The hidden (and enormous) potential for savings in the air you now breathe.
image credit: Dreamstime.com
- Nov 29, 2019 6:00 am GMTNov 27, 2019 2:05 pm GMT
- 349 views
This item is part of the Artificial Intelligence - Special Issue - 12/2019, click here for more
When you consider the impact air conditioning and refrigeration have on energy usage and costs (currently accounting for around 40 percent of a building’s energy use) the potential to reduce building HVAC energy use by up to 30 percent would be enough to make even the most overstretched building facilities manager sit up and listen.
Today’s energy intensive approach to HVAC
Air conditioning is well known as an energy guzzler, and the numbers prove this point: It is responsible for consuming 10 percent of the world’s electricity generation. Based on current emissions associated with electricity generation, air conditioning alone produces 1.3 billion tonnes CO2 annually, or roughly 4 percent of global emissions. By 2050, the IEA predicts over 35,000GW of installed capacity and a doubling of global air conditioning demand in the commercial sector.
It might therefore be surprising to learn that commercial control strategies dictating the staging of chillers and temperature set points have barely improved over the last decade.
Despite significant advances in the efficiency of chillers, pumps and fans, unsophisticated control strategies have left many opportunities for improved system efficiency untapped, largely because these systems generally use a single data point extracted from the building management system as a proxy for cooling load.
You can only improve what you can measure.
Controlling air conditioning energy use in commercial buildings is traditionally seen as complex and costly, restricted to certain systems or manufacturers and to collecting data around usage, rather than measuring and responding to real time inputs. And the current approach only takes a fraction of available information into account when dictating which equipment should be running and at what loading.
Bringing air conditioning controls into the 21st century, calls for data-driven, machine learning enabled, cloud-connected control strategies. An approach that allows for the optimisation of the plant as a whole would mean that equipment loading can deliver peak plant efficiency under all possible conditions. Today, this has become an achievable ideal.
How it works.
Exergenics’ approach begins at the source of the most energy intensive equipment in any commercial building - the chilled plant room - which houses the equipment responsible for delivering chilled water throughout a construction. Machine learning is used to replicate the chilled plant room, creating a mathematical representation of the physical system. With a data-driven digital twin of the system, it can then be controlled in a smart way.
An optimisation algorithm is then applied to the model, before a range of ambient weather conditions and cooling loads are simulated to create an optimised control strategy. (The benefit of this method compared to manual tweaking of the system is that algorithms are unbiased and opportunity agnostic, so will find efficiencies wherever they hide.)
This first stage alone will produce a strategy vastly superior to traditional optimisation. Alone however, such optimisation remains static, unable to integrate additional relevant data sources and maximise efficiency.
Adding real time data changes the game.
For truly dynamic optimisation, these systems need to connect to the cloud. Combining cloud computing with machine learning allows continuous collection of weather and building data to refine a control strategy and further improve efficiency. Such a relationship can predict cooling load throughout the day, based on inputs including temperature, humidity and time. It also improves the control capabilities of the system, for example by precooling a building before a period of high demand. Significantly reducing the peak demand of the building can save money and reduce pressure on the grid at peak times, pressure that is increasingly being driven by air conditioning demand.
Because it’s data driven so applicable across multiple systems, this solution also provides greater visibility: Building Managers will be able to see energy efficiency in action. Building Management Systems (BMS) are also a prime candidate for Internet of Things (IoT) applications. It allows them to run optimally, following a straightforward set of rules that can be stored onsite, while sending data to the cloud so machine-learning algorithms can optimise and update the simple rules stored on controllers.
Revolutionising how we keep our buildings and ourselves cool.
The connection of building controllers to the cloud opens the door to many additional opportunities which are simply not available using traditional building controls. For instance, enabling buildings to participate in demand response markets in a sustainable way, rather than just switching from the grid to a diesel generator (current standard practice in Australia). It also helps stabilise the grid: With weather now dictating the supply of electricity, buildings connected to the cloud can actively participate in energy markets to ensure grid stability, particularly as increasingly common extreme weather events strain grid infrastructure.
Smart buildings stack up.
Research shows that commercial building air conditioning optimisation could save as much as 30 percent of the electricity used to deliver chilled water, compared with traditional control strategies.
If implemented globally, this could decrease electricity demand by around 2 percent worldwide, resulting in 400Mt CO2 less emissions annually. The ability for these smart, cloud-connected systems to be able to load shift to match supply with demand essentially turns every building into a giant battery.
Even greater potential at scale.
The demand for grid stabilising technologies will only increase as the proportion of intermittent, weather dependant generation increases in grids throughout the world.
This system is already being implemented to deliver smart buildings. The next stage - developing smart cities - will take broad implementation and further innovation. Once cloud-connected control systems are commonplace, further opportunities will become available for connecting these assets and using them in ways that can have benefits beyond the energy savings experienced within the buildings.
With control over several energy intensive assets in the same geographic location, there is the potential to aggregate up into a Virtual Power Plant (VPP) and enter the demand response market. This would open new sustainable markets in which building owners could participate. It would also lead to fewer blackouts and a lower requirement for infrastructure investment.
As we move towards 100 percent renewable generation, smart cities will become essential in maintaining a functioning grid with reasonable power prices. While we can’t yet fully estimate their abatement potential, there is no doubt smart cities will help enable a higher penetration of renewable generation, without the need for more expensive firming technologies.