110 Megawatt emergency generator ? 1

[hmchi]

A high-tech company is planning to build a manufacturing facility, when in full operation, will consume 220MW of electricity. The manufacturing process is very sophisticated involving precision control and manipulation equipment and high energy use at the same time. A black out would cause extensive damage and losses.

If a power outage does occur, the company wants to have at least half of the electrical power capacity available within the plant, or 110MW, to keep the plant running at reduced capacity and minimum losses. The question is how to build this 110MW of emergency power ?

The advice they currently received is to use 55 Diesel Generators, each rated 2MW, which is less than ideal --- the facility needs to be a multi-story building where space is a premium, let alone the high civil and mechanical costs.

What would you recommend ? Would you recommend gas turbines running in hot standby ? with DGs as black-start power ? What size machines would you recommend ? They must be able to have the generators picking up the load 10 to 15 seconds after the blackout.

They must be prepared to sustain a blackout lasting a week to 10 days.

Being a hi-tech company, the facility may be obsoleted in 5 to 7 years so they are not interested in investing in a utility class generating plant that may last 30 years.

Your comments will be welcomed.

 

[pablo02]

I'm out on a limb here, but 110 MW in 10 to 15 seconds seems to be a significant amount of power in too short a time span: I think the small DG's can come up to speed in this time frame, but I don't think larger units can do it (too much mass to bring up to speed)...

Instead of running combustion turbines in hot standby, has any consideration been given to partial self-generation with a load shedding scheme? (the critical lines would never even see the outage) [possibly using 3 to 4 aero-derivitive CT's]? [I don't know how fast these start, although 15 seconds sounds not long enough]

Or, since this is a "high-tech" firm, any consideration for a fuel-cell application? [Might even be some research $$$ available to reduce the project costs..]

You might also visit with some of the "micro-turbine" suppliers (they'd love to serve this), but, again, you'll need a significant number of them to get 110 MW...

 

[pablo02]

Another thought: has any discussions go on with the local utility? Possibly, they might be interested in installing local generation -- the customer footing part of the bill for the option to island the customer under conditions that might create blackouts or other disturbances... thus, the customer doesn't pay for utility grade generation, but gets it anyway...

 

[rbulsara]

Why bother buying power from the utility co. at all!

Just build a prime generating plant using turbines rated 25, 50 MW with one or two redundant units and diesel for black start and lifesaftey loads!

No point having a 110MW standby plant. On the contrary work out a deal with the utiltiy co to sell back excess power.

 

[hmchi]

The local utility is a government-owned bureaucratic monopoly who does not want to be bothered. When approached with your very idea, cost-sharing for a local generating station, there were several 'concerns':

[1] There was not a lot of confidence in this gen station in the hands of this bureaucracy --- its very own 'survivability' in the event of a blackout.

[2] The country has seen a number of IPPs and Co-gens, they seem less reliable than the government utility, which the IPP blames on the 'regulations' and operating conditions they were put under the government utility.

[3] The government utility says there is no cooling water available for a power plant.

 

[hmchi]

rbulsara, your idea was the first proposed. However, having problems convincing management :

[1]"We are not power plant people, our focus must be our hi-tech work. Why do we want to build and operate a power plant for 30 years ?"

[2]"Then how about finding a 'power plant people' [IPP] and sign a contract with them to use their generated power ?" Fine, but they have to be 'competitive', and we may close the facility in 5 to 7 years ...

 

[SidiropoulosM]

You have the alternative of strengthening the connection with the utility to improve the reliability of the utility's supply to you. This will depend on economics: how far the main grid substation is and how strong it is electrically.

 

[hmchi]

SidiropoulosM, the basic premise is that the utility system may be blacked out so that there is zero coming from outside ... You're right, it is a matter of dollars and risks ...

However, any thoughts on gas turbines that can pick up loads in 10 to 15 seconds ? or, failing that, 30 seconds ?

 

[Bbird]

There are experts here who may tell you more.

I work only as a civil guy in the power industry and here is my observation.

I think there is a huge problem if you insist 110MW within 10 to 15 seconds response time. At this output the nearest I came across is from spinning a hydro set at sychronising speed in air (empty casing) and then open the inlet valve to let the water through in 10 seconds.

My colleagues advise me that aero-derivative gas turbines (off airplanes) are the fastest but you are still looking at minutes. You can get one gas turbine to supply the full 110MW but it is unlikely avaialble from the lightweight aero-derivatives. Commercial GT sets can go up to 250MW but they are infrequent.

The idea of 55 sets of 2MW high speed diesel generators seems possible and certainly give the power within the specified time but you will have one hack of a job to sychronize the 55 sets. The integration of the standby power to plant main switch boards must be planned well ahead. It also seems to me the emmissions from the protracted standby power generation of 110MW in space-restricted zone may present some environmental issues.

I believe the most likely successful solution would be to package your need in integrated blocks of power to make up for the 110MW using different generating sets at different response times.

From my consulting engineering background it appears that professional help is unavoidable in your case. My company helps clients to install all kinds of power generation (I personally involved up to 4000MW) and transmission systems and is major player worldwide. You will need to seek advice from a company (even if it is not us) that has experience at least in diesel sets, gas turbines, electrical control, supply stability, economics, civil work, structures and enviornmental engineering.

If you need assistance formally I can put you in touch with the right engineers.

 

[aali2208]

hmchi,

I personally think that the solution will be best with at least two portions of Power Generation Partial power you can take from utility company and for the essential power you have to set up a Power Generation. Like in yr case you have requirement of 250 MW total, combined on 110 Essential & 140 non Essential loads. If you can set up an agreement with some power producer (IPP) to provide your essential power and get the rest from utility (Best way will be through HV Net work (110,132 or what ever transmission is there). And set up distribution in such a way by using Intelligent load management scheme that in case of failure of one source you can the load management scheme can shut down the normal load within 3-4 seconds and the couplers between two sources can connect the load buses togather.
The details can be further workout with some good consultation in yr country.
The other option we had worked for one of the suphisticated chemical plant power solution was loop in Loop out arrangement with Utility company. The No.s of circuits are normally connected from Power stations to the distribution grids and most of them supplying to the buses which can be coupled at load grids. ( different configurations for different locations) If you can have the option to connect your consumer Grid from one of the circuit break in such a way that one of the feeder you will get from Generating station and other you will get from the distribution Grid then you will have quite secure condition. The only chance of failure will be in the case when utility loss the whole Grid.
Hope this will be help full for you,

Regards,
aali

%-

 

[peebee]

I'd investigate larger units than 2MW. I understand that units of up to 5 or 10MW are available, although I've never actually seen that documented.

Paralleling for capacity is just about out of the question given the 10-15 second response time you're looking for. It takes time for gensets to parallel. For any quantity of units in parallel, you're probably talking minutes, not seconds, at best. Increasing the size & reducing the quantity of paralleled units will help.

Paralleling for redundancy (not capacity), is a great idea, though.

Breaking the generation system down into smaller separate chunks which pick up only small portions of the load will increase the response time. But possibly at the cost of reliability.

I agree that agreements with the utility to provide reliable power should be rejected. They appear to make great economic sense, but can only lower reliability. Similarly, I'd reject continuous long-term prime generation as an option, unless you go completely self sufficent (2N generator redundancy or better), I'd be too scared that the utility capacity would not be there in a pinch when you needed it.

If you did decide to go prime generation, only then would I say that waste heat recovery made sense.

Turbines generally don't make sense for standby applications.

Fuel cells don't yet seem to make sense in any application (unless marketing is playing a BIG role).

Can you back off on the 10-15 seconds? That would help a LOT. Maybe cooling water capacity can be increased to provide a longer duration of emergency cooling prior to chillers coming back on line.

Good luck.

 

[BJC]

hmchi
I would check the load the company is proposing. 220 MW sounds very large. I have designed three chip plants (one of which is still one of the 10 biggest in the world.) The biggest connected load on any of them was less than 30 MVA.
I would bet a large lunch that 220 MW would power ever chip plant (and the ancillary support plants) within this area. That would include three large Intel plants, one IDT plant, an LSI wafer fab and several others.
It takes a very detailed review to determine what tools and equipment in a plant can take a power outage and for how long.
The first thing I would recommend is separating life safety from standby power. Life safety is the power needed to run the exit lighting, fire detection and protection, and those devices needed to get people safely out of the building. In a chip plant this could include systems and component not found in a "normal" facility including some tools that might release poisonous gases and the components in the gas distribution system as well.
The rest of the system is not legally required stand-by power. Everything else in the plant needs to be evaluated and to determine the consequences of that particular device loosing power.
Loosing power can result in loss of product in process or contamination of the "clean room" part of the facility; both can be very expensive.
Here are a couple of examples of where we have saved the requirement for "emergency" power in the past.
1.Many tools have more than one connection point for more that one level of power. They may accept a short outage for 480-volt power but require an uninterrupted 120-volt source. Look at each tool to see what it needs to keep operating, shut down with out loss and decide which of the above is acceptable (it may be neither).
2.) Clean room fans can sometimes be shifted to 1/2 speed and use less power in a power loss. It may be possible to turn 1/2 of them off if they have a common plenum.
I put "emergency" in quotes above because if you survey production and operations people everything they have (including their coffee pot) is "emergency".
Typically chip plant require the following systems:
1.) 480 volt on DG back-up
2.) 480 volt on CPS (CPS are typically fly-wheel or Holec gensets.
3.) 480 volt on UPS (no interruption)
4.) 120 volt on DG back-up.
5.) 120 Volt on UPS.
6.) 4.16 kV -may be needed on DG (essential chillers may be 4.16 kV or 480v).
I would also recommend "training" all electrical systems. That is dividing them in two separate systems ( "trains" or "divisions&quot, each of which will function alone and in theory keep 1/2 the plant running.
If you really need 220 MWs, gas turbines are the cheapest, your would still need some Diesel generators for loads that need to be picked up in 10 seconds or so.
Where is the plant? There are several threads going on in eng-tips.com about jobs going overseas. I may be shot for offering this limited advise.

 

[peebee]

Here's a thought on how to configure the system: Try to break everything into small cells or modules that all have about the same power rating as a generator, say 2, 5, or 10MW. Provide UPS systems rated at 100% or 50% of that cell size. Similarly, provide cooling system modules with power requirements of 50 or 100% of that cell size.

Then, two options:
1. Provide 50% UPS and 50% cooling in each cell, total 100% load equals 100% of genset rating.

2. Provide 100% UPS on one genset, provide 100% cooling on another genset.


Parallel lots of generators together, but provide bypass breakers around the paralleling gear. All generators are normally in bypass, picking up only its own dedicated cell's load. Normally, no need for instantaneous paralleling, all gensets would be up to speed within about 10 seconds.

In the event of a genset failure, only then would paralleling need to be attempted.


Better have some good long talks with your generator and switchgear vendor before you try that one out, though.

 

[hmchi]

peebee, Thanks for your input.

First of all, what I did not elaborate is that they cannot withstand more than 10 milliseconds of loss of voltage. That is why there will be a UPS totalling 110MW bridging between the utility power and the generators. The UPS can last 30 seconds at full load, 60 seconds at half load.

You said :

Breaking the generation system down into smaller separate chunks which pick up only small portions of the load will increase the response time. But possibly at the cost of reliability.

This is what they are thinking --- breaking down the essential loads into 10 MW blocks and run 5 generators in a bus. The other thought is --- to use the UPS to provide a sych signal for the generator to start and sync. Question, do you guys think this will keep all the 11 groups of gen's synchronized ?

 

[tomatge]

Phew - some detailed responses above. May I add my thoughts.

Assuming the demand off 220MW is accurate (please read BJC's comments carefully), and you require to maintain 110MW of emergency power, then I don't believe any method of conventional generation in standby mode will help. The possibility of starting, synchronising and controlled loading of 110MW in less than 15 seconds is positively mind boggling. Attempting to do it with 55x2MW diesels is, in my opinion, impossible.

You can get multi-MW UPS systems, but I've never heard on any UPS going up to 110MW. Physically it would be absolutely enormous.

You are faced with the solution of running 110MW of generating plant in parallel with the grid all the time. Should the grid infeed trip, you can then maintain you emergency demands with the assistance of some fast acting load shedding. However, you are now into the realms of becoming a base load generator, which as a precision manufacturer is possbly not your field of expertise.

It would be advisable to consider types and size of plant. You would not want one generator, since failure or maintenance of that would make you entirely dependent on the grid again. Conversely, you would not want to be running 55x2MW generators in this mode, the maintenance would be enormous.

If you have requirement for process heat (either on your site or on others nearby) you could consider cogeneration plant - something like 4x30MW gas turbines with heat recovery. If you don't, then base load plant operating on cheapest fuel such as heavy fuel oil or coal.

When you consider the difficulties, it may provoke you to review the problem :

1. Are the power estimates realistic.
2. Is the "uninterruptible" requirement actually that demanding.
3. Can the problem be solved by greater levels of reinforcement from the grid.

Hope this helps ... but I bet is doesn't really.

 

[Bbird]

Our diesel engineer told me that upto 60MW diesel generator is possible but the amount metal involved will not enable it to run properly in 10 to 15 seconds. Neither any owner would want to hear from me how much the foundation will cost him.

Can someone here gives us an indication on the number of staff needed to run a 110MW generation?

The fuel system for supplying 110 MW generation will be substantial. Even gas is right at the door step some kind of conditioning station will still be needed to provide the throughput needed by the gas turbines.

Currently cost to the owner of building a single cycle GT station is about US$400 per kW if he has fuel pipeline at the door step.

 

[hmchi]

BJC,

You are no doubt an old hand at these hi-tech plants. Then you would know that these very secretive guys usually bark up very high requirements for 'electrical pipes' and then settle down to earth later. However, even if the actual requirement is half, it is still a giant system and all the discussions we had so far would still apply.

You must also know that these hi-tech guys always want the plant built yesterday, so you do not have time to engineer it right, as you indicated, going through all the loads and group them according to your meticulous classifications.

They usually just copy what they built the last time ... You should know that well ..

This is probably why it is less expensive for them to over-spend on UPS and emergency generators than taking the time to do it right as you suggested .. [The process guys are not available to you for this kind of analysis for IP protection reasons ..]

This plan is for one CUP for at least 2 phases, why the 110MW.

Almost all the latest fabs in Asia has the 1st floor completely occupied by gen sets and dynamic UPSs, making them looking like a power plant from the outside. They usually have up to 20 gen sets, and up to 10 sets of DUPS.

Exporting jobs ? The biggest fab design house is in Europe and they are all over Asia in each major cities. US investment in the chip semi industry overseas is a very small part of the market.

I would agree that the software migration to India is totally another matter ...

 

[tomatge]

The biggest diesel engines in the world are about 50MW. But they are BIG. They certainly cannot provide start, synchronise and loading in a few seconds (more like a few hours). If you want information on large diesels, check Wartsila (which own Sulzer) or B & W.

Simple cycle GT is possible, and at £400 per kW sounds not too expensive, But running costs will be high with an efficiency of less than 35% (even if gas is cheap and plentiful).

Soembody commented that aero-derivative GTs are fastest, and I dispute this. Aero-derivatives are typically relatively lightweight in construction, and don't like having rapid or large step load changes stuffed on them. Faster and more robust GTs are the industrial types (e.g. try

http://www.industrial-turbines.siemens.com)

but these are naturally more expensive.

Regarding number of staff, it depends upon type of plant, fuel, maintenance contracts etc. The problem here seems to be that the customer only wants the plant to last a few years, so may be buying used plant, so warranties etc. may not be available.

 

[ScottyUK]

Tom at GE (?)

I agree that the big industrial turbines handle step load changes better than the small aero-derivatives, but they have a long start-up time. The aero engines can be at synch speed much more quickly than the frame machines, but as you say, do not like big load steps. For example, our MW-701DA's (170MW shaft output industrial turbine) have a 20min acceleration sequence to get to synch speed of 3000rpm, and then another 5min to get to 110MW on the fast load rate. For comparison, the GE LM6000 aero-derivative we use as a black start turbine can be up and accepting load in less than half that time.

Both are too slow for this application, unless a source to hold up the load for a few minutes can be provided. 110MW off a UPS? That is going to be *big*.

 

[Laplacian]

I would definitely re-verify the true requirement is 220MW. We run a 500K barrel/day refinery and chemical plant at ~400MW. It is mind boggling to conceive a "high-tech" industry plant coming anywhere near that size of industrial complex.

I've seen people size UPS systems for the power supply ratings of the computers they backed up which is not realistic.

If the power requirement is accurate and outages are really that unacceptable, then cogeneration is your only option from a technical/maintenance/reliability standpoint.

Good luck in your quest for an answer, this is a tough requirement to meet. It is not feasible to have 110MW on tap at the flip of a switch soley for backup purposes.