Waste Management xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Waste Management
journal homepage: www.elsevier.com/locate/wasman

Assessment of municipal solid waste settlement models based on

field-scale data analysis
Christopher A. Bareither ⇑, Seungbok Kwak
Civil & Environmental Engineering, Colorado State University, Fort Collins, CO 80523, USA

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

Article history: An evaluation of municipal solid waste (MSW) settlement model performance and applicability was con-
Received 23 July 2014 ducted based on analysis of two field-scale datasets: (1) Yolo and (2) Deer Track Bioreactor Experiment
Accepted 13 April 2015 (DTBE). Twelve MSW settlement models were considered that included a range of compression behavior
Available online xxxx
(i.e., immediate compression, mechanical creep, and biocompression) and range of total (2–22) and opti-
mized (2–7) model parameters. A multi-layer immediate settlement analysis developed for Yolo provides
Keywords: a framework to estimate initial waste thickness and waste thickness at the end-of-immediate compres-
Field-scale
sion. Model application to the Yolo test cells (conventional and bioreactor landfills) via least squares opti-
Landfill
Modeling
mization yielded high coefficient of determinations for all settlement models (R2 > 0.83). However,
Municipal solid waste empirical models (i.e., power creep, logarithmic, and hyperbolic models) are not recommended for use
Settlement in MSW settlement modeling due to potential non-representative long-term MSW behavior, limited
physical significance of model parameters, and required settlement data for model parameterization.
Settlement models that combine mechanical creep and biocompression into a single mathematical func-
tion constrain time-dependent settlement to a single process with finite magnitude, which limits model
applicability. Overall, all models evaluated that couple multiple compression processes (immediate,
creep, and biocompression) provided accurate representations of both Yolo and DTBE datasets. A model
presented in Gourc et al. (2010) included the lowest number of total and optimized model parameters
and yielded high statistical performance for all model applications (R2 P 0.97).
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction Conceptual models of MSW settlement for scenarios of (i) inhib-

ited and (ii) complete organic waste decomposition are shown in
Settlement of municipal solid waste (MSW) commonly is sepa- Fig. 1. The two settlement curves in Fig. 1 are drawn with respect
rated into three compression processes: (1) immediate compres- to laboratory- and field-scale testing of MSW (e.g., Bareither et al.,
sion, (2) mechanical creep, and (3) biocompression (El-Fadel and 2010; Gourc et al., 2010; Bareither et al., 2012a) and apply to a
Khoury, 2000; Hossain et al., 2003; Marques et al., 2003; Gourc given MSW composition under the same constant vertical stress.
et al., 2010; Bareither et al., 2012a). Immediate compression is The lower-bound settlement curve applies to a scenario where
stress-dependent and occurs rapidly with an increase in vertical MSW decomposition is inhibited (e.g., dry, conventional landfill)
stress. Mechanical creep and biocompression are time-dependent and settlement primarily is attributed to physical compression
processes that occur under constant vertical stress. Mechanical processes of immediate compression and mechanical creep. The
creep involves physical yielding and reorientation of MSW compo- upper-bound settlement curve applies to a scenario where MSW
nents, whereas biocompression is attributed to anaerobic decom- decomposition is optimized (e.g., anaerobic bioreactor landfill)
position of the MSW organic fraction. After biodegradation of the such that organic waste is decomposed to the extent possible. An
organic waste fraction is exhausted, MSW compression will con- increase in total MSW settlement occurs with active biodegrada-
tinue as mechanical creep of the residual waste (i.e., final mechan- tion due to removal of solid mass. Settlement curves between
ical creep). the lower- and upper-bound curves in Fig. 1 exist for scenarios
where a fraction of the MSW organic waste decomposes.
Settlement predictions of MSW require an applicable model and
appropriate model parameters. Model parameterization typically
⇑ Corresponding author.
is conducted with one or a combination of the following
E-mail addresses: [email protected] (C.A. Bareither),
[email protected] (S. Kwak). approaches: (1) field-scale data are analyzed to obtain best fit

http://dx.doi.org/10.1016/j.wasman.2015.04.011
0956-053X/Ó 2015 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Bareither, C.A., Kwak, S. Assessment of municipal solid waste settlement models based on field-scale data analysis. Waste
Management (2015), http://dx.doi.org/10.1016/j.wasman.2015.04.011
2 C.A. Bareither, S. Kwak / Waste Management xxx (2015) xxx–xxx

Nomenclature

a compressibility parameter due to immediate compres- Sult ultimate settlement

sion SSR sum of squared residuals
A air phase compression coefficient SST total sum of squares
b compressibility parameter due to creep and biocom- t elapsed time
pression tB transition from mechanical creep to biocompression
B interactive constant related to biodegradation tc elapsed time for duration of waste placement
c rate of time-dependent compression tF transition from biocompression to final mechanical
ct first-order rate coefficient for coupled mechanical creep creep
and biocompression tM transition from immediate compression to mechanical
C0c immediate compression ratio creep
Cm organic matter methane yield tr reference time for power creep model
C0r recompression ratio ua excess pore gas pressure
C0aB biocompression ratio uo initial pore gas pressure
C0aM mechanical creep ratio Vo initial total waste volume
C0aMF final mechanical creep ratio after completion of biocom- w water content (dry mass basis)
pression a fitting parameter for logarithmic model
D constrained modulus a⁄ rate of increase in parameter a as biodegradation occurs
e void ratio b fitting parameter for logarithmic model
eo initial void ratio eB biocompression strain
H0 initial waste thickness eBIO total amount of strain due to biodegradation
HEOI waste thickness at end-of-immediate compression ee elastic strain
k first-order rate coefficient eI immediate compression strain
Lo methane generation potential eMB ultimate strain due to mechanical creep and biocom-
m reference compressibility pression
mv coefficient of volume change eMC mechanical creep compression strain
M frictional constant ep plastic strain
Mso organic fraction of the biodegradable solid in MSW ev vertical strain
N rate of compression g stress ratio
p0 mean effective stress j recompression or swelling index
p0 o initial mean effective stress k compression index
R2 coefficient of determination qo initial rate of settlement
SB biocompression settlement q⁄p initial specific density of MSW paste
SI immediate settlement qs solid density of MSW
Sj measured settlement q⁄s initial specific density of MSW solids
Ŝj modeled (predicted) settlement rvo initial vertical stress at mid-depth of waste
SMC mechanical creep settlement rvc preconsolidation stress
ST total settlement Drv change in vertical stress at mid-depth of waste
STD time-dependent settlement

model parameters; (2) model parameters are obtained from labo- (Bareither et al., 2012b). Findings from these model simulations
ratory experiments; or (3) model parameters are obtained from were used to develop recommendations for MSW settlement
empirical relationships with waste characteristics (El-Fadel and modeling.
Khoury, 2000; Park et al., 2007; Bareither et al., 2013). Model
parameterization techniques generally are discussed that accom-
2. Mathematical formulations for MSW compression processes
pany a given MSW settlement prediction; however, selecting an
appropriate settlement model requires an understanding of how
A common formulation for total MSW settlement (ST) is
compression phases are represented mathematically as well as a
comparison between available models to support model selection. ST ¼ SI þ STD ð1Þ
A thorough review of MSW settlement modeling is presented in
El-Fadel and Khoury (2000); however, MSW settlement modeling where SI is immediate settlement and STD is time-dependent
has been the focus of significant research since 2000 settlement, which can be further separated into the summation of
(e.g., Marques et al., 2003; Hossain and Gabr, 2005; Machado settlement due to mechanical creep (SMC) and biocompression (SB)
et al., 2009; Babu et al., 2010a; Chen et al., 2010; Gourc et al., (i.e., STD = SMC + SB). These contributions of settlement typically are
2010; Chen et al., 2012) and only limited comparisons and related as strain:
discussions of currently available settlement models have been ST ¼ H0  eI þ HEOI ðeMC þ eB Þ ð2Þ
conducted (Babu et al., 2010b; Babu et al., 2011).
The objective of this study was to evaluate the performance and where H0 is initial waste thickness prior to immediate compression,
applicability of available MSW settlement models via analysis of HEOI is waste thickness at the end-of-immediate compression, eI is
field-scale data. Two field-scale case histories were used to evalu- immediate compression strain, eMC is mechanical creep strain, and
ate model performance in regards to (1) conventional versus biore- eB is biocompression strain.
actor landfill behavior (Yazdani et al., 2006) and (2) composite Most MSW settlement models employ similar mathematical
settlement as the summation of settlement of multiple waste lifts functions for immediate compression, mechanical creep, and

Please cite this article in press as: Bareither, C.A., Kwak, S. Assessment of municipal solid waste settlement models based on field-scale data analysis. Waste
Management (2015), http://dx.doi.org/10.1016/j.wasman.2015.04.011
 C.A. Bareither, S. Kwak / Waste Management xxx (2015) xxx–xxx 3

tM tB tF
Elapsed Time (log scale)

Phase 1 Phase 2 Phase 3 Phase 4

Inhibited Waste
Decomposition
Settlement

Complete Waste
Decomposition

Possible Compression Phases

Phase 1: Immediate compression (stress-dependent)
Phase 2: Mechanical creep (time-dependent)
Phase 3: Biocompression (time-dependent)
Phase 4: Final mechanical creep (time-dependent)
Note: Phases 3 & 4 do not apply to scenario with inhibited waste decomposition

Fig. 1. Conceptual models of municipal solid waste settlement with (i) inhibited and (ii) complete waste decomposition. Temporal definitions: tM = elapsed time for end-of-
immediate compression, tB = elapsed time for onset of biocompression, and tF = elapsed time for completion of waste biodegradation and transition to final mechanical creep.

biocompression. The primary differences between models are the Marques et al., 2003; Vilar and Carvalho, 2004; Bareither et al.,
mathematical functions used and what compression processes 2012c).
they represent. The mathematical functions incorporated into the Recent MSW settlement modeling efforts (e.g., Machado et al.,
various settlement models initially are discussed in regards to 2002; Babu et al., 2010a) have adopted the use of k and j from crit-
the compression process they represent, and subsequently in ical state soil mechanics (Wood, 1990) to separate immediate com-
regards to how the functions are combined to represent MSW set- pression into elastic and plastic components as follows:
tlement due to coupled compression processes.
SI ¼ H0  ðee þ ep Þ
 0    !!
2.1. Immediate compression k po þ Dp0 kj M2 þ g2
¼ H0   ln þ  ln ð5Þ
1þe p0o 1þe M2
Immediate compression occurs due to an increase in vertical
stress and is shown as Phase 1 in Fig. 1. Commonly used parame- where ee is elastic strain, ep is plastic strain, p0 o is initial mean effec-
ters to estimate immediate compression include the following: tive stress, Dp0 is the change in mean effective stress, M is the fric-
(1) constrained modulus (D), expressed as the ratio of change in tional constant (estimated by the slope of the critical state line), and
vertical stress (Drv) to change in vertical strain (ev) g is the stress ratio (Wood, 1990). Application of Eq. (5) was simpli-
(D = DrvDev); (2) coefficient of volume change (mv), which is fied assuming an isotropic stress state, whereby p0 o was approxi-
equivalent to 1/D; (3) compression (Cc0 ) and recompression (Cr0 ) mated as r0 vo and Dp0 was approximated as Dr0 v. In this study,
ratios, expressed as the change in vertical strain to change in log- stress in MSW is expressed as total stress and assumed equal to
arithm of vertical stress (Cc0 or Cr0 = DevDlogrv); and (4) compres- effective stress due to the experiments having free-drainage that
sion (k) and recompression (j) indices, expressed as the change prevented positive pore pressure.
in void ratio (e) to change in natural logarithm of mean effective
stress (p0 ) (k or j = DeDlnp0 ). 2.2. Time-dependent compression
Application of mv and D to estimate immediate compression of
MSW are interchangeable, since mv = 1/D, and can be applied as Time-dependent compression is attributed to mechanical creep
SI ¼ H0  eI ¼ H0  Drv  mv ð3Þ and biodegradation of the MSW organic fraction, which occur
simultaneously during the period of active biodegradation. There
where Drv is the change in vertical stress for which immediate are three general approaches to modeling time-dependent com-
compression occurs. Beaven and Powrie (1996) report that D pression: (1) applying unique mathematical functions to three
increases with an increase in applied stress and waste density. compression phases (mechanical creep, biocompression, and final
Thus, mv would decrease with increasing stress and waste density, mechanical creep); (2) applying two mathematical functions to
which indicates that both parameters (D and mv) are stress- represent mechanical creep and biocompression; or (3) using a sin-
dependent. gle mathematical function to represent all time-dependent
The most broadly adopted approach for computing SI is based compression.
on Cc0 :
  2.2.1. Mechanical creep
rv o þ Drv
SI ¼ H0  C 0c  log ð4Þ Time-dependent compression due to mechanical creep (Phase 2
rv o
in Fig. 1) most commonly is estimated via (1) a mechanical creep
where rvo is initial vertical stress at the mid-depth of an MSW layer compression ratio (CaM0 ), expressed as the change in strain to
subjected to Drv. Use of Cc0 is common since a single Cc0 has been change in logarithm of time (DevDlogt), or (2) a rheological model
shown applicable for a broad range of vertical stress (e.g., that incorporates compression and rate parameters (Gibson and Lo,

Please cite this article in press as: Bareither, C.A., Kwak, S. Assessment of municipal solid waste settlement models based on field-scale data analysis. Waste
Management (2015), http://dx.doi.org/10.1016/j.wasman.2015.04.011
4 C.A. Bareither, S. Kwak / Waste Management xxx (2015) xxx–xxx

1961). Settlement due to mechanical creep (SMC) for a given time for the time period following a transition from biocompression to
(t) since final waste placement can be expressed as final mechanical creep (tF in Fig. 1).
 
t
SMC ðtÞ ¼ HEOI  C 0aM  log ð6Þ
tM 2.2.4. Empirical models
Empirical models have been applied as single mathematical
where tM is elapsed time for the transition from immediate com- functions to represent complete time-dependent settlement
pression to mechanical creep (Sowers, 1973; Bjarngard and (El-Fadel and Khoury, 2000). Common mathematical functions
Edgers, 1990). A single CaM0 can be applicable to all time-dependent include logarithmic, power creep, and hyperbolic functions.
MSW compression when biodegradation is negligible (Ivanova Parameters for these models are site specific and settlement data
et al., 2008; Bareither et al., 2012a; Siddiqui et al., 2013). are required to obtain relevant model parameters.
The rheological model is adopted from Gibson and Lo (1961) Yen and Scanlon (1975) present a logarithmic model to predict
and SMC is expressed as time-dependent settlement assuming the settlement rate
SMC ðtÞ ¼ H0  Drv  ½bð1  ect Þ ð7Þ decreases linearly with logarithm of the median fill age of the
waste. Time-dependent settlement is computed as
where b is the time-dependent compression parameter and c is the      
rate of time-dependent compression. Initial waste thickness (H0) is b tc tc
STD ðtÞ ¼ HEOI  a þ  ln t  1 t ð10Þ
used in Eq. (7) for consistency with the original model (Gibson and lnð10Þ 2 2
Lo, 1961) as well as settlement models that integrate Eq. (7) into a
full-composite model (i.e., Marques et al., 2003; Babu et al., 2010a). where a and b are fitting parameters and tc is elapsed time for the
duration of MSW placement (Ling et al., 1998; El-Fadel and Khoury,
2.2.2. Biocompression 2000; Park et al., 2007).
Settlement due to biocompression (Phase 3 in Fig. 1) can be The power creep law is a relationship for time-dependent defor-
estimated via analogous approaches to mechanical creep. A similar mation under constant stress represented by transient creep
expression to Eq. (5) can be written for SB as behavior (Edil et al., 1990; Ling et al., 1998; Park et al., 2007).
  Total time-dependent MSW settlement is computed as
t
SB ðtÞ ¼ HEOI  C 0aB  log ð8Þ  N
tB t
STD ðtÞ ¼ HEOI  Drv  m ð11Þ
tr
where CaB0 is the biocompression ratio and tB is elapsed time for the
transition from mechanical creep to biocompression. The biocom-
where m is reference compressibility, N is rate of compression, and
pression ratio (CaB0 ) is identical to CaM0 (i.e., DevDlogt), but com- tr is reference time, which typically is taken as 1 d to make time
puted and applied within the time period that biodegradation is
dimensionless.
active. The elapsed time tB has been linked to a break in slope of Tan et al. (1991) present a hyperbolic function that has been
time-dependent settlement plotted on a semi-logarithmic plot
applied to settlement scenarios where geomaterial properties are
(e.g., Fig. 1), the onset of methane generation (e.g., Bareither et al., difficult or impossible to determine (Tan et al., 1991; Ling et al.,
2010; Gourc et al. 2010), and a decrease in leachate volatile fatty
1998; El-Fadel and Khoury, 2000; Park et al., 2007). Time-depen-
acid concentration or chemical oxygen demand (e.g., Olivier and dent MSW settlement is computed with the following hyperbolic
Gourc, 2007; Ivanova et al., 2008; Bareither et al., 2012a, 2013).
equation:
A biodegradation-induced compression model based on first-
order kinetics is as follows: t
STD ðtÞ ¼ ð12Þ
kðtt B Þ 1=qo þ t=Sult
SB ðtÞ ¼ HEOI  eBIO  ð1  e Þ ð9Þ

where eBIO is total strain due to biodegradation and k is the first- where qo is initial rate of settlement and Sult is ultimate settlement
order decay coefficient (Park and Lee, 1997; El-Fadel and Khoury, as time approaches infinity.
2000; Park and Lee, 2002; Gourc et al., 2010). Applying Eq. (9) to
time-dependent compression can be executed via (1) coupling
mechanical creep and biocompression such that mechanical creep 3. Composite settlement models
is represented as a continuous process and eBIO only applies to
biodegradation-induced compression (e.g., Gourc et al., 2010; Composite MSW settlement models combine different mathe-
Siddiqui et al., 2013) or (2) decoupling time-dependent compres- matical formulations from those described in the previous section
sion such that during active biodegradation Eq. (9) is used to predict to represent multiple compression phases (Fig. 1). A summary of
both mechanical creep and biocompression (Park and Lee, 1997; relevant MSW parameter ranges for all composite settlement mod-
El-Fadel et al., 1999). The former method is preferable for long-term els is in Table 1.
predictions and is adopted herein; the latter method is instructive
for determining time at which biocompression ends to estimate tF 3.1. Gibson and Lo (1961)
in Fig. 1 (Bareither et al., 2013).
Gibson and Lo (1961) present the following rheological model
2.2.3. Final mechanical creep that couples Eqs. (3) and (7) to represent immediate and time-de-
As shown in Fig. 1, there is a transition in the rate of waste com- pendent compression:
pression that follows the end of biocompression. Although this
transition has been documented in laboratory experiments ST ðtÞ ¼ H0  Drv  ½a þ b  ð1  ect Þ ð13Þ
(Bareither et al., 2013), there is a lack of field-scale data document-
ing this change due to long time durations necessary to capture the where a is analogous to mv in Eq. (3). One caveat with the Gibson
transition. The main approach to predict final mechanical creep is and Lo model is the use of H0 for computing both immediate and
to employ a final mechanical creep ratio (CaMF0 ) similar to Eq. (6) time-dependent compression strain.

Please cite this article in press as: Bareither, C.A., Kwak, S. Assessment of municipal solid waste settlement models based on field-scale data analysis. Waste
Management (2015), http://dx.doi.org/10.1016/j.wasman.2015.04.011
 C.A. Bareither, S. Kwak / Waste Management xxx (2015) xxx–xxx 5

Table 1
Summary of ranges for settlement model parameters.

Model Parameter Unit Range (Average)

Gibson and Lo (1961) a 1/kPa 5.32  107–3.80  104 (1.90  104)
b 1/kPa 0.0001–0.0095 (0.0048)
k/b 1/yr 0.0336–15.3 (7.67)
Marques et al. (2003) Cc0 – 0.073–0.28 (0.18)
b 1/kPa 0.0001–0.0095 (0.0048)
c 1/yr 0.0336–15.3 (7.67)
eBIO – 0.0016–0.44 (0.22)
k 1/yr 0.245–17.9 (9.07)
Sowers (1973), Bjarngard and Edgers (1990), Cc0 – 0.073–0.28 (0.18)
Hossain and Gabr (2005) CaM0 – 0.017–0.1 (0.059)
CaB0 – 0.057 – 0.36 (0.21)
CaMF0 – 0.008–0.22 (0.11)
Machado et al. (2009) CaM0 – 0.017–0.10 (0.059)
a⁄ – 1.40–102 (51.7)
k 1/yr 0.245–17.9 (9.07)
Babu et al. (2010) k – 0.091–0.18 (0.13)
b 1/kPa 0.0001–0.0095 (0.0048)
c 1/yr 0.0336–15.3 (7.67)
eBIO – 0.0016–0.44 (0.22)
k 1/yr 0.245–17.9 (9.07)
Chen et al. (2010) Cc0 – 0.073–0.28 (0.18)
eMB – 0.098–0.24 (0.17)
ct 1/yr 0.1–0.365 (0.233)
Gourc et al. (2010) CaM0 – 0.017–0.1 (0.059)
k 1/yr 0.245–17.9 (9.07)
eBIO – 0.0016–0.44 (0.22)
Chen et al. (2012) k 1/yr 0.245–17.9 (9.07)
ka m/yr 3.65–54.8 (29.2)
m3 m2/N 106 to 107 (5.5  107)

Compiled from Fredlund and Hasan (1979), Edil et al. (1990), Park and Lee (2002), Marques et al. (2003), McDougall and Pyrah (2004),
Park et al. (2007), Sharma and De (2007), Babu et al. (2010), Chen et al. (2010), Gourc et al. (2010), Bareither et al. (2012a), and Bareither
et al. (2013).

3.2. Sowers (1973), Bjarngard and Edgers (1990), Hossain and Gabr and allow for a lag time between waste placement and the onset
(2005) of biocompression.

A complete version of the settlement model initially presented

in Sowers (1973) and Bjarngard and Edgers (1990) that accounts 3.4. Machado et al. (2009)
for all possible compression phases (Fig. 1) is
  Machado et al. (2009) propose an MSW settlement model that
r v o þ Dr v incorporates mass loss due to biodegradation that is related to
ST ðtÞ ¼ H0  C 0c log
rv o methane generation as follows:
      
t t t ( )
0
þ HEOI C aM log þ C 0aB log þ C 0aMF log ð14Þ 
tM tB tF qs Lo ð1 þ wÞk kt Lo ð1 þ wÞð1  ekt Þ
deB ¼ e 1 þ a dt ð16Þ
where Eqs. (4), (6) and (8) are combined with an additional term for
qp C m ð1 þ eo Þ
 Cm
final mechanical creep (Hossain and Gabr, 2005). Application of this
model requires a transition from CaB0 to CaMF0 since biocompression where deB is change in biocompression strain, q⁄s is initial density of
will not occur indefinitely, and CaMF0 can be assumed equal to CaM0 MSW solid material, q⁄p is initial density of MSW paste (non-plastic
(Bareither et al., 2013). Temporal terms for time-dependent com- waste), Lo is methane generation potential, w is water content, eo is
pression on the right hand side of Eq. (14) are specified for the dura- initial void ratio, a⁄ is a biodegradation rate parameter, and Cm is
tion that each process is active (e.g., tB 6 t 6 tF for biocompression). organic matter methane potential. Integrating Eq. (16) with respect
to time and combining with Eq. (6) yields
3.3. Marques et al. (2003)  
t
STD ðtÞ ¼ HEOI C 0aM log
Marques et al. (2003) present the following composite MSW tM
 
settlement model: qs Lo ð1 þ wÞ a Lo ð1 þ wÞ
þ HEOI  1þ ð1  ekt Þ
    qp C m ð1 þ eo Þ Cm
rv o þ Drv 
ST ðtÞ ¼ H0 C 0c log þ bDrð1  ect Þ þ eBIO ð1  ekðttB Þ Þ a Lo ð1 þ wÞ
rv o  ð1  e2kt Þ ð17Þ
2C m
ð15Þ

which combines immediate compression, mechanical creep, and which is used to compute settlement due to mechanical creep and
biocompression strain that are all related to H0. A modification to biodegradation with a function to account for methane generation.
the Marques et al. (2003) model was made to incorporate (t–tB) A detailed derivation of Eq. (17) is in Kwak (2014).

Please cite this article in press as: Bareither, C.A., Kwak, S. Assessment of municipal solid waste settlement models based on field-scale data analysis. Waste
Management (2015), http://dx.doi.org/10.1016/j.wasman.2015.04.011
6 C.A. Bareither, S. Kwak / Waste Management xxx (2015) xxx–xxx

3.5. Babu et al. (2010a) 16

Control Cell
Babu et al. (2010a) present the complete settlement model Enhanced Cell
15
shown in Eq. (18).
" !
(a)
 0   

Waste Thickness (m)

k po þ Dp0 kj M 2 þ g2 14
ST ðtÞ ¼ H0 ln þ ln
1þe p0o 1þe M2

þbDrð1  ect Þ þ eBIO ð1  ekðttB Þ Þ ð18Þ 13

The first two terms on the right-hand-side of the Eq. (18) corre- 12
spond to immediate compression and the last two terms corre-
spond to time-dependent processes of mechanical creep and
11
biocompression. A detailed derivation of Eq. (18) is in Kwak (2014). t = 235 d t = 3980 d

3.6. Chen et al. (2010) 10

Chen et al. (2010) present the following settlement model that 0.0
combines Eq. (4) for immediate compression and Eq. (9) for all (b)
time-dependent compression:
0.5
 
rv o þ Drv

Waste Settlement (m)

ST ðtÞ ¼ H0  C 0c log þ HEOI eMB ð1  ect t Þ ð19Þ
rv o 1.0

where eMB is ultimate strain due to mechanical creep and biocom-

pression and ct is a first-order rate coefficient for coupled mechan- 1.5
ical creep and biocompression. Time-dependent compression is
simplified in this model to a single first-order rate process. 2.0
t = 235 d
3.7. Gourc et al. (2010)
2.5
Control Cell t = 3980 d
The time-dependent waste settlement model presented in Enhanced Cell
Gourc et al. (2010) combines Eqs. (6) and (9) to predict settlement 3.0
0.1 1 10 100
based on CaM0 and k as shown in Eq. (20).
    Elapsed Time (yr)
t
STD ðtÞ ¼ HEOI C 0aM  log þ eBIO ð1  ekðttB Þ Þ ð20Þ Fig. 2. Temporal trends of waste thickness and settlement for the Yolo Enhanced
tM
and Control cells.

3.8. Chen et al. (2012)

mechanical creep (ekn t ) and biodegradation (ekt ). Total settlement
Chen et al. (2012) present a settlement model based on unsat- can be computed using Eq. (22), assuming 1-D compression and
urated consolidation (Fredlund and Hasan, 1979). Volumetric summing settlement along a vertical profile consisting of multiple
strain is related to changes in excess pore gas pressure (ua). The waste layers:
("
spatial and temporal change in ua is presented in Liu et al. (2006) X
H0 1 
X 
4uo 4B
for MSW as follows: ST ðtÞ ¼ H0 m3 ekn t þ ðekt  ekn t Þ
z¼1 n¼1
ð2n þ 1Þ p ð2n þ 1Þ pðkn  kÞ
X1    
4uo 4B M so
ua ðz;tÞ ¼ ekn t þ ðekt  ekn t Þ sinðpzÞ sinðpzÞ  u0 þ ð1  ekt Þ ð22Þ
n¼0
ð2n þ 1Þp ð2n þ 1Þpðkn  kÞ qs V o
ð21Þ where m3 is the coefficient of gas volume change, Mso is mass of
2 biodegradable waste, and qs is density of the MSW solids. A detailed
where kn ¼ Ap , A is the air phase compression coefficient, B is a
derivation of Eq. (22) is in Kwak (2014).
p
constant related to organic waste biodegradation, p ¼ ð2nþ1Þ
2H0
, and
uo is initial pore gas pressure. Eq. (21) is used to express the change 4. Field-scale landfill data
in discharged gas volume as a function of compression due to
4.1. Yolo County Pilot Project
Table 2
Average waste properties of the Control and Enhanced cells, Yolo Co. Pilot Project
(Mehta et al., 2002; Yazdani et al., 2006). The Yolo County Pilot Project (Yolo) included a Control and
Enhanced cell constructed to similar dimensions (i.e., surface area
Property Control cell Enhanced cell
and waste thickness) and filled with waste of similar composition
Surface area (m2) 930 930
and water content (Mehta et al., 2002; Yazdani et al., 2006).
Dry weight water content (%) 20.3 20.3
Total initial mass (Mg) 7932 7772
Leachate recirculation was conducted in the Enhanced cell to accel-
Green (yard) waste (%) 16.7 15.6 erate waste decomposition, but was not conducted in the Control
Approximate initial total unit weight (kN/m3) 7.0 7.0 cell. Average waste properties of the Control and Enhanced cells
Average waste thickness on Day 235 (m)a 15.0 13.6 are in Table 2. Waste was compacted in lifts to replicate full-scale
a
Day 235 = first elevation survey; thickness was averaged from more than 20 operations, and green waste (i.e., yard waste) was used as interim
surface elevation measurements on each cell. cover between successive waste lifts to promote liquid

Please cite this article in press as: Bareither, C.A., Kwak, S. Assessment of municipal solid waste settlement models based on field-scale data analysis. Waste
Management (2015), http://dx.doi.org/10.1016/j.wasman.2015.04.011
 C.A. Bareither, S. Kwak / Waste Management xxx (2015) xxx–xxx 7

0.0

Geomembrane
(a) Cover 0.5

Total Waste Settlement (m)

Leachate
Recirculation 1.0
Gravel 0.9 m

Settlement Geonet 1.5

Plate 4 Waste
Layer 4 (b)
(WL4)
Plate 1
Settlement
2.0 Plate 2
Plate 3 Plate 3
Plate 4
2.5
Waste Layer 3 Settlement
(WL3) Plate 2
0.0

Individual Waste Layer Settlement (m)

8.2 m
0.2

Waste Layer 2
(WL2)
0.4
Settlement
Plate 1

0.6
(c)
Leachate Steel
Collection Waste Layer 1 Cylinder Waste Layer 1
(WL1) 0.8 Waste Layer 2
Gravel
Waste Layer 3
Waste Layer 4
Leachate 1.0
0.5 m 0.0001 0.001 0.01 0.1 1 10
Outflow
2.4 m Elapsed Time (yr)

Fig. 3. (a) Schematic of the Deer Track Bioreactor Experiment and temporal relationships of (b) total waste settlement measured for each settlement plate and (c) settlement
of individual waste layers.

distribution. A geomembrane cover was placed over the cells after Daily average waste settlement measured for each settlement
waste filling to enhance gas collection. plate is shown in Fig. 3b. Waste settlement is representative of
Average waste thickness and average waste settlement for the the total waste thickness below a given settlement plate. Total
Control and Enhanced cells are shown in Fig. 2. Surface elevations waste settlement increased with increasing waste thickness for
of the geomembrane covers were measured periodically to moni- Plates 1, 2, and 3; however, total settlement measured for Plates
tor settlement. An average waste thickness and corresponding set- 3 and 4 was comparable and suggests there may have been differ-
tlement were computed for each elevation survey, which included ential settlement within the waste column.
more than 20 survey points on each cell. The first survey was con- Daily average waste settlement for individual waste layers is
ducted seven months following final waste placement (t = 235 d). shown in Fig. 3c. Individual waste layer settlement was computed
The elapsed time between waste placement and the first eleva- as the difference between total settlement measured for adjacent
tion survey (t = 235 d) suggests that some component of waste set- plates. Similar settlement behavior was observed for waste layers
tlement due to immediate compression and mechanical creep was (WL) 1, 2, and 3, which exhibit distinct phases of immediate com-
not measured (Fig. 2). The measured data are sufficient for evalu- pression, mechanical creep, and biocompression (Fig. 3c). Although
ating the efficacy of time-dependent settlement models; however, immediate compression and mechanical creep also were observed
an estimate of H0 and HEOI are needed for implementation of all for WL4, a pronounced transition from mechanical creep to bio-
models. Thus, a multi-layered immediate settlement (SI) analysis compression was not observed.
was conducted to (1) evaluate the practicality of estimating SI via A summary of waste characteristics for each waste layer in the
Eq. (4) and (2) justify an estimate of HEOI for use in subsequent DTBE and the composite waste column is included in Table 3.
time-dependent settlement analyses. Settlement models were applied to individual waste layers after
initial filling (i.e., waste thicknesses and properties are representa-
tive of the waste state prior to placement of subsequent waste and
4.2. Deer Track Bioreactor Experiment the top gravel layer) to assess variation in model parameters that
may be encountered in a given waste profile. Settlement predic-
The Deer Track Bioreactor Experiment (DTBE) was a field-scale tions were then completed on individual waste layers that consti-
experiment conducted in a drainage lysimeter (8.2-m height, 2.4- tuted the composite waste profile at the end-of-filling condition
m diameter) to assess the physical, chemical, and biological (Table 3) to assess the efficacy of summing individual waste layer
response of MSW with leachate addition (Bareither et al., 2012b). settlement to predict composite settlement measured at the top of
A schematic of the DTBE is shown in Fig. 3a. Settlement was mon- the waste column. For the composite waste profile analysis, waste
itored via four settlement plates placed at different depths in the settlement occurred due to placement of the top gravel layer
waste column. Settlement was measured during waste placement (Fig. 3a) and self-weight of the waste. Moisture content and solid
and following waste placement to capture complete compression waste chemical characteristics in Table 3 were measured on sam-
behavior (Bareither et al., 2012b). ples collected during filling (Bareither et al., 2012b).

Please cite this article in press as: Bareither, C.A., Kwak, S. Assessment of municipal solid waste settlement models based on field-scale data analysis. Waste
Management (2015), http://dx.doi.org/10.1016/j.wasman.2015.04.011
8 C.A. Bareither, S. Kwak / Waste Management xxx (2015) xxx–xxx

Table 3
Waste properties of the Deer Track Bioreactor Experiment for waste layers after initial filling (prior to subsequent waste placement) and the composite waste at the end-of-filling
(prior to placement of final gravel layer) (Bareither et al., 2012b).

Waste state Waste layer H (m) w (%) ct (kN/m3) ro (kPa) Dr (kPa) Cellulose (%) Hemicellulose (%) Lignin (%) Volatile solids (%)
Initial filling 1 1.80 33.0 9.16 8.26 48.7 17.0 4.6 30.8 49.1
2 2.23 26.4 6.29 7.03 33.4 14.7 4.0 24.5 42.6
3 2.00 31.7 6.62 6.63 19.7 18.3 6.2 31.4 57.9
4 1.76 40.8 7.42 6.5 6.5 28.4 6.7 28.3 62.0
End of fillinga Composite 6.85 33.0 8.32 28.5 19.5 19.6 5.4 28.8 52.9

Notes: H = waste thickness; w = dry-weight water content; ct = total unit weight, rvo = initial vertical stress, computed at mid-depth of a given waste layer; Drv = increase in
vertical stress.
a
Composite waste profile constitutes all four waste layers and is representative of the condition at the end-of-filling of waste and prior to placement of leachate
recirculation gravel (Fig. 3a).

Table 4
Summary of waste settlement models used in this study.

Model name Eq. Phases Total Optimized Optimized References

No. includeda parameters parameters variables
Full composite Gibson and (13) IM, MC, BIO 5 3 a, b, and k/b Gibson and Lo (1961)
models Lo
Sowers (14) IM, MC, BIO 11 4 Cc0 , CaM0 , CaB0 , Sowers (1973), Bjarngard and Edgers (1990), Hossain
and CaMF0 and Gabr (2005)
Marques (15) IM, MC, BIO 10 5 Cc0 , b, c, eBIO and Sowers (1973), Gibson and Lo (1961), Park and Lee
k (1997), Marques et al. (2003)
Babu (18) IM, MC, BIO 13 7 k, j, g, b, c, eBIO Gibson and Lo (1961), Park and Lee (1997), Babu et al.
and k (2010)
Chen-2010 (19) IM, MC, BIO 7 3 Cc0 , eMB and ct Sowers (1973), Chen et al. (2010)
Chen-2012 (22) IM, MC, BIO 22 4 uo, k, ka, and m3 Park and Lee (1997), Chen et al. (2012)
Time-dependent Park and (9) BIO 3 2 eBIO and k Park and Lee (1997)
modelsb Lee
Logarithmic (10) MC, BIO 4 2 a and b Yen and Scanlon (1975)
Power (11) MC, BIO 5 2 M and N Edil et al. (1990)
Creep
Hyperbolic (12) MC, BIO 2 2 qo and Sult Tan et al. (1991)
Machado (17) MC, BIO 11 3 CaM0 , a⁄, and k Sowers (1973), Machado et al. (2009)
Gourc (20) MC, BIO 6 3 CaM0 , eBIO and k Park and Lee (1997), Gourc et al. (2010)
a
IM = immediate compression, MC = mechanical creep, and BIO = biocompression.
b
Immediate compression ratio (Cc0 ) used to determine SI for all time-dependent models.

4.3. Settlement model implementation and analysis was based on observation of anticipated settlement beyond the
range of measured settlement. Extrapolating predictions for long-
A summary of the settlement models used in the analyses con- term waste settlement is relevant to evaluate progression of
ducted herein is included in Table 4. All settlement models listed in organic waste stabilization, duration of post-closure, and settle-
Table 4 were used in the Yolo settlement evaluation to encompass ment during site reclamation (e.g., Sharma and De, 2007;
the breadth in MSW settlement modeling. Settlement models used Abichou et al., 2013).
in the DTBE analysis were selected based on the Yolo analysis The coefficient of determination was computed as
(described subsequently).
SSR
Parameters included in the settlement models were differenti- R2 ¼ 1  ð23Þ
ated into four categories: (1) measured, (2) fixed, (3) computed, SST
and (4) optimized. The total number of model parameters and where SSR is sum of squared residuals and SST is total sum of
number of optimized parameters for each settlement model is in squares (Berthouex and Brown, 2002). The total sum of squares
Table 4. Measured parameters were based on waste properties was computed as
and characteristics of the experiments. Fixed parameters were
identified as parameters that could be constrained based on previ- X
N
2
SST ¼ ðSj  Sj Þ ð24Þ
ous experience or fixed to simplify model application. Computed j¼1
parameters were calculated based on measured and fixed parame-
ters (e.g., e computed via cd and Gs). Optimized parameters were where N is the number of observations in the data set, Sj is mea-
determined via least squares optimization in the model implemen- sured settlement, and Sj is the arithmetic mean of all Sj. The sum
tation procedure. Identification of the four model parameter cate- of squared residuals was computed as
gories for each settlement model and accompanying final model
parameters are included in Tables S1-S4. X
N
2
SSR ¼ ðSj  b
Sj Þ ð25Þ
Settlement model performance was evaluated via statistical j¼1
assessment, practicality of model application, and extrapolation
to represent future settlement. Statistical performance was evalu- where bS j is modeled settlement that corresponds to measurement
ated by computing the coefficient of determination (R2). The prac- Sj.
ticality assessment included a comparison of R2 versus the number Model implementation was conducted in either Excel or
of total and optimized parameters. The extrapolation assessment MATLAB. Excel was used to the extent possible with the Solver

Please cite this article in press as: Bareither, C.A., Kwak, S. Assessment of municipal solid waste settlement models based on field-scale data analysis. Waste
Management (2015), http://dx.doi.org/10.1016/j.wasman.2015.04.011
 C.A. Bareither, S. Kwak / Waste Management xxx (2015) xxx–xxx 9

Control Cell
20 Enhanced Cell

WL9
Waste Thickness (m)

15 WL8 WL8
WL9 WL9 WL9
WL7 WL7 WL8 WL8 WL9
WL8 WL8 WL8
WL8
WL8
WL6 WL6 WL7 WL7
WL7 WL7 WL7
10 WL7 WL7 WL8
WL5 WL6 WL6
WL5 WL6 WL6 WL6
WL6 WL6 WL7
WL5 WL5 WL5
WL4 WL4 WL5 WL5 WL6
WL5 WL5
WL4 WL4 WL4 WL4 WL5
WL4 WL4 WL4
5 WL3 WL3
WL3 WL3 WL3 WL4
WL3 WL3 WL3 WL3
WL2 WL2 WL3
WL2 WL2 WL2 WL2 WL2 WL2
WL2 WL2
WL1 WL1 WL1 WL1 WL1 WL1 WL1 WL1 WL1 WL1

Ho HEOI H235 H500 H3980

Maximum possible initial Waste thickness of end-of- Waste thickness on Waste thickness at onset of Waste thickness on
thickness, H0 = ho· n, immediate compression at Day 235 = initial survey biocompression, Day 500 Day 3980 = last survey
where ho = initial thickness for Day 15
each waste layer, n = the
number of layers
Elapsed Time (d)

Fig. 4. Total waste thicknesses and individual waste layer thickness for the Enhanced and Control cells at Yolo. Dashed lines indicate individual waste layer thicknesses; these
thicknesses for H235, H500, and H3980 are averages from the model simulations. The waste thickness corresponding to H0 and HEOI are based on assumptions outlined in the
multi-layer settlement analysis. Waste thicknesses corresponding to H235 and H3980 are based on final cover elevation surveys conducted at Yolo.

function used to determine optimized parameters via least squares The number of 2-m-thick waste layers for the settlement anal-
analysis. MATLAB was used to implement the model in Chen et al. ysis in the Enhanced cell was eight, whereas nine layers were used
(2012), which required multiple imbedded equations and numer- for the Control cell (Fig. 4). The additional waste layer for the
ous iterations to obtain the optimized parameters. All model Control cell was due to a larger H235, which is consistent with a
parameter optimizations were conducted via minimizing the SSR. greater mass of waste compared to the Enhanced cell (Table 2).
The number of waste layers for each cell was constrained to be a
5. Results whole number, and the anticipated magnitude of immediate com-
pression strain was assumed in the range of 0.10–0.30 based on
5.1. Yolo immediate settlement analysis previous studies (e.g., Kavazanjian et al., 1999; Hossain et al.,
2003; Olivier and Gourc, 2007; Bareither et al., 2012c).
A schematic of the multi-layered settlement analysis conducted A target SI was estimated for both the Control and Enhanced
for Yolo is shown in Fig. 4. This analysis included the following cells based on extrapolating mechanical creep compression back
assumptions: (i) both cells were filled incrementally with 2-m- to an anticipated time for completion of immediate compression
thick waste layers; (ii) immediate compression only occurred due (tM in Fig. 1). Measured settlement in the Yolo test cells between
to induced stress from waste placed above a previously deposited Day 235 and Day 523 (i.e., second cover elevation survey, Fig. 2)
waste layer (i.e., self-weight settlement is assumed negligible); and can be attributed predominantly to mechanical creep due to
(iii) the same compression parameter (e.g., Cc0 ) was applicable for methane generation and the onset of biocompression occurring
all waste layers. The first assumption is relevant for sequential fill- approximately on Day 500 (Yazdani et al., 2006; Bareither et al.,
ing of MSW that is typical of full-scale landfills. The second 2010). Thus, a mechanical creep ratio (CaM0 ) was computed for
assumption reduces settlement of the top waste layer to zero. the Control and Enhanced cells between Days 235 and 523 and
The third assumption is based on observations that Cc0 approxi- used to back-calculate anticipated settlement due to mechanical
mately is constant over a broad stress range (e.g., Bareither et al., creep between tM and Day 235. For this analysis, tM was assumed
2012c). equal to 15 d based on observations of field-scale immediate com-
Implementation of the multi-layered SI analysis required waste pression (Bareither et al., 2012c). Incorporating these assumptions
thickness constraints to guide the analysis and aid in defining the and calculations leads to an estimate of HEOI. The target SI is the dif-
number of waste layers for the Control and Enhanced cells. Waste ference between H0 and HEOI.
thickness constraints included the measured thickness on Day 235 A summary of three SI analyses and corresponding compression
(H235) (Fig. 4) and the maximum potential waste thickness prior to parameters, SI, and waste thickness is in Table 5. The analyses per-
any settlement (H0). The waste thickness on Day 235 corresponds tain to the following considerations: Analysis 1 – optimized Cc0 to
to the first elevation survey following waste placement; thus, the achieve the target SI assuming only virgin compression; Analysis
waste thickness on any previous day must be greater. The maxi- 2 – computed SI with Cc0 based on an empirical relationship with
mum potential waste thickness (H0) was computed as the product the waste compressibility index (WCI); and Analysis 3 – assume
of the assumed individual waste layer thickness (h0) and number of SI is attributed to recompression and virgin compression and opti-
waste layers (n). This thickness, H0, is theoretical as settlement will mize a precompression stress (rvc) to achieve the target SI. In
occur during filling and a waste thickness equal to H0 will not actu- Analysis 3, Cc0 was estimated from the WCI and the recompression
ally be achieved. The waste thicknesses H0 and H235 impose con- ratio (Cr0 ) was assumed equal to 1/10Cc0 based on data presented in
straints on the multi-layered SI analysis such that H0 > HEOI > H235. Bareither et al. (2012c). The WCI is a function of w, dry unit weight,

Please cite this article in press as: Bareither, C.A., Kwak, S. Assessment of municipal solid waste settlement models based on field-scale data analysis. Waste
Management (2015), http://dx.doi.org/10.1016/j.wasman.2015.04.011
10 C.A. Bareither, S. Kwak / Waste Management xxx (2015) xxx–xxx

Table 5
Summary of the multi-layer immediate compression analyses conducted for the Control and Enhanced cells at Yolo.

Analysis Description Cell Cc0 Cr0 rvc SI eI (m/ HEOI H235

(kPa) (m) m) (m) (m)
1 Only virgin compression, target SI, optimized Cc’ C 0.196 – – 2.96 0.164 15.04 15.0
E 0.154 – – 1.95 0.122 14.05 13.6
2 Only virgin compression, Cc0 = f (WCI), estimate SI C 0.232 – – 3.50 0.194 14.50 15.0
E 0.232 – — 2.93 0.183 13.07 13.6
3 Recompression & virgin compression, Cc0 = f (WCI), Cr0 = Cc0 /10, target SI, C 0.232 0.0232 10.2 2.96 0.164 15.04 15.0
optimize rvc
E 0.232 0.0232 15.1 1.95 0.122 14.05 13.6
0
Note: Cc = immediate compression ratio; Cr0
= recompression ratio, rvc = pre-compression stress; SI = settlement due to immediate compression; eI = immediate compression
strain; WCI = waste compressibility index; C = Control cell; E = Enhanced cell; – = not applicable.

and the percent contribution of organic waste (Bareither et al., provides a means to evaluate long-term model performance. All
2012c). The contribution of organic waste was assumed equal to settlement models, except for the logarithmic and power creep
the U.S. national average (Staley and Barlaz, 2009), and factoring models, provide reasonable extrapolations of MSW settlement to
in estimates of w and dry unit weight yielded Cc0 = 0.232. t = 100 yr (Fig. 5). Extrapolation of these empirical models does
Analysis 1 is a straight-forward parameter optimization proce- not agree with anticipated long-term MSW settlement (Fig. 1)
dure, whereas Analyses 2 and 3 build on current knowledge of and will lead to an overestimation of future settlement
MSW immediate compression. The Cc0 determined for Analysis 1 (Fig. 5c and d). Thus, these two empirical models are considered
was 0.196 for the Control cell and 0.154 for the Enhanced cell non-representative of actual compression behavior in MSW
(Table 5), which agree with past studies (Table 1). Applying landfills.
Cc0 = 0.232 in Analysis 2 leads to an overestimation of SI in both Extrapolation of all other models in Fig. 5 will lead to either a
cases as HEOI < H235, which is not possible. Applying Cc0 = 0.232 in finite magnitude of settlement or a constant rate of final mechan-
Analysis 3 combined with the optimization of rvc leads to ical creep on a semi-logarithmic plot. The constant rate of final
rvc = 10 kPa for the Control cell and 15 kPa for the Enhanced cell mechanical creep is apparent in the Sowers, Machado, and Gourc
(Table 5). Bareither et al. (2012c) reported an apparent rvc in lab- models (Fig. 5b, h, and k), and is attributed to the use of CaM0 to rep-
oratory-prepared MSW specimens due to compaction. Thus, a rvc resent mechanical creep. In contrast, the Gibson and Lo, Park and
in a full-scale landfill due to compaction efforts is reasonable and Lee, Marques, Babu, Chen-2010, and Chen-2012 all approach a
may need to be accounted for when calculating SI. finite magnitude of settlement near t = 100 yr (Fig. 5). These mod-
The multi-layered SI analysis outlined herein is applicable to els incorporate constraints on the magnitude of mechanical creep
estimate SI in solid waste landfills. Assumptions included in this (Eq. (7)) and/or biocompression (Eq. (9)). Long-term settlement
analysis were based on relevant full-scale landfill operations or behavior in MSW landfills has not adequately been measured
findings from recent studies on MSW compression behavior. due to long monitoring requirements. However, settlement predic-
Thus, the analysis reflects a practical SI estimation procedure. tions with models that constrain settlement to a finite value may
Similar analyses can also be completed for predicting SI based on lead to underpredictions, whereas models that include a constant
Eqs. (3) and (5). However, the focus has been on Eq. (4) and the rate of final mechanical creep may lead to overpredictions. Use of
use of Cc0 as this method is more prevalent and believed to be more multiple models may be useful to identify a range of anticipated
applicable to landfill practitioners. long-term MSW settlement (e.g., Bareither et al., 2013).
The hyperbolic model also includes a finite value of potential
5.2. Yolo time-dependent settlement analysis long-term settlement (i.e., Sult). Although this model appears to
yield practical long-term extrapolations (Fig. 5e), the hyperbolic
Temporal settlement data from the Control and Enhanced cells model is difficult to implement in the absence of measured data
at Yolo and settlement model results are shown in Fig. 5. A sum- that are required for parameterization. These limitations suggest
mary of model parameters for the Yolo simulations and accompa- that the hyperbolic model may not be applicable for predicting
nying R2s are in Table S1 for full composite models and Table S2 for MSW settlement.
time-dependent models. Settlement models were implemented via Relationships of R2 versus the total number of model parame-
a multi-layered analysis (Fig. 4), whereby settlement of individual ters and number of optimized parameters for the Control and
waste layers was summed to yield total settlement. In general, all Enhanced cells are shown in Fig. 6. The majority of the settlement
settlement models evaluated in this study capture compression models yielded R2 > 0.95, indicating that these models were cap-
behavior for the Control and Enhanced cells within the range mea- able of explaining greater than 95% of the variation in the time-de-
sured. The R2s range from 0.832 to 0.971 for the Control cell and pendent settlement data. This level of statistical significance was
0.841–0.990 for the Enhanced cell. Larger R2s for the Enhanced cell anticipated due to parameter optimization. Data in Fig. 6 suggest
are attributed to a broader range in measured settlement data, that the majority of the settlement models are capable of simulat-
which leads to larger SST and higher R2. ing unique compression phases in both conventional (Control cell)
All time-dependent simulations were implemented with the and bioreactor (Enhanced cell) landfills.
constraint that SI equals the target SI discussed previously All settlement models with the exception of the three empirical
(Table 5). Assumed temporal constraints of tM = 15 d and tB = 500 models and the Park and Lee model were used in the subsequent
d were included in all simulations, where applicable. These con- DTBE modeling exercise. Although these omitted models can lead
straints were used to reduce the number of optimized model to statistically significant settlement simulations (Tables S1 and
parameters, improve consistency between time-dependent settle- S2), the other available models are believed more applicable due
ment simulations, and enhance model comparisons. to encompassing a broader range of settlement behavior and
Extrapolating settlement models to elapsed times greater than including physically significant model parameters. The Park and
the last elevation survey conducted on Day 3980 (t = 10.9 yr) Lee model primarily is applicable to biocompression and is

Please cite this article in press as: Bareither, C.A., Kwak, S. Assessment of municipal solid waste settlement models based on field-scale data analysis. Waste
Management (2015), http://dx.doi.org/10.1016/j.wasman.2015.04.011
 C.A. Bareither, S. Kwak / Waste Management xxx (2015) xxx–xxx 11

Data (C) Data (E) Control Cell Enhanced Cell

0
Gibson and Lo Sowers Logarithmic

2
Settlement (m)

(a) (b) (c)

6
0
Power Creep Hyperbolic Park and Lee

2
Settlement (m)

(d) (e) (f)

6
0
Marques Machado Babu

2
Settlement (m)

(g) (h) (i)

6
0
Chen-2010 Gourc Chen-2012

2
Settlement (m)

(j) (k) (l)

6
0.1 1 10 100 0.1 1 10 100 0.1 1 10 100
Elapsed Time (yr) Elapsed Time (yr) Elapsed Time (yr)

Fig. 5. Measured settlement and simulated settlement models for the Yolo Control and Enhanced cells.

incorporated in the Marques, Machado, Gourc, and Babu models to dependent settlement (Eq. (19)). Thus, omission of the Park and
represent the biocompression phase. Also, the Chen-2010 model Lee model in the DTBE analysis primarily is due incorporation of
uses the Park and Lee formulation to simulate complete time- this model in other settlement models.

Please cite this article in press as: Bareither, C.A., Kwak, S. Assessment of municipal solid waste settlement models based on field-scale data analysis. Waste
Management (2015), http://dx.doi.org/10.1016/j.wasman.2015.04.011
12 C.A. Bareither, S. Kwak / Waste Management xxx (2015) xxx–xxx

1.00
g f k
c k i l
g i l
b
j a j
2 2
0.95 R = 0.95 e R = 0.95
a
e f h

d
2
b
R 0.90

h c
0.85 d

(a) (b)
Control Cell Enhanced Cell
0.80
0 5 10 15 20 25 0 5 10 15 20 25
Number of Total Variables Number of Total Variables

1.00
k g k
v f
b g
i i
f
a&j l
l
0.95 2
R = 0.95 e 2
R = 0.95
a&j h
e

d
2
b
R 0.90

0.85 c
d
h
(c) (d)
Control Cell Enhanced Cell
0.80
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
Number of Optimized Variables Number of Optimized Variables

Fig. 6. Coefficient of determination versus total number of model parameters for (a) Control cell and (b) Enhanced cell and versus number of optimized model parameters for
(c) Control cell and (d) Enhanced cell. Settlement model references: a = Gibson and Lo; b = Sowers; c = Logarithm; d = Power Creep; e = Hyperbolic; f = Park and Lee;
g = Marques; h = Machado; i = Babu; j = Chen-2010; k = Gourc; and l = Chen-2012.

5.3. DTBE settlement simulation and prediction (Fig. 7) and the other three waste layers (Figs. S1-S3), which indi-
cates that the models used in the DTBE evaluation have the ability
Settlement measured in the DTBE (Fig. 3) can be attributed to to capture settlement behavior under different applied vertical
immediate compression, mechanical creep, and biocompression. stress. Dashed lines in Fig. 7 are predictions using average, opti-
Measured settlement and model simulations for WL1 in the mized model parameters based on the four waste layers simulated
DTBE are shown in Fig. 7. Similar plots for the other three waste in the DTBE (Tables S3 and S4). There exists greater difference
layers are included in Figs. S1, S2, and S3. A summary of model between measured settlement and the predictions as compared
parameters for the DTBE simulations and accompanying R2s are to the fitted models due to variation in settlement among the four
in Table S3 for full composite models and Table S4 for time-depen- waste layers in the DTBE (Fig. 3c).
dent models. Full composite models that incorporate all three The largest difference between measured settlement data for
phases of compression behavior (i.e., Gibson and Lo, Sowers, WL1 in the DTBE and settlement predictions based on average,
Marques, Babu, Chen-2010, and Chen-2012; Table 4) were optimized model parameters was observed for the Gibson and
implemented directly to simulate the DTBE data. Immediate Lo, Marques, and Babu models (Fig. 7). The Gibson and Lo model
compression, computed with a Cc0 (Eq. (4)), was added to all (Eq. (13)) includes the coefficient of volume compressibility (a or
time-dependent models (i.e., Machado, Gourc; Table 4) such that mv) and time-dependent compression parameter (b), which are
these models also were applicable to simulating all compression both multiplied by the increase in vertical stress (Drv).
phases. Temporal constraints of tM = 15 d and tB = 164 d were Mechanical creep is modeled in the Marques (Eq. (15)) and Babu
included to reduce the number of optimized parameters. (Eq. (18)) models with the time-dependent component of the
These time constraints are based on temporal behavior observed Gibson and Lo model. Thus, Drv has direct influence on settlement
in the DTBE (Bareither et al., 2012b). model parameters and calculated settlement for these three
Settlement models, whereby optimized parameters were models.
obtained via least squares analysis, are shown as solid lines in Relationships between a and immediate compression strain
Fig. 7. In general, all settlement models fit the DTBE data for WL1 versus Drv are shown in Fig. 8a for the Gibson and Lo model.

Please cite this article in press as: Bareither, C.A., Kwak, S. Assessment of municipal solid waste settlement models based on field-scale data analysis. Waste
Management (2015), http://dx.doi.org/10.1016/j.wasman.2015.04.011
 C.A. Bareither, S. Kwak / Waste Management xxx (2015) xxx–xxx 13

Measured Data Fitted Predicted

0.0
Gibson and Lo Sowers
0.2

0.4
Settlement (m)

0.6

0.8

1.0

1.2
(a) (b)
1.4
0.0
Marques Machado
0.2

0.4
Settlement (m)

0.6

0.8

1.0

1.2
(c) (d)
1.4
0.0
Babu Chen-2010
0.2

0.4
Settlement (m)

0.6

0.8

1.0

1.2
(e) (f)
1.4
0.0
Gourc Chen-2012
0.2

0.4
Settlement (m)

0.6

0.8

1.0

1.2
(g) (h)
1.4
0.1 1 10 100 0.1 1 10 100
Elapsed Time (yr) Elapsed Time (yr)

Fig. 7. Cumulative settlement versus time for Waste Layer 1 (WL1) of the DTBE data as well as fitted and predicted settlement. Predicted settlement was based on applying
average, optimized model parameters from the four waste layers in the DTBE to WL1 conditions.

The coefficient of volume compressibility decreases with increas- DTBE (Fitted Strain in Fig. 8a) displayed a modest increasing trend
ing applied stress (Fig. 8a), which is consistent with observations with increasing Drv; however, applying a single, average a
from large-scale compression tests conducted by Beaven and (Table S3) to all waste layers leads to an overestimation of imme-
Powrie (1996). The actual immediate compression strain in the diate compression at higher Drv and underestimation of

Please cite this article in press as: Bareither, C.A., Kwak, S. Assessment of municipal solid waste settlement models based on field-scale data analysis. Waste
Management (2015), http://dx.doi.org/10.1016/j.wasman.2015.04.011
14 C.A. Bareither, S. Kwak / Waste Management xxx (2015) xxx–xxx

represent all time-dependent compression. Thus, the b parame-

ters for the Gibson and Lo simulations generally are larger rela-
tive to the other two models, which reflect a greater magnitude
of time-dependent settlement.
The normalized difference between model parameters opti-
mized for each WL in the DTBE and the arithmetic average
DTBE model parameter (Tables S3 and S4) is shown in Fig. 9.
The normalization was completed with respect to the average
DTBE parameters and allows for comparison among model
parameters of varying order of magnitude. Normalized differ-
ences >0 indicate that WL-specific parameters are larger than
the DTBE average, whereas normalized differences <0 indicate
the opposite. The median (i.e., center line of the box) for each
WL lies within ±0.2; however, parameter sets for WL1 and
WL4 are biased negative, indicating a greater prevalence of
smaller model parameters relative to the average, whereas the
(a) parameter set for WL2 is biased positive, indicating a greater
prevalence of larger model parameters relative to the average.
This comparison suggests that variability in model parameters
exists within a controlled experiment designed to be representa-
tive of field conditions. The variation is attributed to different
compression behavior measured for the four waste layers
(Fig. 3c).
Settlement measured for the composite waste column in the
DTBE and multi-layer settlement predictions are shown in
Fig. 10. Data in Fig. 10 are representative of settlement measured
at the top of the waste column (Plate 4 in Fig. 3a and b).
Settlement model results in Fig. 10 include two analyses: Fitted –
optimized parameters determined for individual waste layers were
used to estimate settlement for the respective layers that was
summed to represent total settlement; and Predicted – settlement
for each waste layer was predicted using average, optimized DTBE
model parameters and settlement of the four layers was summed
to represent total settlement. The Fitted analysis was anticipated
to fit measured data well based on the use of waste-layer specific
(b) parameters. The Predicted analysis was to evaluate the efficacy of

Fig. 8. Relationships between (a) coefficient of volume compressibility and 3

immediate compression strain for Gibson and Lo model and (b) time-dependent
(DTBE - DTBE ) / DTBE
compression parameter versus vertical stress increase for the Gibson and Lo, WL# AVE AVE
Normalized Difference in Model Parameters

Marques, and Babu models. Model parameters are from the four waste layer
settlement simulations conducted for the Deer Track Bioreactor Experiment. 2

immediate compression at lower Drv (Predicted Strain in Fig. 8a).

Thus, a stress-dependent function for a may be needed in the 1
Gibson and Lo model to predict SI for waste layers subjected to dif-
ferent Drv. The vertical offset between measured settlement and
the settlement prediction for the Marques and Babu models was
attributed to variation in the immediate compression parameters 0
(Table S3). A similar, but smaller vertical offset of measured versus
predicted settlement is noticed for all models in Fig. 7, which also
is attributed to variation in immediate compression parameters
among the four waste layers (Table S3). -1
The relationship between b and Drv is shown in Fig. 8b for
the Gibson and Lo, Marques, and Babu models. Although scatter
exists in the b parameter with respect to Drv, there is no defini-
tive trend that suggests b has similar stress-dependent behavior -2
WL1 WL2 WL3 WL4
compared to a. The variation in b with respect to Drv is a func-
tion of the magnitude of time-dependent settlement simulated Fig. 9. Box plot of the normalized difference between settlement model parameters
for the four waste layers. Both the Marques and Babu models for a given waste layer (WL) in the Deer Track Bioreactor Experiment (DTBE) and
incorporate the same mathematical functions for mechanical the arithmetic average of model parameters for all four WLs in the DTBE. The box
represents the middle 50% of the data; the central line in the box represents the
creep and biocompression, which supports the similar magnitude
median; the outer boundaries represent the interquartile range (i.e., 25th and 75th
of b for both models (Fig. 8b). However, the Gibson and Lo model percentile); and the upper and lower whiskers extending from the box constitute
was implemented with a single mathematical function to 5th and 95th percentile of the data.

Please cite this article in press as: Bareither, C.A., Kwak, S. Assessment of municipal solid waste settlement models based on field-scale data analysis. Waste
Management (2015), http://dx.doi.org/10.1016/j.wasman.2015.04.011
 C.A. Bareither, S. Kwak / Waste Management xxx (2015) xxx–xxx 15

Measured Data Fitted Predicted

0.0
Gibson and Lo Sowers
2 2
R = 0.878 R = 0.999
0.5

1.0
Settlement (m)

1.5

2.0

2.5
(a) (b)
3.0
0.0
Marques Machado
2
2
R = 0.974 R = 0.999
0.5

1.0
Settlement (m)

1.5

2.0

2.5
(c) (d)
3.0
0.0
Babu Chen-2010
2 2
R = 0.947 R = 0.979
0.5

1.0
Settlement (m)

1.5

2.0

2.5
(e) (f)
3.0
0.0
Gourc Chen-2012
2
2
R = 0.999 R = 0.911
0.5
Settlement (m)

1.0

1.5

2.0

2.5
(g) (h)
3.0
0.1 1 10 100 0.1 1 10 100
Elapsed Time (yr) Elapsed Time (yr)

Fig. 10. Measured settlement at the top of the DTBE (Plate 4), fitted based on used of layer-specific optimized model parameters, and predicted based on use of average,
optimized model parameters for each waste layer.

using a single set of model parameters in a multi-layered analysis Babu, Chen-2010, and Chen-2012 models. This overprediction pri-
to predict settlement attributed to immediate compression, marily was attributed to overestimating SI for certain waste layers
mechanical creep, and biocompression. (e.g., Fig. 7), and decreasing the predicted SI in all models would
The general shape of the time-dependent settlement measured lead to an improved fit to the measured data. As discussed previ-
in the DTBE was simulated effectively by all eight models using ously, the Gourc, Sowers, and Machado models will all lead to con-
average, optimized DTBE model parameters (Predicted, Fig. 10). tinuous increasing settlement with time due to inclusion of CaM0 ,
In no cases was settlement underpredicted; however, a modest whereas the remainder of the models constrain long-term settle-
overprediction was obtained for the Gibson and Lo, Marques, ment to a finite magnitude.

Please cite this article in press as: Bareither, C.A., Kwak, S. Assessment of municipal solid waste settlement models based on field-scale data analysis. Waste
Management (2015), http://dx.doi.org/10.1016/j.wasman.2015.04.011
16 C.A. Bareither, S. Kwak / Waste Management xxx (2015) xxx–xxx

6. Settlement model recommendations provides a framework for estimating the initial waste thickness
and waste thickness at end-of-immediate compression. The imme-
Municipal solid waste settlement models were evaluated with diate compression ratio (Cc0 ) optimized to yield the target immedi-
respect to statistical significance, practicality of model application, ate settlement (Cc0 = 0.15 to 0.20) was less than Cc0 predicted from
and extrapolation to represent future settlement. An ideal settle- available empirical relationships (Cc0 = 0.23). A precompression
ment model should be straight-forward, have the ability to stress (10 to 15 kPa) was required to obtain the target immediate
represent immediate compression, mechanical creep, and biocom- settlement with Cc0 = 0.23 and recompression ratio = 1/10 Cc0 . A
pression, and include physically-significant model parameters. precompression stress should be included when using empirical
Although a case may be made that all settlement models used in estimates of Cc0 to predict immediate settlement under small verti-
the DTBE analysis are applicable, the Gourc model can be argued cal stresses.
to be the most practical. This model includes a low number of total Simulation of the Control and Enhanced cells with all settle-
and optimized model parameters (Table 4), has the ability ment models via least squares optimization yielded high coeffi-
to represent a broad range of compression behavior cient of determinations (R2 > 0.83) and accurate fits within the
(Figs. 5k, 7g, and 10g), and model parameters can be related to range of measured data. However, empirical models (power creep,
waste characteristics (Bareither et al., 2013) and gas generation logarithmic, and hyperbolic) are not recommended for MSW set-
behavior (Gourc et al., 2010). Also, an immediate compression tlement modeling due to non-representative behavior, limited
component can be added onto the Gourc model to expand model- physical significance of model parameters, and required settlement
ing capabilities to all compression processes. data for model parameterization.
The Machado and Sowers model have similar advantages to the Settlement models (Gourc, Machado, and Sowers) that include a
Gourc model. The Sowers model has been used extensively; how- mechanical creep compression ratio (CaM0 ) yield continuous settle-
ever, one challenge with this model is defining tF in Fig. 1. This ment for long-term extrapolations, whereas all other models (not
elapsed time can be identified using a dual-model approach as dis- including empirical models) include mathematical functions that
cussed in Bareither et al. (2013). Although the Machado model has constrain long-term settlement to a finite magnitude. Inclusion
similar advantages to the Gourc model, the Machado model of CaM0 may lead to overestimation of long-term settlement,
includes twice the number of model parameters (Table 4) to whereas models that constrain settlement due to mechanical creep
achieve a settlement prediction with similar statistical may lead to underestimation of long-term settlement.
performance. The Gourc model included the lowest number of total and opti-
The Gibson and Lo and Chen-2010 model, as implemented, mized model parameters and yielded high statistical performance
combine time-dependent compression due to mechanical creep for the DTBE prediction (R2 = 0.99). In addition, model parameters
and biocompression into a single mathematical function. This included in the Gourc model have been shown to have physical sig-
approach is convenient to reduce the number of model parameters nificance (e.g., Gourc et al., 2010; Bareither et al., 2013) such that
(Table 4), but limits the ability to uniquely represent different parameters can be estimated from waste characteristics and/or
time-dependent compression behavior. The Gibson and Lo model empirical relationships. Overall, the Gourc model was found to
includes a stress-dependent model parameter for immediate com- be the most applicable and straightforward to implement and is
pression that yielded inaccurate predictions when using a single recommended for use in practice. The other models that included
parameter with varying applied stress (Fig. 8a). Also, both models unique mathematical functions for immediate compression,
constrain settlement to a finite value. Thus, the Gibson and Lo and mechanical creep, and biocompression (Machado, Sowers,
Chen-2010 may not be applicable for predicting settlement due to Marques, Babu, and Chen-2012) were all capable of yielding satis-
multiple compression processes. factory MSW model fits and predictions; however, additional
The Marques and Babu models employ similar mathematical model parameters or constraints are necessary to implement these
functions for mechanical creep and biocompression, but use differ- models.
ent soil mechanics parameters to represent immediate compres-
sion. The critical state framework introduced in the Babu model
Acknowledgements
was not found to provide an improved prediction of SI relative to
the Cc0 approach included in the Marques model.
The authors would like to acknowledge the support of Colorado
Finally, the Chen-2012 model was difficult to implement and
State University in completing this work. All finding, inferences,
included the highest number of model parameters (Table 4). The
and opinions that have been reported are solely those of the
increased number of model parameters did not improve the settle-
authors.
ment prediction in the DTBE relative to the other models (Fig. 10).
An improved fit likely could have been obtained via increasing the
number of optimized model parameters or expanding the range in
Appendix A. Supplementary material
which parameters could be optimized; however, this approach is
not ideal and may lead to model parameters that have limited
Supplementary data associated with this article can be found, in
physical significance. Overall, the unsaturated consolidation theory
the online version, at http://dx.doi.org/10.1016/j.wasman.2015.04.
used to develop the Chen-2012 model was not found to provide
011.
advantages relative to the other settlement models.

References
7. Summary and conclusions
Babu Sivakumar, G.L., Reddy, K.R., Chouskey, S.K., 2010a. Constitutive model for
municipal solid waste incorporating mechanical creep and biodegradation-
An evaluation of MSW settlement model performance and
induced compression. Waste Manage. 30 (1), 11–22.
applicability was conducted based on analyses of field-scale data: Babu Sivakumar, G.L., Reddy, K.R., Chouskey, S.K., Kulkarni, H., 2010b. Prediction of
(1) Yolo Co. Pilot Project (Yolo), which included a Control and long-term municipal solid waste landfill settlement using constitutive model.
Enhanced cell, and (2) Deer Track Bioreactor Experiment (DTBE), Practice Periodical Hazardous, Toxic, Radioactive Waste Manage. 14 (2), 139–
150.
which included four settlement plates at varying waste depths. A Babu Sivakumar, G.L., Reddy, K.R., Chouskey, S.K., 2011. Parametric study of MSW
multi-layer immediate settlement analysis developed for Yolo landfill settlement model. Waste Manage. 31 (6), 1222–1231.

Please cite this article in press as: Bareither, C.A., Kwak, S. Assessment of municipal solid waste settlement models based on field-scale data analysis. Waste
Management (2015), http://dx.doi.org/10.1016/j.wasman.2015.04.011
 C.A. Bareither, S. Kwak / Waste Management xxx (2015) xxx–xxx 17

Bareither, C.A., Benson, C.H., Barlaz, M.A., Edil, T.B., Tolaymat, T.M., 2010. Kwak, S. 2014. Assessment of municipal solid waste settlement models based on
Performance of North American bioreactor landfills: I. leachate hydrology and field-scale data analysis, MS Thesis, Dept. of Civil & Environmental Engineering,
waste settlement. J. Environ. Eng. 136 (8), 824–838. Colorado State University, Fort Collins, CO, USA.
Bareither, C.A., Benson, C.H., Edil, T.B., Barlaz, M.A., 2012a. Abiotic and biotic Ling, H.I., Leshchinsky, D., Mohri, Y., Kawabata, T., 1998. Estimation of municipal
compression of municipal solid waste. J. Geotech. Geoenviron. Eng. 138 (8), solid waste landfill settlement. J. Geotech. Geoenviron. Eng. 124 (1), 21–28.
887–888. Liu, C., Chen, R., Chen, K., 2006. Unsaturated consolidation theory for the prediction
Bareither, C.A., Breitmeyer, R.J., Benson, C.H., Barlaz, M.A., Edil, T.B., 2012b. Deer of long-term municipal solid waste landfill settlement. Waste Manage. 24, 80–
Track Bioreactor Experiment: a field-scale evaluation of municipal solid waste 91.
bioreactor performance. J. Geotech. Geoenviron. Eng. 138 (6), 658–670. Machado, S.L., Carvalho, M., 2009. Constitutive model for municipal solid waste
Bareither, C.A., Benson, C.H., Edil, T.B., 2012c. Compression behavior of municipal incorporating mechanical creep and biodegradation-induced compression.
solid waste: immediate compression. J. Geotech. Geoenviron. Eng. 138 (9), Waste Manage. 30 (1), 11–22.
1047–1062. Machado, S.L., Carvalho, M.F., Vilar, O.M., 2002. Constitutive model for municipal
Bareither, C.A., Benson, C.H., Edil, T.B., 2013. Compression of municipal solid waste solid waste. J. Geotech. Geoenviron. Eng. 128 (11), 940–951.
in bioreactor landfills: mechanical creep and biocompression. J. Geotech. Marques, A.C.M., Filz, G.M., Vilar, O.M., 2003. Composite compressibility model for
Geoenviron. Eng. 139 (7), 1007–1021. municipal solid waste. J. Geotech. Geoenviron. Eng. 129 (4), 372–378.
Beaven, R.P., Powrie, W., 1996. Determination of the hydrogeological and McDougall, J.R., Pyrah, I.C., 2004. Phase relations for decomposable soils.
geotechnical properties of refuse in relation to sustainable landfilling. In: Geotechnique 54 (7), 487–493.
Proc. 19th Int. Madison Waste Conf. University of Wisconsin-Madison, Madison, Mehta, R., Barlaz, M.A., Yazdani, R., Augenstein, D., Bryars, M., Sinderson, L., 2002.
WI, USA, pp. 435–454. Refuse decomposition in the presence and absence of leachate recirculation. J.
Berthouex, P.M., Brown, L.C., 2002. Statistics for Environmental Engineers, 2nd ed. Environ. Eng. 128 (3), 228–236.
Lewis Publishers, Boca Raton, FL. Olivier, F., Gourc, J.-P., 2007. Hydro-mechanical behavior of municipal solid waste
Bjarngard, A., Edgers, L., 1990. Settlement of municipal solid waste landfills. In: subject to leachate recirculation in a large-scale compression reactor cell. Waste
Proc. 13th Int. Madison Waste Conf. University of Wisconsin-Madison, Madison, Manage. 27, 44–58.
WI, USA, pp. 192–205. Abichou, T., Barlaz, M.A., Green, R., Hater, G., 2013. The Outer Loop bioreactor: a case
Chen, Y., Ke, H., Fredlund, D.G., Zhan, L., Xie, Y., 2010. Secondary compression of study of settlement monitoring and solids decomposition. Waste Manage. 33
Municipal solid wastes and a compression model for predicting settlement of (10), 2035–2047.
municipal solid waste landfills. J. Geotech. Geoenviron. Eng. 136 (5), 706–717. Park, H.I., Lee, S.R., 1997. Long-term settlement behavior of landfills with refuse
Chen, K.S., Chen, R.H., Liu, C.-N., 2012. Modeling municipal solid waste landfill decomposition. J. Res. Manage. Technol. 24 (4), 159–165.
settlement. J. Environ. Eng. Sci. 66 (8), 2301–2309. Park, H.I., Lee, S.R., 2002. Long-term settlement behavior of MSW landfills with
Edil, T.B., Ranguette, V.J., Wuellner, W.W., 1990. Settlement of municipal refuse. In: various fill ages. Waste Manage. Res. 20 (3), 259–268.
Landva, A.O., Knowles, G.D. (Eds.), Geotechnics of Waste Fills-Theory and Park, H.I., Park, B., Lee, S.R., Hwang, D., 2007. Parameter evaluation and performance
Practice, STP 1070. ASTM, West Conshohocken, PA, USA, pp. 225–239. comparison of MSW settlement Prediction models in various landfill types. J.
El-Fadel, M., Khoury, R., 2000. Modeling settlement in MSW landfills: a critical Environ. Eng. 133 (1), 64–72.
review. Crit. Rev. Environ. Sci. Technol. 30 (3), 327–361. Sharma, H.D., De, A., 2007. Municipal solid waste landfill settlement: post closure
El-Fadel, M., Shazabak, S., Saliby, E., Leckie, J., 1999. Comparative assessment of perspectives. J. Geotech. Geoenviron. Eng. 133 (6), 619–629.
settlement models for municipal solid waste landfill applications. Waste Siddiqui, A., Powrie, W., Richards, D., 2013. Settlement characteristics of
Manag. Res. 17 (5), 347–368. mechanically biologically treated wastes. J. Geotech. Geoenviron. Eng. 139
Fredlund, D.G., Hasan, J.U., 1979. One-dimensional consolidation theory: (10), 1676–1689.
unsaturated soils. Can. Geotech. J. 16 (3), 521–531. Sowers, G.F. 1973. Settlement of waste disposal fills. In: Proc. 8th Int. Conf. on Soil
Gibson, R.E., Lo, K.Y., 1961. A theory of soils exhibiting secondary compression. Acta Mech, and Found, Engr., Balkema, Rotterdam, vol. 22, pp. 207–210.
Polytech. Scand. C10, 1–15. Staley, B.F., Barlaz, M.A., 2009. Composition of municipal solid waste in the U.S. and
Gourc, J.P., Staub, M.J., Conte, M., 2010. Decoupling MSW settlements into implications for carbon sequestration and methane yield. J. Environ. Eng. 135
mechanical and biochemical processes – modeling and validation on large- (10), 901–909.
scale setups. Waste Manage. 30 (8–9), 1556–1568. Tan, T.-S., Inoue, T., Lee, S.-L., 1991. Hyperbolic method for consolidation analysis. J.
Hossain, M.S., Gabr, M.A., 2005. Prediction of municipal solid waste landfill Geotech. Eng., ASCE 117 (11), 1723–1737.
settlement with leachate recirculation. In: Proc. Geo-Frontiers, GSP No. 142, Vilar, O.M., Carvalho, M.F., 2004. Mechanical properties of municipal solid waste. J.
Waste Containment and Remediation. ASCE, pp. 1–14. Test. Eval. 32 (6), 438–449.
Hossain, M.S., Gabr, M.A., Barlaz, M.A., 2003. Relationship of compressibility Wood, D.M., 1990. Soil Behavior and Critical State Soil Mechanics. Cambridge
parameters to municipal solid waste decomposition. Geotech. Geoenviron. University Press, United Kingdom.
Eng. 129 (12), 1151–1158. Yazdani, R., Kieffer, J., Sananikone, K., Augenstein, D. 2006. Full scale bioreactor
Ivanova, L.K., Richards, D.J., Smallman, D.J., 2008. The long-term settlement of landfill for carbon sequestration and greenhouse emission control, final report,
landfill waste. Water Resour. Manage. 161 (WR3), 121–133. Award No. DE-FC26-01NT41152, U.S. Department of Energy, Washington, DC.
Kavazanjian, E., Jr., Matasovic, N., Bachus, R.C. 1999. Large diameter static and cyclic Yen, B.C., Scanlon, B., 1975. Sanitary landfill settlement rates. J. Geotech. Eng. Div.,
laboratory testing of municipal solid waste. In: Proc., 7th Int. Waste ASCE 101 (5), 475–487.
Management and Landfill Symp., CISA, Environmental Sanitary Engineering
Centre, Cagliari, Italy, pp. 437–444.

Please cite this article in press as: Bareither, C.A., Kwak, S. Assessment of municipal solid waste settlement models based on field-scale data analysis. Waste
Management (2015), http://dx.doi.org/10.1016/j.wasman.2015.04.011