![Anisotropic material properties for the CLT layers [6] 130](https://www.researchgate.net/profile/Erdem-Coleri/publication/335571994/figure/tbl1/AS:799007509082114@1567509643469/Anisotropic-material-properties-for-the-CLT-layers-6-130_Q320.jpg)
Content uploaded by Erdem Coleri
Author content
All content in this area was uploaded by Erdem Coleri on Sep 03, 2019
Content may be subject to copyright.
Haynes et al. 1
Selection of the Most Effective Pavement Surfacing Strategy for the Glenwood Cross
Laminated Timber Parking Garage
Matthew Haynesa*, Erdem Colerib, and Mostafa Estajic
a*(Corresponding Author) Graduate Research Assistant, Department of Civil & Construction
Engineering, Oregon State University, Oregon 97331, USA; Email: haynesm@oregonstate.edu
bAssistant Professor, Department of Civil & Construction Engineering, Oregon State University,
Oregon 97331, USA; Email: erdem.coleri@oregonstate.edu; Tel: +1-541-737-0944; Fax: +1-
541-737-3052
cGraduate Research Assistant, Department of Civil & Construction Engineering, Oregon State
University, Oregon 97331, USA; Email: estajim@oregonstate.edu
Abstract
The use of cross-laminated timber (CLT) is becoming increasingly common in large scale
construction projects as a sustainable alternative. This study assesses the effectiveness of several
overlay surfaces which are to be installed on top of floor slabs in a proposed CLT parking garage
structure. Overlay surfaces are necessary to protect the CLT from moisture and vehicle loadings
within the parking garage. First, finite element modeling was performed to determine stress, strain,
and deformation distributions and factors for use in the laboratory tests. Then, three overlay
systems were evaluated using laboratory tests (at the stress-strain levels identified by FE modeling)
where cracking resistance, interlayer bond strength and rutting (flow number) were the main
parameters used to quantify the performance of each overlay strategy. One of the three overlay
systems evaluated consisted of a flexible polyurethane membrane applied directly to a CLT panel.
The other two overlay systems consisted of dual-layered sheet membranes applied to a CLT panel
and then paved over with a modified asphalt pavement. Based on the laboratory test results, a
strategy with the asphalt and sheet membrane layers on top of the CLT layer was selected to be
the best option due to significant reduction in stress-strain distribution achieved when compared
to the other strategies.
Keywords: Cross laminated timber; asphalt concrete; performance testing; finite element modeling;
waterproofing membrane
Haynes et al. 2
1. Introduction
1
1.1 Scope and background information
2
Precast and cast-in-place reinforced concretes are the most common materials used in parking
3
garage construction. Steel structures are also commonly used for parking garages and generally
4
implement corrugated steel paneling with cast-in-place concrete to act as the floors and ceilings of
5
the structure. While its use has been primarily restricted to midsize structures, cross-laminated
6
timber (CLT) has been growing in popularity as a building material in the United States. Marketed
7
as being sustainable and efficient, CLT is starting to be implemented in larger structures such as
8
high rises, bridges and parking structures.
9
Cross-laminated timber is considered to be a sustainable building material and a renewable
10
resource if timber is harvested responsibly. Previous studies have shown that when considering
11
the production and use phase of a material, CLT consumes less energy and produces fewer CO2
12
emissions as compared to concrete. CLT can also provide energy as biofuel at the end of its service
13
life [1]. Despite its benefits, there are a number of challenges associated with the use of CLT as a
14
primary building material. Many contractors lack experience working with CLT, making
15
developers weary to use it for large scale constructions. Additionally, the wood industry in the US
16
has not been as active with funding for CLT related research as compared to other countries
17
currently implementing the material. Finally, the cost saving benefits associated with the use of
18
CLT are not fully know and may not be competitive enough to justify the risk involved with using
19
a novel material [2].
20
Haynes et al. 3
The Glenwood CLT Parking Garage, which is set to be constructed in Springfield, Oregon,
21
would be the first cross-laminated timber parking garage constructed in the United States. The
22
successful completion of this project would promote the feasibility of CLT as a sustainable
23
building material for large scale structures such as high rises and bridges. This study seeks to
24
identify overlay surfacing strategies which would be best suited to protecting the CLT floor panels
25
from vehicle loadings present in a parking garage structure. An ideal overlay surfacing has high
26
cracking and rutting resistance against vehicle loadings as well as a high bond strength to the CLT
27
panel.
28
In other parts of the world, asphalt pavements have commonly been used as an overlay for
29
wooden bridges. Mastic asphalt and rolled asphalt are two overlay strategies commonly used for
30
this purpose [3]. Mastic asphalt is composed of gravel, sand filler and bitumen. It has a high heat
31
resistance, flexibility and fatigue strength and is waterproof. Mastic asphalt must be placed at high
32
temperatures (up to 230°C), which can result in vapor pressure build-up from moisture in the
33
underlying layer and bubbles, or blisters, to form on the asphalt road surface if care is not taken.
34
Rolled asphalt is a high density, low air void asphalt mix which is placed at lower temperatures
35
than mastic asphalt. Because of this, rolled asphalt does not have issues with blistering, however
36
it is not a fully waterproof surfacing. A study by Müller and Scharmacher [3] used a shear test to
37
investigate factors affecting the shear strength of rolled and mastic asphalt constructed on a
38
laminated wooden slab. It was determined that the sealer used to adhere the asphalt layer to the
39
wooden slabs had a significant impact on sample shear strengths. Additionally, it was suggested
40
that the required compaction energy for typical rolled asphalt cannot always be applied on wooden
41
bridges, but modified binders could be used to improve compatibility and construction quality. For
42
this study, a specially modified asphalt mixture developed by Haynes et al. [4] was utilized for
43
Haynes et al. 4
some of the tested surfacing strategies. This mixture provided an overlay which achieved both
44
impermeability and a high compactability through a specialized dense gradation, high binder
45
content and a modified binder.
46
In this study, two types of overlay systems were analyzed in order to evaluate their
47
resistance to rutting and cracking when constructed over a wooden CLT panel and subjected to the
48
maximum vehicle loadings typically found in a parking garage. The overlay systems developed by
49
Company 1 and 2, were installed on a series of CLT block samples by construction companies
50
licensed by the overlay manufacturers. Overlay systems for Company 1 and Company 2 are
51
denoted as Strategy 1 and Strategy 2, respectively. First, a finite element model was developed to
52
determine the strains to be used in laboratory testing for both overlay strategies. Then, two
53
variations of the Company 2 strategy were tested- one having a lower quality but more cost-
54
effective membrane layer (Strategy 2a) and the other having a thicker and higher quality membrane
55
layer (Strategy 2b). For each overlay strategy, axial, torsional and bending tests were conducted to
56
evaluate cracking and rutting performance.
57
1.2 Evaluated surfacing overlays
58
The first overlay strategy (Strategy 1) evaluated in this study was a four-layer system which
59
consisted of a primer, membrane layer, basecoat and topcoat applied to the CLT block samples.
60
Each layer was poured onto the CLT samples and distributed using a roller or notched squeegee.
61
Aggregate was broadcast over the basecoat layer prior to the application of the topcoat in order to
62
create additional friction on the driving surface. The Strategy 1 system is a standalone system
63
commonly used in concrete parking garages and vehicular traffic can be allowed directly onto the
64
surfacing overlay (i.e. no asphalt was constructed on top of the overlay).
65
Haynes et al. 5
The second overlay strategy (Strategy 2) consisted of a fabric textile layer and two
66
membrane layers bolted to the CLT block samples using circular brackets. The fabric layer created
67
no bonding at the CLT- membrane interface, instead relying on the brackets to anchor the overlay
68
system. Two variations of the Strategy 2 system were tested, with the Strategy 2b variation having
69
a thicker and stronger bottom membrane layer than Strategy 2a. An asphalt overlay was then
70
compacted on top of the Strategy 2 membrane. The asphalt overlay used in this system is
71
significantly thinner than the asphalt thicknesses used for pavement constructions to avoid any
72
excessive dead loads on the CLT structure. Since the use of large aggregate sizes to construct thin
73
asphalt layers may cause compaction issues during construction, a modified high-density gradation
74
and mix design was used in this study to allow for adequate impermeability and compactability.
75
The mix uses a high binder content, modified binder and fine gradation to create a pavement which
76
is both impermeable (less than 1.0% air void content) and resistant to cracking and rutting damage
77
[4]. To achieve the mix design, a binder grade of PG76-22 was used along with a binder content
78
of 8.5%. The aggregate target gradation consisted of stockpile percentages of 0% coarse
79
aggregates, 44% medium aggregates, 36% fine aggregates and 20% recycled asphalt pavement
80
(RAP). This gradation for the specialized mix design is significantly finer than that of a typical
81
dense gradation, making it ideal for use with the thin asphalt layer constructed on top of the
82
Strategy 2 membranes. For both the Company 1 and Company 2 strategies, the CLT panels had
83
dimensions of 400mm x 260mm x 175mm during the time of overlay construction and were then
84
sawn as needed for each laboratory test.
85
2. Numerical Modeling
86
In this study, finite element (FE) modeling is used to simulate the vehicular loads in a parking
87
garage and to determine the impact of applied loads on the pavement structure. Model outputs
88
Haynes et al. 6
from the simulations were used to conduct the laboratory experiments. As expected, due to the
89
absence of a structural layer, stress and strain outputs for Strategy 1 were significantly higher than
90
those for Strategy 2a and Strategy 2b, respectively. The asphalt layer on top of the CLT layer for
91
the Strategy 2 option distributed the applied loads to a larger area and reduced the critical stresses
92
and strains experienced by the CLT layer. Model details, factorial, and the results are given below.
93
2.1 Model details and modeling factorial
94
Two vehicle types were considered to develop the moving load models. Figure 1 shows the loading
95
configuration of the car and F450 truck used in the factorial. Car model consisted of two single
96
axles spaced at a distance of 2.75 m from each other. The F450 is a gas-operated vehicle with dual
97
tires in the rear axle spaced 5.1 m away from the front single axle. The model for a traditional
98
semi-truck consists of a tandem axle with dual tires. The loads on the front and rear axles of the
99
car were 1039kg and 748kg, respectively. The loads on the front and rear axles of the F450 were
100
2,617kg and 4,123kg, respectively. Tire pressures for the car and the F450 were 262kPa and
101
814kPa, respectively.
102
103
(a)
104
105
(b)
106
Figure 1. Loading configuration (a) Car (b) F450
107
(The measurements are in cm)
108
Haynes et al. 7
A dynamic finite element (FE) model is developed to investigate the response of CLT slabs and
109
asphalt sections. AbaqusTM software was used for the development of the finite element models
110
[5]. In the developed viscoelastic FE model, only the linear behavior is considered (small strain
111
domain). No nonlinearity (fatigue, permanent deformations, and cracks) is taken into account.
112
Several factors were considered in the numerical factorial to assess the structural response under
113
various speed, temperature and load conditions. Table 1 shows the numerical factorial followed
114
for this study. Since speed and temperature are not expected to have a significant effect on the
115
CLT-only (Strategy 1) pavement system’s response due to the absence of the viscoelastic asphalt
116
layer, simulations for CLT only models were conducted with one speed and one temperature level.
117
For this reason, a total of 20 numerical models were developed.
118
Table 1. Numerical factorial for sections with asphalt layer
119
Factor
Subcategories
Load
Car, F450
Speed
5 mph, 15 mph
Temperature
10 ºC, 30 ºC
Layer
configuration
CLT, CLT+Thin (1.5 in) AC, CLT+Thick (3 in) AC
120
Each CLT slab is comprised of 5 wood layers with a certain alignment pattern matching the
121
alignment of the actual CLT samples that are planned to be used for the construction of the parking
122
garage. The first two and last two layers are aligned in longitudinal direction and the central layer
123
is laid in lateral direction. Each layer of CLTs are represented by a single plane of finite elements
124
with similar thickness. The interaction of CLT layers are considered as fully bonded. The CLT
125
slab is characterized by orthotropic material behavior. In other words, the material follows
126
anisotropic elasticity. In Abaqus, the anisotropic material behavior is defined in different directions
127
Haynes et al. 8
using distinctive local coordinate system for each layer. The anisotropic wood material parameters
128
measured by Paolini et al. [6] were used in this study for modeling (given in Table 2).
129
Table 2. Anisotropic material properties for the CLT layers [6]
130
Material Property
Parameter
Value
Young’s Modulus
11000.0 [N/mm2]
311.695 [N/mm2]
366.7 [N/mm2]
Shear Modulus
483 [N/mm2]
69 [N/mm2]
690 [N/mm2]
Poisson’s ratio
0.42
0.3
0.014
Density
455 [kg/m3]
* 1, 2, and 3 for moduli parameters denote the longitudinal, lateral and vertical alignments, respectively.
131
132
The tire is represented by a square stiff thin deformable body that moves over the surface of the
133
pavement (dynamic loading). The projected contact pressure from tires are applied on the
134
simplified square body. In order to simulate moving loads in the viscoelastic FE model, a
135
displacement boundary condition was applied to the loading body to simulate the constant-speed
136
passage of axles over the pavement section. The distribution of contact pressure on the tire is
137
assumed constant.
138
In order to characterize the viscoelastic behavior of the asphalt layer, the generalized
139
Maxwell model is used in this study to simulate the time dependency of asphalt materials used in
140
the surface layer. The model consists of two basic units, a linear elastic spring in series with
141
multiple Maxwell elements [7,8].
142
Mathematically, the behavior of this model follows Prony series and is described as:
143
Haynes et al. 9
(1)
Where is the equilibrium modulus of the elastic spring, and are the relaxation modulus
144
and relaxation time of the ith member among n Maxwell elements.
145
The viscoelastic behavior of the asphalt models were characterized using the falling weight
146
deflectometer (FWD) test (a nondestructive field test to determine material properties through
147
backcalculation) results from a pavement section in California. Using the relaxation modulus
148
outputs, the constants of Prony series were back-calculated and then imported into the FE models
149
as the inputs to model the asphalt behavior [7,8].
150
The temperature dependency of the asphalt mix is defined by using the Williams-Landel-
151
Ferry (WLF) equation, given as follows [9]:
152
(2)
where
153
aT = the time-temperature shift factor,
154
C1 and C2 = regression coefficients,
155
Tref = the reference temperature, and
156
T = test temperature.
157
158
To optimize the regression coefficients C1 and C2, the shear modulus data were first fitted to a
159
sigmoid function, in the form of:
160
logexp1
Glog
(3)
where
161
= regression coefficients, and
162
ζ=reduced time.
163
Haynes et al. 10
Shift factors are calculated by fitting the measured or back-calculated modulus to the sigmoidal
164
function (Eqn. 3). Finally, the dynamic modulus master curve for each section was obtained by
165
processing the laboratory dynamic modulus test results for the asphalt material used in this study.
166
2.2 Modeling results
167
Outputs for all the models are presented in this section. The critical vertical displacement and
168
vertical strain at the top of the pavement under the vehicle tire were used as the model outputs to
169
determine the laboratory bending beam fatigue test strain levels. Figure 2 shows the displacement
170
distribution on the surface of the pavement layer with a thin asphalt layer (Strategy 2) and without
171
a thin asphalt layer (Strategy 1). It can be observed that displacements experienced by the structure
172
will be about 4.3 times higher when an asphalt layer is not used. A typical highway deformation
173
under a heavy commercial truck (about 40 tons of weight) is around 0.2mm on a high performance
174
structure, while this number increases to about 0.5mm when the roadway structure is not strong
175
enough. A 3 mm deformation on the pavement surface for Strategy 1 can be considered to be
176
extremely high and can create damage in the CLT layer. However, it should be noted that the
177
purpose of this study is to only evaluate the performance of the surface layers (surface epoxy-sand
178
mixture for Strategy 1 and asphalt layer for Strategy 2). Evaluation of CLT performance under
179
vehicular loads is not within the scope of this study and additional analyses and testing are required
180
to evaluate the CLT performance.
181
182
183
184
185
Haynes et al. 11
(a)
(b)
Figure 2. Vertical displacement field along the pavement surface under an F450 vehicle (a)
186
CLT only (Strategy 1) (b) CLT+Thin AC (1.5 in)
187
188
Figure 3 shows the displacement outputs for all 20 cases. By comparing Figure 3a to Figure 3b
189
and Figure 3c, it can be observed that asphalt layer is distributing the vehicular loads to a larger
190
area and significantly reducing the critical displacements under car and F450 vehicles. Modeling
191
of the vertical strains also showed that critical vertical strains reduce significantly when an asphalt
192
layer is used as the surfacing (Strategy 2). For the CLT-only option, increasing speed increases the
193
displacements and strains. For the strategies with thin (1.5 inch) and thick (3 inch) asphalt layers,
194
increasing speed reduces the strains and displacements due to the viscoelastic nature of asphalt
195
(time-temperature dependency). Similarly, increasing temperature increases displacements and
196
strains for the sections with asphalt layers since asphalt gets softer with increasing temperature.
197
Haynes et al. 12
(a)
(b)
(c)
Figure 3. Critical vertical displacement outputs for all models (a) CLT-only model
198
(Strategy 1) (b) CLT+ Thick AC (3 in) (Strategy 2) (c) CLT+ Thin AC (1.5 in) (Strategy 2).
199
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
Car-5mph Car-15mph F450-5mph F450-15mph
Vertical displacement (mm)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
Vertical displacement (mm)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
Vertical displacement (mm)
Haynes et al. 13
Strain levels for the bending beam fatigue (BBF) tests were determined by using the critical
200
vertical strain outputs. Strain levels under the F450 vehicle for 5mph and 15mph and 10oC and
201
30oC were averaged to determine BBF test strain levels. Since BBF tests were conducted until the
202
number of repetitions reach 350,000, the strain level used for testing can be considered to be higher
203
than strain levels that will be experienced by the actual CLT parking garage structure. However,
204
using the high strain level for F450 vehicles is expected to result in a more conservative and
205
reliable evaluation. For BBF testing, the average test strain levels for the alternatives with and
206
without asphalt layers were determined to be 492 and 1,130, respectively.
207
3. Materials and Methods
208
3.1 Materials
209
This section provides information regarding the virgin binders, virgin aggregates and RAP material
210
used for the asphalt overlay with the Company 2 surfacing systems. Aggregates, RAP and binder
211
materials were obtained from local sources. Specific materials used for the overlay systems were
212
provided by the system manufacturers and are discussed in Section 3.2.1.
213
3.1.1 Aggregates
214
Virgin aggregates were obtained from a local quarry in Salem, Oregon. The virgin aggregates were
215
delivered in three gradations, namely coarse (1/2” to #4), medium (#4 to #8), and fine (#8 to zero).
216
To determine the gradation of each stockpiled aggregate, wet-sieve and dry-sieve analyses were
217
performed on multiple samples of each stockpile following AASHTO T 27-11 [10].
218
3.1.2 Recycled Asphalt Pavement (RAP)
219
RAP materials were sampled from a plant in Salem, Oregon. Gradation, binder content and
220
theoretical maximum specific gravity (Gmm) of RAP materials were provided by the plant.
221
AASHTO T 308-10 was followed to determine the RAP binder content [11]. The quantity of binder
222
Haynes et al. 14
in RAP materials was determined as 6.22%. AASHTO T 30-10 was followed to determine the
223
gradation of extracted RAP aggregates [12].
224
3.1.3 Binders
225
A local producer in Oregon provided the virgin modified PG76-22 binder for this study.
226
Temperature curves, mixing temperatures and compaction temperatures were provided by the
227
producer as well. Laboratory mixing and compaction temperatures were estimated by using the
228
viscosity-temperature lines following the procedure described in Asphalt Institute guidelines and
229
AASHTO T 316-11 [13,14].
230
Asphalt mixtures in this study used for the Strategy 2 CLT samples were prepared with an
231
8.5% binder content. This binder content is the percentage of the total binder by the weight of the
232
mix, which includes the recycled binder from RAP materials. In this study, it was assumed that all
233
the RAP binder was completely blended with the virgin binder (100 % blending).
234
3.2 Test Sample Preparation
235
3.2.1 Strategy 1 installation
236
When installing the Strategy 1 overlay system on CLT block panels, the CLT was first sanded to
237
remove surface debris and imperfections. A primer layer was then mixed using an electric mixer
238
and poured onto the CLT blocks. It was distributed using a roll-on brush to a target application
239
rate of 37.2 to 55.7 m2/unit of primer (Figure 4a). The primer was then allowed to cure for five
240
hours prior to applying the membrane layer. The polyurethane membrane layer is a waterproof,
241
highly flexible layer and provides cracking resistance to the overlay system. The membrane layer
242
was applied on top of the primer at a coverage rate of 23.2m2/unit. A 0.8mm notched squeegee
243
was used to distribute the membrane over the block samples. The membrane layer was then
244
allowed to cure for a minimum of 12 hours. A basecoat layer was poured over the membrane layer
245
Haynes et al. 15
and spread using a notched squeegee (Figure 4b). The target basecoat application rate and layer
246
thickness were 23.2m2/unit and 0.6mm, respectively. After allowing the basecoat to cure for 15
247
minutes, quartz sand was broadcast over the basecoat to refusal at an application rate of 3.1kg/m2
248
(Figure 4c). The sand was allowed to set for two hours and excess sand was removed using a leaf
249
blower. The basecoat layer and sand provided texture and skid resistance, and an additional layer
250
of impermeability to the system. The topcoat layer was then applied to the block sample using the
251
same method as previous layers (Figure 4d). The topcoat served to hold the quartz sand in place
252
and also acted as a barrier to prevent UV damage to the other layers. The target topcoat application
253
rate and layer thickness were 0.5m2/unit and 0.2mm, respectively. The topcoat reaches full cure
254
after 12 hours although the manufacturer’s recommend waiting period is 24 hours after the
255
application of the topcoat before subjecting the complete system to vehicle loadings. Figure 4
256
outlines the steps in the Strategy 1 surfacing installation process.
257
(a)
(b)
(c)
(d)
Figure 4. CLT block sample membrane installation (Strategy 1) (a) Primer applied to CLT
258
blocks (b) Membrane and basecoat layers applied on top of primer (c) Quartz sand
259
broadcast over basecoat layer (d) Topcoat applied using rollers over sand and basecoat.
260
261
3.2.2 Strategy 2 installation
262
When installing the Strategy 2 overlay systems on CLT block panels, the blocks were first swept
263
to remove surface debris. The cloth fleece reinforcing layer was then placed on the blocks and
264
stapled down to hold it in place (Figure 5a). Edges of the cloth were placed with a 5.1cm overlap
265
Haynes et al. 16
to ensure full coverage. One of two different base layer variations were then placed on top of the
266
fleece with a 15.2cm overlap at the edges (Figure 5b). Areas of overlap were welded together using
267
a torch to prevent water from infiltrating through the overlaps. A finishing spade was used to
268
smooth out welded edges. The base layer was held in place using circular seam plates screwed into
269
each corner of the CTL block. Strategy 2 membranes were not bonded to the CLT layer using any
270
adhesives to allow the CLT layer to breath during the use phase.
271
The only difference between Strategy 2a and Strategy 2b was the type of base layer used
272
in the system. The Strategy 2a layer was thinner and had less strength than the Strategy 2b layer,
273
however it was a lower cost option. The top layer was then rolled over the base layer. As the top
274
layer was rolled onto the base layer, the underside of the top layer was heated using a torch to
275
partially melt it and allow the top layer to adhere to the base (Figure 5c). In full scale applications,
276
the overlaps on the top layer are positioned between the overlaps on the base layer to further
277
prevent water infiltration. The coarse sand on top of the top layer provided UV protection until
278
asphalt was constructed on top of the membrane (Figure 5d).
279
280
(a)
(b)
(c)
(d)
Figure 5. CLT block sample membrane installation (Strategy 2) (a) Fleece fabric on CLT
281
block sample (b) Base layer installed on top of fleece (c) Top layer welded on top of base
282
layer (d) Completed block sample- Strategy 2 system.
283
284
285
286
287
Haynes et al. 17
3.2.3 Asphalt surface preparation for Strategy 2- Hydraulic Roller Compaction
288
The Strategy 2 overlay systems in this study utilized a thin asphalt overlay to protect the membrane
289
layers and distribute traffic loads. For Strategy 2 OFTT and BBF samples, a 22mm asphalt overlay
290
was compacted over the membrane layers. A 38mm asphalt overlay was used for FN samples. A
291
laboratory hydraulic roller compactor was used to compact asphalt on top of each Strategy 2
292
sample. For asphalt mixture preparation, aggregates and RAP were batched to meet the final
293
gradation obtained from the design method outlined in Section 1.2 [4]. Then, batched samples
294
were mixed according to the AASHTO T 312-12 procedure [15]. Two hours prior to compaction,
295
a high-performance tack coat was applied to the surface of the Strategy 2 membrane layer at an
296
application rate of 0.32L/m2 as recommended by Coleri et al. [16] (Figure 6a) Blocks were then
297
placed into the roller compactor mold. A mold with a height of 60mm was used for BBF samples,
298
whereas a 100mm mold was used for the preparation of OFTT and FN samples. Once the CLT
299
block was in place, the loose hot asphalt mix was poured into the mold and spread around such
300
that an even amount of asphalt was covering all parts of the CLT block (Figure 6b). Molds were
301
then placed in the hydraulic roller compactor. Compaction pressure was increased with each pass
302
of the hydraulic roller and compaction ceased once the mold height was reached (Figure 6c). After
303
compaction, the sample was allowed to cool overnight prior to cutting or coring (Figure 6d).
304
305
Haynes et al. 18
(a)
(b)
(c)
(d)
Figure 6. Asphalt compaction for Strategy 2 samples (a) Tack coat applied to Strategy 2
306
membrane layer and CLT block sample placed in roller compactor mold (b) Loose asphalt
307
poured into mold and evenly (c) Compaction of asphalt by roller compactor (d) Compacted
308
Strategy 2 CLT block sample.
309
310
3.3 Experimental Design
311
In this study, three different overlay strategies were evaluated for cracking and rutting
312
performance. The primary purpose was to evaluate the performance of these strategies when
313
constructed on top of a CLT block panel. Evaluation of the strategies would allow one to determine
314
an ideal strategy for use in conjunction with CLT. Overlay strategies were constructed on CLT
315
blocks by manufacturer representatives. Asphalt was compacted on top of Strategy 2 samples, and
316
all testing of prepared samples occurred at the Oregon State University Asphalt Materials
317
Performance Laboratory. Table 3 shows the experimental plan followed in this study. In order to
318
fulfil the above-mentioned objectives, specimens from three overlay strategies (Strategy 1,
319
Haynes et al. 19
Strategy 2a and Strategy 2b) were tested with Bending Beam Fatigue (BBF), Oregon Field Torque
320
Test (OFTT) and Flow Number (FN) tests to determine bending resistance (cracking), bond
321
strength, and deformation resistance (rutting), respectively. Sample preparation methods and
322
testing procedures for the tests conducted in this study are discussed in Section 3.4.
323
Table 3. Experimental plan for the evaluated strategies
324
Test type
Strategies
Temp.1
Repl.3
Total tests
BBF
Strategy 1
Strategy 2a
Strategy 2b
20 oC
6
18
OFTT
Strategy 1
Strategy 2a
Strategy 2b
RT2
6
18
FN
Strategy 1
Strategy 2a
Strategy 2b
35 oC
4
12
Note:
325
1 Temp. = Temperature
326
2 RT = Room Temperature (~25 oC)
327
3 Repl. = Replicate
328
329
3.4 Test Methods
330
3.4.1 Bending beam fatigue (BBF) test
331
The BBF test, or four-point bend test, is used to estimate fatigue life of pavement layers under
332
repeated traffic loading. In this test, failure is defined as the load cycle at which the specimen
333
undergoes a 50 percent reduction in stiffness relative to the initial stiffness. The BBF test uses a
334
software to continually log the stiffness of the beam using the parameters in Equation 4. This test
335
follows the AASHTO T321 specification [17]. The test setup is shown in Figure 7.
336
3.4.1.1 BBF sample preparation
337
For Strategy 1 samples, the surfacing overlay was applied by Company 1 and no asphalt layer was
338
compacted on top of the sample. For the Strategy 2 samples, two different interlayer strategies
339
were installed by Company 2, and asphalt layers with a thickness of 22 mm were compacted over
340
Haynes et al. 20
the interlayer membranes using a hydraulic roller compactor. The dimensions of the prepared slab
341
specimens were 400 mm (length) by 260 mm (width) by 57 mm (thickness). For both strategies,
342
prepared slabs were cut into beam specimens of 400 mm in length, 57 mm in height and 63 ± 6
343
mm in width using a high-precision saw. Two or three BBF replicate test specimens could be
344
obtained from one slab.
345
3.4.1.2 BBF testing procedure
346
Samples were then placed in the environmental chamber set at 20 ± 0.5ºC for 2 hours to ensure the
347
specimen was at test temperature prior to beginning the test. Clamps of the BBF device were raised
348
and the sample was slid into position. Once the specimen was placed, clamps were lowered and
349
the LVDT was initialized. The desired strain and loading frequency were entered into the test
350
software. In this study, 1,130 microstrain and a 2 Hz loading frequency (simulating 5 to 10 mph
351
speed for the parking garage) were selected for testing Strategy 1 beam samples based on the
352
results of the numerical simulations given in Section 2. For Strategy 2 beam samples, 492
353
microstrain and a 2 Hz loading frequency was used according to the results of the numerical
354
simulations. The test was terminated when the stiffness of the sample reduced to 50 percent of its
355
initial value or when the number of repetitions exceed 350,000. The percent reduction in stiffness
356
versus number of cycles was plotted and condition of the surface layers after testing were reported.
357
Flexural stiffness for each cycle is automatically calculated with software using the following
358
equation [18]:
359
360
Where;
361
Es is the flexural stiffness (Pa);
362
Haynes et al. 21
σt is the maximum tensile stress (Pa)
363
εt is the maximum tensile strain (m/m)
364
a is the space between inside clamps (0.119 m)
365
P is the applied load (N)
366
b is the average beam width (m)
367
h is the average beam height (m), and
368
L is the beam length between outside clamps (0.357 m)
369
370
(a)
(b)
Figure 7. Bending beam fatigue test; (a) Strategy 2 beam specimens prepared with roller
371
compactor; (b) four-point bending test apparatus
372
373
3.4.2 Oregon field torque test (OFTT)
374
For the research conducted, the bond strength at the interface between the CLT and the surfacing
375
layer was determined using the OFTT in-situ torque tester developed by Coleri et al. [16]. The
376
hardware of the device consists of an automatic step motor, planetary gearbox, transducers, torque
377
sensor and amplifier, data acquisition and control systems, and an adjustable frame as shown in
378
Figure 8. Error! Reference source not found. A specialized software was developed to control
379
the loading and rotation speed of the OFTT system. The software automatically logs the applied
380
torque over the duration of the test. Previous research has suggested that shear, not tension, is the
381
primary mechanism controlling bond failures in the field [19]. A study by Coleri et al. also
382
Haynes et al. 22
determined that test results from the OFTT conducted on asphalt samples had a very strong
383
correlation to interlayer shear strengths obtained from laboratory samples [16].
384
3.4.2.1 OFTT sample preparation
385
After preparing full block samples using the methods described in Section 3.2, Strategy 1 and
386
Strategy 2 samples were cored just below the surface of the CLT. For Strategy 2 samples, the cores
387
were glued back into core holes using a fast setting epoxy since the membrane for the Strategy 2
388
system was not bonded to the CLT layer (in order to allow the CLT layer to breath after
389
construction). Platens were glued to the surface of the core using a fast setting epoxy.
390
3.4.2.2 OFTT testing procedure
391
The OFTT test system was then placed on the platens and the torque frame was adjusted to match
392
the platens (Figure 8a). Using the control software, the torque transducer was rotated at 2o per
393
second until the bond was completely broken (Figure 8b). The OFTT experiment was performed
394
to measure the peak torque stress (strength) at the interface between pavement layers. Using the
395
rotation angle versus applied torque, torque strength was determined. Applied torque was
396
continually measured at 0.02 second intervals. Applied torque over the duration of the test was
397
plotted in a spreadsheet, allowing the peak torque strength at interlayer bond failure to be easily
398
identified. After obtaining the peak torque strength for each sample, the measured torque strength
399
(Nm) was converted to OFTT shear strength (kPa) using the equation given below [20]:
400
(5)
Where;
401
τ is the interlayer shear strength (OFTT shear strength) (kPa);
402
M is the peak torque at failure (N.m), and
403
D is the diameter of the core (mm).
404
405
Haynes et al. 23
(a)
(b)
Figure 8. OFTT setup (a) OFTT positioned over top of platen (b) Transducer rotated until
406
breaking of bond.
407
3.4.3 Axial load test- Flow number (FN)
408
The flow number (FN) test is a performance test for evaluating rutting (permanent deformation)
409
resistance of asphalt concrete mixtures [21]. In this test, while a constant deviator stress is applied
410
at each load cycle on the test sample, the permanent strain at each cycle is measured. Permanent
411
deformation of asphalt pavements has three stages: 1) primary or initial consolidation, 2)
412
secondary and 3) tertiary or shear deformation [22]. FN is taken as the loading cycle at which the
413
tertiary stage begins following the secondary stage.
414
3.4.3.1 FN sample preparation
415
Full block samples were prepared using the methods described in Section 3.2. Block samples
416
were then cut in half using a high-precision saw to produce FN samples with dimensions of 200
417
mm (length) by 260 mm (width) by 57 mm (height). Two FN samples could be obtained from
418
one full block sample.
419
3.4.3.2 FN testing procedure
420
In this study, testing conditions and criteria for FN testing described in AASHTO TP 79-15 for
421
unconfined tests were followed [23]. The test temperature was predicted based on pavement
422
temperature in a shaded area (i.e. a parking garage) for cities in Oregon with high populations and
423
Haynes et al. 24
at a depth of 20 mm (0.79 in) for surface courses [24]. Tests were conducted at a temperature of
424
35°C with an average deviator stress of 600 kPa and minimum (contact) axial stress of 30 kPa. For
425
conditioning, samples were kept in a conditioning chamber at the testing temperature for 12 hours
426
prior to testing. To calculate FN in this study, the Francken model was used [25]. The application
427
of this model as a rutting indicator was validated by Dongre et al. [26] through correlation of FN
428
to field data. Coleri et al. [27] developed a code which was used in this study to analyze data and
429
determine the regression coefficients of the Francken model to determine the FN. AASHTO TP
430
79-15 provides recommendations for minimum FN values for different traffic levels [23]. Based
431
on these recommendations, a mixture with a minimum FN of 740 will have adequate rutting
432
resistance.
433
4. Results and Discussion
434
4.1 Bending Beam Fatigue (BBF) Test
435
Beam samples for each overlay strategy constructed on CLT were tested using strain levels
436
equivalent to those produced by a Ford F450- the design vehicle for the parking garage structure.
437
Strains were obtained through finite element modeling (See Section 2). Average reductions in the
438
initial stiffness for each overlay strategy can be seen in Figure 9. After 350,000 cycle repetitions,
439
none of the samples tested using the BBF test reached 50% of their initial stiffness. Thus,
440
comparisons were made by plotting the percent reduction in initial stiffness over the course of a
441
350,000-cycle analysis period. At the conclusion of the tests, none of the tested samples showed
442
signs of cracking, deformation, or delamination between the CLT and overlay. The results from
443
the BBF tests for each strategy are presented below.
444
Beam samples with a Strategy 1 overlay surfacing saw the lowest reduction in initial
445
stiffness. On average, the Strategy 1 samples had an initial stiffness of 7,745.2 MPa. At the end of
446
Haynes et al. 25
the analysis period, several Strategy 1 samples saw no significant reduction in stiffness, with the
447
highest observed reduction being a 7.9% reduction in initial stiffness. The average stiffness of
448
Strategy 1 samples after 350,000 cycles was 96.7% of the initial stiffness, corresponding to a
449
reduction in stiffness of 3.3%.
450
Beam samples for Strategy 2a (thin interlayer) saw the highest reduction in initial stiffness
451
as compared to the Strategy 1 and Strategy 2b (thick interlayer) samples. On average, Strategy 2a
452
samples had an initial stiffness of 2695.8 MPa. Stiffnesses for the Strategy 2a beam samples after
453
350,000 test cycles ranged from 90.5% to 98.7% of the initial stiffness. The average stiffness at
454
the end of the analysis period was 94.8% of the initial stiffness, which corresponded to a reduction
455
in stiffness of 5.2%. On average, the Strategy 2b samples had an initial stiffness of 2355.4 MPa.
456
Stiffnesses for the Strategy 2b beam samples after 350,000 test cycles ranged from 92.4% to 99.0%
457
of the initial stiffness. The average stiffness at the end of the analysis period was 95.5% of the
458
initial stiffness, which corresponded to a 4.5% decrease in stiffness over the test duration.
459
In all cases, it was observed that all strategies are highly resistant to fatigue cracking.
460
Haynes et al. 26
461
Figure 9- Average BBF results for percent reduction in initial stiffness (error bars
462
represent one standard deviation)
463
464
465
466
467
4.2 Oregon Field Torque Test (OFTT)
468
The Oregon Field Torque Test was used to evaluate the interlayer shear strengths of the three
469
overlay strategies. For the Strategy 1 samples, the bonds of particular interest were the bonds at
470
the primer-membrane interface, membrane-basecoat interface and basecoat-topcoat interface. For
471
Strategy 2 samples, the bonds of interest were at the interface between the base membrane and the
472
top membrane, and at the interface between the top membrane and the asphalt overlay. The peak
473
torque obtained from the OFTT was used to calculate the interlayer shear strength of tested
474
strategies using Equation (5) [20]. Tests were run at room temperature where asphalt is quite stiff,
475
and sides and tops of cores for Strategy 2 tests were covered with epoxy, thus very little shear
476
deformation was expected to occur in the asphalt itself due to the applied torque from the test.
477
Additionally, the bond strength of the membrane layer was much weaker than the internal shear
478
Haynes et al. 27
strength of asphalt. These factors allowed one to assume that any amount of shear deformation
479
occurring in the asphalt during the test was minimal and did not significantly impact the test results.
480
Figure 100 shows the interlayer shear strengths obtained from the Strategy 1 OFTT tests.
481
Shear strengths obtained for the Strategy 1 system were consistent for all six tests. The average
482
interlayer shear strength for the Strategy 1 samples was 1123.5 kPa. When the sample replicates
483
did fail, the failure occurred at the membrane-basecoat interface (Figure 111a and Figure 11b).
484
Since the membrane layer required a 12-hour cure time, it was fully hardened prior to the basecoat
485
being applied, whereas the topcoat was applied while the basecoat was still slightly wet, possibly
486
resulting in better adhesion. Additionally, the sand broadcast on top of the basecoat likely
487
improved the bond strength between basecoat and topcoat layers. These factors may have
488
contributed to the failure occurring at the membrane-basecoat interface.
489
490
Figure 10. Strategy 1 OFTT interlayer shear strengths
491
Strategy 2a possessed the weaker of the two inner membranes. When conducting OFTT tests on
492
six replicate samples, two of the samples failed at the membrane-membrane interface, however the
493
Haynes et al. 28
remaining four samples failed at the epoxy-membrane. Thus, the average interlayer shear strength
494
was obtained using only Replicates 1 and 2. From the OFTT tests, it was determined that the
495
average interlayer shear strength for Strategy 2a was 328.9 kPa. One test which failed from the
496
membrane-membrane interface can be seen in Figure 11c. Figure 11d shows a Strategy 2a test
497
which failed from the epoxy-bottom membrane interface. Because the epoxy was applied
498
specifically for this test and served only to bond the membrane to the CLT, no conclusions could
499
be drawn regarding membrane shear strengths when the failures occurred at the epoxy interface.
500
In several tests which failed from the epoxy interface, it was observed that the membrane-
501
membrane interface was also partially delaminated (Figure 11d). The underside of the bottom
502
membrane had very little texture where the epoxy was applied as compared to the Strategy 1
503
samples or a traditional asphalt sample. This lack of texture likely created a poor bond with the
504
epoxy and resulted in the failure at the membrane-epoxy interface.
505
The same conclusions were derived after testing Strategy 2b, which possessed the stronger
506
of the two inner membranes. When conducting the OFTT tests on six replicate samples, all six
507
samples failed at the epoxy-membrane interface. Like the Strategy 2a samples, there was very low
508
texture on the bottom of the Strategy 2b base membrane, producing a poor bond with the epoxy
509
and resulting in early failures at the epoxy-membrane interface. After testing the first replicate
510
sample, texture was artificially created in the membrane to improve bond strength with the epoxy.
511
Despite some increase in bond strength, failures for the remaining replicates still occurred at the
512
epoxy-membrane interface. A stronger epoxy or a method of increasing texture on the bottom
513
membrane and CLT panel should be investigated to properly evaluate the interlayer shear strength
514
of the Company 2 surfacing strategies.
515
516
Haynes et al. 29
(a)
(b)
(c)
(d)
Figure 11. OFTT tested samples (a) Strategy 1 samples (b) Strategy 1 sample broken at
517
membrane-basecoat interface (c) Strategy 2a sample broken at membrane-membrane
518
interface (d) Partial delamination at membrane-membrane interface of Strategy 2a.
519
520
4.3 Axial Load Test- Flow Number (FN)
521
Flow number tests were conducted for the three overlay strategies. Based on the AASHTO TP 79-
522
13 recommended criteria for a traffic level between 10 and 30 million ESALs, tests which
523
produced a flow number less than 190 were to be considered failing. For the highest traffic levels
524
(greater than or equal to 30 million ESALs), a flow number less than 740 was to be considered
525
failing [23]. From the results, it was observed that all three of the overlay strategies met the flow
526
number criteria for all traffic levels. These results are outlined in Figure 12. Strategy 2a and
527
Strategy 2b samples had the highest FN values, at 8,502 and 8,375, respectively. Strategy 1 had a
528
lower average FN value of 4,689, however this was still significantly higher than the failure criteria
529
for the highest traffic level. Following the tests, no observable deformation could be seen in either
530
the Strategy 1 or Strategy 2 samples.
531
Haynes et al. 30
532
Figure 12. Average Flow Number results for all strategies (error bars represent two
533
standard deviations)
534
535
536
5. Summary, Conclusions and Recommendations
537
The goal of this study was to assess the viability of constructing several overlay surfaces on top of
538
a CLT slab to be used in the construction of a parking garage structure. Material properties and
539
characteristics were evaluated using laboratory tests for one overlay system produced by Company
540
1 and two overlay systems produced by Company 2. Stiffness, interlayer shear strength and flow
541
number were the main parameters used to quantify the performance of each overlay strategy.
542
Additionally, finite element modeling was performed to determine factors for use in the laboratory
543
tests that were specific to the location and smaller load levels experienced by a parking garage as
544
compared to a traditional highway. Bending beam fatigue (BBF) tests, Oregon Field Torque Tests
545
(OFTT), and flow number (FN) tests were used to evaluate the bending resistance, bond strength
546
and deformation resistance, respectively, of each overlay strategy.
547
548
Haynes et al. 31
A summary of the results from this study are presented as follows:
549
During OFTT tests, Strategy 1 samples consistently failed at the membrane-basecoat interface
550
at an average interlayer shear strength of 1,123.5 kPa. This high average bond strength for
551
Strategy 1 suggested that it is unlikely to have delamination issues under car and F450 loads
552
for this strategy.
553
Two Strategy 2a samples failed at the membrane-membrane interface at an average interlayer
554
shear strength of 328.9 kPa. Remaining OFTT samples failed at the epoxy-membrane interface
555
due to a smooth membrane surface resulting in poor bonding with the epoxy. Artificially
556
creating texture on the bottom membrane slightly increased the bond strength with the epoxy.
557
When constructed on top of a CLT panel, Strategy 1 produced an average flow number much
558
higher (4,689) than the minimum required flow number (740) for traffic levels greater than 30
559
million ESALs, indicating that rutting is not a concern for Strategy 1.
560
When constructed on top of a CLT panel and used in conjunction with the modified BDWSC
561
asphalt overlay, Strategy 2a and Strategy 2b produced average flow numbers much higher
562
(8,502 and 8,375, respectively) than the minimum required flow number (740) for traffic levels
563
greater than 30 million ESALs, indicating that rutting is not a concern for either strategy.
564
565
The major conclusions drawn from the results of this study are as follows:
566
From BBF testing, very small reduction in stiffness was observed for both Strategy 1 and
567
Strategy 2 samples, suggesting that the strains created by the Ford F450 design vehicle are not
568
sufficient to produce significant fatigue damage in any of the overlay structures.
569
Having an asphalt layer on top of the CLT layer (Strategy 2) creates a significant reduction in
570
surface deformation and strain since the asphalt layer distributes the load from heavy vehicles
571
Haynes et al. 32
more. This reduced surface deformation creates a better protection for the CLT layer. However,
572
it should be noted that CLT cracking resistance under vehicular loads was not investigated in
573
this study.
574
Strategy 2 membranes were not bonded to the CLT layer using any adhesives to allow the CLT
575
layer to breath during the use phase. It should be noted that although having a cloth fleece on
576
the CLT layer and allowing the CLT layer to breath is an effective method to maintain the
577
performance of the CLT layer, having independent layers and not having a bonded-monolithic
578
pavement structure can result in early asphalt layer failures. For this reason, bonding the
579
Strategy 2 membranes to CLT by using heat-activated adhesives is recommended.
580
Since FE modeling suggested significantly lower deformation and strain levels on the CLT
581
layer for Strategy 2 and laboratory testing indicated high rutting and cracking resistance for all
582
strategies, Strategy 2 is recommended as a more effective surfacing strategy for a CLT parking
583
garage. Significantly lower strain levels observed by the CLT layer is expected to create better
584
longevity for the entire CLT parking garage structure.
585
586
Acknowledgements
587
This study was supported and funded by the U.S. Economic Development Administration as a part
588
of the TallWood Design Institute’s Oregon State Engineered Wood Building Products
589
Commercialization Project (Grant Number EDA X0188A). The funding received for this project
590
is gratefully acknowledged. The contents of this paper reflect the views of the authors and do not
591
reflect the official views or policies of the TallWood Design Institute. The authors thank the
592
members of the TallWood Design Institute (Juliana Ruble, Lech Muszynski, Iain Macdonald, and
593
Frederik Laleicke) for their advice and assistance. The authors would also like to thank John
594
Haynes et al. 33
Mikkelson for his assistance with cutting CLT panels, and Jacob Virell, Erick Moreno and
595
Jonathon Schwartz for their assistance with sample procurement.
596
References
597
1. Liu, Y., H. Guo, C. Sun, W. Chang. Assessing Cross Laminated Timber (CLT) as an
598
Alternative Material for Mid-Rise Residential Buildings in Cold Regions in China – A
599
Life-Cycle Assessment Approach. Sustainability, 8(10), 2016.
600
2. Pei, S., J. Berman, D. Dolan, J. Lindt, J. Ricles, R. Sause, H. Blomgren, M. Popovski, D.
601
Rammer. Progress on the Development of Seismic Resilient Tall CLT Buildings in the
602
Pacific Northwest. World Conference on Timber Engineering, Quebec City, Canada, 2014.
603
3. Müller, A., F. Scharmacher. Asphalt Pavements for Road Bridges Made of Wood. Bern
604
University of Applied Sciences, 2012 (in German).
605
4. Haynes, M., E. Coleri, S. Sreedhar. Impermeable Asphalt Concrete Layer to Protect and
606
Seal Concrete Bridge Decks. Transportation Research Record: Journal of the
607
Transportation Research Board, 2015.
608
5. Abaqus 6.14/CAE User’s Manual and Documentation. ABAQUS, Dassault Systems,
609
Vélizy-Villacoublay, France, 2014.
610
6. Paolini, A., S. Kollmannsberger, C. Winter. A High-Order Finite Element Model for
611
Vibration Analysis of Cross-Laminated Timber Assemblies. Building Acoustics, 24(3),
612
2017.
613
7. Coleri, E., J.T. Harvey. A Fully Heterogeneous Viscoelastic Finite Element Model for Full-
614
scale Accelerated Pavement Testing. Construction and Building Materials, 23, 2013, pp.
615
14-30.
616
8. Coleri E., J.T. Harvey. Impact of Pavement Structural Response on Vehicle Fuel
617
Consumption. Journal of Transportation Engineering, Part B: Pavements, 24(3), 2017, pp.
618
135-58.
619
9. Ferry, J. D. Viscoelastic Properties of Polymers, John Wiley & Sons, Inc., New York, 1980
620
10. AASHTO T 27-11. Standard Method of Test for Sieve Analysis of Fine and Coarse
621
Aggregates. Washington, DC: American Association of State Highway and
622
Transportation Officials, 2011.
623
11. AASHTO T 308-10. Determining the Asphalt Binder Content of Hot Mix Asphalt (HMA)
624
by the Ignition Method. Washington, DC: American Association of State and Highway
625
Transportation Officials, 2010.
626
12. AASHTO T 30-10. Standard Method of Test for Mechanical Analysis of Extracted
627
Aggregate. Washington, DC: American Association of State Highway and Transportation
628
Officials, 2010.
629
13. Asphalt Institute Executive Offices and Research Center. Determining Laboratory Mixing
630
and Compaction Temperatures. 2016.
631
14. AASHTO T 316-11. Viscosity Determination of Asphalt Binder Using Rotational
632
Viscometer. Washington, DC: American Association of State and Highway Transportation
633
Officials, 2011.
634
15. AASHTO T 312-12. Preparing and Determining the Density of Hot Mix Asphalt (HMA)
635
Specimens by Means of the Superpave Gyratory Compactor. Washington, DC: American
636
Association of State Highway and Transportation Officials, 2012.
637
Haynes et al. 34
16. Coleri, E., D. Covey, A. Mahmoud, J. Batti, N. Anisimova. HMAC Layer Adhesion
638
through Tack Coat. Final Report-SPR782. FHWA-OR-RD-17-05. Salem, OR: Oregon
639
Department of Transportation, Research Section, 2017.
640
17. AASHTO T 321-17. Standard Method of Test for Determining the Fatigue Life of
641
Compacted Asphalt Mixtures Subjected to Repeated Flexural Bending. Washington, DC:
642
American Association of State Highway and Transportation Officials, 2017.
643
18. Kallas, B., V. Puzinauskas. Flexural Fatigue Tests on Asphalt Paving Mixtures. American
644
Society for Testing and Materials, 1972, pp. 47-65.
645
19. Mohammad, L.N., M.A. Elseifi, A. Bae, N. Patel, J. Button, and J.A. Scherocman.
646
Optimization of Tack Coat for HMA Placement. In National Cooperative Highway
647
Research Program, No. NCHRP 712. National Cooperative Highway Research Program
648
(NHCRP), Transportation Research Board, National Research Council, Washington, D.C.,
649
2012.
650
20. Muslich, S. Assessment of Bond Between Asphalt Layers. The Univeristy of Nottingham.
651
2009.
652
21. Bonaquist, R.F., D. W. Christensen, W. Stump. Simple Performance Tester for Superpave
653
Mix Design: First-Article Development and Evaluation. Transportation Research Board,
654
2003, Vol. 513.
655
22. Biligiri, K., K. Kaloush, M. Mamlouk, M. Witczak. Rational Modeling of Tertiary Flow
656
for Asphalt Mixtures. Transportation Research Record: Journal of the Transportation
657
Research Board 2001, 2007, pp.63-72.
658
23. AASHTO TP 79-15. Standard Methods of Test for Determining the Dynamic Modulus and
659
Flow Number for Asphalt Mixtures Using the Asphalt Mixture Performance Tester
660
(AMPT). Washington, DC: American Association of State Highway and Transportation
661
Officials, 2015.
662
24. Rodezno, M.C., R. West, A. Taylor. Flow Number Test and Assessment of AASHTO TP
663
79-13 Rutting Criteria: Comparison of Rutting Performance of Hot-Mix and Warm-Mix
664
Asphalt Mixtures. Transportation Research Record: Journal of the Transportation
665
Research Board, (2507), 2015, pp.100-107.
666
25. Francken, L. Pavement Deformation Law of Bituminous Road Mixes in Repeated Load
667
Triaxial Compression. Presented at Fourth International Conference on the Structural
668
Design of Asphalt Pavements, University of Michigan, Ann Arbor, 1977.
669
26. Dongré, R., J. D'Angelo, A. Copeland. Refinement of Flow Number as Determined by
670
Asphalt Mixture Performance Tester: Use in Routine Quality Control—Quality Assurance
671
Practice. Transportation Research Record: Journal of the Transportation Research Board,
672
(2127), 2009, pp.127-136.
673
27. Coleri, E., S. Sreedhar, S. Haddadi, B. Wruck. Adjusting Asphalt Mixes for Increased
674
Durability and Implementation of a Performance Tester to Evaluate Fatigue Cracking of
675
Asphalt Concrete. Final Report. SPR 785, Draft report for Oregon Department of
676
Transportation, Salem, Oregon, (under review by TAC), 2017a.
677
678
