Fidelity Matters: What “High-fidelity” Really Means
While all simulation vendors claim that they can provide high-fidelity power plant simulators, what does the term “high-fidelity” really mean? All the buzz today is around the “Digital Twin”, and the soft copy of your plant that enables a wide variety of engineering and optimization applications. However, not all twins are identical.
In the nuclear power realm, the ANSI/ANS 3.5 standard essentially defines the meaning of high-fidelity for operator training simulators. The simulator must perform with certain tolerances, and no negative training can occur because of differences between simulator and plant response. Moreover, performance must be in real-time and repeatable. For fossil fuel power plant simulators, the ANSI/ISA 77.20 standard provides functional requirements and addresses fidelity in terms of both physical and functional realism.
Realistically, however, the question of fidelity is often tied directly to purse strings. Some plants opt for low to medium fidelity simulators because they initially appear to be a lower cost option. But, as Richard W. Vesel notes in his article “Power Plant Training Simulators Explained” (Power Magazine 12/01/2013), “Cost notwithstanding, many users choose high-fidelity simulators to train operators so that the trainees get as deeply exposed to the plant as possible, without actually touching it.”
If you require a high-fidelity power plant simulator, either by law or by virtue, it is important to note that not all high-fidelity simulators are created equal. Fidelity is essentially a combination of three elements:
- Technology – the engineering quality of the equations used in the modeling tools;
- Engineering Rigor – the thoroughness of the model and the scope to which plant systems are modeled;
- Experience – the experience of the simulation engineer and the “test operator” to assure realism in real-time.
These elements combine to provide the user’s desired steady state and dynamic performance during normal and, in particular, off-normal operations.
When it comes to the engineering quality of a simulator’s modeling tools, it’s important to consider whether the simulator is using its equations to predict results or using empirical correlations that model known data.
Many high-fidelity nuclear power plant simulators are designed to model known data. This method complies with the ANS 3.5 standard, but it doesn’t account for predictive modeling. When simulating phenomena that have not happened in real life (thankfully), the models must have the engineering muster to accurately predict behavior so that operator actions provide a realistic response to the situation. Therefore, you should question whether your model technology uses six equations to accurately models mass, energy and momentum, in both gas and liquid phases, or simply uses some correlated phenomena.
For example, let’s take a not-so-simple combined cycle gas turbine plant with a triple pressure heat recovery steam generator (HRSG). Solving the pressure, temperature and enthalpy as a single solution for the high, intermediate and low pressure stages of the HRSG provides the right response for the operator during start up. Make the plant a 2x1 or 3x1 configuration, and the integrated operation complexity of steam mixing, headers, startups/shutdowns and transients demand effective model equations. For utilities with supercritical or ultra-supercritical coal units, high quality equations combined with extended steam tables are required for model performance.
Ideally, you want to trust that the technology can give accurate performance across a wide range of operating conditions from cold iron to rated power. To make that happen, the technology must be based upon engineering principles and not simply tuned to a limited set of data to get the right responses. The engineers that apply the technology must also understand the plant design and operations and nodalize the model to account for accurate and seamless response across the full range of operation.
The fidelity of your simulator is influenced by how much of the plant is modeled. In fact, you gain additional uses for your simulator beyond training when more areas and functions of your plant are properly modeled. Your return on investment in a training simulator increases significantly when engineering, human factors, safety analysis, I&C and other parts of the organization can benefit from the tool. When deciding on a simulator vendor, it is important to ask yourself and others in your organization if additional plant systems should be modeled or just those sufficient enough to provide correct information for training the control room operator.
While providing correct information to the operator in the control room is the goal of training, a true high-fidelity simulator can be used to virtually commission new control logic and equipment and to target plant efficiency improvements. For example, a nodalized, matrix based solution gives a more accurate response to complex thermodynamics (or electrical) systems because of its simultaneous equation solver. The model should tell you what is happening in the plant, not the other way around. Because pressure, temperature, enthalpy, etc., are calculated at nodal level, you often can see what’s happening in the plant in areas that are not instrumented, giving engineering a better understanding of plant performance.
Simulation technology has grown leaps and bounds from the early days of flight simulation, but one of the most important and impactful aspects of an effective simulator are the simulation engineers who build the models. The experience of the engineer to understand plant data and its interactions, determine how detailed the model needs to be and implement the high-fidelity solution is paramount to ensuring the simulator is capable of addressing the user’s performance requirements and expectations. Are your vendor’s engineers really writing control loops to make the simulator behave, or are they mechanical engineers applying thermohydraulic principles within the technology to create the true dynamic effects?
Simulators will likely remain, first and foremost, your plant’s most valuable resource for training operators, engineering, technical and maintenance personnel. However, innovative power plants and process plants are expanding the use of their high-fidelity simulators beyond training to include engineering aides, virtual commissioning, analytics, cyber security, and verification and validation. To take advantage of your “Digital Twin” it is simply not good practice (in fact, isn’t it risky?) to think you can base critical operating decisions on weaker fidelity models.
As you look to invest in a new power plant simulator or consider expanding the application of your simulator beyond training, remember that fidelity matters. Is your simulator truly high-fidelity?
Learn how high-fidelity simulation allowed for extensive distributed control system testing and operator training prior to plant commission of a combined cycle power plant.