You are on page 1of 14

Science of the Total Environment 816 (2022) 151541

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

An integrated life cycle multi-objective optimization model for


health-environment-economic nexus in food waste management sector
Zuchao Lin, Jun Keat Ooi, Kok Sin Woon ⁎
School of Energy and Chemical Engineering, Xiamen University Malaysia, Jalan Sunsuria, Bandar Sunsuria, 43900 Sepang, Selangor, Malaysia

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• A life cycle optimization model is devel-


oped for the food waste (FW) sector.
• Human health, ecosystems, and eco-
nomic impacts are optimized in the
model.
• Anaerobic digestion of FW has the low-
est impacts on human health and eco-
systems.
• If cost is included, high-temperature
drying sterilization exceeds other op-
tions.
• Results are sensitive to the prices of
electricity, cooking gas, and livestock
feed.

a r t i c l e i n f o a b s t r a c t

Article history: Food waste is a universal problem in many countries. In line with Sustainable Development Goals 7 and 12, it is
Received 6 July 2021 crucial to identify a cost-effective food waste valorization management framework with the least human health
Received in revised form 4 October 2021 and environmental impacts. However, studies on the synergistic effect of life cycle assessment and mathematical
Accepted 4 November 2021
optimization interconnected with human health, environment, and economic are relatively few and far between;
Available online 10 November 2021
hence they cannot provide holistic recommendations to policymakers in developing environmental and eco-
Editor: Deyi Hou nomic feasibility of food waste management frameworks. Taking Malaysia as a case study, this study proposes
a simple and deterministic model that integrates life cycle assessment and multi-objective mathematical optimi-
zation to unpack the health-environment-economic wellbeing nexus in food waste management sector. The
Keywords: model evaluates the life cycle human health, environmental, and economic impacts of five food waste disposal
Life cycle assessment and valorization technologies: open landfill, sanitary landfill, aerated windrow composting, high-temperature
General algebraic modeling system drying sterilization, and anaerobic digestion, and identifies the optimal food waste valorization configuration so-
Pareto-optimal solution lution in Malaysia. Based on the results modeled by SimaPro 9.0 and General Algebraic Modeling System with
Hotspot analysis
augmented ε-constraint, valorization of food waste into electricity via anaerobic digestion is the most favorable
Valorization technology
option, with 146% and 161% reduction of human health and ecosystems, respectively, as compared with open
Malaysia
landfill. If cost is combined as an objective function with human health and ecosystems, high-temperature drying
sterilization is the most attractive scenario due to the high livestock feed revenue. Among the 10 Pareto-optimal
solutions, 9% sanitary landfill, 3% aerated windrow composting, 30% high-temperature drying sterilization, 30%
anaerobic digestion to electricity, and 28% anaerobic digestion to cooking gas, is recommended as future food
waste management configuration. The sensitivity results demonstrate that prices of electricity, cooking gas,
and livestock feed affect the optimal configuration food waste management system.
© 2021 Elsevier B.V. All rights reserved.

⁎ Corresponding author.
E-mail address: koksin.woon@xmu.edu.my (K.S. Woon).

https://doi.org/10.1016/j.scitotenv.2021.151541
0048-9697/© 2021 Elsevier B.V. All rights reserved.
Z. Lin, J.K. Ooi and K.S. Woon Science of the Total Environment 816 (2022) 151541

deciding the feasibility of different FW valorization technologies to valu-


Abbreviations
able byproducts:
AD anaerobic digestion
Question 1: How can we scientifically and holistically analyze the rel-
AUGMENCON2 augmented ε-constraint version 2
ative impacts of various environmental factors (e.g., climate change,
DALY disability-adjusted life years
acidification, eutrophication) for different FW valorization technologies?
FW food waste
Question 2: How can we determine the optimal combination solu-
GAMS general algebraic modeling system
tion of different FW valorization technologies based on the SDG 2030
GHG greenhouse gases
Agenda, which emphasizes the synergistic effect of people (human
LCA life cycle assessment
health), planet (ecosystems), and prosperity (cost)?
LFG landfill gas
Question 3: How can we adopt an integrated assessment to achieve si-
MILP mixed-integer linear programming
multaneous life cycle environmental evaluation (Question 1) and optimal
MYR Malaysia ringgit
configuration (Question 2) of different FW valorization technologies?
SDG sustainable development goal
To resolve Question 1, life cycle assessment (LCA) can be used as an
Mathematical notations analytical tool for establishing a scientific analysis of FW management
(Chiu et al., 2016); hence the potential life cycle environmental impacts
σs quantity of food waste sent into each scenario
associated with the process can be assessed and the hotspot areas can
σs,p quantity of pth type product from each scenario
be identified (Albizzati et al., 2021). For Question 2, mathematical
in inflow of food waste into each scenario
optimization modeling can be applied to minimize or maximize the ob-
out outflow of food waste from each scenario
jective functions under certain constraints to obtain an optimal solution.
βs binary parameter indicate if scenario is selected
LCA tool can only evaluate the life cycle impacts of different FW valori-
βs,p binary parameter indicate if pth type of product can
zation technologies without providing an optimal solution. In contrast,
produce from each scenario
mathematical optimization does not analyze the life cycle impacts of
CCs,p conversion rate of food waste to pth product in each
the FW valorization technologies. To better judge the decision-making
scenario
process as captioned in Question 3, an integrated assessment model
CAPs capacity of each scenario
should be developed to bridge the limitations of both tools.
Es electricity generated in each scenario
Table A.1 in the supplementary material summarizes LCA and opti-
ED renewable energy demand
mization studies that evaluated the environmental feasibility of differ-
PD product demand
ent FW valorization technologies. Keng et al. (2020) performed an LCA
CAPEX total capital expenditure
study on a composting system at the University of Nottingham
OPEX total operational expenditure
Malaysia. Choong et al. (2019) evaluated the GHG emissions of recycled
REV total revenue from product
wood waste for biomass-fired fuel in Malaysia using LCA. Yet, no study
ACAPEXs annualized capital cost of each scenario
integrates the LCA and optimization model to identify an optimal FW
ns number of plant from each scenario
treatment configuration solution based on the local condition in
UOPEXs unit operational cost of each scenario
Malaysia. Guven et al. (2019) presented the mid-point results of four
PRICEp price of pth type product
major processes: transportation, wastewater treatment, FW treatment,
Hhealths human health impact of each scenario
and sludge treatment. Yeo et al. (2019) evaluated the environmental
Ecos ecosystems impact of each scenario
hotspots of transportation and production of wooden biochips to co-
compost with FW at ReCiPe mid-point level. These studies did not
categorize the input and output data according to the individual sub-
process, hence unable to identify the key emissions from each sub-
1. Introduction process and hard to provide specific recommendations for technology
improvement. Meanwhile, the mid-point categories make the results
Food waste (FW) constitutes the highest composition, accounting interpretation complex, rendering a hard decision to be made as the re-
for 44% of global solid waste (Kaza et al., 2018). FW consists of perish- sults are multi-faceted. Aggregating mid-point categories can obtain
able, odorous, and high moisture content characteristics. Thus, inappro- end-point categories to provide a more convenient and consolidated in-
priate FW disposal and treatment methods may result in severe terpretation of results (Pérez-Camacho et al., 2018).
environmental impacts, economic costs, and human health threats (Jin For mathematical optimization, cost minimization is the most fre-
et al., 2021). The Food and Agriculture Organization underlined that quently studied objective (Chinchodkar and Jadhav, 2018; Garibay-
about 30% of food production (equivalent to 1.3 billion tonnes) was Rodriguez et al., 2018). Some recent studies examined the economic
wasted, leading to 3.3 billion tonnes of greenhouse gas (GHG) emissions and environmental aspects simultaneously using a multi-objective opti-
and 750 billion USD direct economic cost (FAO, 2013). Chen et al. mization model (Hoang et al., 2019; Rizwan et al., 2020). The weighting
(2017) elucidated that untreated FW is considered a misplaced resource approach was used to integrate two objective functions due to its sim-
as valuable byproducts such as livestock feed, compost, electricity, plicity. Yet, the weights of each objective function are highly influenced
cooking gas, and biogas vehicle fuel could have been valorized from by the authors' preferences, making the results less deterministic. Aug-
FW. Turning FW into valuable byproducts not only contributes to the mented ε-constraint version 2 (AUGMENCON2) method was proposed
economic benefits (e.g., biogas vehicle fuel could save 0.45 USD/Nm3 by Mavrotas and Florios (2013) to eliminate the weighting method in
as petrol substitute) but also minimizes adverse impacts on human solving the multi-objective model.
health and the environment through air and water pollution (Lo and Some studies combined both LCA and mathematical optimization to
Woon, 2016). The FW valorization is in line with Sustainable Develop- compensate for the limitations of each approach (Chen et al., 2017;
ment Goal (SDG) 7 Affordable and Clean Energy (Hettiarachchi et al., Falahi and Avami, 2020; Laso et al., 2018; Malmir et al., 2020;
2018) and SDG 12 Responsible Consumption and Production (Beretta Movahed et al., 2020; Rizwan et al., 2020; Roberts et al., 2018). Laso
and Hellweg, 2019). et al. (2018) combined the environmental results (i.e., water, energy,
Considering the complexity of FW physical and chemical character- and climate change) and the nutritional content using the weighting
istics, local government policies, national circumstances, and invest- method and optimized the FW management process, yet the economic
ment budget constraints, there is no “one-size-fits-all” formula when factor was not considered. Most studies in Table A.1 neglected the im-
treating the FW (Ooi et al., 2021). Several issues are raised when pact of human health on FW management. Falahi and Avami (2020)

2
Z. Lin, J.K. Ooi and K.S. Woon Science of the Total Environment 816 (2022) 151541

compared the optimized configuration from life cycle environmental the inputs and output from a life cycle perspective and the different cost
impacts and emergy cost perspective separately. To date, action has components. Life cycle impact assessment is performed in Stage 3 to
not gone far enough to integrate the LCA and optimization model con- characterize the life cycle impacts of human health and ecosystems.
sidering human health, environmental, and economic factors (Roberts Key emissions identification of sub-processes in each FW valorization
et al., 2018). Studies that explicitly explore the interconnectedness of technology is determined and appropriate recommendations are pro-
LCA and optimization models are still in their relative infancy on FW vided to reduce the environmental burdens. In Stage 4, a mixed-
management. Hitherto, no clear and holistic step-by-step methodology integer linear programming (MILP) mathematical model is formulated
framework of integrated life cycle multi-objective optimization has via the AUGMENCON2 method by integrating the LCA end-point results
been developed to enhance the study's consistency and accuracy. The and the total cost. A three-dimensional set of Pareto-optimal solutions is
methodology framework should be simple and deterministic to en- generated via GAMS software to help policymakers identify FW valori-
hance the synergistic effect on LCA and optimization model. This can zation technologies that provide environmental benefits/burdens and
be done by employing the AUGMENCON2 method to optimize the LCA cost impact. Post integrated LCA-optimization analysis on the sensitive
end-point results (i.e., human health and ecosystems) and the economic parameters is conducted in Stage 5.
perspective (i.e., total cost of the FW management system).
Considering the above-mentioned limitations, this study's main ob- 2.2. Stage 1: goal and scope definition
jective is to develop a model that integrates LCA and mathematical opti-
mization for FW management. The life cycle optimization model This study aims to determine an optimal structure of FW allocation
identifies FW management's optimal configuration solution considering by considering the life cycle impacts of human health, environmental,
human health, environmental, and economic factors. Malaysia is se- and economic perspectives to provide a holistic view for stakeholders
lected as a case study to apply this integrated assessment model. In in developing sustainable FW management. Despite Malaysia's abun-
Malaysia, 16,688 tonnes/day of FW (or 0.53 kg/person/day) was pro- dance of fossil fuels, the government has committed to becoming a
duced, in which 44.5% from household sector in 2018 (Bernama, carbon-neutral nation by 2050 (Abdul Hamid, 2017). Under the Na-
2020). The current FW management in Malaysia is a pressing burden tional Renewable Energy Policy, the Malaysian government strives to
with high dependence on landfills. Out of 170 landfills, only 14 are san- decrease carbon emissions intensity of economic activity by 45% by
itary landfills with landfill gas (LFG) recovery systems (Bong et al., 2017; 2030 relative to the 2005 level and raise the renewable energy mix to
Shamsul, 2016). Given the recent enabling policies for carbon trading 20% by 2025 (MESTECC, 2018). Therefore, FW valorization can be an at-
(Bernama, 2021), valorizing FW to different FW-based byproducts can tractive option for Malaysia in waste reduction and renewable energy
reduce carbon emissions, environmental pollution, and human health production.
impacts and provide economic benefits due to revenue generation To apply the integrated life cycle multi-objective optimization
from valuable byproducts (Woon et al., 2016). The Malaysian govern- model in this case study, some major assumptions are made as listed
ment explores other FW valorization technologies such as sanitary land- below.
fills with LFG recovery systems, composting, high-temperature drying
sterilization, and anaerobic digestion (AD) for turning FW into different 1. The FW amount is formulated based on 2016 and 2018 data
byproducts. At Ulu Tiram, Johor, a FOLO (feed our loved ones) farm (i.e., 15,000 tonnes/day, and 16,688 tonnes/day, respectively). This
composted 3.0 tonnes/day of FW to produce 1.1 tonnes/day of organic FW amount is forecasted from 2016 to 2046 based on a compound
fertilizer for application on-site (Lim et al., 2019). Mentari Alam EKO annual growth rate of 5.63%;
(MAEKO) has utilized composters to convert more than 7500 tonnes 2. The operating life year of the FW valorization technology is 30 years;
of FW to organic fertilizer since 2015 (MAEKO, 2021). Worming Up, a 3. The transportation of FW collection in different scenarios is as-
social enterprise in Kuching, upcycled 450 tonnes/year of FW as live- sumed to be similar and negligible due to the small geographical
stock feed for the maggots of black soldier flies, in which dead flies are area of Malaysia. This assumption is in line with integrated
harvested and sold to fish farms and poultry breeders (Worming Up, waste management facilities proposed under Twelfth Malaysia
2021). The renewable biogas generated from FW can also be valorized Plan (EPU, 2021) and a centralized system for FW treatment
as cooking gas and electricity. The Malaysia Energy Commission under the National Strategic Plan on FW Management (Bong
(2018) reported that 5.56 × 105 tonnes of liquefied petroleum gas and et al., 2017); hence relative impacts due to transportation for dif-
3.11 × 104 GWh of electricity were consumed for residential use in 2016. ferent technologies can be omitted;
This study also bridges the limitations in the previous literature by 4. The capacity of each FW valorization technology is capped at
performing hotspot analysis of sub-processes of each studied FW valori- 5000 tonnes/day (30% of total FW generated per day) to ensure the
zation technology to identify the key emissions from the hotspot area; diversity of valorization technologies and obtain higher product vari-
hence appropriate recommendations can be provided to reduce the bur- ety. This can reduce the reliance of FW management on one single
dens. After the integrated LCA-optimization evaluation, the robustness of technology or one single product market. Thus, risks such as inade-
key parameters affecting the optimal configuration is investigated quate labor force or technical experts, imbalance byproduct supply
through parametric sensitivity analysis. While Malaysia is selected as a and demand can be minimized.
case study, it is hoped that the results can provide a comprehensive and
data-driven decision to the policymakers in other countries when design- Four proposed scenarios encompassing different FW valorization
ing a sustainable FW management framework through enhancing the technologies, namely sanitary landfill, aerated windrow composting,
interlinkages of three main nexuses (i.e., people, planet, and prosperity) high-temperature drying sterilization, and AD, are compared to the
based on the SDG 2030 Agenda. existing FW disposal scenario (i.e., open landfill). These proposed tech-
nologies are chosen because they are appropriate for Malaysia's current
2. Methodology situation and policy needs (Woon et al., 2021). A functional unit is de-
fined as 1 tonne of FW (wet basis) treated in each scenario to provide
2.1. General framework of integrated life cycle multi-objective optimization a fair comparison. The system boundary is defined as “bin-to-cradle”,
model ranging from the treatment process in different FW disposal and valori-
zation scenarios to valuable byproducts. Fig. 2 shows the schematic pro-
A life cycle multi-objective optimization framework is developed as cess flow diagram of each FW valorization scenario with its respective
shown in Fig. 1. Stage 1 defines the system boundaries and problem individual sub-process. The descriptions of each scenario are shown in
statement. Stage 2 is the inventory analysis that compiles and quantifies Table 1.

3
Z. Lin, J.K. Ooi and K.S. Woon Science of the Total Environment 816 (2022) 151541

Fig. 1. The framework of the integrated life cycle multi-objective optimization model.

4
Z. Lin, J.K. Ooi and K.S. Woon Science of the Total Environment 816 (2022) 151541

Fig. 2. A schematic process flow diagram of each FW valorization scenario.

2.3. Stage 2: life cycle inventory composting are provided by Keng et al. (2020), as shown in Table A.8
in the supplementary material.
Input and output data, such as the elementary composition of FW (as For the byproducts production phase of the four proposed FW valo-
shown in Table A.2 in the supplementary material), air and leachate rization, the valuable byproducts replace the conventional products.
emissions, electricity generation, organic fertilizer generation, livestock Avoided environmental impacts for electricity production from S1 and
feed generation, cooking gas, and biogas fuel generation, are collected. S4a lead to offsets in the pollutant emissions from the current electricity
The collected data are converted by the defined functional unit and cat- generation based on the electricity generation mix (i.e., 43% coal, 40%
egorized based on the individual sub-process of each scenario in natural gas, and 17% hydro) in Malaysia (Energy Commission, 2018).
Tables A.4–A.12 in the supplementary material. The data are mainly Organic fertilizer in S2 substitutes the chemical fertilizer, in which the
sourced from on-site surveys (for open landfill), Malaysian Government quantity of organic fertilizer is obtained from Keng et al. (2020). Live-
official websites and reports, peer-reviewed journals, calculations using stock feed in S3 replaces the conventional livestock feed (i.e., 85% of soy-
International Panel on Climate Change guidelines provided in Table A.3 bean feed and 15% of fishmeal) based on a study conducted by Kim and
in the supplementary material, and Ecoinvent Database 3. The maxi- Kim (2010). For S4b, the biogas is upgraded into cooking gas to replace
mum and minimum ranges of certain data are included in the study to the liquefied petroleum gas, while the upgraded biogas in S4c substi-
provide the results' standard deviation. These ranges include the LFG tutes natural gas as biogas vehicle fuel. The avoided emissions of lique-
component by volume (i.e., 45–60% of CH4 and 40–60% of CO2) fied petroleum gas and natural gas are obtained from Edwards et al.
(ATSDR, 2001) in S1 and composition of biogas (e.g., 50–75% of CH4 (2014) and Jahirul et al. (2007), respectively. The emissions data for
and 25–50% of CO2) in S4 (Anukam et al., 2019). The air emissions in avoided products such as electricity mixed by fuel, chemical fertilizer,
S2 and the amount of organic fertilizer during aerated windrow soybean feed and fishmeal, liquefied petroleum gas, and natural gas

Table 1
The descriptions of the FW disposal and valorization scenarios.

Scenario Description

Scenario 0 (S0): open landfill (baseline This is the business-as-usual scenario in Malaysia, in which FW is discarded in the open landfill without any proper environmental
scenario) protection measures such as LFG and leachate collection treatment facilities. The FW bio-decomposition process produces uncol-
lected LFG and leachate, resulting in pollutant emissions to the atmosphere and groundwater. This scenario is constructed based on
an existing open landfill site in Bukit Beruntung, Hulu Selangor (with a treatment capacity of 1500 tonnes/day).
Scenario 1 (S1): sanitary landfill The landfill is equipped with proper LFG recovery and leachate collection systems. There are insufficient LFG emissions for electricity
generation in the first year of the landfill operation, so the LFG recovery rate is zero. From the third to tenth year of operation, the LFG
recovery rate increases to 35–70%, while the rest of the operation period (from the eleventh to thirtieth year), the recovery rate
reaches 90% (Woon and Lo, 2013). The collected LFG is sent to a gas turbine with 40% efficiency for electricity generation (Johari et al.,
2012). The electricity generated is used on-site, while excess electricity is sold to a national power plant based on a feed-in-tariff of
0.25 MYR/kWh (SEDA, 2021). Meanwhile, excess collected LFG is burned in a flaring system and then released into the atmosphere.
The leachate emissions standard follows Malaysia's Environmental Quality Regulation (DOE, 2010). 0.188 m3 of leachate is estimated
to be generated from 1 tonne of FW (Malakahmad et al., 2017).
Scenario 2 (S2): aerated windrow A 4.5 kW wood chipper is used to shred wood into sawdust with a 0.1 tonne/h (Keng et al., 2020). Then FW is mixed with the
composting sawdust in a mass ratio of 1:0.179 to control moisture and reduce the toxicity of FW (Kim and Kim, 2010). Then, the mixture
performs bio-reaction under the aerobic condition to enhance fermentation. The byproduct of S2 is sold as organic fertilizer to replace
existing commercial fertilizer.
Scenario 3 (S3): high-temperature FW is injected into a storage tank, followed by crushing and screening to control FW quality for livestock feed. The FW is then
drying sterilization homogenized to about 100 μm, and the water content is reduced to less than 12% via a pulse combustion drying technology at 140 °C.
Finally, the byproduct is stored in a vertical silo to maintain its quality before selling as a livestock feed (San Martin et al., 2016).
Scenario 4 (S4): anaerobic digestion FW is converted into biogas in an anaerobic biogas digester at mesophilic temperature (around 35 °C) (Jouhara et al., 2017). The
biogas generated is directly sent to a gas turbine by a 12 kW blower to generate electricity in S4a. For S4b and S4c, the biogas is
further upgraded via water scrubbing technology with 1% of CH4 fugitive emissions (Banks et al., 2018). Then, the purified biogas
(with 98% of CH4 composition) is used as cooking gas (i.e., S4b) and biogas vehicle fuel (i.e., S4c).

5
Z. Lin, J.K. Ooi and K.S. Woon Science of the Total Environment 816 (2022) 151541

are provided by Ecoinvent 3-allocation at point of substitution in the FW valorization technology capacity : ∑σs ≤ ∑CAPs  βs ð5Þ
s s
SimaPro 9.0 database.
The economic data are collected from Malaysia's governmental offi-
cial reports published by the Ministry of Housing and Local Government Renewable energy target : ∑Es ≤ ED ð6Þ
s
Malaysia. The same data source is used to ensure the data are standard-
ized by the methodology and year. Table A.13 in the supplementary ma-
Byproduct demand : ∑∑σs, p ≤ PD ð7Þ
terial summarizes the capital and operating cost of different FW s p

valorization technologies. The yield and market price of the byproducts


valorized from FW can be found in Table A.14 in the supplementary ma- There are three objective functions in this mathematical model.
terial. All costs are discounted to 2021 with a discount rate of 4%. The Eq. (8) represents the first objective function in the model which high-
proposed model can be readily adopted to other countries/regions by lights the total cost minimization of the FW management. The cost com-
using their specific data to create a more sustainable circularity FW ponents comprise the annualized capital and operation cost of the FW
management sector. valorization technologies and the revenue generated from the main
products. Details of each cost component are shown in Eqs. (9) to
2.4. Stage 3: life cycle impact assessment and interpretation (11). Eq. (12) illustrates the second objective function, which minimizes
the human health impact. Eq. (13) is the third objective function that
ReCiPe 2016 End-point (E) in the SimaPro 9.0 software is adopted to minimizes the ecosystems impact. The parameters of human health
categorize and characterize the data collected in Stage 2 into different and ecosystems are obtained from the end-point categories of LCA re-
mid-point and end-point categories. The mid-point category is defined sults as obtained in Stage 3.
as “problem-oriented approach” to evaluate the impact category at the
mid-point level, while the end-point category is a “damage-oriented ap- Min ðF1Þ ¼ OPEX þ CAPEX − REV ð8Þ
proach” obtained by aggregating the mid-point category (Pérez-
Camacho et al., 2018). 12 mid-point categories and two end-point cate- OPEX ¼ ∑ðUOPEX  σsÞ ð9Þ
s
gories (i.e., human health and ecosystems) are evaluated in this study as
provided in Table A.15 in the supplementary material. These mid-point
categories are frequently investigated for the studied FW valorization CAPEX ¼ ∑βs  ACAPEXs  ns ð10Þ
s
technologies (Guven et al., 2019). Eight mid-point categories, namely
global warming (human health and terrestrial ecosystems), fine partic-
REV ¼ ∑ðPRICE  σs, pÞ ð11Þ
ulate matter formation, human carcinogenic toxicity, human non- s
carcinogenic toxicity, terrestrial acidification, and freshwater eutrophi-
cation, and marine ecotoxicity, are selected for further analysis as these Min ðF2Þ ¼ ∑ðHealths  σsÞ ð12Þ
s
eight mid-point categories are contributing significant results with ac-
counting for more than 2% of environmental impact for the studied
end-point categories. After characterizing the data into different mid- Min ðF3Þ ¼ ∑ðEcos  σsÞ ð13Þ
s
point and end-point categories, hotspot analysis is conducted to identify
the individual sub-process of each scenario that incurs the highest envi-
The proposed integrated LCA-optimization model adopts a multi-
ronmental burdens; hence appropriate recommendations are provided.
objective MILP to solve problems as MILP can handle constraints that the
variables can be both integers or non-integers. It is programmed using
2.5. Stage 4: mathematical model formulation
GAMS and solved with the CPLEX solver (GAMS, 2019). There are three
conflicting objective functions; therefore, the AUGMENCON2 method, as
The mathematical model comprising objective functions and con-
defined in Eq. (14), is used to solve the problem. AUGMENCON2 imple-
straints is developed in this stage. As indicated from Eqs. (1) to (7),
ments lexicographic optimization on all the objective functions to identify
four major constraints are formulated in this mathematical model.
the alternative optima and form the pay-off table (Mavrotas and Florios,
Eq. (1) emphasizes the availability of ith type FW that is allocated to dif-
2013). The range of the objective functions is calculated and divided into
ferent valorization technologies. Eq. (2) represents the flow of FW in the
equal intervals using intermediate equidistant grid points. Total grid
process stream. Eq. (3) shows that the product output leaving the FW
points are obtained to vary the right-hand side of the objective function.
valorization technologies. Eq. (4) indicates the mass balance between
Ultimately, a set of Pareto-optimal solutions, which consists of equally
input and output streams. Eq. (5) ensures the total FW received is distrib-
good alternatives with different trade-offs, is obtained.
uted among the valorization technologies without exceeding their capac-
ity. Renewable energy target constraint is described by Eq. (6). The total   
capacity of the electricity generating FW valorization technologies built max f 1 ðxÞ þ eps  s2 =r2 þ 10−1 s3 =r3 þ . . . þ 10−ðp−2Þ sp =rp ð14Þ
shall not exceed the Malaysia renewable energy mix target (i.e., 20% of
electricity generation mix by 2025) to avoid over-investment. Eq. (7) em- such that f2(x) − s2 = e2, eps = [10−6, 10−3].
phasizes that the supply of the valuable byproducts generated from FW
valorization technologies should not exceed the consumption of that 2.6. Stage 5: post-optimization analysis
byproduct in Malaysia to prevent overproduction.
The optimal solutions are analyzed following the current policy
FW availability and mass balance : W ≥ ∑σs  βs ð1Þ needs and local conditions in Malaysia. An optimal mix of FW valoriza-
s
tion technology configuration is proposed to assist the policymakers in
in ¼ ∑σs  βs ð2Þ developing a circular FW management. To tackle the uncertainties in
s
the key parameters, the effect of the variation in byproduct price on
the FW allocation on the scenario is evaluated through parametric sen-
out ¼ ∑∑CCs, p  ∑∑ðβs, p  σsÞ ð3Þ
s p s p sitivity analysis to accommodate the high volatility of byproduct price in
the future. The capacity of each FW valorization technology is adjusted
to +50% and +100% to observe the changes in the optimal configura-
in ≥ out ð4Þ tion of the FW management network.

6
Z. Lin, J.K. Ooi and K.S. Woon Science of the Total Environment 816 (2022) 151541

3. Result and discussion natural gas) in the current Malaysia electricity generation mix to biogas
via AD for electricity generation avoids significant emissions reduction in
3.1. Life cycle impact assessment and interpretation pollutants, including a reduction of 419 g PM2.5/tonne FW, 733 g SO2/
tonne FW, and 11.8 g Zn/tonne FW. Converting FW to livestock feed in
Fig. 3 indicates eight mid-point and two end-point categories results S3 (i.e., high-temperature drying sterilization) shows the least environ-
for the sub-processes of five scenarios. The other four mid-points are mental burden for global warming (human health and terrestrial ecosys-
provided in Fig. A.1 in the supplementary material because they only ac- tems). This result is in line with Laso et al. (2018), in which valorizing FW
count for 0.01–0.14% of the studied end-point category (i.e., ecosystems). into livestock feed has comparatively lower GHG emissions than sanitary
The quantitative results of each sub-processes can be found in Table A.16 landfill. The rank order of five scenarios for 12 mid-point categories is
in the supplementary material. Negative values indicate that the envi- tabulated in Table A.17 in the supplementary material.
ronmental benefit is gained by the corresponding sub-process, whereas To identify the key emissions from these five scenarios, hotspot
positive values refer to the environmental burden incurred in the respec- analysis is conducted by investigating each scenario's sub-process that
tive sub-process. The black square dots represent the net environmental incurs the highest environmental burdens for respective mid-point cat-
impact (i.e., the sum of each scenario's environmental burdens and envi- egories. In Fig. 3(a) and (e), fugitive LFG emissions in S1 (i.e., sanitary
ronmental benefits). The error bars indicate the standard deviation due landfill) incur lower impacts in global warming (human health and ter-
to the variability for emissions and production associated with each restrial ecosystems) than LFG with no collection system in S0. S1 has
scenario. significantly reduced 71.3% of CH4 emissions to the atmosphere as
Among the studied mid-point categories, S0 (i.e., open landfill) indi- compared to S0. This is because the installations of LFG recovery sys-
cates the highest-burden with the largest positive values in global tems for electricity generation and flaring systems prevent polluting
warming (human health and terrestrial ecosystems), human non- gases (e.g., CH4 and non-methane volatile organic compounds) into
carcinogenic toxicity, and marine ecotoxicity. S2 (i.e., aerated windrow the atmosphere during the lifespan of sanitary landfills (Malakahmad
composting) incurs the worst result for fine particulate matter formation et al., 2017). Besides, leachate emission in S0 contributes the highest en-
and terrestrial acidification due to NH3 emissions to the atmosphere. S4b vironmental burdens in human non-carcinogenic toxicity and marine
(i.e., cooking gas generation from AD) provides the most burden on ecotoxicity, mainly attributing to 44.4 g Zn/tonne FW emissions as
human carcinogenic toxicity and freshwater eutrophication because of depicted in Fig. 3(d) and (h). The installation of high-density polyethyl-
the electricity consumption in the biogas upgrading process. S4a ene layers as lining systems for leachate collection in S1 prevents heavy
(i.e., electricity generation from AD) contributes the highest environmen- metals (e.g., Zn, Ni, Hg, and Cd) emissions to groundwater (Hussein
tal benefits on six mid-point categories, including fine particulate matter et al., 2019), leading to a 76.2–99.2% reduction of heavy metals emis-
formation, human carcinogenic toxicity, human non-carcinogenic toxic- sions compared to S0. Hence, S1 does not cause any significant environ-
ity, terrestrial acidification, freshwater eutrophication, and marine mental burdens in human non-carcinogenic toxicity and marine
ecotoxicity. The fuel switching from conventional fuel (i.e., coal and ecotoxicity. The error bars in Fig. 3(b) and (f) indicate that S0 and S1

2.0x10-3 2.0x10-4
Fine particulate matter formation (DALY)

(a) (b)
Global warming, Human health (DALY)

1.0x10-3 1.0x10-4

0.0
0.0

-1.0x10-3
-1.0x10-4
-2.0x10-3
-2.0x10-4
-3.0x10-3

-3.0x10-4
-4.0x10-3

-5.0x10-3 -4.0x10-4

-6.0x10-3 -5.0x10-4
Scenario 0 Scenario 1 Scenario 2 Scenario 3 Scenario 4a Scenario 4b Scenario 4c Scenario 0 Scenario 1 Scenario 2 Scenario 3 Scenario 4a Scenario 4b Scenario 4c

5.0x10-4 3.0x10-2
Human non-carcinogenic toxicity (DALY)

(c) (d)
Human carcinogenic toxicity (DALY)

2.5x10-2
0.0

2.0x10-2
-5.0x10-4
1.5x10-2

-1.0x10-3 1.0x10-2

5.0x10-3
-1.5x10-3
0.0

-2.0x10-3
-5.0x10-3

-2.5x10-3 -1.0x10-2
Scenario 0 Scenario 1 Scenario 2 Scenario 3 Scenario 4a Scenario 4b Scenario 4c Scenario 0 Scenario 1 Scenario 2 Scenario 3 Scenario 4a Scenario 4b Scenario 4c

Fig. 3. Mid-point results (per tonne of FW on wet basis) generated by each scenario (a) global warming human health; (b) fine particulate matter formation; (c) human carcinogenic
toxicity; (d) human non-carcinogenic toxicity; (e) global warming terrestrial ecosystems; (f) terrestrial acidification; (g) freshwater eutrophication; (h) marine ecotoxicity. End-point
results (per tonne of FW on wet basis) generated by each scenario (i) human health; (j) ecosystems.

7
Z. Lin, J.K. Ooi and K.S. Woon Science of the Total Environment 816 (2022) 151541

Global warming, Terrestrial ecosystems (species.yr)


4.0x10-6 3.0x10-7
(e) (f)
3.0x10-6

Terrestrial acidification (species.yr)


2.5x10-7
2.0x10-6
1.0x10-6 2.0x10-7
0.0 1.5x10-7
-1.0x10-6
-6 1.0x10-7
-2.0x10
-3.0x10-6 5.0x10-8
-4.0x10-6
0.0
-5.0x10-6
-6.0x10-6 -5.0x10-8
-7.0x10-6 -1.0x10-7
-8.0x10-6
-9.0x10 -6 -1.5x10-7

-1.0x10-5 -2.0x10-7
-1.1x10-5
-2.5x10-7
Scenario 0 Scenario 1 Scenario 2 Scenario 3 Scenario 4a Scenario 4b Scenario 4c Scenario 0 Scenario 1 Scenario 2 Scenario 3 Scenario 4a Scenario 4b Scenario 4c

2x10-8 1.5x10-5
(g) (h)
Freshwater eutrophication (species.yr)

1x10-8

Marine ecotoxicity (species.yr)


0
1.0x10-5
-1x10-8

-2x10-8

-3x10-8
5.0x10-6
-4x10-8

-5x10-8

-6x10-8 0.0
-7x10-8

-8x10-8

-9x10-8 -5.0x10-6
Scenario 0 Scenario 1 Scenario 2 Scenario 3 Scenario 4a Scenario 4b Scenario 4c Scenario 0 Scenario 1 Scenario 2 Scenario 3 Scenario 4a Scenario 4b Scenario 4c

3.00x10-2 2x10-5
(i) (j)
Ecosystems (species.yr)
Human health (DALY)

1x10-5
1.50x10-2

0.00

-1x10-5

-1.50x10-2
Scenario 0 Scenario 1 Scenario 2 Scenario 3 Scenario 4a Scenario 4b Scenario 4c Scenario 0 Scenario 1 Scenario 2 Scenario 3 Scenario 4a Scenario 4b Scenario 4c

Fig. 3 (continued).

have high standard deviations in fine particulate matter formation biochar is effective due to its porosity and high nutrient absorbability
(i.e., 3.12 × 10−5 DALY/tonne FW for S0 and 2.48 × 10−5 DALY/tonne (Awasthi et al., 2020). Fig. 3 (c) and (g) illustrate that the shredding
FW for S1) and terrestrial acidification (i.e., 8.59 × 10−8 species·yr/ process in S2 causes the highest environmental loads in human
tonne FW for S0 and 8.36 × 10−8 species·yr/tonne FW for S1). This is carcinogenic toxicity and freshwater eutrophication. The dominant
administered by the high volume ratio variation of NH3 in LFG emis- contributor is fossil fuels combustion due to electricity consumption
sions (i.e., 0.1–1%). At the maximum point of error bars in S0, the aggre- from shredding machines. Despite the aerobic bio-reaction process
gated results of S0 generates higher values in fine particulate matter emits 0.15 kg N2O/tonne FW to the atmosphere as depicted in Fig. 3
formation (i.e., 8.53 × 10−5 DALY/tonne FW) and terrestrial acidifica- (a) and (e), S2 outperforms 62% in global warming (human health and
tion (i.e., 2.35 × 10−7 species·yr/tonne FW) than the net results terrestrial ecosystems) as compared to S0, providing a similar finding
(i.e., black square dots) of S2 in fine particulate matter formation with Lim et al. (2019).
(i.e., 6.97 × 10−5 DALY/tonne FW) and terrestrial acidification As indicated in Fig. 3(a) and (e), high amounts of GHGs are emitted
(i.e., 2.15 × 10−7 species·yr/tonne FW). in S3 due to high electricity (i.e., 6.76 kWh/tonne FW) and natural gas
As displayed in Fig. 3(b) and (f), the aerobic bio-reaction process is the consumption (i.e., 261.6 kWh/tonne FW) in the pulse combustion dry-
hotspot area in S2 that incurs the worst results in fine particulate matter ing process, contributing to 31% of the total electricity used in the plant.
formation and terrestrial acidification, mainly due to 528 g NH3/tonne FW San Martin et al. (2016) reported that a rotatory oven could be applied
emissions to the atmosphere. To reduce the NH3 emissions in S2, adding to avoid high energy consumption in the drying process, in which 34%

8
Z. Lin, J.K. Ooi and K.S. Woon Science of the Total Environment 816 (2022) 151541

of energy consumption can be reduced compared to pulse combustion human health. Meanwhile, S4b contributes a significant environmental
drying. Although a high electricity consumption in the drying process, benefit (i.e., −2.39 × 10−7 species·yr/tonne FW) in global warming ter-
the livestock feed production in S3 compensates the environmental restrial ecosystems, making S4b (i.e., −2.13 × 10−7 species·yr/tonne
burdens by replacing soybean feed and fishmeal, rendering a negative FW) ranks better than S4c (i.e., −1.46 × 10−7 species·yr/tonne FW)
value (i.e., environmental benefits) to global warming human health in ecosystems.
(i.e., −3.34 × 10−3 DALY/tonne FW) and global warming terrestrial
ecosystems (i.e., −6.68 × 10−6 species·yr/tonne FW). 3.2. MILP optimization results with three-dimensional Pareto frontier
For S4, the environmental burdens in global warming human health
and global warming terrestrial ecosystems are mainly caused by anaer- Based on the end-point results described in Section 3.1 and the total
obic bio-reaction due to the fugitive CH4 emissions in S4a and electricity cost collected in Stage 2, 10 Pareto-optimal solutions for FW manage-
consumption by biogas upgrading process in S4b and S4c. IRENA (2017) ment in Malaysia are generated using the augmented ε-constraint
reported that chemical scrubbers as biogas upgrading technology could method via GAMS software. The 10 Pareto-optimal solutions are non-
be adopted due to low electricity demand (i.e., 0.09 kWh/m3 biogas). dominated alternatives with trade-offs from the most economical to
Fig. 3(i) and (j) depict the two end-point categories, namely human least damaging burden (i.e., human health and ecosystems). Since
health and ecosystems, for all studied scenarios. It is worth noting that there are three interdependence objective functions, the solutions are
human non-carcinogenic toxicity affects the most under the human plotted into a three-dimensional Pareto frontier, as shown in Fig. 4.
health end-point category, while global warming terrestrial ecosystems The exact values for each Pareto-optimal solution can be found in
shows the highest contribution to the ecosystems end-point category. Table A.19 in the supplementary material.
S4a is the most environmentally friendly scenario, while S0 is the When the Pareto-optimal solutions transit from the most economi-
worst performing scenario for both end-point categories. The results in- cal solution (1st solution) to the least damaging solution (10th solu-
dicate that valorizing FW into electricity via AD (i.e., S4a) contributes tion), the Pareto-optimality changes are not linear as the distance
146% and 161% reduction of human health and ecosystems, respec- between solutions narrows down. This finding is supported by Ooi
tively, compared with S0. and Woon (2021), which reported the non-uniform gaps of the Pareto
The detailed rankings of the scenarios for these two end-point cate- curve. Starting from the most economical solution, an increase in the
gories are provided in Table A.18 in the supplementary material. All five total cost of 0.43 MYR/tonne FW reduces the human health and ecosys-
scenarios show similar ranking locations in both end-point categories, tems of the FW management system by 1.8% and 2.0%, respectively, in-
only a swap between S4b and S4c. S4b incurs higher overall adverse im- curring a significant gap between the 1st and 2nd Pareto-optimal
pacts on the fine particulate matter formation, human carcinogenic tox- solution. When it is approaching the least damaging solution (i.e., taking
icity, and human non-carcinogenic toxicity. The production of cooking the 8th and 9th solutions as an example), the gap is narrower since an in-
gas is insufficient to compensate for the environmental burden caused crease in the total cost of 0.68 MYR/tonne has a minimal impact of 0.7%
by electricity consumption, rendering a worse ranking than S4c in and 0.3% on reducing human health and ecosystems, respectively.

5
6
7 8 9 10

Fig. 4. Three-dimensional Pareto frontier of 10 Pareto-optimal solutions.

9
Z. Lin, J.K. Ooi and K.S. Woon Science of the Total Environment 816 (2022) 151541

Fig. 5 demonstrates the amount of FW allocated into different FW energy mix to 20% by 2025. The proposed FW management optimal
valorization technologies to achieve optimal conditions, as shown in configuration can contribute to the carbon-neutral target by 2050, as
the 10 Pareto-optimal solutions. The result shows that four types of sce- stated in the Twelfth Malaysia Plan (EPU, 2021).
narios, namely S4a, S4b, S3, and S1, should be adopted in the FW man-
agement system in Malaysia. S0 (i.e., open landfill), which is currently 3.3. Sensitivity analysis
practiced in FW management in Malaysia, has ceased to exist in all
Pareto-optimal solutions as open landfill lacks adequate leachate and 3.3.1. The effect of capacity
LFG collection systems; hence no valuable byproduct is generated. It is The capacity of each proposed FW valorization technology is capped
interesting to note that S4c is not chosen as one of the configurations at 5000 tonnes/day, as mentioned in Section 2.2. Since S1, S3, S4a, and
in the Pareto-optimal solutions, mainly attributed to the high cost in- S4b are selected for the blueprint for the FW management system in
curred in biogas upgrading unit and lower avoided emissions in pollut- Malaysia, the capacity constraints of these four scenarios are relaxed
ants than the other scenarios. to observe the effect of capacity on the outcome. The maximum capacity
S4a and S3 are both at maximum capacity throughout all the Pareto- is increased by 50% and 100%, and the results are demonstrated in Fig. 6.
optimal solutions, indicating that they are the most cost-effective, S1 appears to be the most sensitive FW valorization technology as the
human-health-friendly, and ecosystem-friendly alternatives. Therefore, optimization model excludes it after increasing the maximum capacity
sensitivity analysis on their capacity is carried out to provide a more in- by 50% and 100%. This illustrates that S1 is the least attractive one
depth discussion for both scenarios. A prominent trend is observed among the four scenarios.
when the solutions move from the most economical solution (i.e., 1st As shown in Fig. 6(b), S3 is the most attractive valorization technol-
solution) to the least damaging solution (i.e., 10th solution), the capac- ogy as it occupies almost all Pareto-optimal solutions in full capacity ex-
ity of S4b decreases by 2665 tonnes/day (16% of total FW) and S1 in- cept the last three Pareto-optimal solutions (i.e., 8th, 9th, and 10th) of
creases by 3205 tonnes/day (19% of total FW). This suggests that S1 is 100% maximum capacity increment. This is mainly due to the lucrative
less damaging but more expensive compared to S4b. This is because of profit and high avoided emissions of livestock feed production from
the high yield of biogas from AD and the high selling price of cooking S3. When maximum capacity is set to 10,000 tonnes/day, the percent-
biogas, providing a higher revenue generated for S4b. It is important age of S3 drops accordingly for the last three Pareto-optimal solutions
to note that the price of byproducts is region-specific, and hence locally and is replaced by S4a by 5.11%, 19.12%, and 33.12%, respectively. This
representative data is essential. Therefore, sensitivity analysis on shows that S4a is the second attractive valorization technology from
byproduct selling price is discussed in Section 3.3.2 to investigate the ro- the cost, human health, and ecosystems perspective. S4a has a higher
bustness of this parameter. total cost but generates more benefits to human health and the ecosys-
Since human health and ecosystems show negative values tems than S3. When comparing S4a and S4b, S4a shows an increasing
(i.e., environmental benefits) for all Pareto-optimal solutions, the 1st so- trend from the most economical solution towards the least damaging
lution (i.e., the most economical solution) is recommended as the blue- solutions (i.e., from the 1st to 10th Pareto-optimal solution), whereas
print for the FW management system in Malaysia. The configuration of S4b has a decreasing trend. This demonstrates that S4b is more econom-
the 1st solution is 9% sanitary landfill, 3% aerated windrow composting, ically favorable, whereas S4a is more human-health-friendly and
30% high-temperature drying sterilization, 30% AD to electricity, and ecosystem-friendly.
28% AD to cooking gas. This result is echoed to the waste management
policies in several European countries (e.g., Finland and Germany), 3.3.2. The effect of the byproducts selling price
which limit landfill activity since the 1990s (EEA, 2013). Sanitary land- A sensitivity analysis is performed with all the selling prices varied
fills and AD cut fossil fuel consumption, avoiding considerable GHG by ±50% to accommodate the impact of the volatile selling price of
emissions due to renewable energy generation. The bioenergy products byproducts. The results in Fig. 7 show that the price fluctuations of elec-
(i.e., cooking gas and electricity from S1, S4a, and S4b) correspond to tricity, cooking gas, and livestock feed give a different degree of influ-
Malaysia National Renewable Energy Policy of achieving the renewable ence. The variations in vehicle fuel (S4c) and organic fertilizer (S2)

Fig. 5. Configuration of FW valorization technologies of each Pareto-optimal solution.

10
Z. Lin, J.K. Ooi and K.S. Woon Science of the Total Environment 816 (2022) 151541

Fig. 6. The effect of increasing in capacity constraint by 50% and 100% on the proposed FW management system (a) S1: sanitary landfill; (b) S3: high-temperature drying sterilization;
(c) S4a: anaerobic digestion to electricity; (d) S4b: anaerobic digestion to cooking gas.

prices do not affect the optimal configuration as the degree of reduc- show that S4a (AD to electricity) outperforms other FW valorization
tions in total cost are insufficient to compensate for the human health technologies in both end-point categories (i.e., −1.28 × 10−2 DALY/
and ecosystems impact. tonne FW in human health and −1.03 × 10−5 species·yr/tonne FW
An increment of electricity price by 50% does not alter the optimal in ecosystems), while open landfill (i.e., business-as-usual scenario)
configuration of the FW management system. Nevertheless, if the incurs the most burdens in both end-point categories. Compared to
electricity price decreases, the optimal configuration changes signif- S0 (open landfill), S4a contributes 146% and 161% reduction of
icantly, where S4a is selected in the last three Pareto-optimal solu- human health and ecosystems, respectively. For hotspot analysis,
tions (i.e., 8th, 9th, and 10th), occupying 9.25%, 23.23%, and 29.96% the leachate emissions in S0 (open landfill) has the most significant
in the total FW amount respectively. This is because the profit of burden for both end-point categories due to heavy metals (e.g., Zn,
electricity-generating valorization technology can no longer cover Ni, Cu, and Hg) emissions to groundwater, followed by the drying
its high capital and operational cost. In this case, S4b and S0 are cho- process in S3 (high-temperature drying sterilization) due to high
sen. S0 is selected as an alternative in the most economical solution electricity consumption in the dryer.
due to its low capital and operational costs. However, it should be By integrating the LCA end-point results with the MILP optimiza-
emphasized that S0 is not a viable option for treating FW due to the tion, 10 Pareto-optimal solutions that optimize three objective func-
significant human health and ecosystems burdens as described in tions (i.e., human health, ecosystems, total cost) are generated. The
Section 3.1. optimal configurations of FW management of all Pareto-optimal so-
As the cooking gas price increases, S4b has become one of the lutions include four types of most favorable scenarios, ranking from
most favorable scenarios in the least damaging solution. However, S3 (high-temperature drying sterilization), S4a (AD to electricity),
the decrease in cooking gas price renders S4b to be wholly replaced S4b (AD to cooking gas), S1 (sanitary landfill), and S2 (aerated
by S1 and S2. For S3, the increase in livestock feed price has no effect windrow composting). Since all of the Pareto-optimal solutions ex-
on the result as the capacity for all solutions is full. In contrast, the hibit human health and ecosystems benefit, the most economical so-
decrease in livestock feed price causes S1 and S2 to be more cost- lution, namely 9% sanitary landfill, 3% aerated windrow composting,
effective than S3. Hence, S3 only appears in the last four solutions, 30% high-temperature drying sterilization, 30% AD to electricity, and
occupying 3.57%, 12.38%, 21.37%, and 29.96% in the total FW amount, 28% AD to cooking gas, is recommended for future FW management
respectively. policy planning in Malaysia. The decrease in electricity price signifi-
cantly impacts the FW management system, followed by cooking gas
4. Conclusion and livestock feed prices. The proposed integrated life cycle multi-
objective optimization model is expected to provide comprehensive
A simple and deterministic life cycle multi-objective optimiza- and data-driven insights into FW management planning. This, in
tion model that integrates LCA and mathematical optimization is de- turn, creates a more sustainable circularity FW strategy, resource-
veloped to foster synergies and maximize benefits across human efficient, and greener economy. The explicit relationship between
health, environment, and economic nexus for sustainable FW man- the modeling results with Malaysia's national targets and policies
agement planning. The developed framework is applied to five FW can be investigated for further research. Parametric sensitivity anal-
disposal and valorization scenarios by assessing 12 mid-point cate- ysis related to the cost of FW valorization technologies can also be
gories, 2 end-point categories (i.e., human health and ecosystems), studied to investigate the cost variation impact on the modeling
and cost impact based on Malaysia's circumstances. The results results.

11
Z. Lin, J.K. Ooi and K.S. Woon Science of the Total Environment 816 (2022) 151541

Fig. 7. The effect of adjusting the byproducts selling price by +50% and −50% on the proposed FW management system (a) electricity price increases by 50%; (b) electricity price decreases
by 50%; (c) cooking gas price increases by 50%; (d) cooking gas price decreases by 50%; (e) vehicle fuel increases by 50%; (f) vehicle fuel decreases by 50%; (g) compost increases by 50%;
(h) compost decreases by 50%; (i) livestock feed increases by 50%; (j) livestock feed decreases by 50%.

CRediT authorship contribution statement Conceptualization, Methodology, Software, Formal analysis,


Writing - original draft. Kok Sin Woon: Supervision, Conceptu-
Zuchao Lin: Conceptualization, Methodology, Software, alization, Validation, Writing – review & editing, Funding acqui-
Formal analysis, Writing - original draft. Jun Keat Ooi: sition.

12
Z. Lin, J.K. Ooi and K.S. Woon Science of the Total Environment 816 (2022) 151541

Declaration of competing interest FAO, 2013. Food Wastage Footprint - Impact on Natural Resources (Summary Report).
Food and Agriculture Organization of the United Nations.
GAMS, 2019. General Algebraic Modeling System V28.2.0. GAMS Development Corp.
The authors declare that they have no known competing financial Garibay-Rodriguez, J., Laguna-Martinez, M.G., Rico-Ramirez, V., Botello-Alvarez, J.E., 2018.
interests or personal relationships that could have appeared to influ- Optimal municipal solid waste energy recovery and management: a mathematical
programming approach. Comput. Chem. Eng. 119. https://doi.org/10.1016/j.
ence the work reported in this paper.
compchemeng.2018.09.025.
Guven, H., Wang, Z., Eriksson, O., 2019. Evaluation of future food waste management al-
Acknowledgments ternatives in Istanbul from the life cycle assessment perspective. J. Clean. Prod. 239.
https://doi.org/10.1016/j.jclepro.2019.117999.
Hettiarachchi, H., Meegoda, J.N., Ryu, S., 2018. Organic waste buyback as a viable method
The authors would like to thank for the financial support from the to enhance sustainable municipal solid waste management in developing countries.
Ministry of Higher Education Malaysia through the Fundamental Int. J. Environ. Res. Public Health 15, 1–15. https://doi.org/10.3390/ijerph15112483.
Research Grant Scheme (FRGS/1/2020/TK0/XMU/02/2). Hoang, G.M., Fujiwara, T., Pham Phu, T.S., Nguyen, L.D., 2019. Sustainable solid waste
management system using multi-objective decision-making model: a method for
maximizing social acceptance in Hoi An city, Vietnam. Environ. Sci. Pollut. Res. 26.
Appendix A. Supplementary data https://doi.org/10.1007/s11356-018-3498-5.
Hussein, M., Yoneda, K., Zaki, Z.M., Othman, N.A., Amir, A., 2019. Leachate characteriza-
Supplementary data to this article can be found online at https://doi. tions and pollution indices of active and closed unlined landfills in Malaysia. Environ.
Nanotechnol. Monit. Manag. 12. https://doi.org/10.1016/j.enmm.2019.100232.
org/10.1016/j.scitotenv.2021.151541. IRENA, 2017. Biogas for Road Vehicles Technology Brief. IREA International Renewable
Energy Agency.
References Jahirul, M.I., Saidur, R., Hasanuzzaman, M., Masjuki, H.H., Kalam, M.A., 2007. A comparison
of the air pollution of gasoline and CNG driven car for Malaysia. Int. J. Mech. Mater.
Abdul Hamid, Z., 2017. Towards a carbon-neutral Malaysia by 2050 [WWW Document]. Eng. 2, 130–138.
URLNew Straits Times (accessed 4.21.21) https://www.nst.com.my/opinion/ Jin, C., Sun, S., Yang, D., Sheng, W., Ma, Y., He, W., Li, G., 2021. Anaerobic digestion: an al-
columnists/2017/11/302480/towards-carbon-neutral-malaysia-2050. ternative resource treatment option for food waste in China. Sci. Total Environ. 779,
Albizzati, P.F., Tonini, D., Astrup, T.F., 2021. High-value products from food waste: an en- 146397. https://doi.org/10.1016/j.scitotenv.2021.146397.
vironmental and socio-economic assessment. Sci. Total Environ. 755, 142466. https:// Johari, A., Ahmed, S.I., Hashim, H., Alkali, H., Ramli, M., 2012. Economic and environmental
doi.org/10.1016/j.scitotenv.2020.142466. benefits of landfill gas from municipal solid waste in Malaysia. Renew. Sustain. En-
Anukam, A., Mohammadi, A., Naqvi, M., Granström, K., 2019. A review of the chemistry of ergy Rev. https://doi.org/10.1016/j.rser.2012.02.005.
anaerobic digestion: methods of accelerating and optimizing process efficiency. Pro- Jouhara, H., Czajczyńska, D., Ghazal, H., Krzyżyńska, R., Anguilano, L., Reynolds, A.J.,
cesses https://doi.org/10.3390/PR7080504. Spencer, N., 2017. Municipal waste management systems for domestic use. Energy
ATSDR, Agency for Toxic Substances and Disease Registry, 2001. Landfill Gas Primer - An 139. https://doi.org/10.1016/j.energy.2017.07.162.
Overview for Environmental Health Professionals - Chapter 2: Landfill Gas Basics Kaza, S., Yao, L.C., Bhada-Tata, P., Van Woerden, F., 2018. What a Waste 2.0: A Global
[WWW Document]. URLAgency Toxic Subst. Dis. Regist (accessed 4.20.21). https:// Snapshot of Solid Waste Management to 2050, What a Waste 2.0: A Global Snapshot
www.atsdr.cdc.gov/HAC/landfill/html/ch2.html. of Solid Waste Management to 2050. https://doi.org/10.1596/978-1-4648-1329-0.
Awasthi, S.K., Sarsaiya, S., Awasthi, M.K., Liu, T., Zhao, J., Kumar, S., Zhang, Z., 2020. Keng, Z.X., Chong, S., Ng, C.G., Ridzuan, N.I., Hanson, S., Pan, G.T., Lau, P.L., Supramaniam,
Changes in global trends in food waste composting: research challenges and oppor- C.V., Singh, A., Chin, C.F., Lam, H.L., 2020. Community-scale composting for food
tunities. Bioresour. Technol. 299, 122555. https://doi.org/10.1016/j.biortech.2019. waste: a life-cycle assessment-supported case study. J. Clean. Prod. 261, 121220.
122555. https://doi.org/10.1016/j.jclepro.2020.121220.
Banks, C., Heaven, S., Zhang, S., Baier, U., 2018. Food waste digestion: anaerobic digestion Kim, M.H., Kim, J.W., 2010. Comparison through a LCA evaluation analysis of food waste
of food waste for a circular economy. IEA Bioenergy Task, p. 37. disposal options from the perspective of global warming and resource recovery. Sci.
Beretta, C., Hellweg, S., 2019. Potential environmental benefits from food waste preven- Total Environ. 408. https://doi.org/10.1016/j.scitotenv.2010.04.049.
tion in the food service sector. Resour. Conserv. Recycl. 147, 169–178. https://doi. Laso, J., Margallo, M., García-Herrero, I., Fullana, P., Bala, A., Gazulla, C., Polettini, A., Kahhat,
org/10.1016/j.resconrec.2019.03.023. R., Vázquez-Rowe, I., Irabien, A., Aldaco, R., 2018. Combined application of life cycle
Bernama, Malaysian National News Agency, 2020. Food waste drops during MCO, rises assessment and linear programming to evaluate food waste-to-food strategies: seek-
soon after [WWW document]. URLHarakah Dly (accessed 9.20.21). https:// ing for answers in the nexus approach. Waste Manag. 80, 186–197. https://doi.org/
harakahdaily.net/index.php/2020/10/20/food-waste-drops-during-mco-rises-soon- 10.1016/j.wasman.2018.09.009.
after/. Lim, L.Y., Lee, C.T., Bong, C.P.C., Lim, J.S., Klemeš, J.J., 2019. Environmental and economic
Bernama, Malaysian National News Agency, 2021. Tuan Ibrahim: Govt’s approach to cli- feasibility of an integrated community composting plant and organic farm in
mate change issues outlined in MyCAC [WWW document]. URLNew Straits Times Malaysia. J. Environ. Manag. 244, 431–439. https://doi.org/10.1016/j.jenvman.2019.
(accessed 4.15.21). https://www.nst.com.my/news/nation/2021/04/682056/tuan- 05.050.
ibrahim-govts-approach-climate-change-issues-outlined-mycac. Lo, I.M.C., Woon, K.S., 2016. Food waste collection and recycling for value-added products:
Bong, C.P.C., Ho, W.S., Hashim, H., Lim, J.S., Ho, C.S., Peng Tan, W.S., Lee, C.T., 2017. Review potential applications and challenges in Hong Kong. Environ. Sci. Pollut. Res. 23,
on the renewable energy and solid waste management policies towards biogas de- 7081–7091. https://doi.org/10.1007/s11356-015-4235-y.
velopment in Malaysia. Renew. Sust. Energ. Rev. 70, 988–998. https://doi.org/10. MAEKO, Mentari Alam E.K.O., 2021. Maeko food waste counter [WWW document]. URL
1016/J.RSER.2016.12.004. https://www.maeko.com.my/ (accessed 9.20.21).
Chen, T., Shen, D., Jin, Y., Li, H., Yu, Z., Feng, H., Long, Y., Yin, J., 2017. Comprehensive eval- Malakahmad, A., Abualqumboz, M.S., Kutty, S.R.M., Abunama, T.J., 2017. Assessment of
uation of environ-economic benefits of anaerobic digestion technology in an inte- carbon footprint emissions and environmental concerns of solid waste treatment
grated food waste-based methane plant using a fuzzy mathematical model. Appl. and disposal techniques; case study of Malaysia. Waste Manag. 70. https://doi.org/
Energy 208. https://doi.org/10.1016/j.apenergy.2017.09.082. 10.1016/j.wasman.2017.08.044.
Chinchodkar, K.N., Jadhav, O.S., 2018. Optimal Planning for Aurangabad Municipal Solid Malmir, T., Ranjbar, S., Eicker, U., 2020. Improving municipal solid waste management
Waste through Mixed Integer Linear Programming. 13, pp. 11883–11887. strategies of Montréal (Canada) using life cycle assessment and optimization of tech-
Chiu, S.L.H., Lo, I.M.C., Woon, K.S., Yan, D.Y.S., 2016. Life cycle assessment of waste treat- nology options. Energies 13. https://doi.org/10.3390/en13215701.
ment strategy for sewage sludge and food waste in Macau: perspectives on environ- Mavrotas, G., Florios, K., 2013. An improved version of the augmented s-constraint
mental and energy production performance. Int. J. Life Cycle Assess. 21. https://doi. method (AUGMECON2) for finding the exact pareto set in multi-objective integer
org/10.1007/s11367-015-1008-2. programming problems. Appl. Math. Comput. 219. https://doi.org/10.1016/j.amc.
Choong, J.E., Onn, C.C., Yusoff, S., Mohd, N.S., 2019. Life Cycle Assessment of Waste-to- 2013.03.002.
Energy : Energy Recovery from Wood Waste in Malaysia. 28, pp. 2593–2602. MESTECC, 2018. Malaysia Third National Communication and Second Biennial Update Re-
https://doi.org/10.15244/pjoes/93925. port to the UNFCCC. Ministry of Energy, Science, Technology, Environment and Cli-
DOE, Department of Environment, 2010. Environmental Requirements: A Guide for Inves- mate Change, Putrajaya, Malaysia.
tors. Environmental Quality (Control of Pollution from Solid Waste Transfer Station Movahed, Z.P., Kabiri, M., Ranjbar, S., Joda, F., 2020. Multi-objective optimization of life cycle
and Landfill) Requlations 2009. assessment of integrated waste management based on genetic algorithms: a case
Edwards, R., Karnani, S., Fisher, E.M., Johnson, M., Naeher, L., Smith, K.R., Morawska, L., study of Tehran. J. Clean. Prod. 247. https://doi.org/10.1016/j.jclepro.2019.119153.
2014. WHO Indoor Air Quality Guidelines: Household fuel Combustion - Review 2: Ooi, J.K., Woon, K.S., 2021. Simultaneous greenhouse gas reduction and cost optimization
Emissions of Health-Damaging Pollutants from Household Stoves, pp. 1–42. of municipal solid waste management system in Malaysia. Chem. Eng. Trans. 83.
EEA, European Environment Agency, 2013. Municipal Waste Management in Germany. https://doi.org/10.3303/CET2183082.
European Environment Agency. Ooi, J.K., Woon, K.S., Hashim, H., 2021. A multi-objective model to optimize country-scale
Energy Commission, 2018. Malaysia Energy Statistics 2018. Putrajaya. municipal solid waste management with economic and environmental objectives: a
EPU, Economic Planning Unit, 2021. The Twelfth Malaysia Plan, 2021-2025. Prime Minis- case study in Malaysia. J. Clean. Prod. 316, 128366. https://doi.org/10.1016/j.jclepro.
ter's Department, Malaysia. 2021.128366.
Falahi, M., Avami, A., 2020. Optimization of the municipal solid waste management sys- Pérez-Camacho, M.N., Curry, R., Cromie, T., 2018. Life cycle environmental impacts of
tem using a hybrid life cycle assessment–emergy approach in Tehran. J. Mater. Cycles substituting food wastes for traditional anaerobic digestion feedstocks. Waste
Waste Manag. 22. https://doi.org/10.1007/s10163-019-00919-0. Manag. 73. https://doi.org/10.1016/j.wasman.2017.12.023.

13
Z. Lin, J.K. Ooi and K.S. Woon Science of the Total Environment 816 (2022) 151541

Rizwan, M., Saif, Y., Almansoori, A., Elkamel, A., 2020. A multiobjective optimization Woon, K.S., Lo, I.M.C., 2013. Greenhouse gas accounting of the proposed landfill ex-
framework for sustainable design of municipal solid waste processing pathways to tension and advanced incineration facility for municipal solid waste management
energy and materials. Int. J. Energy Res. 44. https://doi.org/10.1002/er.4884. in Hong Kong. Sci. Total Environ. 458–460. https://doi.org/10.1016/j.scitotenv.
Roberts, K.P., Turner, D.A., Coello, J., Stringfellow, A.M., Bello, I.A., Powrie, W., Watson, 2013.04.061.
G.V.R., 2018. SWIMS: a dynamic life cycle-based optimization and decision support Woon, K.S., Lo, I.M.C., Chiu, S.L.H., Yan, D.Y.S., 2016. Environmental assessment of
tool for solid waste management. J. Clean. Prod. 196. https://doi.org/10.1016/j. food waste valorization in producing biogas for various types of energy use
jclepro.2018.05.265. based on LCA approach. Waste Manag. 50. https://doi.org/10.1016/j.wasman.
San Martin, D., Ramos, S., Zufía, J., 2016. Valorization of food waste to produce new raw 2016.02.022.
materials for animal feed. Food Chem. 198. https://doi.org/10.1016/j.foodchem. Woon, K.S., Phuang, Z.X., Lin, Z., Lee, C.T., 2021. A novel food waste management frame-
2015.11.035. work combining optical sorting system and anaerobic digestion: a case study in
SEDA, Sustainable Energy Development Authority, 2021. Feed-in tariff -renewable energy Malaysia. Energy 232, 121094. https://doi.org/10.1016/j.energy.2021.121094.
Malaysia [WWW document]. URL http://www3.seda.gov.my/iframe/ (accessed Worming Up, 2021. Worming up about us [WWW document]. URL http://wormingup.
4.21.21). com/donation/ (accessed 9.20.21).
Shamsul, A., 2016. Landfills is the way forward for Malaysia, says Abdul Rahman Dahlan Yeo, J., Chopra, S.S., Zhang, L., An, A.K., 2019. Life cycle assessment (LCA) of food waste
[WWW document]. URLSun Dly (accessed 4.20.21). https://www.thesundaily.my/ treatment in Hong Kong: on-site fermentation methodology. J. Environ. Manag.
archive/1686458-MSARCH349131. 240. https://doi.org/10.1016/j.jenvman.2019.03.119.

14

You might also like