Case Study 2.36: A Risk-Based Condition



Case Study 2.36: A Risk-Based Condition

Assessment Policy

Dunedin City Council developed a Policy to guide its condition assessment approach for the three waters activities. The Policy aims to provide direction on the required outcomes and key roles and responsibilities for the Condition Assessment Programme. This fostered a sense of ownership across both the Asset Planning and Operations teams.

The Policy recognises that condition assessment is a prerequisite to renewal decisions and maintenance planning. The reasons for undertaking condition assessment are therefore:

To achieve service levels and operational objectives.

To better understand actual condition of the pipelines.

To improve confidence in asset data.

To refine renewals planning and potentially remove peaks.

To possibly extend the service life of the pipeline (i.e. reducing water pressure where feasible).

To ensure a co-ordinated approach with the Council’s best practice AM objectives.

The Policy outlines a condition assessment approach primarily based on asset criticality. The probability of failure is also considered where relevant. The Policy also identified how to optimise the data collection and refine the existing data.

The key commitments documented in the Policy document were as follows;

To carry out planned condition assessment on above and below ground assets across the three waters activities.

To develop an asset sampling programme as per the planned condition assessment process diagram.

To carry out opportunistic condition assessment through routine activities and site visits (for above­ ground assets) or in coordination with any work that requires excavation/exposure of an asset (for below ground assets).

Condition assessment to be carried out as per industry best practice.

Critically Assessment

• Asset priority

Condition assessment standards, procedures,

training, methods and policies will be documented.

Engineering judgment shall be used where necessary to determine quantitative sampling requirements.

Condition grading scores and other relevant condition information will be stored in corporate AM systems.

A commitment was made to recognise the value of condition assessment, but also it was important to acknowledge that there is a degree of subjectivity and uncertainty involved. It was necessary to ensure that staff would be trained and become competent in condition assessment and provided with additional specialist support when required. DCC would use in-house

resources to collect and assess condition data wherever possible, however specialist advice or analysis would be sought where necessary.

The pipe criticality rating was based on hydraulic significance (number of customers affected) and others consequences of failure (including location relative proximity to important/sensitive sites etc). The figure below shows how asset criticality was then used to plan the condition assessment programme.

Essentially all “lifeline assets” (Criticality 5) would have an individual plan developed. Key Assets (Criticality 3 and 4) would have at least one sample taken per asset, and non-critical (Criticality 0-2) assets would be assessed using the maintenance history, opportunistic sampling and a “cohort” approach.

It was acknowledged that collecting this volume of data would have to occur over a period of time and a forward programme was established.

For “lifeline” assets a system approach was adopted. For each of the 6 major systems (major water supply systems, the foul sewer trunk system etc) a “Critical Asset Management Plan” is being developed, focussing in on specific performance issues, asset condition knowledge, sampling plan and the long-term operating and renewals strategy. The first of these has been developed and Council targets creating one new plan each year until complete, followed by a 5-yearly review.

Courtesy of Dunedin City Councll

Crltlcallty s: Ufellne assets

Crltlcallty 3 -4: Key assets Crltlcallty o – 2:

Network I plant I clvll

Crltlcallty o – 2

… Existing maintenance history?

If the asset Is Crltlcallty s. a separate plan will be developed specific to that lndlvldual asset. This plan shall Include condition assessment requirements as well as detail for how the results will be used to manage the asset. At least 1 sample per critical area

Use engineer!! Judgement to determine additional number of

samples. btsed on risk

Choose additional sample I assessment PQlnts

… Results recorded

Capital programming

amt Use cohort approach

Are there eno!h data points II from critical or existing assets? T

D f Use engineering Choose additional points based Judgement to

on highest probability of poor …– determine number condition of samples

2182 IIMM International Infrastructure Management Manual 2015

If the asset Is Crltlcallty 3 – 4, at least one sample per critical area will be collected and assessed; subject to engineering Judgement. If the asset Is a Crltlcallty o – 2. representative samples will be collected from pre-determined cohorts and assessed to extrapolate condition grades to the assets In these cohorts.



Intermediate data collection. Increased risk and/ or criticality. Important

to network operation. moderate condition.

operating within performance limits.

Core data collection. Lower r1sk and criticality, limited

Advanced asset management.

Impact on network operation. bridge operating well within performance capabilities. good condition. Core asset management.

Advanced data collection. High r1sk or criticality, very important to network operation, poor condition, close to performance capabilities. Advanced asset management.

Bridge criticality

Figure 2.5.3: A Risk and Crltlcallty Data Collection framework (Ref2123)

Basic and Core Monitoring Techniques

Basic and core approaches will typically use visual inspections on the identified sample sets. These inspections will provide a general overview of the current state of the asset, its performance and what is required to maintain the asset in order to deliver levels of service. Basic and core monitoring approaches might be used when <Ret2124J, the organisation is small and where that level of data is suitable for the decision making frameworks used by the organisation.

In situations where there are a large numbers of assets or costs are being minimised, a sampling approach may be used. However, in situations where sampling is used, a

:;; CII ‘O “‘ u Oil

“‘ cc

c ,a – Ul


EoE � ., o.:!::! CII c ., ‘t ct; 0. CII O >, u Vl c Q. � II) 1: Vl ” Vl e 00 “‘ .9 c c (.? .::

E TI :,; 00 “‘ r. c ;!l Q) c 0. .::

0 c: E, I!’ .8 E

0 00 c � 00 0 0 w c E :::; -;;; > “‘ w ., !!? (l) r. (l) _J u r. 00 <:: 3 ‘ >, I- c: :Jl c I!’ Q) 0 TI ” 0 � c: Q) “‘ :::,

� Q) “‘

Asset 0. “‘ Q) � w

E :::, u

i :, t JJ ii= 0 Groups > � 0 w .s I 0.

Dynamic Assets

Mechanical Plant ,/ ,/ ,/ ,/ ,/ ,/ ,/

Electrical Plant x ,/ ,/ x ,/ ,/ ,/

Instruments x ,/ x x ,/ ,/ ,/

Control Systems x ,/ x x ,/ ,/ ,/ (electronic)

Passive Assets

Roads x x x x ,/ x x

Reservoirs/Dams x x ,c x ,/ x x


Concrete I Steel ,/ ,c x x ,/ x ,c

Earthern ,c ,c ,c x ,/ x x

Buildings ,c x ,c x ,c ,c ,/


Gravity Conduits ,c ,c ,c ,/ ,/ ,c ,c

Pressure Conduits x x x ,/ ,/ ,/ ,c

“” .s

E “‘ V)

!!? 0 u ‘O c “‘ “” c: ti � c 0 :;:; .;; 0 0.

E 0 u w Q)

vi ‘ 2l !!? u c: 0 u












>, I!’ x













I: Poles x x x ,/1,Cl,C • I· Wires x ,c ,c ,c x x x x x Denotes Not Applicable ,/

more sophisticated asset performance assessment system is recommended if sampling shows assets to be performing poorly, particularly if this relates to condition of the asset. Sample selection is covered in Section 2.5.3.

Intermediate Monitoring Techniques

Intermediate approaches will employ more advanced non-destructive testing techniques in combination with visual inspections. This will improve the accuracy of the data and lower the level of uncertainty in the decision making process. Typically more advanced techniques will be used:

when visual inspections provide only limited insights into potential failure mechanisms (e.g. fatigue, weather tightness);

during the development of specific mitigation projects;

for advanced programming;

in deterioration model development; and

for assets that have an increased criticality or failure risk.

Advanced Monitoring Techniques

An advanced performance data collection programme will incorporate visual inspections, non-destructive testing and asset health monitoring methods. Advanced systems may allow the performance to be assessed on many different parameters.

c 0 :;:; 0 � 2l

s e 0. (!!

c:’ 0. :;; Q. E l1 “‘ c: c 0 Q) � :;:; 0 E c Q) �,: g Q) � 0

E Vl Q)

B Vl

!!? Vl -;;; <t :, :::; .2 Vl r. :, -;;; � � Q) IQ c: :, Vl � “‘ !!l Vl ., 0 c: ,: :::; > Q) t; �

g ,: c � :c -�— -&, Q) I- 0 t; 00 B Q) Vl :, � c § � :, “‘ c 0 0 “” w r. >, .Q “‘ c Q) c 00 Vl �

V) 0 � ., 0 B � c: ‘O 00 c: u .:, c: 0 00 0 s(l) ti 0:: ‘5 “‘ Q) V) “‘ g � :::; c: µ c: u 0. ii= µ ., �

€ J c: 0 � (l)

“‘ t_

c: ·.;; .s

� Q) fi 0. E ., Q) � � x E > “‘

� E w -;;; w “‘ I- 0 0. “” ]i ‘” � ., TI u !!? E “‘ “‘ � “‘ > :i: Q) :;,: > j &u 0. V) u _J .s 0. � V)

,/ ,/ x ,/ ,/ x x ,/ ,/ ,/ ,/ x x x

x x x ,/ x x x ,/ ,/ ,/ x x x x

x x x x x x x ,/ ,/ ,/ x x x x

x x x x x x x ,/ ,/ ,/ x x x x

x x x ,/ x ,c ,c x ,/ ,/ ,/ ,/ ,/ ,/

,c ,c ,c ,/ ,/ ,c x ,c ,/ ,/ ,/ x ,c ,c

,c x ,c ,/ x ,c ,c ,c ,/ ,/ ,/ x ,c ,c

x ,c ,c ,/ x ,c ,c ,c ,/ x ,/ x ,c ,c

,c ,c x ,/ x x ,/ ,c ,/ ,/ ,/ ,/ ,c ,c

,/ ,c ,/ ,/ ,/ ,/ x x ,/ ,/ ,/ x ,c ,c

,/ ,/ x ,/ ,/ ,/ ,/ ,/ ,/ ,/ ,/ ,c ,c x

x x ,c ,c ,c ,c ·1·1,/ ,/ ,/1·1· x x x ,c ,/ ,c ,c x ,/ ,/ ,/ ,/ x ,c x

Denotes Applicable

Figure 2.5.4: condition and Performance Assessment Techniques for a Range of Asset Types

2178 IIMM International Infrastructure Management Manual 2015




2.5.5 Predicting Future Condition and Performance

ISO 55001 Cl 10.2 requires that: The organisation shall establish processes to proactively identify potential failures

in asset performance and evaluate the need for preventive


Condition and performance data can be used to model what the performance level might be at some future time. Models are beneficial for a number of reasons, as they;

assist in long-term planning by providing insight into the long-term performance of the asset;

provide insight into the effects that maintenance actions and different investment scenarios have on the performance of the asset;

provide the building blocks for budget optimisation; and

lead to dialogue between budget setters and operational managers managing the budget.

There are two methods commonly used to model changes in long-term asset performance. The first uses equations to represent the deterioration of asset performance with time and is known as deterministic modelling. The second approach uses more complex probabilistic techniques, but more readily accounts for uncertainty in the decision making process, as a greater level of flexibility is provided by the modelling process. This method is called stochastic modelling.

Deterministic Models

Figure 2.5.5 Illustrates a typical deterministic deterioration profile, which can be represented as an equation or set of equations. T his type of deterioration curve can be developed from first principles when there is sufficient knowledge of how materials respond to the service environment or can be based on data or expert judgement. T hese curves take the viewpoint that no maintenance has been applied. To develop the curve based on data the following process is used:

filter the data to include assets where no past maintenance has occurred;

plot the condition against time for the asset being considered; and

use linear regression to identify the best fit to the data.

f O O r-�����- u Ill

� 200 c ns

� 400 Capacity Limit :


i 600 Minimum Level of Service : :

+i •••••••••••···—————·-···r··-�–



1000�������������������\.._.� 0% 15% 30% 450/o 60% 800/o 90%

Percentage Effective Life Elapsed

Figure 2.5.5: A deterministic asset decay profile

2190 IIMM International Infrastructure Management Manual 2015

If knowledge of the asset’s current state is limited the inputs should be varied between known limits in the sensitivity check. This will provide insight into the range of future outcomes and will help in the development of appropriate management actions.

Case study 2.39 illustrates how deterministic models can be used in AM.

Stochastic Models

In a deterministic model all assets are assumed to be in a given condition at a given age. In reality this will not be the case, as assets can potentially be in any condition at any age. To model this uncertainty stochastic models are used. One method, known as a Markov Chain, defines the probability that an asset or component will stay in a given condition or will change to the next condition state. These probabilities being defined for each of the condition states being used in the rating system.

It is recommended that advice is sought when using stochastic methods, as these methods have specific restrictions. Furthermore, developing optimised programmes is more difficult with stochastic models and often uses a form of Monte-Carlo analysis.

Reliability theory is one use of stochastic models and central to reliability centred maintenance (RCM). RCM is most often used in mechanical and electrical assets, but also under pins the choice of safety factors used in the design and assessment of passive assets, such as roads and bridges. In the case of passive assets, advanced AM techniques can be used to reassess these factors <Rer2129>.

Model Accuracy

Models are generally based on expert judgment (past experience), or asset data. Using past experience is appropriate where the asset and the environment the asset operates within is relatively similar. However, in some cases technology changes and alters the assets performance or the environment the asset operates in changes over time e.g. climate change. As an example, composite materials behave differently to steel and concrete, but in some cases insufficient service data may be available to predict their future condition. In these cases the future performance of the asset cannot solely be based on past experience, as this experience will provide a less accurate prediction. This is not to say models should not be used in these cases, as these models provide important insights into the relative impacts of changing management strategies. However, they will not be able to be used for prediction.

In cases where little data is available or where the system or asset has changed over time incremental updating is appropriate. To use this method expert judgement, or other data, such as manufacturer’s data, is used in place of data derived from a monitoring programme. As time passes this data is more slowly supplanted by the data collected as part of the monitoring programme and so the data starts to more accurately reflect the actual performance of the asset. As more data is collected the increased knowledge leads to improvements in the model’s accuracy.