Adaptive Impact Absorption

The approach proposed in this paper, in contrast to the standard passive solution, focuses on active adaptation of energy absorbing structures (equipped with sensor system detecting, identifying the impact in advance, and controllable semi-active dissipaters) with high ability of adaptation to extreme overloading. The paper presents a concept of multi-folding structures providing the additional value of energy dissipation, due to the synergy of repetitive use of active elements (equipped with so-called structural fuses), according to pre-designed optimal distribution of yield stress levels, triggering the desired sequence of local collapses. The second example, an adaptive wind turbine, demonstrates the idea of controlling the value of peak dynamic force that is transferred to the support by means of a pneumatic system with a controllable on/off piezo-valve. An alternative for developing new blade materials is to introduce a semi-actively controlled connection between blade and hub with controllable characteristics. Finally, the paper discusses the possibility of real-time, dynamic mass identification, which is an important part of the process of load identification. Numerical simulations and experimental results are presented.


Introduction
Motivation for the undertaken research is in response to the requirement for high impact energy absorption in areas such as: structures exposed to risk of extreme blast and light; thin-walled tanks with high impact protection; vehicles with high crashworthiness; protective barriers.Typically, the suggested solutions focus on the design of passive energy absorbing systems.These systems are frequently based on the aluminium and/or steel honeycomb packages characterised by a high ratio of specific energy absorption.However high the energy absorption capacity of such elements they still remain highly redundant structural members that do not carry any load in an actual operation of a given structure.In addition, passive energy absorbers are designed to work effectively in pre-defined impact scenarios.For example, the frontal honeycomb cushions are very effective during a symmetric axial crash of colliding cars but are completely useless in other types of crash loading.Consequently, distinct and sometimes completely independent systems must be developed for specific collision scenarios.
In contrast to the standard passive systems (see advanced methodologies of computer simulation and design for improved crash-worthiness, References [1]- [6]) the proposed approach focuses on active adaptation of energy absorbing structures (equipped with sensor systems detecting and identifying impact in advance and controllable semi-active dissipaters, so called structural fuses) with a high ability to adapt to extreme overloading [7].The following characteristic cases of adaptive impact absorption (AIA) problems can be identified: -the case of a protective structure (for example, a protective barrier) sustaining impact overload with minimised maximum acceleration (or force) -the case of an adaptive structure sustaining impact overload with minimum global measure of plastic-like distortions generated in adaptive dissipaters during the impact scenario The first case corresponds to smoothing down the impact effect (such as for reduction in fatigue effect, as in adaptive landing gear [13]) and coming from service loads, while the second one corresponds to the impact absorption (for example, from critical environmental loads) with minimal cost of induced distortions.In both cases a semi-active or fully active solution can be applied depending on constant or time-dependent modifications realised via controllable dissipative devices.Note, these devices are dissipaters (with no need for an significant power supply) rather then actuators.Feasible, dissipative devices (structural fuses) under consideration in the application discussed below can be based on the following technologies: • adaptive dissipaters based on MR fluids • adaptive dissipaters based on (hydraulic or pneumatic) piezo-valves • adaptive dissipaters based on SMA alloys The present paper demonstrates progress in the AIA research field obtained recently in our research group and is based on the following previously published conference contributions, References [8]- [11].The monograph, Reference [12], under preparations, will soon present more detailed discussion of the problems under consideration.

The concept of the adaptive multi-folding microstructure
The objective of this section is to propose a new concept of adaptive microstructures with high energy absorbing characteristics.Let us discuss the truss-like micro-structure (similar to a honeycomb layout) shown in Figure 1 equipped with specially designed micro-fuses, for example, Figure 2a, with stacked thin washers made of shape memory alloys (SMA) as controllable stickers, Figure 2b.The microstructural response to external loading strongly depends on the yield stress levels applied to micro-fuses and these levels can be controlled by activating proper numbers of SMA micro-washers in each sticker.
The crucial point for obtaining additional energy dissipation value (due to the synergy of the repetitive use of dissipaters) is to pre-design the optimal distribution of yield stress levels in all stickers, triggering the desired sequence of local collapses.Let us call the discussed adaptive microstructure the adaptive multifolding microstructure (MFM).Assuming idealised truss structure model (Figure 3a) is composed of perfect elastoplastic members with the yield stress levels shown (obtained by properly activated fuses), the sequence of collapse is shown in Figs 3b, 3c and 3d, respectively.When compressing forces in all members of the idealised truss-like structure shown in Figure 3a are the same, all members with the stress limit level σ 2 start to yield first, converting the structure into the configuration shown in the Figure 3b.Then the yield stress level for elements marked σ 3 is lower than for tripled elements (with a stress limit of 2σ 1 -δ) and the consequent structural configuration is that shown in Figure 3c.Following the same procedure the next configurations, that is, Figure 3d and Figure 3e can be reached.
The piece-wise linear constitutive model of MFM (applicable in computational simulations), shown in Figure 4, can be proposed.Cyclically loaded and unloaded adaptive members will have their characteristics with high hysteresis (Figure 4b).Additionally, fictitious members (dotted lines in Figure 4a) with piece-wise linear locking properties (Figure 4c) are proposed to model contact problems in the loading scenario.The numerical model for simulation of MFM performance will be required to take into account both physical and geometrical non-linearities.Note that the resultant characteristic (Figure 3f) of the MFM model is not unique.The final shape of this curve can be modified according to our requirements by changing the yield stress level distribution.

Optimal control
To provide optimal energy absorption it is necessary for the microstructure to perform a process of adaptation consisting of three stages: The dynamic load level can be evaluated prior to an impact by measuring velocity and estimating the mass of a colliding body.Alternatively its value might be identified at the beginning of impact by structural loading sensors.In order to dissipate the kinetic energy in an optimal way, one has to apply correct strategy to active elements of the microstructure.Two strategies of the structural control are formulated: In the first case pre-selected triggering stress levels i σ in structural elements remain unchanged during an impact, in the second, the theoretical possibility of realtime changes in control parameters is assumed.Although full real-time control does not seem to be feasible in a real design, one can expect at least a few stress changes during the impact time might be applied.
The problem of optimal control can be formulated as follows: Semi-active control -for a given impact load minimise the difference between acceleration values in selected points of the structure and desired response function ( ) Figure 5: The algorithm searching for maximum load level, safe for adaptive MFM, semi-active control Real-time control -for a given impact load, for every time step minimise the difference between acceleration values in selected points of the structure and desired response function ( ) subject to the following constraints Dynamic load evolutionary algorithms / gradient-based methods

Optimal folding mode
where σ denotes the plastic-like yield stress level and q is the maximum crush distance.Satisfactory solution of the above problem exists if the external load intensity is not higher then the maximum safe load level.Beyond this, a limit control strategy with the highest possible capability for energy dissipation must be applied. .Maximum crush distance must be less than 4H.Contact elements provide correct model behaviour.The structure was subjected to a dynamic load (concentrated mass with initial velocity 0 q ) simulating an impact.Kinetic energy of the load was increased linearly by changing mass value from 100kg up to 240kg with constant velocity 15m/s.The response of the modelled structure strongly depends on the relationship between parameters 1 σ and 2 σ .Different, effective folding sequences are depicted in Figures 8 and 9.In the case of first folding mode, which provides a very smooth response 1 σ is lower than 2 σ .The second one, with 1 σ > 2 σ , gives maximum energy absorption capability.The evolution of stress, strain and plastic distortion for selected element "B", corresponding to the second folding mode 2 is shown in Fig 11.Evolution of elastic and plastic energy is presented in Figure 10.In the numerical example only a semi-active control strategy was considered.Optimal distribution of control stresses in the structure with corresponding extreme acceleration values for the loaded node for different mass values are presented in figure 12.Note that at a kinetic energy value corresponding to mass of 130kg it is necessary to change folding mode form mode 1, which reached its limit load to mode 2. The results obtained are compared to the so-called "passive absorber" (the structure in folding mode 2 with the highest possible control stresses, providing extreme energy absorption).Applied semi-active control gives a very stable response for different loading values combined with the possibility of a high reduction in acceleration level.

Adaptive tower
In order to minimise consequences of the dynamic load, a process of structural adaptation to an impact should be carried out.The process consists of the following, subsequent three stages:

Impact detection
Impact detection is provided by a set of sensors that respond in advance to a danger of collision such as radar and ultrasonic devices, or are embedded into the structure within a small passive crush zone (for example, piezo-sensors).Estimation of the impact energy is then based on an initial deformation of the passive zone.

Structural adaptation
The signal from the system of sensors must be directed to a controller unit, which selects an optimal distribution of yield forces P in active zones containing elements equipped with structural fuses.Figure 2 presents the concept of the fuse, consisting of a stack of thin, shape memory (SMA) alloy washers.The yield force in the active element depends on a friction force generated in the fuse by activating different number of washers.The initial value P 3 (without any activation) can be gradually decreased to P 2 and P 3 respectively, when one or two washers change their original shape.Due to the characteristics of the structural fuse, the active element can be modelled as elastic-perfectly plastic with a controllable yield stress value.

Self-repair (structural recovery)
Increasing requirements for structural durability and low-level operating costs create a need for new, smart solutions.The results of an extreme dynamic load may very often be fatal for a structure.In the case of structures equipped with active elements it is possible to remove residual distortions using low-level vibrations induced by an external or embedded shaker.
The following numerical example illustrates the concept of adaptive impact absorption.A 30m high tower, equipped with an active energy absorber, is considered as an example and depicted in Figure 16.The structure is subjected to the impact of a mass of 3000kg with initial velocity of 8m/s.All structural members have a uniform cross-section area and elastic modulus.
The yield stress level in active elements is adjusted according to the value of the kinetic energy of the impact, while in all passive elements it is equal to 6e8Pa.The maximum allowable strain in controlled elements is constrained to 50% of their initial length.The objective function of the optimisation problem is to minimise the horizontal acceleration of the controlled node at the top of the tower.Results for the problem for two different thresholds of yield stresses in active elements are presented in Figure 17.For the highest threshold of 6e8Pa the response of the structure is elastic while thresholds of 2.6e7Pa (the optimal solution) provide localised permanent plastic deformation.It is can clearly be seen that adaptive strategy provides a very significant reduction in the acceleration level and that control parameters in active elements should be adjusted according to the severity of the impact.
Two strategies of semi-active and active control might be considered.In the first strategy, yield stresses in structural members located in active zones remain unchanged during an impact.In the second one, a possibility of real-time changes in control parameters is assumed.The paper focuses only on the semi-active approach, which provides a good balance between expected results and complexity of control strategy.
A position of active elements follows from an assumption that only selected parts of the considered structure would be exposed to a danger of extreme, dynamic load.A number of active elements and their structural properties should be chosen as a result of the separate optimisation task.
The problem of the control may be formulated as follows: for a given structure and a given set of active elements i E A ∈ , find optimal distribution of control yield stresses p i σ , minimising objective function f defined by the impact index I 2 (providing information about average acceleration level at monitored degrees of freedom (DOF) cntrl i q ): ( ) ( ) with the following constraints, imposed on control stresses and displacements in active elements: Two additional measures of structural dynamic response, describing overall acceleration level in the structure and maximum acceleration values at specified DOF, may be introduced: ( ) where T is time of analysis, N is number of DOF, Ncntr is number of monitored DOF.
Figure 17: Horizontal acceleration of the monitored node, for the passive and the optimal plastic solution After optimal impact absorption, permanent deformation is localised in selected active elements.Assume that residual strain in an active element is equal to R i ε .
Low-level vibrations induced by a shaker (external or embedded into the structure) can generate strains ( ) i t ε , which can be used to recover initial length.When residual and actual strains have opposite signs the structural fuse opens and releases accumulated distortions.In the case of equal signs, the fuse remains closed: In order to ensure stability only one active zone should be recovered at a time.

Motivations and the problem formulation
One of major technological limitations of up-scaling of modern wind turbines is the load bearing capacity of blade materials [14].Connection between blade and hub has to withstand very large bending moments in particular during high winds.An alternative for developing new blade materials is to introduce a semi-actively controlled connection between blade and hub with controllable characteristics.This section demonstrates the idea of controlling the value of peak dynamic force that is transferred to the support by means of a pneumatic system with a controllable on/off piezo-valve.Pneumatic pistons, together with controllable valves, and pressure supply systems are used as smart actuators, typically as handling or positioning devices [15,16].Contrary to applications in the present approach, pressure is withdrawn from the cylinder by using a smart valve and the whole system works in a semi-active way without applying external energy.Piezo-materials are used in pneumatic systems either to allow for very fine positioning precision or as a piezo-actuator for micro-valve opening.Problems encountered for common pneumatic valves are long response times and, especially for low pressure, the need to use a pilot valve.In the present case the solution of a mechanically amplified piezo which drives the valve orifice means there is no need for pilot valve and the response time is of the magnitude of single milliseconds.In the first part of the section a numerical study is presented to demonstrate the idea and assess the effectiveness of solution.The experimental set-up is then presented.
The numerical model of the system consists of two major elements, that is, the slender, elastic part which can be viewed as a demonstrator 'blade' and a pneumatic piston which transfers the dynamic force to the support (Figure 19).The pneumatic spring, together with a controllable outlet, forms a 'smart bond' between the blade and the support.Further, it is assumed that the time it takes a high wind to develop is slow enough, so that the piston velocity does not exceed 1 m/s and the heating effects are negligible.
Figure 19: Numerical model features Also, the working gas pressure in the cylinder should not exceed 1 MPa.Given the above, the process within the cylinder was assumed to be an adiabatic transformation of an ideal gas, at the constant temperature of T = 21ºC, according to the equation: It follows that the piston behaves like a non-linear spring with a hyperbolic characteristic.From the controllability standpoint it is preferable to use a piston with a big diameter due to the fact that given force could be obtained at smaller stroke and thus possibly a long stroke could be used during the active phase of the process.
A φ 63 mm piston with the stroke of 50 mm was modelled in the simulation.
The simulation was performed with Abaqus/Standard code.The piston and remaining part of the pneumatic system were modelled with fluid cavity elements.Before adding control, results were compared with those obtained with non-linear Spring elements (Figure 20).

Controllable piezo-valve
Airflow through the controllable valve was modelled with the equation: Pressure difference p Δ between the pneumatic cylinder and the environment is proportional to the rate of mass flow ) (t q .The flow resistance ) (c C V depends on the control variable c: Flow resistance for the open valve was chosen so that it corresponds to the flow rate of 30 l/min at the pressure difference of 0.1 MPa.

Control strategy
The goal for the control procedure is to keep the value of piston force F P below some critical level F cr .This is achieved by means of opening the valve whenever cr P F F ≥ and closing it if F P drops below the critical value.At some point during the high wind the decay sign of F P is changed and under pressure starts to build up in the cylinder.At this point the valve is opened again in order to allow the air to flow back into the cylinder.The control procedure described above can be summarised in the diagram shown on Figure 21.

Numerical results
The use of piston strokes is depicted in Figure 22.The maximum stroke is 43.5 mm, which means 87% of the allowable strokes are used.The active phase of the process begins at the 22mm stroke approximately, that is, at about 44% of allowable strokes.the valve is instantaneous.This is a reliable assumption since the response time of the piezo-valve is of the order 1 ms.The total mass flow is calculated as: where mf(t) is the mass flow rate, and is depicted in Figure 25.One can observe that final value is not equal to zero, which means that the amount of air in the cylinder at the end of simulation is not equal to the initial amount.Since total mass flow asymptotically approaches zero, the time required for the system to return to its initial state would be too long.

Experimental set-up
The system shown in Figure 26 has been manufactured in order to verify numerical results.An air outlet from the cylinder was connected to the on/off valve (Figure 27).Control of the valve opening was obtained by means of Cedrat Technologies piezo-actuator.Its integral part is a mechanical amplifier of displacements which enables us to obtain displacement of 0.2 mm.The response time of such a piezoactuated valve is less than 1ms.Initial tests indicated that the blocking force of the actuator is enough to withstand forces of magnitude considered in the numerical simulations.
Figure 26: Cedrat Technologies piezo-actuator as an element of the controllable valve In order to overcome the problem of a rate of flow back into the cylinder that is too slow (Figure 27) the system was additionally equipped with an air reservoir connected to the cylinder outlet.During the rising phase of the load, air from cylinder is controllably pressed to the accumulator and returned into the cylinder during the decay phase of loading.This forms a closed system which can work in sequence.
The feedback was established between the pressure transducer monitoring, the pressure/force in the cylinder and the piezo-actuator opening/closing the orifice.Setup was integrated using the Bruel & Kjaer PULSE system.An optimal strategy of structural adaptation depends on the impact load, which has to be identified in real-time.In some cases (for example, when the dropping mass is known in advance) the impact velocity measurement is sufficient to perform the desired structural adaptation.However, in many cases (for example, an unknown hitting object) on-line determination of the impact velocity as well as the mass of the hitting object is the real challenge.Having both of these parameters determined, the further optimal structural response can be performed via an open-loop control process.
In order to perform the feasibility study of a real-time dynamic mass identification technique a drop-testing stand was used.The main objective for the study was determination of the time delays in the mass identification procedure, when we assumed that the process was being performed after the beginning of the impact phenomena.The detected mass value can be used for the impact energy determination if the velocity of the falling mass was also monitored.Having the mass and velocity of the monitored object determined before the impact force reached its maximum value, it would be reasonable to apply a methodology to feed up an adaptive impact absorption system operating in real-time.The objective of this research is to give an answer to the following question: How fast are we able to perform a mass identification after the beginning of an impact phenomenon?

Analytical model
The considered concept of the falling mass identification was analysed analytically on a simple 2 DoF model (see Figure 28a).The parameters of the model were taken close to the values present in the experimental model.The analysis, via MATLAB software, allowed us to come to the conclusion that it is feasible to identify the mass on the basis of its acceleration and impact force.The numerical experiment showed that the mass value is readable 1-2ms after impact instant (Fig. 28b) in case of concluding on the basis of mass acceleration and taking into account inertia of the system.

Experimental verification -drop test stand
The main parts of the drop-test stand (see Fig. 29 b) were a magneto-rheological (MR) damper (1) mounted in a vertical position, a frame and a carriage.The stand was fixed to a foundation plate in order to reduce measurement noise.The lift mechanism enabled to conduct the drop tests up to 700 mm height.The mass (2) was guided by a rail system embedded in the frame, to ensure the stability of the vertical movement.The impact of the dropped carriage took place via a rubber bumper (3) located on the impact surface.During the tests the following signals were acquired (Fig. 29b): a force signal from a sensor fixed to a piston rod of the MR damper (4) in order to measure the full impact history; a signal from an optical switch (5) acting as a trigger and enabling determination of the horizontal speed of the carriage just before the impact.The test procedure also covered the measuring of acceleration at two points: deceleration of the falling mass (6a) and acceleration (6b) of the piston rod of the MR damper (7).

Experimental mass identification
Impact forces acquired during the experiments with several masses had characters showed in Fig 30a .The critical maximum value of the force was recorded after 9-15ms from the beginning of the phenomena dependent on the dropped mass.The process of mass identification was performed in two routines, one on the basis of deceleration of the falling body and the other on the basis of acceleration of the impacted body (piston rod).Accelerations measured in the two locations were characterised by curves presented in Fig 30b .In order to obtain a meaningful signal from the acceleration of the piston rod it was conditioned with a low pass filter (see Fig 30b).Results of the mass identification on the basis of the falling body deceleration were performed for a series of masses.The meaningful mass values were observable after 2-4ms from the impact instant (see Fig. 31a).Mass identification on the basis of the impacted body acceleration was performed on the basis of the filtered signal.The results of the analysis and comparison between both methods are depicted in Fig. 31b.Much better results of the mass identification were achieved for the case where deceleration of the falling body was used.The data enabled us to perform reliable dynamic mass identification at the beginning of the impact phenomena, much sooner than the maximum force value was observed.On the contrary, when the acceleration of the MR dumper piston was used the mass was much less reliably identified and the value was observable after 20ms from the impact instant (see Fig. 31b).It was caused by the fact that the mass identification was possible only when the falling body and piston rod moved jointly.

Conclusions
The following overall approach in design for adaptive impact absorption (AIA systems) can be proposed: • design topological pattern (allowing large deformations in location of structural fuses) and geometry of adaptive structure for variety of all expected extreme loadings • determine optimal yield stress level distribution (constant or time-dependent) and the corresponding dynamic responses for each expected extreme loading scenarios.• apply in real-time the pre-computed control strategy as the response for detected and identified (via an on-line sensor system) impact.The main challenging problems to be solved to develop the proposed methodology of design for AIA are the following: • to develop technologies for structural fuses with short response-time • to develop techniques of on-line impact-load identification • to develop control algorithms for structural adaptation • to apply integrated AIA systems to well-chosen demonstrative case studies.

Figure 6 : 3 Numerical demonstrations of structural adaptation to impact loads 3 . 1
Figure 6: The algorithm searching for maximum load level, safe for adaptive MFM, active control

Figure 16 :
Figure 16: (a) Model of an adaptive structure (b) displacements of the elastic solution (scaled x 1.5) (c) displacements of the optimal solution

Figure 18 :
Figure 18: Structural fuse in recovery mode

Figure 20 :
Figure 20: Fluid cavity and non-linear Spring elements comparison

Figure 23 :
Figure 23: Dynamic force in pneumatic piston

Figure 24 :
Figure 24: Mass flow rate mf through the valve (a) whole-time domain (b) time window 0.5 -1.0 section

Figure 25 :
Figure 25: Total mass flow through the valve

Figure 28 :
Figure 28: Numerical model (a) schema of the model and (b) results obtained

Figure 29 :
Figure 29: Drop test stand (a) general view and (b) main parts and sensors distribution

Figure 30 :
Figure 30: Measured signals for the drop test (a) force under the dissipater and (b) acceleration measured on the falling mass and top of MR damper.

Figure 31 :
Figure 31: Mass identification (a) accelerated measured on the falling mass and (b) accelerated measured on the top of MR damper