In my last blog, I wrote about the forward-looking questions utilities can ask their backward-looking data and improve upon it with real-time information sources.
To some degree, all these sorts of analytic progressions already occur at most utilities, but suspicions about the levels of accuracy often dissuade executives from taking decisive action.
Suspicions about accuracy have dogged decisive action for quite a while now. Consider a 1982 study conducted by the US General Accounting Office for the U.S. Congressí Subcommittee on Energy Conservation and Power Committee titled " Analysis of Electric Utility Load Forecasting."
(As an aside, I absolutely loved reading this report as a piece of utility history. Think about it. The report was written before computing power was dispersed on desktops, when state-of-the-art data storage meant large rooms filled with tape machines. Even the typeface font used in the report is quaint, probably coming right off an IBM Selectric, to my non-expert, but age-experienced eye.)
The report is fascinating because it summarizes the utility energy forecasting methodologies that were considered cutting edge at the time:
Trend Forecasting "predicts future power demand by assuming that the factors that influenced demand in the past will continue to do so in the same way in the future." Certainty was affected by the fact that past results donít guarantee identical future outcomes.
"Econometric forecasting uses mathematical equations based on the relationship between past demand and economic and demographic conditions to forecast future demand." The assumption that the relationship will continue into the future made this method weak.
End-use forecasting breaks electricity consumption into the residential, commercial, and industrial demand profiles. While the data for this type of forecasting was considered in 1982 to be "expensive and time consuming to collect and maintain," the advantage was that it readily reflects changes in consumer tastes, increased efficiency of energy-using products, and changes in the economy, particularly technological shifts in our industrial base. But the ability to utilize this sort of approach was questionable given the technologies of the time.
A sum-of-the-utilities forecast gives regional or national perspective on utility demand as a combination of individual utility forecasts. "[T]his approach, because of its aggregated nature, is of limited use to individual utilities in planning new resources because the service areas are significantly smaller than the area covered by the forecast." Enough said.
So weíve looked at limitations of forecasting as perceived in 1982. Letís compare that to the forecasting capabilities that are considered important 31 years later, and the confidences asserted in their likely accuracy, as contained in a recent press release from SAS, the sponsor of this site, about its new energy forecasting solution for utilities. Here are a few of the productís selling points along with my commentary:
"[The new solution] helps utilities operate more efficiently and effectively by capitalizing on new interval data being returned from smart meters." Smart meters werenít included in the 1982 report. But the commission understood certain modern appliances had energy usage profiles that could be discovered, albeit with expense and time. My how smart meters and the development of energy usage profiles for every appliance have increased the possibility for confidence in energy forecasting.
"Unlike other load forecasting software, SAS Energy Forecasting supports multiple planning horizons -- from the next hour to the next 50 years. Utilities can leverage big data from smart meters, power plants and other sources to produce accurate and timely forecasts of short- and long-term load and demand. This helps the utilities better trade energy on the open market, while optimally managing power plants, generators and other assets." Ah, yes, big data -- a term and concept not even created in 1982. But it addresses so many of the factors that contributed to the low levels of confidence that utility manager once had.
"Utilities have successfully used forecasting in the past. Todayís new challenges, including the added complexity of wind and solar power generation, require even greater attention to the data sets and models that feed those forecasts." Once again, to be effective today, forecasting must include consideration of volatile wind and solar sources, as part of a US priority to include more renewable power sources in its mix.
How times have changed. Each of the three new energy forecasting considerations endeavors to increase confidence in methodology and capability to address modern challenges. I wonder how a similar "Analysis of Electric Utility Load Forecasting" report might be written today. Any ideas?
I was just reading about how the use of visual analytics is coming into play for preventive asset management of big manufacturing systems or really anything big with structure -- aircraft, ships, trains, buildings, bridges, and so on. Sensors are delivering vast amounts of data, as is the case in weather systems. I wonder if being able to explore the data visually will help researchers in finding patterns and understanding results in energy fields as well. I would think the application would be perfect in the energy industry as you describe it.
My thought is "No, alternative energy data sets are not readily enough available." It's my opinion that the data histories for weather are so colossal that the industry is struggling with its Big Data conundrums. Weather science algorithms are incredibly complex and its even possible that major factors, like sun spot influences, aren't even being considered by many scientists over the past decade or so to the degree that some scientists believe they should. The use of analytics to deliver data analysis with regard to alternative energy, in bite-sized chunks, will be a ongoing exercise of incredible value -- and difficulty.
Joe, interesting stuff here. We've come a long way in 30 years, that's for sure. And while I agree with you that new forecasting considerations are boosting confidence and capabilities, I have to wonder if we've come far enough. I'm especially wondering whether alternative energy data sets are readily enough evailable and being taken advantage of in energy modeling. Your thoughts?
While 97% of insurers say that insurance fraud has increased or remained the same in the past two years, most of those companies report benefits from anti-fraud technology in limiting the impact of fraud, including higher quality referrals, the ability to uncover organized fraud, and improve efficiency for investigators.