Active harmonic mitigation with AFE drives
In this article we will once again touch the topic of grid side harmonics, particularly focusing on active harmonic mitigation with AFE drives.
Abbreviations used:
AFE = Active Front End
APF = Active Power Filter
ASD = Adjustable Speed Drive (synonym to VFD)
DFE = Diode Front End
VFD = Variable Frequency Drive (synonym to ASD)
This blog post was motivated as a response to article “Active harmonic mitigation – What the manufacturers don’t tell you” [1], published in IEEE Industry Applications Magazine.
First of all, we are happy that the author(s) of [1], Anthony Hoevenaars, Marek Farbis and Michael McGraw, raised this concern. It is important to have an open discussion on subjects related to variable frequency drives. Our mission is to make the topics associated with variable frequency drives transparent and understandable to a broader community.
At the same time, we would like to address some points mentioned in that article where our opinion is somewhat different. And we can speak here about real industrial practice. Over the last 10 years we have conducted numerous harmonic calculations and several network studies.
Differences between low voltage and medium voltage VFD
Before we dive into the subject let us explain one particular point. Note that our view on active harmonic mitigation mainly reflects our experience with medium voltage VFDs. In low voltage world the principles are similar, but some differences apply. Particularly the grid connection is usually somewhat different. Medium voltage drives almost exclusively use an input isolation transformer (see line side connection [2]) . Quite often this transformer has a multi-winding design with phase-shifted windings for harmonic elimination [3, 4]. The transformer also blocks propagation of a common mode voltage into the public grid. Exception is a direct to line solution that is, however, rather seldom. In contrast, low voltage drives might be transformer-less, connected either directly to the grid or through a line side reactor. Purpose of this reactor is to reduce harmonics. It does not provide galvanic isolation, though.
Medium voltage drives are larger in power and their impact on the grid is often thoroughly studied. Low voltage drive is typically smaller and less critical for integration into the grid (although not always true). At the same time, the projects are smaller and the system studies mostly do not go so deep into detail.
Do VFD manufacturers hide some issues?
While working as a tendering engineer I have seen plenty of cases where the competitors developed strategies that would make their solution look better, either technically or commercially. Sometimes their arguments would not really fit for the application. Other times, they would manipulate some “typical data” to make the other manufacturer look worse.
Dishonest manufacturers throw a shadow on the VFD technology as such. We wish that everyone has certain ethical standards and works in a professional way. It is in the interest of (almost) everyone: equipment manufacturer, system integrator and end user.
Certainly, there are VFD manufacturers who are less honest. They may highlight some advantages and downplay specific drawbacks. However, I trust that most manufacturers do not lie or hide things. At least those who want to stay in business in long term. According to numerous studies, negative publicity spreads up to 6-times faster than positive messages. Therefore, I guess that VFD manufacturers do not want negative publicity. In medium voltage drives this is probably even stronger argument as the community is smaller compared to low voltage VFDs and issues will come on the surface rather fast.
Principle of active harmonic mitigation
The principle of active harmonic mitigation was described in our previous article [4]. It basically uses a suitable modulation – mostly a selective harmonic elimination (SHE) algorithm. The input transformer (at medium voltage) is often just a 2-winding design. However, some AFE drives, particularly in higher power range, may use multi-winding transformers as well. In such case some harmonics are cancelled by the phase shifting of transformer while other harmonics are minimized by modulation. The drive then features low harmonic distortion without the need for too high switching frequency.
Active harmonic mitigation: What the VFD manufacturers don’t tell you…
So let’s look into the article “Active Harmonic Mitigation – What the manufacturers don’t tell you”. We will take some statements out of the article and comment them:
A general comment:
From report [1] you might get the impression that main purpose of AFE drive is the harmonic mitigation. Not sure how about low voltage, but in medium voltage this is definitely not the case. While AFE offers active harmonic mitigation it is not the main reason to choose this topology. Users buy AFE drives because of their capability of regenerative braking or possibility to control the power factor. Active harmonic mitigation is an additional benefit, but seldom the reason for AFE selection. If the customer cannot benefit from regenerative braking and input power factor of 0.95 or better with inductive character is sufficient for him we would suggest a DFE variant. Shall the user be concerned about grid harmonics, the manufacturer can offer multiple solutions based on DFE drives that most likely comply with the requirements.
1. Growing popularity of AFE drives despite higher cost
While we do not have exact market figures we would agree that AFE drives grow in popularity. It is also true that in most cases AFE drive has higher cost than his DFE brother. That remains valid even if you consider somewhat lower cost of AFE input transformer (AFE transformer might be just 2-winding type while DFE transformer utilizes multiple converter side windings with a phase displacement).
As said, active harmonic mitigation is not the main reason to choose an AFE drive. It is rather an additional benefit than a selection argument.
2. AFE drives produce higher harmonics that manufacturers may not tell you about
AFE drives, due to their nature, produce characteristic harmonics and their multiples. In higher frequency range they generally have higher harmonic distortion than DFE drives. Authors of [1] call it a switching noise. So far we agree. Nevertheless, we would not agree that VFD manufacturers try to hide something. Most international standards define harmonic limits up to order of 50 (harmonic order is the integer multiple of the fundamental grid frequency). While the compliance with standards is shown only up to order of 50 (for higher harmonics there is no limit to compare the actual value with), for total harmonic distortion manufacturers may calculate higher orders as well.
According to IEC 61000 the total harmonic distortion (THD) is defined as follows:
where
So for purpose of compatibility levels H is equal 50. However, it does not prevent anyone to either calculate or simulate the current and voltage spectrum to much higher orders and compute the corresponding THD accordingly.
In our company we calculate THD based on harmonics up to 100th order (H = 100), sometimes even higher. For medium/high frequency range the input transformer is quite effective in minimizing the harmonics. Note that medium voltage VFDs generally use lower equivalent switching frequencies than their low voltage counterparts.
3. Shunt connected active power filter (APF)
The authors present a figure of AFE drive having a parallel (shunt) active power filter (APF). Maybe such solution is sometimes used in low voltage range. However, most LV drives work without such extra filter. In medium voltage APF is practically not used (no single case known to us). The AFE module can minimize the harmonics and control the power factor without any additional hardware. Since an APF is not used, other drawbacks, such as complex control structure, are not applicable.
4. AFE drives have much higher current harmonics
This is certainly not true. Yes, AFE drives usually have somewhat higher harmonics at higher orders. However, note that most AFE drives are of a voltage source inverter technology (and the authors at the beginning limit themselves to voltage source converters), the AFE has a character of a voltage source. The transformer and grid reactance scale with frequency. Let’s assume an inductive grid as certain simplification. The higher the order of voltage harmonic the higher the corresponding reactance. As such, there is no reason to expect current harmonics of critical magnitudes at higher frequencies.
An example of current harmonic spectrum of 1600 kW AFE low voltage drive is depicted in figures 1 and 2.
As we can see in above figures, the dominant current harmonics are in the range 2000 – 3000 Hz, i.e. within the range covered by the standards are just slightly above. Performing a harmonic analysis up to harmonic order of 100 seems to be sufficient for most cases.
Spectrum of an AFE drive can be characterized as follows (general statement applicable to almost any AFE):
– Voltage harmonic distortion (THDv) is almost the same at no load and full load (AFE is a voltage source)
– Current harmonic distortion (THDi) is quite high at no load. However, this is no issue. THDi is a relative value. At no load the current is highly distorted, but the magnitude is very small. Before starting to panic look at the current harmonics in absolute value (Ampere) rather than %. The world will look okay again.
5. AFE drives have higher losses and lower efficiency
It is true that AFE drives have slightly higher switching losses. Reason is simple: an AFE has higher losses than an alternative DFE consisting of passive diodes. This is no secret and we have informed about it e.g. in our article on energy efficiency [5]. In case of frequent regenerative braking AFE drives are efficiency-wise a better choice. On the other hand, if regenerative braking cannot be utilized (e.g. utility does not allow it) or happens very seldom (e.g. application practically never stops) then DFE drives outperform AFE in terms of efficiency. To be fair, AFE can compensate reactive power and potentially reduces losses in the grid.
6. Missing frequency band in harmonic standards
Sure, both IEEE 519 and IEC 61000 define harmonic limits for “low frequency” distortion up to order of 50 (2.5 kHz in 50 Hz grid / 3 kHz in 60 Hz grid) and limits for “high frequency” (mainly EMC) limits. There is a missing band in between both of them. At the same time, let’s remember that high order voltage harmonics of AFE do see a high reactance of the input isolation transformer (in medium voltage) or of an AC line reactor (low voltage). Consequently, the current harmonics are limited to fairly low magnitudes.
Formulas below illustrate the situation for an inductive network:
Wile it is true that some higher harmonics may fall into the frequency band currently not covered by standards, the manufacturers can still calculate these harmonics and assess the corresponding risks. Moreover, network harmonic studies can be performed. If the study indicates a risk, a specific mitigation measure can be applied.
In medium voltage applications the higher harmonics will not go far beyond the harmonic order of 50 as it is desired to keep a moderate switching frequency for best power utilization and low switching losses.
7. Systems with resonance in power system shall be thoroughly studied
In this point we totally agree. Every power system will have one or multiple resonances. The question is whether there is a resonance in frequency range that could be excited by the harmonics generated by VFD. If so, a more detailed analysis is required. And it does not matter whether the VFD is a DFE or AFE type. We talk about these things in our network harmonics series [6].
Active harmonic mitigation: One more hint
Before we conclude there is one more hint: If you have any doubt regarding harmonics or if you think that the VFD manufacturers does not tell you the true story, feel free to challenge them. We trust that most of them are honest and do not downplay any harmonic issues as it would hurt their reputation in long term.
Still not convinced? Simply specify on-site measurements as part of acceptance tests. Of course, the measurement requires a dedicated equipment. You may engage a third party to perform such measurement for the sake of objectivity. There is some additional cost involved, but you get a peace of mind.
Summary
The article “Active harmonic mitigation – What the manufacturers don’t tell you” addresses several possible drawbacks of AFE type of drives. The authors claim that some of these issues are not much highlighted by the VFD manufacturers. They also illustrate the issues on few examples that are then extended to all commercially available AFEs. Such conclusion is in our view too narrow minded.
First, we note that the topic mainly addresses low voltage AFE drives, as it is obvious from presented examples and schematics. The arguments are not 100% transferable to medium voltage. Moreover, at medium voltage the DFE type VFDs usually feature higher pulse numbers and can meet the harmonic limits without the need for external passive filter (meaning that the comparison active harmonic filter versus passive harmonic filter becomes less relevant). It is true that AFE drives are somewhat “richer” in higher voltage harmonic orders. On the other hand, the corresponding currents are limited by the system reactance that increases with frequency.
The active harmonic mitigation is not a primary reason to select a AFE drive, at least not in medium voltage. If harmonics are the only concern (regenerative braking and active power compensation is not required), a DFE solution fulfilling the harmonic requirements can most likely be presented.
Finally, we want to stress that well-designed drive system, no matter if consisting of AFE or DFE drive, can be properly integrated into the power system without negative impact on the grid. Renowned VFD manufacturers have extensive know-how, long-term experience and necessary tools to assess the situation, elaborate a harmonic study (if required) and propose a solid techno-commercial solution.
Acknowledgement
We would like to thank Mr. Antonio Solar Cabarga who made us aware of this article and initiated this review. Antonio also provided valuable feedback and comments based on his professional experience.
References
[1] Active harmonic mitigation – What the manufacturers don’t tell you, Article by MIRUS International
[2] How to choose a medium voltage VFD: Line side connection and power quality, https://mb-drive-services.com/how-to-choose-mv-vfd-line-conn/
[3] VFD transformers: Multi-winding design, https://mb-drive-services.com/vfd_transformer_design/
[4] Network harmonics: Harmonic mitigation methods, https://mb-drive-services.com/harmonic_mitigation_methods/
[5] What efficiency can you expect from your drive system?, https://mb-drive-services.com/energy-efficiency-part-5/
[6] Network harmonics (series), https://mb-drive-services.com/category/net-harmonics/
5 Comments
Simone · January 25, 2021 at 10:54 am
This is quite an interesting topic.
I would like to point out that, according to my experience, people that are stating that the active harmonic mitigation is the primary reason to select an AFE drive (MV), usually are manufacturers that have only (or mainly) AFE drives in their portfolio, typically CSI drives. I often see specifications where AFE drives (MV) are required for pump applications, simply because the specification has been issued based on CSI drives. In such case, since no braking capability is required, the AFE is justified by the current harmonic mitigation capability.
On the hand, with LV drives AFE is a typical solution to mitigate harmonics. A reason behind it is that LV drives are usually directly connected to the grid without the input transformer (many drives connected to the same bus bar, with a common distribution transformer). In such case only an AFE drive could do the job.
admin · January 26, 2021 at 5:50 pm
Hi Simone,
Yes, this is a great comment! We believe the same – key difference between LV and MV is that MV drive almost always has its own input transformer while LV AFE drive often has direct to line connection (possibly with a decoupling AC reactor). Therefore, in LV systems AFE drive may very well outperform a DFE alternative in terms of harmonics. In MV, however, the DFE can often reach comparable or even better harmonic performance than AFE due to multi-pulse rectifier and phase shifting input transformer.
Tony Hoevenaars · January 26, 2021 at 10:21 pm
I am one of the authors of the paper sited and agree with many of the comments in this article. Our paper was intended to address LV AFE’s and not MV and maybe it would have been better if we had identified that up front. Certainly the transformer associated with MV AFE’s will provide galvanic isolation which most likely eliminates the common-mode issue and will help lower differential mode high frequency harmonics as well. I also agree with the recommendation that AFE drives are very attractive when regenerative braking is required but for harmonic mitigation alone, there are other, often better options. In LV systems, when the objective is to simply reduce the harmonics generated by a VFD, it is our opinion that options such as a series passive filter can reduce low frequency harmonics as effectively as AFE without introducing higher switching frequency harmonics. And if done correctly, these passive filters do not need to be susceptible to resonance with the power system. Also, we felt it very important to note that existing standards do not adequately address harmonics in the frequency range from 2 kHz to 150 kHz which can lead to troublesome levels of these harmonics. If the article seemed to imply that ALL AFE manufacturers hide the high frequency issue, that was not our intent. We simply wanted to make readers aware that they may not always get all of the information they need to make an informed decision.
admin · January 27, 2021 at 12:55 pm
Thank you for your comment and explanation Tony. It is valuable to hear also from the co-author of the original paper. Basically to have both sides of the story. I think that we can agree on most of the things. Our perspective is mainly from medium voltage side while the paper addresses LV AFE’s. We definitely agree that there is a frequency range not covered by the standards. Some manufacturers may see it as an “advantage”, others would proactively address the point and check it deeper to make sure that there will be no issue later on. Finally, I would like to add that although VFD manufacturers try to make the systems filterless for multiple reasons, a filter is sometimes a good option. At the same time, knowledgeable filter designers are scarce. Therefore, it is a benefit for everyone when there are companies with expertise in harmonic filters that one can reach out to.
Harmonic calculation versus network harmonic study - MB Drive Services · January 29, 2021 at 8:21 am
[…] the way, as discussed recently in the article on active harmonic mitigation – what the manufacturers don’t tell you, the frequency scan is performed up to 5 kHz (harmonic order 100) instead of just 2.5 kHz defined […]
Comments are closed.