Water and wastewater infrastructure managers make decisions every day that are aimed at reducing the risk of costly failures. For most, the decision process is ingrained, based on years of experience and knowledge in system management. But over time systems change, people retire, the knowledge base is lost, assets age, and the probability of costly failures increases. This is especially true in developed countries where underground utilities have been in place for over a hundred years, and the people who manage them are nearing retirement.

As a new generation of managers emerges, they are being asked to manage assets that are nearing the end of their useful life with fewer resources, and tougher regulatory requirements, such as California’s Sanitary Sewer Management Plan, or SSMP. To cope with these challenges, savvy managers are turning to computer applications with physical asset management (PAM) features to allocate limited resources more strategically. Fundamental to PAM is prioritisation of assets based on a risk model. At a minimum, this involves knowing how assets might fail and what would happen if a failure were to occur. In PAM we define ‘how assets might fail’ as the probability of failure (PoF) and ‘what would happen a failure were to occur’ as the consequence of failure (CoF). Risk is simply the product the PoF and CoF:

Risk = PoF x CoF.

Probability of failure

To determine the PoF of any asset we must first determine how it may fail in terms of failure modes. When we categorise how an asset may fail there are at least four failure modes to consider that are common to all assets.

Condition

Condition may be put in terms of a Condition Rating by quantifying the number and extents of defects, or by direct measures such as a vibration analysis. It may be helpful to measure condition in both O&M Condition and Physical Condition. O&M Condition can be addressed through tasks such as cleaning and lubrication, while Physical Condition may call for capital remedies such as overhaul and replacement. Age

For age to have meaning we must first determine the life expectancy of any asset. Life expectancy can be influenced by many factors such as the surrounding environment, construction material, and installation techniques. Although age is often a good predictor of condition, an asset that appears to be in good condition may start to deteriorate rapidly or suddenly fail as it approaches the end of its useful life. Knowing how close your assets are to the end of their life expectancy may influence how often you inspect them or how you develop a replacement strategy to avoid costly failures.

Capacity

Does the demand placed on the asset exceed its original design capacity? Influences such as population increases can certainly affect capacity. You must know what the demands are on your assets to measure capacity. Bear in mind that assets that are substantially under-utilised could also lead to a higher PoF. Level of Service

Perhaps the asset was put into place before new regulatory requirements were enacted. Stakeholder expectations for issues such as noise, odour, and safety may be more stringent now. Or it may be that newer alternatives have been developed that reduce the cost of operation to the point that it will be less costly to replace than to continue to operate. Establishing acceptable levels of service will help you make these determinations.

Your actual list of failure modes will vary depending on the asset types that you are rating, but they will all most likely fall into one of these four categories. As you develop your criteria, take into account that ‘failure’ does not always mean a catastrophic failure, but it does mean that continuing to operate the asset without taking action will be more costly than doing something about it. Quantifying probability of failure

When it comes to age, we humans inherently know that the probability of end of life increases as we grow older, and that probability increases at an accelerating rate. However, we have no way of determining precisely when the end will occur. The same is true for physical assets. But what we can do is apply a probability based on experience and historical data when available. Below is a sample table that shows how one might interpret levels of probability in a risk model.

For the failure mode of age, the graph for static assets such as pipes and manholes where failure rarely occurs early in life can be illustrated in an age based curve.

Mechanical and electrical assets are more prone to failures early in life and hence the probability of failure curve associated with these types of assets is often referred to as a ‘bathtub’ curve.

If reliable historical data is available then the PoF should be based on the percentage of failures actually experienced. Similar curves can be created for other failure modes such as capacity where the PoF may plot as a bathtub curve because an asset that operates significantly under capacity is often more likely to fail than one operating at 50 – 75 per cent capacity.

Consequences of failure

Consequences of failure are often put in terms of the cost to fix and/or recover from a failure. In this sense it would be ideal to measure all consequences in terms of actual costs, but for most it is impractical to accurately forecast the cost of all failures and therefore most systems rate the CoF on an arbitrary scale. Other traits of CoFs are that they tend to reflect the service level expected and the priorities of stakeholders. For instance, the public places a high value on the environment, as does the EPA. Therefore, a sanitary sewer overflow that spills into a natural water body would be highly consequential when one considers environmental impact, aesthetic impact, and other impacts including the cost to contain and clean the spill.

Some CoF examples include:

* Threat to employee life and health * Threat to public life and health * Environmental damage * Regulatory compliance * Disruption of service * Property damage * Cost to repair * Loss of revenue * Public relations

Generally it is more difficult to affect consequences than failure probabilities but factors such as backup and redundancy should be considered when rating them. To develop CoF ratings for the types of assets you manage you should.

Develop a list of consequences that could occur if an asset fails. Typically all assets of the same type should be assigned the same list of consequences for comparative purposes.

Rank the importance of each consequence relative to other consequences in the list. This is done in recognition that some consequences carry higher costs than others, for instance ‘life and health’ would typically be weighted higher than ‘public relations’.

Develop criteria for determining a CoF rating for each asset. For instance, if a sewer manhole is within a certain distance and upstream of a water body then the CoF rating for ‘environmental damage’ will be higher than a manhole located further away from the water body. Quantifying risk

Once you have determined thefailure modes, PoFs, consequences and CoF ratings, you can combine this information to calculate risk in a matrix for each asset. A risk matrix accounts for all of the CoFs and PoFs to calculate the risk for each asset

In the risk matrix the consequences, along with their relative priority, are listed on the left. The asset is then rated according to the potential for each consequence to occur if the asset were to fail. The CoF score is calculated by multiplying the priority by the rating and adjusted to an arbitrary scale of 10 (where 10 signifies the highest consequence). Failure probabilities are developed from measures on the asset and entered into the table for each failure mode. Each CoF score is then multiplied by each PoF to generate risk scores. The highest risk score for each consequence is highlighted in red. The highest risk value falls out of the table as the risk factor.

Developing the risk model requires this same analysis to be performed on each asset. If you are dealing with just a few assets you could perform the calculation by hand. However, if you are dealing with hundreds or thousands of assets then you should consider using a computer application to develop the model. Once the model is developed you should see patterns emerge. Planning for the future Managers have always conducted studies to gather information about their systems for decision making purposes. However, with the advent of PAM we now have a means to utilise this information qualitatively in a risk model as a basis for strategic decision making. When dealing with a large number of assets the use of computer software with PAM capabilities can help develop a risk model and assist in prioritising O&M and capital project activities. Risk assessment should be central to any asset management program as there are many tactics in PAM that use the same information pool generated from the risk model including cost/benefit analysis, triple bottom line analysis, optimised budget forecasting and reliability centered maintenance. For more information asset management