Research Article

Transportation Research Record

2023, Vol. 2677(8) 118–136
Ó National Academy of Sciences:
Enhanced Pavement Design and Analysis Transportation Research Board 2023

Framework to Improve the Resiliency of Article reuse guidelines:

sagepub.com/journals-permissions
Flexible Airfield Pavements DOI: 10.1177/03611981231155909
journals.sagepub.com/home/trr

Paula Sutherland Rolim Barbi1 , Pejoohan Tavassoti1 ,

and Susan Tighe2

Abstract
Changes in climatic conditions can directly affect pavement performance. However, accounting for environmental factors in
airport pavement design remains a challenge since design methods such as FAA rigid and flexible iterative elastic layered
design (FAARFIELD) do not consider moisture and temperature variation as input. Therefore, to address this research gap
and improve the resilience of airport pavements, this research proposes a new methodology for the structural design of flex-
ible airport pavements. The proposed methodology in this research was applied to a case study of an international airport in
Canada, using actual field data. Five scenarios were evaluated including the current climate, temperature increase, lower
matric suction, and two flooding events. The results of the proposed design method showed that the traditional FAARFIELD
analysis can possibly overestimate fatigue damage, and underestimate rutting damage. The outcomes showed that climate
change can increase pavement damage and shorten the service life from 7 to 14 years in the scenarios evaluated. It was also
concluded that the lowering of the matric suction can result in the highest damage levels. Considering the implications of cli-
mate change on transportation infrastructure, the proposed methodology can contribute to designing more resilient airport
pavements in the future, since it accounts for climate variations, temperature, and moisture changes, as well as extreme
events such as flooding over the design life of flexible airport pavements.

Keywords
flexible airfield pavements, pavement design, climate change, cumulative damage factor, resilient pavements

Structural design of airport pavements mainly relies on structure can carry, and each aircraft wheel coverage
the properties of pavement materials, loading conditions, then contributes to the cumulative damage factor (CDF)
subgrade soil strength, structural configuration, the based on Miner’s law.
selected pavement failure models, and the ambient condi- According to Miner’s law, pavement damage can be
tions (1). Accounting for the ambient conditions, how- defined as the ratio of the actual number of traffic load
ever, can be challenging, since the common design repetitions over the allowable number of load repetitions
practices do not consider the variation in climate-related (number of coverages) that a pavement structure can
parameters such as temperature and moisture level as an carry (3). Because of aircraft wander, the number of cov-
input for design practices (2). Failure models (aka dam- erages can differ from the number of passes, and for that
age models) are equations used to convert pavement
mechanistic responses into pavement distresses with the
1
goal of quantifying the damage. To obtain the structural Department of Civil and Environmental Engineering, University of
responses of flexible pavements, most computer pro- Waterloo, Waterloo, ON, Canada
2
McMaster University, Hamilton, ON, Canada
grams use the layered elastic design (LED) approach.
The failure models in the link the computed structural Corresponding Author:
responses to the number of coverages a pavement Paula Sutherland Rolim Barbi, [email protected]
Barbi et al 119

road pavements. Such studies have used the Mechanistic-

Empirical Pavement Design Guide (MEPDG) software
to assess the impacts of climate change on road pave-
ment performance and design (5–7). The main findings
suggest the implementation of strategies that include
upgrading binder grade, increasing layer thicknesses, and
stabilizing granular materials.
Airport pavements require closer attention than road
pavements because they suffer concentrated high-speed
loads, and a structural flaw can lead to serious conse-
quences for the safety of the users. However, research on
the impacts of climate change on airport pavements has
been very limited to date. A survey developed by the
Centre for Pavement Transportation and Technology
(CPATT) at the University of Waterloo investigated sev-
eral Canadian airports and their considerations in rela-
tion to climate change (8). The results of the survey
showed that airport authorities perceive an increase in
pavement distresses because of climate change, and they
all agree that the changing environment already has an
Figure 1. FAA rigid and flexible iterative elastic layered design
(FAARFIELD) solution process.
impact in their current practices to some extent.
Source: Modified from Heymsfield and Tingle (1). Climatic changes have been recently investigated in a
variety of airports across Canada and they found that
adaptation strategies in pavement materials and design
reason, the coverage-to-pass (C/P) ratio is incorporated practices were necessary (9). The need for adaptation stra-
in the analysis. An outline for the pavement structural tegies has motivated the authors of this research to develop
design steps followed by FAARFIELD is summarized in a new methodology that can provide benefits in relation to
Figure 1. quantifying the impact of climate change on pavement per-
As showed in Figure 1, FAARFIELD repeats a cycle formance, which can ultimately help with the design of
by changing the pavement layer thicknesses until it more resilient airfield pavements in the future. Therefore,
reaches CDF = 1, which means that the pavement dam- the main objectives of this paper are to: (a) propose a new
age reaches 100% at the end of its service life. methodology for the structural design of flexible airport
Other airport pavement design software available pavements considering climate variation over the pave-
includes Airport Pavement Structural Design System ment service life; (b) implement the proposed methodology
(APSDS, Australia), ALIZE (France), and Pavement- in a case study using actual field data; and (c) perform the
Transportation Computer Assisted Structural analysis of different climate scenarios, including the cur-
Engineering (PCASE, USA). Similar to FAARFIELD, rent climate, temperature increase, lower matric suction,
the APSDS, ALIZE and PCASE codes for flexible pave- and two flooding events.
ment design are based on the layered elastic theory. All
the software uses transfer functions to convert the calcu-
lated mechanistic responses to pavement distresses, using Methodology
equations obtained through full-scale tests. One of the This research proposes a methodology that enables the
main differences between these design methods is the structural design of flexible airport pavements consider-
way that the calibration parameters are used to construct ing climate variations throughout the year, and climate
failure models (1, 4). change throughout the service life of the facility. This
The consideration of climate variation over the pave- study proposes the following 7-step methodology that
ment service life is essential to incorporate climate can be incorporated in airport design practices and
change predictions in the current design practices, never- includes: (1) obtaining hourly climatic data; (2) determin-
theless, it is not possible to assess the damage caused by ing and processing the pavement temperature profile; (3)
climate fluctuation in FAARFIELD, APSDS, ALIZE or estimating the asphalt concrete (AC) dynamic modulus
PCASE, since these software packages consider that the and the soil resilient modulus throughout the year; (4)
pavement materials have constant stiffness values calculating pavement stresses and strains; (5) determin-
throughout the year and over the design life. There are a ing the number of coverages and the C/P ratio; (6) calcu-
few studies that address the effects of climate change on lating the CDF for rutting and fatigue; and (7) adjusting
120 Transportation Research Record 2677(8)

Figure 2. Proposed solution process.

Note: Dashed rectangles indicate actions that are taken under the same step.

the layer thicknesses if CDF is greater than 1 and repeat- properties (i.e., albedo, emissivity, and absorption coeffi-
ing the process until the cumulative damage equals cients). The climatic, meteorological, and surface albedo
100%. These steps will be further discussed in the follow- data were retrieved from the National Solar Radiation
ing sections, accompanied by the results of a case study Database (NSRDB), which is available to the public for
that is analyzed under different climate change scenarios. free. Then, the temperature gradient was calculated
The flowchart in Figure 2 summarizes these steps. through the finite control volume method (10). To this
end, temperature estimate model for pavement structures
(TEMPS) software was utilized. This aspect covers the
Hourly Climatic Data and Pavement Temperature first two steps of the proposed methodology and provides
Profile the foundation for the next steps.
Although most of the existing airfield pavement practices
rely on the simplified temperature and stiffness scenarios,
the modern computational power capability allows
Asphalt Concrete and Subgrade Modulus
researchers to utilize more accurate and realistic climate Step 3 of the proposed methodology focuses on AC and
and climate-dependent parameters. To determine the subgrade modulus determination. The stiffness of the
stiffness of each pavement layer throughout the year, first layers, combined with the definition of the aircraft fleet
it is necessary to estimate the pavement temperature pro- mix, the layer thickness, and the design life, are a crucial
file. Understanding the temperature gradient at different part in a solution process that can incorporate climate
depths throughout the year can be useful to determine parameters. This study aims to include the temperature
the state of pavement subgrade (i.e., frozen, thawed, or dependency aspect of the pavement materials in the anal-
unfrozen), as well as the temperature of the AC across ysis to improve the currently used constant or single-
the year. The input parameters used for calculating the value modulus values. To determine the dynamic modu-
temperature profile included the climatic data (i.e., air lus of the AC, the modified Witczak’s sigmoidal function
temperature and wind speed), solar radiation as the was used (11). The main model inputs are the loading
meteorological data, and pavement surface radiation rate, temperature, binder performance grade (PG), and
Barbi et al 121

volumetric design information. The effect of loading rate The recovery period (TR ) depends mainly on the per-
can be estimated considering a sinusoidal loading, with centage of fines and the soil plasticity index (PI), and ref-
frequency f , to represent the cyclic application of erence values can be found
 in
 the MEPDG manual. The
dynamic loads on pavements. The duration of the load resilient modulus ratio MR
at any given time is defined
MRopt
on a pavement depends on the aircraft speed, s, and the
tire contact radius, a (12). according to Equation 4:
On the other hand, to determine the variations in the
MR ba
resilient modulus of the subgrade soil with time, this log =a+  b   ð4Þ
research proposes a similar approach to the MEPDG, in MRopt 1 + EXP ln a + km 3 S  Sopt
which an environmental adjustment factor (Fenv ) is used in which a, b and km are parameters that depend on the
to modify the optimum resilient modulus (MRopt ) of the soil gradation, and S  Sopt is the variation in degree of
soil according to variations in moisture and temperature. saturation from the optimum condition, expressed in
The resilient modulus at any given period can be decimal. If ground water levels are stable and there are
expressed by Equation 1 (13). no significant cracks in the AC layer, it can be said that
the saturation S is essentially at a state of equilibrium,
MR = Fenv 3 MRopt ð1Þ
unless freezing or thaw recovery is in progress. The equi-
The Fenv factor adjusts the optimal resilient modulus for librium degree of saturation Sequil is a function of the soil
three main conditions: (1) frozen materials (Ff ); (2) matric suction ðhÞ. The matric suction can be defined as
thawed materials recovering to an equilibrium state (Fr ); the capillary pressure of the soil (i.e., ua  uw , where ua
(3) unfrozen, fully recovered materials or materials that is the pore-air pressure and uw is the pore water pres-
were never frozen (Fu ). The frozen adjustment factor (Ff ) sure). The relationship between the soil suction and the
is a function of the frozen resilient modulus MRfrz and the degree of saturation can be depicted by the soil-water
optimum resilient modulus, according to Equation 2. characteristic curve (SWCC).
The unfrozen adjustment factor (Fu ) is used when the
MRfrz soil is fully recovered from thawing. It takes place when
FF = ð2Þ
MRopt recovery ratio RR.1, which happens when the number
of hours elapsed since thawing started (Dt) is greater than
The recovering adjustment factor (Fr ) accounts for the the number of hours required for the material to recover
decrease in the resilient modulus when the soil is thawed, (TR ). In this case, Fu = MMRoptR
, according to Equation 4,
and the recover in the soil resistance that happens as the and S = Sequil ; that is, the saturation is at equilibrium.
soil drains. Therefore, Fr is a function of the reduction As an example, after a flooding event, the water table
factor (RF), the recovery
 ratio (RR), and the soil resilient level is elevated, and most of the layers can reach satura-
modulus ratio MMRoptR
, in which MR is the resilient modu- tion levels of 100%. As the water drains out of the pave-
lus at a given time after the soil has thawed. The soil RF ment structure, the matric suction is raised until it
can be estimated according to the soil gradation and the reaches its normal suction levels. The damage incurred
plasticity index, and reference values can be found in the during the immediate post flooding period (IPFP) can be
MEPDG manual. The RR is a function of the number of conceptually defined according to Equation 5 (14);
hours elapsed since the beginning of thawing (Dt) and the
number of hours required for the material to recover ðT
(TR ), according to Equation 3. DIPFP = DfLðtÞ, hðtÞgdt ð5Þ
0
Dt
RR = ð3Þ in which LðtÞ is the traffic load and hðtÞ is the matric suc-
TR
tion level at a specific time t. It is not possible to find a
The number of hours elapsed since thawing started (Dt) continuous function that would translate the continuous
can be processed using a monthly-basis approach and recovery of the matric suction; therefore, Equation 5 can
translated into a normalized function based on the mean be adjusted by considering the soil suction to increase
value and standard deviation of each month. The Dt dis- into discrete intervals of time and have one single record
tribution curves are then divided into five sub-seasons, of the matric suction and the respective traffic-load at
representing 20% of the frequency distribution. each interval
Therefore, Dt records corresponding to the standard nor- Evidence shows that the wet-dry cycle does not
mal deviates of 21.2816, 20.5244, 0, 0.5244, and 1.2816 adversely affect subgrade stiffness, that is, the strength of
are determined, which correlate to 10, 30, 50, 70 and the subgrade soil can recover after a flood if it is given
90% accumulated frequencies within a given period, enough time to dry (14). Therefore, this research consid-
respectively. ered that the resilient modulus of the subgrade fully
122 Transportation Research Record 2677(8)

recovers after the water from flooding is completely top of the subgrade, eh = critical horizontal tensile
drained, and the pavement returns to its equilibrium strain at the bottom of the AC layer; S = flexural stiff-
degree of saturation. ness of Hot Mix Asphalt (HMA) mix from the labora-
tory fatigue testing (MPa); AV = the air voids of
mixture (%); Vb = the asphalt content by volume;
Pavement Responses, Coverages, Coverage-to-Pass PNMS = the percent of aggregate passing the nominal
Ratio, and CDF maximum size sieve; PPCS = the percent of aggregate
This section explains steps 5, 6 and 7 of the proposed passing the primary control sieve; and P200 = the per-
methodology by presenting some of the available tools cent of aggregate smaller than 75 mm.
to obtain the pavement responses, introducing how pave- Since the number of coverages can differ from the
ment responses are connected to the number of cov- number of passes because of aircraft wander, a C/P ratio
erages, how to estimate the C/P, and last, by showing the needs to be computed. The C/P ratio can be defined as
procedures to calculate cumulative damage because of the probability that any part of the effective tire width
rutting and fatigue. covers the center point of a given strip, as presented in
Flexible pavement design codes typically use the LED Equation 8 (17):
approach for its structural analysis (1), however, advances
in knowledge about materials behavior and computer tech- X Nt  weq   weq 
nology have greatly increased the interest in numerical tech- ðC=PÞi = P xi  ł xk ł xi + ð8Þ
k =1
2 2
niques, such as the finite element method. Some of the
available LED-based tools for the structural analysis of where Nt = number of tires for the gear; xi = lateral
pavements are: KENLAYER, CHEVRON, DAMA, distance from centerline to the midpoint of strip i; xk =
ELSYM5, BISAR, EVERSTRESS, and NonPAS. The lateral distance from centerline of the runway or taxiway
mechanistic-empirical design of airport pavements is an to the centerline of tire k; weq = effective tire width. Up
iterative process, which may require several iterations until to version 1.4, FAARFIELD design procedure would
the optimal design is found. Although more sophisticated account for the effect of aircraft gears in tandem as part
techniques can potentially improve the accuracy of the mod- of the C/P ratio computation, in which the tandem factor
eling, the required computation time and cost to obtain the was computed as a straight-line interpolation between
necessary input parameters can be an obstacle. Therefore, the number of wheels in tandem (for shallow structures)
selection of the right modeling tool to use in the design pro- and unity (for deep structures). This procedure, however,
cess for flexible pavements will be a trade-off depending on has changed in the most recent versions of
the nature of the analysis and level of accuracy targeted. FAARFIELD, and the cumulative damage caused by an
Airport pavement transfer functions connect pave- aircraft mix is now defined according to Equations 9
ment responses to the number of coverages that the and 10 (18).
pavement can carry. Coverages can be defined as the
number of repetitions of the maximum strain before fail- m  
X C
ure. The number of coverages to rutting failure is defined CDFi = 3 PA 3 DðyiÞ ð9Þ
in FAARFIELD 1.41 by a bilinear model (Equation 6) A=1
P
(15) and fatigue failure is defined based on the ratio of
X
n
1
dissipated energy change (RDEC) model (Equation 7) DðyiÞ = sk 3 ð10Þ
(16). k =1
C ðek Þ

1 In which, m = number of aircraft types in the traffic

C = 10ða + b + eÞ ðcoverages.1000Þ
1 c
or mix; PA = total number of passes of aircraft A in the
 8:1 ð6Þ
0:00414131 design period; (yi) is the damage computed along the off-
C= ðcoverages ł 1000Þ set yi; n is the total number of extremum points of strain;
e
ek are the critical strain values; C is the number of cov-
 1:85 erages according to Equations 6 and 7, and sk is a factor
C = 0:4801 3 44:422 3 eh 5:14 3 S 2:993 3 AV
AV + Vb characterizing the kth extremum strains, as presented in
Equation 11 (18).
 0:4063 !0:9007
PNMS  PPCS
3 ð7Þ ( )
P200 1 if ek is maximum
sk = 1 if ek is minimum ð11Þ
where a =  0:163768916705, b = 185:192806802, 0 if ek is not extremum
c = 1:65054449461, e is the critical vertical strain at the
Barbi et al 123

Case Study: Pearson Airport about 452,800 flights were handled (19). In 2020 and
2021 the number of flights decreased drastically because
Toronto Pearson International Airport is the main inter-
of the COVID-19 pandemic, resulting in 32,899 and
national airport serving Toronto, its metropolitan area,
6,383 flights in 2020 and 2021 respectively. It is believed
and its surrounding region. It has 5 runways and several
that the number of flights will resume its normal levels in
taxiways interconnecting the airport. The soil classification
the next few years, and Toronto Pearson Airport will
was based on the borehole data provided by the Greater
continue to grow in the long term. For this reason, the
Toronto Airports Authority (GTAA), which indicates that
traffic information used in this analysis will be based on
the predominant subgrade material is a Sandy Lean Clay
the total operations in 2019, and the growth predictions
with Gravel with a low to moderate frost susceptibility.
will be based on the airport master plan published in
The subgrade soil composition includes 8.8% of gravel,
2018 (20). The airport master plan predicts that takeoffs
29% of sand, 32.5% of silt and 29.7% of clay. Further
and landings will grow at a rate that ranges from 1.13%
details of the subgrade soil is provided in Table 1, such as
to 1.5%. Considering the impact of the COVID-19 pan-
the plastic and liquid limit, the plasticity index, the percent
demic, the authors assumed the lowest growth rate pre-
passing sieve #200 ðp200 Þ, the grain size corresponding to
diction of 1.13%.
60% passing by weight (D60 ) and the California Bearing
An airport runway design is greatly influenced by the
Ratio (CBR) value. The optimum resilient modulus of the
type of aircrafts serving an airport. To perform the struc-
subgrade soil was calculated based on the CBR results pro-
tural design of a runway, first it is necessary to define the
vided by the GTAA and equals 51.71 MPa.
aircraft fleet mix, that is, the percentage of aircraft, by
In relation to aircraft traffic in 2019, 50.5 million pas-
type or category, that operate at the airport. Runway 05-
sengers travelled through Toronto Pearson Airport and
23, managing 45% of the aircraft operations, is the busi-
est runway in the airport (21). The aircraft fleet mix pro-
Table 1. Subgrade Properties at Pearson posed was developed based on the traffic data records
from 2019, according to data provided by the GTAA
Soil type Sandy lean clay with gravel and is presented in Table 2.
Plastic limit 17.2% It is a common practice to consider the aircraft wheel
Liquid limit 29.6% load to have a circular contact area with a flexible pave-
Plasticity index 12.4 ment. FAARFIELD, however, defines the wheel contact
p200 62.1% as elliptical, following the guidelines given by Huang (22),
D60 0.066 mm
CBR 5%
in which the contact length is about 1.6 times longer than
the contact width. Table 3 presents the loading characteris-
Note: CBR = California Bearing Ratio. tics of the B777-300 ER is the most critical aircraft in the
Source: Provided by Greater Toronto Airport Authority. mix because it causes the highest damage to the pavement.

Table 2. Traffic Inputs: Annual Growth of 1.13% Used to Calculate Total Departures

Aircraft Gross taxi weight (kg) Annual departures Total departures

A320-200 std 73,900 7063 157,222

A321-100 std 83,400 5679 126,415
B767-300 159,665 5301 118,000
Bombardier CL-604/605 21,863 3495 77,799
B777-300 ER 352,441 3056 68,027
B787-9 254,692 2803 62,395
B777-200 248,120 3799 84,566
A330-300 std 230,900 842 18,743
B787-8 228,383 579 12,889
A310-300 142,900 772 17,185
B747-400 397,801 245 5454
B747-400 Belly 397,801 245 5454
A380-800 WV000 562,000 105 2337
A380-800 WV000 Belly 562,000 105 2337
DHC-7 19,867 30,830 686,276
B737-800 79,242 19,706 438,656
CRJ100/200 21,636 8552 190,368
124 Transportation Research Record 2677(8)

Table 3. Loading Characteristics of the B777-300 ER

Gross taxi Tire Percent Dual tire Tandem tire Tire contact Tire contact Tire contact
weight (kg) pressure (kPa) weight on gear spacing (mm) spacing (mm) width (mm) length (mm) area (mm2)

352441 1503 0.475 1397 1463 380 609 182045

Table 4. Input Parameters to Estimate Asphalt Concrete Dynamic Modulus Master Curve

Parameter f (Hz) Va (%) Vbeff (%) r34 (%) r38 (%) r4 (%) r200 (%)

Value 4.3 3.9 12.6 0.00 19.70 45.70 3.00

Existing Runway Structure and Materials pre-set in the program of 1,379 MPa (i.e., 200,000
According to the GTAA, the pavement structure from a pounds per square inch (psi)). The manual affirms that
typical flexible runway in Toronto Pearson Airport consists this modulus was conservatively chosen and corresponds
of three layers, namely an AC layer of 125 mm thickness, to a pavement temperature of approximately 32°C (24),
followed by a granular base of 300 mm and a granular sub- therefore, the software considers that the pavement stiff-
base of 900 mm. The materials have similar characteristics ness would perform as it was under 32°C all year long.
to the standard materials from FAARFIELD. Therefore, the modified Witczak model was used to cal-
According to Steward (23) and GTAA information in culate a modulus that would be more coherent with the
2021, the overall pavement thickness of Toronto Airport AC described in the previous section. This was achieved
runways has not been updated since 1993, and the runways using the mix gradation parameters, PG, initial air voids,
were designed for the critical aircraft is the one that causes effective asphalt content, aircraft characteristics such as
the highest damage to the pavement among the aircraft fleet tire pressure, weight and speed, and lastly, temperature.
mix at the time, which was the B747-400. It is expected that The resulting modulus of the hot mix asphalt at 32°C
newer aircrafts such as the B777- 200LR/300ER, A340-500/ was therefore found to be 1,721 MPa, which is 25%
600, and A380 are all more critical for the determination of higher than the standard AC stiffness used in
a suitable pavement structure, and that failure to improve FAARFIELD.
the pavement structure will result in reduced service life. The required data for the design in FAARFIELD
The AC mix used in this research was collected at consists of the soil resilient modulus or CBR, the layer
Toronto Pearson International Airport and analyzed in the materials, and traffic, all of which were defined in the
Centre for Pavement and Transportation Technology previous sections. Table 5 presents the thickness of the
(CPATT) (9). The binder was a PG 70-28J, and the mix pavement layers for a design life of 20 years, as well as
binder content was 5.2%. The percentage air voids of the the modulus and Poisson ratio attributed in
mix were found to be 3.9% and the voids in mineral aggre- FAARFIELD.
gate (VMA) were 16.5%. The original binder viscosity at The pavement design in FAARFIELD shows the
77ºC was calculated as 0.93 cPoise. A summary of the input need for a more robust structure than that currently in
required to calculate the dynamic modulus of the mix is pre- place, since when maintaining the base and sub-base
sented in Table 4. In this table, f is the loading rate, Va is layer thickness constant, the AC was 179 mm thicker
the air void content, Vbeff is the effective binder content, than the present layer. The CDF results demonstrate
and r34 , r38 , r4 represent the cumulative percent retained that the B777-300 ER airplane contributes about 99% of
on standard sieves #3/4, #3/8, and #4, respectively, and r200 the rutting damage, while the B787-900 contributed 1%,
represents the percent passing through sieve #200. The gran- and the rest of the mix made a negligible contribution to
ular materials were assumed to have similar characteristics the rutting damage. Table 6 presents the cumulative rut-
to the standard materials from FAARFIELD, therefore, ting and fatigue CDF of the aircraft fleet mix from
The base characteristics were defined according to the P-209 FAARFIELD, and its respective pass to coverage ratio
FAA standard material and the sub-base according to the (the inverse of C/P).
P-154 FAA standard material.
Validation of Python Codes
FAARFIELD Design
The equations described in the methodology section to
The hot mix asphalt standard material in FAARFIELD calculate rutting and fatigue damage were implemented
is the P-401/403. This material has a fixed modulus value in Python codes, developed by the authors. To verify the
Barbi et al 125

Table 5. Pavement Design by FAARFIELD

No. Type Thickness (mm) Modulus (MPa) Poisson’s ratio

1 P-401 HMA surface 304 1721.00 0.35

2 P-209 Crushed aggregate 300 485.21 0.35
3 P-154 Uncrushed aggregate 900 138.61 0.35
4 Subgrade na 51.71 0.35

Note: FAARFIELD = FAA rigid and flexible iterative elastic layered design; HMA = Hot Mix Asphalt; na = not applicable.

Table 6. Cumulative CDF and P/C Ratio From FAARFIELD

Aircraft Rutting CDF contribution Rutting P/C ratio Fatigue CDF contribution Fatigue P/C ratio

A320-200 std 0 1.08 0.08 1.91

A321-100 std 0 1.08 0.09 1.84
B767-300 0 1.06 0.08 1.02
Bombardier CL-604/605 0 1.21 0.00 2.31
B777-300 ER 0.99 1.04 0.18 0.72
B787-9 0.01 1.03 0.16 1.14
B777-200 0 1.04 0.06 0.75
A330-300 std 0 1.03 0.03 1.06
B787-8 0 1.04 0.03 1.04
A310-300 0 1.08 0.01 0.95
B747-400 0 1.06 0.01 0.96
B747-400 Belly 0 1.06 0.01 0.96
A380-800 WV000 0 1.04 0.00 1.06
A380-800 WV000 Belly 0 1.03 0.01 0.78
DHC-7 0 1.22 0.01 2.33
B737-800 0 1.09 0.30 1.84
CRJ100/200 0 1.2 0.00 2.22

Note: CDF = cumulative damage factor; P/C = Pass to coverage; FAARFIELD = FAA rigid and flexible iterative elastic layered design.

damage codes, the authors calculated the CDF caused the simulated pavement elements which prevents it from
by the B777-300 ER under the same circumstances con- moving in the horizontal direction but leaves it free to
sidered in FAARFIELD. The first part of the verifica- move in the vertical direction. The bottom of the sub-
tion process was to certify that the P/C values for the grade element was constrained from any movement. This
B777-300 ER were the same as those presented in Table type of boundary condition is usually imposed to repre-
6 (i.e., 1.04 for rutting and 0.72 for fatigue). To that end, sent the confinement of the soil. Figure 3 shows the sub-
the aircraft wander was defined as 178.8 cm, the lateral grade divided in half for a better perspective (ZY view),
distance from the centerline of the runway to the center- with model contours/lines in deformed shape, colors rep-
line between the tires was considered to be 548.5 cm, and resenting the compressive strains scale.
other aircraft characteristics were defined according to The results of maximum compressive strains at the
Table 3, which resulted in P/C values of 1.036 and 0.716 top of the subgrade layer and maximum tensile strains at
for rutting and fatigue, respectively. The P/C values the bottom of the AC layer for both ABAQUS and
resulting from the Python codes, when rounded, are KENLAYER are presented in Table 7.
essentially equal to FAARFIELD. The results for compressive strains at the top of the
The same structural conditions defined in Table 5 were subgrade layer were very similar for both KENLAYER
used to obtain stresses and strains through KENLAYER and ABAQUS, with almost no difference between the
(circular tire contact) and ABAQUS (circular and ellipti- strain for circular and elliptical contact areas (0.15% dif-
cal tire contacts). The model in ABAQUS was built using ference). However, the tensile strains at the bottom of
an 8 cm mesh near the loads and a 30 cm mesh in the rest the AC layer exhibited a considerable difference of about
of the model; the contact between layers was defined as 28% between the two models with circular contact area
rough, and the model dimensions were 12 3 12 m wide, from KENLAYER and ABAQUS. The difference in ten-
by 13.504 m deep (the subgrade being 12 m deep). A sile strain values between the circular and elliptical foot-
boundary condition was defined at the lateral walls of prints in ABAQUS was about 2.5%.
126 Transportation Research Record 2677(8)

Figure 3. Compressive strains at subgrade in ABAQUS: (a) model contours and (b) compressive strains scale.

Table 7. Pavement Responses

Software Tire contact Compressive strain at subgrade top (micro-strains) Tensile strain at AC bottom (micro-strains)

KENLAYER Circular 1280.0 2379.9

ABAQUS Circular 1282.0 2295.9
ABAQUS Elliptical 1280.0 2288.6

Note: AC = asphalt concrete.

design period. The calculations implemented in Python

resulted in a damage of 0.90 for the same aircraft, that is,
a difference of 29%. The authors attribute the difference
in the results to the tire contact shape, which was assumed
to be circular instead of elliptical. Knowing the limitations
in duplicating FAARFIELD procedures, the codes imple-
mented in Python were considered satisfactory.

Climate Analysis
Figure 4. Rutting damage at validation phase. As temperature and saturation changes through the year,
so does the stiffness of the materials. The first step in
performing an analysis that considers climate variations
FAARFIELD uses LED to obtain the pavement criti- throughout the year is to estimate the pavement tem-
cal strains. Generally, when LED is used for flexible perature gradient along the pavement depth. To that
pavement design, it is assumed that each tire has a circu- end, hourly climate data from the past 10 years was
lar contact area, because of the axisymmetric nature of retrieved from the National Solar Radiation Database,
the LED. FAARFIELD, however, considers the contact for the years 2010 through 2019, and averaged. The data
area to be elliptical. Most of the existing LED-based pro- obtained included the air temperature, wind speed, solar
grams only allow for modeling a circular contact area, radiance and albedo at the Toronto Pearson Airport site.
and therefore KENLAYER was used in this study. The To determine the pavement temperature gradient, the
critical pavement responses were then used to estimate finite control volume method was used through TEMPS.
the number of coverages to failure. The CDF was calcu- The material properties necessary to calculate the
lated based on 68,027 B777-300 ER departures in a 20- pavement temperature profile include the conductivity
year design period. The results of the cumulative rutting and density, which were extracted from the existing liter-
damage over the design period can be seen in Figure 4. ature (25), as well as the material colors and specific heat
As previously mentioned, FAARFIELD attributed a capacity, which were defined according to typical values
cumulative damage of 0.99 for the B777-300 ER over the suggested by the TEMPS software. The pavement
Barbi et al 127

Table 8. Material Properties

Material type Identifier color Specific heat capacity (J/kg°K) Conductivity (W/m°K) Density (kg/m3)

AC Black 1298 1.08 2374

Base Gray 1038 3.63 2330
Sub-base Gray 1131 3.98 2446
Subgrade Gray 710 1.2 1550

Note: AC = asphalt concrete.

Table 9. Monthly Surface Albedo elapsed since thawing started (Dt) along the pavement
depth, and subsequently, to calculate the environmental
Month Albedo Month Albedo Month Albedo adjustment factor (Fenv ) per sub-season. Fenv multiplied
January 0.72 May 0.16 Sept 0.16 by the subgrade optimum resilient modulus resulted in
Feb 0.75 June 0.16 Oct 0.16 the seasonal subgrade resilient modulus. The resilient
March 0.41 July 0.17 Nov 0.27 modulus of the AC and subgrade per sub-season are pre-
April 0.20 August 0.16 Dec 0.51 sented in Table 10.
At this point, the critical pavement responses could be
calculated through KENLAYER, considering the change
in the resilient modulus of the AC and subgrade layers at
each sub-season. The responses enabled the calculation
of the number of coverages to failure, and consequently
the estimation of the cumulative rutting and fatigue dam-
ages. The same steps proposed in this section were fol-
lowed to evaluate five climate scenarios including: (1)
current climate; (2) temperature increase; (3) lower matric
suction; (4) flooding event 1, and (5) flooding event 2.

Climate Scenarios
Current Climate
The current climate scenario considers the hourly tem-
perature records from 2010 to 2019, and the matric suc-
tion as 293 kPa. A comparison of the rutting and fatigue
damage caused by a B777-300 ER is presented at Figures
Figure 5. Pavement temperature profile from temperature 6 and 7, considering the proposed methodology versus
estimate model for pavement structures (TEMPS). FAARFIELD. Figures 6a and 7a illustrate the damage
at the end of one year while Figures 6b and 7b show the
estimated damage at the end of 20 years.
material properties are presented in Table 8. The surface It is possible to notice from Figure 6a that the most
albedo was obtained through the NSRDB and averaged significant rutting damage occurs during sub-seasons 20
per month, as presented in Table 9. Lastly, emissivity to 30, which are equivalent to the months of May and
and absorption were both considered to be 0.93. June. The total damage caused by the B777-300 ER over
The pavement temperature gradient was generated the 20-year period is 1.55, compared to 0.99 from
based on a mesh of 1 cm. The results from TEMPS are FAARFIELD, therefore, the rutting damage predicted
presented in Figure 5. from the proposed methodology is 57% higher than the
Figure 5 indicates that the maximum temperature at a predicted damage from FAARFIELD. From the results
depth of 1 cm is 44°C and that the temperature at deeper of Figure 6b it is suggested that current climate assump-
layers changes at a lower rate than that at the upper tions result in a shorter service life than FAARFIELD
layers. The AC temperature was used to estimate the by about 7 years because of the increased rutting damage.
resilient modulus of the AC through the year using the In relation to the fatigue damage, Figure 7a shows
modified Witczak model (13). The pavement tempera- that the most significant damage occurs during sub-
ture profile was then used to obtain the number of hours seasons 20 to 45, which are equivalent to the months of
128 Transportation Research Record 2677(8)

Table 10. Resilient Modulus ( Mr ) of the Subgrade and Asphalt Concrete (AC) per Sub-Season.

Sub-season Month Period Mr subgrade (MPa) Mr AC (Mpa) Sub-season Month Period Mr subgrade (MPa) Mr AC (Mpa)

1 Jan 1 138 15,736 31 Jul 31 55 1026

2 Jan 2 138 16,694 32 Jul 32 59 1605
3 Jan 3 138 17,358 33 Jul 33 61 2005
4 Jan 4 138 18,022 34 Jul 34 63 2406
5 Jan 5 138 18,981 35 Jul 35 65 2985
6 Feb 6 138 16,687 36 Aug 36 68 1203
7 Feb 7 138 18,560 37 Aug 37 71 1781
8 Feb 8 138 19,857 38 Aug 38 74 2181
9 Feb 9 138 21,154 39 Aug 39 75 2581
10 Feb 10 138 23,028 40 Aug 40 76 3159
11 Mar 11 138 8701 41 Sep 41 77 1813
12 Mar 12 138 11,902 42 Sep 42 77 2538
13 Mar 13 138 14,118 43 Sep 43 77 3040
14 Mar 14 138 16,335 44 Sep 44 77 3542
15 Mar 15 138 19,535 45 Sep 45 77 4267
16 Apr 16 134 4636 46 Oct 46 77 3518
17 Apr 17 84 6816 47 Oct 47 77 4846
18 Apr 18 49 8326 48 Oct 48 77 5765
19 Apr 19 49 9836 49 Oct 49 77 6684
20 Apr 20 50 12,016 50 Oct 50 77 8012
21 May 21 34 2063 51 Nov 51 77 7973
22 May 22 37 3451 52 Nov 52 77 10,749
23 May 23 39 4413 53 Nov 53 77 12,671
24 May 24 40 5375 54 Nov 54 77 14,593
25 May 25 43 6763 55 Nov 55 77 17,368
26 Jun 26 46 1349 56 Dec 56 77 19,126
27 Jun 27 49 2204 57 Dec 57 77 21,828
28 Jun 28 51 2796 58 Dec 58 77 23,698
29 Jun 29 53 3387 59 Dec 59 77 25,569
30 Jun 30 55 4242 60 Dec 60 77 28,271

Figure 6. Cumulative rutting damage under current climate scenario: (a) over one year and (b) over twenty years.

May through to the middle of September. Figure 7b fatigue damage caused by the B777-300 ER over the 20-
shows that the total fatigue damage estimated through year period is 0.12, while in the proposed methodology it
the proposed methodology is lower than that proposed is 0.05068. This can be mainly attributed to two reasons:
by FAARFIELD; the software indicates that the total first, the amount of damage during the cooler seasons is
Barbi et al 129

Figure 7. Cumulative fatigue damage under current climate scenario: (a) over one year and (b) over twenty years.

Figure 8. Cumulative rutting damage under temperature increase scenario: (a) over one year and (b) over twenty years.

very low, considering that the pavement layers are frozen application called ‘‘Weather Morph: Climate Change
and, therefore, very stiff during this period; second, the Weather File Generator’’ can be used to generate hourly
proposed methodology was based on the specific charac- future weather data for many locations across the world
teristics of the asphalt mix, such as the percentage of air considering four gas emission scenarios in three time-hor-
voids and the effective asphalt content, which are super- izons: 2020s, 2050s and 2080s (26). The temperature
ior to the default values considered in FAARFIELD, increase scenario of this research used data available at
lowering the total fatigue damage. http://139.62.210.131/weatherGen/, under gas emission
scenario RCP 8.5 in the year 2050. The matric suction
was set at 293 kPa. Combining the proposed methodol-
Temperature Increase ogy with this hourly dataset, a comparison of the rutting
There are a few mathematical methods to project future and fatigue damage caused by the B777-300 ER can be
weather, among which the morphing method is one of made, which is presented in Figures 8 and 9, which show
the most well-known approaches. A web-based the proposed methodology versus FAARFIELD.
130 Transportation Research Record 2677(8)

Figure 9. Cumulative fatigue damage under temperature increase scenario: (a) over one year and (b) over twenty years.

As indicated by Figure 8, the total cumulative rutting lower matric suctions will lead to higher degree of
damage at the end of the 20 years would be 1.95 under saturation in the soil.
the temperature increase scenario, as compared to a dam- The effect of changes in matric suction on the expected
age value of 1.55 from the current climate scenario using pavement damage can be investigated through the use of
the proposed methodology, or 0.99 according to the tra- the proposed methodology based on the adjustment of
ditional FAARFIELD method. Therefore, when consid- the modulus values. Results for rutting and fatigue dam-
ering the potential increase in temperature because of age, based on B777-300 ER operations are therefore pre-
climate change, rutting damage can be 25% higher than sented in Figures 10 and 11.
the current climate scenario, and 97% higher than the By comparing the results from the previously dis-
conventional design based on FAARFIELD, which sug- cussed scenarios and the lower matric suction scenario,
gests a shortening of the service life of the flexible airfield as demonstrated by Figures 6a, 8a and 10a, it can be seen
pavement by about 10 years. that the high damage season was extended when satura-
Figure 9b shows that the total fatigue damage under tion levels were higher. The annual result in Figure 10a
the temperature increase scenario is 0.0723, compared to shows that high rutting damage levels occur in this case
0.05068 from the current climate scenario, or 0.12 through sub-seasons 20 to 35, which is equivalent to the
according to the FAARFIELD method. Even though months of May, June and July, that is, one month more
the total fatigue damage predicted by the proposed than that predicted by the current climate analysis.
methodology is still smaller than FAARFIELD, it Figure 10b shows that the total rutting damage at the end
increased by 43% compared to the current climate analy- of 20 years in this case was calculated to be 3.37, com-
sis. Therefore, the results from a potential rise in tem- pared to 1.55 from the current climate scenario and 0.99
perature, if not associated with changes in moisture from FAARFIELD. It can therefore be concluded that
levels or flooding events, will be more critical in relation increasing the subgrade soil saturation level because of
to fatigue damage. the lowering of the matric suction results in an increase in
rutting damage of 117% when compared to the current
climate scenario, and 240% when compared to the tradi-
Lower Matric Suction tional FAARFIELD analysis, resulting in a pavement
Under the lower matric suction scenario, past tempera- service life of only 6 years.
ture records were considered to be similar to the cur- The fatigue results from Figure 11, on the other hand,
rent climate scenario, while the matric suction was indicate a total fatigue damage of 0.05074, which is less
assumed to be 49 kPa. The matric suction is the soil than 1% higher than the results from the current climate
capillary pressure. The smaller the capillary pressure, scenario (i.e., 0.05068). However, it is still much lower
the higher the soil’s ability to retain water. Therefore, than the damage of 0.12 predicted by FAARFIELD.
Barbi et al 131

Figure 10. Cumulative rutting damage under lower matric suction scenario: (a) over one year and (b) over twenty years.

Figure 11. Cumulative fatigue damage under lower matric suction scenario: (a) over one year and (b) over twenty years.

Therefore, fatigue damage was not significantly influ- would be completely saturated because of intense precipi-
enced by the lowered subgrade soil stiffness because of tation. It is reasonable to assume that this hypothetical
the lowered matric suction. scenario occurs in the summer, starting on August 13th.
In the occasion of a flooding event, the soil would be
completely saturated, and the matric suction would equal
Flooding Events zero (0). As time goes by, the soil would drain, and the
It was demonstrated in the lower matric suction scenario matric suction would slowly recover to the equilibrium
that changes in the saturation level can affect the total level of 293 kPa. To evaluate the effects of the rate at
predicted damage over the service life of flexible airfield which the soil matric suction recover, over pavement
pavements. Therefore, under this scenario the effect of an damage after a flooding event, two scenarios were evalu-
extreme event such as flooding is further evaluated. This ated, one assuming the matric suction increase at 50 kPa/
analysis considered the additional damage from a flood- sub-season until it reaches 293 kPa, and another with the
ing event, by simulating a condition in which the subgrade matric suction increase at 25 kPa/sub-season.
132 Transportation Research Record 2677(8)

Figure 12. Cumulative rutting damage under flooding scenario 1: (a) over one year and (b) over twenty years.

Figure 13. Cumulative fatigue damage under flooding scenario 1: (a) over one year and (b) over twenty years.

In Scenario 1, the soil returns to its equilibrium matric than the rutting damage predicted by the traditional
suction in 36 days, and in Scenario 2, the equilibrium FAARFIELD method.
matric suction is reestablished in 72 days. The results The yearly fatigue damage, on the other hand,
from scenario 1 are presented in Figures 12 and 13. increased less than 1% when compared to the current cli-
The total rutting damage during the year in which mate scenario. It should be noted that this outcome may
flooding scenario 1 happened was found to be equal to be associated with the fact that this scenario only consid-
0.3, as presented in Figure 12a, which is equivalent to ered the weakening of the subgrade because of the flood-
about 6 years of damage under normal circumstances ing and did not account for any stiffness reduction in the
(CDF = 0.99 over 20 years). The total rutting damage base, sub-base, and AC layers. In an actual flood event,
over 20 years, presented in Figure 12b, was accounted at however, those layers may also get affected, which will
1.80. Therefore, the total rutting damage under the most likely increase fatigue damage as well as rutting.
flooding scenario 1 is 16% higher than the damage pre- The results from Scenario 2 are summarized by Figures
dicted by the current climate scenario and 82% higher 14 and 15.
Barbi et al 133

Figure 14. Cumulative rutting damage under flooding scenario 2: (a) over one year, and (b) over twenty years.

Figure 15. Cumulative fatigue damage under flooding scenario 2: (a) over one year, and (b) over twenty years.

In this scenario, the total rutting damage at the end of damage predicted by the current climate scenario. Such
year 1 was 0.43, as can be seen in Figure 14a, which is results are consistent with flooding scenario 1, in which
equivalent to 8.7 years of damage under normal circum- the flooding event presented a neglectable increment to
stances (CDF = 0.99 over 20 years). It can be noted in the fatigue CDF.
Figure 12b that the rutting damage over 20 years was
accounted at 1.90. The rutting damage under the flood-
ing event Scenario 2 is 22% higher than the damage
computed in the current climate scenario and 92%
Discussion
higher than the damage predicted by the traditional The cumulative rutting and fatigue damage, as well as
FAARFIELD method. The yearly fatigue damage under the predicted service life from all scenarios evaluated in
the flooding Scenario 2 was less than 1% higher than the this study is summarized in Table 11.
134 Transportation Research Record 2677(8)

Table 11. Damage and Predicted Service Life Summary

Fatigue Rutting
% Difference compared % Difference % Difference compared Predicted
to current climate Rutting compared to current climate service life
Climate scenarios Fatigue damage scenario damage to FAARFIELD scenario (years)

FAARFIELD 1.200E-01 na 0.99 na na 20.0

Current climate 5.068E-02 na 1.55 57 na 12.8
Temperature increase 7.235E-02 42.8 1.95 97 25 10.2
Lower matric suction 5.074E-02 0.1 3.37 240 117 5.9
Flooding event 1 5.070E-02 0.04 1.80 82 16 11.0
Flooding event 2 5.074E-02 0.1 1.90 92 22 10.4

Note: FAARFIELD = FAA rigid and flexible iterative elastic layered design; na = not applicable.

The fatigue damage estimated through the proposed pavement service life could be reduced to 11 years in
methodology in all the scenarios was found to be smaller flooding scenario 1 and 10.4 years in scenario 2, if no
than the total damage estimated from the traditional rehabilitation measures were taken after the event.
FAARFIELD analysis. This can be explained in light of The results of different climate change scenarios
FAARFIELD considering that the AC stiffness is con- showed that the greatest rutting damage occurred in the
stant throughout the year and equivalent to a tempera- lower matric suction scenario. Lower matric suction
ture of 32°C. However, the damage during the colder directly affects the equilibrium saturation level and low-
seasons was calculated to be much less since the pave- ers subgrade stiffness during both the recovery and
ment layers are frozen during this time. Among all the unfrozen seasons. The results showed that the decrease
climate change scenarios evaluated, fatigue damage in matric suction could increase damage by approxi-
seemed to be little affected by changes in soil saturation, mately 117% compared to the current climate scenario,
which is present in the lower matric suction and flooding and decrease the pavement service life to only a few
events scenarios. However, the effects of the temperature years.
increase scenario indicated fatigue damage that is 43% Given the results of analysis of the lower matric suc-
higher than that in the current climate scenario. tion scenario, it is important to recognize that matric
All the scenarios demonstrated greater rutting damage suction can vary though the year because of variations in
than predicted through FAARFIELD. The current cli- rainfall, evapotranspiration, ground water table levels,
mate scenario was based on past temperature data from and pumping of water from wells, and it is difficult to
2010 through 2019 and a matric suction of 293 kPa, and predict if suction will rise or decrease in certain regions
it resulted in rutting damage that was 57% greater than because of the effects of climate change. Although there
that estimated through FAARFIELD, and a reduction is strong evidence of rising temperatures and an increas-
in service life of 7 years. The effects of higher temperature ing incidence of extreme weather events such as flooding
under the temperature increase scenario suggested rutting in Canada, it would be hard to make predictions about
damage that was 25% greater than the damage predicted future matric suction levels. If an airport experiences a
by the current climate scenario and a service life that was significant rise in subgrade saturation levels in the future,
2.6 years shorter. By comparing the impact of tempera- the airport authorities would typically have to monitor
ture increase on rutting and fatigue damage, when the saturation levels and drain the excess water in a more
scenario is not associated with changes in moisture, it is efficient way, or enhance the pavement structure or
possible to conclude that a potential rise in temperature impose operational limitations for certain periods of
because of climate change can be more critical in relation time.
to fatigue damage. It is important to note that this sce- The fleet mix used in the analysis, although realistic
nario simulated temperature predictions corresponding for Toronto Pearson Airport, would not necessarily rep-
to the year 2050, with average temperatures of about 5°C resent other airports, where the amount and nature of
above the current scenario. traffic may be different. In the case of airports that do
Flooding scenarios 1 and 2 showed an increase in rut- not handle B777-300 ER operation, damage is more
ting damage of 16% and 22% when compared to the likely to be evenly distributed among the aircrafts in the
current climate analysis. The impact was greater for fleet mix. Although the finding that climate change can
flooding scenario 2 as it was based on a slower matric increase pavement damage and shorten service life is
suction recovery rate after flooding than scenario 1. The likely to remain relevant in the analysis of other airports,
Barbi et al 135

the percent increase in damage and the decrease in the Tavassoti; draft manuscript preparation: Paula Barbi, Pejoohan
service life of airport pavements will vary from one air- Tavassoti and Susan Tighe. All authors reviewed the results
port to another. and approved the final version of the manuscript.
It should be noted that this research only considered
the conventional flexible airport pavements (i.e., AC on Declaration of Conflicting Interests
top of unbonded granular layers) of Toronto Pearson
The author(s) declared no potential conflicts of interest with
International Airport in the analysis. The pavement dam- respect to the research, authorship, and/or publication of this
age results would likely change if different layer thick- article.
nesses, stabilized materials, or unconventional structures
were used. Furthermore, the traffic and fleet mix data
used in this investigation were pertinent to the selected Funding
airport facility, and may not represent traffic and fleet The author(s) received no financial support for the research,
mix data for other airports. authorship, and/or publication of this article.

Conclusions ORCID iDs

This research proposed a new methodology to analyze Paula Sutherland Rolim Barbi https://orcid.org/0000-0001-
6337-3523
the structural design of flexible airport pavements which
Pejoohan Tavassoti https://orcid.org/0000-0002-5118-4255
took into account climate-induced impacts on the pre-
dicted pavement performance. The methodology was
applied to the case of Pearson International Airport, References
using actual field data. To assess the possible effects of 1. Heymsfield, E., and J. S. Tingle. State of the Practice in
climatic parameters and climate change on airport pave- Pavement Structural Design/Analysis Codes Relevant to
ment performance, the methodology was first cross- Airfield Pavement Design. Engineering Failure Analysis,
verified by the design of a flexible pavement structure Vol. 105, 2019, pp. 12–24.
using FAARFIELD to ensure that a total CDF of 1 was 2. Whiteley, L. S. T. Pavement Thickness Design for Canadian
achieved. Then, the same layer thickness resulting from Airports. University of Waterloo, Waterloo, 2006.
the FAARFIELD design was used to evaluate current 3. Sun, L., and W. R. Hudson. Probabilistic Approaches for
Pavement Fatigue Cracking Prediction Based on Cumula-
climate and climate change effects on the CDF of the
tive Damage Using Miner’s Law. Journal of Engineering
most critical aircraft (i.e., B777-300 ER) from the traffic
Mechanics, Vol. 131, No. 5, 2005, pp. 546–549.
fleet mix operated at the airport. A total of five scenarios 4. Tamagusko, T., and A. Ferreira. Software Tools for Airport
were evaluated in the research, including current climate, Pavement Design. In A. Rocha, H. Adeli, L. Reis, S. Cost-
temperature increase, lower matric suction, and two anzo, I. Orovic, and F. Moreira (eds) World Conference on
flooding events. Information Systems and Technologies, Springer, Cham,
The results of the proposed design method showed 2020.
that the traditional FAARFIELD analysis can possibly 5. Meagher, W., J. S. Daniel, J. Jacobs, and E. Linder.
overestimate fatigue damage, and underestimate rutting Method for Evaluating Implications of Climate Change
damage. The results for fatigue and rutting damage from for Design and Performance of Flexible Pavements. Trans-
the proposed methodology show that climate change can portation Research Record: Journal of the Transportation
increase pavement damage and shorten the pavement Research Board, 2012. 2305: 111–120.
6. Swarna, S. T., K. Hossain, Y. A. Mehta, and A. Bernier.
service life by 7 to 14 years in the scenarios evaluated.
Climate Change Adaptation Strategies for Canadian
The proposed methodology can yield major benefits in Asphalt Pavements; Part 1: Adaptation Strategies. Journal
quantifying the impact of climate change, which can ulti- of Cleaner Production, Vol. 363, 2022, p. 132313.
mately facilitate the design of more resilient transporta- 7. Mills, B. N., S. L. Tighe, J. Andrey, J. T. Smith, and K.
tion infrastructure including airfield pavements. The Huen. Climate Change Implications for Flexible Pavement
proposed method takes account of climate variations, Design and Performance in Southern Canada. Journal of
temperature increase, and extreme events such as flood- Transportation Engineering, Vol. 135, No. 10, 2009,
ing in the design of flexible airport pavements. pp. 773–782.
8. Vanheule, H., D. Lu, D. Dutton, and S. Tighe. Summary
of Climate Change and Airfield Pavements Survey. In
Author Contributions CAPTG, Halifax, 2017.
The authors confirm contribution to the paper as follows: study 9. Abreu, E. A. Impacts of Climate Change on Canadian Air-
conception and design: Paula Barbi and Pejoohan Tavassoti; port Pavements. Thesis. University of Waterloo, 2019.
data collection: Paula Barbi and Pejoohan Tavassoti; analysis 10. Alavi, M. Z., M. R. Pouranian, and E. Y. Hajj. Prediction
and interpretation of results: Paula Barbi and Pejoohan of Asphalt Pavement Temperature Profile with Finite
136 Transportation Research Record 2677(8)

Control Volume Method. Transportation Research Record: 18. Tuleubekov, K. Replacement of FAARFIELD Tandem Fac-
Journal of the Transportation Research Board, 2014. 2456: tors With Cumulative Damage Factor Methodology. Federal
96–106. Aviation Administration (FAA), 2016.
11. Bari, J. Development of a New Revised Version of the Witc- 19. GTAA. GTAA Annual Report 2019. 2020. https://www.tor
zak E* Predictive Models for Hot Mix Asphalt Mixtures. ontopearson.com/ar2019/_downloads/GTAA-Annual-Rep
Arizona State University, 2005. ort-2019.pdf
12. Huang, Y. H. Pavement Analysis and Design, 2nd ed. Pear- 20. GTAA. Toronto Pearson International Airport Master Plan
son Prentice Hall, Upper Saddle River, NJ, 2004. 2017-2037. 2018. https://cdn.torontopearson.com/-/media/
13. Applied Research Associates, Inc., ERES Consultants project/pearson/content/corporate/our-future/pdfs/gtaa-mas-
Division. Guide for Mechanistic-Empirical Design of New ter-plan.pdf?rev=fe7a67440061412da99af754dbac1e9f&hash
and Rehabilitated Pavement Structures. NCHRP Project 1- =B552D38CDF594C6FDB6BF517E715C72E
37A. Transportation Research Board, National Research 21. Chee, K., and M. Ahmed. Fiber-Enhanced Asphalt at Tor-
Council, Washington, D.C., 2004. onto Pearson International Airport. 2018. http://www.onas-
14. Wang, Y., Y. Huang, W. Rattanachot, K. K. Lau, and S. phalt.org/files/Innovation%20in%20the%20Asphal-
Suwansawas. Improvement of Pavement Design and Man- t%20Industry%20-%20Kevin%20Chee%20and%20Mo-
agement for More Frequent Flooding Caused by Climate hammed%20Ahmed.pdf
Change. Advances in Structural Engineering, Vol. 18, No. 4, 22. Huang, Y. H. Pavement Analysis and Design. Prentice-Hall,
2015, pp. 487–496. Englewood Cliffs, NJ, 1993.
15. Kawa, I. Development of New Subgrade Failure Model for 23. Stewart, C. Best Practices in the Design and Construction
Flexible Pavements in FAARFIELD. Federal Aviation of Airfield Asphalt Pavements. In Canadian Airfield Pave-
Administration (FAA), 2017. ment Technical Group (CAPTG), Calgary, AB, 2010.
16. Loizos, A., A. Armeni, C. Plati, and B. Cliatt. New Chal- 24. Federal Aviation Administration. 150/5320-6F - Airport
lenges in Evaluating Bearing Capacity of Airfield Pave- Pavement Design and Evaluation. 2016.
ments. In Innovation and Sustainability in Highway and 25. Khan, A., D. Mrawira, and E. E. Hildebra. Use of Light-
Airfield Pavement Technology, American Society of Civil weight Aggregate to Mitigate Frost Damage in Flexible
Engineers, 2019, pp. 279–289. Pavements. International Journal of Pavement Engineering,
17. KAWA, I. Pass-to-Coverage Computation for Arbitrary Vol. 10, No. 5, 2009, pp. 329–339.
Gear Configurations in the FAARFIELD Program. DOT/ 26. Jiang, A., X. Liu, E. Czarnecki, and C. Zhang. Hourly
FAA/TC-TN12/47, Federal Aviation Administration Weather Data Projection Due to Climate Change for
(FAA), 2012. Impact Assessment on Building and Infrastructure. Sus-
tainable Cities and Society, Vol. 50, 2019, p. 101688.