Use of Smart Maintenance Model in Electric Utility Industry for Power Grids-Part II
Random asset breakdowns take the business by surprise and makes it suffer with hefty erosion of capital as a significant amount of Industry capital is in the form of assets. The only way to prevent is to gauge the pattern of asset breakdowns and predict it before the damage is done.
After the data is assimilated by means of real time monitoring of assets through SCADA, DFR, IEDs and historical data secured through EAM tool, the next phase is to analyze the data so as to pick up a trend or pattern of assets breakdowns. The ultimate goal is the move from reactive (scheduled, break-fix) and preventive (condition-based, preventive) to predictive maintenance. The data collected so far by means of real time monitoring and Asset management tool will be analyzed to spot repetitive patterns and trends based on which predictive actions to safeguard assets will be taken. Linear Regression analysis is used for predicting the unknown value of a variable from the known value of another variable. The variable whose value is to be predicted is known as the dependent variable and the one whose known value is used for prediction is known as the independent variable. The model uses the linear equation:
y = a +bx
Where y is dependent variable and x is independent variable. This is also known as the 'Line of best fit'.
A base year is taken ahead of which the asset breakdown data is plotted to arrive at a scatter diagram. A hypothetical example will be say to take year 1990 as the base year, so all the asset data collected since 1990 till date will be used to predict the future asset breakdown trend. Here we are going to plot the graph taking X as the 'Number of days from the base date' and Y being the 'Number of times the asset has broken down'. The value on Y axis will be cumulative and keeps on adding. Say from base year to 180 days, asset has broken down 2 times, 360 days from base year asset has broken down 9(2+7) times and so forth. After plotting the data, a line is drawn in such a way so as the distance between the line and the coordinates on the graph is minimum, this is done using Sum of Squared Errors (SSE). This is the 'Line of Best Fit'. With the help of this line we will predict the probable future asset breakdown time window. Using the equation arrived at(y = 0.0375x - 7.5333), we can give the value of x (number of days from base date) and based on the value of x we can find out the value of y which will be the predicted value. The data plotted should be at least for 5 years so as to pick a trend. The more the quantity of data in terms of the number of years, the more reliable and accurate the predictions will be.
The diagram mentioned below shows the asset breakdown data plotted on a graph called as Scatter Diagram. A line of Best Fit is arrived using Linear Regression Equation.
After the likelihood of assets
breakdown is gauged, the next step is to take remedial measures. As soon as the
probable asset breakdown time frame is predicted, a prudent maintenance
activity is carried on that asset before it enters the predicted breakdown time
frame. The prime focus of the maintenance activity will be on the problems
because of which the asset might have been broken down in the past and may see
a downtime again in future if the maintenance is not carried out. So a maintenance
carried on the asset beforehand will practically eliminate the probability of breakdown
because of similar problems faced earlier.
Smart Maintenance Model using Predictive Maintenance by way of Linear Regression is a boon for the electric utilities where the entire industry is reliant on the proper functioning of asset. The collection of data using both real time monitoring and historical data leaves no stone unturned for optimizing asset performance. This gives management a leading edge ahead of its competitors to manage asset lifecycle costs in a better way and make informed and judicious decisions.