Semiconductor Manufacturing
Data Analysis Sparks New Electricity Consumption Forecasting Method
Many of us have experienced the shock of an electric bill that exceeds our expectations—and our budgets. At home, we can respond by turning lights off when not in use, switching to energy-efficient bulbs, and other common ways of reducing energy consumption.
But what if we had to continue our daily activities without these options? Predicting energy costs and creating a realistic budget would become a top priority. That’s exactly the predicament many manufacturers face: in the absence of ways to reduce electricity use while still meeting production goals, they aim to accurately forecast usage so they can make informed financial decisions.
This was the approach taken by a semiconductor plant in Thailand, which used Minitab Statistical Software to examine its fluctuating electrical load. The production of semi-conductive material, which is the foundation for many electronics, is in high demand and the plant’s equipment must run almost continuously. Using data analysis, a project team at the plant set out to determine the most precise electricity consumption forecasting method in order to maximize profits.
The Challenge
Forecasting models identify patterns in data and then use those patterns to predict what a certain variable—in this case, the plant’s electricity consumption—will do in the future. The team needed to analyze three years of historical data from the Metropolitan Electricity Authority (MEA) using six established forecasting models and one new method created for this project. The new method incorporates factors specific to the plant’s semiconductor manufacturing process.
Each forecasting method would produce predictions for the plant’s electrical load six months into the future. These results could then be compared to historical data. By calculating the Mean Absolute Percentage Error (MAPE), which expresses the accuracy of a particular forecasting method as an error percentage, the team could then determine which model provides the most accurate predictions.
The project team needed to evaluate each method and use the model with the lowest MAPE value in order to assess the cost of electricity, create a sensible budget, and make sound financial decisions.
How Minitab Helped
The team examined six different time series analyses that employ simple forecasting and smoothing methods. Each model highlights different components of the data collected over a period of time—including trend, seasonality, cycles, and irregular movement—and then extends the estimates of the components into the future to provide forecasts. The team used each model to predict electricity usage six months into the future.
First, the project team used the trend analysis method, which fits a general trend line to the electrical consumption data. This analysis displays the long-term tendency of the series to rise or fall over a period of time.
Then using the decomposition method, named for its ability to “decompose” a problem into sub-problems, the team was able to examine components of the data that were not included in the trend analysis method. Dividing the time series data for electrical consumption into additional parts—such as seasonality, cycles, and random variations—allowed the team to factor the impact of each component into the model’s predictions.
In order to smooth out short-term fluctuations in the data and highlight the longer-term trends and cycles, the team averaged consecutive observations in the series to compute moving averages. Because the individual data points created by seasonality, cycles, and random variation are approximated in this model, the smoothed data set can reveal important patterns that would otherwise be difficult to spot.
The team gained yet another perspective by displaying the data using the single exponential smoothing method, which assigns exponentially decreasing weights to data points over time. This technique forecasted the plant’s electricity consumption in a way that placed greater emphasis on the more recent data.
Because single exponential smoothing is not ideal when there is a trend in the data, the project team introduced a second component using the double exponential smoothing method, which takes into account the possibility of a trend in the series. This approach also gives decreasing weights to older observations in the times series, but then smooths the trend and slope by using different constants for each.
In order to extend the double exponential smoothing method to include trend and seasonal variations, the team used a technique known as Winter’s method to calculate dynamic estimates for three components: level, trend, and seasonality.
The variety of forecasting and smoothing methods offered in Minitab allowed the project team to easily analyze, display, and evaluate their data using each of these different approaches, as well as the method that incorporated their own plant-specific factors. The accuracy of each method was assessed by comparing each model’s results to actual data.
The forecasting method devised by the team incorporated specific factors relevant to the semiconductor plant’s process—such as machine requirements, machine idle time, and kilowatts used for each state—and adjusted the raw electricity usage data provided by the utility company. The team then compared the results of each established model to those of their own method.
Results
After creating time series plots and calculating the MAPE value for each of the established methods, the team compared each method to the actual energy usage data and found them all to be similar in their forecasting accuracy. However, the MAPE value associated with the team’s new method was 2.48—a value significantly lower than that of the established models, making it the best choice for predicting the plant’s future electricity consumption.
With the ability to build and apply different time series analyses in Minitab, the team could easily compare their new, untested method with the established ones. Data analysis verified that their model was better than other methods, and enabled them to begin using it with confidence.
This story was adapted from an article published in Vol.2, Issue 10, October 2013 issue of the International Journal of Advanced Research in Computer and Communication Engineering.