IBM introduces new wind and solar forecasting system for utilities; big data analytics and weather modeling to predict output
12 August 2013
IBM has developed an advanced power and weather modeling technology that will help utilities increase the reliability of renewable energy resources. The solution combines weather prediction and big data analytics to forecast accurately the availability of wind power and solar energy. This will enable utilities to integrate more renewable energy into the power grid, the company says.
The solution, named “Hybrid Renewable Energy Forecasting” (HyRef) uses weather modeling capabilities, advanced cloud imaging technology and sky-facing cameras to track cloud movements, while sensors on the turbines monitor wind speed, temperature and direction.
When combined with analytics technology, the data-assimilation based solution can produce accurate local weather forecasts within a wind farm as far as one month in advance, or in 15-minute increments, IBM claims.
By utilizing local weather forecasts, HyRef can predict the performance of each individual wind turbine and estimate the amount of generated renewable energy. This level of insight will enable utilities to better manage the variable nature of wind and solar, and more accurately forecast the amount of power that can be redirected into the power grid or stored. It will also allow energy organizations to easily integrate other conventional sources such as coal and natural gas.
State Grid Jibei Electricity Power Company Limited (SG-JBEPC), a subsidiary company of the State Grid Corporation of China (SGCC), is using HyRef to integrate renewable energy into the grid. This initiative led by SG-JBEPC is phase one of the Zhangbei 670MW demonstration project, the world’s largest renewable energy initiative that combines wind and solar power, energy storage and transmission. This project contributes to China’s 5-year plan to reduce its reliance on fossil fuels.
By using the IBM wind forecasting technology, phase one of the Zhangbei project aims to increase the integration of renewable power generation by 10%. This amount of additional energy can power roughly more than 14,000 homes. The efficient use of generated energy allows the utility to reduce wind and solar curtailment while analytics provides the needed intelligence to enhance grid operations.
This project builds upon another IBM smarter analytics initiative at Denmark’s Vestas Wind Systems. Vestas, together with IBM’s big data analytics and supercomputing technology, is able strategically to place wind turbines based on petabytes of data from weather reporters, tidal phases, sensors, satellite images, deforestation maps, and weather modeling research. This insight cannot only deliver improvements in energy generation but also reduce maintenance and operational costs over the life of the project.
The Hybrid Renewable Energy Forecaster represents advancements in weather modeling technology, stemming from other fundamental innovations such as Deep Thunder. Developed by IBM, Deep Thunder provides high-resolution, micro-forecasts for weather in a region—ranging from a metropolitan area up to an entire state—with calculations as fine as every square kilometer. When coupled with business data, it can help businesses and governments tailor services, change routes and deploy equipment-to minimize the effects of major weather events by reducing costs, improving service and even saving lives.
About Time! Seriously I was wondering the other day why the variation of solar and wind is such a huge issue when we have tons of predictive data on weather patterns. Surely enough to alert standby power to ramp up or down in anticipation of their normal loads.
We should be able to utilize as much renewable power as possible. Either through maximizing renewable's mix through actively varying the "base" power to fill-in and/or use extra renewable energy to generate hydrogen or somehow store as large a fraction of it as possible.
Posted by: Trevor Carlson | 12 August 2013 at 06:52 AM
You can predict all you like, but when the wind doesn't blow, there is no energy and so you need an alternative.
We had a run of 12 days in July in Ireland where the wind was < 5% nominal, and it isn't possible to store anything like that amount of electricity, so you have to keep all your old fossil plant and crank it up for the times when there is no wind (or sun).
Like a still night.
If you have Norway on your doorstep (as Denmark does), you can use their hydro resources, but if you don't, you have to keep all the old stuff available, even if you rarely use it.
Posted by: mahonj | 12 August 2013 at 02:17 PM
mahonj - I get that you need to keep the conventional coal and natural gas available and online. What I was trying to get at is that with robust predictions the base load could be the renewable sources with the conventional sources on-call for unexpected demand or unexpected shortfalls in renewable output.
If I'm not mistaken right now we have conventional base power and the renewable sources are treated like a boost that can be idled at less cost.
What I'm saying is we should be looking at conventional sources that can flip that paradigm on its head. For example, fully autonomous or nearly automatic distributed natural gas power plants throughout the grid could fill the minor gaps in power at a local level between what is available and what's demanded. This would also serve to "harden" our grid and make it more flexible in the event of disasters or other external acts/attacks.
As customers add on their own solar, wind and geo-thermal the energy providers with predictive software should understand that a small portion of the base load is getting assistance by distributed private renewable sources. As that fraction gets larger it will begin to add volatility into that base load. At some point the volatility of the renewable sources will be greater than the variation in the demand. When that occurs you can either idle some renewable sources, dump the energy, store the energy or treat it like base power.
Customer power usage is fairly predictable (demand) and with software for renewable wind and solar as described above we can accurately predict supply as well. Most of the U.S. has access to somewhat local sources of natural gas. Therefore transportation and transmission losses would be minimized and smaller distributed semi-autonomous stations could serve a large area with minimal overhead. The side-effect would be that our natural resources would be conserved for the future as we could reduce the rate at which they're burned.
However, my inner cynic says it'll never work that way mostly because we have so much infrastructure already in place that we'd need to connect to. Most of it is old and slow to respond comparatively and the bean-counters would not want to shutter those "investments". So in practice it will be cheaper for a long time to keep large coal plants going at 70% of capacity and just spool up or down based on renewable inputs and customer demand.
The bean counter approach would work best in areas of lazy economic growth. However pockets of the country seeing explosive growth would be wise to consider a different model. Wind, Solar, and natural gas can be deployed much more quickly to match demand in terms of building infrastructure. Scaling supply and demand more accurately would result in energy conversion, transportation and transmission savings in addition to the "free" energy available as a result of the investment in our future.
Posted by: Trevor Carlson | 13 August 2013 at 03:22 PM