Autonomous Power Plant Takes Shape in Japan

The world’s first autonomous combined cycle power plant is currently under construction at the Takasago Machinery Works facility in Japan, and it will be operational by 2020, according to Mitsubishi Hitachi Power Systems (MHPS).

The Yokohama-headquartered power plant equipment and technology manufacturer told POWER in February that it began construction of the unique verification project in late 2017 following several years of engineering efforts. The project, known as “Tomoni Point,” is being built adjacent to the original T-Point power plant, where, since 1998, MHPS has demonstrated and validated a range of modern MHPS gas turbines and associated technologies under real operating conditions, and dispatched the power they produce to the grid.

MHPS reported that Tomoni Point will be a 60-Hz gas turbine combined cycle power plant with an output greater than 600 MW, in a one-on-one configuration. The turbine is the latest enhancement of MHPS’s air-cooled J-Series advanced gas turbine model and will have a groundbreaking efficiency of more than 65%. MHPS began marketing the J-Series in 2011 following long-term validation at T-Point, and recently announced that the J-Series has accumulated 750,000 operating hours with 99.3% reliability. Tomoni Point will also integrate MHPS’s proprietary TOMONI digital solutions in all its controls to allow it to further boost flexibility and performance, as well as optimize operations and maintenance.

Among notable TOMONI components that will use or enable artificial intelligence (AI) are smart sensors; an OSIsoft PI System data management platform; Microsoft Azure cloud computing; and AI-augmented advanced pattern recognition software to confirm normal operation or detect and report anomalies. Tomoni Point will also integrate AI-assisted maintenance and enable predictive maintenance strategies.

“This is different than today, which relies on traditional schedule-based or wait-until-failure-based maintenance strategies,” noted Todd Brezler, vice president of Marketing at MHPS. AI-based technologies will include the latest version of MHPS’s advanced combustion pressure fluctuation auto-tuning system, as well as other “similar advanced controls in development and validation,” Brezler said. However, Tomoni Point’s capabilities will also include machine learning, including from itself and other MHPS turbines. Finally, the entire operation will initially be operated remotely and monitored at an analytics center.

After Tomoni Point has been commissioned, which is expected in 2020, it will initially rely on AI to provide operational recommendations—and these will require operator approval, said Brezler. “Think of it as a plant that can perform like it had an unlimited number of experts on site 24/7 who can recommend and respond instantaneously,” he explained. “Once the AI has acquired enough data and machine learning, the plant will operate autonomously, plan its own maintenance schedule and maintenance activities, order its own maintenance parts and service schedule, and become increasingly capable of self-healing.”

It is unclear when the plant will begin fully autonomous operations. Brezler noted that MHPS plans to run the verification test for a minimum of 8,000 hours, as it does with all MHPS gas turbine technology validations. One reason is to ensure enough digital data is collected to augment autonomous operation. “Instead of teaching AI to play chess based on analysis of thousands of chess matches, we teach the AI to ‘talk power plant,’ ” he explained. “AI will incorporate other data like weather patterns, market level pricing, and electricity demand to schedule generation and dispatch power.”

4. The Tomoni Point project under construction at the Takasago Machinery Works facility in Japan. Courtesy: MHPS

The company’s success could prove transformative. While digitalization is being increasingly adopted by power generators, it is typically limited to various facets and components, or to improve specific operational or performance-based aspects. Running an entire power plant through AI, for example, could reshape how operations and maintenance is performed. As Brezler noted, MHPS already applies AI and advanced analytics in key operational areas; last June, for instance, it verified at a Taiwanese coal-fired power plant that an AI combustion tuning system could reduce fuel costs by up to $1 million annually (Figure 4). “In the near future, AI will take action by itself for grid support and some maintenance tasks like ordering parts or dispatching maintenance personnel at optimum times,” said Brezler.

The potential doesn’t end there. “AI can analyze variables like total plant equipment capability at various ambient weather conditions, parts status and aging, and market demand to optimize performance and plant profitability,” Brezler said. Improvements could include higher efficiency at part-load, lower turndown, or increased megawatts, he added. Along with advanced analytics, AI could also optimize generation scheduling, which could improve grid stability in markets where fuel switching and renewables variability is becoming more commonplace. Brezler concluded by stating, “Our industry is undergoing unprecedented change, and we plan to lead this change in power.”

—Sonal Patel is a POWER associate editor.