Utility scale and behind-the-meter energy storage is fast becoming the natural complement to renewable energy’s intermittent, and at times unpredictable, generation output.
- Greentech Media recently noted that “neither renewables backed by storage nor standalone storage are cost-competitive” vs. gas for peak power, and aren’t projected to be more economical than gas for baseload until 2035.
- NREL has published studies and papers on the economics of energy storage to complement customer-sourced PV or other renewables.
- Wind energy industry participants are seeking ways to collaborate to lower the cost of energy.
The bottom line – getting the cost of energy storage down continues to be the challenge. Improving the economics of energy storage increases possible markets for its use, broadening adoption and market penetration for both renewables and energy storage technologies.
As sales rise, energy storage producers earn economies of scale in production, reduce maintenance costs as service crews can cover more customers, and gain better asset lifecycle data as storage units exist in the field longer and more is known about conditions that can possibly lead to service or failure events.
This last point can be helped with machine learning analytics, but perhaps not in an intuitive way. Machine learning can predict asset failures for asset owners, helping to avoid costs and revenue losses from unexpected downtime. But machine learning can also play a role in helping asset producers – battery manufacturers – understand the probable life span of the batteries they produce.
It’s no small feat to predict today the lifecycle of an asset that’s not expected to run into service issues for years. Advanced analytics can learn from the data of existing assets to predict, with high confidence and accuracy, the storage asset’s expected lifecycle. Armed with this knowledge, the manufacturer can set appropriate pricing to cover their economic exposure – production, service, warranty, etc. – keeping costs down for utilities and prosumers alike. And this will only speed adoption of energy storage technologies, both utility scale and behind the meter.
Machine learning can be transformational for an organization, even an entire industry. Merging artificial intelligence systems with IoT data and broad data from your enterprise, operations, and external sources, machine learning can predict the time when a part or machine will fail. Machine learning can also prescriptively optimize your operations based on the predicted event.
To find out how machine learning has been applied to energy storage lifecycle projections and what advanced analytics can do for your organization, download the white paper (registration required).