IIoT Best Practices: Provides machine condition monitoring and predictive maintenance Duke Energy 30 factories

IIoT (IOT industry) is one of the hottest technologies in manufacturing in recent years, but exactly how IIoT landing practice and play in the expected value of the production site, is widely used to before IIoT must first be addressed. In fact, in some of the leading manufacturing companies, IIoT has been applied and began to receive significant return on investment. Duke Energy is one of the most representative one. Today, we take a look at this nation’s largest power generation company, and how to use IIoT big data analytics provide plant-wide machine condition monitoring 30 factories and predictive maintenance. We hope to give practical application IIoT bring enlightenment. The previous manual monitoring machine condition facing enormous challenges Duke Energy Corporation was founded in 1900, is the largest US power company, with more than 80 factories in the United States, and employs more than 29,000 people. The utility company is headquartered in Charlotte, North Carolina, with 52,700 megawatts of generating capacity, to the southeastern United States and the six Midwest states supply about 7.4 million users. Duke Energy has 1.5 million gas customers to provide services, and operates a variety of power generation equipment in Canada and Latin America, including a series of renewable energy equipment.
Figure 1 Duke Energy facility
Duke Energy and other companies in the power generation industry manual monitoring machine condition are faced with challenges. Condition monitoring costs can be controlled by avoiding downtime caused by equipment failure and improve equipment service capabilities. Manual data collection is very labor intensive; in data collection method based on the route, the predictive maintenance Duke Energy experts need to personally go to each site manually collected hundreds of data samples, and then return the computer to view and analyze the data collected . When Duke Energy analysts each month requires nearly 60,000 secondary data collection, typically spend 80% of time to collect data, while only 20% of the time can be used to analyze the data CONTROL ENGINEERING China Copyright [ 123], leading to inconsistent diagnosis, risk assessment is limited, but it takes a lot of time and energy on walking! In 2010, Duke Energy began centralization project throughout the company, the use of new technologies to meet the increasing requirements of reliability and improve employee productivity. Duke Energy wants to have the ability to detect problems and notify the technical experts toBased on data collected to replace the traditional approach routes so they can spend time on more valuable tasks and work done wherever they are. The need to take several years, the project will need to install additional sensors, with a new architecture design and the purchase of new infrastructure to complement the old infrastructure. Figure 2 NI CompactRIO systems to provide solutions for machine monitoring Duke Energy by connecting the whole plant equipment assets
solutions –IIoT and big data analytics challenges Duke Energy began to consider the use of industrial things (IIoT ) and elemental analysis of big data to address these challenges. For this project, Duke Energy and American Electric Power Research Institute and the NI (EPRI), OSIsoft and InStep (now part of Schneider Electric) to jointly develop customized monitoring and diagnostic infrastructure, the facility known as Smart Monitoring and Diagnostics, referred to “Smart M & D”. EPRI has launched I4GEN (Developing Insights through the Integration of Information for Intelligent Generation, improve insight by integrating intelligent power information) framework to support the generation companies turning to things related technologies and solutions. The Institute partners to share expertise and to support collaboration between multiple organizations generate electricity, not make them like a mess, each doing. Technical and operational investments necessary investment for this project include the installation of additional line sensor, infrastructure, architecture and diagnostic applications. In 2012
, Duke Energy began developing a new architecture to support this project Duke Energy uses NI CompactRIO platform that combines a real-time embedded processor, high-performance field-programmable gate array (FPGA) and a hot-pluggable I / O modules, data can be sent to the real-time collection of networked PC computer. The sensor data is transmitted NI CompactRIO monitoring system for signal acquisition and processing, the result is transmitted to the service facility via a wired or wireless mannerDevice. Using a large number of analog data, NI CompactRIO can alarm , and provide a comprehensive analysis of the waveform data expert Duke. Also , Duke Energy also be used for condition monitoring NI InsightCM to visualize and analyze the data. By real-time processor board FPGA and connected to the sensor, the original analog waveform can be significantly reduced, indicating that the node of the system to retain only the “health” status of data required. This avoids data overload situation, to facilitate domain experts to quickly identify problems. NI InsightCM software is an important tool to make data easier to use, and non-technical personnel can easily use. Duke is in transition from traditional technology to new technologies in order to provide better service to end users. Adding more sensors to drive the motor, pump, gearbox and fan monitoring constrained devices. For machines already equipped with sensors, focused energy Duke expanding its sensing capabilities. For example, a steam turbine generator have been higher sensing capability, since they require expensive machine to avoid costly failures by the alarm. Duke Energy in such devices adding more sensors to capture data to support more advanced vibration monitoring, thereby improving the predictability of future failure. Duke in its facilities to determine the number of assets 10,000, and plans to add these devices to more than 30,000 assets sensors, including accelerometers, temperature sensors, fluid analysis sensors, thermal imager, and proximity probes. These vibration sensors increases, and the hydraulic bearing temperature monitoring function, also monitor transformers, electromagnetic characteristics dissolved gases and other assets generators. It is estimated that 75% of the cost of the sensor or not the software, but in the wiring connecting the sensor to the data acquisition computer. Facilities throughout the data acquisition system of the company, can be connected to as many as 30 or 40 hard-wired sensors. Cable must be connected from the sensor to the local data collection computer; then wirelessly transmitted to the signal acquisition device Duke Energy NI data from the server. In order to collect vibration information, you may need to capture 10,000 to 100,000 samples per second, for a few seconds, the machine can charge statusMeasuring points. In addition , Duke Energy plant using a combination of on-site servers to manage large data. Each factory has its own OSIsoft PI server that recoverable data set, and store a variety of tissue sources. The server is located in the center of the Duke of monitoring and diagnostics, which Instep the Prism pattern recognition and prediction software (for mechanical solutions) and GP Strategy of EtaPRO thermal condition monitoring software can help identify the expected behavior of the deviation. In the monitoring and diagnosis center, these software tools are used by a team of five technical staff. Technical staff through a dashboard warning device to know whether some unexpected behavior. In this way, they can investigate and screening questions to determine whether it is truly abnormal situation and the need for further investigation. If an exception is marked, it will send an email to remind standard procedures related to personnel to solve the problem; and graphically to provide information to them, pointing out deviations and provide a preliminary diagnostic recommendations, so that the operator can check the machine. This information will be sent to EPRI’s asset health management system, and compared based on real-time device data compiled from the feature database of all known faults (from multiple companies) is to identify the problem. EPRI then notify Duke Energy experts into the NI InsightCM Data Explore (r section is designed to help engineers quickly locate, inspect, analyze and report measurement data networking software) to conduct a comprehensive analysis. Currently, Duke Energy to all of its data is stored on the internal server, because IT departments are not currently allowed to use the cloud. So far, Duke Energy has been able to handle the huge amount of data using this method collected. Figure 3 is based on condition monitoring software network (Photo NI InsightCM)
the effectiveness of Duke Energy acquired as of March 2017, nearly 30 factories by Smart M & D infrastructure to deploy and manage nearly 2000 CompactRIO systems . In these plants, Duke Energy with automated data acquisition, 80% of the time so that analysts can be used for analysis rather than data collection; therefore, more reliable analysis results. Within a year, Duke Energy monitoring and diagnostic center average use Prism notice published twice a day; only a quarter of which require corrective action alert. These alerts provide a basis for the experts, so that it can plan and repair the equipment at the lowest cost when, for example, when the device is planned maintenance downtime or less when demand. When these machines will continue to run for several weeks, so experts can choose the lowest cost of maintenance arrangements. For example, although there is a fault bearing, Duke Energy is still able to keep the generator running for three weeks until arrangements for safe and timely maintenance. Before the company can collect data from four data points per year, but now will be able to collect data once every five seconds. Not all of the additional data can be permanently stored; therefore the company through a management agreement to determine what type of data is discarded when, in order to achieve a more intelligent data storage. For four years, Duke Energy by avoiding the high costs brought about by the failure, saving up to 130% of the capital budget. The project’s third year, Duke Energy began to see a significant increase in return. The company by avoiding manual data collection and labor cost savings are calculated. Since the system will continue to analyze the data, it can greatly reduce operator rounds, while significantly increasing the frequency of data collection. Data collection is no longer a once a month; but gather several times a day, the number of TB of data collected every week, so more frequently and consistently discover and track problems. This change improves reliability and reduces operating costs, can help managers meet the needs of higher reliability, and optimize productivity by improving analytical skills. Project implementation challenges launched the solution must overcome many challenges. Early in the project, Duke Energy realized that they can not easily complete the project alone. NI and on cooperation with EPRI and other technology providers through a start, Duke avoids many challenges. 1 One challenge facing the company is Duke Energy employees (many of whom are engaged in career-based course manual data collection) to accept and adopt these new methods and reliance on these technology and information. For example, the experts will still receive the warning data based on using a handheld device to manually check equipment. Therefore, the company has invested to optimize data visualization continuous process. Another major challenge facing 2 Duke is running a large number of advanced moldIdentification (APR) model (more than 10,000) can result in a very large amount of received alerts. To deal with this, Duke Energy is developing a prioritization policy management and alert, because the company does not have enough analysts to handle all models. One method is to alert management to prioritize the importance of assets (such as turbines and fans) based. 3 for the use of the four regions of the solution, Duke Energy in each region assigned to the person in charge, they were together with the maintenance experts measure to help them apply new technology to business. However, no one is all-powerful; so start from a collaboration between the OT and the IT team is very important. That there is a challenge of OT and IT departments have their own priorities (and budgets), these matters sometimes there is competition, it is essential to maintain a balance; in addition
, senior project support management is very important, so ensure adequate and sustained financial support. 4 Because the project has been launched CONTROL ENGINEERING China Copyright in a number of factories, the team must cooperate with each site to develop unique solutions for their specific needs. This requires factory managers to choose according to their main pain points and concerns. Another challenge is that in the past few years, the factory management personnel changes have taken place, the need for continuous training of new employees. Extended direction next year, Duke Energy will complete the deployment of additional sensors; because the company has recently expanded the scope of monitoring hope to include additional equipment such as transformers. Duke Energy appreciated, using more wireless sensors can save more money, since they do not require expensive sensor data acquisition system wiring. After that, Duke Energy hopes to get more information about measures taken by the tool helps to diagnose problems in advance. Duke wants to predictive maintenance solutions currently used in further expansion, not only an expert can tell what the problem is, but also provides advice on how to solve the problem. Given the continual loss of industry experts in the field, as will become particularly important.

Related articles