Calculate demand from periodic power readings

I don't have an exact link to what you want, but I believe you can find an explanation on :

Charles Perry P.E.

I could not find anything on the subject. It was an interesting site though.

I have been searching for days and I can't understand why this information is so hard to find. All digital demand meters must use this formula for demand calculations. Many energy management systems must also use this calculation unless they read demand straight from the meter.

There must be documention on the formula someplace.

Chuck

First of all, if you are reading kW (rather than accumulated kWh), you aren't reading the same thing that revenue metering uses for demand.

If you store the kWh readings every minute, look back in memory for the reading 15 minutes ago, subtract the two and divide by 4, this is a sliding or rolling demand.

If your metering is measuring kW (instead of kWh), you have to integrate the readings. The sample rate used will influence the accuracy of the resulting figures, depending on the characteristics of the actual load fluctuation.

In the end, the definition of figures used for revenue calculations should be traceable from the filed tariffs.

Interesting, I didn't know demand was calculated from energy values.

By the way, mathematically it seems like that should be multiply by 4 rather than divide by 4. Consuming 20kWh in 15 minutes, I would think you would 'project' the use of 80kWh in an hour, for a 'demand' of 80kW.

Where I am, demand is based on kVA rather than kW.

I looked up some very comprehensive tariff schedules

Paul, you are correct that I am trying to calculate demand from periodic kW readings. My concern and something I have seen in real life is that the kWh readings don't change very rapidly so this can cause the calculations to have a low resolution for small loads.

For example if you have a variable load that consumes 5kWh every hour, the kWh meter will only flip over a new number approximately once every

With many digital power meters, the kW reading goes down to a resolution of 0.001 kW or 1 watt. I realize that if a take a set of kW readings every minute, I can only approximate the demand by feeding the periodic readings into a calculation but I feel this is better for analysis purposes than using kWh readings.

Any suggestions would be appreciated.

Chuck

It is for integrating demand meters. Thermal demand meters measure kVA (they are essentially voltage compensated ammeters with a time constant that approximates a 15 minute demand. Back in the old days, these were cheaper than integrating meters, but in my area, there is only one utility that hasn't ripped out all of its mechanical, manually read meters with remote metering systems. Once the microprocessor is in, the cost of integrating metering is essentially zip. They can (but don't yet) report back 15 minute demand figures even for residential meters.

You're right. I think 'divide kWH by time to get kW', but the time in this case is 1/4, not 4.

It depends on what you're being billed for. Large commercial and industrial customers (where I work) are assessed a demand charge per kW, not kVA. They are also assessed a reactive power charge per kVARH. One utility has set its real and reactive demand charges for HV (transmission voltage) service customers to be the same, so there its possible to base charges on kVA. But for customers of this size, the bills are large enough to justify integrating metering.

Actually, what goes on in the meter is as follows: A CL200 meter (mechanical disk type used for most residential services) has a meter constant of 7.2 Wh per disk revolution. So, for a 5 kW load running for

Its better than counting disk revolutions, but not by much.

If you have a digital power meter already, then that's the way to go. But why limit yourself to one reading a minute? Take one every second, or 5 seconds. Computing power is cheap.

I am not interfacing with spinning disk type meters. I deal with sub-metering where the metering function of the meters is not great. These meters are all electronic. Furthermore, I don't care if the readings are kW or kVAR as the calculations would be the same to calculate demand.

And you are correct that the meters can perform volts, amps, and kW, but many times do not produce kWh readings. And if they do, the kWh readings are sufficient for once a month readings but not 1 minute periodic readings because of the depth of the resolution.

Ok, so to summarize, I have determined that I should take one minute or quicker readings of kVA or kW and log them to a database. I would then do the following on the data:

-15 minute kW demand is calculated by measuring the kWh used in 15 minute periods and multiplying by 4. In other words, 20 kWh consumed in one 15

minute period equals 80 kW demand. If this is calculated once every 15 minutes, this is the block demand.

-If you store the kWh readings every minute, look back in memory for the reading 15 minutes ago, subtract the two and divide by 4, this is a sliding or rolling demand.

These two methods sound very easy to implement from a software side. But these mathematical methods don't simulate the old elecro-mechnical demand meters: do they? Is this method legal for billing calculations?

Paul mentioned a "with a time constant that approximates a 15 minute demand". Is there someplace this constant formula is documented? I believe what you are referring to is the 90% reading after fifteen minutes to simulate thermal demand. Is this method not used for billing calculations anymore?

Also mentioned, I can take the readings fairly quickly but there are other considerations like bandwidth between the meter and computer. Also, the local computer/collector may not even be a full computer with a lot of power.

Thanks for all the help. I am new here but this seems to be a very helpful newsgroup.

Chuck

Ah...

But storage or data transmission is not always cheap. Working a project in the DTE area, we had to compromise since the meters we were using used POTS lines to 'phone home'. At just 9600 baud, we wanted to be able to download

daestrom