Combat utility prediction

Stanislas GABROVSEK,Ian COLWILL,Elias STIPIDIS

Vetronics Research Centre,University of Brighton,Lewes road,Brighton,UK

Combat utility prediction

Stanislas GABROVSEK*,Ian COLWILL,Elias STIPIDIS

Vetronics Research Centre,University of Brighton,Lewes road,Brighton,UK

The majority of allied casualties from recent conflicts were caused by blast wave and fragments perforation damage from Improvised Explosive Devices.Survivability to this type of threat is a critical factor to consider for land platform design.This paper proposes an original approach to platform survivability assessment using a combination ofAgent-Based(AB)simulation and FaultTreeAnalysis(FTA)to predict the consequences of IED fragment impacts on the platform operational status.As a demonstration,this approach is applied to the comparison of different platform architectures to gain insight into the optimisation of the platform component topology.

Mission;Survivability;Simulation;Agent-based;Fault-tree analysis;Platform architecture

1.Introduction

Since 2001,Improvised Explosive Devices(IEDs)have been responsible for over 50%of the coalition soldiers’deaths,and IEDs dangerousness continues to intensify[1].IEDs are made of explosive material(typically discarded artillery ammunition) connected to a triggering system.The explosion of such a weapon generates blast wave(primary effect),fragments(secondary effect)and heat that interact with critical components or crew to incapacitate a platform.The interaction may be direct or indirect as in the case of Behind Armour Debris(BADs)generated by impacting fragments.

Add-on solutions that offer increased protection from IED effects such as slat armour and anti-BAD liners exist,but further benefit can be achieved through optimisation of the platform architecture itself.Simulation tools are ideal for modelling and testing architecture topology improvements as part of the platform architecture design process.A wide range of vulnerability modelling and simulation tools are available,with approaches that each provides a specific level of insight. For example,war-gaming techniques provide exploitable insights regarding platform usage doctrine while Finite Element Analysis(FEA)provides detailed analysis of specific component resistance to perforation.

This paper proposes an original approach to platform damage assessment analysis that can be applied from the fleet down to the component level,with benefits in terms of scalability,modularity and reusability of the developed models.Among other IED effects,this paper focuses mainly on the fragments’impacts on structures but the application of the approach to blast damage and shockwave transmission is also under study. Platform crew is not particularly considered either,while the method could be easily extended to human occupants,by considering appropriate fatality energy levels.As a demonstration of the benefits of this approach a comparison of different future platform architectures from their survivability to IED fragments point of view is presented.

2.Background

An existing standard for description of system vulnerability analysis is theVulnerability/Lethality(V/L)Taxonomy1[2]represented in Fig.1.Level 1 describes the initial state of thesystem before the attack.Level 2 describes the status of the components after the attack,with regard to the damage criteria. Level 3 describes the platform remaining capability at the functions level while level 4 describes the platform mission remaining effectiveness.Using a different terminology,the“platform incapacitation process”described in Ref.[3]refers to the same stages.

Fig.1.The V/L taxonomy.

2.1.Threat/target interactions prediction

Three types of techniques are commonly used to predict the effects of a given threat on systems:

1)Knowledge-based methods:this approach of survivability assessment is based on human estimation of damage resulting from empirical experiments or“after-action reviews”of enemy contact.This knowledge can then be implemented in“survivability tables”for simulation wargames[4]or in a database for survivability assessment expert-systems[5].

2)Analytical methods:in this approach,the physical reality of the battlefield damage is described by mathematical formulas.Mathematical equations are essentially correlations with experimental data(e.g.THOR equations). Semi-empirical equations are simplified models of physical phenomena[6].Both empirical and semi-empirical approaches are used extensively in vulnerability and technical-operational studies[2].Analytical methods are usually coupled with target geometry ray-tracing analysis to determine which parts of the system have been hit by the threat[7].

3)Numerical methods:this approach uses subdivision to model macro level problems using numerous small domains (nodes)with individualsolutions (Finite Element Analysis).Then,numerical techniques are used to find approximate solutions to each node.FEA provides improved accuracy at the component level,but requires detailed information about the structure to be modelled and is computationally intensive[8].

2.2.Damage assessment criteria

As depicted in chapter 1,platform components(including crew)can be damaged by different types of weapon effects. This paper considers the damage due to Kinetic Energy(KE) projectiles such as bullets and fragments.Possible metrics to estimate the kill probability Pk of a component after damage are reviewed in Ref.[9](Table 1).As detailed information regarding component kill probability is often classified,there is little publically available data.Piecemeal information on criti-cal levels of deposited energy was found in Refs.[10](Table 2) and[11](Table 3).Some data about critical levels of impact energy per area unit were found in Ref.[12](Table 4).

Table 1 Possible metrics for kill criteria when a component is hit by KE projectiles of fragments[9].

Table 3 Critical levels of deposited energy for platforms[11].

2.3.Combat utility estimation

The ability of a vehicle to successfully perform its mission is called the mission survivability[13],or combat utility[2].It is based on the different capabilities of the platform and its crewwhich are typically categorised as mobility,firepower,C4ISR2Command, Control, Communications, Computers, Intelligence, Surveillance and Reconnaissance.and protection.The combat utility of the platform is reduced if any of these capabilities are lost or degraded due to damage to the vehicle or crew(platform susceptibility3The probability of being hit.and recoverability4The probability of restoring mission capability through reconfiguration or repair.aspects are not considered here).So doing,most of the combat utility prediction techniques are based on:

Table 4 Critical levels of energy per area unit[12].

Fig.2.Examples of deactivation,RBD and FT diagrams of a DC power supply.

1)A list of individual equipment or subsystems critical to the operation of the platform capabilities(Standard Damage Assessment List).

2)The results of the criticality analysis,as a series of logic diagrams that present the contribution of critical components to the different platform capabilities.

There are three major formalisms for these logic diagrams, which are illustrated in the simple example of Fig.2.

1)Deactivation diagrams present the operational relationship between critical components and each capability.As long as an unbroken path can be traced through the diagram,no platform capability has been lost[2].

2)Reliability Block Diagrams(RBD)represent the critical subsystems or components connected according to their function or reliability relationship.They are“mission success”oriented[14].

3)Fault Trees(FT)show which combinations of the components failures will result in a system failure.It is composed of a Top Event(TE)that represents the most undesired event and lower level logical“AND”and“OR”gates or Basic Events(BE)that define the combinations of components failures leading to the occurrence of the TE[15].

2.4.Existing work

While existing implementations of the Agent Based(AB) approach for modelling battlefield complexity are numerous, existing applications that assess vulnerability at the component level are rare.In Ref.[16],anAB approach is used to model the effect of an anti-tank projectile on an AFV.Provided results generally concur with experimental data,but unfortunately no information is given about component damage assessment and no extension to the platform and fleet level vulnerability level is described.A similar approach is used in Ref.[17]to model the BAD generated by a missile explosion close to a military aircraft fuselage.The focus is the relative accuracy of the BAD modelling,but without considering the damage they create on components.

3.Combat utility prediction methodology

In order to predict the impact of IED fragments on platform combat utility,we use a combination of two techniques(Fig.3):

1)Agent-Based modelling is used to simulate the threatplatform interactions,as agents are well adapted to model battlefield complexity,meaning that with a large number of elements in interaction.The“deposited energy criteria”(chapter 0)is used to estimate the damage(probabilities of kill)on individual components.

2)Probabilities of kill of individual components are used in the platform combat utility Fault-Tree to determine how the platform capabilities have been affected by the threat and the probability to perform the mission.

3.1.Agent-Based modelling of the threat–target interactions

This approach uses agents to model the physical representation of objects such as IEDs,fragments and platform components with a 3D space.In contrast to the existing attempts in thisdomain,this work aims to generalise the AB approach to every level of the battlefield(fleet down to component),with expected improvements in terms of:

Fig.3.Graphic illustration of the modelling approach.

1)Abstraction range,from the component survivability level up to the system of system survivability.

2)Arbitraryandappropriate level of fidelity from acommon approach.

3)Modularity,by favouring the development of ready-touse threats,components and platforms libraries.

Agent properties describe critical(in terms of the simulation) dimensions of its existence.For example,at a given instant a fragment agent is described by its position,mass and velocity.

Internally the agents implement appropriate empirical and semi-empirical methods to describe their behaviour and interaction with other agents.Four types of agents are used to represent the problem(Fig.4):

1)An IED agent models the threat and the characteristics of the fragments that are generated by the explosion.

2)Fragment agents model all the fragments.Some of them are going to impact the target.

3)Component agents represent the platform components, with geometric and material characteristics attached.

4)Collision agents are used to detect the position and the angle of incidences of fragment impacts on the platform components.

3.1.1.IED agent model

Upon detonation the IED generates a number of Fragment agents with initial parameters that depend on the IED characteristics.For fragmentation,Mott’s formula(1)generates the average fragment mass,considering the explosive and case parameters,leading to consider a certain number of fragments. Held’s formula(2)provides prediction of the distribution of fragments size and mass,while Gurney’s formula(3)predicts the initial velocity of fragments.

Fig.4.The way the 4 types of agents interact in the model.

n:Fragments number,beginning with the heaviest.

B:Empirical constant(≈10-2),function of

λ:Empirical constant(≈2/3)

M:IED body mass(kg)

3.1.2.Fragment agent model

Fragment agents emanate radially from the IED agent and follow straight trajectories(gravity can be neglected for lowmass/high-speed fragments).Fragment agent velocity is calculated from an initial velocity and transmission medium (typically air)using the drag force equation(4).

ρfragment shape,1.5 for cubic shape) :Air density(≈1.2 kg/m3)

l:Fragment distance to explosion(m)

Real experiments[18]have shown that initial velocity is slightly different according to the fragment initial position (Fig.5)and projection angle(Fig.6)because of the common cylindrical shape of the IED casing,which would not be the case for an ideal semi-spherical casing.But the consideration of individual fragment initial position makes the equations much more complex without significant benefits in terms of fidelity of the prediction.For simplification reasons,this phenomenon is not consideredinthe current research.So,the fragment velocity is considered as uniform inside the solid angle of influence of the IED(Fig.6).

In the event of collision with a Collision agent,the Fragment agent transmits some of its energy to the Collision agent impactedandadaptsitsvelocityandmassaccordingtotheresult ofthecollision computation,considering fourpossible situations described in Table 5 and two fragment velocity thresholds:

Fig.5.Variation of fragment initial velocity as a function of its initial position for a 105 mm shell IED[18]and simplification considered in the current research(blue plot).

1)The ballistic limit velocity,under which the projectile does not perforate the target,

2)The shattering limit velocity,over which the target shatters into multiple Behind Armour Debris.

The ballistic limit velocity is approximated using the Brown equation(5)from Ref.[2]while the shattering limit velocity (Table 6)is based on experimental results related in Refs.[20] and[21].

h:Target thickness(cm)

m:Fragment mass(grams)

θ:Angle of impact(rad)

γ:Empirical constant(0.327 for RHA steel,-0.361 for Aluminium 2024)

Eventhoughthesefoursimplecasesgloballyreflectthedifferent mechanismsobservedduringprojectilepenetration[2],theydonot consider more complex phenomena(e.g.projectile erosion).

As mentioned in chapter 3,deposited energy is the criterion used to estimate the damage to individual components.This iscalculated by considering the different distributions of the projectile kinetic energy detailed in Table 8.

Fig.6.Example of fragment velocity versus projection angle for 105 mm and 155 mm shell IEDs[19]and estimation considered in the current research(blue plot).

Table 5 The 4 impact situations considered.

Table 6 Shattering limit velocities according to different sources.

Table 7 Empirical constants of the target material[2].

In case of target material shattering(Fig.7),the number of generated BADs is estimated on the base of experimental data [22]approximated by the empirical formula(6),assuming that the BADs are all the same mass.

Table 8 Distribution of the projectile kinetic energy versus different impact situations.

Fig.7.Model of the BAD generation process.

Finally,BADs spray is characterized by a maximal angle calculated in equation(7).Individual BAD velocity is calculated with equation(8)with angular deviation following a Weibull distribution,according to Ref.[20](Fig.8).

C:Stress wave speed in target(m.s-1,4.61.103for steel)

Fig.8.Direction of the BAD spray.

θ:Angle of debris with the normal(rad)

3.1.3.Component agent model

As a compromise between complex computation and the targeted level of fidelity,components’geometries are bounded with Axis-Aligned Bounding Boxes(cuboids).For cylindrical and complex-shaped components like wheels or pipes,this approximation leads to slightly overestimate the probability of a hit,which is not the case for other platform components,most often cuboid.Components are also assumed to be hollow in order to support typical platform nested components configurations(Fig.9).Hardness and thickness parameters are attached to every component to calculate the fragment impact situation as developed in chapter 3.1.2).

3.1.4.Collision agent model

Collisions between fragments and components are usually detected using ray-tracing techniques[7,23].In order to maintain the agent modelling approach of encapsulating functionality and subsequent problem scalability,our approach uses Collision agents which are generated at the surface of component agents to detect impacts through efficient agent 3D proximity algorithms.This approach allows the collision detection resolution,and consequently the range of the combat utility prediction,to be controlled by the number and size of collision agents without changing the methodology(Fig.10).

Fig.9.Example of nested components configuration.

Fig.10.Different Collision agents(transparent boxes)according to the level of survivability analysis.

3.2.Fault-Tree analysis

We use the Fault-Tree diagram formalism presented in chapter 0 to model the system-under-study architecture(whatever level it is)and the way components contribute to every capability of the system.The Top Event(TE)of the Fault Tree diagram is the system mission failure,meaning the total loss of the platform combat utility.Basic Events(BA)are represented by the physical components of the system,using the same breakdown developed in the agent-based model of the system. So,there is direct correspondence between the Component agents in the AB model and the Basic Events used in the FTA, as illustrated in Fig.11.

Probabilities individual component kills generated by the AB model are used as failure rates for the Basic Events.During the simulation and fragments impacts,failure rates are updated and propagated through the system capabilities up to the top of the Fault-Tree.Note that only static Fault-Tree mechanisms have been implemented at this stage and so-doing sequential relationships among component failures or components recovery modes are not possible.Implementation of Dynamic Fault Tree description has been identified as a possible future development of the method.

4.Validation

While basic information regarding principles and technologies developed to improve system survivability can be found in open literature,defence system vulnerability information and particularly experimental results remain highly restricted, significantly complicating the validation of modelling and simulation approaches.The approach used here is to partly validate the models with piecemeal information available on components vulnerability,and to compare the results obtained at the component level with other available simulation results.

For individual component vulnerability assessment,various simulation runs of this model with varying projectile-threat configurations have been conducted and analysed in Ref.[24]. Fig.12 shows an example of such a configuration and results obtained(angular dispersion of BADs).

A set of platform configurations has been tested for vulnerability in Ref.[24],with positive results.Fig.13 is an example of results obtained when exposing a simple platform to IED fragments.Energy absorbed predictions match with the type of damage reported in after-action feedbacks[1]and other modelling methods[16].

Fig.11.Example of system breakdown into Component agents(left)and FTA basic events(right).Only the observation sub-system capability branch is developed here.

Fig.12.Example of individual components configuration tested in[24]and results obtained(angular dispersion of BADs)compared to the reality.

Validation is confirmed as the predicted energy accumulated by individual components was comparable to those observed in real world experiments,proving appropriate implementation of the equations.Further validation requires that the high sensitivity of the results to slight changes in the target parameters, especially the shatter limit velocity of which estimation is further investigated.

On the basis of these preliminary conclusions,further investigations on more complex platform configurations were conducted.

5.Implementation and results

The modelling approach is implemented in Anylogic software and is applied to the assessment of the influence of platform architecture modularity on mission survivability.

Initially,three different architecture concepts were defined, on the basis of an original approach of platform modularity (chapter 5.1).The FaultTreeAnalysis of these architectures led to preliminary conclusions about their respective intrinsic vulnerability(chapters 5.2.5,5.3.5 and 5.4.5).

Fig.13.Example of platform configuration tested in[24]and results obtained.Dashed lines represent the critical absorbed energy levels leading to light damages, from Tables 2 and 3.

The architecture options were modelled using the approach presented and were exposed to a HE105 M1 shell IED threat. Predictions of combat utility resulting from numerous runs were compared and analysed(chapter 5.5).

5.1.Platform modularity

A platform architecture intended for reconnaissance mission (no offensive capabilities)is described in terms of three axes:

1)The mechanical architecture,of which roles are:

-To insure the contact of the platform with the ground, providing suspension,power transmission,braking and steering capabilities.

-To provide the mechanical mountings for every platform component.

-To protect the platform inner components and crew from external aggression.

2)The energy(or power)architecture,of which roles are:

-To generate the necessary electric energy for powering the platform components.

-To distribute the electric energy in the platform.

-To transform the electric energy into the mechanical energy required for motion.

3)The C4I(or vetronic)architecture,of which roles are:

-To support the data communications between the UGV and its remote command and control station(e.g. HF radio transmitter).

-To acquire all the information required by the remote operation(e.g.driving camera)and by the mission(e.g.tactical awareness camera mounted on a turret).

-To process the information on-board to facilitate its transmission or its interpretation by the distant UGV operator(e.g.video compression software running on an embedded computer).

-To support the exchange of data between the platform components(e.g.CAN utility data bus).

A module is defined as a component of a larger system that operates within this system independently from the operations of the other components.Modularity is a set of properties that support that independence of operations[25].

We make the assumption that an architecture design is purely modular or purely monolithic,whereas in reality a more graduated degree of modularity is likely.

This leads to the definition of 23=8 possible designs described in Table 9.Three architectures were selected for further survivability investigations,as the most representative of current design options for military land platforms:

1)The “monolithic”platform isthemostcommon architecture in low-range land systems:functions are supported by unique and dedicated components that cannot reconfigure.Some components support multiple functions.

2)The“digitalised”platform represents the current trend in land systems architectures.Digitalnetworking of components and intelligent power management allow implementing redundancy mechanisms to improve the system availability.

Fig.14.Illustration of monolithic mechanical architecture with example of existing implementation(Amstaff UGV-Isra?l).

3)The“extreme modular”platform is an illustration of a possible future modular platform,made of identical modules that can re-arrange autonomously.

5.2.“Monolithic”platform concept

Most of the legacy low-range civil and military vehicles are based on this design concept.It is also a usual type of design for disposable UGVs as it is reliable and relatively not expensive to produce(Fig.14).

5.2.1.Mechanical architecture

The chassis is a solid case mostly made of rigidly assembled parts.Optional parts such as the observation turret can be temporarily mounted.In the implementation(Fig.14),the contact with ground is insured by a 2×3 wheel and rubber tyre configuration.It is assumed that a damaged wheel is ripped off the platform and does not hamper the remaining wheels. Change of direction is affected by skid-steering.Steering is therefore reliant upon two operational wheel trains.

5.2.2.Power architecture

Propulsion energy is provided by a single battery that supplies the motors via a dual Motor Controller Unit(MCU).This unit receives commands from the ECU via an I2C data link. Two mechanical transmissions transmit motor torque to the left and right side drive wheels.Another battery is used to supply the C4I equipment(Fig.15).

5.2.3.C4I architecture

A radio receiver unit receives control signals and in turn sends command frames to the ECU.The ECU checks the integrity of the frames and generates commands to be sent to the appropriate equipment through dedicated point-to-point links. Analogue images are acquired by the 2 cameras and compressed by the ECU before being sent to the radio.Additional sensor data(compass,GPS,battery charge)are formatted bythe ECU and transmitted to the radio for communication to the remote control station(Fig.16).

Fig.15.Illustration of the monolithic power architecture design.

Table 9 The same platform with various modularity considered.Architectures investigated are shaded in grey.

5.2.4.Damage assessment Agent-Based model

As explained in chapter 3.1.3,platform components agents geometries are modelled as cuboids.These cuboids are positioned in the 3D model according to their position in the platform as shown in Fig.17.Agent parameters are set according to the different materials and thicknesses of components cases. Components colours refer to the mechanical(grey),power(red) and C4I architectures(blue)they belong to.

5.2.5.Combat utility Fault-Tree model

Fig.16.Illustration of a monolithic C4I architecture design.

Fig.17.Geometry and position of Component agents in the damage assessment model of the monolithic platform.

The least desirable event(Top Event)is the platform mission withdrawal,which can result from mobility(propulsion/ steering),observation or communication function failures as shown in Fig.18.These intermediate events have been refined down to the component level as explained in chapter 3.2.

Once the fault tree diagram of the system has been defined, minimal cut sets can be used to understand the structural vulnerability of the system.Cut sets are defined as the unique combinations of component failures that can cause the topevent to occur.Specifically,a cut set is said to be a Minimal Cut Set(MCS)when any basic event is removed from the set,the remaining events are no longer a cut set.The minimal cut sets can be seen as“critical paths”leading to the mission failure. The order of the cut set is the length of the path that leads to the undesirable event.So,the order of the MCS reflects the vulnerability of the whole system.

The results of the calculation of the MCS for the“Monolithic”architecture are summarised in Table 10.An important observation is that 8 single different components failures(cut sets of order 1)lead to mission failure,which makes the monolithic architecture intrinsically vulnerable to any damage to its components.

5.3.“Digitalised”platform concept

The second design modelled is named“Digitalised Platform”.Many current AFVs in development are based on this architecture,and a lotofeffortis focussed on the standardisation of the mechanical,power and vetronics interfaces(NGVA,Victory[26])with expected benefits in terms of development,operation,maintenance and upgrade costs.

5.3.1.Mechanical architecture

The chassis is not fundamentally different from the monolithic mechanical architecture.The only difference is the con-sequence of the higher number of inner components that significantly increases the volume of the body,offering a larger apparent surface to fragments.

Table 10 Minimal cut sets for the monolithic architecture platform.

5.3.2.Power architecture

Two energy sources deliver electric energy required by the platform components.The energy is produced and stored on-board by micro-generators combined with batteries or fuel cells.Power redundancy is provided through dual power circuits,providing a dual redundant supply for all equipment. Power management for improved silent watch and intelligent power balance can be realised by a dedicated computer(Energy Management Unit)and monitoring capability in all electrical consumers.Wheels are direct-driven by individual electric motors which are controlled by individual control devices (Fig.19).

5.3.3.C4I architecture

The electronic architecture of the digitalised platform is organised around 2 communications buses.The utility bus is dedicated to the platform command and control while the multimedia bus is dedicated to video data communications.Possible technologies for the utility bus are CAN and MilCAN standards,while multimedia communications can be supported by Gigabit Ethernet technology(Fig.20).

Fig.18.Possible causes of the platform mission withdrawal(first level of the FT only).

Fig.19.Illustration of the digitalised power architecture design.

In a normal mode,control and command information is received from the HF radio and transmitted to ECU1 and ECU2 through the utility bus.ECU1 and ECU2 operate in a parallel redundancy mode.Command frames are decoded,integrity is checked and appropriate data frames are sent to the motor controllers to affect mobility according to command laws and sensors feedback.The 2 digital cameras transmit compressed video frames on the multimedia data bus to the ECUs and the VHF radio transmitter(wireless video link).

If the HF radio transmitter gets damaged,control and command frames can still be sent through theVHF radio transmitter.If the VHF radio transmitter is damaged,low-data-rate pictures can be sent after compression by the ECUs to the remote station via the HF radio transmitter.

5.3.4.Damage assessment Agent-Based model

The assumption is made that the motor controllers are mounted next to the motors they control.The space between each pair of motors is used to install the computer units and the batteries.The individual equipment size is the same as for the monolithic architecture(Fig.21).

5.3.5.Combat utility Fault-Tree model

The minimal cut sets(MCS)analysis yields the events described in Table 11.The minimal cut sets order has been set to 2 to limit the size of the table.The results provide the following insights:

1)There is only one MCS event of order 1(EMU damaged), making the EMU component very critical in the digitalised architecture.

2)The criticality of the mechanical components(no MCS≤2) has been transferred to the vetronic architecture.

3)Digital utility and multimedia networking support redundancies between vetronic components(MCS=2).

Table 11 Minimal cut sets(limited to order 2)for the digitalised architecture platform.

5.4.“Extreme modular”platform concept

This“extreme modular”platform architecture is made of several modules that each provide a set common critical capabilities.This means that they are able to re-arrange themselves depending on the mission to realize and according to their current operational status.Specialised sensors and actuators can be mounted on the modules through a generic interface. This is a futuristic architecture as the required technologies are not all available yet.Some partial implementations exist (Fig.22).

5.4.1.Mechanical architecture

The platform chassis is made of identical modules that mount together with quick link interfaces.These modules can be arranged before the mission according to different configurations,or they can replace each other dynamically in case of module failure.Links between modules can be rigid or flexible, acting as articulations or dampers.

5.4.2.Power architecture

Each module is able to provide its own energy as well as energy to adjacent modules if required regardless of module organisation.Power transmission between the modules is insured by dedicated power plugs of different sides of the module.

5.4.3.C4I architecture

All modules communicate via redundant data networks. Connections with sensors and actuators mounted on modules are insured by multiple pins connectors.In case of connector damage,information and power are transmitted by remaining operational sockets(Fig.23).

Fig.20.Illustration of a digitalised C4I architecture design.

Fig.21.Geometry and position of Component agents in the damage assessment model of the digitalised platform.

5.4.4.Damage assessment Agent-Based model

To be able to compare the survivability of the three architectures,equipment size must remain the same,as well as the platform dimensions.Battery size is divided by the number of modules to still have the same UGV range.A drawback of the modularity is the multiplication of components and a particular effort has to be put on the integration of these components into the modules as shown in Fig.24.

5.4.5.Combat utility Fault-Tree model

Fig.22.Illustration of“extreme modular”architecture and example of existing implementation(Roburoc UGV from Robosoft,France).

Fig.23.Illustration of the“extreme modular”C4I and power architecture design.

Fig.24.Geometry and position of Component agents in the damage assessment model of the“extreme modular”platform.

In the“extreme modular”architecture fault tree decomposition,we assume that the mechanical modules have been configured before the mission and they cannot reconfigure in response to damage.This ignores a substantial expected benefit from this“extreme”modular architecture and highlights a limitation of the static fault-tree analysis approach that does not support sequential relationships among component failures. Consequently,some additional assumptions had to be introduced to simulate the re-configurability of the architecture:

1)Two batteries are enough to supply the rest of the architecture in energy in recovery mode,so that the mission can still be completed,

2)A minimum of three wheels are necessary to preserve mobility.We make the assumption that the modules can reconfigure if the undamaged wheels are all on one side.

3)Two operational modules are enough to move and steer the rest of the platform.

4)The observation camera is mounted on the top of one of the modules while the driving video is always provided by the front module camera.

5)One ECU and one radio can support the data processing and communications for the whole platform in a recovery mode.

Analysis revealed no MCS of order less than four(Table 12). The modularity and reconfiguration capabilities of the platform dictate that a minimum of four components(wheels)need to be simultaneously damaged to foil the mission.

5.5.Combat utility prediction

The FTA of the three platform designs provides a means to compare the intrinsic vulnerability of the architectures.As expected,the“extreme modular”architecture is more resilient to individual internal components failures.

Table 12 Total number of MCS per order for the“extreme modular”architecture.

Fig.25.Top view of the target-threat(“extreme modular”)configuration att=to+2.6 ms.

Considering external threats, the MCS calculated from FTA is not a sufficient reflection of the platform survivability,as FTA does not consider the platform design nor the way the platform components interact together to reinforce(BADs generation)or to mitigate(shield effect)the threat.

In order to get a more valid estimation of their relative vulnerability to IEDs,we applied the combat utility assessment approach described in chapter 3 to the three platform concepts described in chapters 5.2,5.3 and 5.4.They were exposed to the same representative IED threat(HE 105 M1 IED)in the configuration of Fig.25.Fig.26 is a 3D view of the IED fragments impacting the platform.

For each platform concept,a set of 50 simulation runs (Monte-Carlo analysis)with the same target and threat configuration and fragment Weibullangulardistribution resulted in the individual components kill probabilities presented in Figs.27–29.Error brackets correspond to the standard deviation for each probability of kill.

We can observe that:

1)The most impacted parts are the left wheels and the motors driver modules(～50%of kill probability),as they are the most exposed to fragments.

2)Predictions of the wheels kill probabilities do not vary in the three platform concepts,as the same components are used.

The real-time injection of these individual probabilities of kill in every architecture fault-tree(see Fig.30)provides additional results regarding the robustness of the architectures to individual components failures.

Fig.31 synthesizes the capabilities and combat utility predictions for the three concepts under study.

It can be observed that:

1)The“extreme modular”architecture is three times less vulnerable to the IED fragments than the“monolithic”architecture in terms of damage causing mission failure, highlighting the potential of modular architecture design for survivability.

2)The mobility capability is the most sensitive to the IED fragments for all architecture concepts,indicating thatparticular effortshould be placed on the design of this function,in this particular IED-threat configuration

Fig.26.Example of 3D view of the IED fragments impacts on the“monolithic”architecture at t=to+11.5 ms.Penetrating collisions are coloured in red,BAD impacts are coloured in orange.

Fig.27.Probabilities of the“monolithic”architecture components for being killed by IED fragments.

6.Conclusion and future work

IEDs will remain one of the most lethal threats to future land engagements while existing defensive aids are not sufficient enough to fully protect the vehicle,crew and equipment.Meanwhile,the optimisation of the internal platform design remains a critical factor in limiting the impact of IEDs effects on the platform combat utility.

The modelling and simulation approach described in this paper first aims to predict the damage caused by primary and secondary fragments impacts on internal platform components.

ThisAgent-Based method has been validated on elementary plates by comparison with existing experimental results or other simulation techniques predictions.

In a second step and as a demonstration,the approach was applied to the comparison of combat utility of three representative types of platform designs,by injecting component damage into the fault-tree diagrams of the architectures.

Fig.28.Probabilities of the“digitalized”architecture components for being killed by IED fragments.

Fig.29.Probabilities of the“extreme modular”architecture components for being killed by IED fragments.

Fig.30.Animated Fault-Tree diagram for the“monolithic”architecture focused on probabilities of right and left propulsion failures.Other functions are masked for clarity.

Fig.31.Prediction of capability and combat utility failure probabilities for the 3 architectures tested.

In this case,results obtained show the benefit of modularity (threetimesmoresurvivablecomparedtothesamplemonolithic architecture)without considering dynamic reconfiguration capabilities that are expected to further increase survivability.

This application also demonstrates the scalability,modularity and reusability of the developed approaches to vulnerability assessment.This approach enables the rapid generation of quantitative results describing platform combat utility.

Future work will consider the thorough analysis of the vulnerability of modular platform,the modelling of IED blast effect and shockwave damage due to components mechanical interfaces as well as platform reconfiguration capabilities through Dynamic Fault-Tree implementation in the modelling environment.

[1]Centre de Doctrine d’Emploi des Forces.La lutte contre les engins explosifs improvisés.Les cahiers du retex 2010.

[2]Deitz PH,Reed HL,Klopcic JT,Walbert JN.Fundamentals of ground combat system ballistic vulnerability/lethality.Progress in astronautics and aeronautics.Arlington:American Institute of Aeronautics and Astronautics;2009.

[3]Gavril F.A synthesis of methods for simulating vulnerability of armored vehicles.Naval Res Logistics 1995;42:247–65.

[4]Kennedy E.A short report on the manual wargaming of three UGV concepts issued from previous FGMC studies,UK MOD/DSTL, Technical Report 2009.

[5]Helfman RA,Saccenti JC,Kinsler RE,Suckling JR.An expert system for predicting component kill probabilities.Aberdeen(MD):1985.

[6]Ackland K,Buckland M,Thorn V,Van der Schaaf P.A review of battle damageprediction and vulnerability reduction methods,DefenceScience and Technology Organisation,Victoria,Australia,Public report DSTO-GD-0620;2010.

[7]Hou Y,Wang K,Jin T.A simulation system of ship vulnerability to fragment damage.J Inf Comput Sci 2014;11(12):4415–25.

[8]MannJ,FisherD,KrausM,LowndesE,YorkA.Ananalysisofengagement algorithms for real-time weapons effects.JDMS 2006;3(3):189–201.

[9]Hartmann M.Component kill criteria,Defence&Security,Systems and technology,FOI,Swedish Defence Research Agency,SE-164 90 Stockholm,FOI-R-2829-SE;2009.

[10]Singh VP,Singh Y.Generalised model for aircraft vulnerability by different weapon systems.Def Sci J 2000;50:12–23.

[11]Hall J.In:United States Naval Academy,editor.Principles of naval weapons systems.Dubuque(IA):Kendall/Hunt;2000.

[12]Colwill I,Gabrovsek S,Webber T,Watts I,Duncan A,Diskett D,et al. Super-hardened unmanned ground vehicle,Vetronics Research Centre, Brighton,UK,Technical Report RT/COM/4/5008;2012.

[13]Schofield J.Survivability assessment from concept to operational support, Qinetiq,Fife,Technical report Qinetiq/09/00063;2009.

[14]Distefano S,Puliafito A.Dynamic reliability block diagramsVS Dynamic fault trees.In:Reliability and maintainability symposium,2007.RAMS‘07.Annual.Orlando:2007.

[15]Lee WS,Grosh DL,Tillman FA,Lie CH.Fault tree analysis,methods, and applications–a review.IEEE Trans Reliab 1985;34(3):194–202.

[16]Chen Y,Xiao S,Song H,Zhang Y.Agent based modelling of the equipment battle damage.J Comput 2013;8(2):334–40.

[17]Bova MJ,Ciarallo FW,Hill RR.Development of an agent-based model for the secondary threat resulting from a ballistic impact event.J Simulat 2015;10(1):24–35.

[18]Zecevic B,Terzic J,Catovic A,Kadic S.Characterization of distribution parameters of fragment mass and number for conventional projectiles. New Trends in Research of Energetic Materials;December 2011.

[19]Hazell PJ.Armour materials,theory and design.New York(NY):CRC Press;2015.

[20]Verolme JL.Behind-armour debris modelling for high-velocity fragment impact(part2),TNO,PML 1997;1997.

[21]Rosenberg Z,Dekel E.In:Springer,editor.Terminal ballistics.Haifa, Israel:Springer;2012.

[22]Baillargeon Y,Dinovitzer A.Modelling of behind-armour debris with the DeCaM software package,Weapons Effects Section–DRDC–Valcartier, Courcelette,Canada,Technical Report 5070C-2.FR;2004.

[23]Roybal LG,Jeffers RF,McGillivary KE,Paul TD,Jacobson R.Modelling and simulating blast effects on electric substations.In:IEEE international conference on technologies for homeland security.Boston(MA):2009.p. 351–7.

[24]Gabrovsek S,Colwill I,Stipidis E.Agents-Based Simulation of improvised explosive device fragment damage on individual components. JDMS 2016;13(4):399–413.

[25]Nelson EM.Open architectures technical principles and guidelines V1.5.8,IBM Federal CTO Office,Technical Report 2008.

[26]Hee YS,Phoon KK,Ang JY.Interoperable open architecture for land platforms.DSTA 2013.

Received 26 June 2016;revised 23 September 2016;accepted 10 October 2016 Available online 24 October 2016

Peer review under responsibility of China Ordnance Society.

*Corresponding author.Tel.:+33687146672.

E-mail address:s.gabrovsek@vetronics.org(S.GABROVSEK).

1There is a close mapping between the V/L taxonomy levels and the more complex Missions and Means Framework(MMF)extensively used in military operations research analysis.

http://dx.doi.org/10.1016/j.dt.2016.10.002

2214-9147/?2016 The Authors.Production and hosting by Elsevier B.V.on behalf of China Ordnance Society.This is an open access article under the CC BY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

?2016 The Authors.Production and hosting by Elsevier B.V.on behalf of China Ordnance Society.This is an open access article under the CC BY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).