Research on Flexural Behavior of Coral Aggregate Reinforced Concrete Beams

Through the flexural behavior test of coral aggregate reinforced concrete beams (CARCB) and ordinary Portland reinforced concrete beams (OPRCB), and based on the parameters of concrete types, concrete strength grades and reinforcement ratios, the crack development, failure mode, midspan deflection and flexural capacity were studied, the relationships of bending moment-midspan deflection, load-longitudinal tensile reinforcement strain, load-maximum crack width were established, and a calculation model for the flexural capacity of CARCB was suggested. The results showed that with the increase in the reinforcement ratio and concrete strength grade, the crack bending moment (Mcr) and ultimate bending moment (Mu) of CARCB gradually increased. The characteristics of CARCB and OPRCB are basically the same. Furthermore, through increasing the concrete strength grade and reinforcement ratio, Mcr/Mu could be increased to delay the cracking of CARCB. As the load increased, crack width (w) would also increase. At the beginning of the loading, w increased slowly. And then it increased rapidly when the load reached to the ultimate load, which then led to beam failure. Meanwhile, with a comprehensive consideration of the effects of steel corrosion on the loss of steel section and the decrease of steel yield strength, a more reasonable calculation model for the flexural capacity of CARCB was proposed.


Introduction
Tropical islands are far from the continent, thus causing the scarcity of normal aggregates and fresh water resources as well as inconvenient traffic. If normal aggregates and fresh water are transported by ships from the mainland, the freight is bound to be higher. And limited by conditions such as wind and waves, the time limit is difficult to guarantee. However, coral and seawater resources are rich on the tropical islands. Corals are porous, strong in water absorption, with CaCO 3 content up to more than 96%, so they belong to the natural light aggregate. For the purpose of reducing the cost, finishing on time and increasing raw material source availability, coral aggregate reinforced concrete beams (CARCB) made of corals, coral sands, seawater and cement can be widely applied to the construction of ports, levees, airports and roads to achieve practical engineering benefits. Clis the main reason for reinforcements corrosion in marine environments, which leads to concrete com-ponent failure (Jin et al., 2004;Song et al., 2008;Angst et al., 2009). Owing to the natural porous structure of corals and the abundant supply of Clin seawater and corals, corrosion easily occurs to the reinforcement in CARCB (Da et al., 2016a(Da et al., , 2016c. Additionally, the flexural behavior of beams is one of the important indicator to ensure the building structural safety (Zhang et al., 2010). Therefore, research on flexural behavior of CARCB is of great practical significance and high practical value to marine engineering structures.
At present, scholars at home and abroad generally focused on the study of ordinary Portland reinforced concrete beams (OPRCB) and lightweight aggregate reinforced concrete beams (LARCB), but studies of coral aggregate concrete (CAC) were few, mainly including its material properties. Liu et al. (2006) explored the flexural behaviors of high-strength LARCB. It was found that strong brittleness in high-strength LARCB during failure and its ductility were worse than those of the OPRCB of the same strength grade. Zhou et al. (2008) and Ji et al. (2012) probed into the differences between the recycled concrete beam and OPRCB on their deformation and failure characteristics. The research demonstrated that the flexural failure mechanisms of the recycled concrete beam and OPRCB were basically the same. Howdyshell (1974) investigated the CAC structure proved that it was feasible to prepare CAC using the coral coarse aggregate and seawater. Rick (1991) explored the three CAC buildings in the Pacific Bikini Atoll, and found that the strength of CAC could meet the design requirements of the engineering structure. Yodsudjai et al. (2003) found that in the Pacific Rim, CAC was used in many USA military facilities, but the study on its mechanical properties was few and its durability did not stood the test of time. Zhang (1995) studied the flexural behavior of CARCB, established a calculation model of CARCB and the rule of selecting its parameters. Whereas corrosion on the common steel adopted had occurred in the mechanical performance tests without any introduced quantitative data. Wang (2013) conducted the feasibility study on CAC of steel tubes, on the basis of the corrosion mechanism of Clon the reinforced quantitative data. Mi et al. (2016) probed into the basic mechanical properties of CAC, and compared with those of OPC and LAC to establish the relationships among the CAC axial compressive strength, splitting tensile strength, flexural strength and cube compressive strength. Da et al. (2016a) carried out the chloride ions diffusion law of CAC, which demonstrated that in the marine environment, diffusion had "three high characteristics"-high C 0 , high C s and high D a . Da et al. (2016b) studied whole stressstrain curves of the CAC. The research demonstrated that CAC had stronger brittleness than OPC and LPC with the same strength grade. Li et al. (2016) studied the flexural behavior of CARCB with composite reinforcements, and obtained the influence rules of the cracking load, ultimate bearing capacity and deflection for the CARCB. Jin (2016) carried out the flexural behavior of CARCB with basalt fiber reinforced polymer (BFRP, and the degradation rule for mechanical property of BFRP-CARCB was deliberated. Da et al. (2016d) and Yu et al. (2017) investigated the durability of CAC in tropical islands, and analyzed the durability states of CAC construction. In consequence, research on the flexural behavior of CARCB was implemented based on the current CAC researches at home and abroad.
In this paper, the flexural behavior of ten CARCB and two OPRCB were tested. The concrete types, concrete strength grades and reinforcement ratios were selected as the parameters. The crack development, failure mode, midspan deflection and flexural capacity were studied. The relationships among the bending moment-midspan deflection, loadlongitudinal tensile reinforcement strain and load-maximum crack width were established. The formulas in JGJ 12-2006 and GB 50010-2010 were proved to be applicable to the calculation model for the CARCB flexural capacity. In addition, the parameter selection rule was also determined. The basic data and theoretical supports were provided for the application of the CAC in marine engineering structures.

Raw materials
Corals  in the tropical islands were adopted, with the Clcontent, apparent density, bulk density and cylindrical compress strength being 0.074%, 2300 kg/m 3 , 1000 kg/m 3 and 5.2 MPa, respectively. Coral sands  in the tropical islands were also adopted, with the Clcontent, silt content, apparent density, bulk density and fineness modulus of 0.112%, 0.5%, 2500 kg/m 3 , 1115 kg/m 3 and 3.5, respectively. It was medium sand. The particle size distribution curves of corals and coral sands are shown in Fig. 1. Other materials were adopted as follows: P·52.5 ordinary Portland cement produced in Nanjing, grade I fly ash (FA) produced in Zhenjiang, S95 grade ground Slag (SG) produced in Nanjing, PCA-I polycarboxylic water reducer and the calcium nitrate inhibitor produced by Jiangsu Research Institute of Building Science Co., Ltd. The seawater was prepared according to the ASTM D1141-2003. Table 1 is the chemical composition of simulated seawater. The reinforcements adopted organic new coated steel developed by Nanjing University of Aeronautics and Astronautics (the coating thickness was 40 μm).

Component design
In this study, the concrete type, concrete strength and reinforcement ratio were selected as the parameters. The flexural behavior of CARCB was tested. Among them, BW1 and BW2 studied the effects of concrete type, BW2-BW4 studied the effects of reinforcement ratio, and BW4-BW6 studied the effects of concrete strength. The dimension of the components was 150 mm×200 mm×1500 mm. The mix proportion of concrete is shown in Table 2. The basic parameters of each component are shown in Table 3 where f cu , f c  Longitudinal reinforcements adopted two rebar steels of Φ12, Φ14 and Φ16. The hanger reinforcements were equipped with two Φ10 rebar steels and stirrups with Φ6 common round steels (double legs) with the spacing between each stirrup as 100 mm. The dimension and reinforcement of the test beam are shown in Fig. 2.

Component fabrication and maintenance
Coral is a natural porous light aggregate which has the characteristics of absorbing water and returning water. After pre-absorption, the coral aggregate could reserve water, which would be released in the process of cement hydration. This would improve the density and bond strength of the cement-coral interface and enhance the resistance to Clpenetration of coral concrete. Therefore, pre-absorption of wa-ter was necessary before preparation of coral concrete. Next, to add the raw materials such as the cements, corals, coral sands, SG and FA into the mixer for 1 min. Then, to add mixed liquor made of water, water reducer and inhibitor into the mixer for 3 min. After discharging, the slump was measured and the pouring and vibration should be done to shape them into concrete specimens. After the beams were completed, they were covered with straws and sprinkled seawater to maintain for 90 d. Related mechanical properties would be tested afterwards. Fig. 3 is a schematic of loading devices. During the loading process, the load changes were measured with a load sensor of 50 t. The readings of the reinforcement strain gauges, concrete strain gauges and displacement sensor were collected by connecting the DH3818-2 static strain meter. With the help of the SW-LW-201 crack observer, the occurrence and development of cracks could be observed and their width could be measured. Five YWC-50 displacement sensors were respectively mounted on both ends of supports of the test beams, at the two loading points and the    MA Hai-yan et al. China Ocean Eng., 2018, Vol. 32, No. 5, P. 593-604 mid-span position to measure the settlement displacement at the two supports and the deflection at the midspan, therefore to obtain the overall deformation information of the test beams.

Loading methods and test point arrangement
In order to measure the steel strain during the loading process, six strain gauges were attached to the longitudinal reinforcements midspan at the beam bottom and loading points, respectively. In order to measure the test beams deformation (concrete strain), five strain gauges were arranged in the pure bending section of the beams to verify whether the plane section assumption was satisfied by the CARCB cross section strain. The tension zone and compression zone were attached with one strain gauge respectively in favor of probing into the concrete strain development in those zones. 10 kN was preloaded to check if every part of the load system was in good contact, and if the instruments are operated properly. During the formal loading, the test beams were loaded at 1 kN per stage before cracking. After cracking, the test beams were loaded at 5 kN per stage until all cracks occurred. The test beams were loaded at 2 kN per stage from the occurrence of all cracks to the tests end.

Test results and analysis
3.1 Crack development Fig. 4 demonstrates the crack development of CARCB and OPRCB. It can be observed from Fig. 4 that flexural failure occurrences of OPRCB and CARCB share the same type of modes. The development laws of cracks in each CARCB were almost the same, and their characteristics were as follows. (1) The cracking was usually accompanied by a slight noise, and the cracks were mostly distributed at the loading point and midspan. (2) The first crack usually ran through 30%-70% of the beam height. As the load increased, the crack extended upward along the beam, and its length increased slowly in the later stage. Meanwhile, the initial crack width slowly increased, and it then rapidly increased to a width of more than 1.5 mm after the reinforcement yielding occurred. (3) When new cracks appeared, the bearing capacity dropped to approximately 2-5 kN.
3.2 Bending moment-midspan deflection curves Fig. 5 displays the bending moment-midspan deflection curves of CARCB and OPRCB. As shown in Fig. 5a: (1) For the same load level, CARCB showed a larger deflection than OPRCB, which indicated that CARCB had a bet-  ter ductility than OPRCB, and this was beneficial for CAR-CB in handling cracks.
(2) For the same reinforcement ratio, the ultimate bending moment (M u ) of BW2 was close to that of BW1, and it could suggest that a C25 CARCB exhibited the same anti-crack performance as a C35 OPRCB. (3) For the same load level, with the increase in the reinforcement ratio, the midspan deflection of CARCB gradually decreased, but M u gradually increased, because the ultimate bearing capacity of CARCB was shared by the tensile reinforcement and concrete. When the bearing capacity of concrete was constant, the bearing capacity of the tensile reinforcement increased, and M u of CARCB increased, which was consistent with the law of OPRCB.
As shown in Fig. 5b, for the same load level, as the concrete strength grade increases, the midspan deflection of CARCB decreases gradually, but M u increases gradually. This is because the ultimate bearing capacity of CARCB is shared by the tensile reinforcement and concrete. When the bearing capacity of the tensile reinforcement was constant, the bearing capacity of concrete increased, and M u of CAR-CB increased, which was consistent with the law of OPRCB.

Characteristics through the process
In Fig. 5, when the reinforcement ratio or concrete strength grade was not too large, the twelve test beams all exhibited the failure characteristics of under-reinforced beams, the deformation characteristics of which could be divided into four stages.
(1) Stage 1: This stage is an elastic stage from the start of loading to beam cracking. As shown in Fig. 4, the results showed that: at the initial stage of loading, the bending moment (M) and strain (ε) of beam were very small. The bending moment-midspan deflection curves ascended linearly, and the concrete was in the elastic working stage. A positive proportional relationship between the stress (σ) and ε existed. The characteristics of CARCB and OPRCB were basically the same. When M reached about M cr , the concrete tensile strain of CARCB in the tension zone reached to the concrete ultimate tensile strain (ε cu ). Small cracks began to appear in the tension zone, and the tensile zone presented a larger plastic deformation.
(2) Stage 2: When M reached M cr , after increasing load continuously, the first crack occurred in the concrete section in the pure bending section or near the loading points. As the load increased continuously, the first crack occurred at the section of the concrete that had a small f t within the pure bending section or near the loading point. The crack ran through 30%-70% of the beam height. The first crack in CARCB was longer than that in OPRCB. The reasons for this were as follows: (a) Although the strength of coral aggregate is low, coral aggregate has the characteristics of "absorbing water and returning water", which makes the interfacial strength higher. Therefore, CAC has no obvious weakness, and the extension of the fracture surface was unaffected by the aggregate. (b) The brittleness of CAC was greater than that of OPC . When M reaches about 40%-80% of M u , the cracks are basically all present, and the maximum crack width (w s ) is 0.1-0.8 mm. When M reaches about 80%-90% of M u , the tensile steel bar yields, and it comes to the end of this stage.
(3) Stage 3: This stage was the failure stage. After the yielding of the steel bar, σ of the reinforcement remained constant or increased slightly, and the increment of M could only be balanced by adding the arm of force. At this time, the cracks extended quickly, the crack width (w) as well as the deflection of the beam increased rapidly. A principle crack of large width was formed in the section where the reinforcement yielded in the pure bending section, developing quickly to the top of the beam.
(4) Stage 4: When the load continued to increase, in a certain area on both sides of the principle crack on the top of the beam, the concrete in the compression zone produced large plastic deformation, forming a concentrated plastic deformation zone. At this time, ε of the concrete compression zone had a great increase, the deflection of the beam increased sharply, then the concrete was crushed in the compression zone, and the beam failed. Table 4 presents the test results of CARCB and OPRCB. In Table 4, M cr /M u of CARCB is between 0.08-0.28, and with the increase of the concrete strength grade and reinforcement ratio, M cr /M u gradually increases. It shows that the cracking time of CARCB can be delayed by increasing the concrete strength grade and reinforcement ratio. Through the analysis of the experimental results of Zhang (1995), it is found that the ultimate compressive strain (ε cu ') of BW4-1 is more reasonable. Therefore, the epsilon ε cu ' is 3470 in the calculation model in this paper.

Bearing capacity analysis
As shown in Table 4, M cr and M u of CARCB with various reinforcement ratios followed the rule: BW4 > BW3 > BW2. M cr and M u of CARCB with various concrete strength grades followed the rule: BW6 > BW5 > BW4. It was shown that M cr and M u of CARCB increased with the improvement of the reinforcement ratio and concrete strength grade. The reasons for this are as follows. (1) Before the cracking, the tensile stress was mainly undertaken by the concrete, and it had a close relationship with f t of CAC. However, with the increase in the concrete strength grade, f t also increased resulting in the increase of M cr . (2) Before the cracking, the tensile reinforcement might share a part of the tensile stress, The reinforcement ratio increased, and the tensile strength of reinforcement was enhanced, so M cr increased. (3) The bearing capacity of CARCB was comprised of that of the reinforcement and concrete. With the increase of the reinforcement ratio, the bending ability of the reinforcement increased gradually, so M u increased. At the same time, with the increase of the concrete strength grade, f t of CAC increased, so M u increased. Fig. 6 shows the load-longitudinal tensile reinforcement strain curves of CARCB. The strain and deformation law of the tensile reinforcement reflected the stress state of the beam. The strain at the loading point and midspan of the tensile reinforcement was analyzed, thus laying a solid foundation for the calculation model of the CARCB flexural capacity.

Reinforcement strain analysis
3.6 Crack width Fig. 7 illustrates the load-maximum crack width curves of CARCB. As shown in Fig. 7, the crack width (w) increased with the increase in load. At the beginning of the loading, w increased slowly, and then it increased rapidly as the load was close to the ultimate load, eventually leading to the beam failure. The relationship between w and the load was similar to the trend of the logarithmic function during the development of the cracks. Besides, for CARCB with the same w, when w = 0.5 mm, the loads of BW2-1, BW3-2, BW4-2, BW5-1 and BW6-2 were 46 kN, 79 kN, 90 kN, 118 1. ρ is the reinforcement ratio; 2. M cr is the crack bending moment, and M u is the ultimate bending moment; 3. f max is the maximum deflection; 4. w s is the maximum crack width, and l m is the average crack space; 5. ε cu is the ultimate tensile strain. The meanings of other parameters are the same as those in Table 3. kN and 103 kN, respectively. It was shown that the development of the crack width was related to the concrete strength grade and reinforcement ratio. Obviously, with the increase in the concrete strength grade and reinforcement ratio, the development of the crack width gradually slowed down. Fig. 8 presents the average crack space (l m ) of CARCB and OPRCB. As shown in Fig. 8a, for CARCB and OPRCB with the same mixing proportion, l m of CARCB was 3%-10% smaller than that of OPRCB. This is mainly because the coral aggregate is natural porous structure material, which has the characteristic of absorbing water and returning water. It reduces W/B around the coral aggregate, and improves the density and bond strength of the cement paste in the interface area. Therefore, the interface bond strength of CAC was higher than that of OPC (Da et al., 2016b. As shown in Figs. 8a and 8b, l m of CARCB decreased with the increase in the reinforcement ratio and concrete strength grade. As the reinforcement ratio and concrete strength grade increased, the flexural capacity and crack resistance of CARCB were significantly enhanced.

Crack space
3.8 Calculation model 3.8.1 Influence of steel corrosion Fig. 9 is the corrosion state of different reinforcements of CARCB, and the test time of which is 90 d. As shown in Fig. 9, the organic new coated steel had local corrosion. Calculation of the mass loss rate of steel based on Standard for test methods of long-term performance and durability of ordinary concrete (GB 50082-2009) (Table 4). At the same time, the mean section loss rate (ω s ) and maximum section loss rate (ω sm ) of BW5-1 are studied by means of 3D scanning technology ( Fig. 9 and Table 5). The research (Zhang et al., 2014) shows that: The effect of steel corrosion on the reinforced concrete beam is mainly the decrease of the steel section and steel yield strength. Eq. (1) is the calculation method of the comprehensive reduction coefficient (α) of the steel section loss and the decrease of the steel yield strength in consideration of steel corrosion (Yuan et al., 2001): (1) Fig. 7. Load-maximum crack width curves of CARCB.   (Zhang et al., 2014) is used to study ω s and ω sm . The specific analysis methods are as follows: 3 non-corrosion and corrosion steels with a length of 30 cm were scanned, and 3D entity model is obtained and the Geomagic Studio software is used to extract the cross section area at different locations. In the end, ω s and ω sm are calculated.
where ω s is the mean section loss rate of steel (%), α 2 is the reduction coefficient of the steel section loss in consideration of steel corrosion, and α 3 is the reduction coefficient of the steel yield strength decrease in consideration of steel corrosion. Fig. 10 is the relationship between the maximum section loss rate (ω sm ) and mean section loss rate (ω s ) of steel. As shown in Fig. 10, there is a good linear relationship between ω sm and ω s : ω sm = 1.301ω s . (2) In the chlorine salt erosion environment, the steel had different degrees of "pit erosion". Therefore, for the steel bar, the minimum section of the steel bar section is the most easily broken, that is, the position of ω sm is the most easily broken. Hence it is more accurate to use ω sm to calculate α of the steel section loss and the decrease of the steel yield strength in consideration of steel corrosion. In addition, the research shows that (Zhang et al., 2014;Yuan et al., 2001;Wang and Zhong, 2005), the mass loss rate (ω) and ω s are roughly equal. Therefore, 3D scanning technology is adopted to get the relationship between the maximum section loss rate (ω sm ) and ω of steel. With a comprehensive consideration of the effects of steel corrosion on the loss of the steel section and the decrease of the steel yield strength, the comprehensive reduction coefficient (α) is expressed as:

Relative height of the limiting compression zone
In Specification for Design of Lightweight Aggregate Concrete Structures (JGJ 12-2006), Code for Design of Concrete Structures (GB 50010-2010) and Zhang (1995), the calculation formulas for the relative height of the limiting compression zone (ξ b ) are as follows.
For LARCB, it is suggested in JGJ 12-2006 that: For OPRCB, it is suggested in GB 50010-2010 that: For CARCB, Zhang (1995) suggested that: where, β 1 is the coefficient of the rectangular stress diagram. f y is the reinforcement tensile strength (MPa). ε cu is the concrete ultimate compressive strain, with ε cu = 0.0033 -(f cu -50)×10 -5 . When ε cu ≥ 0.0033, ε cu = 0.0033. In this situation, f cu is the concrete cubic compressive strength (MPa). E s is the elastic modulus of the reinforcement (GPa). Therefore, considering the effect of the reinforcement corrosion on the decrease of the steel yield strength, the suggested calculation formula for ξ b of CARCB is as follows: where, β 1 = 0.831 (Mi et al., 2016), ε cu =3470 με, α 3 is the reduction coefficient of the steel yield strength decrease in consideration of steel corrosion, and ω is the mass loss rate (%).

Ultimate bending moment
In JGJ 12-2006, GB 50010-2010and Zhang (1995, respectively, the calculation formula for the ultimate bending moment M u is as follows.
For OPRCB, it is suggested in GB 50010-2010 that: 19.4 20.9 Notes: 1. A w is the elongation of steel; 2. ω is the mass loss rate, and ω sm is the maximum section loss rate. The length of reinforcement used to measure ω and ω sm were 10 cm and 30 cm, respectively. The meanings of other parameters are the same as those in Table 3. Fig. 10. Relationship between the maximum section loss rate and mass loss rate.
where, M u is the ultimate bending moment (kN·m); α 1 is the coefficient of the rectangular stress diagram; f c is the concrete axial tensile strength (MPa); b is the width of the rectangular section (mm); h 0 is the effective height of the rectangular section (mm); x is the height of the concrete compression zone (mm); and A s is the sectional area of the longitudinal reinforcement in the tension zone (mm 2 ).
For LARCB and CARCB, it is suggested in both JGJ 12-2006 andZhang (1995) that: Supplementary condition: where f cm is the concrete flexural compressive strength (MPa). Other parameters are the same as those in Eq. (8).
of CARCB was calculated using Eq. (8) and Eq. (10) and compared with of CARCB as shown in Table 6 and Fig. 11. As shown in Fig. 11, of CARCB well coincided with obtained from Eq. (8) and Eq. (10), of which are similar to each other. However, f cm was used in Eq. (10), and its use in the current JGJ 12-2006 and GB 50010-2010 had been eliminated. Therefore, with a comprehensive consideration of the effects of steel corrosion on the loss of the steel section and the decrease of the steel yield strength, the proposed calculation formula for M u of CARCB is as follows: where α is the comprehensive reduction coefficient of the steel section loss and the decrease of the steel yield strength in consideration of steel corrosion.

Crack width
In JGJ 12-2006 and GB 50010-2010, the calculation formulas for the average crack width (w m ) and maximum crack width (w s ) of LARCB and OPRCB, respectively, are as follows: Zhang (1995) suggested that the calculation formulas for w m and w s of CARCB should be: where, α c is the influence coefficient of the crack width, and α c = 0.85; ψ is the nonuniform coefficient of the tensile strain between the cracks. When ψ < 0.2, ψ = 0.2; when ψ > 1.0, ψ = 1.0; ρ te is the reinforcement ratio of the longitudin-al tensile reinforcement (%); A te is the effective section area of the tensile concrete (mm 2 ); σ s is the longitudinal tensile stress (MPa); and E s is the elastic modulus of the reinforcement (GPa).  Table 6 and of CARCB under different calculation models 12-13, it can be observed that a large deviation exists between the curve plotted using Eq. (14), Eq. (15) and the measured one. Therefore, considering the effect of reinforcement corrosion on the loss of the steel section, the proposed calculation formulas for w m and w s of CARCB are as follows: where, α 2 is the reduction coefficient of the steel section loss in consideration of steel corrosion. The calculation method of ψ was consistent with Eq. (14) and Eq. (15).

Conclusions
(1) For the same load level, CARCB had a larger deflection than that of OPRCB, which indicated the better ductility of CARCB. For the same load level, with the increase in the reinforcement ratio and concrete strength grade, the midspan deflection of CARCB decreased gradually, but M cr and M u increased gradually. Meanwhile, by increasing the concrete strength grade and reinforcement ratio, M cr /M u could be increased to delay the cracking of CARCB.
(2) All CARCBs exhibited the failure characteristics of under-reinforced beams, the deformation characteristics of which could be divided into four stages. The characteristics of CARCB and OPRCB are basically the same.
(3) The crack width (w) increased with the increase in the load. At the start of the loading, the crack width increased slowly. And then it increased rapidly when reaching to the ultimate load, thus leading to the beam failure. With the increase in the reinforcement ratio and concrete strength, the extension of w gradually slowed down and effectively inhibited the development of cracks.
(4) The application of new organic coated steels and C60 CAC in the CAC structure was suggested to effectively inhibit steel corrosion and extend service life in the  marine engineering structure.
(5) With a comprehensive consideration of the effects of steel corrosion on the loss of the steel section and the decrease of the steel yield strength, a more reasonable calculation model for ξ b , M u , w m and w s of CARCB was proposed, and the law for selecting the parameters was also determined.