Pilot Demonstration of a Strengthening Method for Steel-Bolted Connections Using Pre-Formable Carbon Fiber Cloth with VaRTM

In recent years, many seismic retrofitting methods have been performed to improve the structural performance and prevent the brittle failure of structural members. In the case of steel structures, slender seismic braces have been widely used for buildings, towers, and bridges. The brace connections should resist the full plastic axial tension load to ensure adequate plastic deformation performance for vibration energy absorption. However, certain connections do not satisfy these requirements. Recently, carbon fiber reinforced plastic (CFRP) has been used extensively to strengthen existing structures because of its high-strength, high elastic modulus, and light-weight characteristics. In this paper, we investigate the applicability of CFRP strengthening for brace connections and gusset plates with stepped surfaces using the vacuum-assisted resin transfer molding technique as a pilot demonstration. Stepped surfaces can be eliminated by using alternative CFRP layers to straighten the structural CFRP layers in order to effectively transfer the axial stress. Eventually, it is shown that CFRP strengthening can improve the connection strength and plastic deformation with 3% elongation, even if the CFRP is molded on the stepped surface.


Research Background
Steel structures have excellent ductile performance due to their plastic deformability, in addition to having good seismic resiliency and energy absorption. Thus, steel materials have been widely used for building structures, latticed towers, large span bridges, highrise buildings, and spatial structures. On the other hand, to prevent structural damage caused by severe earthquakes [1,2], seismic retrofitting has been rapidly applied to existing structures [3,4]. In the case of steel structures, it has been reported that several seismic brace members lack the connection strength-especially in terms of the load-carrying capacity of the net area at the bolt hole and the gusset plate connection-needed to reach axial full plastic capacity [5]. Although energy absorption via plastic deformation is considered in the seismic design of steel braces, brittle failure in the net area without plastic deformation of the gross sectional area often occurs. Therefore, several strengthening methods have been proposed to prevent brittle failure in the net area [6]. Member ductility of 8.5 is recommended to satisfactory perform seismic energy absorption.
To ensure the seismic performance of steel structures, several approaches can be taken. Sarno et al. [3] investigated the effects of concentrated bracing, mega-brace systems, strengthening effects. Finally, it is shown that the proposed CFRP strengthening method can be successfully applied to steel brace connections, even for stepped and bumpy bonding surface, showing that the load-carrying capacity of the connections and ductility of the brace members can be effectively improved.

Strengthening Method Using CFRP by VaRTM
VaRTM has been applied to mold large FRP products, such as the blades of wind turbines and monocoque automobiles and footbridges [22], because of its productivity, quality, and usability. The authors have developed an adhesively bonding and strengthening method by VaRTM through the modification of the VaRTM technique. Figure 1 shows a simplified diagram of the adopted CFRP strengthening method using VaRTM that the authors developed [21]. Usually, metallic frames are used to mold FRP products by VaRTM, and the frames are dismounted after curing. On the other hand, the adopted CFRP strengthening method using VaRTM uses existing steel surface as a molding frame. First, the steel surface is treated to remove paint and make a rough surface for adhesive bonding. It has been recommended that the surface roughness should be higher than 20 micro-strain of ten-point mean roughness [21]. Second, primer resin is applied to the steel surface to avoid galvanic corrosion by carbon fiber and steel. After curing of the primer resin, carbon fiber cloth ais set up onto the steel surface. Next, the molding utilities for VaRTM (peel ply, resin flow material, seal, and bagging film) are attached. Finally, vacuuming and resin injection are carried out. By using this method, carbon fiber cloth can be molded and bonded directly onto various steel surfaces, even if many carbon fiber cloths are required for strengthening. Thus, the molding time can be greatly shortened, and stable mechanical performance can be obtained compared to the hand-layup molding method. Additionally, carbon fiber cloth can be easily molded for vertical surfaces, such as steel brace connections. Moreover, pre-formable carbon fiber cloth has been developed in recent years to allow molding with a large number of carbon fiber cloth layers and to reduce the imperfections caused by cloth slippage. To make the pre-formable carbon fiber cloth, thermoplastic powder was applied onto carbon fiber cloth, as shown in Figure 2. Thus, the cloth can be tentatively formed by heating because thermoplastic can be remolded easily by heating. In this study, we stacked carbon fiber cloth in layers and unified them by heating to ease the setup of the carbon fiber cloth on the steel surface. The details of the strengthening procedure for the targeted structure in this research are introduced in Section 2.3.
Materials 2021, 14, x FOR PEER REVIEW 3 of 17 for steel brace connections. Next, a CFRP strengthening design method using VaRTM is proposed. Then, we demonstrate the proposed CFRP strengthening method and investigate the strengthening effects. Finally, it is shown that the proposed CFRP strengthening method can be successfully applied to steel brace connections, even for stepped and bumpy bonding surface, showing that the load-carrying capacity of the connections and ductility of the brace members can be effectively improved.

Strengthening Method Using CFRP by VaRTM
VaRTM has been applied to mold large FRP products, such as the blades of wind turbines and monocoque automobiles and footbridges [22], because of its productivity, quality, and usability. The authors have developed an adhesively bonding and strengthening method by VaRTM through the modification of the VaRTM technique. Figure 1 shows a simplified diagram of the adopted CFRP strengthening method using VaRTM that the authors developed [21]. Usually, metallic frames are used to mold FRP products by VaRTM, and the frames are dismounted after curing. On the other hand, the adopted CFRP strengthening method using VaRTM uses existing steel surface as a molding frame. First, the steel surface is treated to remove paint and make a rough surface for adhesive bonding. It has been recommended that the surface roughness should be higher than 20 micro-strain of ten-point mean roughness [21]. Second, primer resin is applied to the steel surface to avoid galvanic corrosion by carbon fiber and steel. After curing of the primer resin, carbon fiber cloth ais set up onto the steel surface. Next, the molding utilities for VaRTM (peel ply, resin flow material, seal, and bagging film) are attached. Finally, vacuuming and resin injection are carried out. By using this method, carbon fiber cloth can be molded and bonded directly onto various steel surfaces, even if many carbon fiber cloths are required for strengthening. Thus, the molding time can be greatly shortened, and stable mechanical performance can be obtained compared to the hand-layup molding method. Additionally, carbon fiber cloth can be easily molded for vertical surfaces, such as steel brace connections. Moreover, pre-formable carbon fiber cloth has been developed in recent years to allow molding with a large number of carbon fiber cloth layers and to reduce the imperfections caused by cloth slippage. To make the pre-formable carbon fiber cloth, thermoplastic powder was applied onto carbon fiber cloth, as shown in Figure 2. Thus, the cloth can be tentatively formed by heating because thermoplastic can be remolded easily by heating. In this study, we stacked carbon fiber cloth in layers and unified them by heating to ease the setup of the carbon fiber cloth on the steel surface. The details of the strengthening procedure for the targeted structure in this research are introduced in Section 2.3.

Experimental Methods
This section shows the design requirements for seismic resisting steel braces, not only in terms of load-carrying capacity, but also in terms of ductile performance. The calculation method was used to predict of the number of layers of carbon fiber cloth needed to avoid a stepped surface and to transfer axial force.

Seismic Resisting Steel Braces and Its Seismic Performance Evaluation
The Japan Building Disaster Prevention Association (JBDPA) established an evaluation guideline for the seismic performance of existing steel structures [6]. According to their method, the load-carrying capacity u P of a seismic brace for existing steel structures under tensile force can be evaluated using the following equations: 1  2  3  4 min , , , u P P P P P  (1)


(1) Ultimate shear strength of bolts: where, m is a total number of shear planes; n is a total number of bolts; b A is sectional area of bolt; fu F is tensile strength of bolt (1000 N/mm 2 ).  (2) Tensile strength of brace in net area: where, e A is a net area of brace; u F is tensile strength of brace (400 N/mm 2 for SS400 steel); A is sectional area of brace; where, e is end distance; Thermoplastic powders

Experimental Methods
This section shows the design requirements for seismic resisting steel braces, not only in terms of load-carrying capacity, but also in terms of ductile performance. The calculation method was used to predict of the number of layers of carbon fiber cloth needed to avoid a stepped surface and to transfer axial force.

Seismic Resisting Steel Braces and Its Seismic Performance Evaluation
The Japan Building Disaster Prevention Association (JBDPA) established an evaluation guideline for the seismic performance of existing steel structures [6]. According to their method, the load-carrying capacity P u of a seismic brace for existing steel structures under tensile force can be evaluated using the following equations: • (1) Ultimate shear strength of bolts: where, m is a total number of shear planes; n is a total number of bolts; A b is sectional area of bolt; f F u is tensile strength of bolt (1000 N/mm 2 ).

•
(2) Tensile strength of brace in net area: where, A e is a net area of brace; F u is tensile strength of brace (400 N/mm 2 for SS400 steel); A is sectional area of brace; d 0 , t 2 , h n , and t 1 are shown in Figure 3 (h n = 0.7 for n = 2); n B is a total number of braces.

•
(3) Shear-out strength of brace or gusset plate: m is a number of bolts on axial direction; p is bolt pitch; n is a number of bolts on transverse direction; t is thickness of brace or gusset plate; F u is tensile strength of brace or gusset plate (400 N/mm 2 for SS400 steel).
• (4) Ultimate strength of gusset plate in net area: where, A g is net area of gusset plate.
On the other hand, the JBDPA requires that the connection strength satisfies the following strength to resist the full plastic axial load-carrying capacity and large plastic deformation with hysteresis energy absorption: where, A is sectional area of brace; F is nominal yield stress of steel (235 N/mm 2 for SS400 steel).
where, g A is net area of gusset plate.
On the other hand, the JBDPA requires that the connection strength satisfies the following strength to resist the full plastic axial load-carrying capacity and large plastic deformation with hysteresis energy absorption: where, A is sectional area of brace; F is nominal yield stress of steel (235 N/mm 2 for SS400 steel).
(a) (b) However, several existing steel brace connections designed using older Japanese seismic design codes do not satisfy the JBDPA requirements, especially in terms of the tensile strength of the brace in net area 2 P . This is because satisfaction of Equations (1)-(6) was not required for structural design of the brace connections. In the case of a lack of tensile strength of the brace in the net area, breakage occurs without ductile behavior, as shown in Figures 4 and 5a. Therefore, extensive brittle failure has been reported in severe earthquakes [23]. On the other hand, although sectional loss from bolt holes occurs at the connections, yielding in gross sectional areas can be observed when the breakage load in the net area is higher than the yielding load at the gross sectional area, as shown in Figures 4 and 5b. Thus, several retrofit methods have been proposed, and seismic retrofitting for net areas has been rapidly applied according to the evaluation and retrofitting guidelines [6]. However, previous retrofitting method have required welding or alternative bolt holes being added on site to connect additional steel elements. This means that heavy However, several existing steel brace connections designed using older Japanese seismic design codes do not satisfy the JBDPA requirements, especially in terms of the tensile strength of the brace in net area P 2 . This is because satisfaction of Equations (1)- (6) was not required for structural design of the brace connections. In the case of a lack of tensile strength of the brace in the net area, breakage occurs without ductile behavior, as shown in Figures 4 and 5a. Therefore, extensive brittle failure has been reported in severe earthquakes [23]. On the other hand, although sectional loss from bolt holes occurs at the connections, yielding in gross sectional areas can be observed when the breakage load in the net area is higher than the yielding load at the gross sectional area, as shown in Figures 4 and 5b. Thus, several retrofit methods have been proposed, and seismic retrofitting for net areas has been rapidly applied according to the evaluation and retrofitting guidelines [6]. However, previous retrofitting method have required welding or alternative bolt holes being added on site to connect additional steel elements. This means that heavy materials and tools must be used, especially as welding cannot be conducted in certain factory buildings due to fire risks.
Materials 2021, 14, x FOR PEER REVIEW 6 of 17 materials and tools must be used, especially as welding cannot be conducted in certain factory buildings due to fire risks.

Specimens
Based on the above, we adopted a steel-bolted connection with a lack of tensile strength in the net area, as shown in Figure 6. We adopted SS400 angle steel with a 50 × 50 × 4 mm 3 sectional size as a brace and SS400 steel plates of 16 mm thickness and 100 mm width as gusset plates. Several punch marks of 100 mm pitch were created on the steel brace to evaluate plastic deformation after testing, as shown in Figure 6. M16 highstrength steel bolts were used to connect the brace and gusset plates. The minimum yield stress and tensile strength values for SS400 steel is 245 N/mm 2 and 400 N/mm 2 , respectively [24]. The steel plate around the bolt holes was welded to prevent failure in the net area at the right-side connection. Table 1 shows the load-carrying capacity as evaluated using Equations (2)- (6). Thus, the specimen will break in the net area without plastic deformation in the gross section. Table 2 shows the experimental parameters-without CFRP strengthening specimen (namely NS) and two type of CFRP strengthening specimens (namely CFS-full and CFS-

Specimens
Based on the above, we adopted a steel-bolted connection with a lack of tensile strength in the net area, as shown in Figure 6. We adopted SS400 angle steel with a 50 × 50 × 4 mm 3 sectional size as a brace and SS400 steel plates of 16 mm thickness and 100 mm width as gusset plates. Several punch marks of 100 mm pitch were created on the steel brace to evaluate plastic deformation after testing, as shown in Figure 6. M16 highstrength steel bolts were used to connect the brace and gusset plates. The minimum yield stress and tensile strength values for SS400 steel is 245 N/mm 2 and 400 N/mm 2 , respectively [24]. The steel plate around the bolt holes was welded to prevent failure in the net area at the right-side connection. Table 1 shows the load-carrying capacity as evaluated using Equations (2)- (6). Thus, the specimen will break in the net area without plastic deformation in the gross section. Table 2 shows the experimental parameters-without CFRP strengthening specimen (namely NS) and two type of CFRP strengthening specimens (namely CFS-full and CFS-

Specimens
Based on the above, we adopted a steel-bolted connection with a lack of tensile strength in the net area, as shown in Figure 6. We adopted SS400 angle steel with a 50 × 50 × 4 mm 3 sectional size as a brace and SS400 steel plates of 16 mm thickness and 100 mm width as gusset plates. Several punch marks of 100 mm pitch were created on the steel brace to evaluate plastic deformation after testing, as shown in Figure 6. M16 high-strength steel bolts were used to connect the brace and gusset plates. The minimum yield stress and tensile strength values for SS400 steel is 245 N/mm 2 and 400 N/mm 2 , respectively [24]. The steel plate around the bolt holes was welded to prevent failure in the net area at the right-side connection. Table 1 shows the load-carrying capacity as evaluated using Equations (2)- (6). Thus, the specimen will break in the net area without plastic deformation in the gross section.
ber cloth, BT70-20 (Toray Industries, Tokyo, Japan), to strengthen steel connections. UM46-40P carbon cloth was used for removing steps and bumps, and for stress transfer. The BT70-20 bi-directional carbon cloth was used for improvement of the bonding area on the gusset plate because the gusset plate is usually wider than the brace. UM46-40P carbon fiber cloth is a commercial product, which has an adhered thermoplastic powder on the surface to enable preliminary forming by heating, as shown in Figure 2. Table 3. shows the specifications for the steel and carbon fiber cloths.   Table 2 shows the experimental parameters-without CFRP strengthening specimen (namely NS) and two type of CFRP strengthening specimens (namely CFS-full and CFSloss). We adopted two types of CFRP strengthening-the gross sectional strengthening model and sectional loss strengthening model. The total number of CFRP layers for the gross sectional strengthening model was calculated by the strength of the gross sectional area of steel brace, while the sectional loss model multiplied the sectional area of the bolt hole by the strength of the steel. Thus, CFRP of the gross sectional strengthening model has an equivalent strength of steel brace, while the CFRP strength of the sectional loss strengthening model corresponds to the strength of lost area of steel by bolt hole. Equations (7) and (8) show the calculation method of total number of CFRP layers for stress transfer layer.
Gross sec tional strengthening model;  We adopted a unidirectional, medium-elasticity carbon fiber cloth designated as UM46-40P (Toray Industries, Tokyo, Japan) and a bi-directional, high-strength carbon fiber cloth, BT70-20 (Toray Industries, Tokyo, Japan), to strengthen steel connections. UM46-40P carbon cloth was used for removing steps and bumps, and for stress transfer. The BT70-20 bi-directional carbon cloth was used for improvement of the bonding area on the gusset plate because the gusset plate is usually wider than the brace. UM46-40P carbon fiber cloth is a commercial product, which has an adhered thermoplastic powder on the surface to enable preliminary forming by heating, as shown in Figure 2. Table 3. shows the specifications for the steel and carbon fiber cloths.  [25] 440 Gpa -2400 Mpa BT70-20 [25] 230 GPa -2900 MPa * 1 Assumption (general value), * 2 From manufacturer certificate.
In the steel bolted connection, stepped surfaces are commonly found, as shown in Figure 7. To strengthen a steel connection with a stepped surface, the step between the gusset plate and brace surface should be minimized to reduce the bending force applied to the CFRP layer. Additionally, the CFRP should transfer the axial force to prevent strain concentration around the bolt hole. Thus, we decided on the number of carbon fiber layers needed to minimize step e using Equation (9). Next, the axial and bending stress values for CFRP under axial force N were evaluated by Equation (10). The variables in Equations (9) and (10) are also shown in Figure 7. To determine the number of carbon fiber layers with fiber volume content V f variation, we assumed that the fiber volume content V f would vary by 45% to 55% based on the previous research [21]. BT70-20 bi-directional carbon cloth is used on outermost layers to expand the width of the bonding area and effectively transfer the stress widely because the gusset plate is usually wider than the brace.
where, e is remaining step after CFRP strengthening; t brace is thickness of steel brace; t UM is thickness of carbon fiber cloth UM46-40P; n step is a total number of carbon fiber layers required for step minimization; n bolt is a total number of carbon fiber layers required for bump minimization; V f is carbon fiber volume content; b CFRP is width of CFRP; t CFRP is thickness of CFRP; n st is a total number of carbon fiber layers for required axial force transfer. Eventually, the total number of carbon fiber layers needed to minimize step n step was determined as 33 layers (9 layers to adjust the brace thickness, and 24 layers to adjust the height of the bolt head). The total number of carbon fiber layers needed to transfer the axial force n ST was determined as 13 layers for the gross sectional strengthening model and 4 layers for the sectional loss strengthening model. Figure 8 shows the lamination conditions. Bonding length is important to perform the strengthening effect and to avoid debonding failure. However, only a few hundred-millimeter bonding length can be used for the bonding on gusset plate. Thus, we decided the bonding length should be 150 mm for the gusset plate area because the actual brace connections lack the applicable surface area to adhesively bond the CFRP. However, gusset plates are wider than the brace width, so we applied wider carbon fiber cloth for the step adjustment and alternative bi-directional high-strength carbon fiber cloth on the outermost layer to expand the width of the bonding area shown in Figure 7. On the other hand, the brace members have sufficient length to allow for bonding, while it is known that the bonding capacity will peak with increasing the bonding length [18]. Thus, we decided the bonding length should be 300 mm for the brace side with a 150 mm uniform carbon fiber layer length and a 150 mm taper length based on the previous fundamental bonding test results [21]. Eventually, the total number of carbon fiber layers needed to minimize step step n was determined as 33 layers (9 layers to adjust the brace thickness, and 24 layers to adjust the height of the bolt head). The total number of carbon fiber layers needed to transfer the axial force ST n was determined as 13 layers for the gross sectional strengthening model and 4 layers for the sectional loss strengthening model. Figure 8 shows the lamination conditions. Bonding length is important to perform the strengthening effect and to avoid debonding failure. However, only a few hundred-millimeter bonding length can be used for the bonding on gusset plate. Thus, we decided the bonding length should be 150 mm for the gusset plate area because the actual brace connections lack the applicable surface area to adhesively bond the CFRP. However, gusset plates are wider than the brace width, so we applied wider carbon fiber cloth for the step adjustment and alternative bidirectional high-strength carbon fiber cloth on the outermost layer to expand the width of the bonding area shown in Figure 7. On the other hand, the brace members have sufficient length to allow for bonding, while it is known that the bonding capacity will peak with increasing the bonding length [18]. Thus, we decided the bonding length should be 300 mm for the brace side with a 150 mm uniform carbon fiber layer length and a 150 mm taper length based on the previous fundamental bonding test results [21].  (7) and (8).    (7) and (8). Eventually, the total number of carbon fiber layers needed to minimize step step n was determined as 33 layers (9 layers to adjust the brace thickness, and 24 layers to adjust the height of the bolt head). The total number of carbon fiber layers needed to transfer the axial force ST n was determined as 13 layers for the gross sectional strengthening model and 4 layers for the sectional loss strengthening model. Figure 8 shows the lamination conditions. Bonding length is important to perform the strengthening effect and to avoid debonding failure. However, only a few hundred-millimeter bonding length can be used for the bonding on gusset plate. Thus, we decided the bonding length should be 150 mm for the gusset plate area because the actual brace connections lack the applicable surface area to adhesively bond the CFRP. However, gusset plates are wider than the brace width, so we applied wider carbon fiber cloth for the step adjustment and alternative bidirectional high-strength carbon fiber cloth on the outermost layer to expand the width of the bonding area shown in Figure 7. On the other hand, the brace members have sufficient length to allow for bonding, while it is known that the bonding capacity will peak with increasing the bonding length [18]. Thus, we decided the bonding length should be 300 mm for the brace side with a 150 mm uniform carbon fiber layer length and a 150 mm taper length based on the previous fundamental bonding test results [21].  (7) and (8).    Figure 9 shows the molding process. First, the steel surface was treated to make a rough surface using a blast surface forming power tool (Bristle Blaster ® , G-TOOL Co., Aichi, Japan), as shown in Figure 9a. We set the minimum surface roughness as 20 micro-strain with ten-point mean roughness using a surface roughness tester SJ-210 (Mitsutoyo, Tokyo, Japan). Second, E258R room temperature curable primer epoxy resin (Konishi, Tokyo, Japan) was applied on the steel surface to improve the bonding strength between the steel and CFRP, as shown in Figure 9b. After curing the primer resin, preformed carbon fiber cloth (shown in Figure 9c) and molding materials were attached, and the molding area was covered by vacuum film, as shown in Figure 9d,e. Finally, AUP40T1 low-viscosity, room temperature curable epoxy resin for VaRTM bonding and molding (Toray Building Materials, Tokyo, Japan) was injected after the vacuum pressure and absence of leakage were confirmed. The conditions in Figure 4 correspond to the situation shown in Figure 9e.

Specimen Preparation
Materials 2021, 14, x FOR PEER REVIEW 10 of 17 Figure 9 shows the molding process. First, the steel surface was treated to make a rough surface using a blast surface forming power tool (Bristle Blaster ® , G-TOOL Co., Aichi, Japan), as shown in Figure 9a. We set the minimum surface roughness as 20 micro-strain with ten-point mean roughness using a surface roughness tester SJ-210 (Mitsutoyo, Tokyo, Japan). Second, E258R room temperature curable primer epoxy resin (Konishi, Tokyo, Japan) was applied on the steel surface to improve the bonding strength between the steel and CFRP, as shown in Figure 9b. After curing the primer resin, preformed carbon fiber cloth (shown in Figure 9c) and molding materials were attached, and the molding area was covered by vacuum film, as shown in Figure 9d,e. Finally, AUP40T1 low-viscosity, room temperature curable epoxy resin for VaRTM bonding and molding (Toray Building Materials, Tokyo, Japan) was injected after the vacuum pressure and absence of leakage were confirmed. The conditions in Figure 4 correspond to the situation shown in Figure 9e. Figure 10 shows the specimen with CFRP strengthening applied. The average fiber volume content was 50.4% with a 4.97% coefficient of variation, as measured by a SWT-9000 vortex current coating thickness meter with an Fe-20 probe (Sanko Electric Laboratory Co., Tokyo, Japan) after demolding.    Figure 11 shows the experimental setup, while Table 4 shows the maximum load and failure modes. All specimens were monotonically tested in a 1000 kN Shimadzu universal testing machine, as shown in Figure 11. A load cell built in the testing machine was used to measure the applied load. Displacement transducers were mounted to measure crosshead displacement, and the longitudinal strain in specimens was measured by strain gauges attached directly to the steel and CFRP in several positions shown in Figure 6. It was demonstrated that CFRP strengthening could effectively increase the load-carrying capacity. Figure 12 shows the load-crosshead displacement relations, with the evaluated strength calculated using Equation (6). CFRP strengthening could increase not only the load-carrying capacity, but also the elongation, because the displacement just before breakage in the net area could be significantly increased (shown in Figure 12) in all CFRP strengthening specimens. This means that ductile deformation behavior occurred because CFRP strengthening could cause yielding of the gross section earlier than the net area. On the other hand, a lack of plastic deformation and brittle failure occurred without energy absorption due to hysteresis if the full plastic axial load after hardening in the net area was lower than the yielding load in the gross section, as seen in specimen NS. Thus, the performance of CFRP strengthening reached previously established steel-welded strengthening methods [6] without fire risk and use of heavy tools. Furthermore, no significant changes in elastic stiffness could be found, as shown in Figure 12b. Figure 13 shows load-strain relations at position g1, which is close to the bolt hole. In the case of specimen NS, yielding was observed around 80 kN due to stress concentration, while plastic strain grew rapidly. However, in the case of the CFRP-strengthened specimen, elastic behavior could be maintained until 110 kN, without causing large plastic strain. Thus, it can be confirmed that the CFRP strengthening can effectively decrease the strain nearby the bolt hole and delay the yielding in the net area to perform yielding load in the gross area. After reaching the maximum load, the plastic strain grew due to CFRP debonding, and breakage in the net area finally occurred. Additionally, the loading required for ductility shown in Equation (6) was satisfied without yielding in the net area. Loadings in the breakage section of the net area stayed almost constant for all specimens, because the CFRP was debonded by the plastic strain. This is because the primer resin cannot resist large strains. Thus, there was no difference in mechanical conditions just before breakage in the net area between the NS and CFRP-strengthened specimens. Figure 14 shows loadaverage strain relations at positions g2 and g3. Slight plastic strain occurred in specimen NS because the loading was decreased by the stress concentration in the net area and the associated necking. On the other hand, values of more than 35,000 micro-strain were observed in CFRP strengthening specimens. This means that the gross section position of the steel brace can be elongated due to prevention of the stress concentration in the net area using CFRP strengthening until debonding, as also shown in Figure 13b. Figure 14b shows the yielding behavior of the brace in the gross area. In the case of the specimens CFS-full-1 and CFS-loss-2, yielding was observed at 80kN loading, while specimens CFS-full-2 and CFS-loss-1 yielded at 115 and 100 kN, respectively. This variation of behavior was affected by the strain gauge position, because Figure 13 shows a strain of quite a local position. Eventually, CFRP strengthening specimens resisted almost the same load carrying capacity of 125 kN, as shown in Table 4. Additionally, it was estimated that the CFRP strengthening effect vanished due to plastic strain measuring 35,000-450,000 micro-strain by the CFRP debonding. Thus, it is thought that elongation of the CFRP strengthening steel brace CFRP-strengthened area Strengthening by steel welding Figure 10. Specimen with CFRP strengthening. Figure 11 shows the experimental setup, while Table 4 shows the maximum load and failure modes. All specimens were monotonically tested in a 1000 kN Shimadzu universal testing machine, as shown in Figure 11. A load cell built in the testing machine was used to measure the applied load. Displacement transducers were mounted to measure crosshead displacement, and the longitudinal strain in specimens was measured by strain gauges attached directly to the steel and CFRP in several positions shown in Figure 6. It was demonstrated that CFRP strengthening could effectively increase the load-carrying capacity. Figure 12 shows the load-crosshead displacement relations, with the evaluated strength calculated using Equation (6). CFRP strengthening could increase not only the load-carrying capacity, but also the elongation, because the displacement just before breakage in the net area could be significantly increased (shown in Figure 12) in all CFRP strengthening specimens. This means that ductile deformation behavior occurred because CFRP strengthening could cause yielding of the gross section earlier than the net area. On the other hand, a lack of plastic deformation and brittle failure occurred without energy absorption due to hysteresis if the full plastic axial load after hardening in the net area was lower than the yielding load in the gross section, as seen in specimen NS. Thus, the performance of CFRP strengthening reached previously established steel-welded strengthening methods [6] without fire risk and use of heavy tools. Furthermore, no significant changes in elastic stiffness could be found, as shown in Figure 12b. Figure 13 shows load-strain relations at position g1, which is close to the bolt hole. In the case of specimen NS, yielding was observed around 80 kN due to stress concentration, while plastic strain grew rapidly. However, in the case of the CFRP-strengthened specimen, elastic behavior could be maintained until 110 kN, without causing large plastic strain. Thus, it can be confirmed that the CFRP strengthening can effectively decrease the strain nearby the bolt hole and delay the yielding in the net area to perform yielding load in the gross area. After reaching the maximum load, the plastic strain grew due to CFRP debonding, and breakage in the net area finally occurred. Additionally, the loading required for ductility shown in Equation (6) was satisfied without yielding in the net area. Loadings in the breakage section of the net area stayed almost constant for all specimens, because the CFRP was debonded by the plastic strain. This is because the primer resin cannot resist large strains. Thus, there was no difference in mechanical conditions just before breakage in the net area between the NS and CFRP-strengthened specimens. Figure 14 shows load-average strain relations at positions g2 and g3. Slight plastic strain occurred in specimen NS because the loading was decreased by the stress concentration in the net area and the associated necking. On the other hand, values of more than 35,000 micro-strain were observed in CFRP strengthening specimens. This means that the gross section position of the steel brace can be elongated due to prevention of the stress concentration in the net area using CFRP strengthening until debonding, as also shown in Figure 13b. Figure 14b shows the yielding behavior of the brace in the gross area. In the case of the specimens CFS-full-1 and CFS-loss-2, yielding was observed at 80kN loading, while specimens CFS-full-2 and CFS-loss-1 yielded at 115 and 100 kN, respectively. This variation of behavior was affected by the strain gauge position, because Figure 13 shows a strain of quite a local position. Eventually, CFRP strengthening specimens resisted almost the same load carrying capacity of 125 kN, as shown in Table  4. Additionally, it was estimated that the CFRP strengthening effect vanished due to plastic strain measuring 35,000-450,000 micro-strain by the CFRP debonding. Thus, it is thought that elongation of the CFRP strengthening steel brace can account for 3% plastic deformation without debonding. Figure 15 shows the typical failure mode of the CFRP strengthening specimen. And Figure S1 (supplementary material) is an experimental video of CFRP strengthening specimen. It was observed that debonding began from the edge of the CFRP on the brace, and the debonding area was gradually grown. Figure 16 shows the ductility factor, as calculated by the following equation:

Experimental Results
where, L 1 is gage distance after experiment; L 0 is gage distance before experiment; σ YP is yield stress of the steel brace (370 N/mm 2 ); E is elastic modulus of the steel brace (200 kN/mm 2 ).
In the case of the NS specimen, the maximum value of the ductility factor was only 3.3 at position A5, as shown in Figure 6. On the other hand, the maximum ductility factor was 28 in specimen CFS-full-1, and the ductility factor values for all positions were larger than 11. In the case of CFRP-strengthened specimens, the ductility factor values decreased as the position approached the bolts. The ductility factor at the positions without CFRP strengthening (A4, A5, B4, and B5) reached 20-28. Thus, the total elongation can be calculated (including safety considerations) using the steel length L steel evaluated by subtracting the CFRP strengthening length L CFRP from the brace length L brace , showing a maximum of 3% plastic deformation. can account for 3% plastic deformation without debonding. Figure 15 shows the typical failure mode of the CFRP strengthening specimen. And Figure S1 (supplementary material) is an experimental video of CFRP strengthening specimen. It was observed that debonding began from the edge of the CFRP on the brace, and the debonding area was gradually grown. Figure 16 shows the ductility factor, as calculated by the following equation: In the case of the NS specimen, the maximum value of the ductility factor was only 3.3 at position A5, as shown in Figure 6. On the other hand, the maximum ductility factor was 28 in specimen CFS-full-1, and the ductility factor values for all positions were larger than 11. In the case of CFRP-strengthened specimens, the ductility factor values decreased as the position approached the bolts. The ductility factor at the positions without CFRP strengthening (A4, A5, B4, and B5) reached 20-28. Thus, the total elongation can be calculated (including safety considerations) using the steel length

Conclusions
This paper proposed a CFRP strengthening method for steel brace connections using the VaRTM technique and demonstrated the effects of the proposed method. The VaRTM technique was improved as an adhesion and molding method to strengthen existing steel structures on site. The following findings were obtained: 1. We proposed a strengthening method using carbon fiber cloth for steel brace connections with steps and demonstrated that molding and bonding can be successfully conducted using VaRTM.
2. The load-carrying capacity of the connection and the ductility factor of the steel brace were significantly improved using CFRP strengthening.
3. The total elongation of the steel brace can be estimated using 3% plastic deformation of the length of bare steel position.
This method cannot cover all type of steel brace connections, because the various geometrical condition can be found on the connection. Thus, we will evaluate the applicability of these strengthening methods for the actual structures by finite element analysis and experiment, and the applicability, strengthening effects, strengthening design

Conclusions
This paper proposed a CFRP strengthening method for steel brace connections using the VaRTM technique and demonstrated the effects of the proposed method. The VaRTM technique was improved as an adhesion and molding method to strengthen existing steel structures on site. The following findings were obtained: 1. We proposed a strengthening method using carbon fiber cloth for steel brace connections with steps and demonstrated that molding and bonding can be successfully conducted using VaRTM.
2. The load-carrying capacity of the connection and the ductility factor of the steel brace were significantly improved using CFRP strengthening.
3. The total elongation of the steel brace can be estimated using 3% plastic deformation of the length of bare steel position.
This method cannot cover all type of steel brace connections, because the various geometrical condition can be found on the connection. Thus, we will evaluate the applicability of these strengthening methods for the actual structures by finite element analysis and experiment, and the applicability, strengthening effects, strengthening design

Conclusions
This paper proposed a CFRP strengthening method for steel brace connections using the VaRTM technique and demonstrated the effects of the proposed method. The VaRTM technique was improved as an adhesion and molding method to strengthen existing steel structures on site. The following findings were obtained:

1.
We proposed a strengthening method using carbon fiber cloth for steel brace connections with steps and demonstrated that molding and bonding can be successfully conducted using VaRTM.

2.
The load-carrying capacity of the connection and the ductility factor of the steel brace were significantly improved using CFRP strengthening.

3.
The total elongation of the steel brace can be estimated using 3% plastic deformation of the length of bare steel position.
This method cannot cover all type of steel brace connections, because the various geometrical condition can be found on the connection. Thus, we will evaluate the applicability of these strengthening methods for the actual structures by finite element analysis and experiment, and the applicability, strengthening effects, strengthening design method, as well as experimental variation and stability will be clarified for the typical connection geometries in the future.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest.