Older GTO drive efficiency 2

[jraef]

Oh great gurus...
I have been asked to do a comparison between keeping an old (c. 1988) 250HP 460V Relcon VFD which uses GTOs, and installing a new PWM drive in its place. The utility has indicated that they will contribute to the event IF they are shown that there will be at least a 10% improvement in efficiency. I am assuming they mean throughput efficiency, and although the old Relcon drive currently shows a displacement PF of .65, apparently they do NOT care! I am well versed and capable of describing all of the other benefits to the user, but efficiency is unfortunately the only concern of the Utility.

The question I have for you is this: years ago when we were retrofitting older 6-step and current source VFDs with PWM designs on a regular basis, there were several reputable studies available which showed the benefits of doing so, including the improved throughput efficiency. Since I have not been asked this question in probably 10+ years, I no longer have access to any of those papers. Does anyone else have one floating around or can offer some advice (other than a Google or Thomas Register search) in finding one? It would be greatly appreciated.

"Venditori de oleum-vipera non vigere excordis populi"

 

[cbarn24050]

What makes you think a igbt drive will be any more effecient than a gto drive? Igbts have higher coduction losses and higher switching losses.

 

[aolalde]

Jraef.

250 HP = 186.5 kW
Motor efficiency is around~ 91.2%; input to the motor = 204.5 kW
Assuming 97% efficiency for the actual driver (very bad); Total input= 215.3 kW
Losses in the driver = 10.8 kW

If the expected efficiency for the modern driver is 99.5%
Total input = 204.5/.995 = 205.53 kW
Losses in the new driver = 1.0 kW

By IEE 100, Efficiency= Po/(Po+Pl) *100%

Throughput efficiencies:

Eff. old= 186.5/215.3*100=86.62%
Eff. new= 186.5/205.53*100 = 90.74%

4.12% improvement

I do not see how a 10% improvement could be achieved by changing the driver.

 

[jraef]

aolalde,
I am having my contractor take actual field measurements of the existing drive system, but I think you are right on. It will probably not add up to enough difference.

The one caviat however is that I just found out that someone installed a 500kVA 480-480V shielded isolation transformer in front of this drive some years ago in an attempt to protect other nearby equipment from the noise created by the drive. The impedance of that transformer will be additional losses in the system, and I could make a case for not needing it with a PWM drive. So I will have him take his input power measurements from the line side of that transformer.

cbarn240,
Oops, my bias was showing! You have a good point in that to be truly objective I should not assume that as fact. It was however a factor in my past experience with retrofitting old 6-step inverters. From what I remember, the slower switching time of the thyristors created more losses (per pulse), but the higher switching frequency of the transistors made them roughly equal in switching efficiency. The overall difference came in the necessary series reactors in the thyristor based systems having more losses than say a bus choke and even a line reactor for the PWM drives.

"Venditori de oleum-vipera non vigere excordis populi"

 

[aolalde]

Jraef:

But; if the driver increase the motor losses, due to harmonics, may be a cleaner voltage output will increase the overall efficiency,

Measuring the actual power consumption is a good start and try to get the Motor Efficiency when it operates with a true sinus voltage wave.

 

[jbartos]

Suggestion: The harmonic mitigation technique efficiency has not been considered in the above posting yet. For example, efficiency of harmonic filters or phase shifting transformers should be added in the overall system efficiency.

 

[electricuwe]

I have to object to chbarn24050. In the power and voltage range under consideration an inveter using IGBTs will have significantly lower total losses (at the same switching frequency) compared to an GTO-based inverter.

Since the switching loss per pulse is significantly lower for an IGBT compared to a GTO drive manufacturers increased switching frequency, which may have led chbarn24050 to that conclusion.

But, even if the losses of the inverter are the same the increased swicthing frequency will reduce the losses caused by harmonics in the machine.

 

[jraef]

Sorry for not updating sooner but there are delays in getting accurate measurements. I just wanted to respond to jbartos' issue of Harmonics.

Normally, I would wholeheartedly agree, but this customer is out in the "boonies" and this is the only significant load on the system for miles around. They are not concerned about harmonics and are not intending to do anything about it. They were worried about switching noise on the line with the SCRs, but that will not be a problem with the PWM drive.

"Venditori de oleum-vipera non vigere excordis populi"

 

[cbarn24050]

Hi electicuwe, I did not suggest that the gto drive switch at the same frequency as igbt drives. GTOs have lower conduction losses than igbts, which is significant at these current levels.

 

[jbartos]

Comment: IEEE Std 100 Dictionary does not define "switching noise on line with the SCRs".
PWM drives with Active Front End (AFE) tend to have very low input harmonics so that there will not be anything similar to "switching noise on line with the SCRs".

 

[electricuwe]

chbarn24050,

if you compare recent 1200V-IGBTs to GTOs the advantage of low forward voltage drop is only marginal.

I just searched my files for datasheet of a suitable GTO in that voltage class and found the DGT304SE manufactured from 1988 by Marconi. The company changed their name to GEC Semiconductors later Mitel and now Dynex and this device is still listed on the Dynex-Website:

http://www.dynexsemi.com/assets/DNX_DGT304SE.pdf

This datasheet states a forward voltage drop of 2,2 V maximum at 600 A and 125°C.

If you take a look into the datasheets of modern 1200V-IGBTs you will find forward voltage drops of 2,0 V typical at nominal current and 125°C, e.g. for the FS300R12KE3:

http://www.eupec.com/gb/2_PRODUCTS/2_1_ProductRange/pdf/db_fs300r12ke3_2_0.pdf

The maximum forward voltage at 125°C is not statet in the datasheet but judging from the values at 25°C it will be only slightly higher that the 2,2V of the GTO.

Using two of that Sixpack-Modules in parallel in a PWM-inverter would be a modern alternative to a six-step inverter using 6 disc-type GTOs and I guess such a configuration or a comparable one can be found in several inverters available on the market as long as they have been designed recently.

 

[cbarn24050]

Hi again electricuwe, firstly this is a 250a device, at 250 amp the forward voltage is 1.7v at 125degC. Secondly you are comparing a 20 year old device with the latest new types.

 

[electricuwe]

chbarn24050,

regarding current: you are right with the 250 A nominal current, but that current is the rated average current. Voltage drop has to be calculated from the current as a function of time. For a six step inverter a current if 3x Iav is flowing through the device for 120°.

regarding comparison:
yes, but this is exactly the correct comparison in this case. I guess jraef doesn't want to replace his old GTO inverter with an inverter using first generation IGBTs bought as used part.

In general:

The comparison based just on the device characteristics does give just a fist insight in the topic because the drives under discussion also use different topologies and control methods. I just wanted to show that the low forward voltage drop of the GTO doesn't support the statement that there will be no savings.

Comparing the losses of the differnet drives part by part leads to the following result:

rectifier and DC-link:
PWM drive significantly lower than Six-step (DC-Link chokes tend to have high losses)

Inverter: Conduction losses abut the same, see discussion above, switching losses higher in the PWM inverter due to PWM operation

Motor: significantly lower with PWM-inverter

In total a modern PWM drive will have significantly lower losses. If this justifies a replacement is mostly related to operating scheme and energy cost.

But I strongly recommend that jraef also considers the availability of spare parts. Experience shows that it is often very difficult to keep old power electronic equipment in operation.

This is not only related to the power semiconductor devices itself but also to passive components and control cards. Usually getting spare parts for 16 year old equipment should not be impossible, but if there are major changes in technology as in the case under discussion things might be more severe

 

[ozmosis]

Efficiency can take many forms. If the Relcon drive is part of a critical process then you should also consider the turn-around time of getting a replacement or spares etc if something was to go pop. Especially if it's out in the "boonies".
I'd also pick up on the efficiency proposed by aolade originally. I think you would be mighty pushed to get 99.5% eff out of a PWM drive. More like 97%.

 

[cbarn24050]

Hi electricuwe, if you are going to compare gto loss at 600A then you must also do the same for your IGBT. A gto drive is not 6 step, it's a pwm. The difference in efficiency between the two types is insignificant, it's going to take a very long time, if ever, to recoup the cost of a new drive from the energy saved, assuming that there is any.

 

[electricuwe]

chbarn24050,

comparing at 600A is exactly the thing I did.

I do not know exactly the topology of the Relcon-Drive but from the posted displacement power-factor of 0.65 I have concluded that it is not a PWM-drive.

But if you compare a PWM-drive using GTOs with PWM-drive using recent IGBTs for 400 to 500 V line voltage the result will be quite similar.

All manufacturers of drives in the voltage and power range under discussion have changed their designs from GTOs to IGBTs already some years ago.

 

[cbarn24050]

Hi electricuwe, I guess 1 of us can't read, the data sheet for that igbt shows over 3volts vce at 600A at 125 degC. Drive makers have swiched from gto's to igbt but not because of conversion effiencies.

 

[jraef]

Thanks all for the participation in this discussion. I am getting a lot out of it from everyone's input. For the record, all of the issues about reliabbility etc. are exactly how we convinced the user that he needs to replace this drive, however he is a cheapskate to a certain extent and wants the power utility to help him pay for it, hence the efficiency issue. My belief is that the utility knows we will not be able to show a 10% improvement, which of course gets them off the hook for contributing (and frankly I don't blame them).

I am supposed to get the results of a weeks worth of Dranetz meter readings by the end of this week. I'll post the results.

"Venditori de oleum-vipera non vigere excordis populi"

 

[cbarn24050]

Hi jearef, only time will tell if the new drive will still be going in 20 years time. Sounds to me like you did a nice "job" on your client.

 

[jraef]

cbarn,
You are of course technically correct about how only time will tell.

By the fact that you put "job" in quotes, I think you are implying that I am trying to put a snow-job on this user about buying a new drive. Nothing could be further from the truth. I am a firm believer in the concept of "if is isn't broke, don't fix it". That however is not the case here. This old drive breaks down on a regular basis and has for years. The one and only electrician who has kept it running it is the same one who installed it in 1988. He is also on the forefront of trying to get the customer to replace it now because he is going to retire and nobody else is familiar enough with this drive to repair it, nor can they get support on it from Siemens any longer. The last time Siemens sent out a technician to repair it it cost them nearly $10,000, close to the price of a new drive, and that was 5 years ago. We all doubt that given the latest spate of corporate layoffs that Siemens even has anyone left who can repair it, and even if they did the downtime to wait for them would probably cost too much.

"Venditori de oleum-vipera non vigere excordis populi"