The Process
Some elements of the ORDM process require individual assessment, technical input and economic evaluation but many can be automated, for instance:
The direct maintenance costs for both planned and unplanned activities can be easily recorded in most computerized maintenance management systems (CMMS).
Unplanned maintenance is recorded as a failure history and the direct costs of repair can be easily recorded and identified against this individual asset in terms of:
- Labor
- Spare parts
- Materials
- Specialized equipment
- Contracted services.
The cause of failure and the activities used to overcome the failure are easily recorded and readily available for causal and activity-based analysis.
These CMMS systems can be enhanced to include allowances for the indirect or ancillary costs of failure that should be included in any decision making processes, irrespective of whether these costs have been incurred or not.
These indirect consequences of failure should be economically costed as impacts to the customers in dollar terms assessed against the:
- Number of customers effected by the level of service failure
- Percentage deviation from the contracted level of service — the degree of the failure
- Time or period for which the interruption or deviation existed.
This is applicable to failures that impact primarily on our customers and not on the community in general.
In some cases these failures will result in additional direct costs such as:
- Damage to private property
- Damage to other authority assets
- Injury and trauma claims
- Business loss claims.
These direct additional costs should also be recorded against the failure. They should be used in our assessments of the consequence of failure for predicting failure costs, as they are not always likely to be incurred.
By recording this information, organizations can easily identify the consequences of failure for all of the different causes and relative degrees of failure.
By having these direct linkages through databases, the organization can then identify business costs or risks of various causes of failure such as:
- Operator errors
- System capacity failures
- Structural condition and integrity
- Reliability based failures
- External forces/stresses (for example: wind)
- Accidents
- Other causes.
This way we can extract erroneous data to ensure that we are not making the renewal decision based on costs that are not applicable to the overall condition of the parent asset.
Examples of causal analysis
Example 1
Maintenance costs against a sewer collection system had risen to such a level that they justified the renewal of the main. When the main was renewed, the contractor noted the good condition of the main and queried the decision. An analysis of the cost of maintenance showed that the failures were in fact the decay and failure of the tapping saddles used and were not the fault of the sewer main itself. The optimized renewal strategy for this mode of failure would be to proactively replace all of the defective tapping saddles.
By completing an appropriate cause or fault analysis, using the computer to filter out the cause of failure, an appropriate decision will always be made.
Example 2
Past maintenance costs justified the replacement of a water supply pipeline. An analysis of the cause of failure showed that aggressive ground conditions were corroding the pipeline to a point where it fractured or burst and that many repairs had been made to overcome this failure. The pipeline was some 1 500 m long in a residential area. However when a more thorough analysis was completed it was found that during the original subdivision construction, two watercourses had been filled with imported material and it was in fact these small sections, amounting only to some 150 m of pipe, in which all the failures were taking place.
The maintenance management system in filtering the erroneous failure modes would have still indicated that the renewal was justified.
An automated check on the validity of this type of failure can be achieved through looking at the locations of the failures (work orders) that made up this maintenance cost. This is most effectively done using a Geographic (Spatial) Information System (GIS) to record the failure locations on long network assets.
By using these two tools the asset managers would have realized that only 150 m of pipeline needed to be replaced and that in fact it should be replaced with materials that would be more suitable to the aggressive imported ground conditions.
The key elements in this decision making process therefore become the maintenance system, its costs and causes of failure, together with the locations of the individual failures as recorded on a GIS system or alternatively recorded as locations in the unplanned maintenance work orders in the maintenance management system.
Non critical assets
This process is most effective for long-lived low consequence of failure (non critical) type assets.
These assets will suffer multiple failures before renewal can be justified as the typical direct repair cost and indirect costs are less than the renewal costs.
This is a reactive process in which we wait until the costs reach a level that justifies the renewal. In this way our management and renewal programs become heavily dependent on the performance of the assets and are reactive rather than proactive.
To improve or automate the proactive aspects for this type of asset we can use the maintenance management system to assist us. If we have a non-critical asset that is failing reasonably regularly we can establish the mean time between failures (MTBF).
Because we know the direct and indirect costs of these failures the maintenance system could produce a future picture of failure costs and express this as an NPV at this point in time.
When the NPV reaches a certain percentage of the replacement value of the asset, the computer could indicate that this asset is reaching the point where replacement is justified in the near future and therefore flag this asset for a restricted ORDM analysis.
This analysis would determine when the actual date is justified. In this way we have taken a substantially reactive environment and turned it into a proactive or planned renewal environment using the actual costs of failures recorded.
Macro-aggregations of these predicted costs could give the organization a clear picture of the overall size of their likely renewal program for these types of assets. This process, being bottom driven, would also enable them to identify the actual work required on a year-to-year basis, i.e. impacts on the performance of the overall systems are related to individual asset activities.
If we know the probability of failure of all the individual assets or components within the network or system then we can calculate the probability of the failure of the system itself.
Critical assets
This previous process is not applicable for critical assets where the consequences of one or two failures exceed the direct and indirect repair costs or the replacement value of the asset. In some cases one event may even justify renewal or in lesser cases only a few failures may warrant renewal.
Therefore the key factor involved with these assets is the probability of failure. The maintenance management system is only capable of indicating the mean time between failures for assets that have a failure history. In most cases these assets have no failure history and we need to consider using predictive modeling techniques where condition monitoring and the decay of assets can be related to the probability of failure.
Specialized strategic planning type software can be employed to ascertain the current (and future) business risk exposure represented by the consequences of failure multiplied by their current (and future) probability of failure.
By projecting these risk exposures out into the future we can predict the point at which the risk cost would justify the renewal of the asset. The computer could indicate this approximate date and the associated work could be programmed. Then a more thorough ORDM analysis would indicate the most appropriate strategy and the time for that work to be undertaken.
Filtering processes
There are many inputs into this ORDM analysis. Filtering processes can be used to reduce the detail or sophistication of the analysis used to suit the benefits to be gained. That is, we can use the filters to match the management inputs to suit the benefits to be derived.