Concrete Static Stress Estimation Using Computer Vision-Based Digital Image Processing

Recently, most structures are built of concrete, so identification of safety level of concrete structures becomes a critical issue. In other words, current stress level in concrete is an important factor to check the safety level of the structures in service. Although it is clear that a technique for estimating the static stress level of concrete is essential, the method to identify the stress state of the currently used concrete structure is definitely limited.

Therefore, an objective of this study is to develop a static stress estimation technique that can be applied to real concrete structures. This study proposes a method that can measure the static stress level of concrete by incorporating SRM and computer vision-based image processing. Applying a small damage to concrete specimens can release the current state of stress and induce stress field change inside concrete around the damage.

This strain measurement is used in the static stress estimation algorithm developed in this study.

INTRODUCTION

With a small damage to the concrete surface, the current level of static stress in the concrete can be released and the corresponding released deformation in the vicinity of the damage can be identified using the DIC technique. Using DIC, a field-wide strain measurement can be performed, which is then used to estimate the instantaneous static stress in the concrete. This study presents that it is possible to estimate the static stress in concrete using both the SRM stress technique and digital image processing based on computer vision, while minimizing the size of the damage.

Static stress estimation algorithm was developed to identify the optimal static stress in concrete. Using the algorithm and parametric analysis, the experimental errors can be calibrated which may include uneven and inhomogeneous load distribution and the abnormal condition of the concrete specimen.

RESEARCH BACKGROUND

Previous research for concrete stress estimation

Stress Relaxation Method

The developed method was validated in actual existing concrete bridges and the results demonstrated its feasibility with improved reliability compared to theoretical calculations. Owens [12] first introduced the SRM, which used strain gauges to estimate in-situ stresses in concrete. Because concrete is a heterogeneous material and can cause unintended and random reactions during SRM, the core drilling method in concrete was generally preferred for SRM [13-20] rather than the hole drilling method.

17] used vibrating wire strain gauges to measure released strain and proposed calibration coefficient to estimate in-situ stress of concrete. 18] and McGinnis and Pessiki [19] measured static state of stress with a 10% error on prestressed concrete beams using core drilling method. Ruan and Zhang [20] worked on in-situ stress identification using core drilling and influence function and estimated the stress around 10% error on the concrete specimen.

The size of the core was generally large compared to the size of the hole, and the outer diameter of the core in previous researches ranged from 50 mm to 150 mm [13-20]. This may include some difficulties in recovering from an experiment or may cause structural errors in real structures. In general, a strain gauge was used to measure the released strain and some relationship between the measured values ​​was explained to derive the static stress in concrete [20-26].

However, this method uses a limited number of strain gauges, which may result in lower accuracy due to the use of the average value of the limited measurement points. To overcome these limitations, this study intends to introduce a method to measure the entire field around the inflicted damage using a vision-based image processing technique.

Figure 2. Comparison between hole-drilling and core-drilling [22]

Digital image correlation

In this paper, MATLAB software 'Ncorr' developed by Georgia Institute of Technology [40-42] was selected as the DIC software. Ncorr' is an open source 2D DIC MATLAB software and provides an accessible and intuitive graphical user interface (GUI) as shown in the figure. Since the deformation is obtained by the spatial differential of the displacement, it is very sensitive to the noise around the hole. which may occur during drilling.

To reduce the impact of this noise, the hole size should be increased. However, the aim of this study is to estimate the concrete stress with the smallest damage size. Therefore, displacement measurement is used in this study instead of using strain, which should follow the increasing hole size.

Figure 6. Graphical user interface of ‘Ncorr’ [40]

CONCRETE STATIC STRESS ESTIMATION

A detailed description is shown in Table 2. Thus, released displacement field data around the hole due to hole drilling can be obtained from experiments and FE analysis. In other words, any data set can be displayed in the. a given plane and its normal vector penetrating data as shown in figure. Through this feature, a relationship between two data sets can be obtained by calculating the equation of the plane and its normal vector.

The method of least squares is used to determine the normal vector of each plane that consists of the displacement field as shown in this Figure 11. The least squares method can be used to solve for the normal vector of each data, and the general regression equation is shown in Eq. 1) and this matrix has the size of the number of data points by 3. B is the vector consisting of d in Eq. 4) shows the normal vector of each plane and is used to identify the relationship between the two data. 12 shows the finitely aligned planes and normal vectors using coordinate transformation and finitely transformed displacement field data.

In the second step, the optimal voltage value that has the smallest error can be estimated. In other words, after step 1, the optimal stress can be determined when the difference of two displacement field shapes is minimized as in Eq. 𝑈 (𝜎) is then determined when the difference between FE model (𝑈 (𝜎)) and experimental (𝑈 , ) displacement field is minimized.

13(a) shows both shear field data before data processing and here the reference load stress for FE analysis data is an applied load stress during the experiment as shown in Table 2. 13(b) shows the final matched shear field for both data after performing static stress estimation algorithm. The optimal voltage value can be determined when the difference between them is the smallest. a) Original form of experimental and FE analysis data.

Figure 9. Main idea of static stress estimation algorithm

PARAMETRIC ANALYSIS: Effect of hole diameter and depth

15(a) is the graph showing the average displacement value along the hole depth when the hole diameter is 20 mm. The average displacement released due to hole drilling is increasing with increasing hole depth. However, considering that the most common concrete cover in reinforced concrete and prestressed concrete is 50 mm, the depth of the hole is determined by 40 mm.

15(b) is a graph showing the mean displacement along the hole diameter with a fixed hole depth of 40 mm. Only three types of drill bits used in the experiment are sold on the market, each with a diameter of 12, 20 and 30 mm. Since a smaller diameter is required to reduce the size of the damage, a 12 mm drill bit is preferred.

However, the obtained average displacement of this case (diameter of 12 mm) is too small to be measured by DIC.

Figure 15. Mean displacement along change of parameters

EXPERIMENTAL VALIDATION

Experimental setup

The diameter and depth of the hole in the drill hole is 20 mm and 40 mm, respectively, which is determined in the parametric analysis (in chapter 4). The random dot pattern is painted on the concrete surface to measure the deformation using DIC (See Figure 18). The point pattern is preferred to have more randomness to calculate the correlation of each pattern, and the vision-based deformation measurement is applied.

In this study, a DSLR camera is used to take the photos before and after hole drilling and at this moment the distance from concrete sample to camera is about 1m.

Figure 17. Stress-strain curve of concrete specimen

Results and analysis

The displacement measurement area around the hole is determined empirically through several experiments and is illustrated by the yellow circle in Figure 21(b), which shows the final transformed and matched displacement field shapes of the two data, and the stress level of this concrete specimen is determined when the two data are they match well. Pictures before and after drilling the holes and DIC results - type 1_Test3. a) Original form of experimental and FE analysis data.

22 shows the result numbers of the "Type 2" sample before and after hole drilling and DIC treatment. Please note that only part of the yellow circle in Figures 20 and 22 applies to the measurement area, which is the area from Φ 35 mm to Φ 45 mm from the center of the circle. Pictures before and after drilling the holes and DIC results - experiment type 2_Test10,. a) Original form of experimental and FE analysis data.

Some results have large errors of more than 20% and it is assumed that certain uncertainties of concrete property and experimental condition can be included. In addition, when taking photos during experiment, camera may be slightly moved due to small vibration during hole-drilling. Even if the highest error is more than 30%, the proposed method is validated to be able to estimate the static stress state in concrete while reducing the size of damage.

Figure 20. Pictures before and after hole-drilling and DIC results – Type 1_Test3

CONCLUSION

With the further improvements mentioned, this technique will be expected to investigate a safety assessment of concrete structures. In this study, a DIC-based technique was developed for static stress estimation in concrete including the stress relaxation method. With further improvements, this technique is expected to be useful for assessing the safety of aged concrete structures.

Flatjack tests and inverse analysis for identifying stress states and elastic properties in concrete dams. Identification of elastic stiffness and local stresses in concrete dams by in situ tests and neural networks. Theoretical development of the core drilling method for non-destructive evaluation of stresses in concrete structures.

Analytical and numerical development of the incremental core drilling method for nondestructive determination of in situ stresses in concrete structures. An experimental study of the core drilling method for estimating in situ stresses in concrete structures.