"Noncontinuous" defined?

[iwire]

If the above is true, then there is a lower limit to noncontinuous, which perhaps someday will be addressed in the NEC, and I think it would be better for it.

I really do not understand the problem or issue you feel needs to be resolved. :?

 

[kwired]

All of those "inrush current" applications disregard the inrush current and use the full load current for determining minimum circuit ampacity. One must consider the inrush and trip/time curves however when selecting overcurrent protection, and you can get different trip/time curves from different overcurrent devices with same general rating - take a 20 amp overcurrent device - you get different trip curve from different product lines/types but they are all still considered 20 amp overcurrent devices.

 

[dinos]

I believe that the original poster's question could be rephrased:

"How does the NEC calculate short duration and intermittent loads? Is there an explicit definition for how to calculate loads of very short duration, such as startup transients or inrush currents?"

The 'continuous' versus 'non-continuous' question relates to the calculation of breaker size for a circuit ("100% of non-continuous loads + 125% of continuous loads").

As I mentioned in my previous responses, I don't believe that the NEC explicitly approaches this issue, but rather implies this with the sizing rules for different types of circuit. The rules for sizing the breakers for transformers deal with their inrush currents, similarly motor circuits are sized to accommodate starting currents.

I understand there are sizing rules for 'dedicated' circuits, but it is mixed use feeders that I am trying to get a better feel for.

The genesis of my post relates to an existing feeder that serves many many different loads and the question arose "how much more load can we add?"

Surveying/documenting all of the connected loads and performing an analysis is one path.

I thought perhaps a load recording over a sufficiently long period of time might be an alternative as the combined continuous+noncontinuous loads would be evident by analyzing the data.

I have attached a load recording graph as an illustration.
I realize that one would have to be careful to ensure that the worst case load condition was included (e.g. summer months if significant cooling loads are served) but perhaps reasonable values for continuous and noncontinuous could be gleaned from the data.

This particular recording included maximums based on 12 cycle intervals and based on one minute intervals.
The data could also be massaged to make the intervals 15 minutes, one hour, 3 hours, etc.
This is where I started questioning..."what is the shortest interval I should consider for the noncontinuous load component in the rating calculation?"

I realize this might be a non-standard approach, but I though it might be valid nonetheless.

 

[kwired]

I understand there are sizing rules for 'dedicated' circuits, but it is mixed use feeders that I am trying to get a better feel for.

The genesis of my post relates to an existing feeder that serves many many different loads and the question arose "how much more load can we add?"

Surveying/documenting all of the connected loads and performing an analysis is one path.

I thought perhaps a load recording over a sufficiently long period of time might be an alternative as the combined continuous+noncontinuous loads would be evident by analyzing the data.

I have attached a load recording graph as an illustration.
I realize that one would have to be careful to ensure that the worst case load condition was included (e.g. summer months if significant cooling loads are served) but perhaps reasonable values for continuous and noncontinuous could be gleaned from the data.

This particular recording included maximums based on 12 cycle intervals and based on one minute intervals.
The data could also be massaged to make the intervals 15 minutes, one hour, 3 hours, etc.
This is where I started questioning..."what is the shortest interval I should consider for the noncontinuous load component in the rating calculation?"

I realize this might be a non-standard approach, but I though it might be valid nonetheless.

Did you read 220.87? Kind of on same track of what it allows with what you just described.

 

[dinos]

Did you read 220.87? Kind of on same track of what it allows with what you just described.

kwired...thank you for noting that...I had not seen that section and it speaks to what I was looking for.
I need to consider and digest...

 

[Iron_Ben]

I understand "3 hours or more" for defining continuous loads, but is there a defined limit concerning "shorter duration" noncontinuous loads?

For example, I would expect that transformer inrush (lasting on the order of 0.1 seconds) might not be considered noncontinuous, but would motor inrush be considered noncontinuous?

I believe that you are overthinking this, especially with regard to transformer inrush and motor starting currents. Consider that the standard time interval for electric utility demand recording and billing is 15 minutes.

Let's try an *extreme* example. Not likely, but hopefully illustrative. We have a customer whose peak load (and only load) let's say is 100 kW. But this customer operates infrequently, in fact just once a month. Not only that, but he only operates for 7.5 minutes at this level.

When we bill this customer we bill him for his peak demand based on the highest average power in the worst case (highest usage) 15 minute segment of the month. Here we would bill him not for 100 kW of demand at $10 or so per kW, but for 50 kW as the load was only on for half the 15 minute interval. (He would also be billed for 12.5 kWh of energy, but that's not part of this discussion.)

Perhaps you think electric utility billing and metering and load information is different from the NEC perspective and I agree in many ways it is. However, know that we made all loading and overloading decisions for high and low voltage cabling, transformers, etc. based on this data. This hypothetical customer would be in our system as a 50 kW max demand user. The fact that he peaked at 100 kW and stayed there for 7.5 minutes would be unknown to us unless we had made a site visit and observed the actual load characteristics. And even then, we wouldn't change him in our system model from a 50 kW peaker to a 100 kW peaker.

We would occasionally encounter customers who thought that motor starts contributed to high demand numbers. When we explained the 15 minute recording interval to them (not to mention the fact that a motor start is highly reactive) and how short a start is by comparison they were usually able to see the light.

 

[winnie]

Something else to remember: the NEC is supposed to provide 'practical safeguarding', and is not supposed to be a design manual.

The NEC might not consider extremely short duration transients as a safety issue that you need to calculate, but your users might consider such short duration voltage drops (caused by the transients as supplied through system impedance) as quite a nuisance.


Iron_Ben: these 15 minute segments of which you speak: are they fixed to clock time (eg. 1:00 to 1:15PM, 1:15 to 1:30PM...) or do they 'roll', meaning that if (say) 1:04:55 to 1:19:55 had highest usage, then that would be where demand is calculated?

If your extreme user timed their 7.5 minutes of usage just right, could they have a demand charge for only 25KW?

-Jon

 

[dinos]

I believe that you are overthinking this, especially with regard to transformer inrush and motor starting currents. Consider that the standard time interval for electric utility demand recording and billing is 15 minutes.

It's interesting that you bring up the 15-minute utility demand billing interval...

I think NEC 220.87 echoes that idea. It also appears that the 'continuous vs noncontinuous' distinction goes away as they are essentially "rolled-up" in the 15 minute demand figure. Perhaps the "new load" is also looked at in terms of its 15-minute demand contribution rather than in 'continuous vs noncontinuous' terms.

 

[winnie]

dinos: keep in mind that code makes specific use of the concept of 'continous' versus 'non-continous' for the sizing of OCPD and circuits, and that this usage has nothing to do with short term transients.

It looks to me like 220.87 essentially says that _existing_ load may be measured with 15 minute averaging to account for any short duration transients, but that added load has to be calculated using normal calculation tools.

Note that any of the short duration transients that we've been considering will essentially vanish over 15 minute averaging. Even a 20X normal consumption transient that lasts for a whole second would be a 2% change in average consumption.

So we are back to: the normal calculation tools of the NEC do not consider start-up transients are part of the load that needs to be calculated. For purpose of the NEC calculate based on continuous loading; for specific circuits with known significant transient issues, the NEC requirements for those circuits will accommodate the transients (motor circuits, transformer protection, etc).

Then consider transients are part of good design (not NEC requirements), so that (for example) your computers don't shut down every time the compressor starts.

-Jon

 

[Iron_Ben]

Something else to remember: the NEC is supposed to provide 'practical safeguarding', and is not supposed to be a design manual.

The NEC might not consider extremely short duration transients as a safety issue that you need to calculate, but your users might consider such short duration voltage drops (caused by the transients as supplied through system impedance) as quite a nuisance.


Iron_Ben: these 15 minute segments of which you speak: are they fixed to clock time (eg. 1:00 to 1:15PM, 1:15 to 1:30PM...) or do they 'roll', meaning that if (say) 1:04:55 to 1:19:55 had highest usage, then that would be where demand is calculated?

If your extreme user timed their 7.5 minutes of usage just right, could they have a demand charge for only 25KW?

-Jon

Winnie/Jon:

The answer to your last question, at least at the POCO I retired from, is "Five or more years ago, yes. Today, no." We changed the programming on demand meters from "block" to "rolling". Other POCO's might still use block, which is a better deal for the customer and could allow this hypothetical to use 100 kW for 3.75 minutes at the tail end of interval no. 1 and 3.75 minutes at the beginning of interval no. 2. Good question!

 

[dinos]

dinos: keep in mind that code makes specific use of the concept of 'continuous' versus 'non-continuous' for the sizing of OCPD and circuits, and that this usage has nothing to do with short term transients.

It looks to me like 220.87 essentially says that _existing_ load may be measured with 15 minute averaging to account for any short duration transients, but that added load has to be calculated using normal calculation tools.

Note that any of the short duration transients that we've been considering will essentially vanish over 15 minute averaging. Even a 20X normal consumption transient that lasts for a whole second would be a 2% change in average consumption.

I hear what you are saying, but am still left somewhat confused about the intent/interpretation of noncontinuous per the NEC.
I can get past transformer inrush and motor starts, but at a certain point (undefined at present imo) the magnitude and duration has an impact that has to be taken into account.

I sketched 4 different scenarios (attached), all of which have the same result...size the OCPD at 400A.

I might say of the scenarios to the right (involving the 5 minute noncontinuous loads) that their effect would be small over a 3-hour averaging period, much like a 20X, 1-second transient would vanish into a 15 minute interval, but the end result (at least as far as I can see via the NEC requirement) is no different than those at the left.

Maybe the ones to the left prevent device overheating & time-current-characteristic issues, while the ones to the right just prevent device time current characteristic issues...