Conversion of Distiller's Grains to Renewable Fuels and High Value

Subscriber access provided by RUTGERS UNIVERSITY

Energy and the Environment

Conversion of distiller’s grains to renewable fuels and high value protein: Integrated techno-economic and life cycle assessment Katherine DeRose, Fang Liu, Ryan W Davis, Blake A. Simmons, and Jason Quinn Environ. Sci. Technol., Just Accepted Manuscript • DOI: 10.1021/acs.est.9b03273 • Publication Date (Web): 05 Aug 2019 Downloaded from pubs.acs.org on August 8, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 28

Environmental Science & Technology

1

Conversion of distiller’s grains to renewable fuels and high value protein: Integrated

2

techno-economic and life cycle assessment

3

Authors: Katherine DeRose*1, Fang Liu2, Ryan W. Davis2, Blake A. Simmons3, Jason C. Quinn1

4

1Mechanical

5

2 Biomass

6

United States

7

3Joint

8

CA, 94720, United States

9

*Corresponding Author: Katherine DeRose

Engineering, Colorado State University Fort Collins, CO, 80523, United States

Science and Conversion Technologies, Sandia National Laboratories, Livermore, CA, 94550,

BioEnergy Institute, Deconstruction Division, Lawrence Berkeley National Laboratory, Emeryville,

10

Email: [email protected]

11

Abstract:

12

Distiller’s grains are a by-product of corn ethanol production and provide an opportunity for increasing

13

the economic viability and sustainability of the overall grain to fuels process. Typically, these grains are

14

dried and sold as a ruminant feed adjunct. This study considers utilization of the residuals in a novel

15

supplementary fermentation process to produce two products, enriched protein and fusel alcohols. The

16

value-add proposition and environmental impact of this second fermentation step for distiller’s grains

17

are evaluated by considering three different processing scenarios. Techno-economic results show the

18

minimum protein selling price, assuming fusel alcohol products are valued at $0.79 per liter gasoline

19

equivalent, ranges between $1.65-$2.48 kg protein-1 for the different cases. Environmental impacts of

20

the system were evaluated through life cycle assessment. Results show a baseline emission results of 17

21

g CO2-eq (MJ fuel)-1 for the fuel’s product and 10.3 kg CO2-eq kg protein-1 for the protein product.

22

Sensitivity to allocation methods show a dramatic impact with results ranging between -8 to 139 g CO2-eq

ACS Paragon Plus Environment


Page 2 of 28

23

(MJ fuel)-1 for the fuels product and -0.3 to 6.4 kg CO2-eq kg protein-1 for the protein product. Discussion

24

focuses on the potential impact of the technology on corn ethanol production economics and

25

sustainability.

26

1. Introduction

27

The United States’ corn ethanol industry is full of controversy 1–3. To its advantage, corn ethanol

28

represents a mature technology with a long production history as a first-generation alternative biofuel

29

and provides a means for domestic bio-based fuel production 4,5. However, this process has been widely

30

criticized for its role in the food versus fuel debate and for continuing to have high production costs that

31

are yet to be competitive with conventional fuels largely due to feed stock costs 1,4,6–9. Despite these

32

issues, the corn ethanol industry has experienced significant growth and the associated infrastructure

33

has an expected lifetime of multiple decades 4. Therefore, it is imperative to continue to improve both

34

the economics and sustainability of this process 5,10. One strategy for achieving these goals is to focus on

35

increased utilization of corn ethanol’s largest co-product, distiller’s grains with solubles (DGS). DGS

36

consists of the remaining biomass after fermentation and includes fibers, proteins, ash, and

37

unfermented sugars 11 which are typically sold as a protein source for ruminants

38

proposed a novel second fermentation process to upgrade DGS that resulting in increased fuel yield and

39

protein value in an effort to improve ethanol biorefinery economics and sustainability.

40

11–15.

Researchers have

Researchers at Sandia National Laboratories have engineered a consortium of E. coli-based

41

biocatalyst strains that have been proven to ferment the hydrolyzed sugars and specific amino acids

42

available in the DGS into energy products predominantly composed of a mixture of C4 and C5 alcohols,

43

hereafter referred to as fusel alcohols 12. Additionally, Lui et al. 12 have verified that the fermentation

44

strains preferentially consume high abundance, low value amino acids, producing a protein product that

45

has been enriched with higher value amino acids including valine, proline, alanine, glycine, methionine,


Page 3 of 28


46

cysteine, histidine, and hydroxyproline. While overall protein mass will decrease due to this second

47

fermentation process, it is proposed that the sale of additional fuels and increased value of the

48

upgraded protein product will offset the cost of additional processing with the inherent benefit of

49

increasing overall fuel yields from the starting grain.

50

The purpose of this study is to evaluate the upgrade conversion of DGS to protein and fuel

51

value-add proposition for corn ethanol producers using engineering process modeling coupled with

52

techno-economic analysis (TEA) and life cycle assessment (LCA). This type of analysis allows researchers

53

and stakeholders to evaluate the economic potential and environmental impact of this technology while

54

it is still in development based on results from experimental work. This work focuses on identifying the

55

optimal processing pathway by evaluating three different process scenarios. Modularity in foundational

56

engineering process modeling enables a direct comparison of the proposed processes on a consistent

57

system boundary. Results from these scenarios were used to evaluate the economic viability and

58

environmental sustainability of the proposed upgrading facility with recommendations for future

59

research and commercial deployment presented.

60

2. Methods

61

A modular engineering process model for the conversion of DGS into fusel alcohols including

62

protein recovery was created and used to capture mass and energy balances for three different

63

production pathways, described below. These balances were used as the foundational inputs for the TEA

64

and LCA studies to evaluate different processing options and evaluate the value-add proposition for

65

ethanol producers.

66

Within the engineering process model, two different processing options for both alcohol

67

recovery (solvent or distillation recovery) and protein recovery (protein concentration and drying to

68

create a high protein content (HPC) product and simple drying to create a low protein content (LPC)



69

product) were considered. From these processing options, three different pathways were identified; a

70

Base Case using solvent extraction and HPC production, a Distillation Case using distillation and HPC

71

production, a LPC Case using solvent extraction and LPC production. A process flow diagram of these

72

processing options is shown in Figure 1. The system boundary included the purchasing of DGS from a co-

73

located corn ethanol facility, processing into ethanol and protein product and distribution of products.

74 75 76

Figure 1: Block flow diagram for the DGS upgrading facility. There are two alcohol recovery options including distillation or solvent extraction; and two protein recovery options including ultrafiltration followed by drying and drying.

77

2.1 Process Model

78

The process model was constructed in a modular fashion with sub-process models constructed

79

such that different technology pathways could be evaluated while maintaining a consistent system

80

boundary. The process model was split into two main processes; fermentation and protein recovery. The

81

performance of the fermentation process models was validated based on laboratory scale experimental

82

data provided by Lui et al.

83

based on the whey protein concentrate production process as described in literature 16,17. A complete

84

list of inputs and outputs for the model can be found in the Supplemental Information.

12,

and was modeled in ASPEN Plus. The protein recovery process model was


Page 4 of 28

Page 5 of 28


85

The model assumes the new upgrade facility will be co-located with a corn ethanol facility to

86

reduce costs and eliminate the need for drying, packaging and transporting of the DGS. The bio-refinery

87

was sized to utilize all DGS produced by a 454 M liter per year corn ethanol plant, the most common

88

corn ethanol facility size currently operated in Iowa, USA 18. Using a conversion of 0.71 kg of DGS per L of

89

ethanol, the DGS feed rate was set to 40,909 kg hr-1 of DGS solids 19.

90

The price of DGS was set at $60 per tonne, based on current commodities markets for wet DGS

91

20.

92

reported to be 29% protein, 8% fat, 9.5% insoluble fiber, 4.5% ash and 49% carbohydrates on a dry mass

93

basis, which is consistent with other results reported in literature 11.

94

2.1.1 Pretreatment

95

Biomass composition was provided by Sandia National Laboratories and validated by NJFL, Inc;

Prior to fermentation, the DGS are treated with a dilute acid to solubilize and hydrolyze the

96

carbohydrates and proteins. The pH is adjusted to 6 using lime (CaO) after pretreatment. Experimental

97

work shows total solubility of the carbohydrates, proteins and lipids to be 90% ± 10%, 70 ± 10% and 100

98

± 10% [g soluble after pretreatment per g total].

99

A subsequent enzymatic proteolysis pretreatment step was employed to hydrolyze any

100

remaining non-hydrolyzed proteins, proteolipids, or N-glycans. DGS are treated with a cell-free extract

101

of Streptomyces gresius (Pronase ®) at a ratio of 1 gram per liter of total slurry. The enzyme was not

102

produced onsite and was assumed to be purchase for $1.05 kg-1. Experimental works shows the

103

resulting protein solubility increased to 90 ± 10% [g soluble after pretreatment per g total protein].

104

Products from pretreatment were assumed to be solubilized carbohydrates, proteins and lipids,

105

water and gypsum (CaSO4∙2H2O). The DGS ash and any unrecovered gypsum were assumed to be inert

106

masses flowing through the system with no adverse effects on process equipment or yield. These

107

assumptions would need to be verified with further experimentation.



108

2.1.2 Fermentation

109

The fermentation process modeled was based on experimental work done by Liu et al 12. This

110

process uses an E. coli consortium to ferment hydrolyzed C5 and C6 carbohydrates and a subset of the

111

amino acids into fusel alcohols. Results from experimentation show a utilization of 31% and 95% of the

112

total solubilized proteins and carbohydrates, respectively. Fusel alcohol yields from fermentation were

113

40 g of fusel alcohols per 100 g of hydrolyzed carbohydrates and amino acids, yielding a total conversion

114

rate of 24 g of fusel alcohols per 100 g of total DGS 12.

115

For modeling purposes, fermentation products were limited to fusel alcohols, carbon dioxide

116

(CO2), hydrogen sulfide (H2S), ammonia (NH3), and water (H2O). To predict fermentation product yields,

117

fusel alcohol yields were set using experimental data with all other yields predicted using an elemental

118

mass balance. Hydrogen sulfide and ammonia yields were based on total liberated sulfur and nitrogen,

119

respectively, while carbon dioxide and water yields were based on remaining carbon and hydrogen

120

balances, respectively. Struvite (NH4MgPO4∙6H2O) was also shown to be a product of this fermentation

121

process 12. The hypothesized mechanism for struvite production is magnesium in the ash reacts with

122

liberated nitrogen, phosphorous and water. The struvite was assumed to be recovered in a settling tank

123

and sold as an additional co-product. The proposed model assumes these yields can be maintained at

124

scale; this assumption is reasonable as fermentation represents a mature technology in terms of

125

commercial deployment.

126

2.1.3 Alcohol Recovery

127

The purpose of the alcohol recovery step is to separate the fusel alcohols from the water and

128

remaining biomass. Modularity in the engineering process model supported the evaluation of two

129

different alcohol recovery processes; solvent and distillation extraction.


Page 6 of 28

Page 7 of 28


130

Solvent extraction was employed by experimenters from Sandia National laboratories, who

131

provided inputs and verification for this method. For this process, the fermentation liquor is fed to the

132

bottom of a ten-stage extraction column. Solvent, ethyl acetate, is added at the top at a ratio of 1:5

133

volume of solvent to volume of fermentation liquor. Product recovery was set at 97% of total alcohols.

134

After solvent extraction, the fusel alcohols and solvent were sent to a distillation column to recover the

135

solvent. Alcohol recovery from this distillation column was set at 99.5% of total alcohols. The system is

136

assumed to have a 5% solvent loss, as validated from experimental results and is typical of solvent

137

extraction systems 21.

138

As distillation is currently used for ethanol recovery in corn ethanol refineries, this method was

139

also investigated as it represents a robust technology for alcohol recovery. Low boiling alcohol recovery

140

was set to 99.5% recovery by mass in the distillate. Some of the fusel alcohols have boiling points higher

141

than water and were assumed to be lost in the column bottoms, as recovery was found to be

142

uneconomical. The distillate from the first column was sent to a second rectification distillation column.

143

Alcohol recovery was set at 99% by mass. Distillate from the rectification column was then sent to a

144

molecular sieve dryer for polishing for a final mass purity of 99.5% by mass [kg alcohols per kg total

145

mass-1].

146

2.1.4 Protein Recovery

147

The remaining biomass after alcohol recovery is sent to a protein processing step to recover the

148

protein product. Two different protein recovery strategies were investigated to produce two different

149

products; a concentrated protein product, HPC and a low protein product, LPC. Both of these processes

150

were based on commercial whey protein recovery processes 16,17.

151 152

For the HPC product, the insoluble fibers and fats were removed using a centrifuge. Proteins were then concentrated using ultrafiltration with diafiltration. Concentration of the solids in the



153

ultrafiltration unit was set to 25%, a common target used in the whey industry 16. Sugars and some of

154

the ash were removed by diafiltration washing 16. Operational costs for the ultrafiltration unit were

155

estimated using information from the San Jose Water Company report on Microfiltration Operating

156

Costs and design information from work performed by Hurst et al. 17,22. After concentrating the

157

proteins, the product was then dried in a spray dryer to reduce water concentration to 4 wt%. Product

158

loss after drying was set at 20% of the total solids 17. The spray drier pricing was based on cost curves

159

from work compiled by Hardy 23. To produce the LPC product, the solids from alcohol recovery were

160

centrifuged and dried in a spray dryer to produce a product with 4 wt% water.

161

2.2 Techno-Economic Analysis

162

The mass and energy outputs from the engineering process models were used as the basis for

163

the economic modeling. Capital costs were estimated based on quotes for similar equipment provided

164

in literature and ASPEN modeling 24. Operational costs were estimated based on the energy and mass

165

requirements of the systems.

166

For this analysis, the system boundary was drawn around the DGS upgrade facility. This required

167

the facility to have its own independent utilities equipment, storage, home office, laboratories and

168

management staff assuming the facility would receive no utilities from the co-located ethanol refinery.

169

All capital equipment was assumed to be purchased new, instead of repurposed or retrofitted. Drawing

170

the system boundary in this way allowed for a conservative and clear examination of the costs and

171

benefits accompanying the integration of the upgrade process with existing facilities.

172

An additional Integrated scenario was also considered in the TEA. This scenario modified the

173

Base Case scenario and assumes some equipment, infrastructure and personnel are shared between the

174

new facility and existing ethanol facility. More detailed information for the inputs for this scenario is

175

available in the Supplemental Information.


Page 8 of 28

Page 9 of 28


176

The TEA uses a discounted cash flow-rate of return (DCFROR) model to estimate the minimum

177

protein selling price (MPSP) required for one kg of upgraded protein product 25. The MPSP was

178

calculated as the price required per kg of protein product to achieve a net present value of zero in the

179

DCFROR model. The DCFROR model used N-th plant assumptions, as outlined by the Department of

180

Energy’s Bioenergy Technologies Office 21,24,25. A list of all N-th plant assumptions can be found in the

181

Supplemental Information.

182

Co-products sold from the upgrade facility include the fusel alcohol energy products and

183

struvite. The selling price for the energy products was set at $0.79 per liter of gasoline equivalent (LGE)-

184

1,

185

the amount of fuel required to provide as much energy as a liter of traditional petroleum-based

186

gasoline; 31.7 MJ. The struvite co-product selling price was set at $455 tonne-1 27,28.

187

2.3 Life Cycle Analysis

the target price for energy products set by the Department of Energy (DOE) 26. An LGE is defined as

188

LCA is used to determine the greenhouse gas (GHG) emissions using a 100-year global warming

189

potential (GWP) associated with the production of different products; the protein products and energy

190

products (fusel alcohols or ethanol and fusel alcohols). The LCA was originally conducted according to

191

ISO 14040 standards by considering an expanded system boundary including the agriculture and ethanol

192

refining and using a displacement method for the co-products. However, this approach to LCA may not

193

always be possible or practical if the upstream processes are not as well known or as well defined as

194

corn-ethanol production, leading LCA practioners to employ an allocation method to determine

195

emissions for the co-product now being used as a feedstock. The system being studied here allows for a

196

case study to investigate how different LCA methodologies can affect LCA results when using a

197

bioenergy waste stream as feedstock for another process and considering a contracted system boundary

198

around the new process. The benefit of the contracted boundary allows for more focus to be placed on



199

the process under investigation, which is especially important for process optimization and hot spot

200

analysis. When approaching LCA this way, there are two main complications for this specific study; the

201

first is determining the emissions associated with the DGS production and the second assigning

202

emissions to the upgraded products as an equal mass of each are produced.

203

Quantifying the emissions associated with the DGS production required assigning some burden

204

of the upstream ethanol production emissions to the DGS. In a standard ethanol LCA, a displacement

205

method is used which does not allow for quantifying the emissions associated with DGS. Reversing this

206

methodology and using a displacement method with DGS as the primary product represents a

207

misguided attempt to determine the emissions that should be allocated to this product stream. To

208

address this issue, this work allocates the ethanol facility and agriculture emissions to the DGS through

209

either an energy or market allocation method; a complete description of the energy and market

210

allocation methods is available in the Supplemental Information. Total emissions rate for ethanol

211

production and agriculture were estimated through values found in literature. The reported range was

212

1734 - 2134 g CO2-eq (L ethanol)-1 with an average value of 1974 g CO2-eq (L ethanol)-1 used in this study

213

15,29–32.

214

The total emissions for the upgrade facility modeled in this study were calculated using the mass

215

and energy balances from the engineering process model. Estimates for the emissions associated with

216

energy production and manufacture of chemicals were taken from GREET 32. To calculate the total GWP

217

for each component three different GHG’s were examined (CO2, CH4 and N2O) and combined into a CO2-

218

equivalance (CO2-eq)based on their 100-year global warming potential (1, 25, 298 [g CO2-eq (g GHG)-1],

219

respectively) and summed to provide a total life cycle inventory (LCI) emissions 33. These LCI’s were then

220

multiplied by the upgrade facility’s total consumption to find the total estimated emissions. A list of the

221

LCI’s used in the analysis can be found in the Supplemental Information.


Page 10 of 28

Page 11 of 28


222

The second complication in this LCA is associated with determining which LCA co-product

223

methodology is appropriate to distribute the upgrade facility emissions between the two main products;

224

displacement, energy allocation and market allocation. In total, six different life cycle scenarios were

225

performed for the contracted system boundary by using displacement, energy and market allocation for

226

two different DGS emissions scenarios, energy and market.

227

To employ the ISO 14044 standard displacement method requires LCA results to be generated

228

assuming the system is first a protein production facility and then again assuming the system is a fuel

229

production system, with displacement credits calculated for the complementary product. For both

230

cases, struvite was assumed to displace diammonium phosphate (DAP) on a per kg of phosphorous

231

basis. The products also received a carbon credit based on the embodied carbon as the system modeled

232

has a well to pump system boundary with embodied carbon originating from the atmosphere. A

233

complete summary of the displacement method used in this analysis is available in the Supplemental

234

Information.

235

In addition to the displacement method, other life cycle co-product methodologies were

236

considered; specifically, market and energy allocation methods. For this analysis, all emissions from the

237

upgrade facility and embodied carbon credits were assigned to a product through either an energy or

238

market allocation, as used to determine DGS associated emissions. The results from the different LCA

239

methodology scenarios were then compared to illustrate the impact of methodology on LCA results. The

240

emissions results were also compared to production emissions of traditional products.

241

2.4 Sensitivity Analysis

242

A sensitivity analysis was performed to identify high impact areas for further research focused

243

on reducing cost and emissions. For this analysis model inputs were varied ± 20% and the resulting

244

MPSP and overall GHG emissions were recorded. The results were then used in a Student’s t-test to



245

determine the t-ratio value for the varied variable, with t-ratios falling outside the two-tail 95%

246

confidence interval indicating high impact areas.

247

3. Results and Discussion

248

Economic results were used to identify the optimal processing pathway and recognize significant

249

cost contributors. LCA results were used to determine if the energy and protein products represented

250

more sustainable products than conventional alternatives and highlight the effects of different life cycle

251

co-product methodologies.

252

3.1 Techno-Economic Analysis

253

The TEA results for the different cases are shown in Figure 2 with results broken out for the

254

individual cost components and summed into three main cost categories; capital expenditure (CAPEX),

255

operational costs (OPEX) and taxes. The MPSP for the Base Case was found to be $1.91 kg protein-1,

256

$1.94 kg protein-1 for the Distillation case, $2.48 kg protein-1 for the LPC case and $1.65 kg protein-1 for

257

the Integrated scenario. Markets show protein selling prices between $0.35 - $2.11 kg-1, dependent on

258

protein quality and market opportunity (animal feed, human consumption, etc.). A list of proteins

259

including quality and selling price is located in the Supplemental Information. Initial analysis indicates

260

this process may be feasible if the protein product could be used as aquaculture.


Page 12 of 28

Page 13 of 28


261 262 263 264 265

Figure 2: Breakdown of the major economic contributors as a fraction of the MPSP for the three processing pathways and Integrated scenario modeled using “N-th of a kind” assumptions. CAPEX and OPEX were further broken down into their major constituents.

The largest cost contributor for the Base Case was the operational costs, accounting for 74% of

266

the total MPSP. The greatest cost was from procurement of the feedstock, accounting for 46% of the

267

total annual OPEX and 35% of the total MPSP. A similar trend was seen in the Distillation and LPC cases,

268

with the feedstock accounting for 46% or 42% of the total annual OPEX and 33% or 28% of the total

269

MPSP, for each case respectively. This mimics trends seen in biorefinery TEA’s in general; where the

270

feedstock represents the largest cost burden. In traditional ethanol processes, feedstock costs have

271

been shown to represent over 70% of the total MFSP for corn based ethanol, soybean based biodiesel,

272

and sugarcane based ethanol 34. One advantage of this upgrade system is the feedstock is a waste

273

stream instead of virgin crop, allowing for reduced procurement costs and increased utilization of the

274

original feedstock.



275

TEA indicates the Base Case to be the most economically favorable option as it reduces capital

276

and operational cost while maximizing alcohol recovery by employing the optimal alcohol and protein

277

recovery strategies with the added benefit of potential opportunities for process improvement. While

278

distillation is a proven technology for ethanol production systems, the presence of high boilers and

279

higher water content in alcohol-water azeotropic mixtures leads to much higher energy requirements

280

making this strategy less desirable for this application. Alcohol recovery is maximized through solvent

281

extraction, as high boilers are not lost leading to higher fuel yields, 78.4 M LGE year-1 for the Base Case

282

which employs solvent recovery compared to 76.5 M LGE year-1 for the Distillation Case. There is also an

283

opportunity for process improvement by increasing solvent recovery efficiency to reduce make up

284

solvent costs. The optimal solvent recovery rate was found to be 98.5%, which reduces MPSP by 12% to

285

$1.69 kg protein-1. In addition, solvent extraction reduces the extent of residual alcohols in the protein

286

product. This added benefit was not directly quantifiable in this work but must be considered

287

holistically. These considerations lead to solvent extraction being the optimal choice for alcohol

288

recovery in terms of economic viability.

289

The Integrated Scenario, which is based on the Base Case and assumes synergistic relations with

290

the co-located ethanol plant, shows a reduction in MPSP as expected. By considering this integration

291

with the existing facility, the MPSP is reduced to $1.65 kg protein-1. Significant savings are seen by

292

reducing protein recovery costs by utilizing existing drying and utilities equipment. These savings also

293

led to reduced indirect capital costs, which are calculated as a percentage of total capital costs. Similar

294

optimization of the solvent system further reduces the MPSP to $1.42 kg protein-1. This scenario shows

295

a more optimistic outlook for the process if these assumptions can be realized.

296

The TEA also shows the HPC pathway is more economical than the LPC pathway for protein

297

recovery, with the MPSP for the Base Case being 77% lower than the LPC pathway. While the HPC

298

contains more pieces of processing equipment, e.g. protein concentration equipment, one of the


Page 14 of 28

Page 15 of 28


299

benefits of this step is bulk water removal. By not removing the water prior to drying in the LPC

300

pathway, the dryers had to be sized to evaporate 4.7 times higher volumes of water, or 28,725 kg hr-1

301

for the Base Case compared to 135,127 kg hr-1 for the LPC pathway. This difference in evaporation

302

requirements causes a significant disparity between the two processing pathways with the CAPEX for

303

the LPC pathway 51% higher and the OPEX 9% more than the Base Case. Increased water to the driers

304

also increased the heat energy requirement for this step, which more than offsets any potential labor or

305

electricity energy savings that may have been realized by making the process less complicated. These

306

results indicate that a bulk water removal stage may be required for the LPC to reduce the CAPEX and

307

OPEX; or indicates the requirement to integrate with the existing ethanol facilities’ drying equipment to

308

negate this cost as shown in the Integrated Scenario.

309

3.2 Life Cycle Analysis

310

LCA was completed to estimate emissions associated with the production of the HPC protein

311

product and fusel alcohols. The Baseline results for the expanded system boundary for the protein and

312

fuels-based pathways are 10 kg CO2-eq kg protein-1 and 17 g CO2-eq MJ fuel-1, respectively. When

313

considering the contracted system boundary around the upgrade facility, six different cases were

314

identified for each product, with protein related emissions ranging between -0.3 and 6.4 kg CO2-eq kg

315

protein-1, while fuel production related emissions ranged between -8 and 139 g CO2-eq MJ fuel-1. The

316

breakdown of the associated emissions for the different products, protein and fusel alcohol, for the six

317

cases using the different co-product LCA methods are shown in Figure 3A and 3B, respectively.

318 319

A)



320 321

B)

322 323 324 325 326

Figure 3 A) Protein-related emissions per kg protein produced and B) fuels-related emissions per MJ fuel, as calculated using different LCA methodologies; displacement, market allocation and energy allocation. Emissions associated with the production of the DGS feedstock were calculated using either energy or market allocation methods. The results presented are based on the Base Case.

327

The largest impact on emissions is associated with DGS production, especially when considering

328

the variance associated with the different allocation methods, market or energy. When using the energy


Page 16 of 28

Page 17 of 28


329

allocation method, the DGS production accounts for more emissions than the entire upgrading facility.

330

This highlights how intensive upstream processing is and reinforces the importance of LCA methodology

331

when considering a complex supply chain.

332

Apart from emissions from DGS production, consumable materials is the largest emissions’

333

contributor, accounting for 47% of the upgrade emissions. Increasing the solvent recovery rate from

334

95% to 96% reduces solvent related emissions by 20% but increased heating emissions by 0.5%. Overall,

335

upgrade emissions are reduced by 3%. At the economic optimum solvent recovery rate of 98.5%

336

upgrading related emissions are reduced by 22% to 157 Gg CO2-eq year-1.

337

Carbon credits have a significant impact on the LCA, and some strategies for improving the LCA

338

include increasing these carbon credits through either; decreasing carbon loss or employing carbon

339

capture and storage. During the fermentation processes, approximately 20% of the carbon in the

340

biomass is lost as CO2 which could be reduced through improved carbon utilization. Another 26% is lost

341

during protein recovery and could be reduced through process improvements. An alternative and

342

complementary strategy is carbon capture and storage for the CO2 rich fermentation head gasses.

343

These strategies could be explored to reduce emissions through increased carbon credits.

344

3.2.1 Effects of LCA Co-product Methodology

345

Two different system boundaries were considered; an expanded system boundary including all

346

upstream processes and a contracted system boundary around the upgrade facility. For the contracted

347

boundary, three different LCA co-product allocation methods were used to determine the emissions

348

related to each product, protein or fuels. Sensitivity to LCA methodology is explored by investigating the

349

effects the of system boundary and methodology strategy on product emissions. Here we discuss the

350

results of each analysis and compare results across scenarios.



351

As seen in Figure 3, the LCA results were sensitive to system boundary, especially the protein

352

product. In the expanded boundary scenario, all emissions are assigned to the protein product, even

353

though the production of protein is the not the impetus for the ethanol production facility. This

354

indicates that the expanded system boundary is not appropriate for considering protein production

355

related emissions; and therefore, care must be taken by LCA practioners when considering scope and

356

system boundary. For the fuels related pathway, this expanded system boundary was more applicable,

357

as the intent of all processing steps, agriculture through fusel alcohol production, revolve around their

358

production. In this analysis, we see some increased emissions over the ethanol-alone process, due to

359

increased processing, with results of 8 g CO2-eq MJ fuel-1 for ethanol-alone compared to 17 g CO2-eq MJ

360

fuel-1 for the combined fuels pathway. There were two reasons identified for the increase in emissions;

361

increased processing and decreased protein credits from displacement. In the combined fuels pathway,

362

production related emissions increased by 40%, while fuels production only increased by 29%. In regards

363

to protein credits, DGS credits account for -14 g CO2-eq MJ fuel-1 for ethanol only 29, while the upgraded

364

protein supplied a -5 g CO2-eq MJ fuel-1 credit. These credits may change if more intensive proteins are

365

displaced.

366

The results were also highly susceptible to co-product allocation methods within the upgrade

367

facility system boundary. The highest emissions for each product across the scenarios was found using

368

the displacement method, regardless of DGS allocation strategy. The emissions for producing 1 kg of

369

protein was found to be 1.4 kg CO2-eq kg protein-1 and 6.4 kg CO2-eq kg protein-1 for the DGS market and

370

energy allocation methods, respectively, while the emissions for producing 1 MJ of fuel production were

371

estimated to be 14 g CO2-eq MJ fuel-1 and 140 g CO2-eq MJ fuel-1, using the DGS market and energy

372

allocation methods, respectively. For these scenarios, emissions are not divided amongst the products

373

as they are for the other allocation methods but instead rely on displacement credits to accurately

374

capture the emissions burden. The protein credit was calculated assuming the displacement of corn


Page 18 of 28

Page 19 of 28


375

gluten meal, a protein with relatively low emissions, but if there is an opportunity to displace a more

376

emissions intensive protein the value of this credit would increase. For instance, if we consider

377

displacing beef, the displacement credit compensates for all the upgrading related emissions 35. This

378

variable represents a large uncertainty due to its dependence on the type of protein being displaced and

379

highlights the importance of identifying the correct protein displacement markets and displacement

380

ratios when applying this method.

381

Allocation methods can provide a method to allocate emissions to a variety of products without

382

the requirement of choosing the appropriate displacement product. However, the use of allocation

383

methods introduces other types of uncertainty into the analysis and requires thoughtful consideration

384

on the part of the analyzer. Variability in market price will proportionally affect the division of emissions

385

within market allocation. Reducing DGS cost by 20% reduces DGS related emissions by 18% to 54 Gg

386

CO2-eq year-1. This further affects product related emissions; for example, fusel alcohol emissions

387

calculated using the displacement method reduce from 14 to 9 g CO2-eq MJ fuel-1 through this market

388

price fluctuation. Allocation methods also require practioners to choose the most applicable allocation

389

method. For instance, energy allocation may not be the most appropriate method for allocating

390

emissions to DGS. While the DGS product does contain energy, it is not a traditional energy product,

391

thus market allocation provides a more appropriate approach. Regardless of methodical approach, some

392

uncertainty may be introduced to the analysis.

393

LCA methodology directly effects a product’s ability to meet sustainability metrics. For

394

renewable fuels, one important metric is the Renewable Fuel Standard (RFS) which calls for a 50%

395

reduction of emissions when compared to petroleum-based energy products, approximately -26 g CO2-eq

396

MJ fuel-1 on a well-to-pump boundary 32,36. The ability of the fusel alcohols to meet this standard is

397

dependent on the LCA co-product methodology and is currently not achieved in any of the scenarios.



398

The protein emissions were compared to the emissions required to produce one kg of an

399

equivalent protein; however, this comparison is somewhat subjective and represents an area of high

400

uncertainty. Protein related emissions can range from -1.3 kg CO2-eq (kg protein)-1 for grain-based

401

proteins to over 13 kg CO2-eq (kg protein)-1 for human-grade protein, such as beef, when accounting for

402

embodied carbon in the protein 32,35,37. Without knowing the correct target protein market, we cannot

403

confidently conclude that this protein product represents an emissions savings.

404

From this analysis, we can see how system boundary and LCA methodology can have a

405

significant impact on product related emission and reaffirms the importance of adopting uniform

406

practices within the community, especially when investigating if a product meets certain regulatory

407

goals, such as the RFS. We have also identified the importance of understanding product markets and

408

displacement ratios, as this variable provides great uncertainty for the current state of the model and

409

analysis.

410

While emissions improvements remain uncertain, there are several other environmental

411

benefits to this upgrade process not captured in the LCA. By producing more fuels from the original

412

grain, we can reduce biofuels-based land requirements reducing food versus fuel and land-use change

413

concerns. The upgraded protein may also help meet increased global protein demands, by providing a

414

product enriched in vital amino acids. Increased utilization of the original feedstock would provide

415

environmental benefits by reducing these burdens.

416

3.3 Sensitivity Analysis

417

A sensitivity analysis was also performed to identify the high impact variables in terms of

418

process economics. Results for the top 11 model inputs is presented Figure 4A. Variables whose t-ratios

419

fall outside the t-critical range, ± 2.074, are shown to be statistically significant based on a 95%

420

confidence interval. A sensitivity analysis was also performed on the emissions, with the results shown


Page 20 of 28

Page 21 of 28


421

in Figure 4B. The t-critical ratio for this analysis was found to be ± 2.179.

422

A)

423 424

B)

425 426 427 428

Figure 4: A) Results from the sensitivity analysis for the top 11 inputs on the economics model and B) results from the sensitivity analysis for the top 9 variables on the LCA model. Variables whose t-ratios fall outside the t-critical range, indicated by the dashed vertical lines, are considered statistically significant based on a 95% confidence interval.



429

Protein recovery, solvent recovery and fermentation yield were shown to be high impact areas

430

for improving process economics and reducing emissions. An in-depth discussion of the sensitivity

431

analysis is available in the Supplemental Information.

432

Supporting Information: Table of Techno-Economic assumptions, Table of N-th plant assumptions, Table

433

of Life Cycle Inventories, Table of process model inputs, explanation of “Integrated” scenario

434

assumptions, Life cycle displacement methods, Life cycle allocation methods, Estimates of Base Case

435

capital expenses, Protein market analysis methods and results, Life cycle analysis results, Sensitivity

436

analysis discussion

437

Acknowledgements

438 439 440 441 442 443 444 445

Support for this work was provided by DOE-EERE BioEnergy Technologies Office (BETO) under agreement number 26336. Sandia National Laboratories is a multi-mission laboratory managed and operated by National Technology and Engineering Solutions of Sandia, LLC., a wholly owned subsidiary of Honeywell International, Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525. This work was part of the DOE Joint BioEnergy Institute (http://www.jbei.org) supported by the U.S. Department of Energy, Office of Science, Office of Biological and Environmental Research, through contract DE-AC02- 05CH11231 between Lawrence Berkeley National Laboratory and the U. S. Department of Energy.


Page 22 of 28

Page 23 of 28


446

References

447

(1)

448 449

Controversy. Economist 2008, 156 (4), 507–527. https://doi.org/10.1007/s10645-008-9098-x. (2)

450 451

Keyzer, M. A.; Merbis, M. D.; Voortman, R. L. Notes and Communications: The Biofuel

Luchansky, M. S.; Monks, J. Supply and Demand Elasticities in the U.S. Ethanol Fuel Market. Energy Econ. 2009, 31 (3), 403–410. https://doi.org/10.1016/j.eneco.2008.12.005.

(3)

Yang, Y.; Bae, J.; Kim, J.; Suh, S. Replacing Gasoline with Corn Ethanol Results in Significant

452

Environmental Problem-Shifting. Environ. Sci. Technol. 2012, 46 (7), 3671–3678.

453

https://doi.org/10.1021/es203641p.

454

(4)

Peplow, M. Cellulosic Ethanol Fights for Life. Nature 2014, 507 (March), 4–5.

455

(5)

Farrell, A. E.; Plevin, R. J.; Turner, B. T.; Jones, A. t.; O’Hare, M.; Kammen, D. M. Ethanol Can

456 457

Contribute to Energy and Environmental Goals. Science (80-. ). 2006, 311 (January), 506–509. (6)

Pimentel, D.; Patzek, T. W. Ethanol Production Using Corn, Switchgrass, and Wood; Biodiesel

458

Production Using Soybean and Sunflower. Food, Energy, Soc. Third Ed. 2007, 14 (1), 311–332.

459

https://doi.org/10.1201/9781420046687.

460

(7)

Administration, U. S. E. I. Factors Affecting Gasoline Prices.

461

(8)

Irwin, S. The Profitability of Ethanol Production in 2016. farmdoc Dly. 2017, 7 (18).

462

(9)

Luchansky, M. S.; Monks, J. Supply and Demand Elasticities in the U.S. Ethanol Fuel Market.

463 464 465

Energy Econ. 2009, 31 (3), 403–410. https://doi.org/10.1016/j.eneco.2008.12.005. (10)

Tiffany, D. G. Economic and Environmental Impacts of Ethanol Production From. Fed. Reserv. Bank St. Louis Reg. Econ. Dev. 2009, 5 (1), 42–58.



466

(11)

467 468

Liu, K. Chemical Composition of Distillers Grains , a Review. 2011, 1508–1526. https://doi.org/10.1021/jf103512z.

(12)

Liu, F.; Wu, W.; Tran-Gyamfi, M. B.; Jaryenneh, J. D.; Zhuang, X.; Davis, R. W. Bioconversion of

469

Distillers’ Grains Hydrolysates to Advanced Biofuels by an Escherichia Coli Co-Culture. Microb.

470

Cell Fact. 2017, 16 (1), 1–14. https://doi.org/10.1186/s12934-017-0804-8.

471

(13)

Arora, S.; Arora, S.; Wu, M.; Wu, M.; Wang, M.; Wang, M. Update of Distillers Grains

472

Displacement Ratios for Corn Ethanol Life-Cycle Analysis. Contract 2008, No. September.

473

https://doi.org/10.2172/1004867.

474

(14)

475 476

Pigs. Anim. Nutr. 2016, 2 (4), 262–266. https://doi.org/10.1016/j.aninu.2016.07.008. (15)

477 478

(16)

(17)

(18)

Association, I. R. F. Iowa Ethanol Biorefineries https://iowarfa.org/ethanol-center/ethanolbiorefineries/.

(19)

485 486

Hurst, S.; Aplin, R.; Barbano, D. M. Whey Powder and Whey Protein Concentrate Production Technology, Costs and Profitability. Cornell Progr. Dairy Mark. Policy 1990, No. April, 1–51.

483 484

Pak, T. Dairy Processing Handbook - Whey Processing https://dairyprocessinghandbook.com/chapter/whey-processing.

481 482

Edwards, K.; Anex, C. R. P. Co-Product Allocation in Life Cycle Assessment: A Case Study Extended Abstract 2009-645-AWMA. Analysis 2009, No. 651.

479 480

Adeola, O.; Ragland, D. Comparative Ileal Amino Acid Digestibility of Distillers’ Grains for Growing

Rendleman, C. M.; Shapouri, H. New Technologies in Ethanol Production. Semin. Ser. 2007, Report Num (14), 30. https://doi.org/Agricultural Economic Report Number 842.

(20)

Agriculture, U. S. D. of. National Weekly Feedstuff Wholesale Prices


Page 24 of 28

Page 25 of 28


487 488

https://www.ams.usda.gov/mnreports/ms_gr852.txt. (21)

Davis, R.; Kinchin, C.; Markham, J.; Tan, E. C. D.; Laurens, L. M. L. Process Design and Economics

489

for the Conversion of Algal Biomass to Biofuels : Algal Biomass Fractionation to Lipid- Products

490

Process Design and Economics for the Conversion of Algal Biomass to Biofuels : Algal Biomass

491

Fractionation to Lipid- and Carbohyd. 2014, No. September, NREL/TP-5100-62368.

492

https://doi.org/10.2172/1159351.

493

(22)

Rogers, J. N.; Rosenberg, J. N.; Guzman, B. J.; Oh, V. H.; Mimbela, L. E.; Ghassemi, A.; Betenbaugh,

494

M. J.; Oyler, G. A.; Donohue, M. D. A Critical Analysis of Paddlewheel-Driven Raceway Ponds for

495

Algal Biofuel Production at Commercial Scales. Algal Res. 2014, 4 (1), 76–88.

496

https://doi.org/10.1016/j.algal.2013.11.007.

497

(23)

498 499

Hardy, W. L. COSTS, Process and Construction Spray Drying Costs. Ind. Eng. Chem. 1955, 47 (10), 73A-74A. https://doi.org/10.1021/ie50550a008.

(24)

Humbird, D.; Davis, R.; Tao, L.; Kinchin, C.; Hsu, D.; Aden, A.; Schoen, P.; Lukas, J.; Olthof, B.;

500

Worley, M.; Sexton, D.; Dudgeon, D. Process Design and Economics for Biochemical Conversion

501

of Lignocellulosic Biomass to Ethanol. Renew. Energy 2011, 303 (May), 147.

502

https://doi.org/10.2172/1013269.

503

(25)

504 505

Short, W.; Packey, D. J.; Holt, T. A Manual for the Economic Evaluation of Energy Efficiency and Renewable Energy Technologies. 1995, No. March. https://doi.org/10.2172/35391.

(26)

Energy, O. of E. E. and R. Energy Department Announces $11.3 Million Available for Mega-Bio:

506

Bioproducts to Enable Biofuels https://www.energy.gov/eere/bioenergy/articles/energy-

507

department-announces-113-million-available-mega-bio-bioproducts.

508

(27)

Li, B.; Boiarkina, I.; Yu, W.; Huang, H. M.; Munir, T.; Wang, G. Q.; Young, B. R. Phosphorous



509

Recovery through Struvite Crystallization: Challenges for Future Design. Sci. Total Environ. 2019,

510

648, 1244–1256.

511

(28)

Molinos-Senante, M.; Hernández-Sancho, F.; Sala-Garrido, R.; Garrido-Baserba, M. Economic

512

Feasibility Study for Phosphorus Recovery Processes. Ambio 2011, 40 (4), 408–416.

513

https://doi.org/10.1007/s13280-010-0101-9.

514

(29)

515 516

Wang, M.; Wu, M.; Huo, H. Life-Cycle Energy and Greenhouse Gas Emission Impacts of Different Corn Ethanol Plant Types. 2007. https://doi.org/10.1088/1748-9326/2/2/024001.

(30)

Wang, M.; Han, J.; Dunn, J. B.; Cai, H. Well-to-Wheels Energy Use and Greenhouse Gas Emissions

517

of Ethanol from Corn , Sugarcane and Cellulosic Biomass for US Use.

518

https://doi.org/10.1088/1748-9326/7/4/045905.

519

(31)

Marland, G.; Turhollow, A. F. CO2emissions from the Production and Combustion of Fuel Ethanol

520

from Corn. Energy 1991, 16 (11–12), 1307–1316. https://doi.org/10.1016/0360-5442(91)90004-

521

6.

522

(32)

523 524

Elgowainy, A., Dieffenthaler, D., Sokolov, V., Sabbisetti, R., Cooney, C., Anjum, A. GREET. Argonne National Laboratory 2017.

(33)

Forster P, Ramaswamy V, Artaxo P, Berntsen T, Betts R, Fahey DW, Haywood J, Lean J, Lowe DC,

525

Myhre G, Nganga J., Prinn R., Raga, G, Schultz M., Van Dorland, R. Changes in Atmospheric

526

Constituents and in Radiative Forcing. Chapter 2. In: Solomon, S., D. Qin, M. Manning, Z. Chen, M.

527

Marquis, K.B. Averyt, M.Tignor and H.L. Miller (Eds.), Climate Change 2007: The Physical Science

528

Basis. Contribution of Working Group. 2007, 130–234. https://doi.org/10.20892/j.issn.2095-

529

3941.2017.0150.

530

(34)

Ling, T.; Aden, A. The Economics of Current and Future Biofuels. In Biofuels; Tomes, D.,


Page 26 of 28

Page 27 of 28


531 532

Lakshmanan, P., Songstad, D., Eds.; Springer: New York, NY, 2011. (35)

Beauchemin, K. A.; Henry Janzen, H.; Little, S. M.; McAllister, T. A.; McGinn, S. M. Life Cycle

533

Assessment of Greenhouse Gas Emissions from Beef Production in Western Canada: A Case

534

Study. Agric. Syst. 2010, 103 (6), 371–379. https://doi.org/10.1016/j.agsy.2010.03.008.

535

(36)

Quinn, J. C.; Smith, T. G.; Downes, C. M.; Quinn, C. Microalgae to Biofuels Lifecycle Assessment -

536

Multiple Pathway Evaluation. Algal Res. 2014, 4 (1), 116–122.

537

https://doi.org/10.1016/j.algal.2013.11.002.

538

(37)

Dalgaard, R.; Schmidt, J.; Halberg, N.; Christensen, P.; Thrane, M.; Pengue, W. A. LCA for Food

539

Products LCA for Food Products ( Subject Editor : Niels Jungbluth ) Case Study LCA of Soybean

540

Meal. 2008, 10 (7), 240–254.

541

542


Economic Analysis

Environmental Science & Technology Page 28 of 28 Process Model Distiller’s Grains

Fuels Fermentation

Enriched Protein

ACS Paragon Plus Environment Environmental Analysis

Conversion of Distiller's Grains to Renewable Fuels and High Value

Recommend Documents