Page 38 - Gi March 2019 - Flipbook

igem news – yppc

RISK MODELLING strategy, project delivery, and operations

teams. The criticality factors we chose to
adopt are outlined in Table 1.
Powerco does not have all the data
OF GAS ASSETS that is required to model each of these
factors, therefore assumptions were
made from current data to compensate
for missing factors. For example, pipe
pressure and diameter is used to assume
the number of downstream customers.
With these factors in place, I applied
the model to our ArcGIS system using
By Michael Tong, Gas Asset Intelligence Analyst at Powerco ModelBuilder. This program shows how
and Young Persons Paper Competition (YPPC) 2018 finalist the model is applied in small, distinct
and easy to update steps. The entire
ASSESSING AND MANAGING risk a valve could have a 10 per cent model can be reapplied to all assets at
is a key aspect of all businesses and likelihood of failure in the next year. the press of a button. This tool makes
the importance of risk management This leaking valve could cause many the criticality model easy for new users
is further elevated when a business’s impacts, from a minor leak (requiring to reconfigure and run in the future,
assets are transporting potentially a component tightening at 95 per cent no matter their experience with ArcGIS
hazardous fuel through public streets probability and low cost) to a major leak geospatial tools. This tool also made it
to supply thousands of customers. Prior (that ignites and injures a pedestrian easy to adjust the weightings of factors
to this project, Powerco undertook risk at 0.001 per cent probability with very until all participating parties were in
assessments primarily through regular high financial, safety and reputational agreement with the outputs. As an
workshops, assessing specific events on costs). Here, likelihood is 10 per cent example, our current model for mains
individual assets. and criticality is low since we consider and services is displayed in Table 2.
Actuarial science is the discipline the average impact as being low.
that applies mathematical and Generally, risk models are split
statistical methods to assess risk Criticality into consequence (the impact of
in insurance, finance and other Prior to implementation of these risk
industries and professions. One of models, the prioritisation of projects an event) and likelihood (chance
the key concepts of actuarial science was somewhat subjective, with some of that event occurring). This
is that of pooling risk, whereby the projects being considered highly critical framework is adequate when
risks of similar assets are assessed as a by some and not critical by others. assessing specific scenarios,
group, rather than individually. Powerco needed an objective and
This project was established to apply repeatable criticality model to enable but having all chance pooled
actuarial reasoning and methodologies consistent comparison of projects. into one measure makes it
to assess risk on our gas distribution My idea was to take an actuarial, cumbersome to assess an
assets. With this, we can compare factor-based approach to define and
the probable yearly events against capture criticality across all assets, asset’s risk factor
Powerco’s total acceptance of failures. rather than an engineering approach of
The aim was to develop models to defining all possible consequences and
provide comprehensive preliminary risk the probabilities of these consequences Likelihood
assessments on all assets to complement for each asset type. Development and Next, we needed a measure of our asset
our current in depth assessments. review of this model is simpler and failure curves to quantify how likely our
more straightforward, requiring fewer assets are to fail each year. Polyethylene
Building the models changes than a more complex model. pipes make up over 90 per cent of
Generally, risk models are split into Being simpler, the model has the Powerco’s live network. No reliable
consequence (the impact of an event) advantage of being more likely to be long-term failure curves of polyethylene
and likelihood (chance of that event kept up-to-date. were readily available, so I built the
occurring). This framework is adequate To define the methodology and the asset failure curves from Powerco data
when assessing specific scenarios, factors to create this criticality model, I alone. Fortunately, Powerco implements
but having all chance pooled into hosted a series of workshops involving comprehensive data collection
one measure makes it cumbersome all relevant internal parties from asset methods, which meant I was able to
to assess an asset’s risk factor. This
is because an asset failure could give FIGURE 1 RISK FRAMEWORK
rise to multiple different events with
differing chances of occurrence.
In order to develop an asset specific
risk model, alteration of the framework
was required to separate out the
likelihood of failure from the probability
of each consequence. Criticality is
then developed as the pooled value
of impacts by chance of the possible
events if an asset fails. For example,

38

IGEMNews_YPPC.indd 1 14/02/2019 14:27