Cloud-cam and lidar: wind and solar farms encouraged to do own forecasts

Print Friendly, PDF & Email

Wind and solar farms encouraged to use new technologies like cloud-cam and Lidar and provide their own 5-minute ahead output forecasts.

share
Print Friendly, PDF & Email

Australia’s rapidly increasing number of large scale wind and solar farms are being encouraged to invest in technology such as cloud-cams and “nose-cone lidars” that will improve the accuracy of the output from these wether-dependent and variable energy sources.

The Australian Renewable Energy Agency is providing $10 million to help trial technologies that can deliver the forecasting services, and importantly encore individual wind and solar farms to “self-forecast” output 5-minutes ahead and provide that input to the Australian Energy Market Operator.

Current forecasting measures are more accurate than most people realise. Even on a day ahead basis, the forecasts are more than 90 per cent accurate, and are about as accurate as load forecasts. On short term forecasts such as 5 minute intervals essential for AEMO’s dispatch decisions, the accuracy is usually more than 99 per cent.

But there are still issues. These forecasts are system-wide, not on individual installations. And on those occasions when individual wind and solar farms produce more than the AEMO dispatch allocation, they miss out on revenue.

When they fall short, due to a lull in wind or a cloud passing overhead, they are subject to what are known as “causer pay” penalties. This refers broadly to the cost of making up the shortfall or correcting frequency variations that occur as a result.

These “causer-pay” penalties have amounted to several hundred million dollars in recent years, although it should be noted that wind and solar are not the only culprits here.

Fossil fuel generators also miss their output targets due to sudden trips (and there have been more than 50 of those since the start of summer) or simple errors like pushing the wrong button.

Wind and solar farm operators, however, are keen for the “self-forecasting” because it means that they can manage their own output – and AEMO’s expectations – more accurately and be less subject to either revenue shortfalls or penalties.


Some of the technologies include “cloud-cams” that have been trialled at some solar farms in Australia such as Fulcrum3D’s technology at Uterne in Alice Springs and Karratha Airport. Put simply, it is a camera that monitors approaching clouds and can predict when and by how much the output of solar farms can be affected.

Lidar technology – which stands for light detection and ranging, and is also known as laser scanning – can be placed in the nose-cones of wind turbines to more accurately predict short term wind speeds, protecting wind farm operators from the impact of unexpected short-term lulls or gusts.

Other technologies that will be encouraged under the program may include software and the use of satellite data.

ARENA says that the program is seeking to demonstrate wind and solar farms can provide more accurate forecasts of their output into AEMO’s central dispatch system.

The projects it will fund will deliver ‘5-minute ahead’ forecasts, and also explore the commercial benefits to wind and solar farms of investing in forecasting technology and examine factors that affect the accuracy of forecasts in different weather, operational conditions and geographies.

ARENA CEO Ivor Frischknecht said this initiative originated in ARENA’s A-Lab innovation workshop last year, and could allow wind and solar farms to be better integrated into the grid while simultaneously improving grid security and reducing energy costs.

“As more variable renewables enter the market, we need to improve the accuracy of our short-term forecasts so we can anticipate what will happen as a cloud passes over a solar farm or if the winds change,” he said.

“At present, wind and solar farms can be disadvantaged if their available output doesn’t match the central forecast. If the forecasts are too low, wind and solar farms are restricted in how much electricity they can paid to produce.

“If forecasts are too high, the wind or solar farm may be obliged to pay for the cost of stabilising, which increases the price of electricity and is ultimately passed on.”

AEMO CEO Audrey Zibelman said if successful, the initiative would be another step forward in strategically integrating renewable generation into the National Electricity Market (NEM).

“Accurate short-term forecasts are essential for balancing supply and demand, and avoiding grid instability,” Ms Zibelman said.

“If we can more accurately predict demand and the output of all types of generation, we expect this will reduce the need for additional frequency control services in the future, which the market pays for,” she said.

Expressions of Interest open today and close on May 9, 2018. Successful applicants will be notified in June 2018 and invited to submit full applications.

Print Friendly, PDF & Email

10 Comments
  1. Sir Pete o Possums Reek 2 years ago

    Given your entity is large enough to “play” in the energy market.
    Why wouldn’t you do local predictions as well ?

    Sure there is an up front cost but the financial payback period would be quite quick . [Better return than a fire extinguisher I should think.]

    More importantly the professional ethic , delivered quality of service, builds trust reputatiion (good will) which are stepping stones for longevity and growth.
    [see: pro level networking, aerospace …]
    Perhaps even a QAssurance bonus.

    Even with massive fast reponse storage these technologies would be a valuable management tool.

    Measure it or lose it.

  2. trackdaze 2 years ago

    This sounds like a job for………..BAT*MAN.

    *battery

    • George Darroch 2 years ago

      *management

  3. George Michaelson 2 years ago

    Having some hysteresis in the system inside the boundary you commit to supply from feels like a good idea right now. It doesn’t much matter how you do it, but if you did have some explicit storage of energy *and* lidar or other predictive ability, then you could ride out short term lacks in a smart way, maybe?

    • George Darroch 2 years ago

      Pairing improved prediction with local storage seems like a very smart idea.

      • Kevfromspace 2 years ago

        Fulcrum3D’s technology enables this pairing. They have some very smart technology, and it appears that they will do very well out of this trial

  4. Dr. Nick Engerer 2 years ago

    This article focuses on hardware based solutions for the forecasting problem, a trend that I see across the industry. The assumption that a short-term forecast requires hardware such as a ‘cloud cam’ (solar) or a lidar unit (wind) has not been proven. Presently, the relatively mature Australian forecasting provider sector of the industry offers short-term forecasting solutions for solar and wind, which do not require expensive hardware ($100k+ cloud cams, $200k+ lidar units + the costs of ongoing forecasting services).

    • Deegee 2 years ago

      Fulcrum3D CloudCAM costs ~half your estimate =). Can be significantly less depending on complexity. Its a good solution that I have seen work, however I am also interested in your ‘forecasting provider’ solution. I dont know of any that work for this 5 minute window, and I work in the field. Feel free to elaborate

      • Dr. Nick Engerer 2 years ago

        Great to see an industry expert weighing in! Even if the $50k units are used, how many are required for 50-100MW+ farms? More than one, easily, just based on the field of view these tech options offer & the forecasting horizons they would require to make a 5min prediction. We are back up to $100k+ straight away. Furthermore, such devices have only been proven on ~100kW to 1s of MW solar installations. That is not the scale we are talking about here. The industry/media need to stop making this hardware assumption as solar farm operators are becoming misinformed.

        • Jon 2 years ago

          As Deegee asked who is providing this 5 minute forecasting?

          As for the cloud cam units only one would be needed per site, combined with a layout of the array a program could easily calculate when sections of the array are likely to be covered.

          Sun movement is very predictable, cloud position would be easily predictable 5 minutes ahead with known positions every 10-15 seconds (or more often) for the previous hour.

          The $50k per site wouldn’t take long to recoup its costs on a decent size solar farm.

Comments are closed.

Get up to 3 quotes from pre-vetted solar (and battery) installers.