Data analytics can reduce costs for building operators by providing more accurate measures of chiller and HVAC system performance.
That was the message of Dr Troy Wilson, Affil.AIRAH, Chief Data Scientist at CIM Enviro and a Research Fellow at Macquarie University, when he presented at AIRAH’s Big Data and Analytics Forum earlier this month.
Currently, Integrated Part Load Value (IPVL) is used as a standard industry measure for chiller performance. Wilson explained how this is calculated, and posed questions about its accuracy.
The weighting factors are based on the weighted average of the most common building types and operations using average weather in 29 US cities. However, the weather data is from 1967–1992, when temperatures around the world were measurably lower. Wilson also pointed out that many US cities are not comparable to Australian cities in terms of climate.
He went on to present empirical local data on chiller load density taken over 12 months from 23 chillers in 12 buildings. This real-world data – from current-day Australia rather than 20th-century America – showed that IPVLs are not representative of average efficiency achieved.
The bigger picture
The finding has major implications for building operators.
“Chillers are large and expensive,” Wilson says. “As well as the initial capital cost, operations and maintenance consume a significant portion of a building’s operating budget. So, it makes sense to use data to optimise everything from the acquisition of a chiller through to its operation and maintenance. Any inefficiency has a large cost, energy and carbon impact.
“HVAC systems and chillers use a lot of data in real-time to decide what to do. But once the data is used it’s usually discarded. As the data is rarely in a standardised form that’s easily used there’s no real way to understand how a device is working across time or even how two identical devices are operating in different environments. For example, if the same chiller unit is installed in a building in Darwin and another in Hobart, there’s no way to tell whether either unit is operating optimally. Or even how either is performing over a long period of time.”
Wilson says that, as a result, purchasers are relying on manufacturer specification sheets to make major capital investment decisions.
“Chiller specifications use an efficiency curve to represent how that chiller will work under certain conditions. Those numbers are used to create a single metric using a formula, created almost three decades ago, so purchasers could compare one chiller with another. But that number isn’t representative as usage differs by building type, location, and whether there is just one or multiple chillers in the buildings, and if there are multiple, whether it’s the main or a secondary one.”
A data-based view
According to Wilson, data from the equipment itself offers a more accurate picture.
“The data from these large machines can now be accessed and used to give a clearer picture of their operation,” Wilson says. “When a purchaser chooses a chiller for their building, they can make a better choice based on their specific use-case because the data from a broad number of different chillers operating in different places is accessible. The data tells us which chillers work more efficiently for long periods under a low load or which are better in humid environments or which are optimal for operation under a sustained heavy load.
“With building operators concerned about the cost of energy and their environmental impact, these are extremely important considerations. By accessing that data, they can see clearly when their chillers are costing them the most to operate.”
Wilson notes that, in time, the data could also be used to intelligently detect faults and even prevent them.
“In many cases, there’s no single data point that can be used to diagnose a fault easily. But by collecting, aggregating and analysing data from different parts of the HVAC system, it’s possible to detect hardware faults or opportunities fix bad control strategies in real time to make the system operate more efficiently, and therefore, more cost effectively and with lower carbon emissions.
“The cheapest energy you can buy is the energy you don’t use. The cost of operating chiller units in shopping centres and large buildings is significant, but the intelligent use of data, from the initial purchasing decision through to operations and management, can make a major difference to initial and ongoing expenditure of chiller units.”
Featured image shows (L-R) Brad Schultz, M.AIRAH (Honeywell), Dr Troy Wilson, Affil.AIRAH (CIM Enviro), Chris Stamatis, Affil.AIRAH (CopperTree Analytics), and Tyrone Mawing (BUENO) in a panel session at the Big Data and Analytics Forum 2019. Source: CIM Enviro.