The concept of the digital plant starts with data, most of which has always been there but often underutilized. For years, companies have been offering “Big Data” analytics and diagnostics to provide added value based on software and data science expertise and experience. Targeted integration of data-based capabilities with deep power plant system knowledge and fleet-wide experience has led to new proactive operating and maintenance strategies tailored to individual plants.
For example, today’s advanced pattern recognition technologies (APR), coupled with real-time communications and expert knowledge create fleet-wide knowledge bases of normal and abnormal system behavior under a wide range of conditions, providing advancements in knowledge and predictive analytics capabilities. Combining those with the “Voice of the Customer” (VOC) through significant involvement of all levels of the power plant’s O&M personnel assures cost-effective implementation of elements of the digital power plant for each specific plant.
These are excerpts from a paper on this topic by Mark Bissonnette, Daryl Massey of Mitsubishi Hitachi Power Systems Americas, Kazuyuki Misawa, Yasuoki Tomita of Mitsubishi Hitachi Power Systems, and John Kessinger of Mitsubishi Hitachi Turbine Generator Users’ Groups.
It is difficult to pinpoint the initial origin of the first elements of the digital power plant, but it dates back at least to the 1970’s when control systems began to transform from analog to digital, first on local controls using PLCs and then total system controls. Since then and accelerating in the past two decades, the rapid advance of microprocessors, advanced sensors and sophisticated analytics that use massive computing power has steadily made available new elements that can be incorporated into the digital power plant.
A power plant is a critical infrastructure that must have controllable reliability. Introduction of those newly available elements has proceeded at a measured pace with careful evaluation and qualification at each step along the way. In many cases, these elements were built upon earlier introductions of similar technologies in consumer and commercial applications, to assure that initial introduction in power plants contributed positively to plant reliability.
The initial elements or “building blocks” of the MHPS’ approach to the digital power plant emerged in the 1980’s. These included advanced boiler combustion management systems in the early 1980’s and MHI’s first total plant digital control system in 1983, followed by application of early AI/Expert System in advanced boiler operation optimization support systems in 1987 and in a machinery health monitoring system for automatic diagnosis of abnormal vibration of turbine generator shaft systems in the early 1990’s. An early system-level implementation of massive power plant data acquisition and digitization commenced in 1997 when MHPS commissioned the T-Point power plant at the Takasago Machinery Works in Japan, which is an in-house fully operational and heavily instrumented gas turbine combined cycle power plant dispatching into the Kansai Electric grid.
T-Point’s objective was to go beyond typical industry approach of validating individual pieces of equipment separately for short periods in factory test rigs. Its purpose was to provide long term validation of newly-developed major equipment systems and their integration into a real operating environment. MHPS gains important experience and validation from T-Point before deploying and operating that type of equipment in a customer’s commercial power plant. For example, at the T-Point over 2000 sensors are used to verify reliability and long-term performance of the plant.
T-Point is an MHPS designed total power plant, and consists of an MHPS gas turbine, MHPS steam turbine, Mitsubishi Electric Company generators, MHPS HRSG, MHPS plant DCS and other Mitsubishi designed major equipment. Today, T-Point continues to provide data and knowledge acquisition and long-term validation of MHPS’ latest designs. MHPS T-Point Validation Power Plant T-Point was followed in 1999 with the full-scale implementation of MHPS’s first power plant remote monitoring and diagnostics center, referred to as the Remote Monitoring Center (RMC). This RMC is located in the MHPS Takasago Machinery Works and first began monitoring and providing gas turbine combined cycle plants with real-time early warning, fleet benchmark, and engineering knowledge to improve reliability, reduce unplanned downtime, and implement better outage planning based on predictive analytics.
A second RMC was established in 2001, in Orlando, Florida, USA to increase service coverage in the Americas. In 2016, the third MHPS RMC was opened in Alabang, the Philippines to increase coverage in Southeast Asia and Oceania. Today these centers monitor and provide support to power plants all over the World. An early strategic decision, till today, is to carefully introduce advanced analytics and software to improve response speed and productivity, while keep in mind the value of human expert insight and the importance of teamwork with operations and maintenance personnel at the monitored power plants.
Another important example of an element in the journey towards the digital power plant is the ACPFM (Advanced Combustion Pressure Fluctuation Monitor), which is an advanced adaptive expert system, located in the power plant controls network. As DLN (Dry Low NOx) combustion systems became more common in recent decades, driven by stricter emissions and water use regulations, advanced monitoring and control of combustion pressure fluctuations became necessary.
The first MHPS DLN combustion system entered commercial service in 1984 with an early version of pressure fluctuation monitoring. Subsequent improvements made possible by more advanced sensors, increased data acquisition and analysis speed, and expert system application to create real-time models for individual combustor-specific firing stability estimation, now make possible anticipatory response and automatic combustion tuning.
The current generation of A-CPFM goes far beyond just monitoring and protection. It is a powerful tool for improving power plant reliability and autonomous response to ambient, grid-driven and fuel composition variations when applied locally at the plant. It is even more powerful when combined with the additional modelling, fleet comparisons and human insight available when connected to one of the remote monitoring centers.
Further examples of important steps in the journey towards the digital power plant are upgrades of existing power plants that have become more common in recent years as advances in operating plant data acquisition, thermal performance analysis, and fleet performance comparisons quantify the benefits of retrofitting more recently designed and validated components into existing power plants. At one plant in Asia, these advanced digital technologies and human insight resulted in a pre-outage to post-outage output increase of nearly 9% and a heat rate decrease of more than 4%.
At another plant in the United States a similar approach resulted in approximately 7% pre-outage to post-outage output increase and heat rate decrease of about 2%. In both cases, comprehensive plant data acquisition and analysis by local owner/operator engineers and remote monitoring by MHPS accurately suggested the improvement approaches and then validated the results – which are providing tens of millions dollars of added value per year.