Understanding the rate that the UK’s sewers are deteriorating has, hitherto, been the subject of much expert judgement, strong opinions and statistical modelling based on poor and insufficient data. Here, Matt Wheeldon looks at how quickly the assets are actually deteriorating to find the best approach to fix the situation for the future.

Wessex Water, which is leading the industry in proactive investigation to identify structural problems and invest in them before they collapse, has been ‘consequence of asset’ driven rather than ‘risk’ driven.

This article summarises Wessex Water’s approach to addressing asset deterioration by:

• Quantifying the rate of sewer deterioration through the utilisation of time-spaced observed data.
• Optimising investigation and subsequent investment by assessing each sewer length using a risk-based approach, supported by asset, operational and geospatial data analysis.

Sewers form a network of vital infrastructure in any developed country where public health and environmental protection are considered an important aspect of living conditions in the 21st century.

The length of sewers owned by the ten water and sewerage companies in England and Wales amounts to 325,000 km. Last year, the industry renewal (replacement and rehabilitation) rate was 0.07 per cent of the total length sewer network. At this rate of investment, sewers would be expected to have a lifespan of 1,400 years.

The key question facing the industry is whether this low level of investment is sustainable or whether we are storing up significant problems for future generations to address?

The state of the UK’s sewer network

The structural condition of sewers can be summarised by using Structural Condition Grades (SCG) 1 to 5 (five being the worst). Most sewers in the UK have not been visually observed since the day they were laid. This is because the majority of sewers do not change condition from one decade to another – the pace at which they move through the SCG is, for most sewers, exceptionally slow. On top of this, reported national failure rates continue to be low – around ten collapses per 1,000 km of pipe network per year .

So is the current level of investment adequate? After all, this is a slowly changing asset group – it is highly unlikely that any significant change will be observed in the next few years – so there doesn’t appear to be any ‘burning bridge’ that needs addressing. Or are we missing the point? Is there a massive underlying issue that has yet to visibly materialise – that the general condition profile of the nation's sewers is deteriorating (and therefore risk is steadily increasing), but no-one has quantified the scale of the problem?

Sewer deterioration rate modelling

Time-spaced observed data: The information that has prevented an understanding of how this ageing asset base is deteriorating can only be obtained by recording the movement of sewers through the structural condition grades. Obtaining this data for these buried assets can only reliably be obtained from CCTV surveys repeated over time.

Part of Wessex Water’s commitment to understanding the rate of deterioration of its sewers has been to collect CCTV data over many years and in a consistent format. To date, this extensive dataset contains repeated CCTV surveys on over 10,000 sewer lengths (approximately 600 km or 4 per cent of the company’s total sewer length).

Transitional probabilities between SCG: The data has been analysed to investigate how sewers have deteriorated over time by observing how SCGs have changed in the time between the surveys. The figure and table below show the observed movement of SCGs from CCTV surveys carried out 15 years apart.

The graph and table show the change over time of the structural condition of the sewers. As expected, the largest proportion of sewers showed no change in the 15 year interval between surveys.

Such matrices, combined with statistical analysis methods have enabled sewer deterioration to be visibly observed and recorded and rates of change measured. However, like any model, these are only truly valid for the observed part of the sewer network and for the time interval considered. Therefore, Wessex did not consider it a reliable tool to measure the overall sewer deterioration rate for the whole asset base.

Using separate datasets to substantiate the deterioration model: To overcome the uncertainties described above, the approach used two probability matrices derived from CCTV surveys – one from observed data ten years apart, and the other 15 years apart. The datasets were also from different parts of the sewer network across Wessex Water’s region.

The two datasets were used to calculate two 30-year deterioration estimates (2x15 and 3x10) that were then compared by performing a Z-value test. The results showed a 90 per cent statistical significance level, indicating that the observed deterioration rates in the two datasets were statistically consistent.

Using this sewer deterioration model based on observed data, it becomes possible to predict the rate of decline of sewer structural condition, and therefore predict what would happen to the condition grades of the sewers if varying levels of investment are carried out.

Conclusions: sewer deterioration rates

Based on the observed deterioration rates, the model shows that, for the next few years, approximately 2 per cent of Wessex Water’s sewers will deteriorate into a Grade 5 over five years. This equates to approximately 70 km per year.

From analysis of historical and current rates of renewal, it is clear that the level of investment by the UK water industry is not keeping pace with underlying sewer deterioration rates. In effect, until renewal rates equate or exceed deterioration rates, it is logical to assume that the long-term stability of these assets is by no means secure and risk is increasing. However, with these unseen assets still performing satisfactorily nationally, alongside a constant pressure on sewerage bills, it is unlikely that there is a substantial enough impetus to do anything materially different until asset performance begins to significantly impact customer and environmental service levels.

Optimising investigation using a risk-based approach

Understanding how quickly sewer assets are deteriorating is crucial for understanding the quantum of investment required, but to ensure investment is optimised, regardless of expenditure levels, it is crucial to know where investment is most needed. This is not straightforward for a large unseen asset group for which comprehensive surveying is prohibitively expensive.

Historically, a lack of knowledge of the ‘likelihood’ of asset failure has led to a ‘consequence only’ approach, i.e. investigate those sewers where the consequence of failure would be high – these sewers were categorised as ‘critical’ sewers. Proactive CCTV of ‘critical’ sewers meant that 75–80 per cent of sewers were left to a ‘reactive only’ approach.

The approach adopted by Wessex Water for targeting investigation is a risk-based approach, comprising a set of ‘likelihood’ and ‘consequence’ models built up from all available asset and operational data and information, updated regularly and analysed geospatially.

The result is a current risk score for each individual sewer length, which directs investigation more efficiently towards higher risk sewers than the previous consequence driven approach.

Building the ‘likelihood’ model

Factors affecting the likelihood of failure of a sewer can be a complex process. The likelihood of asset failure is closely linked to the SCG, which can only be obtained from CCTV data. The model uses the SCG as the primary indicator, but since a comprehensive dataset of this information does not exist and is prohibitively costly to obtain (only about 25 per cent of sewers have ever been surveyed), the model uses supplementary indicators – operational and sewer characteristics – which also contribute to the probability of each sewer length failing.

The model uses a dynamic weighting to allow for the real issue of not having comprehensive coverage of all desired asset information.

Density analysis can be performed and presented geospatially for operational and asset failure information. This information contributes to the overall likelihood score.

Building the ‘consequence’ model

Identifying, assessing and scoring the consequence of a sewer failure utilises information abstracted from Wessex Water’s corporate Geographical Information System.

Using a similar process to the likelihood model, a consequence score is calculated for each sewer length.

Visualisation of risk

The product from the likelihood and consequence models becomes an individual sewer length risk score. This can be represented on a risk matrix or histogram for the whole region or any geographical area.

Sewer lengths, towards the top right corner of the matrix have a higher risk than those towards the bottom left corner. The histogram provides a useful view of the sewer risk profile that can then be analysed over time to convey longer term trends in levels of risk associated with this asset group.

As asset and operational data is collected every day, the information added to the models increases. Each month CCTV data is uploaded and the complete risk model is re-run every six months.

Using risk scores to target investigation

The sewer risk scores can also be portrayed geospatially (Figure 9) to enable CCTV surveys to be targeted and grouped efficiently.

Previously, only 10–15 per cent of the length of all sewers surveyed would identify structural issues. The new risk-based approach has increased this to around 60 per cent of surveyed lengths.

The sewer risk model also provides sufficient flexibility to allow add-on models for specific needs and requirements. For example, for the ‘operational hotspot model’ – operational factors like Service Condition Grade and historical service failures such as flooding due to blockages are given higher weightings. These hotspots help steer the operational CCTV and jetting program.

Conclusions

Sewer deterioration rates:

• A better understanding of the rate of deterioration of long-life assets, such as sewers, has only been possible through a commitment to consistent data gathering over a long-time period.
• Evidence shows that investment rates are insufficient to match the rate of deterioration of the nation’s vital sewerage infrastructure and risk is increasing.
• Current asset performance and pressure on sewerage bills are not sufficient to drive a step change in levels of investment for this vital national infrastructure.

Sewer risk scoring:

• Optimising investigation and investment can be improved by adopting a risk based approach as opposed to a consequence driven approach to surveying. The probability of identifying sewer maintenance needs when using a risk based approach are significantly higher than a consequence based approach.
• Incomplete datasets, which add value to deriving likelihood or consequence scores, can still be used using a dynamic weighting approach to scoring and shouldn’t be discounted.
• The sewer risk model doesn’t just highlight asset condition risks. It can steer attention to operational performance issues by adjusting weightings as appropriate.
• Additional information from operational and asset activities can be added regularly to the databases, which drive the models enabling risk scores to be kept as ‘live’ as required.