Application of System Dynamics to Unsealed Pavement Maintenance
Keith T Linard’
linard.keith@gnmil.com.
ABSTRACT: Most pavement maintenance management systems tend to be either non-analytical
databases or statistical correlation models. However, pavement maintenance is part of a complex
system comprising the road pavement, the environment, diverse users, the maintenance authority and
Local/State/Federal Govemments. This system has significant feedbacks, making it a suitable field for
system dynamics enquiry.
This paper discusses a system dynamics based pavement management model that was prototyped
originally by engineering students at the Australian Defence Force Academy (Hyde 1996, Jackson
1997) and refined on contract with the Australian Govemment. The current model was rebuilt in
Powersim Studio and refined in collaboration with a Victorian rural Shire Council. The model
analyses the pavement deterioration over time of 530 individual segments of unsealed rural road,
prioritising rehabilitation treatments based on user preferences and budget constraints and identifies
the consequences of different budgetary approaches. Feedback to the decision makers includes the
number of households served by very rough roads, the number of user complaints and roughness
related accident costs and vehicle operating costs.
Keywords: Pavement maintenance management; pavement life cycle costing; unsealed road
maintenance; transport economics; economic evaluation; system dynamics.
Introduction
With the public sector reforms of the past two decades, Australian Road Authorities have had major
functions trimmed, outsourced or simply chopped. Probably more than most areas of Govemment,
road asset managers are being required to work ‘smarter’. The ‘outsourcees’, road maintenance
companies, are under a twin squeeze - to win maintenance management contracts in a very
competitive environment and to satisfy shareholders concemed with retum on investment. Both road
asset managers and maintenance contractors require tools to assist in ‘whole-of-life’ cost optimisation
in respect of road maintenance.
Over this period two approaches to computer based pavement management system (PMS) have gained
widespread use. The first approach is a database PMS, which catalogues the cunrent state of
pavements and facilitates budget decision-making. The database PMS has little predictive capability
and provides little guidance on altemative policy levers or the implications of such choices. The
second approach utilises sophisticated statistical correlation modelling based on data relating to
diverse factors including pavement type, environment, vehicle loadings, vehicle usage, maintenance
and rehabilitation pattems. The Word Bank's HDM-4 model set the conceptual pattem for this
approach. (Austroads 2008) These models are applied in a predictive sense, based on the assumption
that the identified correlations will persist into the future. They are widely used for highway planning
and top level budgetary planning but, despite urging from Federal Govemment agencies, have found
little favour at the Local Goverment level, where database PMS are more common. In respect of
gravel roads, the situation is even more unscientific. A study between 2000 and 2002 found that fewer
than 15% of Australian Local Councils with at least 50 km of unsealed roads used any form of
pavement management system. (Austroads 2006)
At the Local Govemment level, politics is important, alongside economics and engineering, when
concems arise about road conditions. This highlights another set of stakeholders, the road users,
whose input to the maintenance decision process operates within the much fuzzier and qualitative
political environment, and whose desire for quality roads is balanced by their desire for other public
1 Keith is a former Senior Lecturer in Civil Engineering at the University of New South Weles, Australia, and
Director of the UNSW Centre for Business Dynamics & Knowledge Management
goods and/or lower taxes. ‘Hard systems’ operational research tools, such as HDM-4, are not suited to
this environment. There is a need for analytical and decision support tools for road asset managers
which can address both the ‘hard’ quantitative dimensions and the ‘soft’ qualitative dimensions.
Of hard systems, soft systems and system dynamics ...
Within the diverse systems disciplines the distinction between ‘hard’ and ‘soft’ systems is important to
the understanding the value added of system dynamics modelling techniques.
Hard systems are characterised by:
e —clearand unambiguous objectives;
e widespread agreement with the objectives;
e high degree of agreements on the facts; and
e high degree of knowledge conceming the principles of operation.
In such situations the technical decision paradigm is optimisation and traditional operations research
techniques have a good track record.
Soft systems, on the other hand are characterised by:
e multiple objectives which may be fuzzy or conflicting;
e multiple stakeholders who may have multiple and/or conflicting interests;
e noclearagreement on the objectives; and
e complex interrelationships between system elements which may not be well understood or
which may even be subject to dispute between competent professionals.
In soft systems, human rather than technical issues dominate, and the paradigm is one of mutual
learning between client, project team and diverse stakeholders. An example of a soft systems
problem would be that of urban accessibility. To the highway engineer a freeway may seem an
obvious solution. Some house owners might agree, at least when caught in peak hour traffic -
provided the road is located in someone else’s backyard. Others may be concemed about
environmental issues and support public transport solutions. Y et others may consider the problem to
be one of work place location - bring the jobs to the people rather than vice versa. Whilst economists
might argue that the ‘real problem’ is the lack of an appropriate road pricing strategy.
Road maintenance and system dynamics
At first glance, the maintenance of roads might seem to be a classic ‘hard systen’
e — the objectives are clear and unambiguous - pavements should be safe & smooth;
e there is widespread agreement with the objectives - there is no ‘pothole protection society’ or
‘save the roughness’ campaign;
e high degree of agreement on the facts - both engineers and the public can agree on what
constitutes a rough driving surface, and understand that maintenance reduces roughness; and
e high degree of knowledge conceming the principles of operation - at the least, this is one field
where the public will defer to engineering competence.
However, it’s not that simple. Pavement roughness is the consequence, inter alia, of trade-offs
between routine maintenance decisions, pavement reconstruction and decisions relating to overall
network investment, which influence traffic intensity on particular road links. In addition, there is a
fundamental trade off between roads related expenditure and expenditure on other community
infrastructure and social services. This is illustrated in the causal-loop diagram, Figure 1, below.
(One interprets the causal diagram as follows: An ‘S’ represents a causal change in the [S]ame
direction, whilst an ‘O’ represents a change in the [O]pposite direction. Thus an increase in ‘Routine
Maintenance $$$’ s for Road A’, all else being equal, leads to a decrease (i.e., a change in the opposite
direction) in the ‘Roughness of Road A’.
Figure 1 also indicates two related systems - that of the technical managers, the assets engineering
staff, and the political decision makers, the Shire Councillors. Beyond this are the State and Federal
political systems which have many more resources and which are also subject to influence by the local
residents and their representatives. Investment decisions on road system maintenance and
rehabilitation, based on short term budgetary considerations, can have very significant implications for
diverse social goals, especially in the rural Local Goverment sphere. (Austroads 2007)
Figure 1: Causal Interrelationships Within & Between the Political & Engineering Systems
Civil Engineering System
of technical trade-offs
between competing needs
Local Government Political System
of budget trade-offs between
competing social, environmental
& technical needs
System dynamics is particularly useful in understanding the linkages between the qualitative and the
quantitative aspects of road asset management. System dynamics modelling employs a set of
techniques that allow both quantitative and qualitative factors to be incorporated..
Elected decision maker focus of the model
Local Goverment is applauded because it is the level of Govemment closest to the people. Local
Councillors have the difficult task of allocating scarce resources among many worth competing
demands in both the engineering area and in the human services area. All too often, engineering needs
are presented in a highly technical mathematical fashion, which are much harder for the individual
Councillors to discem than the social data.
One valuable aspect of system dynamics modelling is that it affords a mechanism to communicate the
implications of the technical results, for example, the number of households which will be served
extremely rough gravel roads or the number of residents who are expected to lodge complaints
regarding the state of the roads. More to the point, it becomes possible to highlight which particular
roads are likely to be deficient overtime. A named road has far more impact on a decision maker than
an anonymous road.
As illustrated in Figure 1, the elected official is a critical part of the feedback loop which comprises
the model. The simulation model provides information to the decision maker on the road system
implications of budget options, and on the numbers of taxpayers affected and to what degree.
Background to the development of the simulation model
Preliminary work on the application of system dynamics modelling to road maintenance management
‘was undertaken by the author in the mid 1990's at the Australian Defence Force Academy (ADFA).
The current model grew out of work with a Victorian rural Shire in 2008.
The Shire had participated with the Australian Road Research Board, over a period of 5 years, in the
development of statistical correlation based road pavement deterioration models. (ARRB 2006).
In 2008 the Shire undertook a review of its unsealed road assets, including sampling of pavement
depth, pavement crossfall and the condition of drainage. As indicated in Table 1, the study showed
that the unpaved road system (approximately 40% of all roads in the Shire) has seriously deteriorated.
Table 1: Pavement Thickness, Shape and Drainage Shortfall - Categorised by Traffic Volumes
Average Daily Total Length of PAVEMENT DEFICIENCY Substandard | Substandard
Traffic (ADT) Gravel Pavement APPROACH 2 Surface Drainage
(kms) Substandard Pavement Shape (kms)
Thickness (Less than 50mm) (kms)
Length Volume (Too flat to (Table Drains
(kms) (Cum) shed water) too shallow)
200 - 500 vpd 12.9 2.3 2198 Ta. 19
100 - 200 vpd 36.6 14.8 11207 14.8 4.3
50 - 100 vpd 121.1 59.6 41840 58.8 21.0
20 - 50 vpd 226.9 148.3 88010 135.7 123:
Less than 20 vpd 137.2 105.2 53612 100.1 46.5
TOTAL 534.8 330.2 196867 316.5 146.0
e 103kmof gravel road, or 19% of the gravel road network, effectively have no gravel left -
they are down to the clay sub-base;
¢ 330 kmof gravel road, or 60% of the gravel road network, have less than 60mm of gravel,
which is well below the desirable ‘trigger’ for resheeting;
e 425 kmof gravel road, or 60% of the gravel road network, have at least a 100mm shortfall
below the desirable design thickness;
e 316 kmof gravel road, or 60% of the gravel road network, has lost its surface shape such that,
if it rains, water will pool and the surface will deteriorate;
e 146 kmof gravel road, or 27% of the gravel road network, has inadequate table drains such
that, if it rains heavily, the pavement risks collapse.
This pen-picture of the road network is mirrored in resident dissatisfaction. Over the past 5 years,
State Govemment surveys showed that Shire’s ratepayers have a higher level of dissatisfaction with
roads than with any other service provided by the Shire, and that the Shire performed worst amongst
other Councils in its demographic grouping (Department of Victorian Communities 2007).
Figure 2: Community (Dis)Satisfaction with Roads
WEExcetert = Good EJAdequate, ©] Needs some improvement
oe
aL
Oner
The Shire Counctts tn An Councits,
Group
100 .
* i
© H st had
* s |
= H
2 i 2 2
. H
H = »
i
H
}
H
}
H
0
“20
The Shire _—- Performance Over Time
100.
*
0 “0 0
2 “ 4s
“o J v s ”
i A
we] [2
% Excelieny
% Noods
Improvement
‘% Excell
Good’
Adequate
Noeds
Improvement
In 2007, the most recent Customer Satisfaction Survey, 55% of respondents considered that Shire’s
performance in this area was inadequate. The findings of these surveys are mirrored in the Shire’s
“customer request’ records. An analysis of 5 years data from the Shire’s customer request records
found over 650 complaints per year conceming gravel roads, from 250 to 300 residents per year (i.e.,
some residents raised multiple concems). Table 2 summarises the reasons cited for their concems.
Table 2: Key Roads Concerns In Community Satisfaction Survey
Reasons Cited re Need for Improvement in Roads Number of
Respondents 189
More frequent/ better re-surfacing of roads 40
Improve/More frequent grading etc of unsealed roads 19
Improve standard of unsealed roads (loose gravel, dust, corrugations) 17
More frequent/ better slashing of roadside verges 14
Fix/ improve unsafe sections of roads 12
Fix/ improve edges and shoulders of roads 5
Improve/ Fix/ Repair uneven surface of footpaths 28
Table 3: Analysis of Residents Concerns Re Gravel Roads in the Shire: Mar 2003- J an 2009
Needs Pot Dust Corrugations | Please Allissues
grading Holes Seal re Gravel
Road Roads *
Letters & Emails 321 586 608 515 455 2080
Telephone Calls 555 869 238 684 82 1851
TOTAL 876 1455 841 1199 537 3931
Notes: 1. Some requests identify multiple issues, hence totals for specific issues do not equal the ‘All Issues’ total.
With this level of concem, it was decided to develop a model which could relate community concems
to the technical conditions of the unsealed road network and to the level of resourcing. .
Structure of the model
The model is constructed in Powersim Studio (ver. 7.0). The Powersim model is in five parts:
e roughness progression & roughness rehabilitation modules;
e agravel loss & gravel pavement rehabilitation modules;
e client impact & complaints modules;
e — roughness related accident and vehicle operating costs modules;
e net present value module and budget modules.
The model is designed around an array structure which pennits analysis on a road by road basis. In
this particular example, the Shire identified 530 specific road segments. Some 28 data elements in
relation to each segment was imported into the model from an Excel based road register.
Figure 3: Structure of Road Register Data Imported into Powersim Model
Co ea i Se Le ee a Ww AS
Design Time
Int Graal) int Te Grading | Sng | Nos [Piionty | Prony
Det Thickness] Shapo | int | Ser | Greding | jax Wo [Proporice| Seoro | Scars
Grading | State
[Drainage| 9 ‘| per Year | Ser \Grading|Resheet
Poors | (4ays) |? i
Road Section Section) Section| Int
Asset Longin Wieth
No tm) | (rn)
Design
ESA
2
3 soo_| 24 | 960 szo_| roo | 0 | 100 | 110500] 20 | 0 | 10 | 19 | wo | 2 3 10 504
4 rou_| 25 | 2520 [11731045] 120 | 20.0 | 197 | 083 | 112000] 200 | o | 10 | 19 | 120 | 2 3 5 a3
5 4500 27 | 12030 [e.azo0608| 60 | 20.0 | 9 | 100 | 24500 200 | 7 | 07 | o4 | 80 | 2 3 H 14
6 a7 | 27 | 4617 [10509073[ 180 | 200 | 9 | 100 | 34500 720 | 0 | 10 | 03 | 190 | 2 3 3 23
7 3ac0_| 3 | 10350 [40-190098| 120 | 1000 | 29 | 091 | sas00 | 230 | 28 | 05 | a2 | 120 | 2 3 16 a
3 va70 | 36 | 4082 [10.e03ce2| 120 | 611 | 69 | oat | 1000 | 290 | 10 | 10 | oo | 120 | 2 3 @ 73I
In addition there are a number of other initialisation variables maintained in the supporting Excel
spreadsheet which are imported into the model, including:
e — Climatic data (average monthly rainfall)
e — Soils data (average sub-base bearing capacity - Califomia Bearing Ration or CBR)
e Gravel pavement data (gravel size and plasticity index etc)
e Intervention levels for grading or resheeting
e = Client complaint data
The model has report modules corresponding to each of the above model modules. In addition, the
model exports roughness and pavement condition data to Excel spreadsheet, from where the data is
linked to the Shire’s MapInfo geographic information system. Using thematic layers it is then possible
to produce map overlays, as a means of communicating which roads are likely to suffer distress under
specified budget scenarios.
Prioritisation computations module
As the model is based on modelling the behaviour of individual roads, rather than an amorphous
summary, it was essential to be able to allocate resources to rehabilitating specific roads (for example
by grading or resheeting) according to the decisions on priorities applied in practice.
Because priority ordering changes over time, as maintenance work progresses or rehabilitation is
undertaken, prioritising sub-models were developed for both grading and resheeting, allowing the
model to reassign works priorities at the start of each financial year in a manner consistent with actual
practice. (The author has loaded an ‘unlocked’ version of the prioritisation module into the Powersim
User Group Y ahoo site.)
( hhttp://tech. groups. yahoo.com/qroup/powersimtools/files/Design%20Challenge%20%232/ .)
Priorities are re-computed at the start of each year of the simulation, based on actual practice, taking
into account traffic counts, % heavy vehicles, whether the road is a school bus route, gravel depth
shortfall and number of properties served. Projects are then ‘undertaken’ in the model on a monthly
basis, based on actual work throughput parameters, until the budget is exhausted.
Figure 4: Gravel Resheet - Prioritising Projects Within Resource Constraints
Ine Read
nellbesttin, QO.
Gravel Thickness
pond esa sertr QL
GravelDepth- Shortfall CQ}.
Rood Propertes- C}—
_-—~ ResheetPriority
Roadnum Hierarchy Q)
ResheetEffortRequired pe
Resheet Budget ©). ‘ 7
peshew specs @ i =
ce)
ResheetProgram-
Budget
Gravel loss & gravel resheeting (rehabilitation) module
This part of the model does not purport to introduce any new insights into the pavement engineering
relationships. These are based on Australian Road Research Board (ARRB) cern ata eae
(ARRB 2006) and related research embodied in the HDM-4 models (Austroads 2008).
The specific deterioration algorithms in respect of pavement gravel loss used in the Shire study were
derived from ARRB empirical research. However, the model is designed such that it can incorporate
other pavement deterioration algorithms, such as that used in HDM-4. The data was incorporated into
the model to provide an estimate of monthly gravel loss (in mm of pavement depth) for each of the
530 road segments, based on the respective road segment data on traffic volumes, average rainfall and
gravel characteristics.
This primary stock in this module is pavement thickness, by road. On average, some 10mm to 12mm
of gravel is lost per year as a result of erosion due to traffic, wind and rain. This gravel is replaced by
7
resheeting, subject to budget constraints. Construction practice is that, when resheeting occurs, a
100mm layer is placed. Based on the typical loss rate, this means that, on average, a pavement has a
life of about 8 to 10 years.
Figure 5: Gravel Resheet Module
Time_Since_Grading-TG
Oo
PI_Plasticityindex
i as
Gravel_Thickness O
t z
Resheet_Completed =
Pao PavementDepth_Init
co) Q.
é sheeting
Resheet_Thickness
Gravel_thickness
losing_gravel_depth
Min_Raipfall- resheetProgta
forBrading RoadSec_Lengh RoadSec_Widh Budget
sheeté ffortRequired
GravelPayvement-
DesignDepth
PavementDepth-
StagedOption
Gravel_Thickness
GravelDepth-Shortfal
AddResheetProjects Reshe@t-BeingDone
=
cl
D> Reshegt Timesince- Resheet_Completed
oO GL_IhterventionLevel
Pavement, Resheet_ThisVear
DesignDepth mao
aged
Resheet_Thickness
Resheeting occurs provided the ‘Resheet switch’ is set to 1 for the given road (based on gravel depth.
shortfall, appropriate weather conditions and the intervention levels for gravel depth being met) and
when the prioritisation module identifies that resources are to be allocated to this particular road.
Roughness progression & road grading (rehabilitation) module
The roughness progression algorithms were also derived from ARRB empirical research. (Again, the
module is designed to take similar models from other sources such as HDM-4.) The data was
incorporated into the model to provide an estimate of monthly roughness progression for each of the
530 road segments, based on the respective road segment data on traffic volumes, the percentage of
heavy vehicles, mean monthly rainfall etc.
Figure 6: Roughness Progression & Grading Module
IRI Roughness-ine ©)
Resheet_Completed
Oa Sa
Read ADT ml Des =~ Grading_Beingdone
Road_HV%
-
<IRI_mo
IRI_max
TRIS Ne
- “9 “AN
ae ‘en
The key stock in this module is ‘Roughness’, as measured by the ‘Intemational Roughness Index’
(IRI). Roughness is added each month for each individual road by the flow variable dIRI_mo, based
on the ARRB algorithms. Roughness is decreased either by resheeting of the road or by periodic
grading. In the absence of other research, the effect of grading in decreasing roughness was based on
Paterson 1987.
Economic Evaluation (Net Present Value) Module
The costs of the annual grading and gravel Resheet program are simple to identify. The benefits of
having a smoother rather than rougher road are more problematic to quantify. The ‘benefits’, in
essence, are the avoided costs associated with, for example:
e Roughness related accidents
e Increased vehicle operating costs and increased travel time related to roughness
e Dust nuisance associated with roughness
e Respiratory disease due to dust
¢ Costs associated with denial of access in extreme rainfall events ascribable to pavement
maintenance policies.
This study only incorporated the first two benefits: reduced accident costs and reduced vehicle
operating costs. Figure 7 illustrates the relationship between road roughness and accident rate, and
road roughness and the vehicle operating costs of light trucks.
Relationship Between Accident Rate and Road Roughness Effect of Roughness on Vehicle Op Cost
fc = 7 w Light Truck (incl travel time)
A 8 ss000
Be =o.eeiex v2hae = = |
Eos t T ae BS 1000 |
eo T a BE 12000
Zin | + og
as gt u000
>, <_e = 300.00 = r
~ 2 = ° 2 4 5 5 10
o 1 2 3 4 6 6 7 @ 8
Pavement Roughness (IR!) Roughness (IRI)
Figure 7: Relationship between Road Roughness and Various Social Costs
Figure 8 illustrates the Accident Rate Module, which computes the expected number of casualty
accidents per year, based on the simulated roughness results. These are then compared with that
expected were the roughness to be at an ‘ideal level’ of IRI =5.
2018/1687 *
2013[16.92
2020/16.50
16.59" [= Casualty Accidents
adding casualty | 2021
Sceigents peryr
ain Hierarchy
(Casualty Accidents by
Year
Roushness_calectsr Qt ——)
.
Roughness_Access1 ()~
Roughness _Access2 Q)G
P
Roughness Access3 (jt Roa ADT — RoadSec_Length
C
AccidentRate-IRI-\.
Relationship \
eas ats ze) 2 reslcaly acces
Accidents-by-Road
Figure 8: Casualty Accident Module - ‘Forecast’ increase based on simulated road roughness
Similarly, Figure 9 illustrates the Vehicle Operating Cost Module, which computes the expected
increase in vehicle operating costs (including travel time costs) based on simulated roughness
compared with that expected were the roughness to be at an ‘ideal level’ of IRI =5.
Excess VehicleOpCosts-
fromRoughness
VOC Variation
J ediumCar
CO) Poadsec_Length
Pavement Roughne®
= Variation-
Heavy Truck
PavementRoughness-
sptRaughness- @ Oo
Road_ADT Road_HV%
Marginal VOC
frargetRoughness
Figure 9: Vehicle Operating Cost Module - ‘Forecast’ based on simulated road roughness
Use of simulator to communicate basic ‘science’ of road deterioration & rehabilitation
The simulator can be used as a decision support tool. However, its primary value is in communicating
the underlying ‘science’ of road deterioration and rehabilitation. What factors affect roughness
progression? How does wet weather, or drought, affect roughness progression? How much impact
does grading have on roughness?
Tt also has significant value in identifying exactly who (which taxpayers) will be disadvantaged
because, for the first time, the decision makers can see which roads and which households will be
affected by deteriorating roads.
Effect of grading on roughness
Road roughness is typically measured using the Intemational Roughness Index (IRI), which is a
mathematically defined summary statistic of the longitudinal profile of the road surface. IRI is a scale
of roughness which is zero for a completely smooth surface, 2 for paved roads in good condition, 6 for
10
moderately rough paved roads, 12 for a extremely rough gravel roads, and up to about 20 for
extremely rough unpaved 4-wheel-drive tracks.
Figure 10 provides several qualitative ‘word pictures’ to enable the reader to understand the
implications of the subsequent discussion of gravel road roughness in the Shire.
World Bank ‘guide’
for developing SAFE
nations DRIVING
SPEED
re Impassible 30 kv
18 Victorian rural local
roads study i
7 ‘intolerable’ ROUGH 40 kanvhe
= 6 level of roughness UNGRADED 4:
Ss 4WD TRAC
a
as je Shire's _ food 50
a 43 f Gravel Roads |:
2 ] Just Tolerabie|
£ 12 7
on 60 knvhr
FH [
o 10 ] =
es -~
# / HT Bad
© ¥ ‘OLDER aH Poor
E Zc 0llCES <a” ante
6
2 Adequate
S 5 DAMAGED ————*—_} 100 kewvhr
a4 statep ___ Good |
5 PAVEMENT Very Good
2
1
o
Figure 10: Word Pictures Explaining the International Roughness Index (IRI) Measures”
Gravel roads require regular maintenance grading to ensure adequate ride quality and safety. Periodic
heavy grading is also required to re-instate the cross-section of the road, reshaping the crown to ensure
that surface water does not pond. Heavy grading is also required to remove deep corrugations and
significant potholes. For major or extensive defects, ripping, reworking watering and compaction may
be necessary.
Regular grading has a disadvantage of loosening up the wearing coarse of the unsealed road and as a
result may increase the rate of material loss. Good grading practice, such as grading after rain when
the wearing coarse has higher moisture content, is advisable.
Routine Grading in the dry season is of limited effectiveness as the absence of moisture can prevent
the reshaped material from ‘bedding down’ (unless watered at additional cost). More damage can be
caused by dry grading than not doing the grading at all.
Figures 11 and 12 illustrate the impact of different traffic volumes on roughness and also the effect of
grading.
Figure 11 depicts a maintenance strategy of grading every 6 months, together with periodic resheeting
(every 10 to 12 years). We see a typical ‘sawtooth’ pattem, where the roughness on the low
trafficked road varies from Intemational Roughness Index (IRI) of 5 to IRI of 8, and on the heavier
trafficked road from IRI of 6.5 to IRI of 11.
William Paterson, Road Deterioration and Maintenance Effects - Models for Planning and Management. The
Highway Design and Maintenance. Baltimore: Johns Hopkins University Press, 1987.
11
Figure 11: Effect on Roughness (IRI) of Grading every 6 months
Pavement Roughness (IRI)
ADT = 175 veh/day
ADT= 30 veh/day
tn ty ty. Se a te ee, ep ee, eye ee
%, , “%, %,_ % ~%, %, “%, “4, °%, °%, Z i a Pe
2 2 ‘2, > ‘2 >
ey Pry Pe, Pry Vy Vey My Mg Wy My My My My By By Be By BR Bs
Figure 12 illustrates a maintenance strategy of grading once per year. The consequence is a much
greater roughness range for both traffic volume situations, with IRI varying from 4.5 to 9 for the lower
trafficked road and from 5.5 to 13 for the higher trafficked road.
Figure 12: Effect on Roughness (IRI) of Grading every 12 months
15
144
a iA hAAAAL AAY
wl ) h | ( Nh Al
5 al
Effect of Resheeting Roads
ADT = 175 veh/day
ADT= 30 veh/day
Pavement Roughness (IRI)
These figures also illustrate the fact that grading does not retum a gravel road to a smooth status.
Typically, grading eliminates about 50% of the difference between the roughness prior to grading and
the theoretical minimum roughness achievable from grading (around IRI =2.5)
Figure 13, overleaf, shows the resulting roughness at the end of the 20 year simulation period based on
continuation of current budget allocations. It shows that close on 50% of unsealed roads in the Shire
will have roughness levels categorised as “intolerable”. In fact, this pattem is fairly representative of
every year in the simulation. The current annual budget for grading would have to be increased by
60% to keep the majority of roads below an IRI of 7.
12
Figure 13: Pavement roughness of 530 unsealed roads - Year 20 based on current budget levels
Roughness (IRI) - All Roads
‘i 1 Se
Pavement Roughness (RT)
Gravel loss and gravel resheeting
Gravel loss is mainly due to erosion of fine particles in the road base gravel - either as dust in dry
conditions or washed off in wet conditions. Larger particles break down under traffic, weathering and
grading. The key factors affecting the amount of gravel lost are traffic volumes, rainfall and the gravel
characteristics. Gravel loss is higher on steep grades and curves.
As the gravel wearing course reduces in thickness, other developments such as the formation of wheel
tuts will generate greater impact on subgrades through moisture penetration, further increasing the loss
of gravel. Similarly, loss of shape leads to ponding of water and pot holing, again further increasing
the loss of gravel. Ideally, gravel roads should be resheeted when the remaining thickness is between.
50mm and 75mm before these additional factors become significant.
Based on continuation of the current levels of funding, the model shows that the volume of gravel
placed each year is significantly less than that required for replacement, as illustrated in Figure 14. (In
fact, the diagram understates the extent of the gravel loss because, as discussed below, many roads
have lost all their gravel. There is no more to lose by erosion.)
Figure 14: Gravel Added Compared With Gravel Loss per Year- $500K annual Resheet budget
Volume of Gravel Added (Resheets) & Lost (by Erosion etc) per Year
sean
& 15,000
% 14.000
> 12.000.
§ 10.000
9,000:
3 son:
F 4.000
aon
a
The consequence of this shortfall is dramatically evident in Figure 15, where the number of roads
without any gravel left (i.e., that are down to the clay sub-grade) rises from around the current 80
roads to over 260 roads, or 49% of the network within 6 to 10 years.
13
In fact, the situation is even worse, because a further 100 roads will have only 10mm to 30mm of
gravel remaining. Noting that gravel loss tends to be much higher as pavement depth decreases, most
of these roads would be down to the clay sub-base within a year.
Figure 15: Transformation of gravel roads to unpaved track status - $500K annual Resheet budget
Loss of Gravel Roads to Unpaved Track (Zero Gravel) Status
Ee s
=
The real implication of the loss of gravel will not be felt until there is sustained heavy rain. The Shire
has been suffering prolonged drought, which at least has the beneficial side effect that a clay track can.
carry heavy vehicles. Once wet, however, the clay surface quickly collapses, as illustrated in Figure
16, with two such roads in the Shire after the last heavy rains.
Figure 16: Rain and unpaved (clay) roads do not go together
Use of simulator to communicate the social and political implications of resourcing levels
Analysis of resident complaints show that they are strongly correlated with the prevailing weather
conditions. In months where there is no rain, dust becomes a major problem, especially where the
gravel layer is very thin. On the other hand in wet weather, complaints from households served by
roads that are down to the clay sub-grade sky-rocket. Such roads can become virtually impassable
ovemight, as illustrated in the above photoes.
To capture this characteristic, and to communicate this effect to the decision makers, the model
incorporates the stochastic variations in monthly weather pattems into the Client Feedback reports,
Figure 19. (Of course the ‘prediction’ of a flood event next year after 10 years of drought might raise
some questions. The model is meant illustrate the impact of weather on client complaints, NOT
predict it)
The model pennits simulation of altemative budget scenarios, varying either or both the annual budget
allocation to maintenance grading (addressing roughness) and resheeting (addressing pavement
thickness, and hence strength). For each scenario there are a variety of outputs geared especially to
the political decision makers.
14
System Wide Presentation of Consequences
The outputs are in two categories. Average system -wide outputs which serve to illustrate how the
Shire’s assets overall are faring. This is a useful basis for comparison with other Shires to compare
how well the commumity’s resources are being managed. Figures 17, 18 and 19 are typical system
wide outputs to assist decision makers understand the implications of their decisions.
Figure 17 suggest that within 6 to 10 years, at current budget levels, the number of roads with
effectively zero gravel will increase from 80 (15% of the network) to 260 (49% of the network).
Because of the stock of roads with very low pavement thickness, this shortfall is not eliminated over
time. At $500K annual Resheet budget, the equilibrium level of clay roads will be between 260 and.
280 out of 530 roads.
Figure 17: Roads with minimal remaining gravel depth - Annual budget $500K p.a.
Number of Roads with less than 5 Years Pavement Life
0.
2003 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
This, of course, translates into affected households (and affected voters). Figure 18 indicates the
approximate number of households which will be affected by the loss of all gravel, i.e., by roads that
are down to the clay sub-base. .
Figure 18: Effect on Households of Loss of Gravel Wearing Course - Annual budget $500K p.a.
Total Households on Roads With Zero Gravel Wearing Course
SsseBesb2uy
The loss of gravel, and the resulting increasing road roughness and dust, in dry weather, and boggy
conditions in wet weather will be reflected in the level of customer requests for remedial action. Two
‘wet weather ‘events’ in Dec 2004 and Feb 2005 generated almost 150 customer requests, compared.
with the average annual total of 600. With almost 3 times the number of roads down to the clay sub-
grade, compared with 2005, a dramatic escalation in complaints is expected. Similarly, the general
level of complaints will rise as the gravel wearing course on many roads disappears, and grading has
little lasting effect on roughness.
15
This is illustrated in the charts in Figure 19 which suggest the expected pattem of taxpayer complaints
as the roads deteriorate over time if current budget levels continue. The simulation suggests a steady
increase in resident complaints, with complaints more than doubling over the 20 year time horizon.
Figure 19: Customer feedback on road conditions - Annual budget $500K p.a.
Total Road Condition Complaints per Year
2009 20:0 2011 2012 «2013 «2018 2015 2016 2017 —«2018 +2019 +—«2020~«2021 «2022 +~«2023 «2024 2025 2026 «2027 ~—«2028
‘Personalised’ Presentation of Outputs - Naming the Affected Roads
The model produces as an output not only ‘anonymous’ average results, but the likely outcome for
individual roads (based on the adopted prioritisation criteria). The elected politicians are thus
confronted with the pattem of outcomes in their particular Ward (electorate). This is produced in
graphical and tabular form and also, through linkage of Powersim to MapInfo, in map form.
Thus, Figure 20 shows the expected situation after 6 years, where almost 50% of the roads are down to
the clay base. The difference in the information content, however, compared with Figure 17 is that the
affected roads can be identified.
Figure 20: Scenario 1 - Gravel depth by Individual Road after 6 years - Annual budget $500K p.a.
Gravel Depth by Each Individual Road
: iy] nk A |
25: Ht
In the chart above, one needs a key to link the road number to the road name. However, the model
exports a table to Excel spreadsheet tabulating roads by township by electorate. Figure 21 shows how
the data is presented to Councillors so that they know precisely which roads in their electorate are
likely to be affected by different annual budget levels.
16
Figure 21: Putting names to the consequences - Tabulating the ‘failed’ roads
Scenario 1: Current Funding Levels ($500K per year)
Xxxxxx Ward (88 roads With Zero Gravel (I.e., down to clay surface) by 2027)
Road Township Road Township Road Township Road Township
Mouvt
dd a
Avdpewo Bodhov Boke BepepBoxe Napic Ehawwe Haotiwyo Exepeae
Baddoves | BoXRow Fovpop BepepBoxe | Aevoys | Orcs | Aiddio Mouvt
= = x Eyeprov
Avtthe Mouvt
"
Bpocnaco | Bodkav Toceppevt | BepepBoxe | PidAeto Wwyniorov | Oo oor Eyemov
Tnalistan Tinnat
Finally, the model is set up to export both roughness data and data on remaining pavement depth to
Excel spreadsheet. The spreadsheet is linked to a MAPINFO table, permitting the graphical display of
the simulation outputs. This is illustrated in Figure 22 where the red lines indicate roads that have lost
all gravel, orange lines indicate roads that will be down to the clay sub-base within 2 to 3 years, and
green lines indicate roads with gravel depth greater than 60 mm. This may be even more meaningful
to elected representatives than collections of tables and graphs.
Figure 22: Simulation Roughness Results Linked to GIS
— ” Gravel Road Network
iq: SS
!
ae
<
:
ey
LEGEND
Gravel Depth > 60mm sm
Gravel Depth 10-60mm ==
nw Gravel Depth < 10mm ===
Using the model - Findings from ‘what if’ budget scenarios
The model shows that the Shire faces a very dramatic resourcing problem Continuation of current
budget policies will reduce half the gravel road network to clay track status within 6 to 10 years, with
disastrous consequences for access to many properties when heavy rains retum.
Just to keep the road network in its current condition over the next 20 years requires an annual lift in
budget resourcing to 225% of the current budget ($1,125,000 p.a. compared with $500,000 p.a.).
In order to eliminate the backlog which has resulted from years of underfunding will require an annual
lift in resourcing to 300% of the current budget (1,500,000 p.a.).
17
The model has already been used to revalue the gravel road asset for accounting purposes and to
identify the corresponding depreciation based on replacement cost. Hitherto, the Shire accounts used a
simple straight line depreciation for gravel roads based on a 20 year life. The modelling suggested
that actual depreciation (i.e., loss of gravel thickness) was proceeding at double the accountant’s
depreciation rate.
The model will be used in future budget negotiations to argue for significant increases in both
Model Limitations
This simulation model does not purport to predict the future, especially when the relationships are so
dependent of climatic conditions. It does, however, provide a powerful basis for identifying trends in
outcomes based on altemative policies with respect to resource inputs. The resourcing shortfalls in
current budgets are so significant that the uncertainties in the modelling process pale into
Conclusions
This paper has discussed the application of system dynamics modelling to the management of the road.
maintenance asset. From the work thus far the following advantages can be claimed for SDM over
more traditional statistical correlation modelling:
e By focusing on key stocks (especially amount of gravel on the roads) the implications of
years of underfunding become evident, and the lengthy time frames to redress the situation
can be understood.
e The graphical interface makes apparent the relationships between key variables for the
decision makers;
e “Soft” (qualitative) data, which is important in the decision making, can be readily
e The fundamental feedback relationships in this particular system are the technical advisors
using the simulation model to provide advice, in politically and socially relevant format, to
the elected policy makers, based on scenarios they identify.
Australian Road Research Board. Road classifications, geometric design and maintenance standards
for low volume roads. Melboume:ARRB. 2001.
Australian Road Research Board. Deterioration Models for Unsealed Roads - Victoria.
Melbourne:ARRB. 2006.
Austroads Research Report AP-R267/05: Refinement of Road Deterioration Models in Australia.
Sydney: Austroads. 2005.
Austroads Technical Report AP-T46/06: Asset Management of Unsealed Roads: Literature Review,
LGA Survey and Workshop. Sydney: Austroads. 2006.
Austroads Technical Report AP-T80/076: Process for setting intervention criteria and allocating
budgets: Literature review. Sydney: Austroads. 2007.
Austroads Technical Report AP-T97/08: Development of HDM-4 Road Deterioration (RD) Model
Calibrations. Sydney: Austroads. 2008.
18
Department of Victorian Communities. Local Govemment Community Satisfaction Survey. 2004,
2005, 2006, 2007.
Hyde, K.. (1996), A System Dynamic Model of the Unsealed Pavement Maintenance System. Thesis
submitted for 4” Y ear Bachelor of Civil Engineering Degree, University of New South Wales.
Jackson, J. (1997), A Predictive Pavement Management System for an Urban Road Network. Thesis
submitted for 4" Y ear Bachelor of Civil Engineering Degree, University of New South Wales.
Paterson. W. Road Deterioration and Maintenance Effects. Baltimore: John Hopkins University Press.
1987. p.83-86
19
