my name is Lauren Dao I'm the
communications coordinator for the National Road Research Alliance
and I want to welcome you to the August edition
of Research Pays Off a couple reminders here at the beginning
to keep yourselves on mute uh that will just help prevent any
feedback or unwanted audio noises happening um if you have questions
during the presentation please feel free to write them into the conversation box
and yeah we'll get to those at the end We have a few more
months already scheduled out for Research Pays Off
in September on the 15th we have one of our
invited speakers presenting her her original planned talk for the Pavement
Workshop this year we will have Cheryl Richter presenting on
the "Federal Highway Administration's Turner-Fairbank
Highway Research Center and Pavement Research
and then on October 20th we will have Ben Worel and Steve Henricks from
MnDOT talking about their "Online Systems for
Test Section Tracking at MnROAD." So we've got those already planned out
and you can find more information on the Research
Pays Off website Today's session will
be recorded and you'll be able to find that on the
uh National Road Research Alliance YouTube channel
and links to the video will also be on the Research Pays Off web page.
I'm trying to think if there's
anything else from related to the NRRA that I want to share.
We have announced to the groups who have been chosen for the Call for
Innovation. Those emails went out last week so um
we I'm working on getting those those pages up
and on to the website Today we have Eyoab Zegeye who is here to talk about the Road Doctor and his research with
that as part of work he does with MnDOT and I think various other pooled fund projects I'm
assuming but I'm sure he will cover that more so
welcome, Eyoab, I will turn it over to you and if
you could make your your presentation full screen I think
that would help us.
Oh it's okay let me it's um let me okay can you see now not quite yet Lauren can you see not
quite yet it's loading now it's loading okay there we are perfect okay. Thank you, Lauren for the introduction uh
good morning all, good morning everybody. Thank you for taking the time to tune in
this morning uh my name is Eyoab Zegeye, I'm a research
scientist at Minnesota Department of Transportation
and I'm very happy to share with you our experiences and our the lesson learned
by on using the Road Doctor for the evaluation of pavements affected by
severe crack-heaving also known as tenting before I do that
though uh on behalf of the group of people who work on this project and
myself I would like to thank uh NRRA for
giving us the platform to share uh these these findings these
ideas with you all thank you I'm trying to move to the next slide here oh
let's not move into the slide okay everywhere
um I the work that I will be presenting today it's um it was
accomplished through a numerous collaborations meetings
discussions and consultations with my with my many colleagues at the
uh the Office of Materials and Road Research of MnDOT
um I would like to thank them all and I would also like to thank the leadership
for the continuous support and research on innovative technologies
um I'm it's such a such a blessing to work
in in such an environment and especially
now that we are working all high all from home I appreciate, I
appreciate more their collegiality and friendship
and I would also like to um acknowledge the work
uh the people involved directly in this project Thomas Calhoun,
Steve Olson, Eddie Johnson and my mentors and managers Dai Shongtao and Jeff
Brunner I also would like to thank MnDOT District 1 for helping us for coordinating
for all the field testings discussed in the slides Crack heaving or tenting for those who
don't know is a trans- transverse crack heaving is a payment
distress condition that affects primarily
bituminous surface roads built in cold regions
for those who want to know more about crack heaving
please please do contact me through email or
email or phone I would be able to give you a more comprehensive
literature review um but here just briefly
just to explain what it is it's uh it's a
pavement condition which is caused primarily by
the presence of moisture in the base aggregate freeze-thaw cycles
frost susceptible aggregates deicing chemicals cement treated aggregates and
lately we've seen also that roads built with using rubberized
concrete bases are prone for this type of pavement
distress the main driving force is I try to
explain here with these two sim-simple models simple
scheme it's like in winter when you have
moisture in the base aggregate basically you have this frost heaving
action that moves upward and lifts and lifts the
placement on both sides of an existing or new
transverse crack and so by doing so it creates a a peak
similar to that to the pointed tops of tents so that's
why the name comes from the name tenting comes from again here I give you this is a simple
simply same very simple um I am
description of what it looked like in the in the in the real world
basically you have this tented cracks with different spacing between them and
different height that can cover long stretches of area and and the main impact of this type of
center crack is that they impact the ride
quality and comfort of drivers.
I put here a video just to show you
how how how bad how bad it is when you have a vehicle
driving over these tented cracks it's basically is even it endangers the safety of the drivers. I
don't know if you are able to see the video I see it's very slow to do
things that do do the internet but um it's very hard on the vehicles and it
makes, endangers the the comfort and the quality of the ride
but it's also good to know that it does not affect only drivers it
affects also local inhabitants. Yes uh one thing that we learned during
this testing in in the Duluth area in District 1 is that
this type of um pavement distress affects also local
inhabitants the reason for that is that many of these
highways in the rural areas they are also
designed to serve for mail delivery boxes
or for school buses but you do see some kids waiting for school buses
sometimes and what happens is that drivers
when they are driving on these very rough roads they
try to avoid the tented cracks by driving on the shoulders, this
endangers the local inhabitants.
Another characteristic of this type of
pavement distress is that because it's very visible it's one of the most vis
complained pavement distresses in fact it's very common to see headlines on
on the news on prime time about tainted roads that have been repaired and then
again they still show up winter after winter.
This type of distresses um talking to engineers that have worked
on this on this issue for so many years uh one
thing that you you learn is that there is no
one-size-fits-all treatment approach the phenomenon is
very complex and it's unique to individual projects
however they all know that we do have some possible solutions
from as simple as grinding to level the crack
center cracks to completely removing the and replacing the base aggregate and by
that also the pavement structure so we do have
those treatments that we think could work. What is missing is
that we don't have first of all we don't have an agreed understanding of the
mechanisms that produce tented cracks then most importantly we don't have an
established rapid testing and evaluation methodology
for pavements affected by tenting and then later on we
don't have a structured approach for selecting the best solution
for an individual project.
So if you don't have these tools um
it's very hard to choose the proper treatment and when you don't choose the
proper treatment as is it happens more often than we would we
would like you repair a road and then it comes back
again next year and public is complaining, complaining
even more. Sso this is the
brief introduction to the crack-heaving, tenting problem and especially from the
point of view of Minnesota. Because of this reason at District 1 Office brought the issue to the attention of our office
and it provided us with a long list of roads affected
by tenting that indicated different severity levels based on
visual inspection but also based on the complaints that they
they were receiving from from the public. I know it's very
very small here to see but you can see here some of the roads, is
like US169 one of the most complained by public. So just to give you an idea,
our office is located in the metro area in Maplewood and the roads that we are
talking about they are all in District 1, in the
northern part of the of the state, so
the Minnesota the District 1 office, they provided us this long list of roads
and then they asked us if there is anything that the Road Doctor
can provide to support the treatment decision making
process.
For those who don't know, for those who
are not familiar MnDOT the MnDOT Road Doctor is an
advanced survey van composed of many, equipped with many
advanced sensors, non-destructive sensors. In my view the three most important
items in the Road Doctor are the three data acquisition, data controlling units
and the CamLink from Roadscanner, the SIR30 from GSSI, and the geoscope
from the Examiner. So the Camlink
controls most of the sensors including the positioning and navigation system
GPS, IMU, and DMI. Two video cameras high definition
industrial video cameras high resolution accelerometer
thermal cameras and the LIDAR laser scanner.
While the SIR30 controls the two horn antennas
depicted in the picture here the two horn antennas, I don't know if you can
see my mouse and then the ground couple antenna which
is basically in front of the of the vehicle.
I will speak more about these two
antennas their differences later on, but and then we do have the geoscopes
and the other 3D radar antenna. That's a brief timeline. We purchased
all the sensors for the van from Roadscanner in
2018 and then we assembled and installed all the sensors
in the survey vehicle through a very successful
collaboration between vendor staff and a local contractor. In 2019 it was
completed. We went through a training and
verification with Roadscanner and then since then we continued
learning the system, we developed some implementation plans
for projects, we did some trial field tests on I-90 in Adriano-Adrian, Minnesota
I-94 Saint Paul and then in 2020 we started our, this project the
require which is one one of the most comprehensive projects that we have
ever taken uh so far the district one tenting issue.
We are also running a moisture monitoring test,
and of course in 2020 we had also COVID-19
but in the next we have some other um items that we would like to
investigate in the coming years, including validation of the 20
parameters that we uh propose in this study and moisture
monitoring stripping and some project scoping applications.
Um
Here is a below I put two pictures that shows you,
some gives you some views of the survey when we can
run it in two different configuration. One is using the two horn antennas in
the ground couple the thermal camera and the video cameras
which are inside protected, isolated in this box
and then we have the LIDAR laser on the back of the
of the survey van. Here the second configuration,
we have the 3D radar, everything else is the same. Here is just to give you some views
first to share with you some rear and side and
side views of the van, how we store the equipment
inside this workstation, how it it looks like,
how it looks like when we are testing especially when you're testing it during
nights. So back to the tenting, so when
District 1 approached us with the tenting issue, our perspective,
from our perspective we were very happy because this was a one
a great opportunity to test the Road Doctor in winter conditions.
Then our intention or our objectives if you will,
were to test it to develop a test develop
or explore testing parameters that can be used to detect
measure and measure the tented cracks and then based on these parameters to
develop a criteria for rating and identifying critically tented sections
and then lastly, draft a reliable testing methodology for evaluation of pavement
affected by tenting.
The approach we took was um
obviously it was very difficult to test all the our more than 20 roads that were
that were proposed by District one. Instead we chose to
target a limited number of roads with different tenting severity levels
and then we plan to, we decided to use all the available sensors
and to collect data according to this plan here.
First initial collection in winter 2019 and then monitor if there is any change
between winter and summer by testing in summer or fall 2020 this is going to
happen very soon. And then in winter 2020 um validate
the test parameters that were proposed in this study.
And so the idea was first of all to collect
data with all the available sensors and then look at the data
and explore if we can see if we can find some parameters that can
characterize the issue and that can be used to to basically describe uh
in terms in quantifiable manners the issue.
Here are the roads that we selected for
testing as you can see I colored in red the one the most severely
tented, most complained by the public and then we
have yellow the ones that showed moderate level of
tenting and then we also included Minnesota 37 in here in
green which was recently rebuilt so it didn't
it didn't show any major problem with tenting.
We kept it there
for for control to see to compare with the others.
A total lanes the total of 100 miles of road was surveyed
during two days of testing. We, the target testing speed was 30 miles,
30 miles per hour. This was a balance between the different
requirements of the different sensors and also because we had
the ground couple antenna GPR antenna very close to the surface so we didn't
want to to run into into problems there and
because of this low speed and because we were testing also
highway roads, we we it was required that we coordinate
with traffic control and with the core and that we coordinate
with the winter plowing plans of District one
and District one has been very crucial and very helpful in achieving this safety
in us in a successful manner. Also we also looked at we chose the two
days, two very cold dates for testing. The
reason for that is we wanted to ensure that the tented,
the tented cracks were formed when we test.
This was not planned by but I think it's mostly by plan
not by plan, by luck we end up testing on two days
on the coldest week of the winter. The temperatures were
below 19 degree and without no wndchill. So this was good because it ensured that
tented cracks were formed however unfortunately or
maybe as a first lesson we we found out that the freezing
temperature along with the rough rides end up
damaging some of our cables or sensor cables, especially the cables
which were outside of the van. So this was and this ended up
compromising some of the data that we collected so or not all the data well
well then data processed because they were compromised but at least
this was the first first learned lesson. We need to take more care on choosing um the proper way of testing
when we are especially going on very rough roads and in very
freezing climates. As I indicated before we
the plan was to take all the sensors and test them.
We used all the sensors um we end up not running the 3D radar because
the today the to the vertical temperature and the
long time of testing made it quite impossible for us to go back and then
test again the three radar especially because by the
end of the two dates, we already knew that we had two
cables broken, so for now we kept it with the ground coupled.
This was kind of good because for us was first of
all this was where we learn what we can do and ground
coupling give us a one-dimensional view of the GPR results
and then going forward we can build on that. And once we've completed the data collection we processed the initial
processing was done using the Road Doctor software. The Road Doctor software is a very very very robust software,
in my view, the primary, the main advantage of this Road Doctor software it
can link different data sets and then you can also visualize the data
linked together.
I put here two views of the Road Doctor.
You have the video camera, you have the LIDAR data,
the roughness data, thermal data on the top, on the below you have the
video map GPR data so um I will try
just if it shows so here you can you the video is moving, I don't
know if it shows on your end, and as the video is showing the map is
is move the video is moving, the map is moving,
and so now you can exactly, it helps you to investigate um more
closely when you have anomalies. And then
the another important thing that the Road Doctor software does once
it process the data it process the data the GPR data the iri data everything it
then outputs tabulated files that can be
further investigated.
The views can be shared
with people who have Road Doctor software
viewer, but for our purpose we wanted to investigate further the data and so
we took out the outputted files and then we further processed them using some
PYTHON algorithms, because the data was so large to
to process in Excel. So what we did then is we divided all the data the data from
each road, into smaller lots in lots of one
one one quarter mile and then we plotted and analyze each road separately. Here I
show you I give you an example of, for example, we
have on the top the thermal data, the LIDAR data, the profile measurement,
the IRI data, the GPR data, and another GPR data,
MDI, I will discuss later. So this gives this gave us the opportunity to look
closer at the data and then to start to exploring relevant
parameters that can be used to measure the tented cracks.
And now I will go on by explaining how
each of these data how we looked at each these data sets.
So just to wrap up, we end up testing over 100 miles of road, but only
47 miles of roads were provided, were not compromised, so
we're analyzing this data. This meant, this meant that we end up
looking at more than 300 roads, one by one,
or the videos of the GPS and trying to find out which parameters can explain
can describe or characterize the tenting issue.
So the first is the video we we were very happy with the videos, we
had two videos.
I am showing one of the videos, one of the video was
pointed toward the road, the other is pointed toward the
ditch. Here I'm showing the one pointed
to the road and we were very happy with what the video was showing that
we found that the video is very good for visual inspection of the roads
and it it was essential in identifying sections which were heavily affected by
tenting and but importantly, again importantly, it
also gave us an indication of the degree of tenting
because you could see the vibration on the vehicle and the bouncing of the horn
antennas in front.
This is a video, I will try to show you
but it is it will be very slow probably, you not
appreciate how much the vehicle was bouncing when you see
the tenting. So in that sense it was a very good
tool for us to understand and somehow rate the roads and
what another good thing that does, whatever parameter we came up with,
this was a way to verify the videos were a way to verify if the parameters
the numbers with the parameters, if they make sense. The second data set it came from the
LIDAR laser, the LIDAR laser, it's basically, we have
the laser generator that emits a laser beam toward the pavement surface
and records the reflection. Some of the items
recorded at the travel time the distance between
the laser generator and the target, targeted point, the angle
beam laser positioning strength of reflection.
But the most important thing is that then when you have all these
items you can derive the xyz coordinates, the local coordinates that can be
transformed to longitude latitude and elevation and then based on this
cloud data you can reconstruct the surface
of the roads.
Here I showed one of the the picture is um one LIDAR emission,
the plan view, this is the view of the of the
of one lane of the road you can see here. You can see that the LIDAR revealed
reasonably well the transverse cracks. We looked also we tried also to see at
the elevation data if the elevation data if you could use the elevation data to
measure the tented roads, but because this, because of the, that that effort was not successful
because of the accuracy and density of the data.
But anyway, we were able to see from the LIDAR all the
transverse cracks what we were not able to say if all these transverse cracks
were tented or were just transverse cracks. So
the next data set we looked at is the thermal camera.
The thermal camera is becoming very popular lately because it's
it is very useful in detecting surface, surface defects
and anomalies that could be signs of underlying issues.
The thermal camera they measure the surface, the
infrared thermal camera measure, they measure the surface temperature.
The assumption here is that defects or anomalies possesses different thermal
conductivity and heat retaining capacity compared to the surrounding
bulk area, so if you have an issue you would you
especially in this case where we think that some of the issues are related to
moisture, then definitely an area with
more moist or more salinity or more or more, more ice, it will show up
compared in contrast with what with a with a drier
more less affected section.
So because however because the variation
of the temperature within the roads, within the roads were
very large, we it was required that we did we do
some filtering to especially to enhance and
reveal the short wavelength temperature. We use the normal anti-smoothing filter
based on moving average of of base 16 feet and here you can see
in the in the charts, the blue ones are the short wavelengths and the
uh gray shaded um are the long wave. So when you, this is just a
temperature measurement in one line, one profile but if you
take all these profiles about 300 profiles for the entire width of the
lane, you end up with these two heat maps.
So now please remember these two sections, because I will be talking about
them, I will use them again as an example going forward.
We
have section A, which by looking at the videos and
looking at the LIDAR data it showed very little tenting. On
contrast, section B, it was extremely tented
section. So now when we look at the thermal, thermal dataset from them you can look that the
section A on top, is it shows less thermal segregation
than sections, than section B. Again the section B,
it also shows this anomalies, the defects that
they are occurring almost at regular intervals,
almost close to the matching very closely the transverse cracks that
we've seen in the LIDAR data or the video data.
So this was already another thing it is telling us
when you have where um these warm areas, I'm talking to my supervisor and mentor, Dai, he he also was mentioning that this
great area could be also the one of the plausible reason what could
be yes you have tented area cracks and then probably when you put salt on
them during the de-icing, probably they will
retain the salt more than the others and that could be one of the reason how,
regardless of the reason we were happy to see the thermal
segregation is showing us when you have extreme
tenting compared to no tenting.
Going forward, we look also at
the GPR data, for those who are familiar, not familiar,
GPR is a non-destructive technology commonly used for pavement sub-surface investigation like layer thicknesses anomalies, defects,
utilities, dowel bars, void in asphalt layers, etc.
I will try just for the sake of the discussion I will try to give a
brief simplistic explanation description of what
what is the technology so please bear with me and forgive
my some simplicities in doing this. So the concept consists in transmitting
electromagnetic waves pulses into a medium.
So you have, you
transmit and then you record the back reflected energy.
And the idea is um basically the emitted pulse,
it propagates through the medium and the and then every time it reaches
a medium which is different, which it has different electro
magnetic properties, it reflects back part of the energy,
and part of the and the remaining energy it continues to propagate in the
in the next surface, uh until it attenuates.
So every time there is a contra- has significant contrast between
different mediums, you get back the signal response you get
back the reflection, but so this is what generally looks this is
a very again this is a hornet in a uh idealization very simplistic but just
to give you an idea this is what the uh what the response or signal response
GPR signal repository looks like for one for
one static measurement.
So when you have
static measurement, you you measure the time of arrival of the
response in the y-axis and you measure also the
strands of the reflection so here for example when the pulse
reaches the ground surface you have the ground reflection you see
and then when it reaches the interface between
the AC and the unbound aggregate, you have another
another another significant response here.
So knowing some properties, you can measure the the thickness here or
in other cases you would know that there is a huge contrast
of two different materials in this case.
So this is a starting measurement, what
happens when you move use we move forward with your GPR equipment
is that now instead of one trace, you have
thousands of stacked traces and this that when this hundreds here i have
only 10 or 20 stacked traces but when you have more of
them then you end up with this GPR profile
that you can, most people use binary gray colors to show
them but you can also use different colors. But this is just to give you a brief explanation of what GPR technology is. And at the end is important to know uh what type of GPR you're using.
On the giving for the same pavement
structure for the pavement material, another important um
variable is the antenna frequency so based
on generally, in a simplistic way again, when you have larger antenna frequency you have higher accuracy but you lose in
depth of penetration and vice versa. When you have lower antenna frequency,
you have lower accuracy, but higher depth of
penetration. That's why we have two different configuration or three
different configuration of GPR antennas. We have the ground couple antenna which
is 400 megahertz, which is suitable for investigating the
base, sub base and probably some part of this subgrade,
depending on the pavement condition, depending of the material, depending on
the moisture condition, depending on many
variables and then we have the horn antenna to 2000 megahertz that
are more suitable for the top layer of pavement, probably for the top asphalt layer, because they have higher accuracy but
lower depth of penetration.
And then we do also have the 3D radar
frequency, 3D radar antenna, which operates in a
different technique which is called stair frequency,
basically it collects data in a wide array of frequencies from 100 to 3000
and then during the data processing it chooses the the
responses with the highest resolution. As I say at the, as I indicated before, we
for this study we end up using mostly the ground couple data.
We intend to build on it and then whatever we learn, move it also
build and move it to the other GPRs, especially the
horn antenna and also the 3D data because then you will have a full
picture, full language picture of the of the issue.
So the data that GPR collected from the ground couple data
was processed first with the Road Doctor software,
moisture modeling from Roadscanner.
The Roadscanner moisture modeling gives
you two products from for moisture, one is the moisture map and the second
is the moisture damage index. I will try to explain what these two
things do and first of all the assumption, The
assumption is that when you have the same pavement and when you have for
the same pavement, from the same structure from the same material,
but then you have one spot which is moist and all moist or frozen and one spot
which is dry, then this will show in the GPR signal
responses. And then what the the second assumption
is that the difference in moisture is reflected in small changes
in signal frequency content and moisture impacts are measured
obviously well at the frequencies close to the antenna
frequency so whatever is far from the antenna central
frequency is can be considered as a noise
so to and based on this assumption, you can derive some GPR parameters that
can be used to measure the moisture, moisture
activities in the base aggregate So here is the original for example, just
to give you an example. This is an original, the original measurement
signal frequency and the in time domain and if you
convert it to the, through frequency analysis
to its amplitude and frequency then you can see that
the same signal has many pretty many frequencies so the next step
is…
Sorry. So the next step is to filter the
frequency responses with very close to the antenna in this case
we chose to filter out the from 50 to 400 megahertz
close to frequency and then you get this red signal
response. As on once you have this filter what
Roadscanner has done um we don't have yet a final official
uh manual or paper on this so don't quote me, but just in the
understanding is that then you average the reflections on the on a
fixed window lens of the signals and if you do
this for all the traces, you end up with a profile measurement
where you can, looks like this.
So again these are the GPR parameters obtained derived from the GPR data that
are assumed to measure the moisture
activities in the in the, in the pavements.
I'm again using
the same section A and sections B from before as i as we
mentioned, the section A showed little or no
tenting, while section B showed it was extremely tented.
Here we were very happy to see in fact this was also reflected
on with the GPR data that um the on the top one yes you de have
some moisture activity, but not as um
uh uh as stressed as significant that though as the one
shown in the in the in the plot below. But another thing we showed that
these moisture activities these huge um
moisture activities they were again occurring at the regular interval
and then very closely matching the transverse cracks,
so whatever, wherever it was observed in the surface, now it's also there is
also some something going on beneath the surface. So now we had the
LIDAR, the video, and the thermal telling us,
give, providing us good information on the surface and now
you have the GPR telling you also, confirming you,
that confirming us that there is something going on beneath those
tented cracks.
The MDI moisture model is
basically it takes these GPR parameters and then it come relat-relates them to
moisture contact values, so this relationship was
our understanding is that this relationship was developed through
laboratory experiments using finish finish aggregates um,
so we didn't know if this would work for our for us, for um
for the Minnesota aggregates, so for for now, for the time being,
we chose not to use the MDI, not to consider the MDI for this study.
Instead, what we're doing is that we are working on a
new project, ongoing project, that we're trying to calibrate,
verify the MDI on MnROAD test cells. As you know, MnROAD, we have so many pavement cells, pavement sections with
different structures, all of these cells are instrumented with
different sensors, including moisture sensors
and the temperature sensors are varying at
varying lengths and we've been we've been monitoring this data for
like more than 20 years, so we do have moisture data.
What we're doing now as I show here on the
on the next on the chart in the right for example,
how the moisture varies from winter to summer.
So what we are doing now is we
are monitoring these cells using also GPR and
also FWD to see, to validate the MDI and to see if
we can use the GPR to measure, to monitor the
moisture activities in the pavement. So far we have some very good
promising early, early results. For example, here on the
left I showed the charts from the FWD data as
you ex- as you can expect, in February, when the
base is frozen the pavement is very stiff, so
the deflection, the plate deflection is very low, and then in early spring it
jumps up and then it kind of stabilizes in this plateau.
This is for all different cells with all different pavement types,
but also we observed very encouragingly that the MDI
is showing a very good trend, very good uh
uh very good um, relationship with both the bar
calculated base modeling and also the MDI and parameter that Roadscanner proposed. So we are continuing with
this monitoring, we plan to monitor from winter to winter
and but, these are very encouraging results
that will enable us to use GPR to kind of have a sense of the moisture activity in
the pavement with without having to use moisture probes every time.
Back to tenting, back to tenting, so far we've seen
that the video LIDAR thermal and GPR provided good information surface and
subsurface information of the tented cracks.
What they didn't
provide so far is a parameter that can be used to
measure the frequency the strength and the
frequency of these tented cracks. We don't have a number from
from this receptor so because of that we continue with looking
exploring the data more. Here the plot shows
it was a nice plot here because it shows you it shows the
capability of the data fusion the ability of the Road Doctor
so subray and the product software to link all this data
and if you link them you can see wherever you have transverse cracks you
have also these thermal segregations and then beneath the some
of the transverse cracks you have also significant moisture activities
as derived using the GPR data.
So this fusio,n this linking was
one of the important things of this study.
So we continued looking at the data when we looked at the profile,
profile method data. In particular first we looked at the roughness data, the
international roughness index, as you know, the roughness in the index
is a derived from longitudinal surface
profiles and the vehicle acceleration measurements.
This is commonly used to check the roughness of the pavements so
is very a very popular tool but we found for this, for this purpose
it was yes it is, IRI value are very useful to compare sections,
sections from the same pavement tested at the same
time.
You can compare section A and section B.
Or you can compare the, how the evolution of
tenting between winter and summer but you cannot really
use it to to measure anything because whatever the IRI has
the roughness has many, many competing contributing factors
that it is very difficult to just extract the contribution of the
of the, of the tenting. But here you can see section A
and section B as we and as I indicated before, section A showed very
little to no tenting, while section B had very extreme tenting
and you can see here I put on the right, just one
to give you an idea a scale of IRIs how we, how we commonly measure them so the basically, the section A is falling on
almost a new appointment two to four IRI millimeter two kilometer
while the second one is almost in the edge of unpaved road.
So this is how bad, how bad is tenting on the
on the IRI, but again um IRI has also other contributing factors.
So
the next step, what we did is this is something that we did not
start actually this was very well studied and very well applied in by our
colleagues in Canada, especially I would like to reference this
paper "Evolution of Pavement Winter Roughness."
They use the longitudinal surface profiles to measure
the tenting, so we applied the same technique.
Ohe only difference now we had also other data to backup whatever
we are measuring we have LIDAR data, GPR data, video data to match up, backup, and verify if these parameters were measuring,
but what are these parameters these parameters were derived from the
longitudinal surface profiles.
As you know, logic surface profiles are
composed of long and short wavelengths. The long
wavelengths are generally linked to issues in the subsurface
layers like differential heaving settlements,
while the short wavelengths are due to distress observed in the surface
or they are also responsible for intense vehicle vibration.
So the idea because was to again use the anti-smoothing technique
to remove the long wavelengths and then to reveal
and extract the short wavelengths as I have shown here in the picture.
What was good about this technique is that then
it allows us to come up with an automated way of
picking those peak values and then also to measure the spacing between these
peak values and see if these two parameters now what I call
here PHV and the SBP, by PHV I mean Peak Height Value
and then Spacing Between Peaks, to see if this could
be used to characterize and describe the tenting and
we were so happy to see that those peak and spacing
were were matching very closely the other data.
So the only difference what we added here our contribution, was that
linking these peak and spacing values to the GPR, LIDAR, and
confirmed that whatever we're measuring is close to something which we already
have seen in the other day.
We were very happy,
encouraged by this finding What, why are these two parameters
important in my view? These two important, they are important
because now we have numbers we can assign a number to the tenting issue.
This, we can quantify and characterize the issue.
But more, we can also develop a rating criteria. For example, we
we try to look at the rating criteria based on the IRI we found that obviously because both parameters both peak and IRI are for example obtained derived from the surface profile,
they were, they had a very strong correlation
they are correlation, proportional correlation, while the other spacing
and IRI they have also very strong correlation but in this case,
inverse correlation. So what what this allows you to do is, for example, if you
want to, you can choose what is your acceptable IRI limit for example, in this
case if which is 175 then you find out you find the
corresponding acceptable peak height values and
spacing between between the the tenting cracks.
So by doing this, it
means then you can decide okay if I have average peak heights below
0 12 I am in kay zone, or if I have spacing
between tenting above 40 I am in okay zone. This is a comparative way
but use though you can also look uh use some statistical analysis when you have
a pool of roads that you tested, then you can,
you want to decide which one to start work, where to start the work with
then you can do some statistical analysis. In this case we use the
19 tents and 90th percentile to find what are the height value and spacing
value that leads to severely tented sections. We
found for example in this case, for this
example, of course because this is statistical analysis is very
associated to the data that we have, it may change if I change the data
so but, for these datas 0.17 height was a limiting factor
and then for spacing when you have um spacing below 28 was another limiting
factors. So if you use for example these
parameters then you can create your priority ranking. When we did that for
the rows that we tested, we confirmed that US 169-D
again, as I explained, it was one of the most complained roads
by public and by by indigenous industries one,
we found out that that road was also one of the
based on this criteria that we just proposed you find out that it was one of
the severely tented roads compared to the others.
On the
contrary, Minnesota 37 the the one that we chose just for test control, it was one of the least affected by this tenting issue,
and then you can also do one other important thing, you can then use these
parameters to map them in a geographical map so that
you can locate, identify, where these issues are
more more persistent, more seen, and then you
can locate them and then also measure the
extent of these sections and in my view, this could be a very good
tool to provide to the engineer so they can go on these sections
and do some additional forensic investigation
to find what is a proper solution.
So just to summary the Road Doctor
survey was found suitable for evaluation of pavements are affected by tenting.
However care should be taken when operating the van in rough roads in
extremely cold condition. The video LIDAR thermal GPR and IRI data
provided important surface and subsurface information of tented roads,
but did not provide quantifiable attributes.
On the contrary, the longitudinal surface profile measurement provided two
promising pavement tenting parameters for measuring the
height and spacing between the tented cracks.
The validity of these two parameters, PHV and SBP was verified by examining video
thermal LIDAR and GPR data. An ongoing research at MnROAD for
estimating the moisture content in base aggregates from GPR data is showing good promise. Validate-validation verification scaling
up testing to be conducted soon. So thank you all for your attention and
usually I would stop here but given uh the new working
conditions I would like also to thank my very understanding three office
mates, my two kids and my wife that here they are helping me preparing this
report.
So again, thank you all for your
attention, if you have any question I would be happy to answer. Thank you so much, Eyoab. We just have
a couple more minutes before the top of the hour.
If you have any questions uh feel free to unmute yourself
and you can ask them or you can feel free to type it in to the chat box. Hello if i can this is Timo from
Finland and from Roadscanners if i canI ask one question.
Absolutely, thanks for
joining us. Okay, okay, thanks, thanks and and first
first of all Eyoab, an excellent presentation and you have done good work.
Thank you. Just a few, few comments. First of all, this tenting it's, it's
really interesting, interesting and and uh these tenting problems are also coming, coming to be
more and more problem, problem in in Finland and Nordic countries, we don't
know yet why they are now accelerating but,
but we are facing the same issue not as bad as you guys have but,
still they are.
And about the MDI in
in general what, what you have been seeing
here, here is, we have been doing uh winter testing uh quite much winter
testing also with MDI, MDI
in, in Aurora, this autonomous driving test track in in in the northern
northern flatland, and we have got we have over over there
also pergo station sensors with the moisture sensors and we've been
calibrating and we have got excellent, excellent correlation uh
between the moisture content and everything else and and and
in in when you are talking about MDI and and and GPR data, what you are actually
measuring in the wintertime is the amount of on
the amount of unfrozen water in the frozen ground, ground and and and that is what you see in
in this field basically you see the icelets over there
which is now always containing or containing this
this this unfrozen water and in that way it matches excellent, excellent your
finding with with our findings.
One issue anyway what what I'd like
to and ask also is that you said that you
didn't have, you had a problem with the quantifying attributes on how severe
severe tenting is and and you have to use profilometer data
but uh Road Doctor lancer then has also this uh 3D accelerometer test.
Yes. And, and we have been using uh transverse crack uh classification
from this excellent 3D accelerometer data.
Yeah. In concrete roads and also in transverse crack and
there's a one master's thesis made made for Roadscanners for that,
but how you do that is that you are filtering all the all the
longer frequencies out of accelerometer data, rather than one meter or three foot in in US case so and and you calculate only the
very short wave length and and the amplitude and this
amplitude seems to be matching extremely well.
We've been, we've been doing this also in in couple
of cases in USA also on concrete road when you have a joint joint repair
analysis so that that could be a one issue
you could try and and and just one comment that..
Yes, yes ,please, if I may,
first of all thank you very much for staying up, and yes,
for these excellent comments, but you're right, I need to clarify, maybe I didn't
state it during the test, the profile measurement
were obtained from the 3D accelerator from the raw data, all the data presented
were from the raw data, yes, you're right. In fact, we use that
the only different thing is that we process them outside of the Road Doctor
software, but yes, we use the data from the 3D
accelerometer. Okay yes that's right. But I will leave, a
very, you you gave us very good feedback I
would like to continue to follow up this discussion with you,
since you are working on the same type of issues and
and see if we can interpret better our results by your help.
Yeah, thank you very much. Yeah, thank you.
We do have a few questions in the
conversation box if you want to pull that up.
Oh okay, okay so um thank you and uh..
"What type of sensor were used for the moisture data, from
what type of sensor were used from, for your moisture data?"
We use the GPR derived parameters and also we use the moisture
sensor that are… I don't know if you are referring to the
MnROAD cells, we are using the moisture sensors that are embedded in
the test cells in the in those cells 127, 728, 188, and 189
I believe they are… dieletic moisture sensors. Again, Ken Maser, thank yo. "Can you comment on what you have learned from the GPR LIDAR and
IRIs that you didn't already know from the visual
and profile measurement?" Well what we learned is the visual measurement was a
good confirmation but from the LIDAR, specifically from the
GPR, and the surface profile measurement now
we can't quantify the thinking we can measure the tenting and
give numbers to the tenting issue, I hope.
I hope that answers the questions, Ken Maser. Yeah these were the- I
don't see other questions, Lauren. Yeah, it was just those
couple ones. All right well since we are at just a little past 11, I think we will just wrap it up. Again, thanks, Eyoab for the presentation,
thank you to all of you for joining us and this
has been recorded and will be available online
within about a week if you need to refer back to it.
I think that's all we have so everyone have a great day.
Thank you..