corrosion_revised

corrosion_revised

docs/GPR/Applications/Corrosion.html

@@ -128,11 +128,11 @@ <h2>Introduction to Reinforcement Corrosion<a class="headerlink" href="#introduc
 <p>Figure 1. Type of crack on the concrete bridge due to the corrosion.</p>
 <a class="reference internal image-reference" href="../../_images/Figure_2.png"><img alt="figure 2" class="align-center" src="../../_images/Figure_2.png" style="width: 1000px;" /></a>
 <p>Figure 2. Bridge deck sample located at TFHRC. The right figure shows delamination along the rebars due to the corrosion.</p>
-<p>GPR uses electromagnetic (EM) wave to investigate below the surface. Since EM waves travel at different speeds at different media (the wave speed is slower in higher dielectric, and faster in lower dielectric media), GPR is widely accepted for identifying concrete’s corrosive environment.(Dinh et al., 2017; Faris et al., 2023; Imperatore &amp; Rinaldi, 2019; Solla et al., 2019) Specifically, the GPR signal on corrosive area tend to be attenuated compared to the healthy rebar, because the presence of cracks or corrosion byproducts cause the scattering of GPR signals.(Tešić et al., 2021; Zaki et al., 2018) However, some researches have shown that the signal can be enhanced when the cracking is longitudinal and specific type of corrosion byproduct is stacked along with the crack direction.(Hong et al., 2022) Here we exclude these uncommon cases and consider a general corrosive environment or delamination cases.</p>
-<p>We follow the ASTM D6087 standard method, “Standard Test Method for Evaluating Asphalt-Covered Concrete Bridge Decks Using Ground Penetrating Radar”.(ASTM International, 2022)  In this procedure, the reflection amplitudes among rebars are compared to assess their condition. The assessment is done on a logarithmic scale, where the amplitude differences among the rebar signals are calculated relative to the rebar with the highest reflection amplitude, considered the healthy rebar.</p>
+<p>GPR uses electromagnetic (EM) wave to investigate below the surface. Since EM waves travel at different speeds at different media (the wave speed is slower in higher dielectric, and faster in lower dielectric media), GPR is widely accepted for identifying concrete’s corrosive environment. (Dinh et al., 2017; Faris et al., 2023; Imperatore &amp; Rinaldi, 2019; Solla et al., 2019) Specifically, the GPR signal on corrosive area tend to be attenuated compared to the healthy rebar, because the presence of cracks or corrosion byproducts cause the scattering of GPR signals. (Tešić et al., 2021; Zaki et al., 2018) However, some researches have shown that the signal can be enhanced when the cracking is longitudinal and specific type of corrosion byproduct is stacked along with the crack direction. (Hong et al., 2022) Here we exclude these uncommon cases and consider a general corrosive environment or delamination cases.</p>
+<p>We follow the ASTM D6087 standard method, “Standard Test Method for Evaluating Asphalt-Covered Concrete Bridge Decks Using Ground Penetrating Radar”. (ASTM International, 2022)  In this procedure, the reflection amplitudes among rebars are compared to assess their condition. The assessment is done on a logarithmic scale, where the amplitude differences among the rebar signals are calculated relative to the rebar with the highest reflection amplitude, considered the healthy rebar.</p>
 <p>The reflection amplitudes are converted to a logarithmic scale using the following equation:</p>
 <p><span class="math notranslate nohighlight">\(A_{dB} = 20 \log_{10}(A)\)</span>,</p>
-<p>where <span class="math notranslate nohighlight">\(A_{dB}\)</span> is the reflection amplitude in decibels, and <span class="math notranslate nohighlight">\(A\)</span> is the reinforcing reflection amplitude in data units. After converting the GPR signals to decibels, we subtract each signal’s amplitude from the maximum decibel value among all signals. These subtracted values are then displayed as contour maps on the bridge deck. Locations where the signal is below -6 to -8 dB are suspected to be under corrosive environments.(ASTM International, 2022)</p>
+<p>where <span class="math notranslate nohighlight">\(A_{dB}\)</span> is the reflection amplitude in decibels, and <span class="math notranslate nohighlight">\(A\)</span> is the reinforcing reflection amplitude in data units. After converting the GPR signals to decibels, we subtract each signal’s amplitude from the maximum decibel value among all signals. These subtracted values are then displayed as contour maps on the bridge deck. Locations where the signal is below -6 to -8 dB are suspected to be under corrosive environments. (ASTM International, 2022)</p>
 <p>To achieve our objective, first we need to identify the rebars from GPR B-scan data. Secondly, list all the amplitudes according with the coordinates. Then convert them into decibel scales, and lastly subtract all the decibels from the maximum value to plot the 2D contour map.</p>
 </section>
 <section id="objectives-of-the-case-study">
@@ -147,7 +147,7 @@ <h2>Prerequisites<a class="headerlink" href="#prerequisites" title="Link to this
 </section>
 <section id="corrosion-assessment-results">
 <h2>Corrosion Assessment Results<a class="headerlink" href="#corrosion-assessment-results" title="Link to this heading"></a></h2>
-<p>Executing our code along with proper input variables, it automatically outputs interpolated rebar signal amplitudes in decibel unit. Figure 5 (a), (b), and (c) shows the pinpointed rebar signals on the split GPR B-scans. Our code reads the absolute amplitude values right on the rebar signals, save as a list. Subsequently, the code concatenates the split B-scans and generates 2D scatter plots for the entire data and splits the concatenated B-scans again to plot the 2D scatter plots along the actual four GPR scan lines (Figure 5 (d) and (e)). To check how the rebar signal amplitudes vary, we also output the amplitude distribution histogram on each lane (4 GPR lines), shown in Figure 5 (g).</p>
+<p>Executing our code along with proper input variables, it automatically outputs interpolated rebar signal amplitudes in decibel unit. Figure 3 (a), (b), and (c) shows the pinpointed rebar signals on the split GPR B-scans. Our code reads the absolute amplitude values right on the rebar signals, save as a list. Subsequently, the code concatenates the split B-scans and generates 2D scatter plots for the entire data and splits the concatenated B-scans again to plot the 2D scatter plots along the actual four GPR scan lines (Figure 3 (d) and (e)). To check how the rebar signal amplitudes vary, we also output the amplitude distribution histogram on each lane (4 GPR lines), shown in Figure 3 (f). Lastly, the code outputs the interpolated amplitude contour for each lane (Figure 3 (g)).</p>
 <a class="reference internal image-reference" href="../../_images/Figure_3.png"><img alt="figure 3" class="align-center" src="../../_images/Figure_3.png" style="width: 1000px;" /></a>
 <p>Figure 3. CHARISMA outputs from Mississippi I-10 Region 01 rebar cover depth for the first 4 GPR scan lines. Figure 3 (a), (b), and (c) are the rebar mapping result (3 examples among 20 split B-scans), (d) and (e) show the rebar coordinates, (f) displays the rebar amplitude distribution histogram, and (g) exhibits the interpolated 2D contour map.</p>
 <p>Figure 4 shows the output of processing GPR data with CHARISMA, which is the interpolated 2D contour from the entire lanes in the bridge Zone 01. Figure 4 (a) follows the ASTM standard, stating the rebar signals below -6 or -8 dB is suspected to be under corrosive environment. Figure 4 (b) is a rainbow contour map, which is also used in LTBP InfoBridge™. We repeat this process for the other zones to gather all the rebar points on the entire bridge.</p>
@@ -164,7 +164,7 @@ <h2>Discussion<a class="headerlink" href="#discussion" title="Link to this headi
 <p><strong>How do we use CHARISMA to solve the problem?</strong></p>
 <p>CHARISMA successfully measured the reinforcing reflection amplitude on the concrete bridge at Mississippi I-10 to assess its corrosive environment. Our approach begins by leveraging the XML file containing the actual GPR scan coordinates to correlate with the GPR data. Then, we process GPR data based on the raw data analysis, pinpoint rebar coordinates with the K-means clustering algorithm, and concatenate all the rebar points from different regions with the offset correction. Finally, we interpolate the points to populate the grid space, ensuring it matches the dimensions of the bridge. All the details are organized in <a class="reference internal" href="corrosion_case_study.html#code-explanation-corrosion-section"><span class="std std-ref">Code Explanation</span></a> section.</p>
 <p><strong>What limitations have been reminded of?</strong></p>
-<p>Our corrosion assessment method shares the same limitations as the Cover Depth measurement because the code algorithms are largely the same. While the F-K migration and K-means clustering algorithms produce reasonable results, they are not robust and are difficult to automate for data processing. For more details, readers are encouraged to refer to the <a class="reference internal" href="rebar_cover_depth_case_study.html#discussion-reber-cover-depth-section"><span class="std std-ref">Discussion</span></a> section of the Cover Depth Measurement case study.</p>
+<p>Our corrosion assessment method shares the same limitations as the Cover Depth Measurement because the code algorithms are largely the same. While the F-K migration and K-means clustering algorithms produce reasonable results, they are not robust and are difficult to automate for data processing. For more details, readers are encouraged to refer to the <a class="reference internal" href="rebar_cover_depth_case_study.html#discussion-reber-cover-depth-section"><span class="std std-ref">Discussion</span></a> section of the Cover Depth Measurement case study.</p>
 <p>Additionally, the amplitude map is not depth-corrected. In principle, the signal attenuates more as the wave propagates deeper, resulting in weaker amplitudes for deeper rebars. This means our current methodology may struggle to distinguish between corrosive rebars and deep rebars. Several researchers suggest correcting the signal amplitude for depth using the attenuation equation, but there are multiple methods for this correction, and no standard approach has been established. Furthermore, the rebar depths in a bridge are relatively similar, so depth correction may not make a significant difference.</p>
 <p>Comprehensively, this approach still requires manual raw data analysis and setting appropriate parameters for code input, indicating that the code is not fully automated to get the result. Our focus currently lies on utilizing machine learning tools to locate rebars without relying on migration or the K-means clustering algorithm, aiming for complete automation of the corrosion assessment.</p>
 </section>