesting period, can be accountable for the variation too, since, for example, the planting and harvesting dates directly impact the value of crop coefficient “Kc” (Chapagain and Hoekstra, 2004; Hoekstra et al., 2009).

Besides the inconsistency caused by the above-mentioned factors, the CROPWAT model itself is inherently not that suitable for estimating the WF for biodiesel. The WF of a specific product is evaluated on the basis of its price in the CROPWAT model (Hoekstra et al., 2009), however, there is no inventory for the biodiesel price in the model database, which means the allocation could not be performed by the built-in method, let alone the volatile nature of the price-based allocation. In fact, the commonly accepted method in these papers is to allocate the WF of soybean dry mass towards the final biodiesel production alone (Gerbens-Leenes et al., 2009;

Mekonnen and Hoekstra, 2010) assuming 1:1 conversion of soybean oil to biodiesel. This approach overestimates the WF of soybean-derived biodiesel by neglecting the following facts: 1) only part of the soybean oil is used for biodiesel production every year; and 2) the byproduct, glycerin, also shares WF.

15

Table 2.1 Literature review on WF papers WF for Allocation for Allocation for WF for soybean oil Mass allocation Biodiesel Separated blue WF Study Method Model Scope Data Period soybean soybean oil BioD (m3/ton) alone (gal/gal) for soybean (m3/ton) 1961- Climate 1990 Chapagain and 1997- V(f)=0.34 Dry soybean CWR option CROPWAT USA Yield 1869 3530 1869 10,043 No Hoekstra (2004) 2001 P(f)=0.18 WF/BioD 1997- Kc 2001 NA Hoekstra and CWR option Parameters not 1997- (assuming Dry soybean CROPWAT USA 1869 3530 1869 10,043 No CHapagain (2007) specified 2001 V(f)=0.34 WF/BioD P(f)=0.18) HI(c) 1991 NA Gerbens-Leenes et al. CWR option Climate 2000 (assuming Dry soybean CROPWAT USA 2447 4622 2447 13,149 No (2008) Yield 2005 V(f)=0.34 WF/BioD Kc NA P(f)=0.18) HI(c) 1991 Climate 2000 Gerbens-Leenes et al. CWR option 1997- NA Dry soybean CROPWAT 4.3 Global Yield NA NA NA 7,521 Yes (2009) 2001 WF/BioD 1997- Kc 2001 1996- Climate 2002 1996- Yield Irrigation 1. Grid-based water 2005 V(f)=0.34 Mekonnen and Dry soybean schedule option balance USA Irrigated Fraction 2000 P(f)=0.18 92 177 92 491 Yes Hoekstra (2010) WF/BioD 2. CROPWAT 8.0 1997- Kc 2001 Ky 1979 Ks NA

16

2.2.2 Estimation of agriculture water consumption based on other approaches

In addition to the WF based studies, other researches using different functional units were also reviewed and the results are summarized in Table 2.2. Considering the scope of our study, the papers included were all focused on US alone. As compared with WF studies, the studies in

Table 2.2 are much less transparent in terms of methodology and data sources, especially the latter. This adds to the difficulty in the effort of laying down a comparison between individual studies. King and Webber (2008) performed an life cycle inventory (LCI) study on water intensity of a variety of transportation fuels. The system boundary of the study started with soybean growth and ended with usage of biodiesel in a light duty vehicle (LDV). Three water consumption steps were neglected, which were: transportation of feedstock, transportation of biodiesel, and manufacture and installation of physical capital. For the irrigation step, the data matrix consisted of irrigation data from “USDA 2003 Farm and Ranch Survey”, irrigation loss data from a USGS report (Solley et al., 1998), and allocation factors from Pradhan et al. (2008).

Three irrigation scenarios were considered. In the low irrigation scenario, data of New Jersey was used; i.e. 0.3 acre-ft/acre/yr as irrigation intensity, 41 bushels/acre soybean harvest, 37.3% irrigation water loss. In the high irrigation scenario, data of Texas was adopted; namely, 0.9 acre- ft/acre/yr as irrigation intensity, 33 bushels/acre soybean harvest, 91.9% irrigation water loss.

The third scenario was US average, where the irrigation intensity was 0.8 acre-ft/acre/yr, 48 bushels/acre was the nationwide averaged harvest and the loss of irrigation water was 79.7%. In addition, two allocation factors were applied to simulate the low and high allocation scenarios

(0.18 vs 0.8). In addition to this direct water input, the indirect water involved in the soybean growth stage was that associated with the generation of electricity and fossil fuels, while water consumption related to fertilizer production was not included. In the soybean processing and

17 biodiesel production stage, water consumption was calculated by using the data from Sheehan et al. (1998). The authors also assumed a gallon of soybean oil could produce approximately equal volume of biodiesel. After allocation, the author obtained a range for the water intensity of biodiesel produced from irrigated soybean, which was 0.6-24 gal H2O /mile, based on a mpg of

25.7 for the LDV. On the other hand, if the soybean was not irrigated, the value would be much lower, 0.01-0.02 gal H2O/mile. Similarly, Harto et al. (2010) also studied water consumption for soybean biodiesel from life cycle perspective. In their study, the water consumption for transportation, infrastructure installation, and vehicle manufacture were all accounted for. Also, in soybean growth stage, the indirect water consumption associated with fertilizer manufacture was considered. The irrigation water in this study was cited from the report titled “Energy demands on water resources” by US DOE, and a nationwide average of 6,200 gallons per bushel was used. More specifically, the study divided the soybean producing into low-cost, mid-cost and high-cost categories according to UDSA (20021); and the corresponding averaged soybean yield was 43.6 bushels/acre. Also, the irrigation percentage (the part of soybean that was irrigated) was taken from USDA (20022) and the averaged irrigation percentage was calculated as 4%. The results showed that direct water input consumption for irrigation was around 239 gal

H2O/gal biodiesel (before allocation) on a nationwide averaged basis. Couple with indirect water consumption, the total irrigation water consumption was 262 gal/gal after allocation in their study. However, this caused confusion because in the supporting material of Harto et al. (2010) it had been assumed upfront that the biodiesel yield from soybean was one gallon per bushel of soybean and based on that, the average of 6,200 gallons per bushel could have been directly converted into 6,200 gal H2O/gal biodiesel, which is much higher than the calculated 239 gal/gal.

So these results are self-conflicting as indicated from the supporting material of Harto et al.

18

(2010). For other stages, Unlike King and Webber (2008), soybean crushing related water consumption was not included and biodiesel manufacturing stage was estimated to consume one gallon water per gallon of biodiesel produced. The allocation factors in Harto et al. (2010) were also cited from Pradhan et al. (2008) and three scenarios were proposed, with 0.18, 0.50 and 0.80, respectively. After allocation, the water consumptions for soybean biodiesel were 12.3, 131, and

318 gal/gal for low, mid and high scenarios.

Notably, neither of the above two studies took into consideration of the actual consumption of soybean oil for biodiesel production. However, in fact, only a small part of soybean oil was used for biodiesel production every year (Centrec Consulting Group, LLC, 2010). The neglecting of this factor will introduce bias during allocation of water consumption for biodiesel production throughout the entire life cycle. So in our study, this issue was addressed. In addition, the temporal consistence of data bases were not well maintained in these two studies as well, especially in Harto et al. (2010) where the data for dividing soybean farm (by cost) was compiled in 1997 while the irrigation intensity was taken from the “2003 USDA Farm and Ranch

Irrigation survey”. In our study, on the other hand, the temporal difference in data bases were minimized whenever possible and most of the data were collected from reports published in 2007.

Another innovative aspect of our study is that in addition to nationwide average, we also used the state-level data to calculate the irrigation water consumption, soybean processing water consumption and water consumption during biodiesel manufacturing for each state. This reflected the water intensity of soybean-derived biodiesel in individual states, which could have been obscured by the nationwide average (details are provided in Chapter 3). Mulder et al. (2010) calculated the water consumption of biodiesel based on its market value. The price allocation was performed among soybean meal, biodiesel and raw glycerin. The irrigation water data in

19 their case was cited from Sheehan et al. (1998) which was based on the selected data source from

1994 Farm and Ranch Irrigation Survey. The prices of the three components, on the other hand, were averaged over five-year period (1999-2003) according to the data from USDA and USDOE.

Although both focusing on the irrigated soybean, King and Webber (2010)’s study yielded a lower number than this case, which could be caused by the difference in the data period and allocation method. De Vries et al. (2010) cited the water consumption data from Chapagain and

Hoekstra (2004), which is a total WF rather than blue WF alone, and assessed the water consumption based on the net energy yield (GJ/ha) of soybean biodiesel. For Pate et al. (2008) and the article from Science Magazine, due to the limited information, no further comment and comparison could be made.

Table 2.2 Literature review on other approaches Result Study Model/Method Irrigation Data Period (gal/gal)

King and Webber USDA Irrigation survey NA 2003 205.6 (2008)a (selected states)

Hybrid USDA Irrigation survey Harto et al. (2010) 2003 262 allocation (selected states)

WF of soybean= 1869 m3/ton De Vries et al. (2010) NA 2004 19,000 Chapagain and Hoekstra (2004) Mulder et al. (2010) Price allocation Sheehan et al. (1998)c 1994 762 Pate et al. (2007) NA NA NA 6,500 135,926- Science Magazine NA NA NA 272,963 a LCI, including farming, processing and manufacturing, and end use; functional unit: “gal H2O/mile”; bAll data is converted into “gal/gal”; the high heat value (HHV) of biodiesel is assumed to be 0.133 GJ/gal or 0.035 GJ/L (equal to 126,206 Btu/gal for B100; source: King and Webber 2008) cOnly 9 states are involved: AR, GA, IL, KS, MN, MS, MO, NE, SD; USDA, 1994 Farm and Ranch Irrigation Survey

20

Given the huge variations within the existing literature, it is necessary to conduct a systematic inventory study based on a consistent method and database for irrigation water consumption for soybean biodiesel in the US.

2.2.3 Water consumption in soybean processing stage

The soybean processing stage consists of two major steps: oil extraction and refining. The extraction step is performed via solvent extraction and the source of water consumption is the makeup water for the solvent recovery process. On the other hand, oil refining step typically contains four sub-steps: degumming, FFA removal, bleaching, and deodorization. For soybean oil, depending on the quality, usually, only degumming and FFA removal are performed.

2.2.3.1 Degumming

Degumming helps remove the phospholipids in the feedstock oils. Phospholipids are inherent in some oils, such as crude soybean oil, and they are the strong emulsifiers. Emulsion is detrimental to biodiesel production because it causes difficulty in the separation of crude biodiesel from glycerin and thus reduces the product yield. Phospholipids can be subdivided into two categories: hydratable and non-hydratable. For the hydratable phospholipids, as its name indicates, water washing is the common removal method. Specifically, 1%-3% of water is blended with oil and the whole mixture is mechanically agitated at 70 ºC for 30-60 minutes. After agitation, through gravity settling, or centrifugation, the hydrated phospholipids and gums can be separated from the oil. For the non-hydratable portion, the first step is to convert them into hydratable phospholipids, which can be achieved by means of acid-degumming with citric or phosphoric acid. Commonly, 0.05-0.2wt % of concentrated acid is needed and the temperature is around 60-85 ºC. The contact time is quite short, varying from several seconds to 1-2 minutes depending on the type and quality of the oil (Van Gerpen et al., 2004).

21

2.2.3.2 Removal of FFA

FFA removal is the step following the degumming treatment. The level of FFA varies among individual feedstock oils, depending on their quality. For soybean oil, typically, caustic refining is applied which uses alkaline chemical to react with FFA to form soap. The resulting soap is then washed away by water.

2.2.3.3 Bleaching

The purpose of bleaching is to reduce the darkness of the oil as well as removing some trace contaminants such as trace metals, soap, sulfur, etc. is the typical method for bleaching and the bleaching is often activated by a mineral acid, such as sulfuric acid. The temperature zone for adsorption is 90-120 °C and the retention time is between 10 to 30 minutes.

Another benefit of bleaching is the removal of moisture left by the previous steps and thus reduces its impact on the biodiesel conversion step (Van Gerpen et al., 2004).

2.2.3.4 Deodorization

Deodorization aims to reduce the unpleasant smell from the feedstock, which is especially important for the waste-derived feedstock. Deodorization is essentially a process that utilizes the fact that the odorous components usually have lower point and consequently are easier to be distillated. The temperature for deodorization is quite high (200-260 ºC) and in case of oxidation, oxygen is excluded beforehand (Van Gerpen et al., 2004).

2.2.4 Water consumption in biodiesel production stage

After processing, the refined oil is sent to the main reactor for biodiesel production. Alkaline catalytic transesterification is the typical practice for commercial biodiesel manufacturing considering its high reaction rate, high product yield and low temperature requirement (Meher et al., 2006). Transesterification is a series of stepwise reactions that breakdown the triglycerides by

22 alcohol addition, such as methanol, under the acceleration of catalyst (e.g. NaOH, KOH). Figure

2.3 below shows the overall chemical reaction (Van Gerpen, 2005) and the typical biodiesel production procedure is illustrated in Figure 2.4.

Figure 2.3 Transesterification reaction (source: Van Gerpen, 2005)

Figure 2.4 Schematic of biodiesel production (source: US Department of Energy)

23

2.2.4.1 Transesterification

Alcohol and catalyst will be pre-mixed and injected into the reactor for a better reaction rate during transesterification. In case of methanol evaporation, temperature is normally controlled around 60 °C.

2.2.4.2 Product Separation

After the transesterification, crude biodiesel will be separated from glycerin for further purification and refining. A variety of methods are available in the current industry, such as gravity settling, centrifugation and filtration.

2.2.4.3 Product Purification

After the transeserification process, the resulting biodiesel needs to be purified to make sure the fuel meets the ASTM D6751 standard. So is the glycerin, if the producer wants to make revenue from this important byproduct. Specific to biodiesel, currently there are two categories of purification methods: wet washing and dry washing. Wet washing refers to using water to dissolve the impurities, such as soap, remaining catalysts, and glycerin. The wet washing approach can be subdivided into two groups: deionized water washing and acidic water washing.

Deionized water helps to more efficiently remove the ions, such as and sodium ions, than the regular . Acidic water is usually prepared by adding slight amount of H3PO4, citric acid, HCl, or H2SO4. The acid concentration typically does not exceed 5% to avoid the risk of increased acid value in the refined biodiesel. Acidic water is particularly effective to the removal of soap and residual catalysts and thus, reduces the total amount of water needed for purification. Dry washing technologies mainly refer to using adsorption and exchange. The typical adsorbent used in current biodiesel industry is a synthetic silicate, called

Magnesol®, that is able to adsorb the polar compounds, such as methanol, glycerin and FFAs

24

(Bryan, 2005). On the other hand, the usage of ion-exchange resin is mostly found in water treatment area (e.g. water purification, ), while the purification mechanism for biodiesel is similar to that for water: when the crude biodiesel flow through the compact/fixed bed of ion-exchange resin, the ions of different impurities will react with the reactive group on the surface of the resin bead and the purification process is completed by means of exchanging target ions with the ions from the reactive group according to their own characteristics, such as acidity/alkalinity. Also, other possible working modes co-exist as well, such as filtration, adsorption and soap-glycerin interaction (Biodiesel TechNotes, 2010).

Another type of purification method is purification. This method is well-established in water purification industry but has not yet been adopted in commercial biodiesel production.

The mechanism of this purification technology is to screen the crude biodiesel in a membrane unit with specific pore size. The merit of using membrane separation is that the resulting biodiesel is free of both water and adsorbents. The performance of a membrane separation process is dependent on a variety of factors, such as membrane composition, pressure, temperature, and flow rate (Atadashi et al., 20111,2). The key parameters for the success of membrane purification are the selectivity and permeation rate of the membrane. Selectivity directly affects the effectiveness of impurity removal as well as the biodiesel recovery.

Permeation rate not only determines the purification process, but also influences the membrane durability. Today, however, the membrane purification has only been applied in the lab-scale researches and therefore is not discussed in this study.

As indicated from the above procedure, only biodiesel purification step may literally involve water consumption. However, according to a USDA’s biofuel plant modeling, there can be

25 several other categories of water consumption related to the biodiesel production process (Scott,

2010):

I) Sanitary/ drinking/ equipment washing down water consumption

II) Boiler water makeup

When there is distillation scenario during biodiesel manufacturing, the affiliated boiler water makeup should be taken into consideration. According to the estimation of USDA, the weight percentage of makeup water (consumptive water) versus total steam load is about 3~5%.

III) Cooling tower water makeup

Cooling tower can be used for the event such as methanol recovery in the biodiesel plants. If an evaporative cooling tower is under operation, in the light of USDA’s estimation, 5% wt of cooling tower water will be lost due to the evaporation. However, instead, if the close-loop cooling tower system is put into use, the water loss is expected to be much lower.

Currently for the biodiesel production stage, a uniform recognition of water consumption steps is absent at either academic or industry level. With the increasingly stressed water availability, the impact of biodiesel production on water resource, however, needs to be clarified.

2.2.5 Characterization parameters for waste water quality

For the waste water characterization, a review is presented in this section for the common characterization parameters pertaining to the waste water quality from biodiesel production, such as the conception of Biological Oxygen Demand (BOD) and Chemical Oxygen Demand (COD).

2.2.5.1 BOD and COD

Biological Oxygen Demand (BOD) measures the oxygen needed for the to decompose the organic matters under the anaerobic condition. So it is the most effective parameter to reflect

26 the biodegradability of the waste stream from a biodiesel plant since the residual biodiesel and glycerin all have a significant impact on its value. On the other hand, Chemical Oxygen Demand

(COD) measures the total amount of oxygen needed to convert all the organic matters (including the biologically available part) into CO2 and water. Therefore, COD is always larger than BOD in value. Also, COD measurement takes only a few hours which is far less time-consuming as compared with BOD. A complete determination of BOD (BODu) often is a too long measuring period (e.g. 20 days), therefore currently the most commonly adopted practice for measuring

BOD level is taking the 5-day result as the representative (BOD5). The standard methods for

COD and BOD5 detection are EPA 410.1 and EPA 405.1, respectively (Peterson and Moller,

2010).

2.2.5.2 pH pH is a measurement of the hydrogen ion concentration in the water. The biodiesel washing water may have a high pH value, particularly for the first several washing runs, due to the high level of residual soap and alkaline catalyst in it if the operation is not well controlled (Suehara et al., 2005). The allowable pH range for the municipal waste water treatment is typically from 5 to

9 because a too high or too low pH value may inhibit the activity of the bacteria in the activated (Austic and Lobdell, 2009).

2.2.5.3 Fat, oil and grease

Fat, oil and grease (FOG) is of great concern to the municipal system due to the reason that building-up of the FOG in the pipeline will to backup issues that are very costly and time consuming to control. In addition, the concentration of FOG in the waste water has a direct impact on the effectiveness of the activated sludge (Austic and Lobdell, 2009).

27

2.3 Effect of Free Fatty Acid (FFA) in Biodiesel Manufacturing and Its Control

2.3.1 Formation of FFA

The FFA in waste cooking oil is mainly the product of hydrolysis reactions between triglycerides and water during the cooking process (Figure 2.5). Cooking dehydrates the food, releasing the water into the high temperature oil. The presence of water breaks down the triglycerides into free fatty and glycerol in a stepwise manner, and the reaction is enhanced by the high temperature. Therefore, waste cooking oil from a fryer usually contains high FFA levels. Besides the hydrolysis reaction, oxidation also contributes to the breakdown of triglycerides into FFA, which is partially the reason why aged oil usually contains higher FFA level.

As is mentioned in the previous chapter, the existence of FFA during the biodiesel production process is detrimental if the amount exceeds the recommended threshold (1 wt%). So controlling the FFA level before the feedstock enters the transesterification unit calls for the attention of every biodiesel practitioner.

Figure 2.5 Hydrolysis of fat, oil and grease (source: Satyarthi et al., 2011)

2.3.2 Literature review on FFA reduction by esterfication

On the other hand, esterification is the process that converts the FFA into its corresponding fatty ester, the component of biodiesel, through reacting with alcohol. The reaction can be catalyzed by strong acids, such as sulfuric acid, and methanol is often used as the co-reactant due to its availability and low cost.

28

Esterification reaction, considering its advantage of utilizing rather than removing FFA, is often adopted as a common method of handling FFA-containing feedstocks. Canakci and Gerpen

(2001) performed a comprehensive study on both synthetic and practical greases with high FFA contents and proposed a two-step acid pretreatment for high FFA feedstock oils. The oil was synthesized from virgin soybean oil with palmitic acid and FFA levels were controlled around 20% and 40% wt respectively. In the first part of the study, the one-step esterification was applied.

The methanol-to-FFA molar ratio was fixed at 9:1, and the oil was pretreated under the catalysis by sulfuric acid with different concentration options. The result showed that for 20 wt% FFA, a large amount of alcohol and catalyst was consumed in order to reduce the final acid value down to < 2 mg KOH/g (equal to 1 wt% FFA concentration); however, for the case of 40 wt% FFA, even a 25 wt% (concentration to the FFA) of sulfuric acid was inadequate to lower the FFA level to the targeted range. Then the two-step pretreatment was tested on both synthetic and practical greases. The result showed the yellow grease (12 wt% FFA) and brown grease (33 wt% FFA) required more reaction time, higher methanol-to-FFA molar ratio than the synthetic oils, a difference that might be attributed to the compositional differences (Canakci and Gerpen, 2001).

Berrios et al. (2007) studied the optimum operating conditions and the kinetics of the esterification of FFA of sunflower oil. Methanol-to-FFA molar ratio of 60:1, 60 ºC, and 5 wt%. of sulfuric acid to FFA yielded a final acid value of less than 1 mg KOH/g oil within 120 minutes. Also, Berrios et al. (2010) published a paper on esterification of FFA in waste cooking oil with 2% FFA; and the optimal configuration of parameters was: 60:1 methanol-to-FFA molar ration, 60 ºC and 5 wt% H2SO4. The research conducted by Ramadhas et al. (2005) on high FFA rubber seed oil concluded that the optimal experimental condition was 6:1 (methanol-to-oil

29 molar ratio, approximately 40:1 for methanol-to-FFA ratio), 0.5 wt% of sulfuric acid (relative to oil, approximately 2.9 wt% to FFA) and 45±5 ºC. Thiruvengadaravi et al. (2009) performed a kinetic study on the acid-catalyzed esterification of non-edible pongamia pinnata oil and the optimal condition was determined to be 9:1 methanol-to-oil molar ratio (around 35:1 molar ratio to FFA), 1wt% sulfuric acid with respect to oil (12.5% to FFA), and 60 ºC. Ghadge and

Raheman (2005) converted mahua oil into biodiesel via a two-step process and the optimal condition for acid pretreatment was determined as 0.35 v/v methanol-to-oil ratio (about 14:1 in methanol-to-FFA molar ratio), 1% v/v sulfuric acid (roughly 10 wt% to FFA), 60 ºC and 1 hour reaction time. Berchmans and Hirata (2008) investigated the optimal condition for esterification of FFA during a two-step biodiesel production from jatropha curcas L. seed oil. The best conversion result was accomplished at the methanol-to-oil mass ratio of 0.6:1 (roughly 35:1 molar ratio to FFA), 1wt% sulfuric acid (approximately 6.7% to FFA), 50 ºC and 1 hour reaction time.

The kinetic study of esterication reaction focuses on the activation energy (Ea) of each experimental condition. Ea reflects the easiness of the reaction to proceed towards a certain direction as shown in the reaction formula. The explanation of how to derive the mathematical expression of Ea is elaborated in Chapter 4 and Table 2.3 here summarizes the researches that involved kinetics study (Aranda et al., 2008; Berrios et al., 2007, 2010; Sendzikiene et al., 2004;

Thiruvengadaravi et al., 2009). As is shown, the optimal reaction conditions may vary among individual feedstocks and oil matrix. However, the influence of the operating parameters on esterification of FFA in waste cooking oil, specifically, has not been systematically studied, and either is its kinetics, which entails the research in current study.

30

Table 2.3 Summary on activation energies of esterification reaction from different studies Initial FFA Methanol Acid Case Feedstock (mg Ea (kJ/mol) Ratio (wt %) KOH/g oil) Berrios et al. 5 50.745 Sunflower 3~5 60 (2007) 10 44.558 Berrios et al. Waste cooking 4.26 60 5 28.7 (2010) oil Synthesized oil Sendzikiene (rapeseed+oleic 22~122 NA 1 13.3 et al. (2004) acid) Methanesulfonic Methanesulfonic H SO H SO 2 4 acid 2 4 acid Aranda et al. Pure Palm fatty acids 3.2:1 0.01 0.01 63.04274 42.41118 (2008) FFA 0.03 0.03 42.12626 26.07856 0.05 0.05 27.35232 15.85915

2.4 Utilization of Waste Coffee Grounds (WCGs) for Biodiesel Production

2.4.1Feedstock supply issue for biodiesel production

With the increased concerns about the sustainability of biodiesel industry, more and more biodiesel producers start to switch from dedicated oil crops (e.g. soybean oil) to waste-derived feedstocks (e.g. waste cooking oil). However, to be commercially viable, a waste-derived feedstock should meet the following criteria:

(1) Sufficient oil content. The oil content of a waste-derived feedstock directly decides many aspects of biodiesel facilities, such as storage space, reactor volume, and processing cost. A low oil content (e.g. 5%) may not justify the effort to use this feedstock since extra cost may be inevitable to better accommodate it.

(2) Suitable for current technology. The successful candidate feedstock should not require significant update of the current facilities to process it into biodiesel.

31

(3) Substantial supply. Even feasible in technical aspect, the adoption of waste-derived feedstock will still not be justified if there is no large enough potential supply existing.

2.4.2 Emerging technical challenges in purification

Currently, many biodiesel plants have switched to dry purification technologies using adsorbents and ion-exchange resins (Zeman, 2010). The merits of these technologies include: elimination of wastewater generation, no need for moisture removal in the resulting biodiesel, and shortening the purification time. Adsorbents (e.g. Magnesol®) selectively remove the impurities based on their size as well as polarity, while ion-exchange resins (e.g. Purolite® PD206) change the H ion with Na/K ion to break down the soap and remove the other impurities by the mechanism of filtering (Cooke, 2004; Jacob, 2009; Kram, 2008). The advantages and disadvantages of these two types of materials have been discussed by existing studies (Faccini et al., 2011; Jacob, 2009;

Sims, 2011). Nowadays the commercial drying purification materials in the market work quite well for its purpose for the current ASTM D6751 standard for biodiesel. However, diversification in the feedstock market and increasingly-stringent fuel quality specifications keep calling for the development of novel dry purification materials that can deal with the emerging challenges with respect to dry purification technologies. Sulfur-containing compounds and steryl glucosides are the examples of challenges that need to be addressed. These two impurities are expected to be the focus of the R&D for biodiesel purification field in the near future for the following reasons. Firstly, there is a trend of utilizing low quality feedstocks to reduce the biodiesel production cost; as a result, the risk of elevated sulfur content in the final biodiesel product is concerned for some of the feedstocks. For instance, if trap grease oil is used for biodiesel, it’s likely that the sulfur content in the final biodiesel product will exceed the 15 ppm limitation (Borgese, and Privitera, 20111,2; Chakrabarti et al., 2008; He et al. 2009). The high

32 sulfur level is mainly attributed to the organic/bonded sulfur compounds (e.g. proteins), rather than sulfur containing salts (e.g. Na2SO4, K2SO4) that can be easily removed by water or current adsorbent. Those sulfur-containing compounds are intrinsic to the feedstock types, and currently the most effective way is to apply distillation, which is a highly energy intensive process. Thus, a cost-effective dry purification material that is capable of handling high sulfur feedstock oils/biodiesel is expected to be highly competitive in the purification market. Steryl glucosides

(SGs), on the other hand, increase the risk of filter plugging issue associated with using biodiesel in cold regions. SGs are inherent in the vegetable oils and fats. Before biodiesel conversion process, the SGs are in the acylated form, which enables them to be very soluble in the oil. After conversion, the SGs are deacylated and become easy-to-precipitate due to the elevated melting point (240 °C, Lee et al., 2007). To deal with the cold performance issues, the Cold Filter Clog

Point (CFPP) test has been added to the latest ASTM standard (NBB1, 2011) to determine the applicability of biodiesel in the cold weather (-12 °C). SGs crystallization and precipitation is the major contributor to the failure of biodiesel to pass the CFPP (Lee et al., 2007; Tang et al. 2008;

Wang et al., 2010; White et al., 2010). With the increased awareness of SG issues, characterization of SG precipitation from biodiesel and its removal have attracted much attention of researchers (Haupt et al., 2009; Tang et al., 2010; White et al., 2010); however, to date, the most effective way of removing SG is still the energy-consuming distillation process.

In a word, to address the challenge of sulfur content and SG removal in the biodiesel, a commercially viable technology should be developed.

2.4.3 Cost-effectiveness for purification

For the current dry purification technologies, one major disadvantage against water washing is the relatively high cost of purification materials (Sims, 2011). By using the data available, the

33 result of a simple calculation of purification cost by different purification media is summarized in the table below.

Table 2.4 Summary of costs for selected dry wash materials Cost per gallon Major Composition Material Unit Price ($/lb) Biodiesel ($) Magnesol® 4-5 0.30-0.37 Magnesium silicate Purolite® PD206 NA 0.35-0.5 Ion-exchange resin AmberliteTM 6 0.44 Ion-exchange resin ALX DW-RTM 3 0.22 Ion-exchange resin

The prices of the dry purification materials were collected from the vendors and the calculation was performed based on the following assumptions:

(1) The application rates of the adsorbents and resins are fixed at 1 wt% of the crude biodiesel

(Faccini et al., 2011), unless other values are indicated; and

(2) The cost of using Purolite® PD206 is cited from reference (Sims, 2011);

As is shown, the purification costs range from 22 to 50 cents per gallon biodiesel. Although this cost may be offset by the saving from reduced expense on fresh water use and wastewater treatment, it is always desirable to find a cost-effect way to purify crude biodiesel and thus, maximize the margin for a biodiesel company, especially under an unstable circumstance of subsidy policy (NBB2, 2012).

2.4.4 Non-hazardousness and environmentally-friendliness for post-purification waste

Last but not least, another significant concern about the existing dry purification materials is their disposal. The dry purification materials are usually non-flammable. But the biodiesel, methanol and glycerin entrapped are a potential fire hazard if the used purification materials are not appropriately stored or disposed. Exposure to direct heat (e.g. sun shine) and/or heat built-up in a confined space (e.g. congested area in a landfill) may all lead to the spontaneous combustion of these after-use purification materials, especially when the methanol concentration is high.

34

EPA Test Methods 1030 and 1050 under guidance document SW-846 can be used to determine if solid waste (e.g. dry purification materials) from a biodiesel plant has the spontaneous combustion potential. Several occurrences of spontaneous combustion have been reported in both biodiesel facilities and landfill, which are caused by improper management of used purification materials (Nebraska Department of Environmental Quality, 2009; US EPA, 20082).

The Land Disposal Restrictions (LDR) requires the hazardous wastes to be treated physically and/or chemically before being sent to the non-hazardous waste landfill. For the used purification materials, the removal of methanol and other ignitable components are required. This removal not only adds extra cost to the biodiesel production but also leaves potential safety issues in the landfill if the treatment is not thorough enough. In addition to a fire hazard, the issue of land occupation by the expansion of the landfill should also be of concern if an increasing number of biodiesel purification wastes are dumped.

Another option is to send the used purification materials to an incinerator, where the mass reduction of the waste materials is significant under high temperature. Though with less hazardousness and negative environmental impact, incineration still poses an extra fee to the biodiesel plants.

The best way to manage the waste is to not generate it. Therefore, an alternative to the above two options is to use the waste purification materials as the fuel in the boiler of the biodiesel plant.

By doing so, it may also reduce the operational cost of the plant by lowering the fuel purchase.

However, knowledge is absent in the emission from burning those used purification materials, due to the lack of practice in the commercial-scale biodiesel plants and lab-scale researches.

Another concern is the heating value. Considering their compositions, the purification materials

35 in the current market are expected to have a low heating value that may not be enough to justify the investment for the potential boiler modification.

2.4.5 Coffee grounds as a solution

From the literature review and our preliminary study, WCG has the potential to be an all-in-one solution to the problems stated above. Our purification experiments show that WCG has comparable cleaning ability to commercial materials for most impurities and thus is expected to be a good alternative polishing material for meeting the dry purification needs. In addition, the moderate oil concentration in WCG enables it to be a potential oil source for biodiesel production. Coffee, as world’s most consumed beverage, leaves behind a substantial supply of

WCG every year. For instance, in 2008, every American consumed 24.2 gallons of coffee on average, which translates into 3.38 billion lbs of WCG generated in total (Population Reference

Bureau, 2012; US Census Bureau, 2012). If assuming 10 wt% recoverable oil in WCG, the oil supply for biodiesel could be 44 million gallons per year (based on 2008 data). Therefore, the coffee oil from WCG can be a commercially feasible option for biodiesel industry, considering its low/zero feedstock cost. Additionally, this huge supply coupled with its high heating value

(around 9, 000 Btu as measured by third-party laboratory for our sample) also indicates the suitability of WCG as adjunct fuel for heat/power generation.

2.4.5.1 WCG as an oil source

Concentration and composition of oil in coffee beans and grounds have been explored by several studies (Table 2.5). Soxhlet extraction is the traditional and most frequently used process for oil extraction while supercritical fluid extraction with CO2 is one of the emerging focuses of research in recent years. As is shown, oil (containing both lipids and FFAs) concentration varies individually, dependent upon the coffee species, preparation procedures and extraction

36 conditions. In most cases, oil concentration is within the range of 10 to 20 wt% in both coffee beans and grounds. Compared with the traditional soxhlet process, supercritical fluid extraction with CO2 (with/without co-solvent) takes much less time to finish the extraction, though the recovery rate (wt%) of oil is usually lower. For soxhlet extraction, a variety of solvents have been tested, among which n-hexane is most commonly used. Polarity of the solvents plays a vital role in the recovery rate of oil components. Non-polar solvents, such as n-hexane, are very effective in dissolving tri-, di- and mono-glycerides while polar solvents, such as isopropanol, are suitable for FFAs and phospholipids. Therefore, finding a combination of solvents for an optimal polarity can be the point of interest for research. Besides, the composition studies show that both lipids and FFAs from coffee beans and grounds mainly consisted of long oxygenated carbon chains, such as C16:0, C18:0, C18:1, and C18:2.

37

Table 2.5 Summary of oil level and composition from literature for coffee beans and grounds Case Oil Concentration (wt%) Composition Extraction Method Coffee Beans Raw: 14.71 Bengis and Anderson Oil:NA Soxhlet with petro-ether as the initial and ethyl ether as the Roasted: 16.10 (1934) FFA: C18:2, 18:1, 16:0, 18:0 subsequent solvents Stale: 15.97 Pure CO2: 10.48 De Azevedo et al. CO2/EtOH (5wt%): 10.75 NA Supercritical fluid extraction (2008) CO2/Iso (5wt%): 9.43 Mazzafera (1999) 9-15 NA Soxhlet with hexane Oil: C18:2 (44%), and C16:0 (34%), C18:1(9%), 18:0(7%), 18:3, 22:0, Before roasting: 9.9 20:1 and araquidic acid FFA: Oliveira et al. (2006) 4.9±0.4 g oleic acid/100g oil, which translates into 9.7±0.8 wt% Solvent extraction and screw pressing oleic acid of oil Roasted: 10.3 Unsaponifiables: (BUT the roasted coffee beans have lower Solvent extraction: 9.2 g/100g oil base weight) Screw pressing: 12.8 g/100g oil

Oil: Same as their 2006 paper above Soxhlet with hexane 10-12 FFA: Oliveira et al. (2008) Healthy: 2.62±0.29 Defected: 10.04±0.03

Ratnayake et al. C. Arabica: 15.5 (82.8% as TG) NA Blend mixing with solvents (1992) C. robusta:9.8 (81.7% as TG) Oil: C18:2, 18:1, 16:0, 18:0, 14:0, 20:0 Schutte et al. (1934) 2.7 Soxhlet with petrolic ether FFA: 7.05 (acid number) WCG Al-Hamamre et al.. FFA: Soxhlet with different solvents 8.6-15.28 (2012) 3.25-6.4 wt%

Soxhlet with hexane: 18.3 Oil: Soxhlet with hexane Couto et al. (2009) C12:0, 14:0, 16:0, 18:0, 18:1, 18:2, 18:3, 20:0 Supercritical CO2 (composition varies among different extraction conditions) Supercritical: 15.4 without EtOH Supercritical with EtOH (6.5wt%):19.4

Khan and Brown Oil: C16:1, 18:0, 18:2, 18:1, 16:0 NA Soxhlet with trichloroethylene (1953) FFA: 0.27% Hexane:13.4 Kondamudi et al. Diethyl ether:14.6 FAME: C18:0, 18:1, 18:2, 16:0, 20:0, 20:5 Soxhlet with different solvents (2008) DCM:15.2 Ratnayake et al. 8.51 (84.4% as TG) NA Blend mixing with solvents (1992)

38

Specific to biodiesel production, existing literature also reported several successful efforts.

Oliveira et al. (2008) analyzed the composition of the oil from both healthy and detected coffee bean grounds and produced biodiesel from the extracted oils through transesterification reaction.

The study focused on the optimization of the operational parameters (temperature and reaction time) for the biodiesel production and the highest yield (70.1%) was achieved at 25 °C and 1 hour reaction time. In the US, the team from University of Nevada at Reno has successfully made biodiesel that meets all the ASTM D6751specifications by means of extracting oil from

WCG. In their study, Kondamudi et al. (2008) obtained oil from WCG by using solvent extraction in a 1 liter flask and the characterization showed that the averaged extractable oil content in the coffee ground was around 15 wt%. The WCG (after extraction) was proposed to be the adjunct fuel for combustion and the oil was converted into biodiesel through the regular alkaline-catalytic transesterification. The yield in their case was close to 100%, which may be due to the better quality of the oil (e.g. less free fatty acids, and unsaponificables). Although the above-mentioned the studies have proved the feasibility of making biodiesel from coffee grounds

(beans and WCG), the focus was mainly placed on the “producing step” of biodiesel; in other words, detailed study on the oil extraction process has not been stressed enough. In effect, the type(s) of the solvent used and the optimization of extraction time are crucial to the success of commercialization of this WCG-to-biodiesel pathway and therefore need to be clarified.

2.4.5.2 WCG as adsorbent

Research efforts have been put in developing adsorbents from coffee waste, coffee beans and coffee grounds. Hirata et al. (2002) studied adsorption of dyes onto the activated (ACs) prepared by microwave treatment of coffee grounds. The research team led by Franca et al.

(Franca et al., 2010; Nunes et al., 2009) conducted adsorption studies for the ACs made from

39 thermal treatments of coffee bean grounds that had undergone oil extraction for biodiesel production. The studies showed that the resulting ACs demonstrated moderate uptake capacity

(14.9 mg/g by oven carbonization and 69 mg/g by microwave) of cationic dyes in the aqueous solution and thus could be an inexpensive and readily available source of adsorbents for the wastewater treatment.

Boonamnuayvitaya et al. (2004) used coffee residues from instant coffee production as the raw material for AC. A variety of parameters were tested to determine the suitable condition for preparation the AC. Under the optimal condition, coffee residues were bound with clay (CC- adsorbent) at the ratio of 80:20 under 500 °C. The resulting CC-adsorbent was featured by high fraction of mesopores and a granular size of 4 mm in diameter. Franca et al. (2009) used spent coffee grounds as the adsorbent directly (without activation), which indicated that spent coffee grounds can be a suitable adsorbent for removal of cation dyes like methylene blue. Escudero et al. (2008) performed research on the effect of EDTA to interfere with the adsorption of divalent metal ions by coffee wastes and grape stalk. The coffee wastes in their study were from the instant coffee production process without activation. EDTA had the inhibitive influence on the adsorption of metal ions in both adsorbents, while coffee wastes were less susceptible to this negative effect.

Boonamnuayvitaya et al. (2005) also studied activation of coffee residues by different agents

(ZnCl2, N2, CO2, and steam). The prepared ACs were tested for the effectiveness of adsorbing formaldehyde vapor. It turned out that the coffee residue treated by ZnCl2 impregnation and N2 activation (CZn-N2) demonstrated highest adsorption capacity, which was ascribed to its featured surface functional groups. Namane et al. (2005) also studied chemical activation of coffee grounds by ZnCl2 and H3PO4; and the prepared ACs were investigated for their capability of

40 removing phenol and dissolved dyes. In addition, Boudrahem et al. (2009) investigated the lead ion removal by coffee ground-derived ACs that were impregnated by ZnCl2. It was determined that a 100% impregnation ratio yielded the highest micropore volume (0.772 cm3/g) and surface area (890 m2/g). In terms of lead uptaking capability, however, a 75% impregnation ratio was considered as the optimal condition since it displayed the same adsorption ability as the AC from

100% impregnation. Reffas et al. (2010) prepared the ACs from coffee grounds by H3PO4 impregnation and pyrolysis under high temperature. It was reported that impregnation process determined the porosity, pore structure and surface chemistry of the resulting ACs. The removal efficiency of resulting ACs on the methylene blue and nylosan red was studied. The high impregnation ratio (180%) lead to the formation of exclusively mesoporous ACs that are highly effective to organic pollutant removal. Kante et al. (2012) studied the effectiveness of AC prepared from waste coffee ground on H2S separation. In their study, zinc was used as the activation agent and the surface features of the AC prepared were analyzed. It was shown that the resulting ACs possessed a large amount of pores with sizes between 10 and 30 Å, which is an optimal size range for H2S adsorption. Khenniche and Benissad-Aissani (2009, 2010) also studied the effectiveness of ACs from coffee residue on the removal of salicyclic acids and phenol. The ACs were prepared from ZnCl2 activation and the optimal impregnation ratios was found to be 25% with the highest microporosity. The influence of impregnation ratio was also studied by Ching et al. (2011). The coffee grounds in their case were impregnated by H2SO4 and the optimal ratio was determined as 50% under which the highest pore surface area (224.7 m2/g) and micropore volume (0.07986 cm3/g) were achieved. Castro et al. (2011) used water vapor and

K2CO3 as the activation agents for AC preparation from spent coffee ground. The prepared ACs showed high microporosity and surface area by both agents, with chemical activation (K2CO3)

41 yielding greater values in these two parameters (950 m2/g and 0.38 cm3/g). capacity of ACs prepared by KOH impregnation was studied by Akasaka et al. (2011). The waste coffee beans were activated by KOH at different ratios and the highest surface area (2070 m2/g) was obtained at the weight ratio of 5:1. The hydrogen storage capacity was found to be proportional to the surface area of the resulting ACs and waste coffee beans were proved to be a suitable raw material for preparing hydrogen storage carbon.

Tokimoto et al. (2005) also used spent coffee grounds as the media for AC production and conducted experiment on lead ion removal. Their study investigated the relationship between adsorption characteristics and different inherent features (e.g. levels of caffeine, fat and protein) in the spent coffee ground. It was found that the lead ion adsorption was explicitly proportional to the protein content in the coffee. Kaikake et al. (2007) investigated the metal ion removal by degreased coffee beans. The degreased coffee beans were found to be effective to the aqueous solution with small concentration of metal ions. It was also proposed that between the metal ions and cations in the degreased coffee beans was the major removal mechanism.

Degreased coffee beans were also studied by Baek et al. (2010) for the removal of Malachite green. The specific surface area and porosity of the coffee beans increased after the degreasing process and so for the removal capability.

In order to further diversify the beneficial use of WCG, it is necessary to see whether the WCG can be comparable to the existing dry wash materials in terms of removing the impurities such as trace glycerin, soap and moisture in the crude biodiesel. However, to date, the application of

WCG (after extraction) for biodiesel purification has not yet been explored and thus is investigated in this study.

42

2.4.5.3 Spent WCG (after purification) as adjunct fuel

Coffee grounds have attracted a certain level of attention as an adjunct fuel. In Japan, used coffee grounds have been tested as an to generate power (ABC, 2012). In addition, as was estimated from an LCA study that looked into the burning of biomass, combustion of 1 kg of coffee grounds in municipal incineration could generate 0.53 kWh electricity and 3.92 MJ of useful heat (ESU-services Ltd., 2011). In our study, the preliminary analysis showed that dry coffee grounds had a heating value of 8,795 Btu/lb with 0.88% ash content, which indicates its suitability as a good resource to generate heat.

43

Chapter 3

Influence of Biodiesel Production on Water Resources

3.1 Goal and Scope

The first goal of this part of the thesis is to provide a systematic and up-to-date characterization of water consumption throughout entire soybean biodiesel production cycle. The second goal is to provide a review on the waste water generated during biodiesel production stage and corresponding regulation concerning biodiesel waste water discharge. Water consumption in this study is defined as the portion of water that is not recycled or further used after application in a specific stage. For example, in the feedstock growth stage, water consumption is regarded equal to the irrigation water that is applied to the particular feedstock. In respect to the specific feedstock, considering its dominant role in the market, soybean biodiesel is chosen as the target of interest in this study. Accordingly, the boundary of the study starts from soybean growth and ends by the refining of a biodiesel final product, as shown in Figure 3.1. The entire soybean biodiesel production cycle is divided into three main stages: soybean growth, soybean processing and biodiesel production; and the water consumption steps are identified correspondingly in each stage. In Figure 3.1, the vertical flows (dash lines) demonstrate the input/output of water in each stage. For example, in the “soybean growth stage”, the water input includes both rain water and irrigation water. On the other side, the output of water is runoff and evapotranspiration. Similarly, the horizontal flows (solid lines) represent the flow of soybean between different stages. For instance, in the “crushing & extraction stage”, the flow denotes the amount of soybean oil that finally goes to a biodiesel production facility. On a yearly-basis, “irrigation water consumption

(W1)”, “refining water and process related water consumptions for soybean processing (W2)” and “biodiesel purification water and process-related water consumptions for biodiesel

44 manufacturing (W3)” were the targets of interest and hence quantified in this study, while other water input and outputs were beyond the scope of this study. The criterion for data source selection is based on a combination of the factors such as spatial and temporal representativeness, availability, and transparency of the data collection and analysis processes. However, due to the limited access and proprietary reasons, some of the water consumption data was in aggregated form and no specific information about the technique used was available. In addition, to compensate for the limited data availability, the water consumption from industry survey for the biodiesel purification step involved multi-feedstock biodiesel plants instead of those operating on soybean oil alone. Specific to irrigation water consumption, unlike the literatures found, in this paper, state-level data was used whenever possible instead of national averages to reflect the local situations. In addition, mass-based allocation was performed for all the stages to obtain a more accurate allocation of water consumption.

Figure 3.1 Schematic of water consumption stages of biodiesel production from soybean

45

3.2 Methodology

3.2.1 Soybean plant growth stage: irrigation water consumption (W1)

The growth of soybean requires water input, either from rainfall or irrigation. The practice of irrigation for soybean growth is determined by the local climate conditions. In current study, the irrigation water was considered as completely consumptive and corresponding annual irrigation water consumption (W1), in the unit of “million gallons per year (MMgy)”, was determined on a state-level basis for this stage by using the data from “USDA Farm and Ranch Irrigation Survey”

(USDA, 2008) after mass-based allocation (discussed in later section).

3.2.2 Soybean processing stage: crushing & extraction, crude oil refining (W2)

After harvest, the soybean is transported to the refining plant for oil extraction and refining. The water consumption consists of: (1) equipment-operation related water consumption, such as cooling tower makeup water; and (2) water used as the purification media for refining steps such as free fatty acid removal (United Soybean Board 2010; Van Gerpen et al. 2004). Considering the transportation cost, the soybeans harvested were assumed to be processed locally within the same state; thus, the W2 (MMgy) was assumed to have the same spatial scope as W1 in this study. Since water consumption involved in the soybean processing is highly case-specific depending on the practice of the plants, it was hardly possible to collect the state-level data. Thus, a consumption factor was calculated from the result of a survey conducted by National Oilseed

Processers Association (NOPA) among its member plants in 2008 (United Soybean Board 2010).

This factor was assumed to be uniformly applicable to all the soybean processing plants in each state and W2 was calculated based on the harvest of soybean in the individual state and same allocation method.

46

3.2.3 Biodiesel production stage: crude biodiesel purification, cooling tower makeup (W3)

Water use in the stage of making biodiesel (refers to the product from the transesterification process) consists of three categories: washing water to purify the crude biodiesel, cooling tower makeup and boiler makeup. The water use in biodiesel manufacturing is highly process dependent for the following reasons. Nowadays more plants are adopting dry wash technology, i.e. using resins or sorbents to remove glycerin, color and impurities instead of water wash. For the water wash, more plants are reusing the washing water instead of discharging after one use.

Boiler water makeup should be considered when a plant uses distillation to separate glycerin from biodiesel, and the consumptions vary depending on the distillation processes used in the facilities. As an example, vacuum distillation is widely used and the resultant boiler water makeup can be much lower than steam distillation. In addition, cooling tower makeup can be the largest source of water consumption in a biodiesel plant if the plant makes use of evaporative cooling towers to condense process vapors (such as for methanol recovery) and cool liquid process streams. The actual consumption can vary considerably depending upon the process system setup and the extent of heat economization used in the plant (Scott 2010; Smith 2011).

Similar to W2, it’s impossible to look into every biodiesel plant and calculate the water consumption for each state. In this study, the washing water consumption data was collected from a literature review and an industrial survey of biodiesel producers. The cooling tower makeup, as mentioned before, was highly case-specific. Due to the limited access, only data from two biodiesel companies was presented as the reference. The data for boiler makeup, however, was not available so far since no information could be found from the literature review or industry survey. Based on these data collected, three scenarios were calculated to approximate different water consumption situations and an averaged factor was taken to represent the most-

47 likely water consumption situation in a biodiesel plant. Also, this factor was assumed to be universally applicable to all the biodiesel plants in the US. Unlike W1 and W2, the water consumption data for W3 calculation was reported in the unit of “gallon water per gallon biodiesel (gal/gal)” directly and hence no allocation was performed for this stage. W3 here was calculated based on the total annual productivity of biodiesel plants within a specific state and thus was also in the unit of MMgy.

3.2.4 Allocation and normalized irrigation water consumptions

As mentioned earlier, mass-based allocation was applied in current study. Firstly, water consumption was allocated between soybean meal and soybean oil. In current study, oil content of soybean was estimated to be 19.5% (United Soybean Board, 2010). Next, it was acknowledged that only a small portion of soybean oil was used for biodiesel production. So, corresponding data of soybean oil consumption for biodiesel production was cited from a consulting publication (Centrec Consulting Group, LLC, 2010), which contained the statistical data from US Census Bureau and USDA/ERS. Finally, during the biodiesel manufacturing process, glycerin is also produced and usually sold as value-added product. To correctly allocate the water consumption between biodiesel and glycerin, a mass ratio of 89% vs 11% was adopted in current study (United Soybean Board, 2010). After allocation, water consumptions were quantified on a yearly-basis for W1, and W2. Also, normalization was performed to characterize the state-level irrigation water consumption intensity (Wni) and total water consumption (Wnt) in the unit of “gal/gal”. In addition, the nationwide averaged Wnt from current study could be directly compared with values in the existing literature.

48

3.2.5 Sample calculation

A sample calculation for State of Ohio was performed to demonstrate the mass-based allocation in current study. The data (Table 3.1) and calculation procedures are provided below.

Table 3.1 Major data sources for current study (Ohio) Data type Value Unit Period Irrigation intensity 3.2 Acre-feet/acre 2008 Irrigated area 1,056 Acre 2007 Soybean harvested 191,559,567 Bushel 2007 % oil in dry soybean 19.5 % 03-07 % oil used for biodiesel production 17 % 2007 % oil converted to biodiesel 89 % 2010 * Irrigation related data listed in Table 3.1 is specific for Ohio as an example ** Other conversion factors such as bushel-to-pound conversion, density of biodiesel are not listed

According to Table 3.1, the total irrigated area for soybean through primary water distribution methods was 1,056 acres in Ohio and the average acre-feet applied per acre was 3.2. Therefore, the total irrigation water for soybean in 2007 was:

VT 1,0563.2  3,379.2 acre  feet

Since one acre-foot equals 325,851 gallons:

9 VT  3,379.2325,851  1.1010 gallons water per year

Mass-based allocation:

According to the assumptions above, 19.5% of soybean was oil, about 17% of the soybean oil in

2007 was used for biodiesel production and 89% oil was eventually converted into biodiesel. So:

Therefore, W1 for the State of Ohio was estimated to be 32.5 MMgy.

For calculating Wni for the State of Ohio, normalization was applied as following:

49

The total harvested soybeans in bushel were 191,559,567 in Ohio, hence the normalized irrigation water consumption per bushel of soybean was:

9 R1 1.1010 /191,559,567  5.75 gallons water /bushel soybean

So, with the same allocation method, the water consumption for all the biodiesel from one bushel of soybean was:

R2  5.7519.5%17%89%  0.17 gallons water /bushel soybean

Generally, one bushel soybean weighs about 60 pounds (US Commercial Bushel Sizes, 2001) and the density of soybean biodiesel is 7.4 lb/gallon (United Soybean Board, 2010). Therefore, the irrigation water consumption intensity (Wni) for Ohio, based on every gallon of biodiesel was:

Wn  0.71 gallons water / gallon soybean biodiesel i

Following the same principle, W1 and Wni for soybean growth were calculated for each state by using state-specific data from USDA reports (Census of Agriculture, 2007; Farm and Ranch

Irrigation Survey, 2008).

3.2.6 Waste water characterization

For the quality of waste water affiliated with biodiesel production, a summary on BOD and COD values of biodiesel washing water was performed. In addition, EPA documents regarding the regulations for industrial waste water discharge that pertain to biodiesel production were reviewed.

.

50

3.3 Results and Discussion

3.3.1.1 State-level irrigation water consumption intensity (Wni) and annual consumption

(W1) for soybean biodiesel

Figure 3.2 shows the result of normalized irrigation water intensity (Wni) in the US for the soybean dedicated to biodiesel production. A total of 35 states were plotted and the rest 13 states, namely Alaska, Arizona, California, Florida, Hawaii, Idaho, Montana, Nevada, New Hampshire,

New Mexico, Oregon, Pennsylvania, Rhode Island, Utah, and Wyoming were excluded due to either negligible soybean growth or data deficiency. Based on soybean harvest, the weighted nationwide average of Wni was 61.78 gal/gal, and the range was from 1058.20 gal/gal

(Washington) to literally 0 gal/gal (e.g. Vermont). The irrigation water use varied significantly from region to region. The States of Washington, Arkansas, and Colorado, Mississippi and

Nebraska were the top five states with the highest irrigation water consumption intensities of

1058.2, 674.3, 611.5, 285.9 and 199.7 gal/gal, respectively. However, if soybean productivity was taken into account, it was found that Iowa, , Minnesota, Indiana and Ohio were top 5 states accounting for 56.37% of US while their irrigation water intensities were only 1.88, 4.01,

8.6, 7.85, and 0.71gal/gal, respectively. Additionally, if looking at the biodiesel production capacities (Biodiesel Plant List, 2012), Texas, Iowa, Missouri, Illinois, and Ohio were the top 5 states in the US and the average Wni was 51.7 gal/gal for these states. On the other hand, the summary of irrigation water consumption (W1) of each state was plotted in Figure 3.3. Range of

W1 varied from 0 to 15,953 MMgy, with a weighted nationwide average of 1,812 MMgy. Since

W1 was directly related to soybean growth area, it can be found that although States of

Washington and Colorado had very high irrigation water consumption intensities, their W1s ranked at the lower end of all the 35 states due to the relatively smaller growth scale. Similarly,

51 in the top 5 soybean producing states, the weighted average was 298.79 MMgy; and for top 5 biodiesel producing states, the weighted average was 483.92 MMgy. Both were much lower than the nationwide average. However it is also indicated that even the normalized irrigation intensity is low, the annual irrigation water consumption for soybean-derived biodiesel could still be quite high in total.

52

10000

1000

100

10 Wni (gal/gal) Wni

1

0.1 WA AR CO MS NE TX DE KS LA GA OK MO MD NC NJ SC MI AL VA WI MN SD IN KY TN IL ND IA OH CT ME MA NY VT WV State Abbreviation

Figure 3.2 Normalized irrigation water intensity (Wni) for 35 states

100000

10000

1000

100 W1 (MMgy) W1

10

1 AR NE MS KS MO LA MN IN IL NC MI SD GA IA TX WI DE MD OK ND SC KY VA OH TN NJ CO AL WA CT ME MA NY VT WV State Abbreviation Figure 3.3. Annual total irrigation water consumption (W1) for 35 states

53

3.3.1.2 Profile of irrigation methods/water sources applied by state

Figures 3.4&3.5 show the summary of two primary irrigation methods, pressure system and gravity system, for soybean growth in selected states. The evaluation of these two methods (e.g. advantages/disadvantages) and their implications on water intensity of soybean biodiesel are beyond the scope of this study and the information provided below is intended for illustrative purpose alone. Figure 3.4 displays the ratio of application between these two methods in the top five states with highest irrigation areas, from which it is indicated that the practice is highly case- specific. The similar phenomenon could also be observed when the target of interest becomes feedstock market share of soybean (Figure 3.5).

On the other hand, the profile of irrigation water sources was more consistent among individual states (Figures 3.6&3.7). Among all three types of water sources, “ground water from wells” was the dominant options for farmers. Also, it is noteworthy that irrigation was typically exerted by using on-site water resources, since for each case the “water from off-farm suppliers” accounted for less than 5% of the total water consumed.

54

Ratio between Irrigation Methods in Top 5 States with Highest Irrigation Areas 2,500,000

2,000,000

1,500,000 Gravity system

Pressure system 1,000,000

Irrigated AreaIrrigated(acres) 500,000

0 NE AR MS MO KS

Figure 3.4 Ratio between irrigation methods in top 5 states with highest irrigation areas

Ratio between Irrigation Methods in Top 5 States with Highest Feedstock Market Share of Soybean as Feedstock 600,000

500,000

400,000 Pressure System 300,000 Gravity System

200,000 Irrigated AreaIrrigated(acres) 100,000

0 IA MO IN SC IL

* Data of MN is not included due to data deficiency Figure 3.5 Profile of Irrigation Methods in Top 5 States (Market Share)

55

Profile of Irrigation Water Sources in Top 5 States with Highest Biodiesel Production Capacity 2,500,000

Water from off-farm suppliers 2,000,000

On-farm 1,500,000

1,000,000 Ground water from wells

Irrigated AreaIrrigated(acres) 500,000

0 TX MO IL AR MS

* Data of MS is used in the place of IA due to the data deficiency Figure 3.6 Profile of Irrigation Water Sources in Top 5 States (Capacity)

Profile of Irrigation Water Sources in Top 5 States with Highest Feedstock Market Share of Soybean as Feedstock

600,000

500,000 Water from off-farm suppliers 400,000

300,000 On-farm surface water

200,000 Ground water from wells

Irrigated AreaIrrigated(acres) 100,000

0 MO IN SC MN IL

* Data of IL is used in the place of IA due to the data deficit Figure 3.7 Profile of Irrigation Water Sources in Top 5 States (Market Share)

56

3.3.2 Soybean processing and refining (W2)

Table 3.2 below shows the aggregated data of soybean crushing, extraction and degumming steps from NOPA, as well as the water consumption during caustic refining (United Soybean

Board, 2010). Same assumption for soybean oil usage for biodiesel production (17%) was applied. Figure 3.8 summarizes the water consumption for soybean processing (W2). 12 states were excluded either due to no soybean growth or data deficit (States of Alaska, Arizona,

California, Hawaii , Idaho, Nevada, New Hampshire, New Mexico, Oregon, Rhode Island, Utah, and Wyoming). The range of W2 varied from 0.003 to 112 MMgy, with a weighted nationwide average of 59.08 MMgy based on soybean harvest.

Table 3.2 Water consumption during soybean processing and refining stage Process Water (kg/1000 kg Water (gallon/gallon Water (gallon/gallon oil) oil) biodiesel) Crushing, extraction 1,164 1.047 0.16 & degumming Caustic refining 65.9 0.059 0.009 * Assume oil loss between these steps is 4%; water density = 8.35 lb/gallon; soybean oil density = 7.51 lb/gallon; 1 kg = 2.2 lb; 17% soybean oil is used for biodiesel production; 89% wt of soybean oil is converted into biodiesel

57

1000

100

10

1

W2 (MMgy) W2 0.1

0.01

0.001 IA IL MN IN OH NE MO SD ND AR KS MI WI MS KY NC LA TN VA MD GA SC NY OK DE AL TX NJ PA WV FL CO VT WA ME MT CT MA State Abbreviation *This figure presents the soybean processing water consumption (MMgy) for 38 soybean growing states. The horizontal line indicates the nationwide averaged value based on these 38 individual numbers. Figure 3.8 Annual soybean processing water consumption for 38 states (W2)

58

3.3.3 Biodiesel purification and cooling tower makeup (W3)

3.3.3.1 Biodiesel washing water consumption

Overall, the water consumption in the biodiesel production stage was much less than that of irrigation while still remained highly case-specific. Table 3.3 summarized the washing water consumption data collected from several existing studies and personal communication with expert from National Biodiesel Board (Scott, 2010). United Soybean Board (2010) conducted a life cycle assessment for soybean-to-biodiesel process, where water washing was quantified as

0.26 gal/gal. This number was cited from a survey conducted by National Biodiesel Board, though the date and scope of the survey were not specified in the report. A similar result was provided by Mr. Don Scott, the Director of Sustainability for National Biodiesel Board.

According to him, one pound of water was consumed for four pounds of biodiesel, which translated into 0.22 gal/gal on a volumetric basis. Haas et al. (2005) and Zhang et al. (2003) used simulation software to estimate the energy and cost inputs for biodiesel production for soybean and other feedstocks. The results were close to each other (0.03 and 0.01 respectively) and both were much lower than the results from existing industry summary, which might be explained by the difference between simulation and practical operation.

Table 3.3 Washing water consumption data from literature review and personal communications Washing Water Production Capacity Data Sources Feedstock (gal/gal) (Million Gallons) United Soybean Board Soybean 0.26 NA (2010) Scott (2010) Multi-feedstock 0.22 NA Zhang et al. (2003) Multi-feedstock 0.01 2.4 Haas et al. (2005) Soybean 0.03 10

In the current study, an industry survey was also conducted to obtain an up-to-date summary of the current practice of biodiesel water washing in the US biodiesel market. The data from

59 industry survey is summarized in Table 3.4. A total of 123 inquiries were sent to the commercial biodiesel companies and 21 replies were received, among which 7 reported applying water washing, 11 indicated dry purification and the rest did not specify the technology used. Six out of these 7 companies offered an estimation of washing water consumption and based on their capacities, a weighted average was calculated based on plant capacity and the result was found to be 0.12 gal/gal. The variation in the data reported by these companies can be attributed to a number of factors, including water reuse practices, washing water properties (e.g. acidic/warm), plant size, as well as water availability and price. As compared to the existing industrial estimations (Scott, 2011; United Soybean Board, 2010), this value was about 50% less. This can be explained by the difference in scope and time of the surveys as well as technology development and increased awareness of water resource protection.

Table 3.4 Washing water consumption from industry survey in current study Washing Water Production Capacity Data Sources Feedstock (gal/gal) (Million Gallons) Company 1 Multi-feedstock 0.1 3 Company 2 Animal Fats 0.0125~0.015 1.25 Company 3 Waste Cooking Oil 0.84 4.5 Company 4 Multi-feedstock 0.25~0.375 12 Company 5 Multi-feedstock 0.09~0.1 180 Waste Cooking Oil Company 6 0.06 1.5 Animal Fats * Company names are omitted due to proprietary reason

3.3.3.2 Cooling tower makeup water

It was proposed that both cooling tower makeup water and boiler makeup water should be collected from the industrial inquiry. However, since most of the participants did not have a detailed breakdown of water bill for those two numbers, only limited data about (evaporative) cooling tower makeup water consumption was presented (Table 3.5). Evaporative cooling tower is a commonly used type of condensation practice for gas recovery in a variety of industries.

60

Table 3.5 summarizes the evaporative cooling tower makeup data from two specific biodiesel plants. From Case A, it could be found that using low quality feedstock usually resulted in higher makeup water input, which could be attributed to the need to recover excessive amount of methanol (e.g. for the esterification reaction). Another noteworthy phenomenon is that even when the same feedstock was used, the dry wash method often consumes less water during distillation as compared with recycled water wash. This could be explained by the increased water evaporation due to the distillation of recycled water. On the other hand, Case B had much smaller makeup water consumption, which is achieved by integrating other cooling approaches such as air chiller. This once again indicated the highly process-specific feature of the water consumption in a biodiesel plant.

Table 3.5 Cooling tower makeup water consumption Data Sources Feedstock Purification method Gal water/ gal biodiesel Virgin oil Dry wash (silicate) 0.12-0.15 Virgin oil Recycle water wash 0.19-0.21 A Waste cooking oil Dry wash (silicate) 0.27-0.3 Waste cooking oil Recycle water wash 0.33-0.36 B Multi-feedstock Dry wash (silicate) 0.03-0.05

Currently, no information is available on to what extent dry wash and water wash are adopted among biodiesel producers in either nationwide level or state level. Therefore, three scenarios were calculated for estimating the biodiesel production related water input (W3) in this study

(Table 3.6). For water washing, two scenarios were set up. In the high case scenario, the value of

0.26 gal/gal from United Soybean Board (2010) was used for washing water consumption because it is a documented value and close to the estimation given by NBB staff (Scott, 2010). In the low case scenario, the washing water consumption was assumed to be 0.12 gal/gal, as calculated from our survey results. Accordingly, the cooling tower makeup for these two

61 scenarios were set as 0.275 gal/gal, using the averaged estimation for “recycled water wash”

(Smith, 2011). On the other hand, for dry wash, the washing water consumption was literally zero and cooling tower makeup was also lower (0.153 gal/gal, Smith, 2011). In order to better reflect the real-world situation, an average of these three scenarios was adopted to represent a mixed application condition of water and dry wash as well as their related cooling tower makeup consumptions for the biodiesel industry. The resulting value was assumed to be applicable to each state during the calculation of W3.

Table 3.6 Different W3 estimation scenarios Washing Water Cooling Tower Makeup W3 Scenario (gal/gal) (gal/gal) Water wash 0.26 0.275 (High case) Water wash 0.12 0.275 (Low case) Dry wash 0 0.153

Figure 3.9 presents total annual water consumption during the biodiesel manufacturing stage

(assuming the same purification and process water consumption rate apply) for the applicable states. Four states were excluded (Colorado, Montana, Vermont, and Wyoming) due to the fact that there was no biodiesel plant available in those states. Range of W3 varied from 0.09 to 208

MMgy, with a weighted nationwide average of 77.8 MMgy based on biodiesel plant capacity.

However, since in practice the selection of biodiesel purification method varied among individual biodiesel companies, the actual number and ranking of W3s of the states could be different from this figure. Moreover, considering the trend of switching from water washing to dry wash technologies in the industry, the values are expected to decrease in the near future.

62

1000

100

10

1 W3(MMgy)

0.1

0.01 TX IA MO IL OH IN AR MS WA PA SC NJ ND CA MN GA KY TN MI AZ FL OK WI NY NC LA OR VA AL UT MD SD NH ID DE NE HI KS CT WV RI ME NM NV MA AK State Abbreviation *This figure represents the W3 (MMgy) of 46 states having biodiesel plants. The horizontal line indicates the nationwide averaged value based on these 46 individual numbers Figure 3.9 Annual water consumption during biodiesel manufacturing stage (W3) for 46 states

63

3.3.4 Annual water consumption in total

It is important to see the total water consumption in each stage on a yearly basis. The total quantity (Wtot) of consumptive water as the sum of water consumption from three stages, i.e. the sum of water consumption in irrigation (W1), soybean –to-soybean oil processing (W2) and biodiesel production (based on capacity, W3), was summarized for each state.

Figure 3.10 shows the total consumptive water (Wtot) for soybean to biodiesel process for each state (in million gallons per year, MMgy). There were totally 49 states included in this figure.

The only state excluded was Wyoming, which had neither soybean growth/processing nor biodiesel plants. The average water use in the US was estimated as 808.7 MMgy, and the states with the top 10 water use are listed in Table 3.7. In the states having high irrigation water intensity, such as Arkansas, Nebraska, Mississippi, Missouri, Kansas, and Louisana, irrigation accounted for more than 95% of water use. The biodiesel production capacities in these states were relatively low.

Table 3.7 Top 10 states with highest total annual water consumption (Wtot) State Wtot (MMgy) W1/Wtot (%) W3/Wtot (%) Arkansas 16,022.29 99.57 0.27 Nebraska 9,108.01 99.44 0.02 Mississippi 3,770.65 98.52 1.11 Missouri 2,566.48 95.73 2.58 Kansas 2,537.82 99.09 0.06 Louisiana 655.47 97.96 1.06 Minnesota 626.24 85.37 3.81 Indiana 495.08 80.06 8.83 Illinois 493.87 68.63 12.69 Iowa 416.39 46.62 26.41

Another way of assessing the water consumption of entire soybean biodiesel production life cycle in individual states is to normalize all the results down to “gal/gal” scale and the resulting total water consumption Wnt can be a straightforward value for comparison with results in

64 existing literature. In Figure 3.11, normalized total water consumption is plotted for 46 biodiesel producing states. The nationwide average was approximately 62.31 gallons water per gallon of soybean biodiesel produced. States of Nebraska, Kansas, and Arkansas were the three states that were above the nationwide average and the rest states consumed much less water for producing one gallon of biodiesel. The value of 62.31 gal/gal is still much lower than the results in the existing literature (Table 2.1, 2.2), which can be attributed to the fact that (1) state-level irrigation data was used; (2) a more accurate allocation method was applied (taking into consideration the percentage of soybean oil usage for biodiesel production); (3) consistency among data bases was stressed (temporal and spatial consistency) and (4) primary data was used

(up-to-date industry survey).

65

100000

10000

1000

100

10

1

Wtot (MMGy)Wtot 0.1

0.01

0.001 AR NE MS MO KS LA MN IN IL IA TX MI NC SD GA WI DE ND OH MD OK SC KY NJ WA VA TN PA CA CO AL AZ FL NY OR UT NH ID HI CT WV RI ME NM NV MA AK VT MT State Abbreviation

*This figure represents the Wtot (MMgy) of 49 states having soybean growth/processing capacity and/or biodiesel plants. The horizontal line indicates the nationwide averaged value based on these 49 individual numbers Figure 3.10 Total annual water consumption in 49 states (Wtot)

10000

1000

100

10 Wnt (Gal/Gal) Wnt 1

0.1 NE KS AR SD LA MS DE MD NC MO MN MI WI IN GA OK VA IL ND IA AL KY OH SC TN TX NJ NY WA WV MA PA ME FL CT AK CA ID NV OR RI AZ HI NH NM UT State Abbreivation

Figure 3.11 Normalized total water consumption for 46 biodiesel production states (Wnt)

66

3.3.5 Water-stressed areas

Although the definition of water-stressed areas varies, in current study, it is highly relevant to the plant location, i.e. to investigate whether the biodiesel plant is located in a potentially (or currently) water-stressed area. Table 3.8 summarizes three publications which involved assessment of water consumption/withdrawal for the water-stressed areas defined in their scope.

Table 3.8 Summary of water-stressed states from literature Reference States Criteria sustainability index EPRI report AL, AZ, CA, FL, GA, ID, LA, NM, (integrated with projection; mostly (2003) NV, TX, WA based on withdrawal) Hurd et al. AZ, CA, CO, KS, NM, NV, TX, Watershed-level (1999) UT Scown et al. Palmer Drought Index Southwestern US (2011) impact Yang (2010) AZ, CA, CO, FL, GA, NV Decrease in precipitation by projection

The EPRI report (2003) projected the water sustainability stress for US in 2025. In the study, the precipitation that was not lost due to evapotranspiration (ET) was quantified and used as an approximate measurement of available renewable water. The precipitation and potential evapotranspiration (PET) data was collected from 344 climate divisions to cover continental US and was averaged from 1934 to 2002. Based on 1995 data, significant total freshwater withdrawal occurred in the areas such as AR, CA, FL, ID, LA, MO, eastern TX, and eastern WA.

The calculation of withdrawal as a percentage of available renewable water (surface water part) showed that in some regions the ratio was over 100%, which indicated that supplementary water sources (such as natural or manmade flow structures) were often needed. This phenomenon was most notable in southwestern regions of US. In terms of groundwater, the ratio between groundwater withdrawal and available renewable water (groundwater part) indicated the degree of exploitation of this precious reservation of water. A percentage over 100%, in many cases,

67 indicated the occurrence of unsustainable withdrawals; and those over-100% ratios were found mainly in parts of AZ, CA, FL, ID, KS, NE, and TX. From the data above, the authors described a few scenarios based on the increases in population and electricity generation to predict and compare the water demand in 2025. The results showed that the above-mentioned regions were susceptible to the constraints by increased water demands. In addition to limitation by quantity of water, the authors also incorporated several regulatory constraints to develop a Water Supply

Sustainability Index to evaluate the water supply constraints in the US based on the projection.

Six criteria were included, which were: (1) extent of available renewable water development.

The water use was not supposed to exceed 25% of the total available renewable water; (2) sustainable groundwater use. The ground water withdrawal was not expected to exceed 50% of the total available renewable water (groundwater part); (3) environmental regulatory constraints.

No more than two endangered aquatic species were identified in the specific region where water use occurred; (4) susceptibility to drought. The region was considered to be susceptible to drought if its summer deficit during low precipitation years was greater than 10 inches; (5)

Growth of water use. If the “business as usual” water use requirements to 2025 increased current freshwater withdrawal by more than 20%, the region triggered this sustainability concern; (6)

Growth in demand for stored water. If the summer deficit increased more than one inch over

1995-2025, this criterion was triggered. Based on this index and the county-level data, if a county meets any 2 of these criteria, it is defined as “somewhat susceptible” to an unsustainable water supply practice. If three criteria are met, the county is “moderately susceptible”; and if four or more criteria are met, the county is considered as “highly susceptible”. Once again, according to the results, the susceptible areas were mainly located in the southwestern part of the US such as AZ, CA, NM and NV. Other susceptible regions were AL, FL, GA, ID, LA, TX and WA.

68

Hurd et al. (1999) developed a matrix of indicators for assessing the vulnerability of water supply, distribution and consumptive use for 204 watersheds in the US. The indicators included: level of development, natural variability, dryness ratio, groundwater depletion, industrial water use flexibility and institutional flexibility. For in-stream use, water quality and ecosystem support, the authors also proposed an array of indicators to evaluate the changes in flood risk, navigation, ecosystem thermal sensitivity, dissolved oxygen, low flow sensitivity and species at risk. The detailed definition and calculation principles of these indicators can be found in the paper and hence are not elaborated here. From their study, it can be found that western US, specifically AZ, CA, CO, KS, NM, NV, TX, and UT, are vulnerable to water stress. Scown et al.

(2011) studied both water withdrawal and consumption for biofuels. In their study, “drought- prone” areas (for surface water) were defined based on the Palmer Drought Index (NOAA, 2012).

The Palmer Drought Index measures the long-term drought patterns, their duration and intensity, in a specific region. The county-level data for drought occurrence was collected by NOAA and the calculated index was used for mapping the US drought conditions. There are five categories reflecting the different severeness of drought, which are: “Abnormally Dry (D0)”, “Moderate

Drought (D1)”, “Severe Drought (D2)”, “Extreme Drought (D3)” and “Exceptional Drought

(D4)”. In Scown et al. (2011), the areas with D2 or worse for more than 10% of the time in its last 100 years were selected as drought-prone areas (for surface water). For groundwater, 27 states were identified as susceptible to either significant decline in levels, subsidence or both. Accordingly, the maps plotted by the authors for drought-prone areas and groundwater impacts showed that southwestern US was more vulnerable to both of the two water constraints.

Yang (2010) performed the projection of precipitation variability for contiguous US by using historical precipitation data from 1207 climatic stations. The results indicated that States of

69

Arizona, California, Colorado, Florida, Georgia, and Nevada were susceptible to the potential of decreased precipitation in the future.

As is shown above, although the definition of water-stressed areas varies, there are some states in common that are deemed as vulnerable to water constraints. Hence, the States of Arizona,

California, Colorado, Florida, Georgia (Southern Georgia), New Mexico, Nevada, and Texas

(Western Texas) were selected as water-stressed areas for analyzing the water consumption of soybean biodiesel production.

Accordingly, the summary of total annual water consumption (Wtot) was made for water- stressed states in Table 3.9.

Table 3.9 Total annual water consumption (Wtot) for the states in the water-stressed areas State Wtot (MMgy) W1/Wtot (%) W3/Wtot (%) Arizona 17.33 0 100 California 28.33 0 100 Colorado 21.75 98.82 0 Florida 14.16 0 99.46 Georgia 228.54 89.06 10.03 New Mexico 0.54 0 100 Nevada 0.36 0 100 Texas 365.75 56.99 42.76

For the States of California, Arizona, Florida, New Mexico and Nevada, more than 99% of the water consumption was in biodiesel production, since there was very limited soybean growth in these areas, whereas in Colorado and Georgia irrigation still constituted of dominant amount of water due to soybean growth. Specifically, Colorado and Georgia accounted for 0.32% of the total soybean harvested in the US and 0.65% of total water consumption. Colorado had a high value for irrigation intensity, 611 gal/gal, which was 3rd in the US. State of Georgia ranked the

10th with irrigation water intensity of 106.8 gal/gal. Texas, on the other hand, had a more evenly- distributed consumption of water between irrigation and biodiesel manufacturing stages, which

70 was attributed to its high biodiesel production capacity. Texas accounted for 0.13% of total soybean growth in the US and its normalized irrigation water intensity was approximately 190 gal/gal.

Overall, the total water consumption of these states accounted for only 1.7% of the total, which was low considering their total biodiesel capacity being approximately 27.61% of the total.

3.3.6 Limitation of current study

Current study has a breakdown of water consumption in three major stages of biodiesel production life cycle. Each stage was quantified by using corresponding state-level data and uniform allocation factors, with the acknowledgement that these factors did not necessarily reflect the real-life situation. For instance, it was assumed that 17% usage of soybean oil for biodiesel production (Centrec Consulting Group, LLC, 2010) was uniformly applicable for all states. However, in practice, the amount of soybean oil going to the biodiesel plant is not only affected by the needs of local biodiesel plants (e.g. capacity and feedstock preference), but also dependent on whether there are trans-boundary trades of soybean oil for biodiesel production between states. In current study, this uncertainty was not addressed due to insufficient data availability. Similar uncertainties, as mentioned earlier, also existed in W2 and W3 calculations due to the adoption of uniform water consumption factors. To overcome these uncertainties, more in-depth and specific data needs to be collected for the corresponding steps; for example, to conduct a survey for biodiesel purification options adopted by biodiesel plants and collect related water consumption data that could serve as a database for a better estimation of W3 in the specific state.

71

3.3.7 Comparison of current biodiesel purification methods

Comparison among current biodiesel purification methods have been performed by several studies (Berrios and Skelton, 2008; Berrios et al., 2011; Canoira et al., 2008; Faccini et al., 2011;

Karaosmanoglu et al., 1996; Predojevic, 2008). Karaosmanoglu et al. (1996) compared three wet washing media: hot , petroleum ether followed by distilled water and acidic water

(by H2SO4). The authors indicated that washing with distilled water at 50ºC provided best purification results for the three targeted specifications: refined biodiesel recovery (86.3%), water content (0.095%) and acid value (0.4). Canoira et al. (2008) tested magnesium silicate and acidic water for purification of biodiesel made from a mixture of animal fat and soybean oil. The results indicated almost an identical performance of the two methods in removing contaminants while water washing offered higher biodiesel recovery. Berrios and Skelton (2008) studied cleaning efficiency of water washing, Magnesol® adsorption and ion exchange resin. For water washing, three types of water were used: tap water, deionized water and acidic water (5%

H3PO4). It was found that ion exchange resins had little effect on the removal of methanol and glycerides. Also, an increased acid value after purification was observed, which could be the result of ion exchange between resin and soap. The removal efficiency of soap and free glycerin by resins was evaluated and in the best case, the free glycerol concentration was reduced down below the EN 14214 specification. Besides, the loading for resins ranged from 500 L/kg to 720

L/kg. Magnesol® in their study demonstrated similar performance as the ion exchange resins and was also able to bring the level of free glycerol down to the acceptable range while not effective to the glycerides. Magnesol® was also effective in removing methanol, though not able to lower the concentration down to the threshold value by EN 14214. The recommended condition for

Magnesol® was found to be ≥ 0.75 wt% loading and at least a 10-minute purification time. Water

72 washing was the only process that could reduce the amount of both methanol and free glycerin down to the acceptable level. But still, it was not efficacious to removing the glycerides. The authors studied the influence of temperature, agitation speed and water/biodiesel ratio on the purification outcomes. It was shown that the types of water and purification conditions did not affect removal of methanol and glycerin in a significant way. For soap cleaning, however, acidic water demonstrated the highest removal of soap among the three approaches and agitation speed played a role in the removal as well. Based on the findings, the optimal condition for water washing in their study was determined as: ambient temperature, tap water, 200 rpm and water/biodiesel ratio of 0.5/1. Berrios et al. (2011) studied purification of biodiesel made from waste cooking oil. For the adsorption perspective, both Magnesol® and bentonites were tested.

Magnesol® was found to have a positive effect on soap, methanol and glycerol removal. The best condition for Magnesol® was found to be 1 wt% and 600 rpm, and the corresponding removal efficiencies for soap, methanol and glycerol were 92%, 98% and 55% respectively. Bentonites were also capable of treating these impurities plus reducing the acid value. The optimal condition for bentonites were determined as 1 wt% and 200 rpm, the corresponding purification effects for soap, methanol, glycerol and FFA are 40%, 98%, 20% and 30% respectively. For water washing, the two-step washing procedure with 10 wt% water was found to be the optimal practice. There was minimal difference between distilled water and tap water in removing most of the impurities, except for soap removal (94.6% vs 82.0%). Under optimal condition, the removal efficiencies for soap, glycerol and methanol were around 82%, 84% and 99% respectively for both types of water. One downside of the water washing was that the acid value was increased due to the hydrolysis reaction. Glycerol, a new medium for wet washing, was also tried and it displayed comparable removal effects for soap (69.6-81.8%) and methanol (99.2-99.8). In addition,

73 glycerol showed capability of removing water, probably due to their close polarity. However, the increased acid value (a 27.5% increase) and glycerin level could be a concern for its application.

The recommended condition for glycerol purification process was 15 wt% and a two-step contact procedure as well. Thirdly, for the ion exchange experiments, the tested resin was found to have a low resin capacity (205 L biodiesel/kg) but still had a moderate purification effect on the major impurities. Predojevic (2008) studied the influence of different purification methods on biodiesel recovery) and fuel properties made from waste cooking oil. The author performed purification experiments by using silica gel, 5% H3PO4, and hot distilled water. The results showed all three methods provided similar cleaning effects while silica gel and H3PO4 treatments gave a higher biodiesel recovery, and thus were recommended by the author. Faccini et al. (2011) tested two adsorbents (Magnesol® and silica) and two ion exchange resins (Amberlite BD 10 DRY® and

Purolite PD 206®) for their drying washing performances, and as a control group, water washing was also performed, in which acidic water washing (2% H3PO4) was performed at 55ºC first, followed by hot water washing. For soap and water removal, the two adsorbents demonstrated highest purification capacities while for methanol, only samples treated by water washing and

Magnesol® met the specification. The two adsorbents were also more competitive in , bonded and free glycerol. This result indicated that Magnesol® (1%) could be the preferred polishing material for biodiesel purification.

Overall, the advantages of wet washing are: easy to operate, low cost, and capable of dealing with almost all the impurities except for bonded glycerin and glycerides. But its disadvantages are equally clear: emulsion formation, needing post-washing heating, long purification time, and water consumption and wastewater generation. Emulsion forms during the contact between biodiesel and water and breaks down slowly during the settling period. Emulsion not only

74 prolongs the purification time but also leads to the loss of biodiesel product. Also, after wet washing, the biodiesel usually needs to be heated to get rid of the remaining moisture content, which adds extra energy consumption and cost to the producers. More importantly, wet washing consumes considerable amount of water and generates wastewater that contains high concentrations of contaminants. With the increased concerns about biodiesel sustainability, water consumption and wastewater generation are becoming an urging issue for biodiesel industry.

Therefore, dry washing technologies have been developed to solve the problems stated above.

Dry washing process possesses the merits of reduced purification time, water-less process, and increased final biodiesel recovery. But downsides also exist for dry washing methods. For adsorbents like Magnesol®, one concern is the increased purification cost since it cannot be regenerated. Also, it has been found that some fine particle of Magnesol® will stay in the purified biodiesel, potentially causing the friction and deposition issues in the engine. For ion-exchange resins, the elevated acid value after purification is an inherent problem, considering its purification mechanism. So the FFA level should be low enough before ion-exchange resin can be applied.

Besides the above methods, some other technologies and materials are also expected to gain the market share of biodiesel purification in the near future.

Distillation, though more energy-consuming, has gained certain attention in the industry. The final product from distillation process is crystal clear, unlike the biodiesel product from other purification techniques bearing the yellowish color. Also, distillation helps to remove the trace contaminants, such as sterol glucosides, which could enhance the cold-start performance of the biodiesel (Schultz, 2010). Another merit of applying distillation is that it is effective in removing the bound sulfur components that could not effectively be handled by other methods (Borgese

75 and Privitera1, 2011).

Saw dusts and wood chips are the other two types of materials, which are still primarily in lab- scale development or applied merely in community-scale biodiesel production. Saw dusts are the fine powders from lumbering process while wood chips are larger pieces from construction, agriculture, and landfill other than sawmills. These two kinds of materials are both good filtration stuffing. Wall (2009) conducted a research of comparing the effect of different biodiesel purification methods, where a packed bed of saw dusts was tested. The result indicated that sufficient soap removal could be achieved at a biodiesel to saw dust mass ratio no higher than 14.6:1. Although further information, such as characterization of the saw dust and quantitative data on removal of other contaminants, was not available, this study still sheds some light on a very promising and eco-friendly purification method for the biodiesel industry.

Fillingham and Rushworth (2010) submitted a patent application for the design of a purification apparatus for biodiesel using a mixture of wood fragments (chips, shavings and turnings) and diatomaceous earth. The preferred source was oak wood and this novel design was expected to lower the cost for the biodiesel purification step while maintaining a good cleaning effect.

3.3.8 Characterization and regulation of waste water generated through biodiesel purification

3.3.8.1 BOD5 and COD of the waste water

As far as process waste water is concerned, washing water is usually the major part from a biodiesel plant and the major contaminants within the washing water are: residual biodiesel and glycerin, soap, and methanol. Also, cooling tower blowdown, boiler blowdown and water softener discharge are another main portion of process waste water effluent (US EPA2, 2008). If the wash water is directly drained into the sewer pipes, the remaining oil and grease may clog the

76 pipe and cause a lot of backup problems. Also the presence of soap may elevate the pH of the washing water beyond the permitted range, which may damage the infrastructure under certain condition. Another concern is total dissolved solids (TSS) and salt, which may inhibit the bacteria activity as well as the treatment efficiency of the water treatment plant. Finally, these major contaminants are all strong contributors of BOD5 in the wastewater discharge (Austic and

Lobdell, 2009). The result of the literature view (Table 3.10) shows that BOD5 and COD are of highest concern among the research institutes and regulation agencies.

Table 3.10 Comparison of waste water COD, BOD5 data from different refining processes (biodiesel, bio-ethanol, and petroleum diesel) COD (mg/L) BOD (mg/L) Case 5 Biodiesel manufacturing process Dunn (US EPA) N/A 10,000~15,000 US EPA1 (2008) N/A 4,500~37,000 Austic and Lobdell (2009) 16,000~110,000 9,800~75,000 Kato et al. (2005) 60,000 2,900 Jaruwat et al. (2010) 312,000 168,000 Satianpattanakul et al. (2001) 12,110~28,167 875 Merrick & Company (1998) Effluents from different process for a bio-ethanol plant Enzymatic 27,000 13,400 Softwood 37,000 18,300 Counter-current 54,000 29,400 Crude oil refining (petroleum diesel) Sheehan et al. (1998) 11,000 1,300

Table 3.11 COD, BOD5 data of different biodiesel products (Peterson and Moller, 2010)

Biodiesel COD (mg/L) BOD5 (mg/L) Rapeseed Methyl Ester 2,430,000 1,500,000 Rapeseed Ethyl Ester 2,640,000 1,700,000 Soybean Methyl Ester 2,630,000 1,700,000

There are vast differences between individual data of both COD and BOD5 in biodiesel wash water (Table 3.10). Highly variable as the values of COD and BOD5 are, all of them are evidently higher than the normal limitations required by publicly owned treatment works

77

(POTW). The possible reason for such a high level of COD and BOD5 is that the data may be collected from the raw effluent (first several batches of after-washing water) which contains great amount of impurities. Also, factors such as operational parameters, system configuration

(batch/continuous), washing methods (water recycling or not) and manufacturing capacity all have a direct impact on the impurity concentration in the crude biodiesel and hence, the strength of the washing water, especially for the first several batches. In addition, another factor of significance is the management of the crude glycerin. The BOD5 of crude glycerin could be as high as 1,000,000 mg/L (Dunn, US EPA), so if the separation and purification technologies are effective enough, less glycerin will escape into the sewer pipes and accordingly BOD5 in the effluent can be much lower. Alternatively, if the plant could not afford or does not plan to spend time and energy on glycerin refinery, several options may exist depending on the local policy: (1) to discharge crude glycerin with process wastewater to a POWT. Usually, pretreatment is required considering the high BOD and COD of the effluent. Nowadays, this option is rarely available due to the increasingly stringent regulation of the municipal sewage discharge. (2) To transport the crude glycerin to a centralized waste treatment facility. When discharging glycerin with wastewater is not allowed in some case, biodiesel plants could send their crude glycerin to a centralized waste treatment facility (US EPA1, 2008). Besides glycerin, methanol also plays an important role in waste water quality since it’s found that the residual methanol in the washing water causes an increase of BOD and COD as well (Austic and Lobdell, 2009). Escaping biodiesel, though in very low amount, could contribute to the BOD5 and COD too (more than one order of magnitude higher; Table 3.11).

78

Another finding from Table 3.10 is that biofuels (biodiesel and bio-ethanol) usually have much higher BOD5 and COD level. However, due to limitations on the number of data sources (subject to the lack of statistical significance), missing of underlining information associated with where the waste water samples were taken and inconsistency in reporting the values, such as in the unit of “lbs BOD5/1000lbs crude oil” (EERE, 2007); no quantitative conclusion and explanation could be drawn so far.

3.3.8.2 Review on current policies and regulations

There are EPA documents (US EPA1, 2008; US EPA2, 2008) that provide guidance and regulations pertaining to a biodiesel plant discharge, from which the applicable water-related policies are summarized below.

Clean Water Act

Clean Water Act (CWA) regulates the permits for direct/indirect discharge of waste water and waste water land application. As a matter of fact, most of the waste water discharge associated with a biodiesel plant is categorized to indirect discharge (discharge to waste water treatment facility instead of directly going to the water body again). On account of their capacities, many biodiesel plants are considered as significant industrial user (if the wastewater discharge >25,000 gallons/day or it takes up 5% of the receiving POTW’s treatment capacity), which is subject to the general pretreatment regulation, 40 CFR Part 403.Therefore, if a biodiesel plant is to discharge its wastewater to a POTW, in most cases a permit is mandatory, which requires the effluents to be pretreated (either by the plant or by the POTW, depending on the local regulation) before being discharged into the waste water collection and treatment systems, in case of the shock caused by overloading. In practice, the standards may be different among individual

POTWs, which results in the fact that no uniform limitation for wastewater discharge from

79 biodiesel plants exists throughout the US (Dunn, USEPA). Another point of concern is the storm water discharge and this portion of water effluent is subject to the uniform standards under

NPDES regulations since in this case biodiesel manufacturing facilities are generally considered under SIC code 2800 that stands for “Chemical and Allied Products” (US EPA1, 2008). Land application is another possible way of managing the waste water from biodiesel plants. However, there are currently no particular federal regulations that cover this part of practice and usually, an

NPDES permit is required.

For the direct discharge (to water body), though seldom applicable to biodiesel industry, is subject not only to the technology-based effluent limitations but also the water quality-based limitations, since the former is not stringent enough. Hence, typically NPDES permit writers would develop site-specific limitations that integrate both of the two concerns above.

The CWA also requires a biodiesel plant to prepare fully for the potential spill issues through

“Spill prevention, control and countermeasure regulation” (40 CFR Part 112). The criteria for the compliance requirement for a biodiesel facility are: (1) the facility is non-transportation related;

(2) the total above-ground oil storage capacity is larger than 1,320 gallons or the buried storage capacity is greater than 42,000gallons; (3) reasonable expectation of the oil spill exists, which is judged upon the location of the facility. Once the plant is regarded subject to this regulation, spill prevention plan must be prepared and implemented when necessary.

Safe Act

If the biodiesel plant provides drinking water to more than 25 people per day for at least 60 days per year through its own drinking water supply source, Public Water System Supervision

80 program (PWSSP) will apply. PWSSP is part of the (SDWA) that regulates the construction, modification and operation of the water supply system. Besides the drinking water, PWSSP may also cover other water-supply issues such as cooling water and industrial processing water supply in the state-level. The examples are: (1) water use permit that regulates the withdrawal or using the water from surface or underground source requires; (2) well construction permit that allows the drilling or modification of a well as the water source; (3) operator certification that secures the public water system is operated properly; (4) routine testing that ensures the water quality is in compliance with the relevant standards. The

Underground Injection Control Program (UICP) is another part of the SDWA that may be applicable to a biodiesel plant in a certain circumstance. The applicability criteria are: (1) the disposal of storm water, cooling water or other fluids is performed by means of injection well; (2)

On-site sanitary waste disposal system exists with the capacity of serving 20 or more people; (3) the on-site sanitary waste disposal system is receiving more than the sanitary waste stream alone;

(4) a remediation process is undergoing where fluids are entering the subsurface through an injection well for the purpose of facilitating the cleanup process. The function of UICP is to secure that the injected fluids would not endanger the underground drinking water sources.

81

Chapter 4

Kinetic Study on Esterification of FFA in Waste Cooking Oil by H2SO4

4.1 Goal and Scope

The goal of this part of the study is to: (1) investigate the effect of the operational parameters on the outcome of the esterification reaction; (2) study the kinetics of the reaction; and (3) provide an recommendation for the optimal configuration of the operational parameters based on the experimental data.

4.2 Methodology

4.2.1 Materials

The waste cooking oil was collected from the restaurants in the Cincinnati Zoo & Botanical

Garden. The FFA concentration in the waste cooking oil was 5±0.5 wt%. Chemicals used were sulfuric acid (HPLC grade, 99.8%, Pharmco-Aaper), methanol (HPLC grade, 99.9%, Pharmco-

Aaper), toluene (HPLC grade, 99.8%, Tedia) and isopropyl alcohol (HPLC grade, 99.8%,

Pharmco-Aaper), KOH pellet, and phenolphthalein (Fisher Scientific).

4.2.2 Experiment setup

Figure 4.1 Experiment setup

82

A 1000 ml round bottom flask was used as the reactor for the experiment and a 4000 ml beaker was used to hold the flask in a water bath (Figure 4.1). A hotplate with magnetic stirring was used to supply heat and agitation. The agitation speed was set at 600 rpm to secure an acceptable mass transfer during the reaction (Berrios et al., 2007). A condensation apparatus was placed on top of the flask to prevent the methanol vapor from escaping. A thermometer was inserted in the flask to monitor the temperature and glass pipettes were used to take samples. The sample size was 3 ml and the sample vial was put in the freezer immediately after the sample was taken.

4.2.3 Experiment matrix

The amount of methanol, dosing of catalyst, and temperature were varied to determine the optimal condition. The amount of methanol used for each experiment run was determined in the unit of “methanol-to-FFA molar ratio” and the range was 20:1 to 60:1 with 10:1 increment. The dosing of H2SO4 ranged from 5-15 wt% (to FFA) with 0.5 wt% increment. Temperature was varied from 35 °C to 65 °C with 10 °C increment. The reaction time was fixed at 120 minutes and the sampling schedule was: 5, 10, 20, 30, 60, 90 and 120 minutes after the initial mixing of the reagents and catalyst.

4.2.4 Analytical methods

4.2.4.1 Acid value

The FFA level in the waste cooking oil was expressed in the form of acid value (mg KOH/g oil).

A titration method was applied to determine the acid value of the sample by using the AOCS method Cd 3d 63 (AOCS, 2007). The solvent for titration was a mixture of toluene and isopropyl alcohol at the blend ratio of 1:1 (v/v). A 0.1 N KOH solution was used as the titrant and 1% (w/v) phenolphthalein solution (in isopropyl alcohol) was the indicator. The acid value was calculated by the following equation:

83

56.1 Acid Value  (A  B) N  W (Eq.1)

Where:

Acid Value: mg KOH/g of sample;

A: volume, ml of KOH solution used in the titration;

B: volume, ml of KOH solution used in the neutralization of solvent;

N: normality of KOH solution;

W: mass, grams of the sample

4.2.4.2 Calculation of average FFA molecular weight

Since the amount of methanol to be used for each batch was calculated based on methanol-to-

FFA molar ratio, it’s necessary to know the molecular weight of FFA. But in reality, it’s not cost-effective to analyze the composition of the FFA for every batch in the lab; so instead, an average molecular weight of FFA was calculated in the following way:

1) Assume the molecular weight of triglycerides to be 872.4 g/g before being hydrolyzed into FFA (Freedman et al., 1986)

2) As can be seen from Figure 2.5, after hydrolysis, the resulting FFA would have a molecular weight of 278.4 g/g mole.

This approximation was very accurate, which agreed well with the measured value in a newly published paper (278.102 g/g mole, Jain et al., 2011).

4.2.4.3 Conversion between acid value and FFA mass percentage

Acid value (mg KOH/g) is a scientific expression of the FFA level in the oil, however, the mass percentage (wt%) is more intuitive in some cases. So it would be helpful to achieve an empirical conversion between these two expressions:

Example:

84

So,

4.2.4.4 Kinetics

Theoretically, the acid catalytic esterification is a reversible reaction. However, the presence of significantly excessive methanol will suppress the backward reaction and thus the reverse reaction (hydrolysis) could be neglected when doing the kinetics analysis (Berrios, 2007). Hence, the conversion of FFA could be considered as pseudo-first order and the reaction rate could be represented in the following equation:

d[FFA]   k[FFA] (Eq. 2) dt

Where:

[FFA]: instant weight percentage of free fatty acid in the oil, wt%;

t: time, minute

After integration:

C ln  kt C 0 (Eq. 3)

Where:

C: instant weight percentage (concentration) of free fatty acid in the oil, wt%;

C0: initial weight percentage (concentration) of free fatty acid in the oil, wt%;

85

Thus, the rate constant k could be obtained from the slope of the graph “ln(C/C0) vs t”. The activation energy (Ea) and pre-exponential factor (A) of the reaction under different parameter configurations were determined via the Arrhenius equation:

Ea  k  Ae RT (Eq. 4)

Where:

k: rate constant, min-1;

A: pre-exponential factor, min-1;

Ea: activation energy, J/mol;

R: gas constant;

T: Temperature, K

After arrangement:

(Eq. 5)

Therefore, from the “ln(k) vs T-1” plot, based on the given pairs of k and T, Ea and A could be determined from the slope and Y-axis intercept of the graph, respectively.

4.2.4.5 Conversion rate

A cut-off value of 1 wt% FFA was chosen for a successful FFA conversion, which was equal to approximately 2 mg KOH/g in the form of acid value. Another expression of this threshold was the conversion rate, which was calculated in the following way:

(Eq. 6)

For the oil with 5±0.5 wt% FFA, an 80% conversion would be needed for the final FFA to be 1 wt% (2 mg KOH/g).

86

4.2.5 Procedure for material preparation

4.2.5.1 Titration materials preparation

1. Prepare the solvent by mixing 400 ml toluene and 400 ml isopropyl alcohol in a 1000 ml flask.

2. Prepare the 1% w/v phenolphthalein solution by dissolving 2 g phenolphthalein powder in 200 ml isopropyl alcohol. Stirring and slight heating is needed; stop when the solution is clear.

3. Prepare the 0.1 N KOH solution by dissolving 5.61 g KOH pellet in water.

4.2.5.2 Reactant and catalyst preparation

1. Acid value determination (using 40% of the standard solvent and indicator amount) a) Add 50 ml solvent in a 250 ml flask b) Add 0.8 ml phenolphthalein solution c) Add certain volume of KOH solution which serves as the blank value B in Eq. 1. The criterion for a proper amount is that after 5 minutes the color of the liquid still remains faint pink. d) Add 3 ml of waste cooking oil e) Start titration and keep adding KOH solution until the color of the liquid turns peach and remains stable for at least 30 seconds f) Use Eq. 1 to obtain the acid value; the corresponding mass concentration of FFA is approximately half the acid value (explained above).

2. Amount of chemicals for each batch a) According to the acid value and its corresponding wt% of FFA, calculate the mass of FFA in mole b) Determine the volume of methanol to be used by Eq. 7.

(Eq. 7)

c) Determine the volume of sulfuric acid to be used by Eq. 8.

87

(Eq. 8)

4.3 Results and Discussion

4.3.1 Characterization of original samples

Table 4.1 Major composition of waste cooking oil samples from Cincinnati Zoo Components Sample 1 (wt %) Sample 2 (wt %) FFA 5.1 4.9 Water 0.16 0.15 Triglycerides (TAG) 75.59 78.05 Diglycerides (DAG) 16.96 17.63 Monoglycerides (MAG) 0.94 0.96

Original oil samples were sent to a third-party laboratory to perform the characterization analysis.

As shown in Table 4.1, mass concentrations of FFA, TAG, DAG, MAG and water are tested.

FFA measurement (AOCS Cd 3d 63) is in good agreement with the results in obtained in the lab

(5±0.5 wt %). All the tests for other components are performed through the ASTM standard methods.

4.3.2 Effect of Temperature

Figures 4.2~4.5 reflect the effect of temperature on FFA conversion. As is indicated, FFA reduction rate increased with the rise of temperature. For instance, in the case of 40:1, 10 wt%

H2SO4, the final FFA level of 35 °C was 2.286 mg KOH/g while that of 65 °C condition reached down to 0.374 mg KOH/g. The similar phenomenon was also observed for other methanol-to-

FFA molar ratio conditions, which implied a positive effect of temperature on the FFA reduction process. Moreover, the time for the reaction to reach the cutoff value was also observed to be related to the temperature. For instance, the time was about 20 minutes at 65 °C and 60 minutes at 45 °C. Therefore, it would be possible to make a balance between heating temperature and time duration to achieve the cost-effectiveness from an energy saving standpoint. In addition, it

88 was indicated that a threshold of methanol-to-FFA molar ratio might exist since in cases of 20:1 molar ratio even the highest temperature (65 °C) was not able to bring the acid value down to 2 mg KOH/g.

Figure 4.2 Effect of temperature on the FFA reduction (methanol-to-FFA molar ratio: 50:1, H2SO4 concentration: 10 wt%; dashed horizontal line: cutoff line for FFA ≤ 2 mg KOH/g)

Figure 4.3 Effect of temperature on the FFA reduction (methanol-to-FFA molar ratio: 40:1, H2SO4 concentration: 10 wt%; dashed horizontal line: cutoff line for FFA ≤ 2 mg KOH/g)

89

Figure 4.4 Effect of temperature on the FFA reduction (methanol-to-FFA molar ratio: 30:1, H2SO4 concentration: 10 wt%; dashed horizontal line: cutoff line for FFA ≤ 2 mg KOH/g)

Figure 4.5 Effect of temperature on the FFA reduction (methanol-to-FFA molar ratio: 20:1, H2SO4 concentration: 10 wt%; dashed horizontal line: cutoff line for FFA ≤ 2 mg KOH/g)

90

4.3.2 Effect of Catalyst Concentration

Figure 4.6 reflects the influence of catalyst concentration on the conversion of FFA at a given set of methanol-to-FFA molar ratio, which is 40:1 in this case. As can be found, the conversion rate went up with the increase of H2SO4 concentration from 5 wt% to 12.5 wt%; however, after the catalyst concentration went beyond 12.5 wt% the conversion rate started to drop, which indicated the existence of an optimal concentration. The phenomenon could be attributed to the residual catalyst that caused the need of more KOH for neutralization. A similar observation was obtained by Ramadhs et al. (2005) who found in his study the highest conversion was achieved at around 2.9 wt% (to FFA) and beyond that point the conversion was reduced.

100.00%

80.00%

60.00%

65 °C hour FFA hourFFA Conversion

- 55 °C 2 40.00% 45 °C 35 °C

20.00% 5 7.5 10 12.5 15 Acid Catalyst Concentration (wt%) Figure 4.6 Effect of acid catalyst concentration on 2-hour FFA conversion rate (methanol–to- FFA molar ratio 40:1, the dashed horizontal line: cutoff line for conversion of 80%)

4.3.3 Effect of Methanol-to-FFA molar ratio

Figure 4.7 illustrates the influence of methanol-to-FFA molar ratio on the conversion rate. The average molecule weight of FFA in our study was determined as 278 g/g mol, which agreed pretty well a literature value (Jain, 2011). The conversion rate, as shown in the figure, rose up

91 proportionally with the methanol-to-FFA molar ratio, which was more pronounced from 20:1 to

40:1. For example, when the methanol-to-FFA molar ratio was 20:1 under 65 ºC the conversion was 71.43 %, while that of 40:1 case was 96.29%. As mentioned previously, none of the experiments under 20:1 proceeded to the acceptable 80% conversion, which indicated the existence of lower limit for methanol-to-FFA molar ratio. On the other hand, the increase of the molar ratio did not show significant increase in the conversion after 40:1, which might partially be due to the high conversion rate under those ratios that made the difference less distinguishable.

An ANOVA single factor analysis was performed to evaluate if there were statistically significant differences of conversion rates among various methanol-to-FFA molar ratios. The p- values for 40:1 to 60:1 ratios at 45 ºC, 55 ºC, and 65 ºC were 0.35, 0.40, and 0.08, respectively.

All of the values were all larger than the significance level at 0.05, indicating the differences of conversion rates were not statistically significant. Therefore, the optimal methanol-to-FFA molar ratio in this case was determined as 40:1.

100%

80%

60% hour FFA ConversionFFA hour

- 65 ºC 2 40% 55 ºC 45 ºC

35 ºC

20% 15 25 35 45 55 65

Methanol-to-FFA Molar Ratio Figure 4.7 Effect of methanol-to-FFA molar ratio on 2-hour FFA conversion rate (10 wt% sulfuric acid, the dashed horizontal line: cutoff line for conversion of 80%)

92

4.3.4 Kinetic Calculation

Table 4.2 Activation energy (Ea) and pre-exponential factor (A) of the experiments Methanol-to-FFA Sulfuric acid Ea (J/mol) A R2 molar ratio (wt % to FFA) 40 15% 38798.94 19747. 85 0.99 40 12.5% 20746.76 43.25 0.99 40 10% 24440.67 207.41 0.96 40 7.5% 34370.91 6666.83 0.96 40 5% 42007.32 97831.32 0.95 60 10% 29235.35 1241.90 0.96 50 10% 23132.04 128.32 0.97 40 10% 24440.67 207.41 0.96 30 10% 43106.43 150843.8 0.95 20 10% 45936.51 179512.5 0.96

Table 4.2 summarizes the activation energy (Ea) of whole experimental array. The R2s indicated a quite good fit of the pseudo-first order model. The range of Ea was 20.7-45.9 kJ/mol. This range is also within the range reported in the existing literature (Aranda et al., 2008; Berrios et al.,

2007, 2010; Sendzikiene et al., 2004; Thiruvengadaravi et al., 2009). As is seen in Table 4.2, when the catalyst concentration (10 wt%) was fixed, the Ea was reduced proportionally with the methanol-to-FFA ratio, except for the condition of 60:1, the reason of which still remained to be investigated. On the other hand, when under a fixed methanol-to-FFA molar ratio (40:1), the Ea decreased with the increase in catalyst concentration until it reached 12. 5 wt%. Further increase resulted in a rise of the Ea, which was also considered as being influenced by the excessive

H2SO4 leftover. From the result of kinetic calculation, an optimal configuration of the methanol and H2SO4 could be found as 40:1, and 12.5 wt%.

93

Chapter 5

Utilization of WCGs for Biodiesel Production

5.1 Goal and Scope

The ultimate goal of this part of the study is to utilize waste coffee grounds (WCG) in a closed- loop manner for biodiesel production: oil extraction from WCG for feedstock supply, purification of crude biodiesel by using WCG (after extraction) as a polishing material, and heat/power generation by burning WCG (after purification). However, due to the tight schedule, only oil extraction and purification effectiveness were tested in current study and the results presented in the thesis are based on the preliminary experiments. Considering the wide scope of this research, a detailed future work plan is added at the end of this chapter instead of Chapter 6 to better connect the current result with the proposed future work.

5.2 Methodology

5.2.1 Materials

WCGs were collected from Starbucks’s “grounds for your garden” program in the West Campus of University of Cincinnati. Solvents, such as n-hexane (99.9%, HPLC grade), isopropanol

(99.9%, HPLC grade), and heptane (99.9%, HPLC grade) were purchased from UC chemistry stockroom. Crude biodiesel for analysis was from BlueGrass Biodiesel® (KY, USA). The commercial adsorbent, D-Sol (C400)TM, was purchased from Dallas Group of America, Inc (TX,

USA) and the ion-exchange resin was from ALX Enterprises, LLC.

94

5.2.2 Analytical methods

5.2.2.1 Total oil extraction ratio

Total oil extraction ratio was defined as the mass percentage of extracted oil (containing both lipids and FFAs) to the initial weight of WCG. The mass of the extracted oil was measured on an analytical scale after solvent recovery.

5.2.2.2 Composition analysis (oil and FFA)

Characterization (composition of both oil and FFA) and quantification were performed in reference to the methods provided in Kondamudi et al. (2008).

5.2.2.3 ASTM specifications

Acid Number (AN)

The acid number was determined through AOCS Cd 3d 63, as described in Chapter 4.

Free glycerin

The free glycerin was measured by a HP gas chromatography (Model 5890) equipped flame ionization detector (FID) and an auto-sampler (Model 7673) in accordance to ASTM D6584 method. The column used was Restek Rtx-Biodiesel TG column (10 m*0.32 mm*0.1 um) and column temperature was programmed as following: 1) initial temperature: 50 ºC, hold 1 min; 2) rate 1: 15 ºC/min to 180 ºC; 3) rate 2: 7 ºC/min to 230 ºC; 4) 30 ºC/min to 380 ºC, hold 10 min.

The temperature of FID is set at 380 ºC. The carrier gas is helium and the flow rate was 3 mL/min.

Moisture

The moisture content in biodiesel was determined by volumetric Karl Fischer Titration method.

The pyridine-free Karl Fischer titrant was used and the measurement was carried out by following ASTM D 4377 method.

95

Methanol (MeOH)

Methanol content was measured by GC using method EN 14110.

Elements

Elements, such as Na, Ca, K, Mg, P and S were determined by ICP-OES following EN 14538,

ASTM D5453 and D 4951, respectively.

5.2.3 Oil extraction

Oil was extracted by soxhlet process and the general procedure was as following: the WCG was baked in the oven overnight to remove the moisture. Oil extraction was then performed through soxhlet process with different solvents under the designated time and ratios. After extraction, the oil was then separated from the solvent through the rotary evaporator and the solvent was recovered for the next batch. The resulting oil was analyzed in the lab.

5.2.4 Post-extraction WCG as the purification material

The designated amount of WCG (after extraction; baked to remove trace solvent) was packed in the graduated column. The purification experiment was performed by flowing crude biodiesel through the column. In the end of each purification run, a sample was withdrawn to check the purification effect for each ASTM specification.

5.3 Results and Discussion

5.3.1 Oil extraction

Total oil extraction ratio

Seven batches of oil extraction with heptane were performed under short extraction time. The averaged extraction ratio was 8.37%, i.e. 8.37 g oil out of 100 g WCG. The prolonged extraction time improved the extraction ratio to a certain extent, 8.88%, 9.23% and 11.62%, with different solvents or combination. This could also be observed in Figure 5.1, where the concentration of

96 oil in the solvent increased with extraction time in the case of hexane/isopropanol mixture. The blend of n-hexane and isopropanol offered the highest oil extraction performance (Table 5.1).

This could be ascribed to the fact that oil was more soluble in the non-polar solvent (n-hexane) while the solubility of FFA was higher in the polar solvent (isopropanol), and the blend took advantage of both properties (Berezin et al. 1996). In addition, the higher extraction ratio by pure isopropanol than pure n-hexane might indicate considerable amount of FFA present in the WCG oil. Yet, due to the tight schedule and system maintenance, only three configurations of solvents were tested and the extraction time was fixed at seven hours. The detailed characterization (the complete fatty acid profiles) of extracted oil and FFA is still underway and thus quantitative conclusion will be available later.

Table 5.1 Results of extraction Solvent Extraction Time (hr) Extraction Rate (wt%) Note Averaged from 7 Heptane 0.5-0.75 8.37 batches Hexane 7 8.88 Isopropanol 7 9.23 Hexane/Isopropanol 1:1 v/v mixture 7 11.62 (1:1)

14

12

10

8

6

4

2

0 Oil Concentration in Solvent Concentration the (wt%) Oil 3 4 5 6 7 Time (hr)

Figure 5.1 Oil concentration in the solvent vs. extraction time (solvent: hexane/isopropanol 1:1 v/v mixture)

97

Composition

Depending on how the WCG was collected (e.g. time between disposal and collection, environment for disposal), water content in our WCG samples varied from 20% to 50%. On a dry basis, WCG in our study had 8-12% oil content. Ultimate analysis (by third party laboratory) showed that dry WCG was composed of 54.26% carbon, 7.3% hydrogen, 2.38% nitrogen, 35.3% oxygen and 0.15% sulfur. The remaining constitutes included proteins, sugars, cellulose and hemicellulose. The metal and other element test results on the extract coffee oil are listed in

Table 5.2. Our preliminary composition results were comparable to other studies (Barkenbus et al. 1927, Daglia et al. 2004, Kondamudi et al. 2008, Mussatto et al. 2011). Table 5.2 shows the elemental analysis of one coffee oil sample from current study.

Table 5.2 Elemental analysis of selected coffee oil sample Oil P Ca Mg Na K S Instrument Coffee Oil 66 188 30 2 10 18 ICP-OES * All the values are in the unit of ppm

5.3.2 Purification effect on crude biodiesel

Three purification experiments were performed with the WCG-to-crude biodiesel ratio fixed at

1:10 (wt/v). As a preliminary study, three conditions of WCG, air dried alone, baked, and baked

& extracted, were tested to see the difference in the purification performance. Four major parameters (acid number, moisture, methanol and free glycerin) were measured in the first run, while in the second run two more parameters (Na+Ca content) were included to obtain a better understanding of the purification effect. In the third run, the pool of parameters was further enriched by including K, Mg, P and S contents. All the results are summarized in Table 5.3.

98

Purification with WCG (no oil extraction)-1st run

The first run consisted of two parallel experiments. The WCG used in the first run did not undergo the solvent extraction. The “Air-dried” WCG was dried in the hood for a week while the

“Baked” WCG was prepared via both air drying and subsequent baking. Acid number (AN) is the reflection of the FFA concentration in the biodiesel. It’s noteworthy that although none of the

ANs in the purified biodiesel exceeded the ASTM limit (bottom row in each section), still, the purification through WCG tended to slightly increase the acid level in the resulting biodiesel.

This could be attributed to the leaching out of the FFA from the oil contained in the WCG, since in these experiments the WCG was not extracted by solvent in advance. The changes of moisture after purification show that baking is necessary because air-drying was not efficient enough to remove the moisture in the WCG and thus the moisture of the biodiesel after running through the column was raised from 0.040% to 0.066%. On the other hand, the baked WCG showed capability of absorbing the residual moisture in the crude biodiesel by reducing the moisture from 0.040% down to less than 0.02%, which meets the specification in ASTM standard.

Methanol (MeOH) and free glycerin removal effect were found to be significant in the first run.

Both air-dried and baked WCGs displayed considerable ability to improve the two specifications.

For methanol, the final concentration was 0.01% and 0.07%, both of which are lower than the threshold in the ASTM standard. Similarly, for free glycerin, the concentration was reduced to

0.019% and 0.025%, respectively. The free glycerin levels of the two samples were very close to the limit (0.02%), which indicated the opportunity to improve the removal by increasing the usage of WCG (higher WCG-to-crude biodiesel ratio).

99

Purification with extracted WCG-2nd run

Based on the previous results, the second run of the purification study was conducted with the after-extraction WCGs and a commercial adsorbent (C400). Hereafter, “H-1” labels the WCG from n-hexane extraction while “I-1” represents the data of WCG from isopropanol extraction.

This time the crude biodiesel was literally free of FFA and thus the AN was zero. As indicated in the first run, the FFA within the WCG could contribute to the rise in the final AN in the purified biodiesel, which was resonated by the results of second run. After going through the column of baked WCG, the AN in the biodiesel was raised to 0.1. The similar phenomenon was observed in the case of I-1, though the WCG in the I-1 experiment has been extracted by isopropanol. This may be explained by the leaching out of remaining FFA that has not been fully extracted by isopropanol during the soxhlet process. But the AN of biodiesel purified by H-1 did not show any increase, which disagreed with the conclusion in the reference (Berezin et al. 1996) and findings above. This uncertainty needs to be clarified through further studies (triplicate experiments). Fortunately, the raise of AN by after-extractin WCG was not too high so the final biodiesel product still met the ASTM standard. For the moisture, it should be stated that since the initial moisture of the crude biodiesel was lower than the detection limit (0.02 v%), the effectiveness of WCGs and C400 to remove the carry-over moisture from the crude biodiesel was not clearly proved in the second run. Also, similar significant reductions for MeOH and free glycerin were observed in the second run. But neither of the two after-extraction WCGs managed to eliminate the residual MeOH down to the acceptable level, unlike the C400. This disagreement (with the finding in the first run) could be attributed to the much higher initial

MeOH residue than that in the first run (0.720 vs 0.185 wt%). An increased WCG-to-crude biodiesel ratio (i.e. higher WCG dosing) may help to better address this problem. Similarly, for

100 free glycerin removal, the two after-extraction WCGs were able to lower the concentration down to a level slightly over the ASTM. Finally, the level of Na and Ca ions indicates the residual catalyst from biodiesel producing step and residue of Ca/Mg-containing purification materials, if applicable. In this run, the reduction of metal ions (Na+Ca) was significant in all cases.

Interestingly, in this run, baked-only WCG (not extracted) seemed to be more effective in terms of purification than the after-extraction WCGs, which needs to be verified through future study.

Overall, the WCGs demonstrated comparable capability with the commercial counterpart again in this purification run.

Table 5.3 Summary for three purification runs Acid Free number Moisture Methanol # glycerin (mg (%) (%) (%) KOH/g) Crude Biodiesel 0.35 0.04 0.185 0.055 Air-dried WCG 0.4 0.066 0.01 0.019 1 Baked WCG 0.4 0.02 0.07 0.025 ASTM 0.5 0.05 0.2 0.02 Na+Ca

(ppm) Crude biodiesel 0 0.02 0.72 0.065 160 C400 0 0.02 0.15 0.019 6.1 WCG-hexane 0 0.027 0.3 0.028 6.3 2 WCG-isopropanol 0.2 0.02 0.32 0.04 4.5 Baked WCG 0.1 0.02 0.2 0.027 4 ASTM 0.5 0.05 0.2 0.02 5 Na+K Ca+Mg P S

(ppm) (ppm) (ppm) (ppm) Crude Biodiesel 0 0.02 5.7 0.185 225 32 12 1 ALX 0.5 0.03 0.24 0.01 7.3 4.1 4 1 3 WCG-mixture 0.5 0.05 0.8 0.08 21 17 6 1 ASTM 0.5 0.05 0.2 0.02 5 5 10 15

Purification with extracted WCG-3rd run

In the third run, the WCG underwent the soxhlet process by a solvent mixture of hexane and isopropanol and is thereafter named as “M-1”. Another commercial purification material, ALX, was used as the control group. As is shown in Table 5.3, the ANs in this case increased for both

WCG and ALX. Since ALX is a type of ion-exchange resin, it is likely that FFA was formed by

101 replacing H ion with Na in the crude biodiesel during purification, which could be responsible for this rise. In addition, AN increase by WCG still existed in this run, indicating special treatment may be needed to completely solve the acidic leaching issue.

The elevated moisture level (0.05 %) by WCG was observed and the reason for this uncertainty in moisture removal needs to be clarified in future study. The removal effect on MeOH and free glycerin by WCG was still evident, though neither of the two specifications was brought down to the ASTM threshold. The reason for this could still be high initial concentrations. ALX, on the other hand, also demonstrated considerable capability in removing these two impurities. As for the elements, both WCG and ALX resulted in a significant reduction in the concentration, though

ALX was a little more effective. The purified biodiesel by WCG could successfully meet the specifications of P but failed the Na+K and Ca+Mg concentrations. This may be explained by the influence of initial concentrations too. Similar to WCG, ALX managed to decrease the concentrations of P, and Ca+Mg to the acceptable level too. Due to the detection limit and low initial concentration of S, the effectiveness of removing S by both materials were not clearly revealed.

Table 5.4 summarizes the purification effects of the three materials in a qualitative manner based on the preliminary results. The denotations are set by the following criteria: (1) if a material is positive in removing a specific impurity in all the runs it was involved, a “+” mark is assigned for this specification; (2) similarly, if a material is negative in removing a specific impurity in all the runs it was involved, a “-” mark is assigned for this specification; (3) if conflicts exist in purification effect for a specific impurity, a “/” mark is assigned; (4) if no applicable data is

102 available, a “NA” mark is assigned. So, except for the AN and moisture, WCG was found to be a good purification material for most of the impurities.

Table 5.4 Qualitative summary of purification effects by different materials based on the preliminary results Free Material AN Moisture MeOH Na+K Ca+Mg P S Glycerin WCG - / + + + + + + C400 + / + + NA NA NA NA ALX - - + + + + + + “+”: the material has a positive effect on removing the specific impurity; “-”: the material has a negative effect on removing the specific impurity; “/”: the material’s effectiveness on removing the specific impurity is not clear; “NA”: no applicable data available;

Table 5.5 gives a quantitative summary of the purification effects for five specifications by different materials. Since the set of specifications were being completed during course of the preliminary study, only in the case of M-1 and ALX all five specifications were addressed. As is shown, ALX showed highest performance in all the specifications. Overall, WCGs provided comparable purification capability to the commercial materials for those impurities. Among three

WCGs, M-1 worked better than the other two, especially for MeOH removal, though further work is needed to validate this phenomenon. Moreover, M-1 displayed very close capability to

C400 in MeOH and free glycerin removal. For a more definitive conclusion, however, additional experiments in future study will be necessary to generate more statistically significant data for the comparison.

* The removal of SGs is not tested in the preliminary study and hence will be one of the focuses in the future work.

103

Table 5.5 Removal rates of selected impurities by different purification materials Free Material MeOH Na+K Ca+Mg P Glycerin H-1 58.33% 56.92% NA NA NA I-1 55.56% 38.46% NA NA NA M-1 85.96% 56.76% 90.67% 46.88% 50.00% C400 79.17% 58.56% NA NA NA ALX 95.79% 94.59% 96.76% 87.19% 66.67% “NA”: no applicable data available

Recovery

Recovery rate is another important factor to evaluate a purification material. The recovery rate of biodiesel is calculated as the ratio between the volume of biodiesel received after purification and the initial volume of crude biodiesel. It was observed that for both purification materials tested (C400 and WCG), a “prime” stage usually existed where the recovery rates were only about 80%. After the prime stage, the recovery rates of both materials reached up to almost

100%. With more data collected in future study, a statistic analysis will be performed to see if there is significant difference between WCG and C400 in terms of recovery rate.

5.4 Future work

Based on the preliminary results, future work for utilization of WCG for biodiesel industry is expected to accomplish the following tasks:

Extraction of coffee oil for biodiesel production

(1) Clarifying the uncertainties regarding oil extraction from WCG, such as solvent selection vs extraction outcome, influence of solvent on oil composition, optimal extraction time, etc. More solvents will be tested individually and in different mixtures as well. Under the optimal combination of solvent(s) and extraction time, the maximum oil extraction ratio will be reported and corresponding lipid and FFA composition will be analyzed.

104

(2) Analyzing the biodiesel made from coffee oil. Parameters such as density, viscosity, heating value, and all the ASTM D6751 specifications will be measured.

Purification of crude biodiesel by WCGs (after extraction)

(1) Clarifying the uncertainties that are identified in this preliminary study regarding purification effectiveness. The effect on moisture removal will be verified. Also, the influence of initial concentration of impurities on removal efficiency will be investigated.

(2) Studying the purification effect of individual WCGs that are extracted by different solvents to see if there are statistically significant differences. Comparing the purification effectiveness of

WCG with other purification materials, such as commercial dry purification materials, , wood chips and saw dusts.

(3) Testing the effectiveness in removing S and SGs. Crude biodiesel made from high sulfur- containing feedstocks, such as trap grease, will be polished by using different purification materials to check their removal efficiency on S. For SGs, analytical methods will be developed based on our instrument setup to facilitate the purification experiments.

(4) Studying the properties and purification mechanisms of WCG. More experiments will be performed to obtain the representative values of certain parameters of WCG, such as heating value, BET surface area, heating value, elemental composition, and size distribution. In-depth study will look into several potential purification mechanisms of WCGs, such as adsorption and filtration.

(5) Finding the influence of dosing (WCG-to-crude biodiesel ratio; wt/v) on the purification.

(6) Finding a regression model to link the dosing of WCG, either by itself (100%) or in coalition with commercial materials, with the initial concentrations of the target impurities to achieve a

105 desired purification effect; this model can serve as a useful tool for guiding the economic application of WCG as a polishing material.

Using spent-WCGs (after purification) as an adjunct fuel

(1) Performing combustion tests on burning spent WCG (after purification) to obtain the knowledge of its emission profile to the environment.

Life cycle energy consumption for comprehensive usage of WCG for biodiesel production

(1) To upgrade lab-scale system to pilot-scale;

(2) To collect data from pilot-scale, details include:

Table 5.6 Major data collection and co-product crediting methods Life cycle stages Co-product crediting methods Note WCG Pretreatment Electricity input in oven baking Moisture content Oil Extraction Life cycle energy of solvent Mass-based allocation Between coffee oil and grounds electricity input in soxhlet Solvent evaporation loss Cooling water evaporation loss Extractable oil content Esterification Life cycle energy of chemicals Dosing of chemicals Process energy Reaction efficiency Mass-based allocation Between biodiesel and glycerin Transesterification Life cycle energy of chemicals Dosing of chemicals Process energy Reaction efficiency Biodiesel Purification Energy saving credit due to the Life cycle energy of materials Displacement displacement of dry purification Dosing of materials Process energy material WCG Burning Energy saving credit due to the heating value of spent WCG Displacement displacement of fuel Energy saving from burning WCG

(3) To perform full LCA study based the life cycle energy consumption study above.

106

Chapter 6

Conclusions

This study investigated two major aspects relevant to the sustainability of biodiesel produced from transesterification process.

The study firstly looked into consumptive water use of soybean-derived biodiesel and broke down water use into the following three stages: plant growth (irrigation water use), soybean processing into soybean oil and biodiesel production. Generally, irrigation water use (plant growth stage) tended to be much larger than the soybean processing and biodiesel production stages. The irrigation water use for soybean growth varied significantly from state to state, as some states irrigated the soybean plants while some states did not.

Mass-based allocation was used throughout the calculation and irrigation water intensity was calculated based on state-level data. Totally 35 states were included in calculating irrigation water consumption and the rest were excluded due to either no soybean growth or lack of data from USDA report. The result showed that for one gallon of biodiesel produced from soybean, the normalized irrigation consumption intensity (Wni) was 61.78 gallons on average in the US.

State-level analysis showed that irrigation practice varied significantly throughout the whole US, ranging from literally “zero irrigation” in the states like Maine and Massachusetts, to as high as

1,058 gal/gal in the State of Washington. Although the normalized irrigation intensity was low, in terms of total irrigation water, the highest annual irrigation water consumption was 15, 953

MMgy, with 1,812 MMgy as the nationwide average. This indicated the necessity to switch to less “irrigation-demanding” feedstocks or waste-derived feestocks to reduce the burden on water supply, if biodiesel production capacity continues to grow. The investigation on the selection of irrigation water sources found that the dominant choice was “ground water from well” and few

107 irrigations were exerted through “water from off-farm suppliers”. On the other hand, soybean processing and biodiesel production stages contributed less to the total water consumption while innovations were mostly applied in these stages, such as applying dry wash technologies (e.g. adsorption, ion-exchange). However, according to the feedback from industrial survey, the current status of dry washing technology application was absent, the acquisition of which will be of great value to the promotion and further development in this field. In addition, the water consumption in the biodiesel purification and cooling tower makeup stages was related to the feedstock quality; i.e. the waste-derived feedstocks were often associated with higher water consumption, which was mainly due to the need for extra cleaning efforts and methanol recovery.

Moreover, the water consumption in biodiesel production stage was highly process-dependent and thus remained case-specific as shown by the literature review and industrial survey, which consequently requires efforts to set up an inventory and data update mechanism (e.g. 3 or 5 year survey interval) to come up with a thorough understanding of the impact of biodiesel production on water resource. Overall, the normalized total water consumption for making one gallon of soybean oil was around 62.31 gallons, which is much lower than the values in existing literature.

This could be explained by the fact that more consistent, detailed and up-to-date data bases and accurate allocation methods were applied. In water-stressed areas, since the scale of soybean growth was relatively small, the irrigation and total water consumption for soybean biodiesel produced in these areas were lower than nationwide average. However, if biodiesel productivity is going to expand in these areas, water supply issues should be a concern for the decision makers.

108

The second sustainability aspect investigated in this study was the utilization of waste materials for biodiesel production. Two sub-topics were explored. The first one was the parametric study of acid-catalytic esterification of FFA in the waste cooking oil. In most of the experiments, the

FFA level was reduced down to 1 wt% or less. Various parameters were examined to check their influence on the reaction rate. Temperature was found to have positive impact and the trade-off between temperature and reaction time existed. For the reaction to remain pseudo-first order and irreversible, excessive methanol was needed. The FFA conversion rate was proportional to the methanol-to-FFA molar ratio and the trend was more pronounced from 20:1 to 40:1. Catalyst concentration played an important role in FFA reduction but the rate did not increase monotonously with the increase of H2SO4 concentration, which might be attributed to the excessive catalyst leftover. The kinetic calculation indicated the optimal condition for the parameters could be 40:1-50:1 methanol-to-FFA molar ratio, and 12.5 wt%. Combined with the temperature influence, the optimal condition for esterification of FFA in waste cooking oil could be methanol-to-FFA molar ratio 40:1, catalyst concentration 12.5 wt% and 65 ºC.

The second sub-topic is the utilization of WCG for biodiesel production. Specifically, oil extraction and crude biodiesel purification were tested in the preliminary study. The results showed that WCG contained around 10 wt% extractable oils; coupled with its potential large supply, WCG might become a supplementary feedstock for biodiesel industry. Purification results indicated that WCG possessed comparable purification capacity to commercial polishing materials for most of the target impurities, and thus could be a good alternative to the commercial products.

109

Future work

Extended work for assessing influence of biodiesel production on water resources will be focused on the following:

(1) Detailed “in-state” analysis needs to be performed. Water supply and consumption patterns are often unevenly distributed within a state. For example, in the State of Texas, the precipitation is much higher in east Texas than in west region. Also, in the State of Georgia, the northern part has higher water consumption than the southern Georgia. So an in-state analysis is necessary to better assess the impact of biodiesel production in that state by investigating the factors such as precipitation patterns, water consumption hotspots, and location of biodiesel plants. To illustrate, if a biodiesel plant is located in a region where surface water supply is limited by low precipitation rate, it's likely that groundwater aquifer would be the preferred water source. This however may increase the vulnerability of that region to the potential impacts, such as over pumping and subsidence. Similarly, if a biodiesel plant is to be built in the area where water demand is already high due to human or industrial activities, requirement of additional water supply may lead to the over exploitation of existing surface and groundwater sources and hence put burden on local water resource protection. The states of interest are those with distinct water supply and/or water consumption patterns. For instance, States of Texas and Washington are the two states that have significantly uneven distribution of precipitation. Also, States of Georgia and Florida are the states that have clear consumption-driven patterns. So the in-state analysis of biodiesel production in these states can give a more insightful understanding about the impact of local water resources. To fulfill this future goal, the state specific data such as precipitation and evaporation rates, distribution and regional water consumption patterns as well as biodiesel plant distribution will be collected.

110

(2) Tracking the trans-boundary flow of soybean oil. As mentioned in Section 3.3.6, one limitation of current study is that there is not enough information to track what percentage of local soybean oil is exactly used for biodiesel production by local biodiesel producers. While a uniform factor of 17% was used, it is always preferable to have a more specific number for each state. For example, States of Arizona and California had no soybean growth, so if the biodiesel producers in these two states want to use soybean oil as the feedstock, the trans-boundary import of soybean oil is inevitable. This may result in the increase in percentage of soybean oil for biodiesel in the state that exports soybean oil. More importantly, considering the transportation and storage costs, it’s likely that some of the states have to import soybean oil from nearby states that have high irrigation consumption, which could bring adverse impact on local water supply in those oil exporting states. Therefore, a clear tracking of soybean oil trans-boundary flows can be helpful to obtain a more complete picture of the impact of biodiesel production on water resources in each state.

(3) Data base for biodiesel manufacturing process needs to be enriched. As discussed in Chapter

3, the boiler water makeup was not included in current study. Also, the makeup water consumption was from two biodiesel companies, which is a fairly small data pool. For purification water, although our survey had a pretty high response, the profile of purification options (percentage of dry/water wash application) in an individual state is still desirable for achieving a more accurate estimation of state-level purification water consumption.

The future work for reducing FFA in waste-derived feedstocks could be:

(1) To clarify the uncertainties identified during the kinetic study.

111

(2) To increase the FFA concentration to a higher level to explore the effect of the water on the kinetics of the reaction;

(3) To derive a regression model based on the data from current and potential extended work; the input parameters could include the initial FFA concentration to better predict the outcome of the reaction.

The future work for utilization of WCG for biodiesel production has been stated in Chapter 5,

Section 5.4; and hence, is not iterated here.

112

References

ABC news, Coffee grounds give jolt to crippled Japanese energy grid,

http://abcnews.go.com/Business/coffee-grounds-supply-power-japan-shuts-nuclear-

plants/story?id=13770412#.T6FM-Xg3_Ko, accessed on May 2012.

Akasaka, H., Takahata, T., Toda, I., Ono, H., Ohshio, S., Himeno, S., Kokubu, T., and Saitoh, H.

2011. Hydrogen storage ability of porous carbon material fabricated from coffee bean

wastes. International Journal of Hydrogen Energy. 36, 580-585.

Al-Hamamre, Z., Forester, S., Hartmann, F., Kroger, M., and Kaltschmitt, M. 2012. Oil extracted

from spent coffee grounds as a renewable source for fatty acid methyl ester

manufacturing. Fuel. 96, 70-76.

Allen, R.G., Pereira, L.S., Raes, D., Smith, M. 1998. Crop evapotranspiration - Guidelines for

computing crop water requirements - FAO Irrigation and drainage paper 56. Food and

Agriculture Organization of the United Nations, Rome, Italy.

American Oil Chemist’s Society (AOCS). 2007. AOCS methods for biodiesel feedstock quality.

Technical services AOCS (CD-ROM), AOCS press, Urbana, IL.

Aranda, D. A. G., Santos, R. T. P., Tapanes, N. C. O., Ramos, A. L. D., and Antunes, O. A. C.

2008. Acid-catalyzed homogeneous esterification reaction for biodiesel production from

palm fatty acids. Catalysis Letters. 122, 20-25.

Argyopoulos, P. 2008. EISA 2007: renewable fuels standard program.

http://www.energy.ca.gov/2009_energypolicy/documents/2009-01-

13_workshop/presentations/Paul_Argyopoulos_EISA.PDF, accessed on Jan. 2011.

113

Asbridge, D. 2010. Market production analysis and market outlook-Planting impacts on markets.

http://archive.unitedsoybean.org/expert_advice/the_market_edge.aspx?bid=87869414934

18858837, accessed on March 2011.

Atadashi1, I.M., Aroua, M.K., Abdul Aziz, A.R., and Sulaiman, N.M.N. 2011. Refining

technologies for the purification of crude biodiesel. Applied Energy. 88, 4239-4251.

Atadashi2, I.M., Aroua, M.K., Abdul Aziz, A.R. 2011. Biodiesel separation and purification: A

review. Renewable Energy. 36, 437-443.

Austic, G., and Lobdell, S. 2009 Treatment, Disposal and Reuse of Biodiesel Derived Waste

Products. http://www.docstoc.com/docs/22776999/Treatment-Disposal-and-Reuse-of-

Biodiesel-Derived-Waste-Products, accessed on March 2011.

Baek, M. H., Ijagbemi, C.O., O, S.J., and Kim, D.S. 2010. Removal of malachite green from

aqueous solution using degreased coffee bean. Journal of Hazardous Materials. 176, 820-

828.

Barkenbus, C.; Zimmerman, A. J. 1927. Kentucky coffee nut tree seed oil. J. Am. Chem. Soc. 49,

2061–2064.

Benefits of Biodiesel, National Biodiesel Board:

http://www.biodiesel.org/pdf_files/fuelfactsheets/Benefits%20of%20Biodiesel.Pdf,

accessed on Nov. 2010.

Bengis, R.O., and Anderson, R.J. 1934. The chemistry of the coffee-bean II. The composition of

the glycerides of the coffee-bean oil. http://hwmaint.jbc.org/cgi/reprint/105/1/139,

accessed on May 2012.

Berchmans, H. J., and Hirata, S. 2008. Biodiesel production from crude Jatropha curcas L. seed

oil with a high content of free fatty acids. Bioresource Technol. 99, 1716-1721.

114

Berezin, O.Y., Tur’an Y.I., Kuselman, I., and Shenhar, A. 1996. Rapid and complete extraction

of free fatty acids from oilseeds for acid value determination. JAOCS. 73, 1707-1711.

Berrios, M., Martin, M.A., Chica, A.F., and Martin, A. 2011. Purification of biodiesel from used

cooking oil. Applied Energy. 88, 3625-3631.

Berrios, M., Martin, M. A., Chica, A. F., Marin, A. 2010. Study of esterification and

transesterification in biodiesel production from used frying oils in a closed system. Chem.

Eng. J. 160, 473-479.

Berrios, M., and Skelton, R.L. 2008. Comparison of purification methods for biodiesel. Chemical

Engineering Journal. 144, 459-465.

Berrios, M., Siles, J., Martin, M. A., Martin, A. 2007. A kinetic study of the esterification of free

fatty acids (FFA) in sunflower oil. Fuel. 86, 2383-2388.

Biodiesel plant list, biodiesel magazine: http://www.biodieselmagazine.com/plant-list.jsp,

accessed on May, 2012.

Biodiesel Production, US Department of Energy:

http://www.afdc.energy.gov/afdc/fuels/biodiesel_production.html, accessed on Nov. 2010.

Biodiesel TechNotes. 2010. Ion exchange resins for cleaning biodiesel: how do they work?

http://www.uiweb.uidaho.edu/bioenergy/ionexchange.pdf, accessed on May 2010.

Borgese, C., and Privitera, M1. 2011. “Biodiesel reaction and separation technology”. Biodiesel

Magazine. September Issue.

Borgese, C., and Privitera, M2. 2011. “Quick and dirty feedstock characterization”. Biodiesel

Magazine. August Issue.

115

Boonamnuayvitaya, V., Sae-ung, S., and Tanthapanichakoon, W. 2005. Preparation of activated

carbons from coffee residue for the adsorption of formaldehyde. Separation and

Purification Technology. 42, 159-168.

Boonamnuayvitaya, V., Chaiya C., Tanthapanichakoon, W., and Jarudilokkul, S. 2004. Removal

of by adsorbent prepared from pyrolyzed coffee residues and clay.

Separation and Purification Technology. 35, 11-22.

Boudrahem, F., Aissani-Benissad, F., and Ait-Amar, H. 2009. Batch sorption dynamics and

equilibrium for the removal of lead ions from aqueous phase using activated carbon

developed from coffee residue activated with zinc chloride. Journal of Environmental

Management. 90, 3031-3039.

Bryan, T. 2005. “Adsorbing it all”. Biodiesel Magazine. March issue.

Canakci, M., Van Gerpen, J. 2001. Biodiesel production from oils and fats with high free fatty

acids. T. ASAE. 44, 1429-1436.

Canali, M., and Aragrande, M. 2010. Biofuels versus food competition for agricultural resources:

impacts on the EU farming systems. Studies in Productivity and Efficiency. 7, part 3,

191-209.

Canoira, L., Rodríguez-Gamero, M., Querol, E., Alc_antara, R., Lapuerta, M., and Oliva, F. 2008.

Biodiesel from low-grade animal fat: Production process assessment and biodiesel

properties characterization. Ind. Eng. Chem. Res. 47, 7997–8004.

Casman, G. K., and Liska, J. A. 2007. Food and fuel for all: realistic or foolish? Biofuels,

Bioproducts and Biorefining. 1, 18-23.

116

Castro, C.S., Abreu, A.L., Silva, C.L.T., and Guerreiro, M.C. 2011. Phenol adsorption by

activated carbon produced from spent coffee grounds. Water Science & Technology. 64.

10, 2059-2065.

Centrec Consulting Group, LLC. 2010. Economic impacts of biodiesel production on the

soybean sector, revisited.

http://unitedsoybean.org/images/uploads/multimedia/Economic_Impacts_of_Biodiesel_P

roduction_on_the_Soybean_Sector_Jan_2011.pdf, accessed on March 2011.

Chakrabarti, A. R., Hake, J. M., Zarchi, I., Gray, D. M. D. 2008. Waste Grease Biodiesel

Production at a Wastewater Treatment Plant. WEFTEC Report, 2770-2789.

Chapagain, A.K., and Hoekstra, A.Y. 2004. Water footprints of nations. Volume 1:Main report.

Value of water research report series no. 16.

Ching, S.L., Yusoff, M.S., Aziz, H.A., and Umar, M. 2011. Influence of impregnation ratio on

coffee ground activated carbon as landfill leachate adsorbent for removal of total and

orthophosphate. . 279, 225-234.

Cooke, B.S. 2004. Purification of biodiesel with magnesium silicate adsorbent treatment.

http://www.eurofedlipid.org/meetings/archive/gothenburg/5870/5870_0016.pdf, accessed

on April, 2012.

Couto, R.M., Fernandes, J., Gomes da Silva, M.D.R., and Simoes, P.C. 2009. Supercritical fluid

extraction of lipids from spent coffee grounds. Journal of Supercritical Fluids. 51, 159-

166.

Daglia, M.; Racchi, M.; Papetti, A.; Lanni, C.; Govoni, S.; Gazzani, G. 2004. In-vitro and ex-

vivo antihydroxyl radical activity of green and roasted coffee. J. Agric. Food Chem. 52,

1700–1704.

117

Dallas Group of America Inc. D-SolTM . http://www.dallasgrp.com/web/products.aspx?id=31,

accessed on May, 2010.

De Azevedo, A.B.A, Mazzafera, P., Mohamed, R.S., Vieria de Melo, S.A.B., and Kieckbusch,

T.G. 2008. Extraction of caffeine, chlorogenic acids and lipids from green coffee beans

using supercritical and co-solvents.

http://www.scielo.br/scielo.php?pid=S0104-66322008000300012&script=sci_arttext,

accessed on May 2012.

De Vries, S. C., van de Ven, G. W. J., van Ittersum, M. K., and Giller, K. E. 2010. Resource use

efficiency and environmental performance of nine major biofuel crops, processed by

first-generation conversion techniques. Biomass and Bioenergy, 34, 588-601.

Dunn, J. Management of Crude Glycerin from the Production of Biodiesel for NPDES Programs.

www.great-.net/lists/p2tech/2008-05/doc00000.doc (accessed on June, 2010)

EIA. 2004. Soybean Oil Prices as a Function of Soybean Oil Use for biodiesel Production, 2004-

2013 (2002 Dollars per Gallon).

http://www.eia.doe.gov/oiaf/analysispaper/biodiesel/pdf/tbl1.pdf, assessed on Jan., 2011.

Energy, Efficiency and Renewable Energy (EERE). 2007. Energy and environmental profile of

the U.S. petroleum refining industry.

http://www1.eere.energy.gov/industry/petroleum_refining/pdfs/profile.pdf, accessed on

May 2011.

EPA Renewable fuel standard GHG lifecycle assessment. US EPA,

http://www.crcao.com/workshops/LCA%20October%202009/Session%201/2-

Larson,%20Bob.pdf (accessed on Sept., 2010)

118

EPRI technical report. 2003. A survey of water use and sustainability in the united states with a

focus on power generation.

http://my.epri.com/portal/server.pt?space=CommunityPage&cached=true&parentname=

ObjMgr&parentid=2&control=SetCommunity&CommunityID=404&RaiseDocID=0000

00000001005474&RaiseDocType=Abstract_id, accessed on June, 2012.

EREC-European Renewable Energy Council. 2010. Biofuels: sustainability criteria.

http://ec.europa.eu/energy/renewables/biofuels/sustainability_criteria_en.htm, accessed

on Jan., 2011.

Escudero, C., Gabaldon, C., Marzal, P., and Villaescusa, I. 2008. Effect of EDTA on divalent

metal adsorption onto grape stalk and exhausted coffee wastes. Journal of Hazardous

Materials. 152, 476-485.

ESU-services Ltd. 2011. Life Cycle Assessment of Burning Different Solid Biomass Substrates.

Faccini, C.S., da Cunha, M.E., Moraes, M.S.A., Krause, L.C., Manique, M.C., Rodrigues,

M.R.A., Benvenutti, E.V., and Caramao, E.B. 2011. Dry washing in biodiesel

purification: a comparative study of adsorbents. J. Braz. Chem. Soc. 22, no.3, 558-563.

Feuge, R. O., Kraemer, E. A., Bailey, A. E. 1945. Modification of vegetable oils: IV.

Reesterification of fatty acids with glycerol. Oil and Soap. 22, 202-207.

Fillingham, A. J., and Rushworth, M. S. 2010. Purification method and apparatus. US Patent No.

20100275510A1.

Franca, A.S., Oliveira, L. S., Nunes, A. A, and Alves, C. C. O. 2010. Microwave assisted thermal

treatment of defective coffee beans press cake for the production of adsorbents.

Bioresource Technology. 101, 1068-1074.

119

Franca, A.S., Oliveira, L. S., Ferreira, M.E. 2009. Kinetics and equilibrium studies of methylene

blue adsorption by spent coffee grounds. Desalination. 249, 267-272.

Franca, C. L., Maddigan, K., and White, K. 2006. Sustainability opportunities and challenges of

the biofuels industry.

http://www.bth.se/fou/cuppsats.nsf/all/7bcebda836530f2fc1257295004b7b50/$file/3_Fra

nca_Maddigan_White_2006_Biofuels-1.pdf, accessed on Jan. 2011.

Freedman, B., Pryde, E. H. 1982. Fatty esters from vegetable oils for use as a diesel fuel. in

Vegetable Oils Fuels: Process of the Intl. Conf. on Plant and Vegetable Oils as Fuels, pp.

117-122, St. Joseph, Mich.: ASAE.

Fridley, D. 2010. Nine challenges of alternative energy. In: The post carbon reader: managing

the 21st century’s sustainability crises. http://www.postcarbon.org/Reader/PCReader-

Fridley-Alternatives.pdf, accessed on Jan. 2011.

Gerbens-Leenes, P. W., Hoekstra, A. Y., and van der Meer, Th. 2009. The water footprint of

energy from biomass: a quantitative assessment and consequences of an increasing share

of bio-energy in energy supply. . 68, 1052-1060.

Gerbens-Leenes, P.W., Hoekstra, A.Y., van der Meer, Th. 2009. The water footprint of

bioenergy. PNAS vol. 106 no. 25: 10219-10233

Gerbens-Leenes, P.W., Hoekstra, A.Y., van der Meer, Th. 2008. Water footprint of bio-energy

and other primary energy carriers. Value of water research report series no. 29.

Ghadge, S. V., Raheman, H. 2005. Biodiesel production from mahua (Madhuca indica) oil

having high free fatty acids. Biomass Bioenerg. 28, 601-605.

Haas, M. J. 2010. Alternate feedstocks and technologies for biodiesel production. IN: the

Biodiesel Handbook, 2nd edition, AOCS publishing, Urbana, IL. pp. 47-66.

120

Haas, M. J., Forglia, T. A. 2005. Alternate feedstocks and technologies for biodiesel production.

In: G. Knothe, J. Krahl and J. V. Gerpen (Eds.), The Biodiesel Handbook, AOCS press,

Illinois, pp. 42-61.

Haas, M. J., McAloon, A. J., Yee, W. C., and Foglia, T. A. 2005. A process model to estimate

biodiesel production costs, Bioresource Technology. 97, 671-678.

Haupt, J., Brankatschk, G., and Wilharm, T. 2009. Sterol glucoside content in vegetable oils as a

risk for the production of biodiesel-study of the technological chain impact. Final Report.

http://www.agqm.de/downloads/pdfs/ApplicationToAmericanSoybeanAssociation_Final

Report.pdf, accessed on April, 2012.

He, B.B., Van Gerpen, J. H., and Thompson, J. C. 2009. Sulfur content in selected oils and fats

and their corresponding methyl esters. Applied Engineering in Agriculture. 25, 2, 223-

226.

Hirata, M., Kawasaki, N., Nakamura, T., Matsumoto, K., Kabayama, M., Tamura, T., and

Tanada, Seiki. 2002. Adsorption of dyes onto carbonaceous materials produced from

coffee grounds by microwave treatment. Journal of Colloid and Interface Science. 254,

17-22.

Hoekstra, A.Y., Chappagain, A.K., Aldaya, M.M., and Mekonnen, M.M. 2009. Water footprint

manual-state of the art 2009. Water footprint report. http://www.waterfootprint.org/.

Accessed on March 2011.

Hoekstra, A. Y. and Chapagain, A. K. 2007. Water footprints of nations: water use by people as

a function of their consumption pattern. Water resource Management, 21, 35-48.

121

Hurd, B., Leary, N., Jones, R., and Smith, J. 1999. Relative regional vulnerability of water

resources to climate change, Journal of the American Water Resources Association,

Vol.35, No. 6, pp 1399-1409.

Jacob, W. 2009. Comparison of methods for the purification of biodiesel. Master’s Thesis,

University of Idaho.

Jain, S., Sharma, M. P., and Rajvanshi, S. 2011. Acid base catalyzed transesterification kinetics

of waste cooking oil. Fuel Processing Tech. 92, 32-38.

Jaruwat, P., Kongjao, S., and Hunsom, M. 2010. Management of Biodiesel Wastewater by the

combined process of chemical recovery and electrochemical treatment. Energy

Conversion and Management. 51, pp. 531-537.

Kaikake, K., Hoaki, K., Sunada, H., Dhakal, R.P., and Baba, Y. 2007. Removal characteristics of

metal ions using degreased coffee beans: adsorption equilibrium of cadmium (II).

Bioresource Technology. 98, 2787-2791.

Kante, K., Nieto-Delgado, C., Rangel-Mendez, J. R., and Bandosz, T.J. 2012. Spent coffee-based

activated carbon: specific surface features and their importance for H2S separation

process. Journal of Hazardous Materials. 201-202, pp.141-147.

Karaosmanoglu, F., Baris Cigizoglu, K., Tuter, M., and Ertekin, S. 1996. Investigation of the

refining step of biodiesel production. Energy & Fuels. 10, 890-895.

Kato, S., Yoshimura, H., Hirose, K., Amornkitbamurung, M., Sakka, M., and Sugahara, L. 2005.

Application of microbial consortium system to wastewater from biodiesel fuel generator.

IEA-Waterqual (Singapore), CD-ROM.

Khan, N.A., and Brown, J.B. 1953. The composition of coffee oil and its component fatty acids.

JAOCS. 30, 12, 606-609.

122

Khenniche, L., and Benissad-Aissani, F. 2010. Adsorptive removal of Phenol by coffee residue

activated carbon and commercial activated carbon: equilibrium, kinetics and

thermodynamics. Journal of Chemical & Engineering Data. 55, 4677-4686.

Khenniche, L., and Benissad-Aissani, F. 2009. Characterization and utilization of activated

carbons prepared from coffee residue for adsorptive removal of salicylic acid and phenol:

kinetic and isotherm study. Desalination and Water Treatment. 11, 192-203.

King, C.W., and Webber, M.E. 2008. Water intensity of transportation. Environ Sci & Technol

42, 21: 7866-7872.

Kondamudi, N., Mohapatra, S., Misra, M. J. 2008. Spent coffee grounds as a versatile source of

green energy. Agric. Food Chem. 56, 11757-11760.

Kram, J.W. 2008. “Cleaner and clearer”. Biodiesel Magazine. January Issue.

Lee, I., Pfalzgraf, L.M., Poppe, G.B., Powers, R., and Haines, T. 2007. “The role of sterol

glucosides on filter plugging”. Biodiesel Magazine. April Issue.

Liu, K. 1994. Preparation of fatty acid methyl esters for gas-chromatographic analysis of lipids

in biological materials. Journal of American Oil Chemist’s Society. 71, 1179-1187.

Mazzafera, C. 1999. Chemical composition of defective coffee beans. Food Chem. 64, 547-554.

McMartin, C. and Noyes, G. 2010. America advances to performance-based biofuels.

http://www.stoel.com/webfiles/RFS2_White_Paper_Stoel%20FINAL.pdf, accessed on

Jan. 2011.

Meher, L. C., Vidya Sagar, D., Naik, S. N. 2006. Technical aspects of biodiesel production by

transesterification—a review. Renew. Sustain. Energ. Rev. 10, 248-268.

123

Mekonnen, M.M., and Hoekstra, A.Y. 2010. The green, blue and footprint of crops

and derived crop products. Volume 1: main report. Value of water research report series

no. 47.

Merrick & Company. 1998. Wastewater Treatment Options for the Biomass-to-ethanol Process.

http://www.p2pays.org/ref/22/21201.pdf, accessed on May 2011.

Mulder, K., Hagens, N., and Fisher, B. 2010. Burning water: a comparative analysis of the

energy return on water invested. AMBIO, 39, 30-39.

Mussatto, S.I., Carneiro, L.M., Silva, J.P.A., Roberto, I.C., Teixeira, J.A. 2011. A study on

chemical constituents and sugars extraction from spent coffee grounds. Carbohydrate

Polymers. 83, 368-374.

Namane, A., Mekarzia, A., Benrachedi, K., Belhaneche-Bensemra, N., and Hellal, A. 2005.

Determination of the adsorption capacity of activated carbon made from coffee grounds

by chemical activation with ZnCl2 and H3PO4. Journal of Hazardous Materials B. 119,

189-194.

National Biodiesel Board (NBB). Production statistics.

http://www.biodiesel.org/production/production-statistics, accessed on June 2012.

National Biodiesel Board (NBB)1. Specification for biodiesel (B100-ASTM 6751-11a.

http://www.biodiesel.org/docs/ffs-specs/b100-specification.pdf?sfvrsn=4, accessed on

May 2012.

National Biodiesel Board (NBB)2. Biodiesel industry disappointed in senate energy votes.

http://www.nbb.org/news/nbb-press-releases/press-release-display/2012/03/13/biodiesel-

industry-disappointed-in-senate-energy-votes, accessed on April, 2012.

National Biodiesel Board (NBB). 2008. Proceedings from the biodiesel sustainability symposium,

124

St. Louis, MO.

National Oceanic and Atmospheric Administration. Palmer Drought Index.

http://www.ncdc.noaa.gov/oa/climate/research/prelim/drought/palmer.html, accessed on

June 2012.

Nebraska Department of Environmental Quality. 2009. Waste tips for biodiesel fuel plants.

http://www.deq.state.ne.us/Publica.nsf/a9f87abbcc29fa1f8625687700625436/a85102870

a5b10bf862575d90051c55f/$FILE/08-022.pdf, accessed on April, 2012.

Nunes, A. A., Franca, A. S., and Oliveira, L. S. 2009. Activated carbons from waste biomass: An

alternative use for biodiesel production solid residues. Bioresource Technology. 100,

1786-1792.

Nye, M. J., Southwell, P. H. 1984. Conversion of rapeseed oil to esters for use as diesel fuel.

Fifth Canadian Bioenergy Research and Development Seminar, 487-490, , U. K.

Oliveira, L.S., Franca, A.S., Camargos, R.R.S., and Ferraz, V.P. 2008. Coffee oil as a potential

feedstock for biodiesel production. Bioresource Technol. 99, 3244-3250.

Oliveira, L.S., Franca, A.S., Mendonca, J.C.F., Barros-Junior, M.C. 2006. Proximate

composition and fatty acids from profile of green and roasted defective coffee beans.

LWT. 39, 235-239.

Pate, R. 2008. “Biofuels and the Energy-Water Nexus,” Sandia National Laboratories,

http://www.sandia.gov/energy-water/docs/EW_RoadmapSum_Solar8-06.pdf , accessed

on July, 2010.

Peterson, C.L., and Moller, G. 2010. Biodiesel fules: biodegradability, biological and chemical

oxygen demand, and toxicity. In: the biodiesel handbook, second edition. pp. 231-246.

AOCS Press, Urbana, IL.

125

Population Reference Bureau. 2008 World population data sheet.

http://www.prb.org/publications/datasheets/2008/2008wpds.aspx, accessed on May 2012.

Pradhan, A.; Shrestha, D. S.; Van Gerpen, J.; and Duffield, J. 2008. The Energy Balance of

Soybean Oil Biodiesel Production: A Review of Past Studies. Trans. ASABE. 51 (1),

185–194.

Predojevic, Z.J. 2008. The production of biodiesel from waste frying oils: A comparison of

different purification steps. Fuel. 87, 3522-3528.

Ratanayake, W.M.N., Hollywood, R., O’Grady, E., and Stavric, B. 1993. Lipid content and

composition of coffee brews prepared by different methods. Fd Chem. Toxic. 31, 4, 263-

269.

Reffas, A., Bernardet, V., David, B., Reinert, L., Lehocine, M.B., Dubois, M., Batisse, N., and

Duclaux, L. 2010. Carbons prepared from coffee grounds by H3PO4 activation:

characterization and adsorption of methylene blue and Nylosan Red N-2RBL. Journal of

Hazardous Materials. 175, 779-788.

Ramadhas, A. S., Jayaraj, S., Muraleedharan, C. 2005. Biodiesel production from high FFA

rubber seed oil. Fuel. 84, 335-340.

Satianpattanakul, T., and Ounjitti, M. 2001. Treatment of Biodiesel Processing Wastewater. King

Mongkut' s Institute of Technology North Bangkok

Satyarthi, J. K., Srinivas, D., Ratnasamy, P. 2011. Hydrolysis of vegetable oil and fats to fatty

acids over solid catalysts. Applied Catalysis A: General. 391, 427-435.

Schultz, A. 2010. Biodiesel purification options. Biodiesel Magazine. March Issue.

Schutte, H.A., Cowley, M.A., and Chang, C.Y. 1934. The characteristics and composition of

coffee bean oil. J. Am. Chem. Soc. 56, 10, 2085-2086.

126

Scott, D. 2010. Personal communication, National Biodiesel Board.

Scown, C.D., Horvath, A., and McKone, T.E. 2011. Water footprint of U.S. transportation fuels.

Environ. Sci. & Technol. 45, 2541-2553.

Sendzikiene, E., Makareviciene, V., Janulis, P., Kitrys, S. 2004. Kinetics of free fatty acids

esterification with methanol in the production of biodiesel fuel. Eur. J. . Lipid Sci. .

Tech.106, 831-836.

Service, R. F. 2009. Another biofuels drawback: the demand for irrigation. Science. 326, no.23,

516-517.

Sheehan, J., Camobreco, V., Duffield, J., Graboski, M., and Shapouri, H. 1998. Life cycle

inventory of biodiesel and petroleum diesel for use in an urban bus. National Renewable

Energy Laboratory, NREL/SR-580-24089.

Sims, B. 2011. “Finding the right purification approach”. Biodiesel Magazine. March issue.

Smith, W. 2011. Personal communication. Pacific Biodiesel Inc.

Solley, W. S., Pierce, R. R., and Perlman, H. A. Estimated Use of Water in the United States in

1995; U.S. Geological Survey Circular 1200; United States Geological Survey: 1998.

Suehara, K., Kawamoto, Y., Fujii, E., Kohda, J, Nakano, Y., and Yano, T. 2005. Biological

treatment of wastewater discharged from biodiesel fuel production plant with alkali-

catalyzed transesterification. J. of biosci. and bioengineering. 100, no.4, 437-442.

Tang, H., De Guzman, R.C., Salley, S. O., and Simon Ng, K.Y. 2010. Comparing process

efficiency in reducing steryl glucosides in biodiesel. J Am Oil Chem Soc. 87, 337-345.

Tang, H., De Guzman, R.C., Salley, S. O., and Simon Ng, K.Y. 2008. Formation of insolubles in

palm oil-, yellow grease-, and soybean oil-based biodiesel blends after cold soaking at

4 °C. J Am Oil Chem Soc. 85, 1173-1182.

127

The Royal Society. 2008. Sustainable biofuels: prospects and challenges.

http://royalsociety.org/Sustainable-biofuels-prospects-and-challenges/, accessed on Jan.

2011.

Thiruvengadaravi, K. V., Nandagopal, J., Sathya Selva Bala, V., Dinesh Kirupha, S.,

Vijayalakshmi, P., Sivanesan, S. 2009. Kinetic study of the esterification of free fatty

acids in non-edible Pongamia pinnata oil using acid catalyst. India J. Sci. Technol. 2, 20-

24.

Tilman, D., Socolow, R., Foley, J., Hill, J., Larson, E., Lynd, L., Pacala, S., Reilly, J.,

Searchinger, T., Somerville, C., Williams, R. 2009. Beneficial biofuels-the food, energy,

and environment trilemma. Science. 325, no. 5938, 270-271.

Tokimoto, T., Kawasaki, N., Nakamura, T., Akutagawa, J., and Tanada, S. 2005. Removal of

lead ions in drinking water by coffee grounds as vegetable biomass. Journal of Colloid

and Interface Science. 281, 56-61.

US Biodiesel Production, Exports and Consumption, US Department of Energy:

http://www.afdc.energy.gov/afdc/data/fuels.html, accessed on Oct. 2010.

US Census Bureau, Per capita consumption of selected beverages by type: 1980 to 2008.

http://www.census.gov/compendia/statab/2011/tables/11s0211.pdf, accessed on May,

2012.

US Commercial Bushel Sizes. 2001. http://www.unc.edu/~rowlett/units/scales/bushels.html,

accessed on March, 2011.

USDA. 2007. 2007 Census of Agriculture. AC-07-A-51, U. S. Department of Agriculture.

USDA. 2008. 2008 Farm and Ranch Irrigation Survey. AC-07-SS-1, U. S. Department of

Agriculture.

128

USDA1. 2002. Characteristics and production costs of U.S. soybean farms. Statistical Bulletin

No. 974-4.

USDA2. 2002. 2002 Census of Agriculture. AC-02-A-51, U. S. Department of Agriculture.

USDA. 2003. 2003 Farm and Ranch Irrigation Survey. AC-02-SS-1, U. S. Department of

Agriculture.

USDA & US DOE. 2008. Sustainability of biofuels-future research opportunities.

http://www.ree.usda.gov/ree/news/Sustainability_041609_HR.pdf, accessed on Jan. 2011.

US DOE. Biodiesel production.

http://www.afdc.energy.gov/afdc/fuels/biodiesel_production.html, accessed on Jan. 2010.

US EPA1. 2008. Applicability of Effluent Guidelines and Categorical Pretreatment Standards to

Biodiesel Manufacturing.

http://www.cwacs.com/Downloads/Newsletters/EPA%20Applicability%20Statement%20

for%20Biodiesel%208-11-08.pdf, accessed on June 2010.

US EPA2. 2008. Environmental Laws Applicable to Construction and Operation of Biodiesel

Production Facilities.

http://www.epa.gov/region7/priorities/agriculture/pdf/biodiesel_manual.pdf, accessed on

June 2010.

United States Geological Survey (2005) Water use in the United States.

http://pubs.usgs.gov/circ/1344/pdf/c1344.pdf, accessed on March 2011.

United Soybean Board. 2010. Life cycle impact of soybean production and soy industrial

products. http://www.soybiobased.org/sustainability/life-cycle-profile/, accessed on Jan.

2011.

129

Van Gerpen, J., Shanks, B., Pruszko, R., Clements, D., and Knothe, G. 2004. Biodiesel

Production Technology. National Renewable Energy Laboratory. NREL/SR-510-36244.

Van Gerpen, J. 2005. Biodiesel processing and production. Fuel Processing Technology. 86,

1097-1107.

Wall, J. 2009. Comparison of methods for the purification of biodiesel. (Master’s Thesis,

University of Idaho). http://www.uiweb.uidaho.edu/bioenergy/ionexchangethesis.pdf,

accessed on May 2010.

Wang, H., Tang, H., Salley, S., and Simon Ng, K.Y. 2010. Analysis of sterol glycosides in

biodiesel and biodiesel precipitates. J Am Oil Chem Soc. 87, 215-221.

Western Organization of Resource Councils (WORC). 2006. Biofuels sustainability criteria

adopted by WORC board.

http://www.worc.org/userfiles/Biofuels%20Sustainability%20Criteria.pdf, accessed on

Jan. 2011.

White, J.M., Shah, P., Sanford, S., Valverde, M., and Meier, G. 2010. Evaluation of low

concentration of steryl glucosides in biodiesel by gas chromatography.

http://www.soypower.com/pdfs/Customer%20Newsletters/REG%20Steryl%20Glucoside

s%20Study.pdf, accessed on April, 2012.

Wiltsee, G. 1998. Urban waste grease resource assessment. National Renewable Energy

Laboratory. NREL/SR-570-26141.

Yang, Y. J. 2010. Topographic factors in precipitation dynamics affecting water resource

engineering in the contiguous U.S.: Notes from hydroclimatic studies. World

Environmental and Water Resources Congress 2010: Challenges of Change, Providence,

RI, 2010.

130

Zeman, N. 2010. “Crystal Clear Fuel” Biodiesel Magazine. April Issue. 26-31.

Zhang, Y., Dube, M. A., Mclean D. D., Kates, M. 2003. Biodiesel production from waste

cooking oil: 1. process design and technological assessment. Bioresource Technology. 89,

1-16.

131

Appendix

Water Consumption Survey for Commercial Biodiesel Plants

1. BACKGROUND INFORMATION

 Company Name ______ Location ______ Feedstock ______/______(Primary, if using multi- feedstock)  Capacity ______(Producing)/______(Maximum)

2. WATER CONSUMPTION SCENARIOS (Please check the box(s) if applicable)

2.1. Feedstock Processing (gallon(s) water/gallon oil)  Degumming ______ Others ______

2.2. Biodiesel Purification (gallon(s) water/gallon oil)  Water washing □ Water recycling: ______

□ Once-through: ______

Water source(s): ______

 Ion-exchange resin □  Adsorbent □  BioD distillation □

2.3. Process-related consumption (% or gallon(s) water/gallon oil)  Cooling tower □ Makeup water: ______□ Blowdown: ______

132

Cooling tower type: ______ Boiler □ Makeup water: ______□ Blowdown: ______ Air cooling □  Others □ ______

3. COMMENTS AND SUGGESTION

133