Predictive Analytics a precursor to Smart Asset Management in Utilities
Traditional Enterprise Asset Management in asset intensive utilities has grown into much wider dimension with the introduction of networked Smart Grids and installation of intelligent devices on the grid network. Utilities that are in the asset intensive business have the challenging task of maintaining and monitoring their critical assets effectively and efficiently with high availability and reliability. Sophisticated and advanced networking technologies are helping utilities to achieve this goal to a large extent as they enable gathering real time data much more effectively and accurately. Traditional asset management techniques are no longer sufficient enough to handle the additional complexities that are introduced due to the recent technological advancements. Usage of predictive analytics in asset management business can be an enabler for these Infrastructure heavy utilities to optimize their asset life cycle management costs and to effectively implement Smart Asset Management. This paper tries to look into how these utilities can effectively use Predictive analytics techniques in smart asset management area to realize some of the key benefits such as asset life cycle cost reduction and improved accuracy of decision making.
Fig 1: Asset Management Dimensions
Business Drivers for SAM
Ultimate objective of traditional or smart asset management in any utility organization is to reduce / minimize / optimize asset life cycle costs through the efficient management of one or more of the asset dimensions. Asset Life cycle encompasses through various phases like asset investment planning, asset network design, asset procurement, asset installation and commissioning, asset operation and maintenance, asset decommissioning and finally asset disposal / replacement. Optimizing the costs associated with each of these lifecycle phases is one of the key aims of asset intensive utility organizations. The real-time data coming in from the field devices along with historical incidents and failures when churned through the predictive analytics engine(s) should ideally help organizations in making the right decisions about asset failures, risk, utilization, reliability and availability.
How Predictive Analytics can help?
Modern day predictive analytics engine(s) can combine data from multiple disparate systems to enable analysis of input variables. The crux of the solution is to find the correlation between multiple input variables and the outputs that one is interested in. Predictive analytics encompasses a variety of statistical techniques (e.g. Regression models, time series models, neural networks etc.) that analyze current and historical facts to make predictions about future value of output variables. Predictive analytics engine(s) can convert large volume of business data into actionable results which are futuristic in nature. In asset intensive industries, effective decision making becomes much easier if an appropriate predictive analytics solution methodology is chosen and implemented with right kind of algorithms applied on the input data coming in from the field sensors. Some examples can be pressure variations on the gas and water pipelines impacting the bursts and leakages, load fluctuations in the electricity network impacting the transformer failures etc.
Most of the modern Predictive analytics packages extract correlation information from historical data and use this correlation data to predict the future trends and behavior patterns. Predictive analytics can be applied to any type of unknown whether it is in the past, present or future and validated with actual outcomes of the predicted variable. Correlations can be derived for critical equipment(s) for identifying suspected causes of failures before the actual failure occurs. It is important to note, however, that the accuracy and usability of results will depend greatly on the level of data granularity and the quality of input data available for analysis. A probability pattern is usually derived along with the results to improve the prediction accuracies.
Fig 2: Components of Predictive Asset Management Analytics
The important questions that arise from the above discussion are: What are the important business cases which will benefit immensely from Predictive outcomes? What future events are most critical for a utility organization? How much monetary savings can be achieved by predicting a certain future event such as a failure ? Does a cost benefit analysis favour predicting failure of each and every equipment class / category? The answers to these may vary from one organization to another organization.
Asset availability, safety and reliability are dependent a lot on accurately predicting the asset failures. Predictive asset analytics has gained a lot of importance in the recent days mainly due to the easy availability of real time data related to smart assets from the field. In the advent of modernization of infrastructure assets lot more data is available to the asset management team from which they can derive quick and accurate decisions for their critical asset characteristics and use this information effectively for their longer term asset investment planning.
Leading utilities are currently engaged in implementing Predictive analytics solutions for management of their critical assets. One of the main advantages of using the predictive analytics solutions will be the asset cost optimization. The percentage of workload that is expended in handling reactive work can be channeled more towards handling predictive behaviors thus reducing the asset down time and increasing the asset availability. This will have direct impact on the Organizations OPEX costs.
Fig 3: High Level Predictive Asset Performance Monitoring Dimensions
The various asset performance areas under which the predictive analytics dashboards can produce useful results are Operational productivity parameters, financial projections, Health and safety forecasts, regulatory forecasts etc.
Asset utilization measures how efficiently assets in the organization are utilized to generate return on investment. Asset utilization can be accurately predicted based on the input variables (e.g. Voltage, Pressure, Flow etc.) that are coming from the field measurement devices. Asset utilization can be different at different time periods of the day and peak load utilization needs to be monitored to improve productivity. The asset can either be under loaded or over loaded when compared to it optimum load. Based on the predictive analysis of the load patterns of utilization of the asset, the asset failure can be analyzed and the return on investment on the network asset can be improved.
Modern day predictive analytic software can analyze the root cause of the utilization issues on critical network assets and produce analytic dashboard(s) that can inform well in advance to the operational staff about the actions that needs to be taken. E.g. Asset utilization for a particular time period is low because of the increased transmission and distribution losses, leakage or theft etc. These correlations can be found by intelligent data analysis using various statistical models and techniques.
Asset failures occur due to aging infrastructure, improper maintenance of assets, network overload etc. Best case scenario is to prevent the failures from occurring before they can cause interruptions in the supply network. There should be a conscious initiative within the organization to identify and network critical assets. Network critical assets should ideally be those assets whose failure or outage can cause tremendous financial impact and customer supply interruptions. The failure behavior of these assets assumes lot more importance.
If the asset failures can be predicted accurately, right type of maintenance action can be initiated at the right point in time with right kind of skilled technicians / operators. This is definitely an improvement over the traditional time and frequency based preventive maintenance programmes. Condition monitoring data inputs into enterprise asset management systems is a right step in this direction.
Improving the adherence to industry standard reliability indices (e.g. IEEE standards SAIDI, SAIFI, CAIDI, CAIFI and MAIFI) has assumed greater importance these days. Unusual or unplanned events in the network can be better detected by the trends and patterns generated by the predictive analytic dashboards. Predictive analysis solutions that are implemented in order to track and monitor the asset reliability measures should ideally focus on:
· What-Ifs: Financial benefits that can be achieved through improvement in relevant indices
· The root cause of the problems that are causing the reliability issues
· The degree of influence that each of the input variables have on the reliability variance
Fig 4: Asset Reliability
An asset is available when it is ready for use. Asset down time is the time when the asset is not utilized due to a forced outage or due to planned maintenance and operation on the asset. Increasing the asset availability and reducing the asset downtime is a key objective of utility asset management professionals as it has direct impact on asset management costs. The historic asset downtime behaviors of various assets can be analyzed and the future asset down time patterns can be predicted if the appropriate scenarios can be modelled effectively.
Asset risk calculations have gained importance with the introduction of networked smart grid concepts. Asset risk or Asset Health index is an index that helps to predict the probability of failure of assets. With the availability of real time data from smart sensors in the field, the asset risk calculations can be performed in real time which can help improve the accuracy of asset investment / replacement decisions. Asset risk will have direct impact on deciding the priority with which the asset replacement projects needs to be scheduled. Assets with high risk need to be replaced / reinforced at a higher priority compared to the assets with lower risk. The cost of future asset replacement projects can be accurately predicted if the asset risks can be predicted accurately. The techniques employed by the asset managers for prioritization usually involve sorting and visualization based techniques. The assets can be sorted in ascending or descending order of their risk or health index scores and can be then visualized on spatial maps to group them together. Replacement projects can then be initiated either to the Direct Labor or to the contract staff for these grouped assets. Based on the accuracy of the asset risk scores arrived, potential savings can be achieved in initiating and executing these replacement projects / programmes.
Asset efficiency is the amount of use of asset for the time period for which it is intended to be used. Predictive analysis techniques can effectively forecast the asset efficiencies even for future time periods with the help of advanced forecasting techniques. The asset efficiencies can be accurately predicted based on the smart metering data that is coming in from the field. The key factors impacting the efficiency of operation of the assets are siloed set of processes, applications and data. When the input data based on which key actions and decisions are taken is in itself inaccurate, the efficiency of operation of network assets gets reversely impacted. Predictive analytics can devise statistical models which can take into account these errors in base data inputs to some extent.
Traditional Vs SAM with Predictive Analytics Comparison
The table below detail the shortcomings of traditional asset management techniques and how these can be improved using the recent developments in predictive analytics packages and availability of data smart asset data measurement devices in the network.
Asset Management Parameters
Traditional Asset Management
Predictive Analytics + Smart Asset Management
Utilization gaps are often identified after several days.
Smart device measurements are often analyzed in real time to reduce the utilization gaps for critical assets.
A large number of asset failures are managed through reactive / emergency maintenance after the asset failure occurs.
Causes of asset failures are analyzed in real time and failure are predicted before they actually occur.
Asset downtimes cause customer interruptions and underperformance in adhering to the key reliability indices.
Better adherence to the key reliability indices due to less customer interruptions and lower asset downtimes.
Asset availability is the amount of productive working time availability of an asset vs the amount of unproductive down time.
Improved asset availability due to the actions taken in real time. e.g. Real-time flow adjustments in the network to control the overload in a transformer to improve its availability.
Asset risk scores are calculated usually a week afterwards resulting in inaccurate decisions.
Variance in asset risks are predicted in advance through the forecasting techniques helping in better asset investment planning
Asset efficiencies are analyzed post the actual day of operation resulting in delayed decision making.
Real time data analysis helps in improving asset efficiencies.
Table 1: Traditional Vs Smart Asset Management with Predictive Analytics
Usage of Predictive analytics along with data retrieved through smart asset management devices on the network is fast emerging as a theme of great importance. In the near future, asset intensive utility organizations will be forced to adopt the use of predictive analytics tools / techniques to predict various dimensions of asset management using the real-time data retrieved from the smart devices.
Fig 4: Predictive Analysis enabled outcomes
The real benefit for these infrastructure intensive utility organizations will be derived from the automated generation of predictive analytic dashboards to optimize their asset maintenance costs. For Predictive analytics tools to produce effective and accurate outputs the input data sets also need to be accurate. Asset Data Quality monitoring and improvement solutions along with predictive analytics will play a critical part in optimizing the asset maintenance costs and achieving the goal of smart asset management in the near future.
This article reflects the personal views of the author, and not necessarily those of his employer Wipro Technologies.
No discussions yet. Start a discussion below.