Best of Both Worlds

by Bogdan Kasztenny and Mangapathirao (Venkat) Mynam, Schweitzer Engineering Laboratories, Inc. USA

Digital Distance Relays Following Analog Principles

At the heart of any distance element is the “IZ – V” operating signal, polarized with voltage (mho) or current (quadrilateral).

As we unsurprisingly began focusing more on life-cycle challenges, cost, and ease of use, we gradually started taking protection for granted. We accepted that the performance of modern relays came close to an unbreakable ceiling, and too often we dismissed improvements as diminishing returns. After all, we have made considerable progress over the last five decades since electromechanical relays ruled the world. Or have we?

This article shows that progress sometimes takes detours. We discuss how early microprocessor-based relays gave up implementation methods of static relays not because they were inadequate, but because they needed a new implementation method that would fit the processing power available at the time. We argue that going back to analog principles and implementing them in digital relays allows us to unlock performance that is otherwise hidden from us.

A Brief History of Distance Elements
Imagine it is the late 1920s and you are a protection engineer working on line protection. Overcurrent relays and fuses are commonly used, but they are not very selective. Pilot wire schemes have just been invented, but they are expensive. Would it not be beneficial if one could devise a relay that sees faults on the line but not beyond the remote bus, yet uses only local measurements? This would be a significantly better version of a current-based relay but with reach that is much less dependent on system strength and fault resistance. All you have, however, are iron cores, copper wires, coils, resistors, contacts, springs, and bearings. Vacuum tubes are not reliable, and semiconductors are not yet invented. Your solution may involve a balance-beam relay (Figure 1). The beam tilts toward closing the contact and tripping the breaker if the ampere turns from current overcome the ampere turns from voltage. The V-over-I term in (2) is the apparent impedance, and you just invented a nondirectional distance relay. To make the scheme directional, you supervise it with a directional relay. You allow the users to set the reach by adjusting the R, N1, and N2 parameters.

You may improve on your original invention by using a cylinder unit relay (Figure 2). The cylinder rotates toward closing the contact and tripping the breaker if the torque created by the polarizing and operating signals is positive. You shape the operating signal to be proportional to I • ZR – V and the polarizing signal to be proportional to V. As a result, you obtain a directional mho operating characteristic. The characteristic is a circle that stretches between impedance 0 (relay location) and ZR. ZR is the distance relay reach that users set by adjusting turns and resistances in your new design.

The directional mho operating characteristic is a circle with several advantageous features: it allows load to come close to the characteristic boundary; for external resistive faults, the characteristic bends away from the apparent impedance that intrudes due to infeed effect; the characteristic is directional on its own. But is the mho circle special? Would the industry adopt your design if the resulting characteristic was an ellipse? Absolutely. Fixed reach is the key advantage, not any specific characteristic. Any reasonable shape would work (we now insist on specific shapes for coordination of stepped-distance and pilot schemes).

At the heart of any distance element is the “IZ – V” term or the operating signal, polarized with voltage (mho) or current (quadrilateral). Typically, distance comparators use the 90-degree limit angle originating from electromechanical relays. As a protective relay pioneer once said, “V over I does not make a distance relay.” Our industry has been improving distance relays since their introduction in the 1940s, including better polarization methods to address bolted close-in faults, phase selection, directionality, load blinders, reactance elements, and so on.

Imagine it is the 1970s and semiconductor devices (diodes, transistors, and operational amplifiers) have become reliable enough for protection applications. Now you can implement various characteristics by checking the angle between operating and polarizing signals through coincidence timing (Figure 3). Two sine waves are 90 degrees apart if they have matching polarities for a quarter cycle. In your distance relay, you replace torque-based electromechanical comparators with analog coincidence timers. Because the latter are smaller, lighter, and cheaper, you can use more of them and shape sophisticated operating characteristics. A mho circle is now a choice, not a constraint.

You also realize you are not shackled anymore to the inherent inertia of electromechanical relays, but you can explicitly control the degree of filtering and speed in your solid-state designs. Protective relays have become much faster: perhaps too fast for their own good. Transient simulators are still in their infancy. Fault recorders are not widely available. Static relays have been designed for speed without necessary transient testing. Predictably, security has suffered. High component failure rates, lack of self-monitoring, and electromagnetic interference (barely recognized as a problem at the time) have only added to the challenges. Fast but fragile is a fitting label for the static relay generation. Static relays would have matured with investment fueled by their advantages, but emergence of the microprocessor-based relay with an explosion of functionality arrested the static relay generation.

Imagine yourself in the early 1980s. You have access to analog-to-digital converters, but you can sample just several times a cycle. You have access to a microprocessor but the available processing power allows running only so many instructions in a power system cycle. Can you emulate static relays the same way static relays emulated the angle comparison of electromechanical relays? Absolutely not. You need a new method. You need to slow down the flow of information to keep up. If you barely satisfy the Nyquist principle and sample just a few times a cycle, you need to heavily filter the input voltage and current. The output from the anti-aliasing filter is therefore very clean, and you
can represent the near-sine waves with phasors, even if calculated from just a few samples per cycle. Phasors allow
you to slow down the flow of information to any arbitrary rate you need in order to keep up. You can calculate phasors just once or twice a cycle and still claim instantaneous protection operation. Once you have the phasors, you can implement the tried-and-true protection principles through calculations (Figure 5). The synthesis method becomes an implementation method. You can shape the directional mho characteristic by using torque (3), angle (4), m-calculation (5), or any other equivalent equation. Once you have the phasors, it is just mathematics from there.

Do not underestimate the role of numerical optimization in the early microprocessor-based relays. For example, the m-calculation (5) does not improve performance, it still just plots the mho circle. It is not simple either, but it is attractive because when calculated once, the m-value can be efficiently used in a multizone relay by comparing it to the reach setting of each zone.

BeijingSifang June 2016