Amanda Neuenfeldt is the power administration system (EMS) modeling lead engineer at Actalent, a world firm specializing in engineering and sciences providers, together with expertise options. Neuenfeldt works with rising grid applied sciences associated to synthetic intelligence (AI), machine studying (ML), and cybersecurity. She has studied the appliance of a predictive machine studying algorithm, coupled with a synthetic intelligence optimization algorithm, to cut back grid era deviations whereas automating the management features to cut back management latency and improve reliability.
Neuenfeldt not too long ago offered POWER with perception about her firm’s work, together with its use of information and machine fashions to help grid reliability and resiliency.
POWER: What are some enhancements that may be made to an electrical energy transmission and distribution system and/or an influence plant (thermal or renewable) to enhance its reliability?
Neuenfeldt: A serious enchancment we’re engaged on so far as system reliability is aligning the information throughout completely different methods. Planning, safety and controls, power administration methods (EMS), and supervisory management and knowledge acquisition (SCADA) methods usually function independently, inflicting misalignment that results in delays and additional work. To deal with this, we’re aligning fashions like Normal Electrical’s and different operational machine fashions (ODMs) in addition to collaborating with regional transmission organizations (RTOs) to make sure consistency. Over the previous yr, we’ve accomplished round 20 sub-projects centered on cleansing and aligning knowledge—reviewing SCADA factors, values, scores, and impedances. By making certain all methods share a dependable, synchronized dataset, we’re constructing a basis for extra correct and reliable operations shifting ahead.
POWER: What are a few of the upkeep applications that may be carried out to make an influence plant and/or an influence grid extra dependable? How can this upkeep be automated?
Neuenfeldt: We’re working with knowledge requirements like CIM (Widespread Info Mannequin) 61 970, 301 for normalization and alignment, and the rising 61 850 customary for sensible grids. The purpose is to assemble complete knowledge by means of widespread sensor networks, enabling extra dependable predictions and higher preparedness for capability enlargement. Given the complexity and scale of this knowledge, machine studying and AI are important to precisely forecast and automate upkeep, which is why we have to have all the information aligned. AI is nice, however for those who’re feeding knowledge that’s misaligned or simply unfaithful, your predictive mannequin isn’t going to make very correct predictions.
POWER: How necessary is redundancy as a part of energy plant/energy grid operations?
Neuenfeldt: Probably the most essential level of redundancy is avoiding energy outages. I reside in Texas and was one among many individuals who didn’t have water or electrical energy for per week throughout winter storm Uri as a result of we didn’t have that redundancy. And tragically, folks died as a result of it was too chilly and when the facility went down heating methods stopped working. In order that hit laborious, and it’s why we want to ensure these methods are redundant. As we anticipate points with local weather change and the rise in power calls for that it brings, we have to be sure that we’ve acquired the capability to fulfill these calls for, and the interconnections to redirect power in an emergency, whereas remaining price efficient. An affordable steadiness of redundancy will probably be obtained with correct, well timed, and dependable predictive knowledge that’s then optimized with sensible calculation methods utilizing synthetic intelligence.
POWER: How can operations be standardized to enhance reliability?
Neuenfeldt: Widespread Info Mannequin connectivity and integration. One quick advantage of CIM is the synchronization and normalization of information from varied methods ensuing within the elimination of false entries in methods. All of it comes again to making sure the information is aligned and “talking the identical language”, that’s the north star of reliability at this level.
POWER: What are a few of the particular elements/gear that can be utilized to boost reliability?
Neuenfeldt: Elevated accuracy and verification in knowledge and fashions. All of it comes again to knowledge reliability. The one really viable solution to acquire that is the centralization of information, or interfaces and integrations that end in synchronized knowledge between methods. These methods all work in live performance with each other, however all converse completely different languages. As soon as we are able to get the methods to all converse the identical language by way of knowledge, that would be the greatest part in enhancing reliability.
POWER: How can mannequin predictive management be used to boost reliability?
Neuenfeldt: Introducing a predictive mannequin can decrease delays, end in a discount in gear injury and outages, improve effectivity and cut back prices. It additionally results in higher cross utilization of methods and gear to create redundancies; predictive grid state values may be created through CIM connectivity and integration.
POWER: How can a program be carried out to allow proactive substitute/restoration of older gear (each software program and {hardware}) in energy plant and energy grid methods?
Neuenfeldt: Utilizing machine studying and predictive instruments to calculate the imply time between failure values for particular particular person items of apparatus will probably be big. This, mixed with optimized upkeep applications utilizing AI and gear substitute scheduling will allow proactive substitute with out disruption to the system.
With ageing grid methods, elevated demand, climate dynamics, and varied inexperienced initiatives, reliability will probably be depending on smart-grid elements using synthetic intelligence, real-time knowledge acquisition, centralized knowledge availability, and optimization of methods and elements in any respect ranges. The present strides within the preliminary implementation of CIM-based methods and elements will open the door to the sensible elements and methods that can present the important wants required to offer the reliability demanded by customers and regulatory businesses.
POWER: What are a few of the energy outage administration methods (OMS software program or others) utilized by your organization, or that you’re accustomed to?
Neuenfeldt: Whereas I’m not immediately working with these methods, I’m supporting them by means of work with one among our shoppers, a big utility firm that’s engaged on integrating a number of methods and databases collectively. Mainly, we’re taking all these misaligned methods and streamlining them into one mannequin. The CIM mannequin is a centralized knowledge repository, and each RTO that exists in North America goes to be connected to this CIM as a result of for AI and machine studying to successfully work, you’ve acquired to have knowledge from all over the place. Once we’re machine studying and AI on the grid for automation, you want each RTO concerned. You want the sensors to gather the data, and also you want the central knowledge repository to retailer that data.
POWER: How can software program upgrades assist energy plant operators, and energy grid managers, higher handle energy outages?
Neuenfeldt: Our present venture is an implementation of the CIM. When that is full it’ll function the inspiration for sensible elements to automate and support energy grid managers in dealing with energy outages in addition to elevated effectivity, reliability, and profitability.
POWER: How necessary is having the ability to have real-time data to anticipate outages brought on by excessive climate or excessive demand for electrical energy?
Neuenfeldt: Dependable, steady, and correct forecasting is essential since our present infrastructure and procedures are usually not able to assembly what we predict we’ll want sooner or later. There was a rise in grid complexity and power calls for, however a discount in general grid inertia. Actual-time data will even be important for the implementation of smart-grid elements in addition to machine studying and synthetic intelligence instruments. As soon as the mixing venture my consumer is engaged on goes reside, and as they’re profitable within the knowledge normalization, the flood gates are going to open and also you’re going to see extra initiatives for knowledge integration and extra RTOS getting on board.
POWER: How does your organization use (or what’s your familiarity with) knowledge sources to find out the situation of an outage and the variety of affected clients?
Neuenfeldt: I’m not working in an space immediately coping with real-time reporting of outages, nevertheless the CIM implantation with this integration venture is the bedrock of the gathering and distribution of that knowledge to be used in real-time response. This venture is a centralized knowledge repository, and we’re extra involved with aligning the information in order that sooner or later these knowledge sources are simpler to make use of and extra dependable.
The success of initiatives that purpose to offer centralized and uniform knowledge for a number of methods would be the basis spurring the business to implement the CIM and pave the way in which for a lot of new and improved outage administration and response instruments, methods, and methods.
—Darrell Proctor is a senior editor for POWER.
