VFD and conventional wisdom

VFDs and conventional wisdom

Welcome to our blog about variable frequency drives and drive systems. This time on the topic: VFDs and conventional wisdom. The message we try to convey is that what many people usually believe and what seems logical at first point is not necessarily right. Does it sound interesting? Then make yourself comfortable and keep reading.

What do we mean by conventional wisdom? Well, conventional wisdom is something that most people believe that it’s right. It sounds logical. People believe it. It gets repeated and repeated. Finally it reaches some books and, of course, internet. It sounds so logical that many people just take it for granted. And yet, it sometimes may not be true.

Conventional wisdom according Wikipedia

From Wikipedia, the free encyclopedia

“The conventional wisdom or received opinion is the body of ideas or explanations generally accepted by the public and/or by experts in a field.”

Further Wikipedia writes about accuracy of conventional wisdom:

“Conventional wisdom is not necessarily true. It is often seen as a hindrance to the acceptance of new information, and to the introduction of new theories and explanations, an obstacle that must be overcome by legitimate revisionism. That is, conventional wisdom has a property analogous to inertia that opposes the introduction of contrary belief, sometimes to the point of absurd denial of the new information or interpretation by persons strongly holding an outdated but conventional view. Since conventional wisdom is convenient, appealing, and deeply assumed by the public, this inertia can last even after many experts and/or opinion leaders have shifted to a new convention.”

This articles talks about conventional wisdom associated with VFDs. We will demonstrate few examples that seem right according to conventional wisdom, but actually are wrong.

Let’s look now at some of these “VFDs and conventional wisdom” items that in fact might not be right.

N+1 semiconductor redundancy is a must for high reliability

Power semiconductor is the heart of a VFD. We all feel that this is a very essential component. Therefore, redundancy is a crucial thing and N+1 semiconductor redundancy dramatically improves the reliability and availability. In case of failure of one semiconductor within one string (phase) the VFD continues in operation. Therefore, this feature is fundamental for every VFD design. Sounds logical?

Well, the N+1 redundancy is another example of conventional wisdom. However, the reality is a bit different. If you have a VFD based on medium voltage semiconductors, the total number of power semiconductors within the VFD is fairly low. In popular 3-level NPC topology that most of the market leaders still actively manufacture and promote, you need just 12 active switches inside the inverter. So take the reliability figure of such semiconductors as well as all other components inside the VFD and calculate the mean time between failure. Now take another VFD, this time a multi-cell design, for the same motor power. The multi-cell VFD is based on low voltage cells (~ 600 to 750 V per cell). Each phase consists of multiple such cell connected in series. The number of series connected cells depends mainly on nominal motor voltage. Let’s assume 5 cells per phase. This makes in total 15 cells or 60 semiconductors. 5-times more than mentioned 3-level NPC. Shall the multi-cell VFD have comparable reliability figures as the 3-level NPC, it needs N+1 redundancy.

So the argument for N+1 semiconductor redundancy does not apply to all VFD topologies, but only those with large quantity of semiconductors. That is the true. The conventional wisdom, which is not quite true, is that you need N+1 redundancy no matter what type of VFD you have.

Unlike the mystified N+1 requirement we have a proof from the field. LCI technology exists since decades. The LCI drives work in the industry for more than 40 years (some people consider them as ‘dinosaurs’). LCI puts thyristors in series depending on machine voltage. Therefore, it is similar principle like the multi-cell topology (however, LCI thyristors are medium voltage rated unlike the low voltage cells). One of the LCI features is that it can realize N+1 redundancy. Despite of that, majority of all the LCIs in the field, no matter whether it is ABB, Alstom, GE, Siemens or any other brand, actually do not have this N+1 redundancy! So the LCIs are around us and work fine for decades without a semiconductor redundancy. In fact, many users are happy with them and use the control upgrade options to extend the service lifetime of the LCI.

Ask yourself the questions: Is N+1 semiconductor really necessary? Or does it apply just to low-voltage multi-cell topology? Are maybe the LV semiconductors less reliable as they are made for mass production while MV components tend to be more industry-graded? Finally, most importantly, are specific manufacturers pushing N+1 semiconductor redundancy to limit the competition and make more profit? N+1 was born as a solution to solve the otherwise poor reliability of multi-cell VFDs with plenty of semiconductors inside. Later, it was twisted by smart marketers as a ‘must have feature’.

The true is that smart users actually ask for guaranteed availability rather than specifying type of semiconductor or semiconductor redundancy. N+1 is neither an insurance nor a golden duck. Semiconductor redundancy does not provide you any guarantee in terms of availability.

Liquid cooled VFD is less reliable

How many times have you heard this story: Liquid-cooled drives are less reliable than air-cooled counterparts. Of course, liquid cooling system is more complex unlike the simple and robust air cooling. Liquid cooled VFD will soon run into gigantic leakage issues and your service crew will become team of highly skilled plumbers. You may ask why liquid cooled VFDs are still manufactured? How come that they did not disappear from all manufacturers’ brochures and catalogs? No-one would buy an equipment that is programmed to fail?

In fact, the area of high-power VFDs is clearly dominated by liquid cooled drives. More specifically, all major VFD manufacturers used liquid cooled drives as the first choice in the multi-megawatt range. Also frequency converters for other applications, such as grid interties, high power STATCOMs or HVDC converters are practically exclusively liquid-cooled.

Semiconductors are the most common components to fail

One more dive into reliability topic today. It is believed that semiconductor failure is clearly the most frequent root cause of a VFD to be out of service. Building on this mindset, people focus e.g. on semiconductor redundancy (as described earlier) or try to specify some additional ‘design margins’. Does it actually make sense?

Although it is true that semiconductors account to many VFD trips and downtimes, they are not the only root cause. Look at a part list of a VFD. How many different components can you find? 20? 50? No, there are many more.

Can a capacitor fail? Of course, it can and does. Can a resistor overheat and fail? Definitely! Can a sensor get damaged? Oh yes! Can some of the many fiber optic connections get loose? Well, the list could be much longer, but we don’t want to bother too much.

While semiconductors are essential and critical, there are many other components as well. Reliable VFD is such with all components properly designed, assembled and tested. Make sure that you don’t lose the forest out of your vision because of this bunch of trees (called semiconductors in this case).

Regenerative drives have the best overall efficiency

Regenerative braking is the most energy efficient way of braking. In regen mode your machine runs as generator and the VFD injects this generated power back to the grid. Beautiful. Therefore, regenerative drive must inherently have the best energy efficiency, right?

Another conventional wisdom that may turn to be wrong. We don’t deny that regenerative braking is a great feature that is just perfect in certain cases. But is it always the best choice, particularly from energy efficiency point of view? if you are our regular reader, you may already know the answer…

Whether regenerative drive has advantage in terms of overall energy efficiency depends mainly on the application and process. How frequently is the VFD expected to brake? How long does one such braking take. How much energy can be ‘regenerated’?

The fact is that AFE based VFD will have somewhat higher losses in motor operation compared to a DFE. For simplicity let’s assume following example:

== Example ==

Table 1: Total consumed energy over a duty cycle with seldom braking

In this case the total consumed energy over a duty cycle of 2 hours (120 min) is marginally lower for the DFE drive. Although this drive cannot regenerate the braking power and energy, it has lower losses in drinving mode that prevails. Of course, in case of frequent braking the situation would look different.

AFE drives solve all harmonic issues

Install an AFE drive and any potential issues with harmonics will just vanish. This is a popular argument of some of the VFD manufacturers. But is it really true?

Well, let’s go a step back. When do we experience an issue with increased harmonic distortion? One reason is when the VFD draws very distorted current from the grid or modulates very distorted voltage. Other reason is when the harmonics generated by the VFD hit a parallel resonance in the grid.

AFE drives can shift the harmonic spectrum depending on selected modulation. However, the harmonics are still there and could have the potential to cause an issue.

Much more to discover as premium subscriber

Are you interested in modern drive technology? Do you enjoy our posts? Would you like to get deeper and acquire real knowledge?

Our blog posts are just little apetizers. They basically scratch the surface. In contrast, premium plans give you access to a valuable knowledge and experience that we have been collecting and building for more than 12 years.