Real-time electricity usage in Texas is graphed by ERCOT, the Electric Reliability Council of Texas. Today's graph can be found here, with a screen-grab below:

Note the difference between the lines and the bars. There are three lines: yesterdays forecast for today, today's forecast for today, and today-actual. According to the help menu, the lines are for the "system load". I quote: "The System Load (green line on graph) represents the current system load. It updates at 15 minutes after each hour." (the green line is hiding under the blue one). Fine. Very interesting.

What about the bars? Two sets of bars: forecast vs. actual. I quote: "The Current-Day HSL (lighter blue bars on graph) represents the current HSL. It updates at 7 minutes after each hour." Oh, what's "HSL"? I quote, it is the "Hourly System-wide Load (HSL)" Super duper dandy.

My question is: what's the difference between the system load, and the hourly system-wide load? Why aren't these exactly the same, up to rounding errors? I can't imagine how to integrate, differentiate, average, Laplace transform or apply any other tricks, that would explain this difference, that would explain how they are related.

I do notice that the difference between these two is approximately equal to the solar+wind generation. But only approximately: at 6AM, there is a 20GW difference in this graph, whereas combined wind+solar was generating 12GW. So that's not it.

• The link is inaccessible from EU. But hey, I am curious, too! Our grid operator, for one reason or another, doesn't publish current-day graphs. It does publish both past and forecast ones, but no complex features like ERCOT, just a MW curve and it is way more complex than your near-sinewaves. – fraxinus Jun 16 at 20:07
• Could it be that the bar graph is the transmission lines total load? It may be then higher than the total electricity produced/consumed, because some of the power passes more than one line and participates in the sum more than once. And by the day local generating powers kick in, lowering the percent of the power transmitted from far away? Just speculating. – fraxinus Jun 16 at 20:17
• Make sure you use http not https to access the link! Alternately, you can get to it from the home page. The sinusoid is an accident for today, it's usually more complex. Summertime, it's dominated by mid-day air-conditioning. Wintertime, its more erratic. – Linas Jun 16 at 20:19
• Nope, it says my IP address is not entitled to access even the ERCOT homepage. As if I am a Russian hacker. – fraxinus Jun 16 at 20:26
• @fraxinus ... wow. [political commentary elided.] I just tried TOR, and it loads, because (by luck) my current TOR exit node is in the US. – Linas Jun 16 at 20:53

That appears to be a typo in the help for the graph. From the ERCOT glossary:

HSL:

High Sustained Limit.

High Sustained Limit (HSL) for a Generation Resource:

The limit established by the QSE, continuously updated in Real-Time, that describes the maximum sustained energy production capability of the Resource.

High Sustained Limit (HSL) for a Load Resource:

The limit calculated by ERCOT, using the QSE-established Maximum Power Consumption (MPC).

Compare the difference between the bars and the lines to the chart of available capacity:

Just eye-balling:

• At midnight:
• HSL = 63 GW
• System load = 46 GW
• Difference = 17 GW
• At 1pm:
• HSL = 71 GW
• System load = 62 GW
• Difference = 9 GW

This matches but is a bit lower than the blue line, which is HASL. From the glossary:

High Ancillary Service Limit (HASL):

A dynamically calculated MW upper limit on a Resource to reserve the part of the Resource’s capacity committed for Ancillary Service, calculated as described in Section 6.5.7.2, Resource Limit Calculator. HASL is also included in Section 5.7.4.1.1, Capacity Shortfall Ratio Share, and in the Reliability Unit Commitment (RUC) optimization but is not adjusted for Non-Frequency Responsive Capacity (NFRC) as in Section 6.5.7.2.

This training presentation explains the relationship between HSL and HASL:

• Bingo! I've even looked at that SCED graph months earlier, and read about ancillary services, and surely saw HASL before. Never made the connection! – Linas Jun 16 at 21:22

I'd imagine that the grid can be pretty choppy as loads switch on and off which would cause a lot of variation in the signals.

I think HSL would be peak load for the hour and system load would be the average, this data has been 'binned' by hour and would make more sense if you could see the actual load on a 5 minute or 1 minute basis.

Edit: Here is a sample of the actual data over the course of a year.

• I find this hard to believe. The difference between trough and peak is 30GW. This is almost entirely air-conditioning. At 6AM, it was 70 degrees outside, its about 95 degrees now. So, 30GW of A/C kicking on between 6AM and noon. Fine. But where would a 20GW load appear from, and then disappear, between 6AM and 7AM? I mean, 20GW is huge, compared to the base load of 40GW. – Linas Jun 16 at 19:33
• @Linas I don’t find it hard to believe - some industrial places can have very high loads. A paper mill I worked at had to inform the elec company of its start-up time after a shutdown due to the instantaneous load - they forgot once and there was a power outage... then a large fine :) – Solar Mike Jun 16 at 19:52
• It may not be moving up and down 2GW, but it could with all the variable loads, so the HSL would be the peak value – Voltage Spike Jun 16 at 19:56
• A peak load almost twice the hourly average (at night) is still pretty hard to beleive when averaged over the whole TX. – fraxinus Jun 16 at 20:12
• You are misreading the eia.gov graph -- it is for four years, 2011 to 2015, showing weekly peaks. – Linas Jun 16 at 20:25