stergios mavromatis, assistant professorstergios mavromatis, assistant professor technological...
TRANSCRIPT
![Page 1: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/1.jpg)
Stergios Mavromatis, Assistant Professor
Technological Educational Institute of Athens
Eleonora Papadimitriou PhD, Senior Researcher
National Technical University of Athens
George Yannis, Professor
National Technical University of Athens
Basil Psarianos, Professor
National Technical University of Athens
![Page 2: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/2.jpg)
2 Independent and Mostly Uncorrelated 2D Stages
horizontal alignment
vertical alignment
2D Approach Associated with Design Misconceptions that Influence Design Performance Adversely
typical case: SSD
![Page 3: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/3.jpg)
Inexact
Fragmentary
May Produce
Design Deficiencies
May be Detrimental
to Cost, Performance
and/or Safety
of Divided Highways
![Page 4: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/4.jpg)
2D Approach efforts to overcome this incorrect
SSD determination
establishing some coordination
between the horizontal
and vertical curve positioning
not all design cases
are addressed
![Page 5: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/5.jpg)
Necessity for SSD Adequacy
Emphasized
No Explicit Process Provided
SSDAVAILABLE defined by lateral
clearance and curve radius
valid for circular curves longer
than the sight distance assuming
both driver and obstacle
positioned on circular curve
no assurance whether barrier height
and/or the presence of vertical curve
do not obstruct driver’s line of sight
![Page 6: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/6.jpg)
3D Models
capable of simulating
accurately compound
road environments (3D)
allow the definition
of actual vision field
to driver (3D)
focused in optimizing
the available SSD
introducing
new algorithms
design parameter
combinations
![Page 7: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/7.jpg)
Deliver Reliable Tool for Researchers
and Practitioners in terms of
SSD Assessments
Quantify the Safety Impact
During Emergency Braking Conditions
in Such 3D Roadway Environments
identify areas of interrupted vision lines
between driver and object
examine the interaction
of the utilized design parameters
estimate the probability
of SSD inadequacy
predict the additional object height
in order to grant SSD adequacy
![Page 8: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/8.jpg)
SSDDEMANDED ≤ SSDAVAILABLE
SSDDEMANDED
enriched point mass model
actual values of grade
(vertical curves)
friction variation
(vehicle cornering)
SSDAVAILABLE
driver’s line of sight
towards object height
at certain axis offset
3D roadway environment
![Page 9: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/9.jpg)
A
![Page 10: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/10.jpg)
calculation step
![Page 11: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/11.jpg)
![Page 12: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/12.jpg)
calculation step
![Page 13: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/13.jpg)
SSDDEMANDED = SSDAVAILABLE
![Page 14: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/14.jpg)
SSDDEMANDED = SSDAVAILABLE
![Page 15: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/15.jpg)
SSDDEMANDED = SSDAVAILABLE
![Page 16: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/16.jpg)
Alignment Selection
SSDdemanded
Calculation Step Definition (100m)
Define Intersection Points between
Driver to Object Sightline and Median Barriers Area
SSDavailable = SSDdemanded
Record Points
Vertical Difference between Sightline and Barrier Top Calculate
for the Most Unfavorable Case
Amended Object Height
Length of Hidden Sightline
![Page 17: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/17.jpg)
AASHTO 2011 Design Guidelines
Vdesign =130km/h
variety of horizontal – vertical parameters
passing lane 3.60m
NJ curvature at top increases by 0.22m
NJ median barrier (0.90m high)
crest vertical curve boundary values
+4% and -4% (rolling terrain) ISW
![Page 18: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/18.jpg)
263.0
263.4
265.9
269.7
273.8
282.4
287.1
292.0
297.1
302.6
278.0
306.8
308.3
308.4
0.08
0.17
0.45
0.62
0.64
0.66
0.69
0.71
0.73
0.75
0.75
0.48
0.10
0.07
0.69
1.36
1.90
1.95
2.02
2.09
2.17
2.25
2.34
2.44
2.09
1.86
0.68
0.80
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60 2.80 3.00 3.20
1100
1200
1300
1400
1500
1600
1700
1800
1900
2000
2100
2200
2300
2400
2500
2600
1200 1300 1400 1500 1600 1700 1800 1900 2000 2100 2200 2300 2400 2500 2600 2700 2800
Height (m)
Sta
tio
ns
of
Ex
am
ined
SS
D A
deq
ua
cy (
m)
Distance (m)
SSD
Length of HiddenSightline
Max Vertical Dist.of Sightline BelowNJ Barrier
Object Heightat End of Sightline
Area of Vertical Curve, L=1000m
![Page 19: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/19.jpg)
In-depth Analysis
individual variables effects
interactions
Identify Statistically Significant Effects
model parameters with p-values<0.05
![Page 20: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/20.jpg)
Probability of SSD Inadequacy
logistic regression model
estimate probability SSDDEMANDED > SSDAVAILABLE
Amended Object Height Prediction
lognormal regression model
![Page 21: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/21.jpg)
Road Geometry Design Elements Interaction
in 3D
Calculation Step
100m
Explanatory Variables
horizontal curvature (R)
950m, 1500, 2000m, 2500m, 3000m, 3500m
vertical curvature (K)
125m, 200m, 250m, 300m, 350m, 400m
starting (s1) and ending (s2) grade values
of braking process
-4% ≤ si ≤ +4%
grade difference (Ds=s2-s1)
inner shoulder width (ISW)
1.20m, 1.80m, 2.40m
Every Possible Arrangement Examined
2,772 different alignment cases
![Page 22: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/22.jpg)
Logistic Regression Model
Logit(πij) = Log𝛑𝐢𝐣
𝟏−𝛑𝐢𝐣 = 𝛃𝐢𝚾𝐢 + 𝛆𝐢
where: πij : probability of SSD inadequacy
Xi : explanatory variables
βi : parameters to be estimated
εi : logistically distributed ~[0, μ] error term
![Page 23: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/23.jpg)
Parameter B S.E. Wald test p-value Exp(B) Elasticity
K ,01813703 ,00295862 37,580 <0,001 1,018 3,404
R ,00875999 ,00064639 183,663 <0,001 1,009 1,516
s1 -,16167454 ,05077569 10,138 ,001 ,851 -0,009
Ds -11,56791068 1,42435424 65,959 <0,001 ,000 0,617
ISW -5,15025730 ,31647009 264,845 <0,001 0,006 -0,680
10-3 R2 -,00232342 ,00020009 134,829 <0,001 0,998 -1,006
K * Ds -,00765764 ,00269050 8,101 ,004 ,992 0,115
R * Ds ,00375407 ,00048204 60,651 <0,001 1,004 -0,513
K * R -,00000705 ,00000127 30,590 <0,001 1,000 -3,250
Ds * s1 ,78238421 ,08736706 80,195 <0,001 2,187 -0,036
Null Log-likelihood (LL0) -1864,85
Final Log-likelihood (LL) -392,825
Likelihood Ratio test 2944,05
df 10
pseudo R-squared 0,79
Parameter Estimates and Goodness-of-Fit Measures
![Page 24: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/24.jpg)
Classification of Cases of the Model Predictions
SSD inadequacy Predicted Correct (%)
0 1
Calculated 0 1032 75 93,2
1 80 1585 95,2
Total 94,4
![Page 25: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/25.jpg)
Lognormal Regression Modelling of AOH
Log(AOHi) = 𝛃𝐢𝚾𝐢 + 𝛆𝐢
where: Xi : explanatory variables
βi : parameters to be estimated
εi : normally distributed ~[0, σ2] error term
![Page 26: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/26.jpg)
Parameter Estimates and Goodness-of-Fit
Parameter B Std. Error Wald Chi-Square p-value (Intercept) 1,1172864 ,2487440 20,175 <0,001
K -,0024194 ,0003363 51,754 <0,001
R -,0029807 ,0001715 301,972 <0,001
ISW -,7683734 ,0546478 197,697 <0,001
Ds -,7717293 ,0748511 106,300 <0,001
Kx R ,0000008 ,0000002 23,110 <0,001
K * Ds -,0053688 ,0002712 391,984 <0,001
R * Ds -,0002012 ,0000362 30,832 <0,001
R * ISW ,0002100 ,0000310 45,802 <0,001
[s14=,00] * R -2,4985374 ,2694198 153,768 <0,001
[s14=1,00] * R 0 . .
[s14=,00] ,0022574 ,0001820 86,003 <0,001
[s14=1,00] 0 . .
(Scale) ,1259748 ,0043661
Null Log-likelihood -2198,8
Final Log-likelihood -637,8
Likelihood Ratio test 3122
df 10
pseudo R-squared 0,71
![Page 27: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/27.jpg)
SSD Adequacy Investigation
passing lane of freeways
left-turn curves
overlapped with crest vertical
alignments
Potential Safety Violation
for AASHTO 2011
Vdesign =130km/h
SSDDEMANDED ≤ SSDAVAILABLE
![Page 28: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/28.jpg)
Extensive SSD Shortage Areas
for Control Horizontal and Vertical
Design Values
SSD Sufficiency Evaluation
probability of SSD inadequacy
logistic regression model
prediction of the amended object height
lognormal regression model
![Page 29: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/29.jpg)
Analytical Model is Computationally Demanding
(Although Accurate)
Proposed Modelling Approach
emphasis in identifying interacting
road design elements
quantification of these effects (.
(elasticities)
very useful tool for practitioners
and researchers
![Page 30: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/30.jpg)
Probability Modelling Approach
efficient, suitable for researchers
to correctly assess SSD
in such 3D road alignments
Logarithmic Modelling Approach
somewhat less accurate in AOH prediction
further analysis required
to investigate non-linear relationships
between the examined variables
![Page 31: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/31.jpg)
Additional Work
models validation
examine more speed values
optimize in terms of SSD provision,
the influence of additional parameters involved
median barrier type for certain cases
(e.g. bridge - tunnel areas, etc.)
issues associated to human factors
more realistic deceleration values
night time driving
![Page 32: Stergios Mavromatis, Assistant ProfessorStergios Mavromatis, Assistant Professor Technological Educational Institute of Athens stemavro@teiath.gr Eleonora Papadimitriou PhD, Senior](https://reader034.vdocuments.us/reader034/viewer/2022052017/602f9f1eab41922d680bb044/html5/thumbnails/32.jpg)
Stergios Mavromatis, Assistant Professor
Technological Educational Institute of Athens
Eleonora Papadimitriou PhD, Senior Researcher
National Technical University of Athens
George Yannis, Professor
National Technical University of Athens
Basil Psarianos, Professor
National Technical University of Athens