50/50 vs 100-0-100 generator loading

[Mbrooke]

Ok, say you've got a site that needs 1000kw of power. Would it be better to purchase two 1000kw gensets running each at 500kw (1000kw for the failure or service of the other) or purchase 3 500kw gensets using one as a spare for service or failure of the other two?

Or simpler way to view it: two 1000kw gensets in parallel or three 500kw gensets in parallel. 1000kw load.


Two schemes- can't make up my mind which is better or why. Help.

 

[waross]

Standby or prime power?

Bill
--------------------
Ohm's law
Not just a good idea;
It's the LAW!

 

[davidbeach]

Four 600s or 750s. Start all four on loss of source, parallel all four. If they all start and parallel, shut down the highest hour machine after they’re all up to temperature. Use three to carry the load of two. If one shuts down start the fourth one. If the load gets low enough for just one, shut down the third. That’s the data center way.

I’ll see your silver lining and raise you two black clouds. - Protection Operations

 

[Mbrooke]

Prime power. Though I'm also open to any standby discussion.

@David: I like the way you think!

 

[catserveng]

Not just for data centers, several large biotechs setup similar, and a couple hospitals in my region. Some use all units same size, others have mix of sizes. One biotech just added a natural gas fueled unit for longer outages dues to PSP shutdowns.

Recent mine site had two larger units at 1600kW each, and a 450kW. During the day one or two large units run, after abut 11pm one large unit comes off line, then the small unit comes online and takes the load off the remaining large unit, then at about 0530, process goes other, large unit comes on, small units drops off, other large unit comes back on. If daytime load drops below load demand setpoint, then the second large unit will come off line. System was designed to meet their site load profiles and work schedules, modern controls sure make doing things like this a lot easier.

Some sites have what appears to be excessive capacity, there is a larger drug manufacturer here, average site load is about 4.8MW, when on emergency power the BMS system will trim non-essential loads so site runs about 3.4-3.6MW. When an outage occurs, all 5 2MW units start and come online, lots of extra capacity but helps deal with large inrush and starting loads, then as system stabilizes the load demand start stop becomes active and takes unneeded capacity offline.

Lots of ways to do it, just up to whoever is holding the purse most times.

MikeL

 

[davidbeach]

In a lot of respects a utility outage can be thought of in terms similar to a Taxi ride (anybody still remember those?). There's (at least) two components to the total fare; one is the flag drop charge and one is the mileage charge. In this case the cost of an outage, any outage, starts with the flag drop charge and then continues to accrue at the mileage charge. Different customers have different "fare schedules". Residential and most commercial customers have a very low flag drop charge, and probably don't really start to accrue "mileage" until the outage has gone on for some amount of time; if the room temperature remains acceptable and none of the freezers start to thaw it's much more an annoyance than a pain.

But there are the customers that just can't take any hit. That flag drop charge can be in the millions or tens of millions of dollars, all in seconds or less. That's what the UPSs are for. That initial outage never happened. While the UPSs are chugging along carrying the vital load, generators are starting and ramping up to pick up the UPSs and all of the critical support load (don't think you can run a data center without HVAC, but a few seconds are no big deal). At that point the "mileage" charge is simply the difference in power cost between the self generated power and the utility provided power. For some of those customers that can be a huge number, but that means that the utility power cost is already a very big number.

If your focus is code compliance, the simplest backup generation scheme possible is the only viable option. If even a few cycles of power interruption means flushing weeks of production and starting over there will be UPSs and a back up system that won't let the process ever see that outage. Neither could fathom nor afford the other's system.

I’ll see your silver lining and raise you two black clouds. - Protection Operations

 

[waross]

I once worked in a very large battery room.
It was part of the UPS system for a boiler feedwater pump.
It kept the pump running for the few seconds that it took for the diesel generators to start and come online.
The pump was 10,000 HP in a very large nuclear generating station.
Even a few seconds outage was not an option.

Bill
--------------------
Ohm's law
Not just a good idea;
It's the LAW!

 

[Mbrooke]

UPS? 5 2MW units? You guys get all the fun stuff.

Typically with me a simple 3 source open transition system keeps the taxi driver content

https://files.engineering.com/getfi...5-44d7-9487-ea1d67b3c094&file=Clipboard01.jpg

 

[FreddyNurk]

KAIC is kA interrupting rating right? How come there's all sorts of different ratings on the one board?

EDMS Australia

 

[Mbrooke]

Yup, kA. killo amps interrupting. The numbers correspond to the rating of each individual breaker rather than the board itself.

 

[cranky108]

I would think the kA rating would also apply to the board also, as the magnetic forces acting on the bus bar during a fault would cause a bending of the bus bar. So that bus bar would need bracing to handle that force.

 

[FreddyNurk]

cranky108, it generally does, which means that putting a breaker with less rating than the board is not acceptable.
Also one of the reasons I asked, particularly when there's a 10kA 200A breaker mixed in with a bunch of 42kA breakers.

EDMS Australia

 

[Mbrooke]

You can have lower rated breakers if they are approved as a series rating. Though I think the main must be in the panel itself for that to apply- not 100% sure.