Register for free and continue reading
Join our growing army of changemakers and get unlimited access to our premium content
What you are about to read is a case study application of risk-based prioritisation of sewer maintenance. It addresses the issue of why sewer maintenance planning should be risk based and how this should be linked to strategic investment planning. A proof of concept was demonstrated using data from Dwr Cymru/Welsh Water (DCWW).
A concise method – Investment, Planning and Delivery (IPaD) matrix – for structuring the use of data and their linkages with asset service and performance is also presented.
The results from this study clearly show how maintenance planners could potentially target maintenance in a way that is consistent with UKWIR’s Capital Maintenance Planning – A Common Framework. It is proactive rather than reactive in the way that it looks to pre-empt future performance problems that are linked to service.
To allow the approach to have further-reaching benefits, a more in-depth, robust regression-style analysis is needed, to deal with:
In addition, to fully complete the risk assessment, the inclusion of a measure of severity of consequence is needed.
Background
Why do we need to invest in sewerage networks?
Water and sewerage companies (WASCs) have a statutory duty to maintain serviceability to customers (Ofwat letter to Managing Directors, MD161). This means that it is not enough to demonstrate that the service is being provided; they must also show that the ability of the assets in continuing to provide that service is maintained.
Companies can choose to invest in their assets either reactively (as the assets fail) or proactively (to prevent future asset failure). To do this requires an understanding of how assets deteriorate in condition over time, how this could give rise to failures in asset performance but, more importantly, how this change in performance affects the ability of the asset to serve customers. There are stakeholders other than customers that are affected by the way companies invest in their assets. These include: company shareholders or financiers, the environment, staff, and the general public.
To operate successfully, companies must balance the needs of all the stakeholders. This culminates in the overarching need to manage the assets in a sustainable way that ensures social and economic stability within the corporate and legal framework.
Why should investment planning be risk-based?
The phrase ‘maintain sustainability’ indicates that companies need to understand how the ability of assets to serve customers might change with time. Unfortunately, the future cannot be predicted with certainty and we must therefore assess the likelihood of failures and their associated consequences. In essence, the two parameters of likelihood and associated consequence give us a measure of expected consequence or what the industry frequently terms ‘risk’.
The element of investment planning that relates to understanding how the future is different from what we have observed from the past allows us to demonstrate changes in our maintenance investment needs to the Regulator. In 2002, the industry adopted an approach known as the UKWIR Common Framework – Maintenance Investment Planning (CF). The CF addresses the need for companies to be forward-looking in their plans and explicitly includes a forecasting stage to estimate the likelihood of asset failures and their effect on service and costs. This framework has been used by Ofwat to assess the robustness of companies’ PR04 capital maintenance investment plans.
Companies which failed to adhere to the CF were to varying degrees penalised by Ofwat, in the form of cuts to those companies’ proposed plans. As we now embark on AMP4, companies will be required to demonstrate that they can deliver what they have promised for the investment agreed with Ofwat.
This gives us two issues to deal with during this AMP period:
The simple answer is to embed the principles of the CF in the investment delivery process. In doing so, capital and operational solutions delivered in AMP4 will tie in with the principles adopted in the PR04 planning process, but, more importantly, there will be a consistent set of data and analysis tools that are essential for a less painful and hopefully more successful investment plan for PR09. Embedding the CF into ‘business as usual’ processes during AMP4 is easier said than done, though. The CF concepts and processes need to be assimilated by all internal stakeholders to the investment process. The tools and data to support the processes may not always be available and would therefore need to be developed and collected.
Over the next few pages I shall present an approach which was tested at DCWW for prioritising sewer capital maintenance in AMP4. It follows from the original study to develop the PR04 submission for sewer capital maintenance using a strategic investment planning tool called GAasset. A brief overview is given of the linkages between sewer investment prioritisation and strategic business planning. This is followed by an outline of work carried out for DCWW as an initial proof of concept with suggestions for future improvements.
Investment delivery – prioritising sewers
Level of application
Following the Final Determination from Ofwat for PR04, investment budgets to varying degrees of detail have been identified for each asset group. In general, companies have identified capital maintenance budgets for groups of catchments and need annual plans of how and where to concentrate their efforts first. Many companies have already pulled together their year 1 and year 2 plans. It is essential that companies develop these plans in line with the CF principles to avoid investing inappropriately and risking non-delivery of promised performance and levels of service.
This work concentrates on the initial prioritisation of sewer catchments for capital maintenance consideration by targeting those catchments where risk reduction is maximised most cost-effectively. It aims to provide a ranked
list of catchments for further detailed investigations and scheme definition.
This kind of top-down approach is absolutely crucial if company-level service targets are to be achieved within the resources that have been made available in the wake of PR04.
Criteria for good investment prioritisation
The process for implementing the CF is illustrated in Figure 1. The key aspects of these are:
– historic analysis – to understand the review maintenance expenditure and historic levels of service,
– forward-looking analysis – to forecast future service by predicting future service or risk of failure; and assessing the impact of maintenance on forecast service or risk of failure,
– conclusions – to explain any differences in the level of maintenance between historic spend and forecast spend, and to identify any opportunities for further efficiencies.
By being CF-compliant, companies will be able to
manage their investment programmes to deliver what has been promised in the company business plans. Furthermore, it is widely believed that the CF is here to stay. Therefore, collection and analysis of data during AMP4 that is in line with the principles of the CF will significantly support preparation for PR09 and beyond.
Maintenance agreed with Ofwat is generally based on agreed levels of service and investment to maintain the ability of assets to serve customers and the environment. When prioritising catchments for investment, maintenance managers must make sure the approach is consistent with the company business plan; thus it must be service-driven.
This does not, of course, mean that companies should concentrate only on areas where service has been observed historically to have significant failures. Instead, companies need to understand how asset performance, e.g. collapses, is linked to service, e.g. flooding. In such a way might they determine precisely what needs to be tackled in order to manage service.
Where companies ask for capital maintenance that exceeds historic expenditure, they are required to explain why the future requirement is predicted to be different from the past. A robust analysis would be based on sewer performance predictive or deterioration models linked to service and performance of the assets.
The predictive models can be used to give two types of measures:
expected number of performance and service failures; and probability of performance and service failures. Both are measures of likelihood and can be used interchangeably. For example, if the expected number of flooding incidents in a catchment in any one year is predicted to be 2, the equivalent probability of failure is 2 in 365 (or 0.0055).
To calculate risk of failure, the severity (or consequence) of the service failure must be ascertained. For example, given that a service failure (flooding incident) has occurred, the severity of the incident depends on the extent of the consequence. This could be related to the number of properties affected and whether there has been damage to internal or external property. A common currency for calculating the extent of consequences across all asset types is helpful, and should ideally be monetised. This allows companies to balance expenditure priorities more objectively. For example, the direct and indirect costs associated with internal and external flooding of properties is a consequential cost and can be combined with likelihood of flooding (service failure) to give risk of failure. In the absence of monetised consequence, a structured approach to assessing the extent of consequences using scores will suffice, though this cannot immediately be used to compare one asset type against another.
The priority of maintenance should be targeted at those catchments which offer the highest return for every £ of investment, ie what
DCWW call ‘the most bangs for bucks’! The measure of ‘bangs’ is
inextricably linked to service and the risk reduction associated with the expenditure.
The primary measure for cost-effectiveness is risk-reduction per £, a measure that reflects the level of benefit generated for every £. By ranking the highest values as top priority, companies can concentrate on achieving as much cost-effectiveness early on in their investment programmes. However, sense-checking of this list is important, as certain schemes which are necessary for achieving regulatory and company service targets may require significant lead-in times.
Capital maintenance is only one of the investment activities associated with sewerage. Other investment streams include flooding due to hydraulic overload (enhancement) and CSO schemes (quality).
By identifying areas of overlap for capital maintenance and other investment schemes, potential efficiencies can be gained. Typically, this might be where maintenance can be carried out at the same time as a CSO or flooding scheme offering maintenance benefits at a lower unit cost of intervention. The overlap of schemes is an important consideration as Ofwat would have taken this into account when setting efficiency targets.
The benefits of linking the investment programme for delivery with the planning process are far-reaching. Ultimately, this could mean PR09 ‘at the press of a button’. The other key benefits are:
– consistent purpose for investment to avoid inappropriate targeting of funds,
– alignment with CF, which is considered to be best practice for capital maintenance planning,
– continual updating and learning from ongoing analysis of data that is common to delivery and planning.
Figure 2 shows a framework for linking the planning process to the delivery process; this is a concept known as IPaD and developed by the Ewan Group.
The blue box – Strategic Planning Model (GAasset) – represents the model used to develop the strategic business plan. This gives the high-level budgets for groups of catchments, together with the expected performance targets and associated levels of maintenance activity. To determine these values requires the development of key performance indicator (KPI) models which are used to predict the change in performance over time of the grouped catchments under the following scenarios:
-no intervention or capital maintenance – i.e. ‘do-nothing’,
– specified mix of interventions or capital maintenance.
The GAasset strategic planning model uses the KPI models together with (proactive) intervention and reactive costs to determine the most cost-effective (optimal) plan over a 15-year planning horizon.
Once the high-level budgets and performance targets for the groups of catchments have been determined, a sewer prioritisation exercise is necessary to establish a programme for targeting the investment at the catchment level.
The prioritisation exercise also depends on KPI models. These should be built up from detailed analysis of asset attributes and their relationship with asset service and performance. Ideally, to avoid misalignment of the ‘top-down’ strategic model outputs with the expected performance from the ‘bottom-up’ sewer investment prioritisation, the KPI models should be the same.
Having prioritised the catchments within their strategic level groups to identify the high-priority catchments, the next step would be to determine indicative levels of intervention for each catchment by drilling down to sewer level and examining the predicted likelihood for asset performance problems. This analysis, combined with the high-level budgets and performance targets, can then be used to define a first-cut investment programme.
The intention of the first-cut plan is to target detailed investigations to obtain a better understanding of the potential problems in the high-priority catchments and sewers, thereby providing more accurate costs and estimates of impact of interventions. As knowledge is improved, it is essential to re-run the strategic model with updated information to check that the overall company targets can still be satisfied. This feedback of information should continue throughout the delivery stage through purposeful and focused monitoring of the investments and their beneficial impacts to ensure that the planning models are reflective of latest knowledge.
Application to DCWW – proof of concept
Separating performance and service measures
To justify their investments to Ofwat, companies have to demonstrate that service is at risk should the investment not be available. The driver for investment is therefore asset service (e.g. flooding) and not asset performance (e.g. collapses). However, using service alone to prioritise investment only helps us to identify where to invest. To know how to invest, we need to understand the underlying failure mode or asset performance failure that has given rise to the service failure. Taking this further, if we can forecast any likely changes in asset performance failure by looking at why these failures occur and linking them to explanatory factors, we can determine the future risk to asset service.
Figure 3 illustrates the concept of separating asset service, asset performance and explanatory factors. An example of the linkages between service, performance and explanatory factors are shown in the IPaD matrix (Figure 4). The IPaD matrix is a highly visual and simple means for structuring the data required for modelling asset service and performance by clearly establishing causal links.
This matrix has been used to structure the risk-based modelling approach for prioritising maintenance of DCWW sewers. The process for the modelling approach is shown in Figure 5.
The green box (Problem catchments and sewers incidents) identifies where service problems have been observed in the past. This was derived by combining historic service incidents for each sewer using weightings based on the Ofwat OPA scoring method to give an overall service score. The catchment scores were then generated by aggregating sewer scores and were ranked to identify the worst catchments. However, this only shows where, historically, service problems have been observed.
The following step involved the identification of those catchments that also had a high likelihood of asset performance failure. This was based on an analysis of historic performance failures for blockages and collapses and generating regression relationships with sewer attributes in order to calculate, for the whole asset stock, the likelihood or expected number of collapses and blockages.
The change in likelihood of collapses and blockages occurring in the future were calculated using the regression relationships by simply increasing the age (asset attribute) by five years. This gave an indication of the rate at which sewers were deteriorating and helped identify catchments that would present a problem in the near future.
Finally, additional parameters were generated to aid the maintenance planning process, including the length of sewers that exhibited high likelihood of collapses and blockages together with the coincidence with other investment schemes, i.e. CSOs and flooding HO. The former gives an indication of the extent of the problem; the latter identifies areas that could benefit from additional efficiencies through enhancement schemes or combining maintenance and quality.
Generating relationships for blockages and collapses
A simple multi-linear regression approach was adopted for this initial proof-of-concept study, although it is believed that other modelling approaches more suited to the behaviour of sewer performance should eventually be applied.
Figure 6 shows the factors which, according to the regression study, had a statistically significant effect on blockages and collapses. Although the relationships were not strong, it is believed that this was due to the following factors:
– limited data cleansing (within the time available),
– non-linearity of the relationships,
– lack of environmental-type data eg soil, traffic, groundwater,
– normalisation issues – ie the use of sewer lengths to determine appropriate rates of performance problems did not always give a true reflection of the performance of the sewer.
Illustrative results
Figure 7 shows illustrative results based on the regression relationships generated in this proof-of-concept study. The results show that the majority of historic service problem catchments (red) are most likely to be linked to sewer blockages. However, there are some service problem areas associated with high likelihood of blockages and collapses.
Maintenance planners would be able use these results to target further investigations, by concentrating on those areas with high coincidence of service problems and both blockages and collapses. The final columns in Figure 7 highlight the lengths of sewer mains that are coincident with mains within 500m of CSO and flooding HO schemes. This allows maintenance planners to look for further efficiencies when planning whole catchment solutions.
Conclusions and further study
The results from this study clearly show how maintenance planners could potentially target maintenance in a way that is consistent with the CF. It is proactive rather than reactive in the way it looks to pre-empt future performance problems that are linked to service.
To allow the approach to have further-reaching benefits, a more in-depth and robust regression-style analysis needs to be carried out. This needs to allow for:
– the non-linear relationship between explanatory factors and asset performance,
– the data issues encountered during the course of this study,
– and the linkage between service and performance to be established mathematically.
In addition, to fully complete the risk assessment, the inclusion of a measure of severity of consequence is needed. The current proof of concept shows only the visual linkage of asset performance to asset service. It is strongly recommended that consequence measures be developed in monetary values. This will enable risk measure to be monetised and hence it will be possible to make direct comparisons between sewers in different geographic locations and between different asset groups.
Eventually, the approach should include data relating to the typical costs of dealing with the likely asset performance problems and the associated benefits. This allows an assessment of the cost-effectiveness of schemes, ensuring that companies can additionally target on the basis of ‘most bangs for bucks’.
Please login or Register to leave a comment.