Peter Wung
The scientific method is the foundational basis of our intellectual inquiry and is a fundamental tool, not only in testing our hypothesis, analyzing our assumptions, and understanding the physical realities, giving the design engineer direction in anticipating and predicting future designs, but also in determining the accuracy and precision of our hypothesis through comparison.
The scientific method process steps of observing, modeling, verifying models through testing and experimenting, questioning, and adjustment of the theories/models has been ingrained in the practicing engineer’s intuition and problem-solving ability since our very first formal introduction to the art of the scientific inquiry.
While the process of scientific inquiry has not changed in principle, the implementation of the steps has evolved through the evolutionary and revolutionary changes brought by technological breakthroughs. While the practice of performing small-scale prototyping and experimenting with these prototypes was fruitful and enlightening, the scales of the problems have grown exponentially. The problems have become larger, in physical size, the number of variables necessary to describe the problems, the number of cross-coupling and interaction among those variables, the complexity of the nonlinearities inherent in the problems, and the financial cost necessary to construct and investigate the problems. Indeed, there is the question of whether a scaling down of large problems would indeed be adequate to account for all of the dynamics, both modeled and unmodeled, inherent in the actual problems.
Concomitant to the growth of the problem complexity, the question of accuracy and precision of the experimental results comes into question. On a simple level, the concatenation of measurement errors could accumulate to unacceptable levels, obviating the experimental veracity of the painstaking testing process. On a more complex level, there are a number of questions that arises naturally to the inquisitive mind:
One of the great ideas that has made the verifying step more efficient, in the context of time and cost, is the idea of simulation. Computer simulations overtook the venerable network analyzer decades ago, crossing the digital divide as well as the divide between the physical and the numerical. Indeed, our present-day engineering practices are predicated on the quality of the simulation results—accuracy and precision—all so that we can predict our system response in an infinite number of excitations, variations of the problem structures, and an infinite number of unpredictable variations.
Unfortunately, we can only simulate what we know, as the results are a product of our conscious knowledge of reality; that is, we cannot model what we don’t know or have never experienced. So we circle back to measurements in real-life applications and its attendant questions, listed above.
The mix of simulation uncertainties, modeling inaccuracies, and experimental uncertainties gives the present-day engineer an opaque sense of where their theories lie. This is not to say that our view of reality is becoming less accurate; on the contrary, our technological development has been astounding both in terms of our simulation and metrological advancements. The problem is actually a result of the successes that we have experienced. We have been emboldened by our successes and are taking steps into unknown territories: into broader, more complex problems while also digging deeper into the granularities of the intricacies of the unknown and unmodeled.
I had the privilege of convening and chairing a Special Panel Session during the 2021 IEEE Energy Conversion Conference and Exposition regarding this topic in the motors and drives context. A summary of the panel proceedings was published in the IEEE Industry Applications Magazine (IEEE Industry Application Society 2022) (https://ieeexplore.ieee.org/document/9755228).
The focus of the panel was on electric machines and electric machine drives, a smaller system in scale as compared to the electric grid, but no less complex. From the article:
The fundamental issue at the center of the discussion is: how to verify theoretical engineering calculations? In certain cases, prototyping is convenient, cheap, relatively accurate, and large number of samples are available and economical to test. While in other instances, only limited prototyping is possible — electric machines and drives testing generally become more expensive and time consuming as the machine and drive rating increases, which require massive resources in terms of hardware, software, and manufacturing expertise necessary to conduct valid experiments with confidence.
Ultimately, the question is whether the simulation results are trustworthy enough to be accepted as the proof of the proposed technical solution and design? Have the simulation software attained a level of accuracy which matches and/or surpasses the accuracy of experimental results?
Prof. Lingling Fan was in the audience for the panel, and her enthusiasm for the topic made this issue of IEEE Electrification Magazine a reality. I am honored and excited to be a part of this intriguing and informative foray into this discussion for larger electric power systems.
The primary intent of this issue of IEEE Electrification Magazine is to gather, centralize, and summarize information on the state-of-the-art computer simulation technologies in use today. The summary provides insight into how these simulation packages are applied in their specific applications, identify the salient advantages and shortfalls, give the reader a broad view of the present-day simulation landscape, and provide a roadmap for future comparison.
IEEE Industry Application Society, “Simulation versus experimental verification [Society News] ,” IEEE Ind. Appl. Mag., vol. 28, no. 3, pp. 88–94, May/Jun. 2022, doi: 10.1109/MIAS.2022.3148628.
Peter Wung (pwung@earthlink.net) is with the University of Dayton, Dayton, OH 45469 USA, and Marquette University, Milwaukee, WI 53233 USA.
Digital Object Identifier 10.1109/MELE.2023.3320529
2325-5897/23©2023IEEE