TY - JOUR
AU - Repplinger,, Christian
AB - Abstract This study introduced a fatigue-based approach to design and implement an indicator channel into an in-tank hydrogen valve. It was aimed at providing a mean to point out multiple early valve's damages. To achieve the goal, the study was proposed to handle via three main phases. They included (i) the risk point determinations, (ii) the new valve design and the crack nucleation life estimations, as well as (iii) the simplified crack growth analyses. The obtained results firstly highlighted the construction of the test channel (TC), whose branches were located close to the predicted damage's sites. The damages could be identified either when a crack reaches the TC (then forms a leakage) or indirectly via the crack propagations’ correlation. The results also pointed out that the TC-implemented valve could perform as similarly as the non-TC one in the non-treated condition. More importantly, this new structure was proved to have a capacity of satisfying the required minimal life of 1.5E5 cycles, depending on the combined uses of the specific material and the pre-treatment, among those considered. In addition, the results emphasized the complexity of the TC that could not be formed by the traditional manufacturing process. Hence, direct metal laser sintering was proposed for the associated prototype and the final TC was issued based on the fundamental requirements of the technique. Finally, it was suggested that practical experiments should essentially be carried out to yield more evidence to support the demonstrated results. Graphical Abstract Open in new tabDownload slide Graphical Abstract Open in new tabDownload slide automotive hydrogen valve, fatigue-based design, sustainable design, direct metal laser sintering Highlights A test channel (TC) was designed for a hydrogen valve via a fatigue-based approach. The approach mainly relied on crack nucleation analyses. The TC-implemented valve could perform similarly to the non-TC one. The TC valve could satisfy the required minimal life of 1.5E5 cycles. The valve was designed towards compatibility with direct metal laser sintering. Nomenclature a1 Crack length b Fatigue strength exponent c Fatigue ductility exponent c1 Half of the crack width CJP Material parameter measured by experiments d Diameter E Elastic modulus H Empirical tightening factor J Elastic plastic failure parameter (J-integral) JE, i Elastic part of J-integral, estimated at stage i JP, i Plastic part of J-integral, estimated at stage i |$J_{\mathrm{ P}, \mathrm{ i}}^A$| Amplitude of plastic part of J-integral (stage i) |$J_{\mathrm{ P}, \mathrm{ i}}^{\mathrm{ max}}$| Maximal value of plastic part of J-integral (stage i) |$J_{\mathrm{ P}, \mathrm{ i}}^{\mathrm{ min}}$| Minimal value of plastic part of J-integral (stage i) |${J_{\mathrm{ T}, \mathrm{ i}}}$| Total J-integral, estimated at stage i |$K^{\prime}$| Cyclic strength coefficient |${K_\mathrm{ I}}$| Mode I stress intensity factor |${K_{\mathrm{ I}, \mathrm{ i}}}$| Mode I stress intensity factor, estimated at stage i |${K_{\mathrm{ TREAT}}}$| Surface treatment factor |${K_{\mathrm{ USER}}}$| User surface factor |${N_\mathrm{ C}}$| Cumulative total life for crack propagation |${N_G}$| Crack growth life (propagation life) |${N_\mathrm{ i}}$| Crack initiation life (nucleation life) |${m_{\mathrm{ JP}}}$| Material parameter measured by experiments |$n^{\prime}\ $| Cyclic strain hardening exponent |${P_t}$| Pretension load |${R_a}$| Surface roughness T Tightening torque |$\Delta {a_\mathrm{ i}}$| Crack propagation length in stage i |$\varepsilon _\mathrm{ f}^{\prime}$| Fatigue ductility coefficient |$\vartheta $| Poisson ratio |$\sigma _f^{\prime}$| Fatigue strength coefficient |${\sigma _\mathrm{ Y}}$| Yield strength AF Autofrettage CAD Computer-aided design DMLS Direct metal laser sintering EN Strain-fife FE Finite element LV2 Lattice valve – version 2 M-A Critical plane with SWT mean stress correction M-B Critical plane with Morrow mean stress correction M-C Critical shear plane with SWT mean stress correction M-D Critical shear plane with Morrow mean stress correction SWT Smith–Watson–Topper TC Test channel TPRD Thermal temperature relief device WB-mean Wang–Brown method with mean stress correction 1. Introduction There has been raising the interest of using hydrogen as one of the alternative fuel sources in the transportation sector. It is due to the generation of water as a bypass from the vehicles powered by hydrogen. The foreseen green use of such vehicles has motivated leading automotive companies to focus their product development more on the hydrogen-powered cars for the sustainable future. Some of the cars being available in the commercial market are Toyota Mirai (News – Road Vehicles, 2017a), Honda Clarity (News – Road Vehicles, 2017b), and Audi H-Tron (News – Road Vehicles, 2016). Currently, the hydrogen-powered cars are often equipped with the 70 MPa fuel storage system. This system employs an in-tank valve and other devices (e.g. solenoid actuator, TPRD) to control the distributions of the hydrogen. To guarantee that the system is able to work properly and safely in the high-pressure condition, the in-tank valve (along with other components of the car) needs to be well-functioned during service. Although hydrogen technology has been studied for years, published works focused on investigating an in-tank valve of a hydrogen-powered car are limited. So far, there have been two articles (Sellen et al., 2015, 2016) dealt with the valve, subjected to produce by traditional processes. In the work published in 2016, the body of the valve was studied. The main aim of the work was to identify an appropriate pre-treatment pressure to enhance the fatigue life of the valve. In the earlier work, the largest female thread of the same valve was investigated. A similar procedure was used to investigate the thread to see the positive influences of the suitable pre-treatment pressure on the fatigue durability of the structure. In both of these works, it was proved that fatigue life of the body and the largest female thread of the pre-treated valve could satisfy the required minimal total life of 1.5E5 cycles. However, some issues, which could affect the performance of the valve, still existed. Firstly, the presence of the sharp intersections between flow channels will induce high-stress concentration and potentially lead to early crack nucleation on the critical surfaces of the valve during service. Secondly, since the investigated body was designed towards traditional fabrications, the material within the low-stress volumes of the body could not be well distributed. A recent study (Cao & Kedziora, 2019) addressed the two pointed issues to build up the new valve body, theoretically having better performance than the original. The study employed the optimisation-assisted design methods to innovate the body structure. Specifically, free shape optimisation was utilised to eliminate the sharp intersections between flow channels of the valve. Topology optimisation was then used to redistribute material within the selected design space. Since the developed structure possessed several unexposed features (such as internal cavities, internal fillets, and others), it would be hardly shaped by the traditional manufacturing process. Hence, to guarantee the manufacturability of the designed structure, implementations of permanent lattice and powder release pipes into the body, as well as modifications of the body towards DMLS compatibility were considered. In so doing, the new structure (called LV2), having a longer life and lower mass than those of the original body as well as being able to fabricate by DMLS, was successfully generated. This study continued the earlier work, described in Cao and Kedziora (2019). It dealt mainly with designing a TC for the LV2, following a fatigue-based approach. The primary purpose of the study was to implement the TC into the valve to be able to indicate early fatigue damages at the critical regions, while not weakening the valve structure considerably. Towards the purpose, several phases of design and virtual evaluations were proposed to handle as follows. Firstly, the initial structure was analysed to predict the most critical locations and the associated crack planes. Secondly, the TC-implemented valve was designed, similarly analysed, and compared with that considered in the first stage to evaluate the performance of the new structure. Third, some rough investigations of crack propagation from the critical areas were carried out, aimed at correlating the early fatigue damages at multiple risk locations within the valve. 2. Methodology In this section, the methods, used to design and evaluate the TC-implemented valve virtually, are introduced. An overview of the whole steps of the study is indicated in Fig. 1. The main flow sequences of these steps are listed in Table 1. Figure 1: Open in new tabDownload slide Proposed procedure for designing TC in hydrogen valve and for evaluating TC-implemented valve. Figure 1: Open in new tabDownload slide Proposed procedure for designing TC in hydrogen valve and for evaluating TC-implemented valve. Figure 4: Open in new tabDownload slide Full model setups for FE analyses. (a) Setups for applied internal pressure and supports. (b) Setups for pre-tension force and forces at different ports of hydrogen valve. Figure 4: Open in new tabDownload slide Full model setups for FE analyses. (a) Setups for applied internal pressure and supports. (b) Setups for pre-tension force and forces at different ports of hydrogen valve. Figure 5: Open in new tabDownload slide Load steps setting for FE analyses of hydrogen valve in ANSYS Workbench. (a) Setting for forces at different ports. (b) Setting for internal pressure. Figure 5: Open in new tabDownload slide Load steps setting for FE analyses of hydrogen valve in ANSYS Workbench. (a) Setting for forces at different ports. (b) Setting for internal pressure. Table 1: Summary of the main divided phases of the current study. Phases . Flow sequencesa . Explanations . 1 ①→②→③→④→⑤→⑥ Fatigue life assessments for the (non-TC) valve 2 ⑥→⑦ TC-implemented valve design (fatigue approach) ⑧→⑨→⑩→⑪→⑫→⑥→⑬ Fatigue life assessments for the TC-implemented valve 3 ⑧→⑨→⑭→⑮→⑯→⑰→⑬ Rough estimations of crack growth life for the new valve Phases . Flow sequencesa . Explanations . 1 ①→②→③→④→⑤→⑥ Fatigue life assessments for the (non-TC) valve 2 ⑥→⑦ TC-implemented valve design (fatigue approach) ⑧→⑨→⑩→⑪→⑫→⑥→⑬ Fatigue life assessments for the TC-implemented valve 3 ⑧→⑨→⑭→⑮→⑯→⑰→⑬ Rough estimations of crack growth life for the new valve a The sequences are graphically demonstrated in Fig. 1. Open in new tab Table 1: Summary of the main divided phases of the current study. Phases . Flow sequencesa . Explanations . 1 ①→②→③→④→⑤→⑥ Fatigue life assessments for the (non-TC) valve 2 ⑥→⑦ TC-implemented valve design (fatigue approach) ⑧→⑨→⑩→⑪→⑫→⑥→⑬ Fatigue life assessments for the TC-implemented valve 3 ⑧→⑨→⑭→⑮→⑯→⑰→⑬ Rough estimations of crack growth life for the new valve Phases . Flow sequencesa . Explanations . 1 ①→②→③→④→⑤→⑥ Fatigue life assessments for the (non-TC) valve 2 ⑥→⑦ TC-implemented valve design (fatigue approach) ⑧→⑨→⑩→⑪→⑫→⑥→⑬ Fatigue life assessments for the TC-implemented valve 3 ⑧→⑨→⑭→⑮→⑯→⑰→⑬ Rough estimations of crack growth life for the new valve a The sequences are graphically demonstrated in Fig. 1. Open in new tab 2.1. Fatigue life assessments of the non-TC valve 2.1.1. Models’ setup for non-linear FE analyses The FE analyses dealt with two types of CAD model. The first type was the full model (Fig. 2b) which was used to simulate the real working condition of the valve (Fig. 2a). The remaining type was sub-models. The sub-models were defined within the valve (Fig. 3a) so that they contained the critical regions, pointed out in Cao and Kedziora (2019). The sub-model 1 (Fig. 3b) was employed to investigate the intersection between holes 1, 2, and 4. Whereas, the sub-model 2 (Fig. 3c) was used to analyse the intersection between holes 4 and 6. Figure 2: Open in new tabDownload slide Hydrogen valve and assembled components taking into account in (a) real condition and (b) testing condition. Figure 2: Open in new tabDownload slide Hydrogen valve and assembled components taking into account in (a) real condition and (b) testing condition. Figure 3: Open in new tabDownload slide Sub-models identifications for FE analyses of non-TC valve. (a) Relative positions of sub-models with respect to full model, (b) sub-model 1, and (c) sub-model 2. Figure 3: Open in new tabDownload slide Sub-models identifications for FE analyses of non-TC valve. (a) Relative positions of sub-models with respect to full model, (b) sub-model 1, and (c) sub-model 2. The full model's setup for FE analyses started with importing geometry (Fig. 2b) to ANSYS Workbench. Next, materials were assigned to the model. While AW6082T6 (Sellen, 2014; Sellen et al., 2015) was utilised as the material of the valve, 17–4 PH SS (Yadollahi, Thompson, Elwany, & Bian, 2017) was employed as that of the cup. Properties of these materials are summarised in Table 2. It was followed by discretising the geometry to a finite number of tetra-10 elements. Subsequently, the contact was defined and set up for the connection. In the next step, the applied loads and the boundary conditions were applied to the model to simulate the working conditions of the valve. The considered loads (Fig. 4) included (i) an internal pressure, (ii) those forces at different ports, and (iii) a pre-tension force. This pre-tension force was used to represent the torque of 300 Nm, applied to tighten the valve–cup connection. Equation (1) was utilised to calculate the pre-tension force from the tightening torque. Afterwards, the model was constrained by placing the fixed support on surface D and frictionless support on surface E of the cup (Fig. 4a). It is noted that during the FE analyses the internal pressure and the forces at ports were considered via the six load steps (Fig. 5) whereas the pre-tension one was only taken into consideration during the first step. $$\begin{equation} {P_t} = \frac{T}{{H.d}} \end{equation}$$(1) Table 2: Summary of mechanical properties of the used materials. Materials . E (GPa) . |${\sigma _\mathrm{ Y}}$| (MPa) . |$\sigma _\mathrm{ f}^{{^{\prime}}}$| (MPa) . |$\varepsilon _\mathrm{ f}^{{^{\prime}}}$| . b . c . |$K^{\prime}$| (MPa) . |$n^{\prime}$| . AW6082T6 (Sellen, 2014; Sellen et al., 2015) 74.6 323 a a a a a a AlSi10Mg (Romano, Patriarca, Foletti, & Beretta, 2018) 74.6 304.5 552 5.534 −0.11 −1.017 410 0.092 17-4 PH SS (Yadollahi et al., 2017) 187.3 580 2043 1.8 −0.15 −0.53 1225 0.19 18Ni300 (Branco et al., 2018) 168 910 1798.73 0.327 84 −0.1311 −1.0941 1921.21 0.11 Materials . E (GPa) . |${\sigma _\mathrm{ Y}}$| (MPa) . |$\sigma _\mathrm{ f}^{{^{\prime}}}$| (MPa) . |$\varepsilon _\mathrm{ f}^{{^{\prime}}}$| . b . c . |$K^{\prime}$| (MPa) . |$n^{\prime}$| . AW6082T6 (Sellen, 2014; Sellen et al., 2015) 74.6 323 a a a a a a AlSi10Mg (Romano, Patriarca, Foletti, & Beretta, 2018) 74.6 304.5 552 5.534 −0.11 −1.017 410 0.092 17-4 PH SS (Yadollahi et al., 2017) 187.3 580 2043 1.8 −0.15 −0.53 1225 0.19 18Ni300 (Branco et al., 2018) 168 910 1798.73 0.327 84 −0.1311 −1.0941 1921.21 0.11 a Some confidential data stated in Sellen (2014). Open in new tab Table 2: Summary of mechanical properties of the used materials. Materials . E (GPa) . |${\sigma _\mathrm{ Y}}$| (MPa) . |$\sigma _\mathrm{ f}^{{^{\prime}}}$| (MPa) . |$\varepsilon _\mathrm{ f}^{{^{\prime}}}$| . b . c . |$K^{\prime}$| (MPa) . |$n^{\prime}$| . AW6082T6 (Sellen, 2014; Sellen et al., 2015) 74.6 323 a a a a a a AlSi10Mg (Romano, Patriarca, Foletti, & Beretta, 2018) 74.6 304.5 552 5.534 −0.11 −1.017 410 0.092 17-4 PH SS (Yadollahi et al., 2017) 187.3 580 2043 1.8 −0.15 −0.53 1225 0.19 18Ni300 (Branco et al., 2018) 168 910 1798.73 0.327 84 −0.1311 −1.0941 1921.21 0.11 Materials . E (GPa) . |${\sigma _\mathrm{ Y}}$| (MPa) . |$\sigma _\mathrm{ f}^{{^{\prime}}}$| (MPa) . |$\varepsilon _\mathrm{ f}^{{^{\prime}}}$| . b . c . |$K^{\prime}$| (MPa) . |$n^{\prime}$| . AW6082T6 (Sellen, 2014; Sellen et al., 2015) 74.6 323 a a a a a a AlSi10Mg (Romano, Patriarca, Foletti, & Beretta, 2018) 74.6 304.5 552 5.534 −0.11 −1.017 410 0.092 17-4 PH SS (Yadollahi et al., 2017) 187.3 580 2043 1.8 −0.15 −0.53 1225 0.19 18Ni300 (Branco et al., 2018) 168 910 1798.73 0.327 84 −0.1311 −1.0941 1921.21 0.11 a Some confidential data stated in Sellen (2014). Open in new tab The setup of a sub-model began with constructing the geometry in DesignModeler and importing such geometry to Workbench. Next, the same material as that of the full model was assigned to the sub-model. Then, the second-ordered tetrahedral elements were employed to mesh the model. This mesh was refined so that it was considerably finer than that of the full one, targeting the converged results of stress at the critical regions. It was followed by applying the internal pressure, the constraint, and setting up the load steps for the models. In more details, the pressure was placed on the flow channels of the models. The displacement constraint was applied onto the cut boundary faces, which separated the sub-models from the remaining part of the full one. Such a constraint was originated from the analysis of the full model and could be transferred to the setups of the sub-models. Finally, the same six load steps were used for the FE analyses of the sub-model. 2.1.2. Crack initiation life estimations The crack initiation life was estimated by the use of nCode DesignLife. In nCode, the EN regime was selected for setting up and running the calculation models. In details, the stress–strain history was initially defined by combining the stress–strain results, predicted at the three last time-steps (Fig. 5) in ANSYS. Properties of the assigned material were then defined. Apart from these properties, others in association with surface conditions of the material were additionally determined in nCode via the three parameters symbolised as KUSER, KTREAT, and Ra. Subsequently, the predictions were taken place by using either the standard EN or the multiaxial EN solver engines (nCode, 2016). In the former engine, the four embedded methods including M-A, M-B, M-C, and M-D were employed. In the latter one, the WB-mean method (Wang & Brown, 1993) was utilised for the estimations. The more detailed descriptions of the five methods can be found somewhere in Cao and Kedziora (2019). During the estimations, the beyond cut-off value and the certainty of survival were proposed to be 2 × 108 cycles and 50%, respectively. 2.2. Design of TC and virtual evaluations of the TC-implemented valve 2.2.1. TC design for the LV2 The TC design was handled based on the results of the fatigue life assessments. From the assessments, those nodes having the lowest life were firstly identified and considered as the risk positions, where the earliest crack theoretically nucleated. Next, an assumption was made to hypothesize that a 3D crack would propagate on a plane from the nodes to the solid material, following mode I. The crack planes could be pointed out in Workbench via two steps. In the initial step, the maximal principal stress directions of the critical nodes were determined. In the second step, the crack planes were built up so that they went through the identified nodes and took their maximal principal stress directions as the normal vectors. Upon defining the crack planes, different branches of the TC were proposed and implemented to the LV2 structure. In general, the branches were located close to the critical positions of the valve and were aligned to be intersected by the crack planes. 2.2.2. Fatigue life assessments of the TC-implemented valve Fatigue life assessments of the TC-implemented valve were handled similar to those described in Section 2.1. The assessments contained the FE analyses of the new full model, the analyses of the new sub-models, and the estimations of crack initiation life. 2.3. Rough investigations of crack growth for the TC-implemented valve 2.3.1. Construction of geometries of sub-models holding cracks To construct the new sub-models holding cracks, DesignModeler was employed to define the geometries of these models. There were two types of based geometries, used for the sub-models. The first type was the structure of the sub-model 3, which was defined so that it included the most critical site located between holes 1, 2, and 4 (Fig. 6a). Whereas, the second type was the structure of the sub-model 4 that was determined to contain the riskiest area found between holes 4 and 6 (Fig. 7a). To complete the constructions of the geometries of these sub-models, 3D cracks having different crack depths were introduced to the based geometries of these models. Figure 6: Open in new tabDownload slide Defining and meshing sub-model 3. (a) Determinations of sub-model 3 and associated 3D crack. (b) Mesh of sub-model 3. Figure 6: Open in new tabDownload slide Defining and meshing sub-model 3. (a) Determinations of sub-model 3 and associated 3D crack. (b) Mesh of sub-model 3. Figure 7: Open in new tabDownload slide Defining and meshing sub-model 4. (a) Determinations of sub-model 4 and associated 3D crack. (b) Mesh of sub-model 4. Figure 7: Open in new tabDownload slide Defining and meshing sub-model 4. (a) Determinations of sub-model 4 and associated 3D crack. (b) Mesh of sub-model 4. The cracks’ geometries, which would be implemented to the based structures, were parametrically defined in DesignModeler. For the constructed crack, it had the aspect ratio |$( {{a_1}/{c_1}} )$| of 1 and composed of four flank surfaces. The two larger surfaces were designed in parallel to each other and formed a gap of 50 µm. The remaining two surfaces were aligned to build up the crack front. This crack front was located on the crack's middle plane, which assumed to be also the crack plane. In addition to the features, the front edge also took the most critical node, at which the crack was predictably nucleated, as the central point. For each type of the sub-models (in Fig. 6 or 7), the crack holding different depths, ranging from 0.4 to 4.1 mm, was implemented to the based structure to create the geometries of the models. 2.3.2. Rough estimations of crack propagation life for the new valve The crack propagation life was estimated via two sub-stages, including (i) FE analyses of the sub-models holding cracks and (ii) theoretical calculations of life. The FE analyses were handled to estimate the values of |${K_\mathrm{ I}}$| and J for the fronts of the constructed cracks. These analyses were done separately for different types of sub-models (3 and 4), possessing different crack depths. The setup of a sub-model began with importing the geometry to Workbench. Next, AW6082T6 was assigned as the material of the model. Then, the model was discretised. Within the small volume, surrounding the crack front, the hexahedral elements were employed to generate the mesh contours of the crack (Figs 6b and 7b). The remaining volume of the model was meshed with the tetrahedral elements. Fracture setup was subsequently performed. In the setup, those nodes, located on the crack front, the cracks’ upper surfaces, and lower surfaces, were identified. The number of contours near the crack front was determined as well for the extractions of the J solutions during the analyses. In addition, the local coordinate system associated with the crack was defined as follows. The system had its origin, coinciding with the most critical location of the sub-model 1’ or 2’. Its Y-axis was the crack plane normal. Whereas, the local X direction was the assumed crack propagation way (Figs 6a and 7a). The remaining setup steps, such as applications of the working pressure, the cut boundary constraints, and the assigned load steps, were similarly handled as the ones described in Section 2.1.1. Within the second sub-stage, the life calculations were performed based on the obtained results of |${K_I}$| and |$\ J$|⁠. Those equations, listed in Table 3 were employed to calculate the crack propagation life |${N_\mathrm{ G}}$| of the specific crack. The crack growth life results together with the crack nucleation life results would provide theoretical evidence to evaluate the performance of the TC-implemented valve. Table 3: Parameters used to estimate crack propagation (growth) life. Parameters . Values estimated via . Notes . |${K_{\mathrm{ I,i}}}$| ANSYS Workbench 19 − |${J_{\mathrm{ T,i}}}$| ANSYS Workbench 19 − |${J_{\mathrm{ E,i}}}$| |${J_{\mathrm{ E,i}}} = \frac{{K_{\mathrm{ I,i}}^2}}{{( {\frac{{ E}}{{1 - {\vartheta ^2}}}} )}}$| (2) |${J_{\mathrm{ P,i}}}$| |${J_{\mathrm{ P,i}}} = {J_{\mathrm{ T,i}}} - {J_{\mathrm{ E,i}}}$| (3) |$J_{\mathrm{ P,i}}^A$| |$J_{\mathrm{ P,i}}^A = \frac{{J_{\mathrm{ P,i}}^{\mathrm{ max}} - J_{\mathrm{ P,i}}^{\mathrm{ \mathrm{ min}}}}}{2}$| (4) |$\Delta {N_i}$| |$\Delta {N_\mathrm{ i}} = \frac{{\Delta {a_\mathrm{ i}}}}{{{C_{\mathrm{ JP}}}.{{( {J_{\mathrm{ P,i}}^A} )}^{{m_{\mathrm{ JP}}}}}}}$| (5) Based on Hutař et al. (2014) |${N_\mathrm{ G}}$| |${N_\mathrm{ G}} = \sum \Delta {N_\mathrm{ i}}$| (6) Parameters . Values estimated via . Notes . |${K_{\mathrm{ I,i}}}$| ANSYS Workbench 19 − |${J_{\mathrm{ T,i}}}$| ANSYS Workbench 19 − |${J_{\mathrm{ E,i}}}$| |${J_{\mathrm{ E,i}}} = \frac{{K_{\mathrm{ I,i}}^2}}{{( {\frac{{ E}}{{1 - {\vartheta ^2}}}} )}}$| (2) |${J_{\mathrm{ P,i}}}$| |${J_{\mathrm{ P,i}}} = {J_{\mathrm{ T,i}}} - {J_{\mathrm{ E,i}}}$| (3) |$J_{\mathrm{ P,i}}^A$| |$J_{\mathrm{ P,i}}^A = \frac{{J_{\mathrm{ P,i}}^{\mathrm{ max}} - J_{\mathrm{ P,i}}^{\mathrm{ \mathrm{ min}}}}}{2}$| (4) |$\Delta {N_i}$| |$\Delta {N_\mathrm{ i}} = \frac{{\Delta {a_\mathrm{ i}}}}{{{C_{\mathrm{ JP}}}.{{( {J_{\mathrm{ P,i}}^A} )}^{{m_{\mathrm{ JP}}}}}}}$| (5) Based on Hutař et al. (2014) |${N_\mathrm{ G}}$| |${N_\mathrm{ G}} = \sum \Delta {N_\mathrm{ i}}$| (6) For Al 6082: |${C_{\mathrm{ JP}}} = 9 \times {10^{ - 8}};\ $||${m_{\mathrm{ JP}}} = 1.2941$| (Hutař et al., 2017). Open in new tab Table 3: Parameters used to estimate crack propagation (growth) life. Parameters . Values estimated via . Notes . |${K_{\mathrm{ I,i}}}$| ANSYS Workbench 19 − |${J_{\mathrm{ T,i}}}$| ANSYS Workbench 19 − |${J_{\mathrm{ E,i}}}$| |${J_{\mathrm{ E,i}}} = \frac{{K_{\mathrm{ I,i}}^2}}{{( {\frac{{ E}}{{1 - {\vartheta ^2}}}} )}}$| (2) |${J_{\mathrm{ P,i}}}$| |${J_{\mathrm{ P,i}}} = {J_{\mathrm{ T,i}}} - {J_{\mathrm{ E,i}}}$| (3) |$J_{\mathrm{ P,i}}^A$| |$J_{\mathrm{ P,i}}^A = \frac{{J_{\mathrm{ P,i}}^{\mathrm{ max}} - J_{\mathrm{ P,i}}^{\mathrm{ \mathrm{ min}}}}}{2}$| (4) |$\Delta {N_i}$| |$\Delta {N_\mathrm{ i}} = \frac{{\Delta {a_\mathrm{ i}}}}{{{C_{\mathrm{ JP}}}.{{( {J_{\mathrm{ P,i}}^A} )}^{{m_{\mathrm{ JP}}}}}}}$| (5) Based on Hutař et al. (2014) |${N_\mathrm{ G}}$| |${N_\mathrm{ G}} = \sum \Delta {N_\mathrm{ i}}$| (6) Parameters . Values estimated via . Notes . |${K_{\mathrm{ I,i}}}$| ANSYS Workbench 19 − |${J_{\mathrm{ T,i}}}$| ANSYS Workbench 19 − |${J_{\mathrm{ E,i}}}$| |${J_{\mathrm{ E,i}}} = \frac{{K_{\mathrm{ I,i}}^2}}{{( {\frac{{ E}}{{1 - {\vartheta ^2}}}} )}}$| (2) |${J_{\mathrm{ P,i}}}$| |${J_{\mathrm{ P,i}}} = {J_{\mathrm{ T,i}}} - {J_{\mathrm{ E,i}}}$| (3) |$J_{\mathrm{ P,i}}^A$| |$J_{\mathrm{ P,i}}^A = \frac{{J_{\mathrm{ P,i}}^{\mathrm{ max}} - J_{\mathrm{ P,i}}^{\mathrm{ \mathrm{ min}}}}}{2}$| (4) |$\Delta {N_i}$| |$\Delta {N_\mathrm{ i}} = \frac{{\Delta {a_\mathrm{ i}}}}{{{C_{\mathrm{ JP}}}.{{( {J_{\mathrm{ P,i}}^A} )}^{{m_{\mathrm{ JP}}}}}}}$| (5) Based on Hutař et al. (2014) |${N_\mathrm{ G}}$| |${N_\mathrm{ G}} = \sum \Delta {N_\mathrm{ i}}$| (6) For Al 6082: |${C_{\mathrm{ JP}}} = 9 \times {10^{ - 8}};\ $||${m_{\mathrm{ JP}}} = 1.2941$| (Hutař et al., 2017). Open in new tab 3. Results and Discussion 3.1. Crack initiation life assessments for the non-TC valve Figure 8 represents the life distributions on the critical areas of the non-treated valve (non-TC). As indicated in Fig. 8a, the most critical node 10 840 was found to locate somewhere on the intersection between holes 1, 2, and 4 of the sub-model 1. This node had an estimated life of 7160 cycles, which was considerably shorter than the life, distributed on the remaining intersection of the model. In Fig. 8b, the representative result highlights the minimal life of 4826 cycles, detected at node 5453 of the sub-model 2. This node was situated on the left intersectional area between holes 4 and 6. Additionally, the figure points out that the right intersectional area of the sub-model also acted as the risk region, on which a crack theoretically nucleated soon after 4826 cycles. Figure 8: Open in new tabDownload slide Estimated life distributions on surface hot spot areas of non-TC hydrogen valve in non-treated condition. (a) Areas belonged to sub-model 1. (b) Areas belonged to sub-model 2. Figure 8: Open in new tabDownload slide Estimated life distributions on surface hot spot areas of non-TC hydrogen valve in non-treated condition. (a) Areas belonged to sub-model 1. (b) Areas belonged to sub-model 2. 3.2. TC design and virtual evaluations of TC-implemented valve The first two parts of Fig. 9 show the local coordinate systems and the crack planes, which were constructed based on the results indicated in Fig. 8. In details, the local origins coincided with the most critical nodes of the sub-models 1 and 2. The Z local directions were the maximal principal stress directions of the identified nodes. The XY planes, which were also the green planes in the figure, acted as the assumed crack propagation planes. Based on the planes, the yellow dash ellipses were proposed, highlighting the potential regions where a part of the TC structure should intersect. In so doing, if a crack is propagated along the plane during the valve's service, the crack will, sooner or later, reach the TC. At this point, the pressure enhancement in the TC due to hydrogen leakage will be detected by a sensor assembled at the port of the TC, which can somehow inform the damage status to a car driver. It is noted that two assumptions have been considered during TC design. Firstly, it was assumed that mode I failure was dominated for propagating the crack from the nucleation sites along the planes towards the solid material. Secondly, the planes were hypothesised to be not altered (or slightly altered) during the crack growths. Figure 9: Open in new tabDownload slide Constructing TC in hydrogen valve. (a) and (b) Defining potential regions, which TC should intersect, on crack planes. (c) Designed TC structure in new valve. Figure 9: Open in new tabDownload slide Constructing TC in hydrogen valve. (a) and (b) Defining potential regions, which TC should intersect, on crack planes. (c) Designed TC structure in new valve. The third part of Fig. 9 demonstrates the temporary structure of the TC which was implemented into the modified LV2. According to the figure, the TC was composed of three branches. Branch 2 was formed so that it could be intersected by the crack plane 2 within the highlighted regions (Fig. 9b). Branch 3 had two ends, which were designed to evaluate the possibility of extending the TC ends to the free surfaces of the valve without considerably weakening it. The remaining branch 1 was a backup and it was generated based on the earlier investigations of the hydrogen valve (Sellen, 2014). All of these branches were connected to each other to construct the completed TC, having an opened end (Fig. 9c) to assemble with a pressure sensor. The highlighted TC possesses the diameter of 1.5 mm and its branches were generally located around 4 mm, being away from the critical sites of the valve. It should be emphasised that despite indicating the potential region 1 in Fig. 9a, there was no branch cut through this region to prevent the valve from being significantly weakened. It is further noted that the extension of branch 3 to one of the free surfaces of the valve was also expected. It was aimed at forming an additional TC port that could promote the release of the un-sintered powder after additively fabricating the valve by DMLS. However, the extension has not been applied yet to the temporary structure of the valve. The minimal life, predicted by the five selected methods for the non-TC valve and the TC-implemented valve, is indicated in Fig. 10. The figure shows that the life values of the non-TC valve were approximate those of the TC-implemented valve in the non-treated condition. This observation suggested that the TC design was appropriately done since it did not result in the weaker valve structure. It was also clear that all predicted life of the non-treated valves was shorter than 5E4 cycles. By applying the AF pressure (230 MPa) to the TC-implemented valve, the life values of the valve were significantly enhanced. Some results, predicted by M-A and M-C, even reached the cut-off value of 2E8 cycles. Whereas, the smallest value of nearly 1.5E5 cycles was obtained when using the WB-mean for the prediction. Hence the WB-mean is seen as the most conservative method that should be used to determine the riskiest points and the crack planes for ‘safely’ designing the TC. It is noticed that during the life prediction KUSER was 1, which was almost half of the value of that considered in Cao and Kedziora (2019). As a result, the predicted life, highlighted in Figs 8 and 10, was much shorter than those demonstrated in the mentioned article. In other words, the selection of KUSER also targeted the safe consideration for designing the TC in the subsequent phase. It is also worth to note that the results shown in Fig. 10 were achieved by handling the calculations only with AW6082T6 (the assigned material). This material was specifically used for evaluating the efficiency of the TC design/implementation process and the newly created valve with respect to the non-TC one. Figure 10: Open in new tabDownload slide Comparisons of minimal estimated life (crack initiation life) of both non-TC valve and TC-implemented valve within two considered treating conditions. Figure 10: Open in new tabDownload slide Comparisons of minimal estimated life (crack initiation life) of both non-TC valve and TC-implemented valve within two considered treating conditions. The positive effect of the AF treatment on fatigue performance of the TC-implemented valve can be explained in view of Fig. 11. In this figure, the stress distributions (at time-step 5) were considered on the tiny area, defined by the XLOCAL range of [−3.3; 0.6] mm and the YLOCAL range of [0.0; 4.0] mm, on the crack plane. From Fig. 11b, it is seen that within the non-treated condition, the surface plot of the stress result was located completely above zero. The highest stress of 262 MPa was detected at the local origin and the adjacent area. Towards the positive local Y direction, the normal stress was gradually decreased. However, in the treated condition, the distributed behaviour of the stress was totally changed. There was a significant reduction of the stress in the region, limited by the XLOCAL range of [−1.0; 0.6] mm and the YLOCAL range of [0.0; 1.0] mm. In this region, the minimum stress value was found and stood at −157 MPa. Following the direction, showing the higher values of YLOCAL, the magnitude of the stress was generally increased. On top of these observations, it can be realised that the mean values of the stresses, estimated in both conditions, were circa 98.5 and 2.5 MPa. The reduction of the mean stress to 2.5 MPa, as a result of the pre-treatment, is believed the main reason, contributing to the life enhancement in the TC-implemented valve. It should be noticed that the weakest point and the crack plane of the TC-implemented valve (Fig. 11a) were identified by employing a similar procedure as the one described in Section 2.2.1. In addition, during the calculations of the normal stresses, it was hypothesised that the maximal principal stress direction of the risk point did not change when alternating the pre-treatment conditions. It should be further noted that the compressive stress (shown in Fig. 11b) might imply the presences of the crack closure effect and the strain hardening effect. These effects could hinder the crack from growing and, therefore, also contribute to the enhancement of fatigue life of the treated valve. Figure 11: Open in new tabDownload slide Comparison of normal stress distributions on a tiny area of assumed crack plane in non-treated and pre-treated conditions. (a) Highlights of crack plane and local coordinating system. (b) Normal stress distributions on defined areas, considered at time-step 5. Figure 11: Open in new tabDownload slide Comparison of normal stress distributions on a tiny area of assumed crack plane in non-treated and pre-treated conditions. (a) Highlights of crack plane and local coordinating system. (b) Normal stress distributions on defined areas, considered at time-step 5. Figure 12 indicates the minimal life results of the critical regions within the TC-implemented valve. These results were estimated by WB-mean, taking into account different assigned materials. There were two types of materials, including the wrought material (AW6082T6) and the printed materials (e.g. AlSi10Mg, 17–4PH SS, and 18Ni300). From the indicated results, it can be seen that the life of the printed aluminium valve was remarkably shorter than that of the one made of AW6082T6, considered in the pre-treated condition. Such life was 1.23E4 cycles, which is far below the minimum required life of 1.5E5 cycles [Commission Regulation (EU) No 406/2010, 2010]. Since the values only represented the number of cycles at which the earliest crack nucleated (not the total life), a practical experiment should be handled to confirm whether AlSi10Mg can be used. It is additionally seen that both the pre-treated steel (17–4PHSS) and the non-treated steel (18Ni300) valves possessed the promising life, which stood far beyond 1E5 cycles. In other words, the two types of steel might be selected as the valve materials. However, it must be pointed out that the above evaluations relied mainly on the durability aspect. The other aspect of compatibilities between the valve materials and hydrogen has not been considered yet. Therefore, further investigations are needed to perform to ensure that the types of steel are among the right candidate materials of the valve. Figure 12: Open in new tabDownload slide Minimal estimated crack initiation life of considered valves, which were ‘made’ of different materials and were considered in either pre-treated or non-treated condition. Figure 12: Open in new tabDownload slide Minimal estimated crack initiation life of considered valves, which were ‘made’ of different materials and were considered in either pre-treated or non-treated condition. 3.3. Crack propagation life assessments for the TC-implemented valve Figure 13 demonstrates the life, required for transiting the middle point of the crack's front to the adjacent position in each propagation step. The figure also represents the J amplitude values, extracted for the point, and the cumulative total running cycles for the transitions. More specifically, Fig. 13a shows the results of the crack propagation process within the sub-model 3; whereas, Fig. 13b indicates the results of the similar process, examined in the sub-model 4. From the figure, it can be realised that the total life required to grow the crack on the crack plane 3 (to the final stage) was considerably longer than that needed to complete the remaining crack propagation. Hence, once the crack front (in Fig. 13b) reaches the TC, the other front (Fig. 13a) would propagate to somewhere near the depth of 4.1 mm. In other words, the TC could be used to indicate the damages not only at the intersection between holes 4 and 6, but also at the intersection between holes 1, 2, and 4. Nevertheless, further practical investigations are essentially handled to confirm this theoretical view. The transitions of the crack fronts on the crack planes could be observed in Figs 14 and 15. Figure 13: Open in new tabDownload slide Estimations of J plastic's amplitude, crack growth life, and cumulative total life for cracks theoretically propagating within (a) sub-model 3 and (b) sub-model 4. Figure 13: Open in new tabDownload slide Estimations of J plastic's amplitude, crack growth life, and cumulative total life for cracks theoretically propagating within (a) sub-model 3 and (b) sub-model 4. Figure 14: Open in new tabDownload slide Representation of different stages of simplified crack (front) growth to depth of 4.0 mm within sub-model 3 (crack front with depth of 4.1 mm was not represented). Figure 14: Open in new tabDownload slide Representation of different stages of simplified crack (front) growth to depth of 4.0 mm within sub-model 3 (crack front with depth of 4.1 mm was not represented). Figure 15: Open in new tabDownload slide Representation of different stages of simplified crack (front) growth to depth of 4.0 mm within sub-model 4 (crack front with depth of 4.1 mm was not represented). Figure 15: Open in new tabDownload slide Representation of different stages of simplified crack (front) growth to depth of 4.0 mm within sub-model 4 (crack front with depth of 4.1 mm was not represented). During the investigations of the crack formation and growth, it should be noticed that the pre-treated condition was taken into account. However, e.g. upon reaching 1.41E5 cycles, the residual compressive stress within layers under the critical surfaces (formed as a result of the pre-treatment) was potentially relaxed. The full or partial relaxation of the stress could lessen the crack closer effect and promote the easier transition of the crack front, under the applied load. Hence, to consider the crack growths in the severe condition for safely designing purpose, it was assumed that a crack was theoretically propagated on the crack plane without influences of the AF pressure. In other words, only the working pressure was applied to surfaces of the flow channels and those of the crack during the FE analyses. It was noted for the last stage of the crack propagation (Fig. 13a) that a significant enhancement of the number of running cycles (required to complete the stage) was observed. This observed behaviour might be derived from the nature surrounding the crack front in the last propagation stage. Within this stage, the leading edge of the 3D crack was estimated to locate somewhere on a region, identified similarly to the potential region 1 (Fig. 9a). On one side, further propagation of the crack on the crack plane 3 led to a closer approach to the channel, marked as C1 in Fig. 9a. On the other side, the maximum working pressure of 87.5 MPa, applied on surfaces of the channel C1, would form the residual compressive stress within the thin layers below the surfaces. This stress stage is believed to result in the crack closer effect and might be the reason, contributing to the observed result. 3.4. The final design of TC in the new TC-implemented valve The final design of the TC-implemented valve is shown in Fig. 16a. From the figure, it can be realised that the updated TC was somehow similar to the initial proposal, which composed of three branches. The main features of branch 1 and branch 2 were almost remained with respect to those demonstrated in Fig. 9c. Branch 3 was modified so that it could be simpler and had an extension to the nearest free surface of the valve. The latter modification targeted a formation of the additional TC port (port 2), which could promote the powder release process after printing. Apart from these features, the new TC was constructed with a constant circular cross-section, having a diameter of 2 mm. The TC feature (large enough cross-section), on the one hand, could also contribute to the powder releasing process. On the other hand, it could help to ensure the successful fabrication of the TC pipe by DMLS, as suggested in (3D Hubs, n.d.), for the valve. Figure 16: Open in new tabDownload slide TC-implemented valve with updated TC structure. (a) Final proposed design of TC. (b) Example of as-printed lattice valve (LV2 structure) holding TC. Figure 16: Open in new tabDownload slide TC-implemented valve with updated TC structure. (a) Final proposed design of TC. (b) Example of as-printed lattice valve (LV2 structure) holding TC. Figure 16b shows a real example of the TC-implemented valve, which was produced by DMLS. The valve was only the as-built structure and would be subjected to post machine to create the threaded features and the sealing surfaces, with the required tolerances. Afterwards, the valve will be inspected via an in-house fatigue test for providing further evidence to support the current work. It should be emphasised here that without consideration of the TC both the designed valve and the printed one were obtained by slightly modifying the called LV2 structure, developed in Cao and Kedziora (2019). In this circumstance, the ‘minor’ channels of the valves, which were only used to connect different types of sensor and therefore would not withstand the working pressure, were proposed to be filled by solid material. In so doing, we could make the valve compatible with our available testing facility, while being able to limit the change in the valve's performance. It should also be noticed that there existed additional four aspects of differences between the current work and the previous one (Cao & Kedziora, 2019), apart from the slight structural differences of the designed valves pointed right above. These aspects have mainly concerned the method of design, the primary purpose of work, the employed boundary condition, and the used material. The aspects are briefly indicated in Table 4. Table 4: The main differences between the current design work and the previous one.a Aspects of difference . Currently considering design work . The previous design work . Method of design • Fatigue-based structural design method (relied on results of fatigue crack initiation and propagation analyses) • Optimisation-assisted design method (relied on free-shape optimisation, topology optimisation, and lattice implementation) Primary purposes of work • Designing a test channel (for the valve) so that it can cut crack planes for early damages detection and its presence does not weaken valve structure considerably • Redesigning valve-body towards enhancement of fatigue life and reduction of mass (with respect to the original valve) Boundary condition • Connection between the valve and a storage tank (or test adaptor) is simulated more realistically by using explicit thread connection • Connection between the valve and a storage tank is simplified by fixing all six degrees of freedom at the connection area Used material • Apart from AW 6082 T6, other printing materials are also considered (e.g. AlSi10Mg, 17–4 PH stainless steel, 18Ni300 steel) • Only considering the fatigue performance of different design structures with AW 6082 T6 Valve configuration • Some holes of the valve which do not practically withstand internal working pressure are filled • All holes of the valve are presented • Containing a TC pipe for early crack detections • Without TC pipe for early crack detection Aspects of difference . Currently considering design work . The previous design work . Method of design • Fatigue-based structural design method (relied on results of fatigue crack initiation and propagation analyses) • Optimisation-assisted design method (relied on free-shape optimisation, topology optimisation, and lattice implementation) Primary purposes of work • Designing a test channel (for the valve) so that it can cut crack planes for early damages detection and its presence does not weaken valve structure considerably • Redesigning valve-body towards enhancement of fatigue life and reduction of mass (with respect to the original valve) Boundary condition • Connection between the valve and a storage tank (or test adaptor) is simulated more realistically by using explicit thread connection • Connection between the valve and a storage tank is simplified by fixing all six degrees of freedom at the connection area Used material • Apart from AW 6082 T6, other printing materials are also considered (e.g. AlSi10Mg, 17–4 PH stainless steel, 18Ni300 steel) • Only considering the fatigue performance of different design structures with AW 6082 T6 Valve configuration • Some holes of the valve which do not practically withstand internal working pressure are filled • All holes of the valve are presented • Containing a TC pipe for early crack detections • Without TC pipe for early crack detection a The previous design work was shown in Cao and Kedziora (2019). Open in new tab Table 4: The main differences between the current design work and the previous one.a Aspects of difference . Currently considering design work . The previous design work . Method of design • Fatigue-based structural design method (relied on results of fatigue crack initiation and propagation analyses) • Optimisation-assisted design method (relied on free-shape optimisation, topology optimisation, and lattice implementation) Primary purposes of work • Designing a test channel (for the valve) so that it can cut crack planes for early damages detection and its presence does not weaken valve structure considerably • Redesigning valve-body towards enhancement of fatigue life and reduction of mass (with respect to the original valve) Boundary condition • Connection between the valve and a storage tank (or test adaptor) is simulated more realistically by using explicit thread connection • Connection between the valve and a storage tank is simplified by fixing all six degrees of freedom at the connection area Used material • Apart from AW 6082 T6, other printing materials are also considered (e.g. AlSi10Mg, 17–4 PH stainless steel, 18Ni300 steel) • Only considering the fatigue performance of different design structures with AW 6082 T6 Valve configuration • Some holes of the valve which do not practically withstand internal working pressure are filled • All holes of the valve are presented • Containing a TC pipe for early crack detections • Without TC pipe for early crack detection Aspects of difference . Currently considering design work . The previous design work . Method of design • Fatigue-based structural design method (relied on results of fatigue crack initiation and propagation analyses) • Optimisation-assisted design method (relied on free-shape optimisation, topology optimisation, and lattice implementation) Primary purposes of work • Designing a test channel (for the valve) so that it can cut crack planes for early damages detection and its presence does not weaken valve structure considerably • Redesigning valve-body towards enhancement of fatigue life and reduction of mass (with respect to the original valve) Boundary condition • Connection between the valve and a storage tank (or test adaptor) is simulated more realistically by using explicit thread connection • Connection between the valve and a storage tank is simplified by fixing all six degrees of freedom at the connection area Used material • Apart from AW 6082 T6, other printing materials are also considered (e.g. AlSi10Mg, 17–4 PH stainless steel, 18Ni300 steel) • Only considering the fatigue performance of different design structures with AW 6082 T6 Valve configuration • Some holes of the valve which do not practically withstand internal working pressure are filled • All holes of the valve are presented • Containing a TC pipe for early crack detections • Without TC pipe for early crack detection a The previous design work was shown in Cao and Kedziora (2019). Open in new tab 4. Conclusions In this study, the fatigue-based approach, employed to design the TC and to virtually evaluate the TC-implemented valve, was demonstrated. The three main stages were proposed to handle the study. They included (i) the identification of the critical locations, (ii) the TC design along with the investigations of crack nucleation in the valve, as well as (iii) the investigations of the simplified crack growth process. From the achieved results, the following conclusions could be drawn up. It was possible to design a complex TC for the hydrogen valve to indicate the status of cracks propagated from the multiple critical sites within the valve. The TC-implemented valve performed as similar as the non-TC one in the non-treated condition. In other words, the valve structure was not weakened by the implementation of the TC. More importantly, depending on the used materials, it was theoretically proved in the study that the new valve structure could satisfy the minimal life of 1.5E5 cycles. For the critical region, where no TC branch was constructed nearby, its damage status could not be directly pointed out. However, it might be possible to indicate the damage via the theoretical correlation between the crack propagation within the region and that within the remaining one. Since the TC structure was complicated, its design targeted the compatibility with DMLS (e.g. 2 mm diameter pipe and two TC ports). In doing so, the TC pipe could be additively constructed by DMLS without difficulty. It was suggested that practical experiments should be handled in the near future to support the shown theoretical results. ACKNOWLEDGEMENTS The authors would like to acknowledge the financial support provided by the University of Luxembourg. Conflict of interest statement Declarations of interest: none. References 3D Hubs . ( n.d .). How to design parts for Metal 3D printing | 3D Hubs . Retrieved January 14, 2019, from https://www.3dhubs.com/knowledge-base/how-design-parts-metal-3d-printing Branco R. , Costa J. D. M., Berto F., Razavi S. M. J., Ferreira J. A. M., Capela C., Santos L., Antunes F. ( 2018 ). Low-cycle fatigue behaviour of AISI 18Ni300 maraging steel produced by selective laser melting . Metals , 8 ( 2 ), 32 . https://doi.org/10.3390/met8010032 Google Scholar Crossref Search ADS WorldCat Cao T. B. , & Kedziora S. ( 2019 ). Innovative designs of an in-tank hydrogen valve towards direct metal laser sintering compatibility and fatigue life enhancement . Structural and Multidisciplinary Optimization , 59 , 1 – 22 .. https://doi.org/10.1007/s00158-018-2174-1 Google Scholar Crossref Search ADS WorldCat Commission Regulation (EU) No 406/2010 . ( 2010 ). Implementing Regulation (EC) No 79/2009 of the European Parliament and of the Council on type-approval of hydrogen-powered motor vehicles , Official Journal of the European Union. https://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2010:122:0001:0107:EN:PDF Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Hutař P. , Kuběna I., Ševčík M., Šmíd M., Kruml T., & Náhlík L. ( 2014 ). Small fatigue crack propagation in Y2O3 strengthened steels . Journal of Nuclear Materials , 452 , 370 – 377 .. https://doi.org/10.1016/j.jnucmat.2014.05.071 Google Scholar Crossref Search ADS WorldCat Hutař P. , Poduška J., Šmíd M., Kuběna I., Chlupová A., Náhlík L., Polák J., Kruml T. ( 2017 ). Short fatigue crack behaviour under low cycle fatigue regime . International Journal of Fatigue , 103 , 207 – 215 .. https://doi.org/10.1016/j.ijfatigue.2017.06.002 Google Scholar Crossref Search ADS WorldCat nCode . ( 2016 ). DesignLife theory guide , HBM United Kingdom Limited: nCode DesignLife Theory Guide Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC News – Road Vehicles . ( 2016 ). Audi unveils h-tron quattro concept car at Detroit auto show . Fuel Cells Bulletin , 2016 ( 2 ), 2 . https://doi.org/10.1016/S1464-2859(16)70003-4 OpenURL Placeholder Text WorldCat News – Road Vehicles . ( 2017a ). Alphabet adds Toyota Mirai cars for fleet of FCEVs in Germany . Fuel Cells Bulletin , p. 2 . https://doi.org/10.1016/S1464-2859(17)30402-9 OpenURL Placeholder Text WorldCat News – Road Vehicles . ( 2017b ). Honda begins Clarity Fuel Cell deliveries in Europe, California . Fuel Cells Bulletin , 2017 ( 1 ), 2 . https://doi.org/10.1016/S1464-2859(17)30002-0 OpenURL Placeholder Text WorldCat Romano S. , Patriarca L., Foletti S., & Beretta S. ( 2018 ). LCF behaviour and a comprehensive life prediction model for AlSi10Mg obtained by SLM . International Journal of Fatigue , 117 , 47 – 62 .. https://doi.org/10.1016/J.IJFATIGUE.2018.07.030 . Google Scholar Crossref Search ADS WorldCat Sellen S. ( 2014 ). Untersuchung der Ermüdungsfestigkeit von komplexen innendruckbelasteten Aluminiumgeometrien und der Lebensdauersteigerung durch Autofrettage (Doctoral dissertation). Campus Kirchberg University of Luxembourg, Luxembourg . Google Scholar Google Preview OpenURL Placeholder Text WorldCat COPAC Sellen S. , Maas S., Andreas T., Plapper P., Zürbes A., Becker D. ( 2015 ). Improved design of threaded connections by autofrettage in aluminium compounds for cyclic high pressure loading: Design calculations and experimental verification . Fatigue and Fracture of Engineering Materials and Structures , 38 ( 6 ), 714 – 729 .. https://doi.org/10.1111/ffe.12270 Google Scholar Crossref Search ADS WorldCat Sellen S. , Maas S., Andreas T., Plapper P., Zürbes A., Becker D. ( 2016 ). Design rules for autofrettage of an aluminium valve body . Fatigue and Fracture of Engineering Materials and Structures , 39 ( 1 ), 68 – 78 .. https://doi.org/10.1111/ffe.12328 Google Scholar Crossref Search ADS WorldCat Wang C. H. , Brown M. W. ( 1993 ). A path-independent parameter for fatigue under proportional and non proportional loading . Fatigue & Fracture of Engineering Materials and Structures , 16 ( 12 ), 1285 – 1297 .. https://doi.org/10.1111/j.1460-2695.1993.tb00739.x Google Scholar Crossref Search ADS WorldCat Yadollahi A. , Thompson S. M., Elwany A., Bian L. ( 2017 ). Effects of building orientation and heat treatment on fatigue behavior of selective laser melted 17-4 PH stainless steel . International Journal of Fatigue , 94 , 218 – 235 .. https://doi.org/10.1016/J.IJFATIGUE.2016.03.014 Google Scholar Crossref Search ADS WorldCat © The Author(s) (2020). Published by Oxford University Press on behalf of the Society for Computational Design and Engineering. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com
TI - Designing an early failure indicator channel for an in-tank hydrogen valve via a fatigue-based approach
JF - Journal of Computational Design and Engineering
DO - 10.1093/jcde/qwaa007
DA - 2020-02-01
UR - https://www.deepdyve.com/lp/oxford-university-press/designing-an-early-failure-indicator-channel-for-an-in-tank-hydrogen-2YeZ2pqrfU
SP - 65
EP - 77
VL - 7
IS - 1
DP - DeepDyve
ER -