Skip to main content

Pre-normative research on resistance to mechanical impact of composite overwrapped pressure vessels

Final Report Summary - HYPACTOR (Pre-normative research on resistance to mechanical impact of composite overwrapped pressure vessels)

Executive Summary:
The main objective of Hypactor was to provide recommendations for Regulation Codes and Standards (RCS) regarding the qualification of new designs of Composite Overwrapped Pressure Vessels (COPV) and the procedures for periodic inspection in service of COPV subjected to mechanical impacts. It leads to a better understanding of the COPV behaviour and integrity when submitted to mechanical impacts.
The consortium started by studying the damages induced by impacts on COPV and the correlation between impact parameters and damage characteristics. From a first screening, conditions that lead to immediate failure were identified and several conditions were selected for further investigation with respect to short and long term residual performance. This leads the consortium to generate an extensive results database, which represents by itself a huge step forward for the COPV open literature. A 3D model was developed to predict impacted cylinder residual strength due to impact and to predict the damage growth due to fatigue.

COPVs are subjected to periodic inspections by Non Destructive Testing (NDT) methods. A wide range of methods has been assessed and compared. Complete investigation and comparison of NDT (Visual, AE, MAE etc.) with respect to the tolerance to impact for the different applications. Acoustic Emission pass-fail criteria have been defined both for real-time and post-processing and have been validated over a wide number (>80 tests) and range of type-4 cylinders (from 25MPa to 95MPa service pressure, from 36L to 513L water volume) as relevant criteria to remove cylinders in service when subjected to critical loss of performance caused by mechanical impacts. Additional work (not planned) has been conducted in order to include a complementary NDT that is pushed to standardization in US mainly in order to compare it with those already in use in the project.

The consortium developed an extensive database, capitalizing all the experimental impacts performed and related characterizations to provide a better knowledge on the relationship between the impact, the damage induced in the composite vessel and the loss of performance of the cylinder at short term and after further pressure loads in service.
Finally, recommendations to standardization committees have been generated from the project experimental results mainly with respect to qualification testing and periodic inspection. For qualifying tests, recommendations based on test results achieved, indicate a rather sharp change in cylinder performance when the impact energy increase above a cylinder specific impact energy value. The criteria for determining the threshold value are no visual fibre breakage and the burst pressure after impact shall not be lower than the minimum burst pressure for the cylinder design.
For periodic inspections, the curve based on burst pressure reduction of empty vessels is relevant to calibrate inspection of type IV cylinders subjected to impact. Indeed, in case of impact the burst pressure reduction is the major failure mode and fatigue has little influence on such cylinders.

The scheduled MAE tests in Hypactor were warmly welcome by the ISO group on MAE to feed their understanding of MAE on composite cylinders. The MAE database built in Hypactor is unique in Europe, as it was the first evaluation of this method on periodic inspection of type IV COPV. Results and comments were communicated to the WG15 through the project workshop (June 2017) and also to AFNOR; a second presentation is planned for the next ISO meeting in November 2017 in the USA.
Project Context and Objectives:
a. Study of damages due to impacts
This part was dedicated to study of damages due to impact and was divided into 5 tasks:
In service incident and operating experience feedback
The goal of this task was to define the most likely and the most severe conditions of impact to help defining the boundary conditions of the impact test matrix. These conditions have been obtained thanks to public literature and feedback experience from AIR LIQUIDE and HEXAGON.
Test matrix definition
The first step of the test matrix consisted in two studies:
- A threshold study to determine the impact energy that leads to immediate failure ,
- A parameter study to provide data on impact damage related to weight, velocity and dimensions of impactor.
This first step was performed on 36L – 70 MPa tanks.
From this first step, 2 or 3 most relevant conditions of impact have been chosen to carry on the experimental investigation. In this second step, the influence of parameters will be studied. These parameters include vessel characteristics (composite thickness, impact location, glass layer) and impact parameters (impactor shape, repeated impact).
Impact testing
The goal of this task was to generate impacts according to the test matrix. Drop tower and pneumatic canon were used as impact means to accommodate with the whole range of impact parameter sets.

Characterisation of impact damage
The goal of this task is to characterize damage induced by impacts. Multiple damage characterisations have been applied to ensure proper description of the impact in volume. A result test matrix integrating all the experimental data collected in the project has been built. These data have been used to select the most relevant impact parameter sets for residual performance assessment and for numerical model development and validation.
Relationship between impact and damage induced and choice of representative impact conditions to be tested in the following part
Two impact conditions for WP3 have been defined based on the first results available and previous experience from AL:
1) Empty vessel; impact energy 1 kJ; hemispherical tup
2) 700 bars N2; impact energy 3 kJ; hemispherical tup.
These damage were interesting as they appear to generate significant internal damage without major modification of the surface. However, these energies do not lead to significant performance reduction at the end and thus another series of impacts with higher energies was also performed in order to create more damage and investigate strong decreases of burst pressures.

b. Residual performance assessment of impacted tanks
This part was dedicated to assess short- and long-term residual performance of impacted tanks, as compared to the behaviour of healthy tanks. Connected to this is also identification and definition of the most critical impact conditions with respect to operational safety of COPV, and development of numerical models to predict the influence of well-known impact defects on composite tank performance and safety.

Short term behaviour of impacted tanks
The goal of this task was to perform impacts on selected vessels, thoroughly characterize the initial state of damage, and asses the short-term behaviour of impacted tanks. The assessment was obtained by gradual pressurizing the vessels to monitor damage evolution with acoustic emission (AE). Prior to that, healthy (unimpacted) cylinders were subjected to the same pressurization tests in order to determine their signature. The final assessment is the residual burst pressure of a vessel.
Long term behaviour of impacted tanks
The goal of this task was to perform impacts on selected vessels, thoroughly characterize the initial state of damage, and asses the long-term behaviour of impacted tanks. The assessment was obtained by exposing the tanks to pressure cycling together with AE monitoring of the evolution of damage. For comparisons sake, the same tests were performed on undamaged vessels.

Modelling
The goal of this task was the development of models that can predict damage growth and residual strength in COPV. A 3D progressive failure model was developed, but simplified models were also investigated.

Definition of the characteristics of the critical damage
The goal of this task was to carry out a complete analysis of the experimental data and reports in order to bring to the fore the most critical impact conditions and the subsequent critical damage with respect to operational lifetime and safety (reduction of performance).
The definition of critical damage depended on the level of safety (or loss of performance) that the COPV should have for its remaining lifetime. Since current standards do not address this question explicitly, suggestions for a definition were developed. The definition of critical damage is based on the requirements of existing standards and suggestions for new approaches were made according to standards and the experience from the field. The value(s) of critical damage threshold were defined in this task. The parameters of the critical damage were used to determine the AE pass – fail criteria. They were also compared to current normative testing protocols to determine if evolutions were required and eventually propose the most appropriate procedures with respect to HYPACTOR results.

c. Inspection methods
Periodic inspection of COPV by Non Destructive Testing (NDT) methods has been assessed. Two NDT methods were studied as they are the most mature, already supported in standards (visual inspection) or in development (Acoustic Emission Testing). Other NDT techniques have been applied in the project as R&D expertise tools or exploratory techniques in order to characterise the characteristics parameters of the damages induced by mechanical impacts and thus allow the calibration of AE and visual inspection with respect to critical damage level.
The objectives of WP4 were to define:
• NDT protocols to be used in the prior activity to assess composite damages (characterisation of damages);
• Pass Fail criteria for visual and Acoustic Emission Testing methods corresponding to the pass-fail criteria in terms of tolerance to damage of composite pressure vessels, identified during the residual performance assessment;
• Recommendations for inspection procedures suitable with operating constraints for considered applications.

d. Dissemination and RCS recommendations
One major outcome of the project are recommendations for periodic inspection of COPVs submitted to mechanical impacts and for ISO qualification testing. These recommendations have been done to different standardization committees.
Project Results:
1. Study of damages due to impacts
In service incident and operating experience feedback
The review of literature & database for incidents in the field leads to the following conclusions:
• Most of accidentology is related to on-board storage of CNG.
• COPV failure due to mechanical damage is not the most critical failure mechanism of COPV.
• A default in integration can contribute to COPV damage: abrasion due to mounting, cylinder fall in a traffic accident, failure of brackets.

Test matrix definition
According to the review, a test matrix has been built in. First step consisted increasing various damages that will be characterized by different NDT methods (T2.4) and finally defining the borders of the impact energy range of the test matrix. In a second step, few impact conditions were selected for further investigation in the second step. These impact conditions are diameter, weight and velocity (kinetic energy) of the projectile and inner pressure of the tank.
For WP2, the reference vessels are 36L@70MPa. However, two vessels 250L@95MPa have also been impacted to study the influence of the characteristics of the vessel.
The first results of impact testing and damage characterization leads to the following conclusions:
- The energy level that leads to immediate failure is lower than expected (for the impact conditions studied),
- Damage overlap requires a decrease of the number of impacts per vessel.
In fact, up to 9 impacts were first planned on each vessel for economic reasons, but the overlap between damages leads to a decrease of number of impact per vessel.
Therefore, three energy levels were studied by using several ways to get the same energy (by changing weight and speed of the projectile). These levels were 3 kJ, 5 kJ and 7 kJ, as the damage created by 1 kJ was hardly visible at the surface and 10 kJ was too high compared to the vessel stiffness.

Then, the following conditions have been validated and selected by partners based on the results that were presented by CEA and ISA to pursue the investigations:
- 1kJ on vessel without pressure;
- 3 kJ on vessel with 700 bar N2.
The same projectile was used, in order to gather comparable data (56 mm hemispherical projectile).

The energy levels have been chosen so that none should lead to immediate failure. Indeed, 5 kJ has been considered too high, regarding the fact that several impacts were performed on each vessel and that a pneumatic burst had occurred at 6 kJ for a vessel pressurized at 875 bars (6th impact) and at 7 kJ for a vessel pressurized at 700 bars (5th impact). That is the reason why 3 kJ has been chosen. It has been decided to study two different energy levels. 1 kJ has been selected because it creates a hardly visible external damage but a noticeable damage in the thickness of the composite (cf. NDT examinations).
The higher energy will be studied with the higher internal pressure.
Regarding the internal pressure, it has been chosen to study the service pressure. Indeed, the behaviour of the vessel pressurized at 875 bars is quite similar to the behaviour at 700 bars and it is rarer. So studying the service pressure appears more relevant to the consortium. Testing a vessel with a low pressure is much more complicated and longer than testing it without pressure. It has been demonstrated that the behaviour of the vessel is similar to those with minimum pressure. So, one of the two impact conditions has been chosen without pressure.
Using these two conditions, the parameter study was carried out on the two following item families:
- Impact parameters : projectile shape ; repeated impact
- Vessel parameters : composite thickness ; shot location (dome) ; impact revelatory
For the impact parameter study, each impact parameter was studied on two vessels. The projectile shapes are 56 mm diameter and three ending (hemispherical, conical and flat).
Repeated impacts were performed 2 or 3 times, and the cumulative energy was kept constant.
Specific configurations were studied:
- Composite thickness influence has been studied with thicker 36L vessels of two designs: thickness + 20% and thickness + 40%. Two vessels of each type are required. Moreover, a larger vessel (255L@950 bars) has been studied (2 vessels).
- Impacts in the dome area were performed at about 45° to be representative of qualification conditions
- impact revelatory influence was studied using dedicated vessels with external glass layer
In that way, a large number of damages has been generated, delivered to feed the study of the influence of impact / vessel parameters on the damage created and finally aggregated in a large experimental database.
Impact testing
An extensive impact test campaign has been carried out according to the test matrix.
Two main testing facilities have been used at CEA to generate the impacts: a drop tower and a gas launcher. A gas pressure device has been used in order to perform impacts on pressurized vessels. All these settling and devices are presented on the following pictures.
All impacts have been recorded using a fast camera that allows to measure the impactor penetration depth through the vessel thickness and the speed of the projectile (incident and rebound). A test sheet has been established for each impact test, giving:
- the name of the test,
- the vessel number,
- the characteristics of the impact,
- pictures at the impact and at the maximum penetration,
- pictures of the external damage and of the video-endoscopy.

141 impacts have been performed on vessels 36L@700 bars and 255L@950 bars and all the damages created have been characterized using NDT.
In order to manage the vessels analysis after impact, a tracking sheet has been established for each vessel, which gives its location and the tests performed on the vessel. All the tracking sheets are shared among the partners of the project and have followed the vessels through numerous shipments between partners.
A correlation between incident energy and rebound energy has been established using results for 43 impacts.
As a conclusion, we can consider that the energy is mainly absorbed by the vessel and the percentage is around 60-70%. All the damages created have been analysed to complement the database and increase the understanding of COPV behaviour under impacts.
Characterisation of impact damage
At the beginning of this project, numerous Non Destructive Techniques (NDT) have been assessed in order to characterise the corresponding damage. Thermography, Shearography, Deflectometry and Scanner 3D were used and proved their ability to detect almost all the damage area (external surface) but without any new information compare to visual inspection.
This is why it was chosen to perform only visual testing for surface damage characterisation and ultrasonic testing for internal damage characterisation on each impact location. As ultrasonic testing is sensitive only to delamination defect, some additional computed tomography has been conducted to complete characterisation (cracks and matrix failure) on main impact conditions.
A large database has been created in xls format. It contains for each impact, impact characteristics, NDT characterisation and associated residual performance. This database will be granted an open access after the end of the project. Impact characteristics are given in terms of impactor shape, impact incident and rebound energy, impactor penetration (measured by the fast camera). Residual performance is given in terms of burst pressure if vessel goes to burst or other kind of residual performance (leak during impact test or during cycling). NDT characterisations are given for each method in terms of sizing and for tomography only by naming the type of damage induced by impact with following terminology:

For sizing, all partners decided to define the damage with the size of an envelope shape. This shape could be in 2D in case of visual testing or in 3D for ultrasonic testing and tomography testing as illustrated in Figure 4 below is according the hoop axis and y is according the main axis of the vessel.
This database includes also some results coming from inspection method (Acoustical Emission and Modal Acoustical Emission).
Relationship between impact and damage induced and choice of representative impact conditions to be tested in the next work package.

The influence of the following parameters was investigated through the WP2 tests:
- Impactor shape
Three shapes were investigated: hemispherical, cylindrical, and conical.
Tests were performed at high impact energy (5 kJ) in order to create significant damage and facilitate the comparison. It was found that conical impactor penetrates deeper inside the composite depth, creating more severe damage, but also visible fibre breakage. Such impact can easily be detected by visual inspection and will lead to rejection of the vessel. On the other hand, damage created by cylindrical impactor is hardly repeatable: if the drop is perfect, the flat surface hits the composite, leading to smaller and less visible damage; but in case of a light tilt (e.g. due to wind) the impact will be done by the edges, making more damage but also very detectable fibre breakage.
As a result, hemispherical impactor was retained for the following tests and the normative recommendations. It enables more repeatable impacts, which are also more hazardous with respect to periodic inspection as they could be missed visually.

- Weight and speed of the impactor
Tests were performed at a same energy level (ca. 3 kJ) using different weight / speed combinations ((i)1.3 kg, 70.2 m/s (ii) 52.7 kg, 10.7 m/s (iii) 102 kg, 7.6 m/s) and two test setups (pneumatic canon and drop tower). The damage created was observed by tomography and no significant difference was found between these three tests: energy of impact matters more to damage created than weight and speed of the impactor.

- Empty vs pressurised cylinders (at 20 or 700 bar)
Visual inspection shows, that for the same level of energy (and same hemispherical impactor) the external aspect of empty and pressurised cylinders is very similar. Once an energy threshold is determined, it is possible to determine the corresponding aspect and calibrate visual inspection. It must be noted that this has been studied only on HEX 36L cylinders, and may be very dependent on the vessel (especially diameter or surface finish).
On the other hand, ultrasonic testing shows that the extent of delamination through thickness depends on internal pressure during impact. The delamination cone is wider when the cylinder is impacted empty, whereas the damage is more concentrated under the impact point at 700 bar. This is not observable when using visual inspection only; which shows that different methods can be complimentary to fully investigate the damage (still, the ultrasonic testing does not give information about fibre breakages, which dominate the decrease of performance).

- Vessel’s characteristics (thickness, geometry)
The influence of geometry was studied by comparing 36 L and 513 L vessels. The same trends were observed on induced damage for both geometries.
In order to investigate the influence of thickness independently of other parameters (especially diameter), vessels with extra thickness were made specifically by HEX for the project. 36 L vessels with 20% and 40% more thickness were made, and their behaviour under a 1 kJ impact (empty) was compared with standard 36 L vessels. There is no significant difference in terms of visual inspection, and ultrasonic testing shows the same behaviour on induced damage with change of thickness (for 1kJ impact, there is very deep delamination regardless of the thickness).

- Repeated impact
Single impacts were compared to repeated impacts at lower energy (for a same total energy). The picture below shows the UT results for a cylinder impacted on three locations (i) 3 times 1 kJ, (ii) 1 time 3 kJ, (iii) 2 times 1.5 kJ. It appears that repeated impacts contribute to induced damage, but the single impact at higher energy seems to create more damage. This observation has been confirmed by residual burst pressures of other cylinders (one time 3 kJ leads to lower residual performance than three times 1 kJ).
- Presence of a glass fibre layer as impact revelatory
Two 36L were manufactured with an extra layer of glass fibre / epoxy on the outside. They were impacted at respectively 1 kJ (empty) and 3 kJ (empty) and the damage was assessed through visual inspection and ultrasonic testing (UT). The glass layer was found to have no effect on the UT signature, and did not prove useful for visual inspection.
(Test are realised on only one specific glass fibre layup. Other ways of making the glass fibre layups might have different effects, but this is not investigated HyPactor.)

- Impact location
Three different impact locations on the cylindrical part were studied (in the middle and at both ends of the cylindrical part). No significant difference of damage created was observed. Sometimes, for impacts close to the ends, it seems as if the beginning of the dome could limit the propagation of the delamination in that direction. Damage may be a little deeper when impacts are done in the middle. The picture below shows the UT behaviour of an empty cylinder impacted at 3 kJ on 3 different locations.

2. Residual performance assessment of impacted tanks

Short term behaviour of impacted tanks
In this section, two types of vessels were subjected to tests: standard 36L vessel with nominal working pressure (NWP) of 700 bar and 513L vessel with NWP of 248 bar. Two relevant impact conditions were selected for 36L/700 bar vessels, based on previous evaluation:
- 1 kJ impact on empty (unpressurized) vessel,
- 3 kJ impact on vessel pressurized with nitrogen to 700 bar (NWP).

These impact conditions were used for most vessels as it was presumed such impacts may lead to decrease of performance without obvious superficial damage. 3 vessels were dedicated for each of those conditions. Because it was deemed valuable to test also higher impact energies for comparison, following impact conditions were also tested, with one vessel each:
- 3, 4, 5, 7 kJ impacts on empty vessel,
- 4, 5, 6 kJ impacts on vessel pressurized with nitrogen to 700 bar,
513L/248 bar vessels were impacted at following conditions:
- 3 kJ impact on empty vessel,
- 1.5 and 3 kJ impacts on vessel pressurized to 176 bar.

As means of performing the impact, a drop tower test has been selected, with a 50 kg impactor dropped from height corresponding to the desired energy. The impactor tup is hemispherical with a diameter of 56 mm. Each vessel was impacted only once, on the cylindrical part. The impactor was striking perpendicularly to the vessel surface. All these impacts were performed at CEA/TEE facility in Le Barp, France. For more details about impact tests, refer to WP2 § and/or deliverables.
Impacted vessels were subjected to non-destructive ultrasonic (US) examination by ISA to characterize the extent and shape of impact-induced damage (mainly delamination). After the investigation of damage, vessels underwent the short-term performance test. Full short-term performance test consisted in an acoustic emission (AE) test and a burst test. Vessels after severe impacts were subjected only to burst test. AE is further elucidated in section dedicated to WP4.
Both at CEA and WRUT, the burst tests of the 36L/700 bar vessels were carried out in a dedicated lab using water as medium. The burst test at ambient temperature was performed according to regulation ISO 11439 (Annex A - A.12). An hydraulic pump increased the pressure in the vessel until burst at ~5 bar/s. A sensor allowed monitoring and recording the evolution of the pressure in the tank. This sensor was located as close to the vessel as reasonable to avoid any damage during the burst of the vessel. At CEA, a high speed camera recorded the burst of the vessel.

Following figures present short-term performance residual burst pressures of 36L/700 bar vessels, normalized to mean burst pressure of healthy vessels, as a function of incident and absorbed energy of impact. It is evident, that there is no much difference between the two pressure changes.
According to these results and only for our study configuration, no significant decrease of short term performance for impact energy (incident) up to around 2 kJ for empty vessels or 3 kJ for pressurised ones could be observed. For impact energy equal or higher than these thresholds, the damage is clearly visible and should not be missable during visual inspection, especially if the impact occurred on a pressurized vessel (as the threshold impact energy is higher).
Incident and absorbed energy is correlated to a high degree, especially for given internal pressure. However also for empty and pressurized vessels taken together, the correlation coefficient is high.

According to these results and only for our study configuration (reservoir type and size + impact conditions), we can assess that there seems to be no significant decrease of short term performance of 513L/248 bar vessels for impact energy up to around 2-2.5 kJ both for empty vessels or for pressurised ones.

Long term behaviour of impacted tanks
The impact conditions selected were the same as for short-term performance (see the relevant sub-section for details), with some additions. The additional testing encompassed the repeated-impact tests which, for each impact conditions, distributed the total impact energy to 3 separate impacts to the same area instead of concentrating it in one impact (e.g. 3x1 kJ impact instead of a 3 kJ impact). All other conditions remained equal. Altogether, following conditions were tested:
- 1 kJ impact on empty (unpressurized) vessel,
- 3x0.33 kJ repeated impact on empty (unpressurized) vessel,
- 3 kJ impact on vessel pressurized with nitrogen to 700 bar (NWP),
- 3x1 kJ repeated impact on vessel pressurized with nitrogen to 700 bar (NWP),
- 3, 4, 5, 7 kJ impacts on empty vessel,
- 4, 5, 6 kJ impacts on vessel pressurized with nitrogen to 700 bar,
All these impacts were performed at CEA/TEE facility in Le Barp, France.

Impacted vessels were subjected to non-destructive ultrasonic examination by ISA to characterize the extent and shape of impact-induced damage (mainly delamination) as a baseline for assessment of damage evolution during long-term loads.
After the ultrasonic investigation of damage, vessels underwent the long-term performance test at WRUT. Long-term testing campaign consists of following steps:
• Acoustic emission (classic and modal) testing of vessel after impact to 105 MPa
• 50 pressure cycles to 87.5 MPa to remove excessive emissivity
• Acoustic emission to 105 MPa testing after 50 cycles
• 5000 pressure cycles to 87.5 MPa
• Acoustic emission to 105 MPa testing after 5000 cycles
• 10,000 pressure cycles (15,050 cycles in total) to 87.5 MPa
• Final acoustic emission testing to 105 MPa
• Residual burst test.

Pressure cycle tests at ambient temperature conditions were based on relevant regulations. First, the vessel to be tested is filled with a non-corrosive fluid. Most often a high quality hydraulic oil, or a mixture of water and glycol is used. Next, according to the standards, vessels are pressure cycled with minimal pressure of 2 MPa (20 bar) and maximal pressure of 87.5 MPa (875 bar) at a rate of not more than 10 cycles per minute.
Seven additional vessels were impacted at elevated energies to test behavior above the critical impact energy. These vessels were subjected to abbreviated long-term performance tests. Abbreviated long-term testing campaign consists of following steps:
• Acoustic emission (classic and modal) testing to 105 MPa of vessel after impact
• 50 pressure cycles to 87.5 MPa
• Acoustic emission testing to 105 MPa after 50 cycles
• Residual burst test

Following the completion of the cycling trial plan, residual performance was evaluated by burst test. Most of these tests were performed at WRUT; some were burst tested at CEA. The equipment used and protocol followed are the same as in the case of short-term performance.
Following graph presents the comparison of burst pressures of vessels without cycling and after 15,000 cycles for different impact conditions.
According to these results and only for our study configuration (reservoir type and size + impact conditions), we can assess that there seems to be no significant decrease of long-term performance (both as a number of cycles in predicted lifetime and as a residual pressure after cycling) for impact energy up to 3 kJ. For impact energy equal or higher than 3 kJ, the damage is clearly visible and it should not be missed during visual inspection, especially if the impact occurred on a pressurized vessel. 3kJ may be a threshold and sensitive to impact conditions.
The following graphs present in graphical form the results for vessels without cycling and cycled for 50 cycles and ca. 15,000 cycles.
According to these results and only for our study configuration, we can assess a marginal reduction of performance (burst pressure) for vessels cycled for 50 cycles at 2-87.5 MPa for vessels impacted while empty. The same is not true for COPV impacted at 700 bar – burst pressures for such vessels are even higher than burst pressures of vessels burst directly after impact.
Given that even after 15,000 cycles there seems to be no significant decrease of long-term performance (both as a number of cycles in predicted lifetime and as a residual pressure after cycling) for impact energy up to 3 kJ, it would seem that the long term performance of composite pressure vessels is not significantly different from short term performance.

3. Inspection methods
Industrial constraints for NDT in operation
The goal of this task is to give the operating conditions for the implementation of Non Destructive Testing (NDT) methods on type IV COPV. Two NDT methods have been studied, as they are the most mature already supported in standards and potentially used for periodic inspection of COPV: visual inspection of accessible areas and Acoustic Emission Testing during pressurization.
To this aim, HEXAGON and Air Liquide worked together in order to provide respectively experience in pressure vessels manufacturing and operating & maintenance.
The operating conditions to consider for the implementation of AE monitoring and visual inspection directly on site were provided. In order not to dismount vessels from the frame, the best way to achieve Acoustic Emission Testing is to use a filling bench in a filling centre or directly at the on-site application, allowing to pressurize at least to maximum service pressure.
Another possibility could be to perform Acoustic Emission Testing in a dedicated AE workshop (eventually in an existing retest centre) or at an accredited agency. This solution would lead to more perturbation in the supply chain process for an industrial gas provider, but could be a backup solution.

Definition of characterisation NDT protocols
A list of potential NDT was established with expected performance in terms of damage characterization, threshold and data needed for simulation. NDT techniques are therefore dispatched on different partners (ISA, CEA and WRUT).
The first impacted cylinders have been inspected at CEA with X-Ray technique and videoendoscopy, and at ISA with the following techniques: Ultrasonic, Laser 3D scanning, thermography, deflectometry, shearography.

In order to limit discrepancies, NDT procedures were written by ISA. These procedures include calibration, setting-up and reporting format for each NDT. The relevant NDT was then selected in accordance with all partners, and applied for damage characterisation on all vessel experienced.

Collection of AE data
Taking into account the AE trial plan, some AE tests have been performed at different state of the vessels: after impact, after 100 pressure cycles, after 5100 cycles and after 15100 cycles.
First, three healthy vessels (not impacted) have been AE tested in order to evaluate their acoustic signature and to check a possible dispersion in acoustic responses. Results are shown in Figure 23:
A first result is that AE decreases with the number of cycles, and is quite similar after 5100 cycles and after 15100 cycles, indicating that the state of the vessel (in terms of damage) has undergone any evolution between these two cycling periods.
The second result is that there is a very low AE dispersion between the three vessels at a given number of cycles. It suggests no need to have initial AE on all vessels, and then we have taken those three AE as a baseline signature for all 36L COPVs. The AE trial plan with impacted vessels was updated several times in order to take into account the AE results, and to try to properly adjust the critical impact energy to be screened with AE.
With the approval of the European Commission, another method called Modal Acoustic Emission (MAE) was found relevant by the consortium and investigated in this project. This method was under project ISO DIS 19016 and was implemented in Hypactor by ISA, in September 2016.

3.a. AE protocol
3.a.a. AE systems
AE tests on 36L vessels were performed using the WRUT AMSY6 system (8 channels) with resonant VS150 sensors. Six spring-loaded sensors (number 2, 3, 4, 5, 6 and 7) were placed on the cylindrical part by means of three belts, and acoustically coupled with vacuum grease. Sensors number 1 and 8 were placed on bottoms.
For 255L and 513L vessels, AE was performed using ISA AMSY5/6 systems (36 channels) with resonant R15 sensors and AEP3N preamplifiers.

3.a.b. Sensors location
A planar configuration was used in order to cover the overall vessels and provide an accurate localization of damage. The number of sensors was defined to ensure a similar sensitivity for all vessels: 8 sensors for the 36L, 20 sensors for the 255L and 18 sensors for the 513L. For 255L and 513L, sensors distance is around 40 cm in the longitudinal direction and 50 cm in the hoop direction. For the 36L, the hoop distance is the same as before (50 cm) but only 24cm in the longitudinal direction.

3.a.c. Localization performance
A verification of events localization accuracy has been carried out by way of pencil lead breaks (PLB) in close neighbourhoods of the impacted zones. That means if the localization accuracy result is bad, waves velocity should be adjusted. This verification allows also determining if planar localization could be adopted for the vessel monitoring. The obtained result is shown in Figure 25. As it can be seen, it is satisfactory despite the heterogeneous characteristic of the composite. Polygons are superimposed on sensors distribution schematics (blue, purple and orange squares in the same figure) for delimitating the impacted regions. Hence, during the monitoring period, the AE events, which will be located inside these polygons, should be associated to damages growth.

3.a.d. Pressure Cycle
When using AE for periodic inspection of vessels, the loading used to be gradually increased up to the maximum pressure (Pmax) which will be at least 110% of the maximum applied in service (MAP) during the reference period before the test (12 months).
Assuming that vessels were filling various times by year (with hydrogen), the maximum pressure applied is then equal to the pressure filling (estimated to 875 bars= 1.25Ps for service pressure equal to 700 bar in the case of 36L). Therefore, the maximum pressure during the AE monitoring must be at least equal to 1.35Ps. It is important to note that the pressure sequence has to start at a pressure less than MAP and gradually increased until the maximum pressure, by applying various bearings. This enables to identify the damage state and the evolution but also to avoid a sudden burst. To avoid too sudden stress relaxation, strain rate during depressurization should be similar to that during pressurization. It should not exceed 5% (50bars/min) of the maximum pressure test per minute and should be as regular as possible.

3.a.e. Criteria according to the Guideline
The only code that provides AE procedure for periodic inspection of pressure vessels is the French Guideline for acoustic emission testing of pressure equipment. This document was established by the French association of pressure equipment engineers and all procedures were defined by the acoustic emission working group (GEA). The appendix 7 which was approved in 2014 by the French authority provides the methodology to be applied to define a procedure applicable to composite pressure vessels. Although the appendix 7 does not cover transportable vessels, the methodology remains applicable and was applied on this project. The aim was to assess compliance of criteria thresholds defined in this appendix for Hypactor vessels and therefore to rely on this solid reference to establish pass/fail criteria.

3.b. MAE protocol
Hypactor Modal Acoustic Emission application provided a database of 55 tests, carried out simultaneously with AE:
• 2 on Healthy vessels = 2 vessels
• 16 impacted vessels at 1 kJ empty = 7 vessels
• 24 impacted vessels at 3 kJ 700 bar = 6 vessels
• 7 impacted vessels at 3,4,5,7 kJ empty = 4 vessels
• 6 impacted vessels at 4, 5,6 kJ 700 bar = 3 vessels

3.b.a. MAE equipment
The acoustic emission system was an 8 channels DM system manufactured by Digital Wave Corporation. Eight wide-band sensors B1025 (50 kHz–2.0 MHz) were used, and the sampling rate for the system was set to 5 MHz to ensure all sensor output data was being recorded. 8192 data points for each waveform was recorded including 2000 pre-trigger points. Threshold voltage was set to 0.1V and the gain to 48 dB.

3.b.b. Sensors location
For 36L vessels, the same sensors position has been adopted in order to compare MAE & AE results.

3.b.c. Criteria of the ISO/DIS 19016 and Performance

The three following criteria shall be applied to the waveforms recorded during the MAE test pressurization. If any of the following criteria is met, the cylinder shall be rejected.

• Rejection due to partial fibre bundle rupture criteria
The criteria 1, 2 and 3 in Annex A (of ISO19016) shall be applied to determine if fibre bundle breakage has occurred during the MAE test. If all the criteria (1, 2 and 3) in Annex A have been met, the fibre bundle rupture rejection criteria have been met. When the energy of a single fibre bundle rupture event on any channel exceeds UFBBAE rejectable fibre bundle rupture damage has occurred during the MAE test.
In the absence of testing or calculation that demonstrates otherwise, a value of F1 = 100, which is equivalent to stating that 100 average strength fibres ruptured in a tow (partial bundle rupture), shall be used.

• Rejection due to single event energy
If the energy of a single event on any channel exceedsF2*UFBAE, the cylinder shall be rejected.
In the absence of testing or calculation of the allowance factor, a value of F2 = 10,000 shall be used.
In the above, UFB is calculated using the average breaking strength found in literature, from the manufacturer’s data or independent test data.

• Rejection due to background energy (BE) and background energy oscillation (BEO)
The cylinder shall be rejected if the background energy increases to the point that U(BE) >=M1 (U(QE)), where U(QE) is the quiescent background energy. A value of M1 = 2 shall be used unless test data or calculation demonstrates that a larger (less conservative) value is appropriate to the type of material and vessel under inspection. If oscillations in the background energy (difference between the maxima and minima values of background energy from neighbouring cycles) greater than M2 are observed to occur at any time during the test, the vessel shall be depressurized immediately and the cylinder rejected. A value of M2 = 2 shall be used unless test data or calculation demonstrates that a larger (less conservative) value is appropriate to the type of material and vessel under inspection.

3.c. Development & Validation of pass-fail criteria by NDT (visual inspection & AE)

3.c.a. ISA AE CRITERIA
ISA AE criteria have been built in 2016 in the framework of Horizon Hydrogen Energy (H2E) project on type IV 140 L transportable vessels. Mechanical impacts have been carried out using a dedicated bench. The configuration of impact was in the dome area at 45°. The same AE criteria developed have been used in Hypactor even if the design, impact configuration and operating pressure are different. The aim is to evaluate their application on 36L Hypactor vessels.
It has been demonstrated that post-analysis criteria of the appendix 7 are too conservative. The same conclusion is also available for real time criteria. These criteria are triggered quickly and lead to a premature shutdown of most of vessels. Indeed, parameters to monitor are numerous and require a solid expertise of the operator.
The alternative proposed by ISA, for real time monitoring, is based only on two criteria, one zonal and one using localization analysis. These criteria are very fast to determine and simple to automate. As all AE tests were carried out by the WRUT, in Poland, and in order to avoid a sudden burst during AE tests, ISA has communicated criteria to WRUT operators and formed them to the identification of critical vessels.
For all healthy vessels, impacted at 1kJ@empty and impacted at 3kJ@700 bar, only three vessels were stopped in real time during the AE test:
- Impacted at 1kj@empty: 2 vessels
- Impacted at 3kj@700 bar: 1 vessel

Visual inspection of these vessels has reinforced the AE analysis by indicating an external damage located in the critical area identified by AE. After verification with CEA, it has been found that damage was induced by the rebound of the projectile after impact. In the case of the vessel 2670-008, the induced damage is very severe (cuts induced by edges of the impactor).
All vessels impacted at high energy were also rejected by real time criteria during the AE test. Therefore, all these vessels are considered as dangerous and were stopped in real time during the AE test. For post analysis criteria, pass-fail criteria have been built, based on the appendix 7 through thresholds adjustment. The application of ISA post analysis criteria leads to a less severe but more accurate classification based on the correlation with burst pressure.

With healthy vessels
36L: 11AE tests => no rejection after ISA AE criteria application
255L: 3 AE tests => no rejection after ISA AE criteria application
513L: 2 AE tests => 1 rejection after ISA AE criteria application (seems to have a leak)
With impacted Vessels

36L:
1 kJ@empty: 28 AE tests => 2 rejections after ISA AE criteria application, due to rebound impact
3 kJ@700 bar: 27 AE tests => 1 rejection after ISA AE criteria application, due to rebound impact
3,4,5,7 kJ@empty: 7 AE tests => All rejected after ISA AE criteria application
4,5,6kJ@700 bar: 6 AE tests => All rejected after ISA AE criteria application
255L : 3kJ_250 bar@255L: 1AE test => no rejection after ISA AE criteria application
513L : 3kJ@empty: 1 AE test => rejection after ISA AE criteria application

AE criteria developed have proven to be relevant whatever the cycling period and were validated for a large database (86 Classical Acoustic Emission tests). Questions remain concerning the validity of criteria achieved on vessels filled with water to pneumatic tests. This topic has been investigated in this work by performing pneumatic AE tests in TEE Bordeaux. Two vessels have been carried out, one healthy and one impacted at 3kJ@700 bar. The application of criteria leads to the acceptance of the two vessels. This result is coherent with burst pressure obtained for 3kJ@700bar. Acoustic signals from gas pressurization was comparable to acoustic signals from hydraulic pressurization, nevertheless, a particular attention should be paid for noise filtering, due to acoustic parasite echoes generated by gas flow near the inlet.
An example of gas noise generated during the pressurization sequence and a solution of filtering were proposed in Figure 31. In the absence of continuous activity, guard sensors can be relevant for noise filtering. Otherwise, other solutions can be used depending of the noise features and location.

3.c.b. Performance of MAE (ISO/DIS 19016) criteria
In the absence of testing or calculation, the ISO/DIS 19016 recommends the use of default values for the calculation of rejection criteria. Four default values are then available in this standard:
- F1=100
- F2= 10000
- M1=2
- M2=2
It has been found that default factors are too conservative and not adapted for type IV Hypactor 36L vessels. However, optimized factors recommended by DWC lead to a better classification:
With healthy vessels
36L: 2 MAE tests => no rejection using optimized MAE criteria (1 MAE test not analysed: data lost)
With impacted Vessels
36L:
1 kJ@empty: 16 MAE tests => no rejection using optimized MAE criteria
3 kJ@700 bar: 27 MAE tests => 12 rejections using optimized MAE criteria impact
3,4,5,7 kJ@empty: 7 MAE tests => all rejected using optimized MAE criteria
4,5,6kJ@700 bar: 6 MAE tests => all rejected using optimized MAE criteria

3.c.c. Comparison of AE & MAE classification
An example of one impacted vessel was chosen to compare the relevance of the two methods. The case of study is one vessel impacted at 3kJ@700 bar and monitored by both methods: AE & MAE.
In Zonal analysis, and according to AE ISA criteria, it is well indicated that the area near the sensor n°5 is rejected. This area is distinguished by a high activity during holds and also high amplitude events recorded during all the pressure sequence. All the others parameters are also in category 3 (rejected). For this kind of analysis, the area near the sensor n°5 could be critical and the vessel is then rejected (see table below).

For a more accurate localization, 8 sensors were considered to build 12 planar triangles, permitting the localization of any AE sources which can occur throughout the entire COPV. As it is shown previously, the accuracy of source localization was validated successfully. The planar localization analysis is based here on the concentration of clusters (group of events) and amplitudes of the located events. It can be noted that the concentration of red cluster (more than 50 located events concentrated in a 10*10 cm square area) is located near the sensor n°5 and more precisely above the impact position. This area is critical according to ISA criteria and the vessel is then rejected (Figure 32).
For MAE analysis according to the ISO/DIS 19016 criteria, using optimized factors recommended by DWC, it was found a significant fibre bundle rupture (energy greater than the limit value fixed by the criteria n°1) near the sensor n°5 (Figure 33). It has been also noticed that the cylinder is rejected due to the background energy oscillation recorded by the sensor n°5 (Figure 34).
Visual inspection (Figure 35) of this vessel was coherent with analysis obtained by both methods AE & MAE. An external damage is located near the sensor n°5 and above the impact position. The damage was induced by the rebound of the projectile after impact. This example argues that both methods are relevant in the determination of the damaged zone and its criticality.

3.d. Conclusion
The approach followed by ISA to develop AE criteria was to use the appendix 7 (of the French Guideline for Acoustic Emission Testing of Pressure Equipments) as a baseline.
First the compliance of appendix 7 criteria has been evaluated, showing that thresholds were too conservative. For this reason, criteria thresholds have be optimized by ISA to feed H2E project vessels (27 AE hydraulic tests on type IV 140L vessels in January-June 2016) and then were applied on Hypactor project vessels.
AE Criteria are divided into two types, real time shutdown criteria and post analysis criteria.
The real time shutdown criteria aims at detecting and rejecting dangerous vessels in real time during the AE test, case of vessels with high loss of performance. ISA based its alternative only on two criteria, one zonal and one using localization analysis. This allows a quick identification of the hot-spot areas.
For all vessels (healthy, impacted at 1kJ@empty and impacted at 3kJ@700 bar), only three vessels were stopped in real time during the AE test and rejection was confirmed by visual inspection and burst pressures reduction.

For vessels impacted at high energy (energy higher than threshold defined in critical damage, see D3.9) all of them were rejected by real time criteria during the AE test.
Post analysis criteria are calibrated to detect and reject vessels with a low loss of performance. The approach followed by ISA was to refine criteria thresholds of the appendix 7 and adapt them for type IV vessels using zonal and cluster localization analysis. The use of both localization methods offers the possibility to obtain a more complete and accurate result.
Post analysis criteria as well as real time criteria have proven to be relevant whatever the cycling period and were validated for a large database, including:
-86 AE hydraulic tests on 36L, 255L and 513L
- 2 AE pneumatic tests.
Acoustic signals from gas pressurization were comparable to those from hydraulic pressurization. Nevertheless, a particular attention should be paid for noise filtering, due to acoustic parasite echoes generated by gas flow near the inlet. In the absence of continuous activity, the use of guard sensors can be relevant for noise filtering. Otherwise, other solutions can be used depending of the noise features and location.

Modal acoustic emission (MAE) is the second method investigated in Hypactor. This method is proposed for periodic inspection of COPVs and was described in the ISO/DIS 19016 stage in 2016 (through ISO TC58/SC4/WG15). With the approval of the European Commission and all partners, this topic was involved in this project and collaboration with DWC was initiated by ISA in September 2016. It is important to note that MAE was applied for the first time in Europe within Hypactor. The application of MAE according to ISO/DIS 19016 was revealed difficult due to the unclear/lack of information in the ISO document. However, thanks to the support of DWC, a great progress has been done on criteria understanding and calculation.
It has been proven that allowable factors (F1, F2, M1, M2), currently defined in the ISO/DIS19016 were too conservative and not adapted for thick type IV Hypactor 36L vessels. However, using optimized factors provided by DWC, results of classification are satisfying.
The advantages of MAE method, are the following:
- A better identification of delamination and fibre fracture, using modal analysis
- Only three rejection criteria are used allowing an automated and fast classification
However, the method is still not mature in Europe and more explanation should be given in the ISO19016. Many comments on draft standard have been transmitted to ISO TC58/SC4/WG15 representatives.

4. Dissemination and RCS recommendations

4.a. Recommendations for ISO periodic inspections
One major outcome of the HYPACTOR project is to propose recommendations for periodic inspection of COPVs submitted to mechanical impacts. This task has been achieved by analysing and describing results from WP3 & WP4 for transcription into recommendations to standardization committees.

Recommendations based on results from Hypactor:
The curve based on burst pressure reduction of empty vessels is relevant to calibrate inspection of type IV cylinders (with plastic liners) subjected to impact. Indeed, in case of impact the burst pressure reduction is the major failure mode and fatigue has little influence on tested cylinders.

In order to determine the impact capacity of a composite cylinder, Hypactor recommends determining the threshold impact each cylinder design can absorb, without reduction of burst capacity (inflexion point on Figure 36 curve). When no better indication is available, Hypactor proposes a formula to make a first estimate of the threshold. Hypactor also recommends an impact test program to be included in qualification testing program (see b.).

Concerning visual inspection:
Based on the burst pressure reduction curve, it is possible to complement the curve with some pictures with respect to impact energy. Using this curve, we can split the recommendations in two:
- A visual aspect showing loose fibres should lead to vessel rejection, whatever the elliptic damaged surface.
- Some pictures of damage after impact may be plotted under the curve, with respect to impact energy level. Visual aspects on the left of the inflexion point represent vessels without burst performance reduction, and visual aspects on the right of the inflexion point should lead to rejected vessels.
These two points should allow rejecting dangerous vessels (i.e. with a decrease of burst pressure) at time of filling.

Concerning Acoustic Emission:
Based on Hypactor results, AE criteria:
- are validated for a wide range of type IV vessels (diameter, NWP, volume)
- have proven to be relevant whatever the cycling period after impact
- may be used with gas (in substitution to water, but particular attention should be paid for noise filtering)
- Need an overpressure to ensure detection of damage created at high internal pressure
Thus, use of AE criteria is more adapted to periodic inspection than to inspection at time of filling.
Calibration of AE criteria should be performed for each vessel design, and connected to the burst pressure reduction curve, as well as visual inspection calibration.
In the same way, others relevant NDT (such as UT or tomography) may be used as complementary methods. The recommendation is still to calibrate any NDT with respect to burst pressure reduction curve. It has to be noted that 3D FE modelling may help the calibration by studying the relationship between damage and residual performance.

4.b. Recommendations for ISO qualification testing

The other major outcome of the HYPACTOR project will also be recommendations for qualification testing of COPVs submitted to mechanical impacts.

Results from Hypactor:
• The impact energy a composite cylinder can absorb depends mostly on the structural composite thickness (driven by minimum burst requirement and diameter) and pressure in the cylinder when the impact occurs.

• Requesting same level of impact energy for all vessels from 1L to 10 000L and pressures from low pressure to more than 1000 bar does not make sense.

• Hypactor results confirm burst pressure reduction as a relevant method to assess the safety level of composite pressure cylinder after impact. Type 4 cylinders have demonstrated little to no reduction in cycle performance up to 15 000 cycles. This might not be valid for other types of cylinders, especially when fatigue on deformed liner is an issue

• An estimation of energy threshold (inflexion point of Figure 36) is given by the following formula:

Eabs/(Pburst*Ri ) = 30.10-6 (units = S.I.)

By using Eabs = 0.65 Eincident (Source = ICHS 2017, paper n°136 “Residual performance of composite pressure vessels submitted to mechanical impacts”), the formula becomes:

Eincident = (Pburst*Ri ) . 30.10-6 / 0.65 . (units = S.I.)

Recommendation:
Cylinder manufacturer should specify an impact energy in relevance with the one estimated with the formula and in accordance with cylinder application, and demonstrate by burst testing that the cylinder can still demonstrate minimum burst capacity and no performance reduction compared to healthy cylinders batch results.
Further testing with higher impact energy might be beneficial for evaluation of impacted cylinder in service (on voluntarily basis- not part of the initial type approval test program- but will help refining the threshold and provide visual pictures for inspection calibration).

Recommended impact test program:
• Manufacturer specifies expected threshold energy
• Demonstrate that the impact does not reduce the burst capacity of the cylinder.
• Acceptance criteria will be to achieve minimum burst capacity.
• Footprint photo of the impact area to be used as reference for visual evaluation of impacts in service.
• Make photo available in service manuals following the product in service

Recommended impact test set-up:
• Boundary conditions
o Testing on neck mounted cylinder (to be able to compare with Hypactor results where neck mounted condition was mostly used).
o Testing on empty cylinders is more conservative than on pressurized cylinders
o Testing on the cylindrical part will give the best reproducible test results.
• Impactor shape
o Test to be done with hemispherical impactor shape to obtain most challenging footprint and reproducible results
• Impactor parameters:
o Diameter 56 mm - steel
o Take precaution to avoid secondary impacts due to rebounds on the vessel
o Impacts generated by a pneumatic canon or from a drop tower create equal effects to the structural composite
o Record impact and rebound speeds (e.g. with high speed camera) may be helpful later inspection

4.c. Exchanges with ISO/TC58/SC3/WG24

AL and HEX participated in 4 meetings, 2 workshops and 3 regulatory ISO TC58 (SC4/WG15, SC3, SC3/WG32 and SC3/WG35) between February 2014 and June 2017. During those meetings, Hypactor was presented, feeding the experts with its outcomes and updates on impact energy.
Air Liquide and Hexagon participated to some ISO meetings on Modal Acoustic Emission (MAE) standard (TC58/SC4/WG15 DIS 16016). Courtesy of the WG15 conveyor, Institut de Soudure attended as an Acoustic Emission expert to one meeting of this WG, in February 2016. Different options to publication of the DIS as an ISO-standard were discussed. One is to release the DIS as a Technical Specification (TS), but the release of a TS will not help the countries that need this methodology the most. The WG ended up with a recommendation to ask for a second DIS ballot to build more confidence in the DIS before release as an ISO-standard.
The MAE database built in Hypactor is unique in Europe, as it was the first evaluation of this method on periodic inspection of type IV COPV. Results and comments were communicated to the WG15 through the project workshops (a webinar in November 2016, a workshop and webinar in June 2017) and also to AFNOR. A second presentation is planned during the next ISO TC58/SC3 (Cylinder Design) yearly plenary meeting in November 2017 in the USA.
Further, HEX is preparing a New Work Item (NWI) application on the topic, as reference document at ISO that should be valid for all relevant composite cylinder designs and applications. This work can be addressed on TC58 level, TC58/SC3 level or directly to TC58/SC3/WG24 Factors of safety for composite cylinders, and most likely in all cases in liaison with TC 58/SC4 Operational requirements for gas cylinders.
Potential Impact:
The main impacts of the project are:
• A review on industrial constraints and operational incidentology.
• A novel & extensive database capitalizing the experimental data from impact testing and damage characterisations.
• Validated equipments, protocols and monitoring techniques for impact testing on COPV up to their nominal working pressure and compatible with pneumatic burst
• Validated equipments, protocols and pass-Fail Criteria for non-destructive examination of type IV COPV subjected to impacts
• A first 3D FE model for quantitative prediction of mechanical behaviour of impacted composite vessels including fatigue.

The key results of HYPACTOR contributing to achieve these impacts are:
• The determination of the structural properties of composite material exposed to impacts:
• Structural properties versus impactor properties: diameter, mass and speed of impactor for a same level of energy; shape, angle and material of impactor
• Determination of the influence of the inner pressure on immediate failure
• Structural properties versus vessel thickness, shot location, boundary conditions
• Structural properties versus repeated impact on the same location
• Determination of mechanical impacts which produce loss of performance
• Structural properties versus sustained load, cycling and combination of both

• The modeling of damages created by mechanical impacts on COPV:
• Modeling of damage propagation for the main failure mechanisms
• Modeling of damage propagation by reducing elastic properties of the finite elements.

More than 200 impact tests have been performed during the project on 3 different vessel types, with 3 impactor shapes and 3 impactor masses. That lead to specific work on vessel test bench to control the test and be representative of industrial lay out. Available NDE techniques have been assessed and 2 pass-fail criteria (real time and post-analysis) have been validated on more than 80 tests mainly hydraulic but also pneumatic.

Specific instrumentation has been defined to be able to measure real incident and rebound speeds, using fast camera records. Impact tests performed on pressurized vessels ensured the importance of safety issues for such tests. In fact, within the test matrix, two impacts lead to pneumatic burst with very significant energy release on the testing field. Infrastructures, equipments and procedures used by CEA have permitted safe testing and complete monitoring data acquisition.
A special attention must be paid to rebound when using drop tower.

An AE based real-time shut down criteria was developed and validated during the project to allow immediate rejection of severely damaged cylinders. A post-processing pass/fail criteria was also calibrated and validated with respect to damage characterization and burst performance.

Thanks to all the impact tests performed, to the NDT characterization of the damages and to the residual performance assessment, a large database has been established giving access to the impact conditions, the damages characteristics and the residual performances of the impacted vessels within HYPACTOR.

Recommendations with respect to qualification testing and periodic inspection were drawn up, presented through international webinars, international conferences and within standardization committees.
List of Websites:
Coordinator: Fabien NONY, CEA
Tel: +33 2 47 34 49 07
E-mail: Fabien.nony@cea.fr
Project website address: http://www.hypactor.eu/