Many power traders and analysts study different factors that impact the dynamics of the electricity market. Factors like seasonality, time of the day, and unusual weather conditions can affect the load, which will in turn affect prices. In this blog, I analyze recent temperature fluctuations in the U.S. to find a correlation between electricity prices and temperature.

The second week of July 2014 was unusually mild in the U.S., and these cooler temperatures reduced pressure on the power grid, as Figure 1 indicates. As we can see, the aggregate power load lowered in this period when temperatures dipped by two degrees Celsius (C). This trend illustrates an industry assumption: that temperature and power consumption are correlated. Let’s examine this assumption in ZEMA, applying relevant formulas and the latest data reports from seven Independent System Operators (ISOs): CAISO, MISO, SPP, ERCOT, NEISO, NYISO, and PJM.1 These ISOs cover about 75% of the U.S. population. In order to better understand this occurrence, the observation will be performed over a larger sample size.

Figure 1: U.S. Aggregate Power Load  vs. Temperature (July 7-11, 2014)

Figure 1: U.S. Aggregate Power Load vs. Temperature (July 7-11, 2014)

Given how temperatures fluctuated in the last six months in the U.S., reaching both ends of the scale, it would be interesting to analyze the correlation between temperature and the aggregate load in this period (from mid-February 2014 to mid-July 2014). Temperature fluctuations based on normal seasonal patterns are expected, but unexpected weather patterns do drive power consumption and cause prices to skyrocket.

Correlation of Temperature Fluctuations and Power Demand

The correlation coefficient tells us how one indicator is changing with respect to another. In this case, I am analyzing the relationship between temperature and power consumption. When put in the context of correlation, the relationship between temperature and power consumption is negative in the winter –that is, when temperatures go down, demand goes up–and positive in the summer–when temperatures increase, demand does too. The graph below (Figure 2), created with ZEMA, represents this relationship from February to July 2014 in the U.S.

Figure 2: U.S. Power Load vs. Temperature (February - July, 2014)

Figure 2: U.S. Power Load vs. Temperature (February – July, 2014)

 

During this period, the land of the free experienced weather conditions where temperatures were reaching their end points. Within the 20-day period from February 1 to February 20, the average “real feel” temperature in the U.S. reached its pinnacle. In this period, temperature movements and power demand were highly linked, as the persistent cold maintained the correlation between the temperature and power consumption above -0.8 levels. Further insight can be gained by examining California and New York in 2014.

Mediterranean Climate – California

Data obtained from CAISO and AccuWeather (Figure 3) reveals that in California, the correlation reached a peak value of 0.901 in the month of May 2014. As expected, Californian temperatures were warm, so the correlation is positive.

Figure 3: California Power Load vs. Temperature (Feb - July, 2014)

Figure 3: California Power Load vs. Temperature (Feb – July, 2014)

Humid Continental Climate – New York State

Residents of the state of New York experienced both high and low temperatures recently. As opposed to California, where the correlation is mostly positive due to hot temperatures in the summer, New York temperatures and consumption in this period correlate similarly, both positively and negatively. Figure 4 reveals that very low temperatures (-10 C) fueled power demand, making the correlation reach beyond -0.8. When the temperature reached 25 C, the correlation between the two indicators was positive, reaching beyond 0.8. Both of these cases prove that the relationship between temperature and power consumption is stronger in the event when temperatures reach their end points.

Figure 4: New York Power Load vs. Temperature (February - July, 2014)

Figure 4: New York Power Load vs. Temperature (February – July, 2014)

ZEMA helps power market participants track correlations in a visually dynamic environment. Further, ZEMA possess in-built analytic formulas that can be fed with data from many database sources. Using ZEMA, testing assumptions like the one described in this blog is easy. To learn more, book a complimentary ZEMA demonstration today at http://www.ze.com/book-a-demo/.