The increasing use of long duration battery storage is prompting the need for better performance monitoring and predictive maintenance.
As the battery energy storage market has rapidly matured over the last two to three years and there has been greater awareness of the applications and benefits, there has been a noticeable shift towards long duration technologies.
“The utilities are right in the midst of that understanding and of the applications such as ramping and time shifting,” says Russ Weed, VP Business Development & Marketing at Washington State storage provider UniEnergy Technologies.
“When you want to have a lot of energy compared to the amount of power, long duration batteries tend to be more economic. Our technology allows both short and long duration applications and the ability to layer the usage enables the market value to be maximised.”
UniEnergy Technologies’ product line is focused on its 4-hour vanadium redox flow technology.
Flow batteries, which circulate electrolytic fluid between tanks to store and discharge energy, have a number of benefits. They can conceivably last for decades because they can be continually replenished. They are easily scalable enabling storage of large amounts of energy. They also provide a full depth depth of discharge without being subject to the ‘runaway thermal reactions’ or fires, that lithium ion technologies are.
That said, it’s been a rocky road for flow batteries. Invented in the 1980s, the technology has only recently been fine-tuned for commercial use and many companies have struggled to bring down the price to better compete with the plummeting price of lithium ion cells.
UniEnergy Technologies is on track for these issues, says Weed.
More importantly, utilities are moving toward a situation where they will apply different technologies for different purposes. Flow batteries are better suited for applications such as telecoms, microgrids, demand response and others where medium to large amounts of electricity from storage might be needed over hours.
Flow batteries are also suited for remote locations where technicians monitor the battery from a control room.
In order to stay at the forefront of the market with a new technology in such a fast evolving environment, UniEnergy Technologies took the decision from the outset to monitor battery system performance, both for their own use as well as for that of their customers.
This has multiple applications, explains Becca Gillespie, Applications Manager at UniEnergy Technologies.
“There are two primary reasons. One is to monitor and evaluate the performance over time and to verify that the energy storage is performing as expected. The other is for health monitoring to pick up any degradation in for example the pump performance or other ancillary components.”
From the customer perspective, they gain the assurance that the system is performing as expected and the opportunity for a predictive maintenance approach. Indeed, to qualify for the ‘no degradation’ warranty, the ongoing automated monitoring is a requirement.
The data is also valuable to present to future customers to demonstrate how the technology works.
Like all modern data projects, the volume of data is significant – the majority at 5 second intervals but some faster at 1 second or even 0.5 second intervals.
“A lot of the data points are collected every 5 seconds but the fast data is useful to catch if there is a problem to see how it manifests and for historical analyses,” says Gillespie. “As a relatively young company and technology it pays us to get into the nuts and bolts of what’s going on.”
Weed points out that currently four strings of data are coming in. These are from deployments at Avista and Snohomish County Public Utility District in Washington State and Terna in Sicily, as well from the company’s own systems.
“In all we have monitored approximately 1GWh of discharge back on to the various grids,” he says.
For the management and analysis of the data, UniEnergy Technologies has selected the PI System from OSIsoft.
Along with the PI System, the company is using PI Vision for dashboarding and PI DataLink for reporting.
The PI Vision dashboards present a global view and allow drilling down into greater levels of detail. For example, the power for each site can be visualised, then at the next level users can check the pump speed or see the temperatures of the pumps or the electrolyte.
PI DataLink on the other hand is used for example to check data such as the power usage of the pumps over time, and it also provides the automated weekly and monthly performance reports.
“We use PI Vision for studies and troubleshooting whereas PI DataLink is more for long term monitoring,” Gillespie says.
Commenting that from a company perspective the studies and analyses “enable us to keep on our toes and be dynamic,” she adds: “They also have been invaluable in being able to expand the user base. Some people want to look at a few points here or there or a particular event and the packages are accessible and easy to learn. Also, changes to what we want to look at can be easily made.”
The third package that is used is PI Notifications. This provides instant notification if there is a problem with any of the systems in the field and is also used to test out possible controls before being passed on for implementation in the control code.
The other significant use of the data is in informing design enhancements to the company’s vanadium flow storage technology.
“Examples of improvements we have been able to make include to the optimisation of the pump, the thermal characteristics of the system and the stack, which we have been able to entirely reshape to improve the capacity and long term performance,” says Gillespie.
She adds that such hardware improvements are introduced in the future systems whereas control system improvements also are fed back to current deployments via a software update.
Weed comments that the systems deployed to date are the company’s initial 0.5MW/2MWh Uni.System storage. A new ‘small’ system, named ReFlex, has been contracted in five US states and the data analyses significantly advanced the design process.
“We were able to do the design and prototyping more rapidly with the PI System data from the systems in the field,” he says.
Likewise the data will feed into the design of a proposed new ‘small’ system for C&I/microgrid applications. For a sense of scale ReFlex is housed in a 6m container whereas the small system will be about the size of a domestic freezer.
“These developments have been possible based on our data collection and analyses and enable us to keep up with this rapidly expanding, ultra-competitive and cutthroat industry,” concludes Gillespie.