Energy efficiency has become an essential part of sustainability in manufacturing during the past few years. The world faces challenges such as climate change, resource depletion, and rising energy costs. With geopolitical and economic conditions, as well as existing regulations, energy costs are constantly fluctuating, making it a challenging task for manufacturers to manage energy consumption. The good news is that manufacturers can leverage Industry 4.0 and Industrial IoT (IIoT) technologies to help reduce their energy consumption and increase sustainability. One such technology is Digital Twins.
A Digital Twin is a virtual replica of a physical system that can be used to simulate and predict its behavior in real time. In manufacturing, Digital Twins can help monitor and optimize energy consumption by providing a digital model of equipment and processes. By using data from sensors and other sources, the Digital Twin can simulate the behavior of machinery, a production line, or a whole factory if required. This enables manufacturers to identify opportunities for improvements including energy savings and optimize their processes to reduce consumption.
One of the most promising trends in Digital Twins for energy consumption is the use of Artificial Intelligence (AI) / Machine Learning (ML) algorithms for optimization. These algorithms can analyze large amounts of data from sensors, equipment, and other sources to identify patterns and optimize energy usage. While Digital Twins can be a powerful tool in managing energy consumption, there are also examples that don’t work well. One reason for this is the use of low-quality data, which can result in inaccurate simulations and predictions. Another reason is the lack of actual understanding of data collected in the development process, which can lead to a Digital Twin not reflecting the actual behavior of the system.
Digital Twin Development
Development of a Digital Twin for energy consumption involves several steps. The first step is to identify the equipment or processes to be monitored and optimized. This can be done by analyzing energy consumption data and identifying areas of high energy usage. The next step is to create a digital model of the equipment or process, which involves collecting data from sensors and other sources and creating a digital replica of the system.
Once the Digital Twin is created, it can be used to simulate and predict the behavior of the system and identify opportunities for energy savings. This involves running simulations and analyzing the data to identify patterns and optimize the system for energy efficiency. But how do we make the first steps and make sure that we use high-quality data that really affects the performance of the system instead of creating some irrelevant digital noise?
So, let’s summarize the questions we should consider:
- How to identify potential areas of energy misusage? How to find bottlenecks and inefficiencies?
- How to analyze data and what approach should be used for automating it?
Digital Twin development requires a thoughtful approach to every detail of the system. And before we start creating a copy of a whole production line or facility, we need to focus on some basics to simulate their role in the whole process.
Motors, bearings, actuators and gearboxes are key elements of almost every machine. If we talk about energy consumption optimization, we should take a closer look at these parts among other areas. Malfunctions usually lead to increased energy consumption in order to compensate for some unbalanced performance. And as a result, component wear increases to the point of total breakage.
Real-time monitoring of these components not only leads to energy efficiency, but actually provides a means of adaptable predictive maintenance, which enables cost savings and accurate process control.
But wouldn’t it be better to simply send a maintenance team to investigate once we detect increased energy consumption?. But how scalable is this approach? How much data will be collected to prevent a breakdown in the future? Can minor problems be observed in the shadow of major and more obvious failures? A smart approach is to use the maintenance team’s time effectively to solve some advanced problems. And of course, they play a crucial role in developing Digital Twins by providing valuable insights about equipment and performance.
So, let’s say we have industrial-grade sensors for monitoring vibration, temperature, noise, humidity, etc and some PLCs controlling motors and gearboxes sending data about performance. What method should be used to automate performance monitoring and provide meaningful analytics to detect abnormal energy consumption early?
Each manufacturing environment is unique. And therefore, it’s almost impossible to predict what methodology will show the best results. Usually, effective energy consumption monitoring solutions combine several algorithms to increase their data-driven insights’ value. We will focus on the most popular ones:
- Long Short-Term Memory (LSTM)
- Vibration Spectrum analysis
- Thermal Models.
Detecting malfunctions in motors, bearings, and gearboxes is crucial to prevent additional energy consumption, potential equipment failure, and downtime. The most straightforward approach is to establish thresholds for various components such as the bearing temperature and motor vibration levels, for example. A sudden increase in temperature or vibration beyond normal levels can indicate potential malfunction. However, this method is effective only for identifying existing issues, and stopping the equipment immediately is necessary to prevent more significant problems. To overcome this limitation, AI / ML algorithms can be implemented to detect malfunctions before they occur, ensuring efficient and uninterrupted operations.
Long Short-Term Memory
Long short-term memory (LSTM) is a popular method used to forecast the behavior of various processes. It is commonly used to predict a motor overheating, for example, without requiring a thermal model. Instead, we provide data to an LSTM model, which then identifies patterns of behavior that are expected to occur. By comparing current data to the model’s prediction, we can detect any discrepancies and alert the maintenance team to check the affected machinery part.
Although this approach appears promising, applying it to an actual motor is not as straightforward. The equipment parts do not operate continuously, and motors start and stop frequently based on the equipment’s mode and load. These input parameters also affect the LSTM model’s output, resulting in false alarms and missed detections. As a result, this method yields only average results.
Despite its limitations, LSTM remains a valuable tool for predictive maintenance in manufacturing, especially in identifying patterns of behavior for equipment that operates continuously. However, other Digital Twins techniques may offer more precise predictions for intermittent operations. Ultimately, choosing the most effective approach requires careful consideration of the manufacturing process and equipment involved.
Vibration Spectrum Analysis
The use of vibration data to monitor the mechanical state of equipment has been well-established in the industry. However, the idea of applying a Fourier transformation to vibration data to identify low or high-frequency components that could provide valuable information is gaining attention. The main challenge lies in collecting data with a high enough sample rate, which is necessary for accurate spectrum analysis. To address this challenge, spectrum analysis can be implemented on a gateway device, instead of sending raw data to the cloud, which would be impractical and costly. Initial results from our proof-of-concept studies indicate that spectrum analysis on the gateway device is effective in identifying existing malfunctions of machinery components. However, the study did not yield any patterns that could predict potential malfunctions in the future. Nonetheless, this approach offers a promising solution to improving maintenance and energy consumption practices in the manufacturing industry. By identifying issues before they become severe, manufacturers can reduce downtime, save costs, and improve overall energy efficiency.
The heating of a physical body is directly proportional to the electrical power consumed, making it possible to use temperature data to identify malfunctions in machinery. By comparing calculations from a physical model with actual data from motors, such as power consumption and operating mode, it is possible to predict when a malfunction may occur. This approach has proven effective in detecting issues at an early stage, allowing for maintenance to be performed during scheduled downtime periods. The sensitivity of temperature changes in mechanical systems has made this method a valuable tool in identifying potential issues and preventing further damage.
The use of Digital Twins for energy savings in manufacturing is becoming more popular as companies seek to increase their sustainability and reduce energy costs. Energy consumption optimization can be started from monitoring of some common equipment parts like motors, bearings, and gearboxes, creating their Digital Twins and gradually coming to higher levels of a production line / facility’s performance. Various methodologies can be used to automate performance monitoring, including threshold setup, Long Short-Term Memory (LSTM), vibration spectrum analysis, and thermal models. Combining different approaches increases the chances of more accurate predictions and more effective energy consumption optimization.
Creating detailed models of basic elements such as motors, gearboxes, actuators, and bearings of the system enables more advanced Digital Twin performance, leading to operational expenditure reduction and effective energy consumption.
Author: Max Ivannikov, Industry 4.0 Expert at DataArt
For more information: www.dataart.com