Comparison of LCI versus VSI Technology: Myths and Truth
We have already discussed the comparison of LCI versus VSI technology quite in detail [1]. However, there is an article [2] in the Internet making a similar comparison of above VFD platforms. Unfortunately, in that article, there are certain statements that seem to be incorrect or misleading. Therefore, we feel prompt to write this post as a response to that article pointing out what is not quite correct. Hopefully it helps to counter some myths and consecutive errors in engineering of drive systems.
Avoid incorrect or misleading statements
Nowadays many places on Earth are blessed with a freedom of speech. People can say, write or manifest whatever they want, wish, believe or simply have to say as long as it is ethical, does not de-honest others, it is free of racism, does not bully or harass etc. Also our company respects that right and privilege of freedom of speech. With regards to VFDs we understand that one manufacturer may prefer one technology against the other. That is perfectly fine. In fact, it is a natural behavior. However, what is not quite fair is to depict a competing technology worse than it is and put some strains. Even worse, making incorrect statements and presenting some benchmark figures that are “produced” to favor one technology against the other.
Our aim here is to provide a fair comparison of LCI versus VSI technology. In my professional career I have worked with both LCI and VSI drives quite extensively. And I appreciate both of them with all their strengths and weak points. Actually, I have not necessarily seen them as competitors. Rather than that, they are complementing each other. I also believe that there is no reason to put one technology in worse light than the other. If one VFD platform is clearly better in all aspects it will evolve and progress while the other VFD platform will decline. My personal opinion is that there is nothing like “One size fits all” [3] when it comes to medium voltage VFDs.
The fact is that the leading VFD manufacturers, such as ABB, GEPC or Siemens, have multiple medium voltage VFD platforms (ABB: 7 platforms, Siemens: 7 platforms). Each of these manufacturers has both VSI and LCI drives in the portfolio.
LCI & VSI Comparison Summary
The mentioned article [2] presents following summary of LCI versus VSI comparison.
As you can see, we have numbered these items 1 to 12 and will comment each of them individually.
1. Line Current Total Harmonic Distortion
In [2] the author states that VSI platform has approx. 2% THD and does not need a harmonic filter whereas LCI systems have up to 12% THD and harmonic filter is required.
First of all, the terms are a bit mixed here! THD stands for total harmonic distortion and normally applies to voltage waveform. For current distortion the abbreviation TDD is used which stands for total demand distortion (see IEEE 519-1992 or IEEE 519-2014). One could also apply THD for current (e.g. denoted as THDi). However, the fundamental current for scaling is generally not the same.
Second, the statement can be assumed to be correct as a “high level” typical figure. The article [2] actually considers a VSI with 36-pulse diode rectifier. Naturally, such VSI has low line current harmonics. However, keep in mind that not every VSI features a 36-pulse rectifier. It is only valid for specific VFD presented in that article.
Finally, current distortion is not the only criteria. It is important to check the grid data and specifically the presence of parallel resonances. Even small current harmonic component can cause significant voltage distortion when matching or being close to a grid resonance frequency.
2. Air Gap Torque Ripple
According to [2] VSI system has 0.5 – 1% air gap torque ripple while LCI has up to 7% ripple. VSI allows standard mechanical system design while LCI requires special mechanical design.
First, the torque ripple numbers seem a bit fishy. There is no definition on the torque ripple provided leaving the interpretation up to the reader. Ripple of 0.5% to 1% sounds extremely low. I could imagine that for a VSI with output sine filter or for a multilevel technology with lot of levels, but not for the topology that [2] is promoting. Something around 2 – 3% would definitely be more realistic. I refer to mean to peak value (0-pk) to be precise.
Nonetheless, the misleading thing is the conclusion on standard and special mechanical design. That is the same like for the line harmonics. You cannot ignore the rotordynamics and eigenfrequencies of the shaft string. It is not so relevant what is the torque ripple as long as it is in a reasonable range (say below 8-10%). What matters is the frequency spectrum of the air gap torque and the correlation with torsional natural frequencies.
Torsional vibration in variable speed drive systems
You would find multiple conference papers dealing with torsional vibration in VFD fed drive systems. When you read the papers a bit carefully, you notice that many of these issues happened in drive systems with VSI technology. Therefore, I disagree with the wording in [2] in this respect. The electro-mechanical integration requires a proper analysis and cannot be waived based on a fuzzy figure of torque ripple.
By the way, looking at figure 9 in [2] the air gap torque ripple seems higher than 1% (considering that one division is 20%).
Generally I agree that VSI tends to have lower torque ripple. Anyhow, what is important for smooth operation is the VFD control. The control system has the capability to either damp or amplify torsional vibration. Refer to our article on torque ripple [4].
3. Input Power Factor
Summary in [2] claims that VSI has input power factor greater than 0.95 and does not need power factor correction. LCI system has power factor 0.5 – 0.92 and “power factor equipment (capacitor) is necessary. True or not?
Yes, VSI with diode rectifier normally reach input power factor 0.95 or greater. VSI with active front end can set the power factor arbitrarily, i.e. 1.0 as default or even leading.
LCI itself has quite low input power factor, particularly at lower motor speeds. However, with the filter system the overall power factor seen by the grid can also reach 0.95. The harmonic filter of LCI always combines both functions: harmonic filtering and power factor compensation. As a result, the LCI system with the power factor correction has as good input power factor as the VSI DFE. Of course, there are some considerations to keep in mind (such as overcompensation at light load) but all of them can be mastered.
4. Power System Stability
At this place it is only mentioned that VSI system is stable whereas stability of LCI system is delicate.
It would be nice to elaborate a bit more on this topic and justify why stability of LCI systems is delicate and for VSI this is no issue.
Generally, I would agree that stability of LCI needs a careful check. VSI with diode front end is rather immune. The rectifier is passive (not actively controlled) and therefore insensitive to variation of grid parameters and interaction with other equipment. In case the VSI is of active front end type the stability becomes more relevant again.
5. AC Power Disturbance
Based on [2] VSI has ride through or automatic restart functionality. LCI can ride through. True or not?
Well, both LCI and VSI have the capability to bridge short-term grid disturbances. I have seen multiple recordings of successful ride through events for both drive technologies.
I don’t understand why LCI should be worse in terms of ride through than the VSI. In fact. LCI is known to have high robustness against grid power disturbances. This was field-proven many times in locations susceptible to more frequent power disturbances such as the Norwegian west coast and offshore installations.
Key for successful ride through is again the control system. We know that LCI of one manufacturer mostly trips during AC power disturbance while other LCI of another manufacturer, located in the same area, remains operational and rides through the disturbance. Therefore, I would not make a direct link between VFD technology and ride through capability. It is about the overall system design, philosophy and maturity of the control system.
Refer also to our article on undervoltage ride through feature [5].
6. Reliability and MTBF
According to [2] VSI has MTBF of 28 years and LCI has MTBF greater 10 years. Both technologies are proven.
What to say to that? I have tried to convey the message within the reliability and availability series.
I would humbly comment that both technologies, when properly designed, manufactured, tested and maintained, have satisfactory reliability. Why shall VSI have almost 3-times higher MTBF is a mystery to me. Especially considering the parts count that is naturally larger for VSI compared to LCI. Moreover, the promoted product does not have so huge field experience to support the high figure. It is on the market for maybe 15 years (status spring 2021)? Reading the whole article it is written that the MTBF “has been estimated to be 28 years”.
When it comes to MTBF lot of manufacturers state shiny figures without any explanation [6]. How was above MTBF calculated? Based on a failure rate of every single part inside the drive? Or based on so called field experience? Does the manufacturer refer to any international standard or do they use some internal rules? As long as this is not ruled we compare apples with bananas.
For end users I recommend to request guaranteed availability as part of you contract with corresponding penalties. Specify it very precisely without leaving any room for interpretation. That will show whether the manufacturer can stand behind those shiny figures or not.
7. Repair time and MTTR
[2] states 0.5 hours for VSI and about 2 hours for LCI.
Same comment here: There is no base for those figures. In the corresponding chapter it is just written that trained technicians and spare modules readily available are assumed. Complete power modules are replaced in the specific VSI solution. What is actually meant with “power module”? That is not explained…
I am not sure whether replacement of a power module inside the VSI is so much faster than replacing a component inside the LCI (e.g. thyristor).
Anyway, it is clear that above MTTR is very optimistic and does not include any troubleshooting or so. Again, there is no good argument why the MTTR shall be shorter for one technology than for the other. The only reasoning I could imagine is that in mentioned VSI platform the entire power module is replaced whereas in LCI one would typically replace the faulty device. Whether this saves time is questionable.
The comparison is anyway not correct. You would have to look at specific LCI model and specific VSI model. What matters is the accessibility of the components from service personnel perspective. Obviously, the better accessibility the faster the replacement. You can have LCI from manufacturer A and LCI from manufacturer B and each of them will have somewhat different MTTR.
8. Equipment size
This comparison is misleading, especially together with Figure 4: Size Comparison of VSI System with LCI System in [2]. It tries to make an impression that LCI requires additional harmonic filter and a PF correction. As we explained, that is one component inherently having both functions.
Here I simply don’t believe that 30’000 hp VSI is 25% shorter than LCI. The LCI depicted in figure 4 looks like a monster and does not correspond to anything what sound manufacturers produce. LCI with 433 inches – that is almost 11 meters. It has nothing to do with reality. I was personally handling a project with 60 MW LCI that had a footprint of 10 x 2.25 m including control, cooling and motor excitation panels [7]. Promoted 25 MW VSI has a footprint of 8.1 x 2.1 m. For 60 MW one would need to parallel at least two (maybe three) of such VSI banks. The equipment footprint of VSI would be way bigger than LCI. Moreover, note that the VSI featured in [2] is rather compact compared to other VSI platforms.
True is that LCI drives require the harmonic filter which is quite bulky. In fact, the harmonic filter often has larger footprint than the LCI itself. On the other hand, the filter system is often located outdoors so it seldom occupies space in the building.
LCI system needs a harmonic filter: True.
LCI drive has larger footprint than VSI drive: False.
The size comparison in [2] is entirely staged to make VSI look better in terms of footprint. However, reality is different and on converter level itself LCI is the most compact solution at high power.
9. Motor design
It is true that LCI requires a dedicated motor design (dual star stator winding, leading PF). On the other hand, the motor manufacturers have mastered such design over the decades and it is not a burden. Some VSI drives actually need a special motor design as well (e.g. 3-level NPC without output sine filter).
LCI has higher motor current harmonics and almost sinusoidal output voltage. VSI has much lower current distortion and a voltage waveform with high dv/dt (exception are VSI drives with output sine filter – those drives typically do not go so high in power). One technology impacts more the thermal design while the other imposes requirements on the insulation system. It is not black & white.
10. Motor type
Yes, VSI works with induction and synchronous motors while LCI requires synchronous motor only. However, in the high power range where LCI actually makes sense it is anyway preferred to use synchronous machine. In medium and low power range induction machine is popular. Is it a drawback for LCI? Not really as LCI does not target such power range.
11. Motor power and drive
This comparison in [2] is very much made up. In fact, the promoted VSI platform can reach max. 30 MVA. By paralleling 4 such drives up to 100 MW can be reached. The disadvantage is that you get a 100 MW system with 7.2 kV output voltage, i.e. large motor current and huge amount of motor cables (or fat busbars). It would be much more reasonable to use a drive with higher output voltage.
One can also parallel multiple LCI converters and increase the power beyond mentioned 120 MW. This is again comparing apples with bananas.
12. Drive output voltage
Not much to say here as there are the limits of specific VFD products. Note that the described VSI technology that is being benchmarked is limited to 7.2 kV. The other VSI type can go up to 11 kV but is limited to much lower power. I wonder why the other VSi type is actually stated in the summary table as it is again quite misleading.
Of course, [2] is already few years old and the technology evolved. There are nowadays VSI platforms for higher power and higher output voltages available on the market.
Remark:
A much more detailed comparison of LCI versus VSI with additional explanations is available for our premium subscribers.
Summary
Both LCI and VSI are mature platforms with their pros and cons. [2] is trying to present VSI as the perfect solution while LCI is depicted as cumbersome. Some advantages of LCI are not even mentioned (regenerative braking, n+1 thyristor redundancy). The comparison was made to promote VSI.
While we have absolutely nothing against VSI drives we believe that comparisons shall be fair. Not every user of a drive technology is automatically a drive expert and may not be able to differentiate the truth from marketing material.
Below once again the summary with our corrections in blue font.
Conclusion
LCI and VSI are popular medium voltage drive technologies in industrial applications. Each of them has its strengths as well as certain drawbacks. Note that most renowned drive manufacturers have both VSI and LCI in their active portfolio. They have a good reason for that. There is nothing like one-size-fits-all.
As you can see, there is certain amount of misleading information circulating in the Internet. I have personally worked with both technologies and can appreciate their pros and cons. VSI can generally “fix” some issues on a product level. Consequently, the platform is relatively easy to integrate into a system. Price for it is an increased complexity of the product itself. LCI solves some of the challenges on a system level (e.g. using a harmonic filter). Therefore, there is an additional system component needed. At the same time the LCI itself remains simple and robust.
References
[1] LCI versus VSI series of articles, https://mb-drive-services.com/category/lci-versus-vsi/
[2] Variable Frequency Drives – a Comparison of VSI versus LCI Systems, https://www.tmeic.com/Repository/Media/Comparison%20of%20VSI%20versus%20LCI%20Systems%20FINAL.pdf
[3] Best VFD topology: One size fit all? https://mb-drive-services.com/best-vfd-topology-one-size-fit-all/
[4] One reason for torsional vibration issue: Torque ripple, https://mb-drive-services.com/one-reason-for-torsional-vibration-torque-ripple/
[5] Undervoltage Ride through, https://mb-drive-services.com/ride_through/
[6] Reliability and availability: MTBF, https://mb-drive-services.com/rel-avail-2/
[7] LCI versus VSI: 60 MW benchmark, https://mb-drive-services.com/lcivsi-5/
