Researchers and business owners eager to cut energy waste are flummoxed by a paucity of decent data, writes Joe Wyndham of the Institute for Sustainable Futures. Without it, what hope is there for fixing the world’s problems?
As energy researchers we thrive on data. It’s what bridges the gap between our fancy ideas and the real world. We can sit at our computers all day coming up with clever ways to save energy and money for consumers, or to alleviate the struggle between supply and demand on the electricity grid, but no amount of cleverness matters until we can verify our ideas against good data.
At the UTS Institute for Sustainable Futures (ISF), we have clocked up a lot of time searching for good data, like prospectors fervently digging for golden nuggets, through the depths of the internet, through official and unofficial repositories.
But, like gold mining, the world of data is fickle, and when our digging has left us destitute we have even begged at the doors of businesses who might offer their generosity. Some have certainly been generous, but not enough to allow us to fully achieve our goals.
The data we so desperately seek is about electricity consumption and solar power. You see, we think we have a suite of clever solutions for tweaking energy consumption so that businesses get the most out of their solar systems in a way that could both save them money and help prevent issues on the grid. We’ve done some site-level modelling and had some very interesting results, but what we’re ultimately interested in is the impact of deploying these solutions en masse across industry. But alas, our dearth of data leaves us unable to form solid conclusions that we can release into the wild, at least without awkward qualifying statements about our great levels of uncertainty.
“What are these clever solutions?” you ask. Well, it’s all about optimising electricity consumption through flexibility.
Optimisation is really just a fancy word that describes a fundamentally simple thing – getting the most out of your effort and resources. Businesses of all sizes engage in optimisation activities all the time, whether they call it optimisation or not.
Cafe owners, for example, might schedule deliveries to arrive at the quietest time in the week when their staff are already rostered on but are less busy than usual. A farmer might stockpile feed because it allows them the flexibility to delay buying more until prices are lower. And construction companies engage in what’s known as resource levelling and smoothing, the practice of spreading project tasks over time to make sure the demand for resources matches resource supply given a specific supply constraint.
All of these activities seek to maximise productivity for the lowest resource cost.
As the rise in electricity prices begins to bite business bottom lines, it’s very interesting to consider how this natural inclination towards optimisation can be applied to energy management.
It may not be obvious, but the examples above are remarkably transferable to energy. The most important feature common to them all is that there is an incentive to invest in flexible operation.
The cafe owners and construction companies are incentivised to minimise peaky work schedules because they result in the need to hire additional people for short periods, which is costly. The farmer is incentivised to spend money on storage infrastructure because feed prices vary enough that waiting for lower prices is worth it.
To transfer this optimisation mentality to energy, you simply need to understand where flexibility can be found in an energy system and the incentives that make it worthwhile to operate flexibly.
In the world of energy, flexibility comes in many forms but can broadly be categorised as either:
- load flexibility, much like our cafe and construction task examples, where energy loads can be supplied at different times, intensity or duration;
- energy storage, like our farmer example, where a resource is stored, dispatched and purchased strategically, and;
- generation flexibility, where the availability or supply of energy can be changed according to demand.
Historically in Australia energy costs have been low, so the financial incentives to be flexible have been limited. But as electricity prices rise and the options for investing in cheaper electricity diversify, there is an emerging case to be made for flexible energy management taking a larger role in Australian businesses.
One simple example of where this makes the most sense is deploying flexible loads to soak up excess solar power generation within businesses.
Match load with solar
In most cases, when a business buys solar the best-case scenario is that their minute-to-minute electricity consumption closely matches their real-time solar generation.
For the business this is a simple economic calculation. They paid the capital for the solar system because, per unit of energy, solar is cheaper than grid electricity and, when surplus solar spills back into the grid, the tariff received by the business is usually less than the cost of solar generation.
So, pushing energy usage into the solar-generating hours to minimise grid consumption is the best way to get value out of the system.
From the perspective of a network business it is also usually best when customers use all of their solar energy. The most manageable scenario for the network is for the aggregate demand profile in a region to be relatively flat over the course of the day. Of course, demand generally swings up and down markedly according to the daily routines of consumers. But when solar generation occurs during low demand times the swing can be much more dramatic, in some cases even reversing the flow of energy in parts of the network. There is a strong case then — as businesses continue to invest heavily in solar — to optimise how energy is consumed with respect to solar generation for the benefit of both businesses and network operators.
Know your tariffs
So, how can businesses optimise energy consumption to get the best out of their solar investment? To answer this you need to look at the incentives and opportunities to operate flexibly. On the incentives side of things for businesses it’s usually about either reducing costs, increasing revenue or both. Electricity tariffs are very influential here.
As far as flexibility is concerned, the story is complicated and highly contextual. But before we look at examples of flexible operation, it’s probably worth covering how electricity tariffs are structured and how this affects electricity bills so that the incentives in each case are clear.
Electricity bills are not just calculated by adding up the amount of energy a business uses — there are fixed and variable tariffs which apply in different circumstances.
Most bills include a daily supply charge, which is the cost of being connected to the grid. Some include time-of-use (TOU) rates for energy consumption occurring at different times of day, commonly called peak, off-peak and shoulder periods.
Many bills include controlled load rates, like hot water systems that are controlled to by the network operator to switch on at the least burdensome time for the grid. And for larger premises there is often a demand charge (otherwise known as a capacity charge), which is a rate applied across a whole period, perhaps monthly or annually, based on the single highest moment of power consumption observed in that period.
Of course, for premises with solar, bills may include a solar feed-in tariff (FIT) that is paid to the consumer for any energy they export back into the grid. The images at the right illustrate how each of these tariffs can affect a bill.
For businesses with solar, the most significant incentives for flexibility can generally be found in TOU tariffs and demand charges, and in the difference between solar generation cost and FITs.
Flexible operation in businesses can mostly be achieved through load flexibility and storage. Some examples of technologies and strategies that support flexible operation are listed in following table.
Let’s take a look at a specific example of how working flexibly with over-cooling might impact an electricity bill of a food warehouse that uses large-scale freezers.
The main idea here is that most frozen foods can be cooled to temperatures much lower than they need for food safety, without any negative consequences. This is known as over-cooling. This means that rather than exclusively running constant temperature of, say, -20°C, sometimes you could over-cool the freezer to -30°C and avoid using the motor as the temperature slowly approaches -20°C again.
What’s the point of this? Well, just like in the cafe, farmer and construction examples, it’s all about flexibly dispatching your resources at the least expensive time. In this case, that time will depend on what electricity tariffs are in place. The images below show how this could be applied to save money by avoiding peak TOU tariffs, avoiding demand charges or by offsetting grid consumption with solar utilisation.
It’s possible that all three of these tariffs could simultaneously be impacted by the over-cooling strategy, and that’s where we think a significant opportunity lies.
We definitely know that this is technically feasible, but is it actually worth it financially? Well, that comes down to whether the costs associated with over-cooling outweigh the potential savings.
Costs might include upgrading the freezer motor and the insulation so that the system can handle over-cooling efficiently. Other costs might include installing automated monitoring and control, or assigning someone to manually configure the freezer settings.
However, there might not be any real costs at all. Many freezers units are in fact over-sized to begin with because they need to handle seasonal weather changes. And many warehouses already have control systems in place to control their freezer loads.
Savings in this example are dependent on the structure of the tariffs that the business is contracted under, their existing electricity consumption profile and whether or not they have solar system is big enough to spill surplus generation to the grid.
For example, if TOU tariffs were in place, the FIT was low and there was a large amount of solar spill, the case for moving consumption away from the afternoon peak would be high. However, if the tariffs were not TOU and the FIT was at parity with grid electricity there would be no savings at all.
Why data matters
To add further complexity to this case, the total amount of energy consumed could go up or down depending on the ambient air temperature at time of over-cooling, which significantly affects efficiency, and on the energy losses typically associated with storing things at a cooler temperature.
In short, to understand whether over-cooling is worth it, you need lots of data about the site, preferably data that represents the situation on an hour-by-hour basis at the very least. At ISF, we rely on a bespoke computer model to churn through hourly electricity consumption and solar generation data while it virtually manipulates the flexible portion of electricity consumption to achieve optimal cost outcomes according to the tariffs in place. We refer to the whole approach as renewable energy and load management, or REALM.
We have investigated the potential value of flexible loads, as shown in the above table, for a variety of specific real and hypothetical businesses cases, but our end-goal is to understand the potential impact of deploying such flexibility en masse at scale across different industries. In particular, we are interested in business productivity and the ability of the electricity grid to host very high penetrations of renewable generation.
As you saw above, it can get a bit complicated to assess an individual site, and data is the key to figuring it all out. As we look at more and more sites we are attempting to formulate simple but effective heuristics that can be applied to “typical” sites. For example, we want to be able to make statements about how much a typical cold warehouse could save by shifting a certain amount of energy into solar hours.
Heuristics like this are invaluable for businesses who want to dip their toes in the water of energy management by alerting them to high likelihood potential savings.
We need numbers
While the flexibility story is crystallising steadily for us at the individual site level, we’ve come up against some big hurdles when trying to estimate the full potential impact across whole sectors of industry. Specifically, a lack of good data has frustrated our analysis. When we started our research we expected the hardest data to find would be descriptions of flexible loads because they are many and varied, and because they haven’t necessarily been documented in a way that is helpful to model. But, to our surprise, there are more basic data deficits that impede our analysis; namely, data that describes electricity loads and solar generation.
Let me explain this issue using the over-cooling example from above and some back-of-the-envelope thinking. Say you want to estimate the total potential that over-cooling has to increase utilisation of existing solar, and to increase the total amount of solar generation that your state’s electricity grid can handle. A first approach might be to just take our warehouse example and multiply the site’s result across the number of freezer warehouses in the state. Easy, right? Well, easy, but probably not very accurate for the following reasons:
- not all warehouses are the same size (in terms of average electrical consumption);
- not all of them have solar, and not all warehouses with solar will spill surplus electricity into the grid, and;
- even sites with exactly the same average electrical load and solar system size will have different daily usage profiles, so the same flexibility strategy may not work in the same way for all warehouses.
A better approach would be to take samples of electricity consumption data from warehouses that fit different consumption categories and identify typical solar installation sizes in each category. Then you could model a typical example site and come up with a per-site rate of impact that can be multiplied across the state’s number of sites in each category.
But here’s where we encounter the data problem. Not only do we struggle to find the hourly energy consumption data for a reasonable number of businesses, we don’t even have enough information to estimate the number of businesses in each consumption category. And, even if we did have enough data to understand consumption better, we don’t have any data on the distribution of solar systems across the consumption categories. Heck, we don’t even know the total number of cold warehouse businesses with solar, let alone the size of their systems and how it compares to their consumption.
This problem is not unique to the cold warehouse example; it’s the same for all industries. The data just isn’t available for us to do a detailed, rigorous analysis.
Willingness to share
It seems like a preposterous notion that we can’t find these types of data. Good data is in fact being generated all the time. Electricity metering has gone digital (think smart metering) and solar systems have fancy hardware that continuously monitors generation. Based on our interactions with businesses and energy management specialists, we suspect that troves of data exist behind walls of commercial in confidence.
The data is likely to be out there somewhere, but the big problem for us is that barely anyone is collecting and storing it in a way that is useful — that is, for wider analysis that could inform for example where governments should focus program spending or grid management.
The electricity network, we are told, is soon to become a smart grid, in which there will be some brilliant control system with such omniscience about real-time demand and distributed generation that it will flexibly balance supply and demand in perfect harmony for the benefit of energy players, grids and consumers alike.
From our perspective it’s hard to envisage such wonderful real-time control when even a retrospective data analysis is impossible. The effort to which we must go to find data sets for a relatively simple analysis would indicate we are a long way off from a truly smart grid. Good smart grids must be built on good data and data availability.
Hypothesis on hold
Where does this leave us? Based on our early REALM modelling we argue that there is potentially a big opportunity to discover significant flexibility in electricity consumption across many industries, but we can’t verify this hypothesis unless we get access to better data.
In the context of the smart grid paradigm that our electricity network is approaching, it’s important that good data is available to support not just our research but all the innovators across the changing landscape of energy and, crucially, the control systems that will keep the whole network running.
As a first step we need to see action from businesses and energy management professionals who can paint an accurate picture of what energy data is out there and how it can be distributed in a commercially acceptable way. Some of the best data we have received as researchers, for example, has been generously provided by our business associates after being de-identified to protect privacy.
As a second step, stakeholders in the electricity industry, including regulators, energy retailers, network operators, businesses, energy consultants and innovators, need to develop a deliberate data plan that recognises what is most important in whatever version of the smart grid we end up with. This plan needs to consider which data sets are potentially valuable, how they should be collected, how they should be stored and how they can be distributed to innovators whilst protecting consumer privacy and commercial sensitivity.
There are, of course, notable efforts being made by many of our colleagues in the energy industry, but the present grab-bag nature of data sets that exists in the wild world of electricity data suggests there is still a long way to go.
Meanwhile, we at ISF will continue our search for those rare golden nuggets of data that sustain our research and help us build our clever ideas.
Joe Wyndham is a researcher and data scientist of energy at the Institute of Sustainable Futures, University of Technology Sydney.