Defence Workforce Transition Simulation and Analysis

∗

Katie Mortimer, Cameron Pike and Vivian Nguyen

Defence Science and Technology Group, 506 Lorimer St, Fishermans Bend, VIC, 3207, Australia

Keywords:

Simulation and Modelling, Workforce Planning, Workforce Transition, Defence.

Abstract:

Simulation plays a key role in the workforce planning of defence workforces, with their strictly hierachical

nature, and complex interactions and interdependencies. In this paper, the Athena tool suite, containing Athena

Lite and Athena Pro, is introduced. The Athena tool suite is the ofﬁcial workforce planning tool of the

Australian Defence Force (ADF), and uses Discrete Event Simulation and hybrid Agent-Based Discrete Event

Simulation to simulate the highly complex ADF workforce. These are web-based, scalable tools that have been

speciﬁcally designed to be used by Defence workforce planners. Their use in the analysis of the Australian

Army Aviation’s planned transition from the Tiger Armed Reconnaissance Helicopter to the Apache helicopter

demonstrates their capability in identifying workforce shortfalls and risks, bottlenecks, and potential solutions

through optimisation.

1 INTRODUCTION

The Australian Defence Force (ADF) workforce is a

large, highly complex system with many constraints,

interdependencies and complex interactions among

people, resources, schools and units. The system

is closed and strictly hierachical, meaning external

personnel cannot be easily brought into the work-

force and the system relies on personnel progressing

through their career and achieving particular proﬁ-

ciencies and skills in order to ﬁll critical roles. Train-

ing and career progression of Defence personnel can

take many years, be very expensive and require large

amounts of resources (both platforms and people). On

top of this, changes in governance, capability require-

ments and operational platforms also regularly occur.

Robust workforce planning is therefore required to

ensure that the right amount of people at the right time

with the right skills are trained and ready to provide

ADF capability.

To assist in this planning Athena was created.

Athena consists of two simulation tools, Athena Lite

and Athena Pro. These are scalable, web-based

decision-support tools that provide the ADF with the

ability to simulate, analyse and forecast workforce

surplus or shortage against workforce requirements,

and determine recruitment, training and promotion re-

quirements to meet future demands.

∗

None of the values and charts used in this paper are

representative of ADF capability, and have been created in

order to demonstrate the use of Athena.

Athena Lite and Athena Pro are complementary

tools that simulate the ADF workforce to different ﬁ-

delities. Athena Lite is a discrete event simulation

(DES) engine, that is less data intensive, computation-

ally faster, and has speciﬁc optimisation algorithms

capable of optimising inﬂow and promotion numbers

for the workforce. Athena Pro, an Agent-Based Dis-

crete Event Simulation (AB-DES) engine, provides

a more detailed analysis on the complex structure of

the ADF workforce, and how requirements for train-

ing and career courses, and speciﬁc prerequisites and

proﬁciencies effect the workforce. Athena Lite and

Athena Pro are available on the Australian Defence

network, and are in active use by ADF workforce

planners from the Royal Australian Navy (RAN), the

Australian Army, and the Royal Australian Air Force

(RAAF).

The Australian Army Aviation (AAvn) are a crit-

ical capability in the Australian Army, and the wider

ADF. This capability contains three separate heli-

copters - the Tiger Armed Reconnaissance Helicopter

(ARH), the MRH90 helicopter, and the Chinook

CH47F helicopter - that provide speciﬁc roles. Each

of these helicopters requires many different highly

trained personnel in specialist areas, including engi-

neers, technicians, pilots, and aircrew. Training these

personnel can be an expensive, time-consuming pro-

cess that requires many resources. Competing needs

between services and between different helicopter

types within AAvn, as well as unforeseen circum-

stances such as particularly high and variable train-

Mortimer, K., Pike, C. and Nguyen, V.

Defence Workforce Transition Simulation and Analysis.

DOI: 10.5220/0011920400003396

In Proceedings of the 12th International Conference on Operations Research and Enterprise Systems (ICORES 2023), pages 249-258

ISBN: 978-989-758-627-9; ISSN: 2184-4372

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

249

ing failure rates, workforce separation or low recruit-

ment numbers, means it is necessary for pilot work-

force scheduling to be resilient enough to withstand

these variations. Added stress occurs when AAvn

undergo complex workforce strains as they transi-

tion their ARH capability from the Tiger ARH to the

Apache Guardian helicopters. This transition is oc-

curring over the period of several years from 2025

and requires retraining of the current Tiger-trained

personnel to the new Apache helicopter while, at the

same time, requiring Australia’s ARH capability to be

maintained. To ensure that the workforce transition

occurs in this time-frame, with as little disruption to

the capability as possible, a robust workforce transi-

tion plan is necessary.

Athena Lite and Athena Pro were used in a com-

plementary way in the analysis of the AAvn work-

force, and the ARH to Apache workforce transition.

The career pipelines of the AAvn operators were mod-

elled in both Athena Lite and Athena Pro. These

pipelines were simulated over 20 years, analysing the

health of the workforce, including personnel short-

ages, bottlenecks, and high-risk positions, with and

without the platform transition.

2 RELATED LITERATURE

Athena Lite and Athena Pro are speciﬁcally designed

for simulating Defence workforces. Defence work-

forces are closed systems with a hierachical structure.

They are reliant on personnel progressing through

their career, meeting particular milestones before be-

ing able to ﬁll certain positions.

Previously Markov chain models have been used

to model Defence workforces, as in

Skulj et al. (2008)

and Filinkov et al. (2011). However, an important

feature in Defence workforce planning is the ability

to run ‘what-if’ scenarios, testing different workforce

and training policies, recruitment, wastage and pro-

motion changes, and capability transitions. Markov

chain approaches are not easily adapted for these

analyses.

System Dynamics (SD) modelling has been used

for this purpose, where the workforce is represented

as a set of stocks and ﬂows. In Thomas et al. (1997) a

SD model is built to analyse the effect of policy deci-

sions on personnel strength in the United States Army.

An SD model was also used to model the pilot occu-

pation in the Royal Canadian Air Force, and deter-

mine the impact of increased production and reduced

budget on the occupation (S

eguin, 2015).

DES models a system as a sequence of discrete

events in time. DES models are very ﬂexible, and

can model a high level of detail. As such, discrete

event simulations have been previously used to model

Marine training (Davenport et al., 2007), the Royal

Canadian Navy (Henderson and Bryce, 2019), and in

various industries, such as in healthcare (Gunal and

Pidd, 2010), and call centres (Mathew and Nambiar,

2013).

Heath et al. (2011) compares different simulation

paradigms, including SD and DES. They argue that

while SD models have much fewer data requirements,

they do not provide the ﬂexibility and detail of a

DES. Heath et al. (2011) also compares combinations

of simulation paradigms, such as SD-DES, AB-DES,

and SD-AB, and argues that AB-DES is a good choice

when resources perform activities and human inter-

actions where individual behaviours effect how these

activities proceed. As such, AB-DES has been used

to assist in disaster planning and evacuation (Na and

Banerjee, 2014) and in modelling emergency medical

services (Anagnostou et al., 2013), as they are capa-

ble of incorporating the complex interactions between

agents. In the Defence context, AB-DES has previ-

ously been used by the authors (Nguyen et al., 2017)

to model the ADF aircrew supply problem.

Athena Lite and Athena Pro both employ DES.

However, to properly simulate the complexity in-

cluded in the higher ﬁdelity Athena Pro, an AB-DES

engine was used. This is necessary in Athena Pro,

as it models the inter-dependencies in the workforce,

where positions and training can be shared across lev-

els and careers, and to model complex personnel ca-

reers, where many possible paths may be available. A

key difference between Athena Lite and Pro and other

simulation tools available is that they are web-based

tools, speciﬁcally designed to allow Defence work-

force planners with little experience in modelling and

simulation to make use of them. They are scalable,

and capable of simulating the entire ADF. They have

intuitive user interfaces, and detailed results analy-

sis, allowing users to create, use and validate their

own workforce simulations, while still maintaining

the complexity of modelling and simulation required

for the Defence context.

3 ATHENA

Athena Lite and Athena Pro are two separate sim-

ulation engines that were used in the AAvn work-

force transition analysis. Both simulation engines

are capable of completing detailed modelling of De-

fence workforces, analysing individuals’ progression

throughout their career and detailing unit, posting,

and rank readiness. These simulation engines were

WAPAT 2023 - Special Session on Workforce Analytics - Practical Application and Theory

250

employed in a complementary manner to provide ex-

tensive what-if analysis of the workforce transition,

and to provide detailed modelling of the effects of

the complex training and prerequisite requirements

present the highly trained AAvn workforce.

In this section, the logic behind these engines will

be explained, as well as the web-based user inter-

face that allows Defence workforce planners with lit-

tle modelling experience to interact with these simu-

lation engines.

3.1 Simulation

Athena Lite and Pro are designed to be detailed

enough to accurately model the complexities of

careers in a Defence workforce, however generic

enough to be able to model many different operator

types to any required ﬁdelity. Both simulation en-

gines employ Monte-Carlo methods to simulate the

randomness is workforce wastage, promotions, and

allocations. Athena Lite is a DES, that simulates indi-

viduals as they progress through their career and com-

plete postings. Personnel attributes, such as current

posting, platform endorsement, level and current time

in posting and level, are used to determine personnel

eligibility to be promoted to the next level or ﬁll a

particular posting. In this case platform endorsement

refers to any platform speciﬁc training or proﬁcien-

cies that a posting may require. Athena Lite is used to

answer workforce questions such as:

• can the workforce requirement be met?

• what is the risk of not meeting workforce require-

ments?

• what are the optimal recruitment and promotion

targets in order to meet demand?

Athena Pro uses an AB-DES engine to simulate

the Defence workforce. Athena Pro is capable of sim-

ulating the workforce at a much higher ﬁdelity than

Athena Lite, modelling training and career courses,

complex prerequisite structures in the workforce, and

more detailed modelling of the workforce’s skills pro-

gression. This allows the user to understand not just

where the personnel shortages are, but exactly which

skills are missing from the workforce and trace it back

to particular bottlenecks. Athena Pro is used to an-

swer questions such as:

• which skills and proﬁciencies are missing from

the workforce?

• what are the bottlenecks slowing progression of

personnel through the workforce?

• what is the effect of dependencies and interactions

between different ranks and careers on the work-

force?

• how do course timings, session sizes and pass

rates effect the progression of personnel?

3.1.1 Athena Lite

Athena Lite creates a workforce structure, using

generic concepts: careers, levels, and units. In a

Defence workforce levels generally represent ranks,

while careers represent operator types. Units are de-

ﬁned as groupings of positions and may include op-

erational squadrons, regiments, training schools or

headquarters positions. Positions are grouped as post-

ings, which are deﬁned as a particular combination of

career, level and unit. Postings have a requirement or

target value, as well as a priority value which deﬁnes

the order in which the postings are ﬁlled.

Individuals will attend postings or be promoted

based on their eligibility. To determine their eligi-

bility, the simulator records and updates a number of

personnel attributes, including current posting, time

in posting, platform endorsement, current level, and

time in level. Personnel may be promoted to a higher

level if they have reached the minimum time in rank

constraint and there is an available posting at the

next level. Wastage is applied stochastically based

on a particular career’s survival proﬁle, where per-

sonnel have a certain probability of leaving the work-

force based on their length of service, as described

by Bartholomew (1971). To ﬁll some postings, per-

sonnel must have the required platform endorsement.

Personnel will either join the workforce as trained in-

ﬂow with a particular platform endorsement or must

attend a transition course to gain a new endorsement.

3.1.2 Athena Pro

Athena Pro is an AB-DES. Personnel are modelled

as individual person agents. They progress through

career pipelines, made up of building blocks: in-

take points, buffers, courses, milestones, and position

groups. These blocks are connected to form levels,

which in Defence workforce modelling generally rep-

resent ranks. A typical career pipeline can be seen in

Figure 1.

A key part of these pipelines is the buffer agent.

Each level in a career pipeline will have one buffer,

that allocates personnel based on position vacancies

and priorities, career progression requirements, and

course vacancies. Upon completion of a course, po-

sition group, or milestone, personnel will return to

the buffer to be re-allocated. Attendance at position

groups, courses, and milestones may give personnel

‘proﬁciencies’. These proﬁciencies can be used in de-

termining whether personnel are eligible to ﬁll vacan-

cies in blocks, using prerequisites. All blocks may

Defence Workforce Transition Simulation and Analysis

251

Figure 1: An example of a typical career pipeline in Athena Pro. In this pipeline, personnel enter the system as new recruits,

before attending courses in a training level. They then ﬂow through the pipeline, ﬁlling positions and completing courses.

have prerequisites, which will restrict the ﬂow of per-

sonnel from the buffer. These prerequisites include:

• Time-based prerequisites: These prerequisites

take the form of ‘time since X’, or ‘time spent in

X’.

• Entity and proﬁciency prerequisites: These pre-

requisites can be mandatory completion of one or

more particular entities or proﬁciencies, a mini-

mum number of particular entities or proﬁcien-

cies, or one or more particular entities or proﬁ-

ciencies having not been completed.

Position groups, courses, and milestones may

be in multiple levels, or careers, so that the com-

plex inter-dependencies in Defence workforces can be

modelled.

3.2 Optimisation

The Athena Lite optimiser is designed to determine

recruitment and promotion plans that help minimise

shortfalls and excesses of personnel within different

careers and levels over time. These shortfalls and ex-

cesses are costed (and can be costed differently for

different workgroups and ranks) in generic units by

the workforce planner, and the optimiser attempts to

minimise the total cost over time.

The optimiser uses a customised heuristic which

involves backpropagating personnel demand from

combinations of level and time, and then using a

Dijkstra-like pathﬁnding algorithm (Dijkstra, 1959)

to ﬁnd level-progression pathways for personnel

(whether already in the system or recruited through

inﬂows) to ﬁll that demand whilst minimising the to-

tal cost. The pathﬁnding must consider several con-

straints such as time in rank limits and recruitment

year-on-year change limits. Wastage is handled de-

terministically, where a person’s supply is a deci-

mal value between one and zero, and application of

wastage reduces this value.

This algorithm has shown favourable solution

quality and speed, when compared with a Linear Pro-

gramming implementation. Further description and

evaluation of the optimisation algorithm will be in-

cluded in future work.

3.3 Web-Based User Interface

Athena Lite and Pro are web-based simulation tools,

designed with intuitive user interfaces. These user in-

terfaces are key components in ongoing Athena use,

as they allow Defence workforce planners with little

modelling and simulation experience to create, use,

and validate the workforce models.

Athena Lite and Pro are currently in active use by

ADF workforce planners, with feedback incorporated

into new releases.

3.3.1 Building Workforce Models

As it is a requirement that users build their own mod-

els, effort has been put in to make this process as

simple as possible. Data is entered via Microsoft Ex-

cel, or via the UI, to facilitate data transfer between

personnel management databases and Athena. Ca-

reer pipelines are built in Athena Pro using ‘drag-and-

drop’, where users drag entities into the pipeline and

drag connections between them based on the ﬂow of

personnel in the workforce. Extensive data validation

with clear and understandable error messages is also

completed.

WAPAT 2023 - Special Session on Workforce Analytics - Practical Application and Theory

252

3.3.2 Visualising Simulation Results

Athena Lite and Pro provide a variety of results vi-

sualisations. These visualisations are speciﬁcally cre-

ated to highlight personnel shortages, workforce bot-

tlenecks, and capability risks.

In both Athena Lite and Pro, charts can be created

that show various statistics about the workforce, per-

sonnel and position ﬁll. Some examples include:

• supply vs. demand statistics,

• frequency of observed undersupply,

• wastage and promotion statistics,

• personnel progressions statistics, such as time in

level, time in posting, and time in service,

• course graduations, session sizes, and input queue

statistics.

Some charts created using Athena Lite and Pro can be

seen in Section 4.3.

The Sankey diagram in Athena Pro shows the ﬂow

of personnel through the career pipeline, at each sim-

ulation timestep. Each node and arc in the career

pipeline, is represented in the Sankey diagram. The

size of each arc represents the amount of personnel

ﬂowing through that arc, while the colour displays

whether the secondary node is above, below, or at ca-

pacity. The bar above each node represents the ﬁll

level of that node, while the colour of the node rep-

resents the type of node. A timeline panel is dis-

played along the bottom of the Sankey diagram. This

panel contains a vertical bar that represents the cur-

rent timestep that the Sankey diagram is displayed

at. This bar can be dragged along the timeline to

set the timestep. The user can also use the ‘play’

button to automatically step through the simulation

timesteps, with the Sankey diagram changing at each

timestep, or the ‘forward’ and ‘backward’ buttons to

click through the simulation timesteps.

3.3.3 What-if Scenarios

‘What-if’ scenarios can be quickly and easily run and

analysed. A scenario in Athena is deﬁned as a par-

ticular representation of a Defence workforce. This

scenario has speciﬁc input parameters, careers, and

events, resulting in speciﬁc simulated results. Sce-

narios are grouped in a study. From some ‘baseline’

scenario, child scenarios can be created, where all the

contents of the ‘baseline’ scenario are copied into a

new scenario. This child scenario can then be mod-

iﬁed and compared to the baseline scenario to de-

termine the effect of the modiﬁcation on the work-

force, thereby answering ‘what-if’ questions, related

to workforce and training policy changes, recruit-

ment, wastage and promotion changes, and capabil-

ity transitions. An example of the scenario tree used

in the analysis of the AAvn workforce transition is

shown in Figure 3.

4 USE IN THE SIMULATION OF

THE AUSTRALIAN ARMY

AVIATION

Seven careers were modelled to provide a complete

analysis of the AAvn capability. These careers in-

cluded engineers, technicians, ground and air support,

pilots, and aircrew. Models were built in Athena Lite

and Athena Pro. In this section, only the pilot model

will be discussed. Three main scenarios were investi-

gated. These were:

• Scenario 1: analysis of the ‘as-is’ workforce

to ﬁnd whole-of system health, workforce bot-

tlenecks, and high-risk positions, assuming no

changes to AAvn workforce capability and re-

quirements.

• Scenario 2: Analysis of the workforce during and

post Tiger to Apache platform transition to exam-

ine the impact of the transition on the workforce

and the ability of AAvn to meet milestone capabil-

ity dates. Various transition plans were modelled

and analysed in Athena Lite, as in Figure 3. Those

of particular interest to AAvn workforce planners

were then modelled to a higher ﬁdelity in Athena

Pro.

• Scenario 3: Optimisation of recruitment and pro-

motion numbers in order to meet AAvn work-

force requirements, while undergoing the Tiger to

Apache platform transition. This involved using

the Athena Lite optimiser to optimise inﬂow and

promotion numbers, and using these results as in-

put to Athena Pro, in order to inform training re-

quirements.

Athena Lite and Athena Pro were used together

in a complementary manner to complete these anal-

yses. The simulations were run over 10 years, with

30 Monte-Carlo trials per scenario. The number of

Monte-Carlo trials can be chosen by users, and in

choosing this the size of the workforce being mod-

elled (and the corresponding simulation size), and the

amount of available computation time should be con-

sidered. In this case 30 trials were chosen as this pro-

vides a reasonable error, while still facilitating com-

putations. All models, and results discussed in this

section are only to demonstrate the use of Athena, and

Defence Workforce Transition Simulation and Analysis

253

Figure 2: The Sankey visualisation in Athena Pro. This visualisation shows the ﬂow of personnel at the timestep May 2026.

The ﬁll level of particular nodes can be seen by the bar on top, while the capacity size of ﬂow, and capacity at the succeeding

node are represented by the size and colour of the arcs. The colour of each node represents the type of node.

Figure 3: The scenario tree in the AAvn workforce transi-

tion analysis. A number of different transition plans, and

inﬂows are investigated.

not representative of ADF capability. Chart sizes are

notional.

4.1 Simulating AAvn in Athena Lite

In all three scenarios, the career and level struc-

ture remained the same. Pilots in ranks Lieutenant

(LT), Captain (CAPT), Major, (MAJ) and Lieutenant

Colonel (LTCOL) were modelled and move through

the workforce as in Figure 4. Pilots were pro-

moted through the ranks when they reached spe-

ciﬁc minimum time in rank requirements, and there

was an unﬁlled position they were eligible for at the

higher level. As Athena Lite does not model train-

ing courses, the training phase of a pilot pipeline was

modelled as a training (TRN) level. Wastage was

modelled using survival curves based on historical

data.

Figure 4: Progression of pilots throughout their career, with

the application of inﬂow and wastage shown.

All AAvn positions were modelled, including

those in AAvn operational units, 1 AVN REGT, 5

AVN REGT, and 6 AVN REGT, and those in generic

school or headquarters units. Personnel entered the

system as inﬂow with a singular platform endorse-

ment, that will deﬁne which units they can ﬁll posi-

tions in.

The model had initial states, based on current per-

sonnel data, and position requirements that changed

over time, based on planned establishment changes.

In scenarios 2 and 3, when the platform transition oc-

curred, 1 AVN REGT was modelled as two separate

units, with different platform endorsement require-

ments. This changing requirement was modelled as

in Figure 5, where the requirement for Tiger trained

personnel was reduced, as personnel were sent to con-

version training. However to ensure no loss of capa-

bility, the requirement for Apache trained personnel

increased at the same rate.

WAPAT 2023 - Special Session on Workforce Analytics - Practical Application and Theory

254

Figure 5: The changing requirement at 1 AVN REGT dur-

ing the transition period. The requirement for ARH trained

personnel reduced, while personnel were sent to conversion

training. At the same time, the personnel requirement for

Apache trained personnel increased.

In scenario 2, a number of ‘what-if’ scenarios re-

lating to transition timing, and personnel inﬂow were

completed. These ‘what-if’ scenarios are fast, and

simple to run using Athena Lite due to its relatively

low ﬁdelity compared to Athena Pro. As such, the

majority of these scenarios were investigated using

Athena Lite. Scenarios of particular interest to AAvn

workforce planners were then modelled in more detail

in Athena Pro. In scenario 3, optimisation for inﬂow

and promotion was completed.

4.2 Simulating AAvn in Athena Pro

Career pipelines of the AAvn workforce were built,

based on career category management requirements

and Defence workforce planner expertise. In this

pipeline, personnel entered the career and become

ARMY-PILOT-TRN, representing the training level.

During this level they complete a number of ﬂying

courses. The ﬂow of personnel to these courses was

restricted by prerequisites.

Flow of personnel is also restricted by the number

of positions in each position group, and the session

sizes and timings in each course. Extra personnel are

stored as overﬂow in the buffer. Positions in position

groups had start and end dates to represent changing

requirement. This changing requirement for scenar-

ios 2 and 3 was as in Figure 5. Similarly session tim-

ings and sizes changed over time, as school schedul-

ing changes were made, and platform transitions oc-

curred. These sessions also had pass rates based on

historical data. These pass rates were particularly rel-

evant for courses in the ‘ARMY-PILOT-TRN’ level,

as upon failure of these courses pilots do not continue

through the career pipeline.

Inﬂow is controlled by the intake points and was

set by the users. Upon inﬂow into the ‘ARMY-PILOT-

TRN’ level personnel may become overﬂow in the

buffer, if the planned sessions were not frequent or

large enough. This occurs similarly for promotion be-

tween levels. This is displayed as a bottleneck in the

Sankey diagram, as well as in the chart statistics.

Detailed initial personnel, position, and proﬁ-

ciency history was used as input to the simulations.

This data was collected using personnel management

systems, cleaned and converted to the required for-

mat.

4.3 Discussion and Comparison of

Results

These models were validated by AAvn personnel, and

the results were used to inform the workforce transi-

tion plan throughout the acquisition process. Due to

the high ﬁdelity modelling in both Athena Lite and

Pro, there was a large number of interesting results

that came from this investigation. Some key observa-

tions found were:

• Multiple transition plans were tested and analysed

in Athena Lite. The variables altered in these

‘what-if’ scenarios were the transition course

length, transition course numbers and transition

course delay. A comparison of the supply vs. de-

mand at 1 AVN REGT, throughout these transition

plans can be seen in Figure 6. It can be seen that

the plan that leads to the lowest risk of undersup-

ply in 1 AVN REGT is transition plan 2.

• As Athena Lite simulates all positions in AAvn,

the impact of the transition on the entire work-

force could be seen, and not just the direct im-

pact on the unit where the transition is occurring.

For example, as an increased number of ARH pi-

lots are required to ﬁll ARH positions in 1 AVN

REGT, attend transition training to Apache, and

ﬁll Apache position in 1 AVN REGT, less are

available to ﬁll school and headquarters positions.

This leads to a shortage in these units in compari-

son to when no transition occurs.

• Using the results from scenario 2 in Athena Lite,

transition plan 2 was modelled in Athena Pro. A

comparison of the results between Athena Lite

and Athena Pro highlighted a number of things.

First, while the overall number of personnel in the

workforce was approximately the same in both

tools, there were differences in the number of

personnel per rank. This is because, in Athena

Defence Workforce Transition Simulation and Analysis

255

Pro promotion eligibility is more restricted than

in Athena Lite, with the additional prerequisites

of proﬁciencies and courses. Figure 7 shows

the number of personnel waiting to attend the

CAPT Careers Course, a prerequisite to being

promoted from ‘ARMY-PILOT-LT’ to ‘ARMY-

PILOT-CAPT’. It can be seen that this course is

unable to meet the student demand, leading to less

people being eligible for promotion to the follow-

ing rank.

• The cost-based optimiser is also used to optimise

the inﬂow and promotion numbers for this sce-

nario. The resulting supply vs demand results in

Athena Lite can be seen in Figure 8. In this ex-

ample, it can be seen that even with optimal re-

sults there is still some undersupply and oversup-

ply, especially in the ﬁrst few years. This is be-

cause progression of personnel through the ranks

is restricted by a minimum time in rank constraint,

and as the ranks start undersupplied, in order to

meet the requirements at higher levels, the LT and

CAPT ranks need to be oversupplied and under-

supplied, respectively.

• The optimisation results in Athena Lite show the

required number of inﬂow and promotion num-

bers, however does not detail the course require-

ments to meet these numbers. To calculate the

required number of graduations per year in pilot

training courses and in pilot promotion courses to

meet this optimal inﬂow and promotion, Athena

Pro was used. Intake points in the pilot pipeline

were set to match the optimal inﬂow, and all

courses were set to ‘on-demand’. This means that

the set sessions for these courses were ignored,

and they were instead allowed to run when the

minimum session size was met. An example of

the results from this are seen in Figure 9 where

the required number of graduates from the ‘Ro-

tary Wing Training’ course at particular times to

meet the required optimal inﬂow is shown. These

results inform the scheduling of courses in order

to meet the personnel requirement.

5 CONCLUSIONS

Athena Lite and Pro represent large improvements in

the ability of Defence workforce planners to accu-

rately represent and simulate the ADF workforce. To-

gether they provide a capability that is able to forecast

personnel shortages, identify bottlenecks in the sys-

tem, and optimise recruitment and promotion targets.

Figure 6: The supply vs demand of pilots at 1 AVN REGT

with Apache platform endorsements, under different transi-

tion plans.

Figure 7: The input queue at the CAPT Careers Course,

a prerequisite for promotion from ARMY-PILOT-LT to

ARMY-PILOT-CAPT. These are the number of students

waiting to complete this course. The increase in this queue

indicates not enough sessions are being held to meet the de-

mand.

Figure 8: The supply vs demand for each pilot rank, af-

ter using the Athena Lite optimiser. Each graph shows the

number of personnel in a particular rank throughout the sim-

ulation time. The requirement line can also be seen. It can

be seen that in order to meet requirement at higher levels,

MAJ and LTCOL, the LT and CAPT ranks must be over-

supplied and undersupplied, respectively.

The relatively lower ﬁdelity simulation tool,

Athena Lite, allows Defence workforce planners to

quickly explore their workforce risks, while still

WAPAT 2023 - Special Session on Workforce Analytics - Practical Application and Theory

256

Figure 9: The number of pilot graduations from ‘Rotary

Wing Training’ in order to meet the optimal inﬂow, as found

using the Athena Lite optimiser.

maintaining enough detail to effectively analyse the

readiness level down to the rank, unit, and career.

Athena Pro is able to simulate the workforce to

a higher ﬁdelity, analysing the ﬂow of personnel

through the system as they complete courses, post-

ings, and gain proﬁciencies.

These capabilities have been demonstrated in the

analysis of the ARH to Apache workforce transition

in the AAvn. In this analysis, various transition plans

were investigated in Athena Lite, before those plans

of particular interest to AAvn planners were modelled

in more detail in Athena Pro. In Athena Pro, the effect

of training courses, complex prerequisite structures,

and more restricted ﬂow through the career pipelines

could be seen. The Athena Lite optimiser was then

used to ﬁnd the optimal recruitment and promotion

strategies throughout this time, in order to meet re-

quirement. Athena Pro was then used to inform the re-

quired training course schedules to meet this demand.

Besides AAvn, Athena Lite and Pro have also both

been used to simulate the entirety of the RAN, includ-

ing over 20,000 personnel, 100 vessels, and 66 ca-

reers. The Athena suite, including both Athena Lite

and Pro, has been accepted as the ofﬁcial workforce

planning tool for the entire Australian Defence Force.

Athena Lite and Pro are both still in active de-

velopment, as ADF users use and request new fea-

tures and changes. The future focus of the Athena

tool suite is on automated diagnosis and reporting.

This involves using machine learning, and data ana-

lytics techniques to analyse the workforce and under-

stand the risks and vulnerabilities. Natural Language

Generation is also being investigated as part of this

process, to produce usable reports of interesting re-

sults, and provide an interface between the tool and

the user. This will improve the interactions that users

have with the Athena tool suite, and the insights that

can be gained.

ACKNOWLEDGEMENTS

We thank Professor Terrence Caelli (Deakin Univer-

sity) for his editing advice and Australian Army Avi-

ation for their ongoing support. We also acknowledge

the Defence Research and Development Canada for

the ongoing collaboration on Athena development.

REFERENCES

Anagnostou, A., Nouman, A., and Taylor, S. J. (2013). Dis-

tributed hybrid agent-based discrete event emergency

medical services simulation. In 2013 Winter Simula-

tions Conference (WSC), pages 1625–1636.

Bartholomew, D. J. (1971). The statistical approach to man-

power planning. Journal of the Royal Statistical Soci-

ety. Series D (The Statistician), 20(1):3–26.

Davenport, J., Neu, C., Smith, W., and Heath, S. (2007). Us-

ing discrete event simulation to examine Marine train-

ing at the Marine Corps communication-electronics

school. In 2007 Winter Simulation Conference, pages

1387–1394.

Dijkstra, E. W. (1959). A note on two problems in connec-

tion with graphs. Numerische Mathematik 1, pages

83–89.

Filinkov, A., Richmond, M., Nicholson, R., Alshansky, M.,

and Stewien, J. (2011). Modelling personnel sus-

tainability: a tool for military force structure analy-

sis. The Journal of the Operational Research Society,

62(8):1485–1497.

Gunal, M. M. and Pidd, M. (2010). Discrete event sim-

ulation for performance modelling in health care: a

review of the literature. In Journal of Simulation, vol-

ume 4, pages 42–51.

Heath, S. K., Brailsford, S. C., Buss, A., and Macal, C. M.

(2011). Cross-paradigm simulation modeling: Chal-

lenges and successes. In Proceedings of the 2011 Win-

ter Simulation Conference (WSC), pages 2783–2797.

Henderson, J. A. and Bryce, R. M. (2019). Veriﬁcation

methodology for discrete event simulation models of

personnel in the Canadian Armed Forces. In 2019

Winter Simulation Conference (WSC), pages 2479–

2490.

Mathew, B. and Nambiar, M. K. (2013). A tutorial on mod-

elling call centres using discrete event simulation. In

ECMS.

Na, H. S. and Banerjee, A. (2014). An agent-based dis-

crete event simulation approach for modeling large-

scale disaster evacuation network. In Proceedings of

the Winter Simulation Conference 2014, pages 1516–

1526.

Nguyen, V., Novak, A., Shokr, M., and Pash, K. (2017).

Aircrew manpower supply modeling under change:

An agent-based discrete event simulation approach.

In 2017 Winter Simulation Conference (WSC), pages

4070–4081.

eguin, R. (2015). PARSim, a simulation model of the

Royal Canadian Air Force (RCAF) pilot occupation

Defence Workforce Transition Simulation and Analysis

257

- an assessment of the pilot occupation sustainabil-

ity under high student production and reduced ﬂying

rates. In Vitoriano, B. and Parlier, G. H., editors,

ICORES 2015 - Proceedings of the International Con-

ference on Operations Research and Enterprise Sys-

tems, Lisbon, Portugal, 10-12 January, 2015, pages

51–62. SciTePress.

Thomas, D., Kwinn, B., McGinnis, M., Bowman, B., and

Entner, M. (1997). The US Army enlisted person-

nel system: a system dynamics approach. In 1997

IEEE International Conference on Systems, Man, and

Cybernetics. Computational Cybernetics and Simula-

tion, volume 2, pages 1263–1267 vol.2.

Skulj, D., Vehovar, V., and Stamfelj, D. (2008). The mod-

elling of manpower by Markov chains - a case study

of the slovenian armed forces. Informatica (Slovenia),

32:289–291.

WAPAT 2023 - Special Session on Workforce Analytics - Practical Application and Theory

258