Skip to main content

Schneider Electric: Daily Energy Model

Schneider Electric has set a goal to reduce its energy intensity by 3.5% each year. The Lexington site was an ideal site in which to implement a daily model, as it uses more energy than any other Schneider Electric site in N. America and has experienced significant detection challenges. Before the implementation of the daily model tool, the energy performance of the site was determined using normalized model data based on calendarized utility invoices, which resulted in a six-week delay in determining performance for any given month.

The primary goal of the daily model was to implement a way to communicate the energy performance of the Lexington plant to the site personnel with the least delay possible. This process is analogous to statistical process control, with visualization using control charts. The site has several electric and natural gas submeters. The most significant barrier for implementation was the technology required to perform the data consolidation, statistical analysis, and visualization in one platform.

A collaboration between members of the Facilities Management (FM), Energy and Sustainability Services (ESS), Analytics and Artificial Intelligence (AAI), and Hosted and Managed Solutions (HMS) teams made this project possible. The ESS team, in conjunction with FM, developed the requirements of the daily model as well as the visualization. The AAI team further developed the project and implemented a platform to consolidate interval data from the site submeters, hourly weather data, and operational features to develop the daily model. The AAI team also developed a live visualization of the model using a control chart format. The visualization was embedded with relevant information, confidence intervals, and color coding to make the chart easy to read and interpret.

Implementation took approximately three months, with a cost of less than $50,000. The AAI team provided consultation and contextual input from ESS (an energy program manager) and feedback from FM (a regional facilities manager). The AAI team consisted of one data analyst intern, one data scientist, and one software architect. A senior information services technician from the HMS team played a key role as well by providing a method to extract interval data from the submeters for use in the daily model platform.

At the conclusion of the development, the daily model was released to the local energy team and relevant training was provided.

Performance Indicator Tool

The daily energy model shows an indicator bar on a graph for each day. A green bar indicates that the goal has been met or exceeded. A yellow bar indicates that the goal has been missed slightly. A red bar indicates that the goal has been missed to a greater degree. The length of the bars provides an indication of the degree to which goals have been exceeded or missed. The user can also get numerical values (expected energy use, actual energy use, weather conditions, and other relevant variables) by hovering over the bars. In addition, the model includes a zoom feature. The color codes mirror those that have been used in the monthly energy model for many years, making it easier for plant personnel to adopt.

Exhibit 1: Snapshot of daily observation and pop-up box with the parameters used in prediction (the parameters considered include the daily average temperature and number of shifts of operation).

Exhibit 2: Implementation of color-coding that aligns with corporate energy reduction goals (the gray bars around each observation indicate the confidence interval of the predicted observation for each day).

Exhibit 3: Daily gas model for 2015, 2016, and 2017.

Exhibit 4: Daily electricity model for 2015–2018.


The success of this model can be seen through its adoption in facilities and its ability to provide a better means of real-time communication of equipment. The daily model helped engage the local facilities management personnel in energy management and allowed them to be in tune with the energy consumption and performance of the facilities. In addition, the model also allows for the early detection of communication errors and other failures of the submeters, facilitating timely repair and minimized loss of data.

Exhibit 5: Gas performance during the first quarter of 2016 (before implementation of the daily model).

Exhibit 6: Gas performance during the fourth quarter of 2016 (after implementation of the daily model).

The tool has increased the site’s focus on energy use, including the amount of discussion regarding energy performance. The biggest benefit of the project was the immediate availability of energy performance data for the facility on a daily interval. Unoccupied weekend consumption was occasionally identified as higher than normal according to the daily model, and this information was used to improve operational control during subsequent weekends. In addition, the tool has been used to identify waste during production periods. Refinements to the tool have also been added since the initial launch.