Why Digital Metering Fails
The primary purpose of digital metering is to enable people to make better decisions about how to manage water. By leveraging data collected on water use, water authorities and facility managers can resolve issues faster and make more efficient infrastructure investments. By sharing data and information, individual people are able to adjust their behaviour and take greater care with this vital resource.
Currently the tools of digital metering; hardware, software, and infrastructure are evolving rapidly to improve the viability of its use and increase its wider adoption. This rapid rate of technological development also presents challenges around the maturity of the technologies being implemented and the lack of experience and skills required to use them effectively.
As with all new concepts and technologies, there are relatively few people with the knowledge and experience required to properly plan for and manage these multifaceted projects. Adopting digital metering requires a high degree of collaboration and integration with many different areas within the organisations and their communities that use them. Planning for and implementing digital metering takes months if not years before outcomes and benefits start to be realised.
Despite the challenges, by gaining awareness and an understanding of how to address them empowers us to make digital metering projects successful and ultimately to gain the many social, economic, environmental benefits that these projects bring to our communities.
Here we look at some of the biggest ‘digital metering killers’, what causes them, and how to avoid them.
Stuck in Pilot Purgatory
Pilot purgatory is the state of a project never progressing beyond the trial, pilot, or proof of concept stage to scale up. This is a costly and dangerous state to be in as the limited benefits gained through small scale pilots rarely exceed the financial investment required to achieve them. The purpose of a pilot is to gain knowledge and influence decisions on the best path forward to larger scale solutions and vastly greater benefits. Pilot projects often carry significant establishment and development costs with the intent that those costs then form the foundation of the following stages such as large-scale roll outs. Therefor it is only once a project successfully scales up that the investment in the pilot fulfils its purpose.
Often at the core of the pilot purgatory problem is not just a lack of planning on how to progress beyond the pilot but also a lack of understanding of what the requirements of a digital metering system should be and the key metrics around their potential benefits, costs, and how to quantify them. There are dozens of economic, social, and environmental benefits of digital metering being deployed across large communities and it can be difficult to assess them all. Value models (e.g. business cases) should be designed to factor in as many of the benefits and costs as possible and to deliver key requirements for the next stages. For example, a purely economic assessment could be; if the pilot outcomes prove a value of X, then the scale solution must cost less than Y and be implemented within Z months to achieve a positive financial outcome. This is where the value of the outcomes are measured through the pilot and then the requirements on cost and timeframe are determined based on that outcome. We can then seek a solution that fits these requirements.
Another cause of pilot purgatory is the phenomenon of always wanting the newest and greatest technology. There will always be a new meter, wireless protocol, or analytics software with greater performance or lower costs that’s release date is just around the corner. This is similar to how we feel when we need a new smartphone. There are plenty of products available now that meet or exceed our needs but there is always the next generation product being teased and previewed even just months after that latest model is released. At some point we must commit.
The solution architecture should be flexible to accommodate various outcome scenarios of the pilot and, enable adoption of the next generations of technologies, and be scalable to meet the requirements. What this tends to look like in practice when done successfully is a system where all components utilise widely supported industry standards. For example, data management and IoT platforms that are agnostic to the various wireless communication protocols and readily integrate with various analytics and user interface software products. Ideally this type of open framework is implemented at the pilot stage to avoid having to go through the effort of establishing new systems all over again to go beyond the pilot.
Lack of Commitment to Analytics
One of the biggest killers of successful digital metering is overlooking the need to continually analyse data to gain insights and drive outcomes. Digital metering systems all follow the same basic process in operation; get data, integrate to enhance data, analyse data to gain insights, act on those insights, and achieve benefits. This process runs on repeat for the life of the project.
Many organisations including private businesses and government water authorities do not allocate sufficient resources to the analysis of data. Many don't think beyond the automated alarms to end-users or automated reports to show to a corporate sustainability manager. Those features of digital metering are greatly important, but they alone do not come close to ensuring that the full potential value is realised. A keen intern who put up their hand to look at the data every few weeks will eventually move on to other jobs. Influxes of automated messages get muted or blocked by busy facility managers. Abnormal usage occurs outside of the automated alarm parameters you initially set. Batteries in meters die and cables get cut causing the flow of data to cease. To this extent, a dedicated resource must be allocated to monitor the system and ensure that insights are gained from the data on an on-going basis.
Analytic don’t need to be highly sophisticated artificial intelligence systems in order to deliver value. This is especially true of smaller scale projects. Artificial intelligence and machine learning are excellent tools and when properly implemented, supported, and continually developed can become one of the most valuable components of a digital metering solution. They can even be vital to the success of larger projects where the human input to analyse large datasets becomes too costly. That said, at this point in 2019 many, if not most, digital metering systems do not include mature AI or ML. At a basic level spreadsheet or business insights programs such as Excel and Power BI can still be used with only novice to intermediate skill to gain mission critical insights and drive actions to achieve outcomes.
Looking further ahead, data from digital metering can be combined with other systems such as network pressure, valve actuation, insurance assessments, human health assessment, and many other related data systems that are not even being considered today. By combining these various data sources of insights together, the value gained from the data and analytics is not just added, but rather it is multiplied.
Major problems can appear in an instant at any time of day. Without regular analytic actions being taken, ideally as soon as new data becomes available, value is being lost. Every moment between when a leak starts, and when it is identified is time and lost water that is adding up. Analytic actions should be taken as often as new data becomes available, ideally with the support of automation and with a view that digital metering is just one part of an increasingly connected and data-rich world.
Data Loss in Wireless Communications
Digital metering is growing rapidly with compound annual growth rates of the industry reported to be around the 12% mark. One of the keys factors in this growth is the rise of Low Power Wide Area Network (LPWAN) technologies. These new wireless technologies enable data to be sent over long distances, through obstructions such as concrete, and with minimal energy use. LPWANs enable digital meters to deliver sufficiently detailed data from difficult locations with low energy consumption and at a low cost.
LPWANs are one of the most important enablers of digital metering but like any technology, they must be properly managed in order to deliver. The common problem with wide area wireless networks including LPWANs is where too many data packets containing meter readings and device data are lost over the air to the point that the data reduces the analytic capability to gain insights. This is caused by a lack of coverage, i.e. connectivity quality, from the wireless networks over the areas where they are needed, and increased usage of networks leading to congestion and interference. These problems stem from a lack understanding of the performance and functionality of the wireless technologies and or a lack of planning to address issues as they arise.
Generally, only one or two LPWAN receivers are needed to cover an area at least the size of a suburb with most meters delivering a high percentage of their expected meter readings. The challenge is the portion of meters that deliver only a small percentage or none of the expected meter readings and effectively throw a spanner in the works of the analytics. Often the problem meters are the ones that have the most difficult obstructions to the wireless transmissions such as metal pit lids or being in underground basements. One of the obvious solutions is to continue to add more receiver stations, however this is often an impractical and costly way to achieve only minor improvements.
Many of the LPWAN technologies that are becoming increasingly common in digital metering are relatively new with many of the fastest growing technologies having only been standardised and becoming commercially available in the last few years. In this rapidly growing and hotly contested marketplace, there is a lot of hype and defamation around what each of these technologies are capable of which often results in confusion and misunderstandings.
To cut through the confusion, do your own research on the technology options and how they have repeatedly proven to perform elsewhere. Validate your expectations with your own local testing and accept that 100% packet retention is almost always not economically viable to achieve, regardless of what technology is used.
To minimise or avoid these issues systems need to be designed and implemented with various potential future horizons in mind. Start by providing coverage to the vast majority of meters with the fewest receivers and accept that a small portion will likely have poor performance. Ensure that the analytics solution has the capability to handle missing data. This could be by averaging or extrapolating similar data to fill in the blanks while maintaining the understanding that the data is not perfect. Remove obstructions and add additional receiver stations only where practical and as required. Deploy larger networks in stages with continual assessment and refinement.
Critical Firmware Faults
Firmware is the software that is installed on electronic devices and defines how they function. For digital metering firmware controls how the meter records water usage, how that data is stored locally, how digital displays show meter readings and other data, and how the meter transmits its data over the air as well as a range of other functions.
Firmware faults, or ‘bugs’, are often apparent before a product is released when they can readily be corrected but sometimes these faults may only be identified after some period of time operating in the field. Faults with firmware may be non-critical and easy to live with e.g. a symbol for ‘low battery’ may appear on the digital display slightly prematurely. The meter still functions perfectly fine and has plenty of energy remaining so while the issue is understood, it has little to no impact on operations. At the other end of the spectrum, a firmware fault could leave the device exposed to critical security breaches or even ‘brick’ the device where it ceases to function at all.
A major Australian utility has recently experienced a critical firmware fault effecting around 15,000 digital water meters installed at customer properties. The fault caused the displays to be blank instead of displaying meter readings or other information. In this case the supplier took responsibility and arranged for the meters to be replaced but the inconvenience and impact on reputation cannot be so easily repaired.
One of the reasons that firmware faults may take time to be realised is that they may only become apparent in ‘edge cases’. For example a memory allocation in a meter may overflow causing some meter reading data to be deleted prematurely if a specific and uncommon combination of alarm conditions occur simultaneously. It can be challenging for engineers and developers to cover every potential uncommon scenario when designing and testing firmware but in order to scale up without issues, serious efforts must be made to cover as many likely scenarios as possible.
Meters can also have new code and updates delivered to them via wireless communications networks where the chosen technology supports it. This process is known as firmware over the air (FOTA). With FOTA, many devices can receive and apply updates within a short period of time, perhaps thousands of meters in a single day, and significant risks are introduced for failure of the updating process or new bugs in the updated firmware being loaded onto large quantities of devices.
Thorough testing is always required to reduce or eliminate firmware bugs. Testing should be done to ensure that the firmware results in the device functioning exactly as expected, not just when the device is in a typical scenario but also in various edge cases and for as long as possible in real-world conditions before commercially releasing firmware on a large scale.
With each update to firmware, a change-log and testing records should be created to document what has been changed from one version to another. Each new version of firmware should be properly evaluated to ensure security is not compromised.
New features and update processes should be first released at a small scale and observed for a reasonable amount of time to reduce the risk of failures occurring at larger scale.
Underestimating Battery Life
Battery life is one of the most influential factors in the total cost of ownership of digital metering systems. When the energy in a battery is depleted past its usable limits, digital meters cease to transmit data or may even cease to register flow at all. There are only two ways to resolve this; replace the battery on site or replace the whole device. Both options may be close to or equal in cost to the initial installation and commissioning of the meter and therefore the frequency this process and the frequency that it is required must be clearly understood.
Many metering devices designed for outdoor use have their internal components protected by potting compounds. This is where materials such as polymer gels are used to fill the casing to protect the electronic components from water ingress. Potting adds to the challenge of replacing batteries as the compounds used are very difficult to remove and replace. They usually need to be dissolved using specialised materials and then thoroughly cleaned which is not viable to do on site. Where devices are potted, paying the price to replace the whole device is almost always the least expensive option.
Even devices that do have user-replaceable batteries may not be worth the effort to maintain if the other critical components do not have a sufficient operating life. Does the LCD only have an expected life of 13 years when you’re relying on being able to keep replacing the batteries and using the device for the next 15 or 20+ years? Does the casing remain watertight after reassembly? These challenges are rarely considered properly in the planning stages. The fact is that most devices will be removed from service when the battery is depleted and replaced with a newer model.
While the latest generation of LPWAN technologies bring benefits in standardisation and long battery life, the reality is that many of them simply haven’t existed for as long as the battery life is expected to be and haven’t been proven in real world testings. That’s not to say that they won’t achieve the 10 or 15 or more years of life that the manufacturers claim. It means that for better or worse, it must be understood and validated in order to take these technologies to scale.
Vendors tend to advertise upper estimates and purely theoretical battery life values. It can be challenging to get vendors to share detailed energy consumption data and calculations for various real-world scenarios. This is most common with meters equipped with new wireless technologies where the emerging technologies are still in their proving stages and where there is a lot of innovation and proprietary information that the vendors want to protect. Some questions to ask here are what battery life and conditions are the vendor willing to warrant? What are the formulas used to calculate battery life? What real world testing has been done with the product or previous models to support the claims?
Consider what the lifecycle of a device will entail. How much data will they be required to send and under what radio scenarios? Will devices be consuming more energy due to lower radio link quality? Many devices support features that increase energy consumption for short periods of time such as configuration changes, fast-logging modes, and firmware updates. How often are these features expected to be used and what impact do they have on total battery life?
In any case, the key is to research, plan, test, and validate for the outcomes that you require. Everyone wants ‘long’ battery life but knowing how long is possible and what is truly required only comes from experience.
Lack of Stakeholder Engagement
Stakeholder engagement both internally and externally can make or break digital metering projects. Success in this area is where all parties involved in or effected by digital metering participate in the system and share in the benefits while the project achieves its goals. Poor engagement on the other hand results in parties being left out and value going unrealised or even lost, potentially resulting in failure of the project.
Energy smart metering in Victoria among other places has often been criticised due to a lack of benefits being realised by customers. This has stemmed from a grid operator focused roll out and challenges (at least initially) in customers accessing their own usage data so that it can be used by them in a meaningful way. Ultimately these issues stem from poor stakeholder engagement. Many lessons must be learned from this and issues avoided for the success of digital water metering.
Ensuring that all parties gain from a project and the overall return on investment is maximised is not easy. With so many potential beneficiaries, it’s important that stakeholders are mapped and consulted from in the planning stages and continually engaged throughout the lifecycle. Someone is interested in improving the billing processes around metering, someone is interested in delivering sustainability outcomes and a total reduction in water use, while someone else is interested in reaching customer satisfaction targets, and so on and so on. These various wants and needs may have different requirements and a solution must achieve a positive outcome for all.
Taking utilities as an example, an academic researcher seeking to understand customer behaviour right down to a single drop may want highly detailed data on water use in millilitre detail recorded every 5 seconds. This is challenging to achieve as it requires high battery capacity and management of large volumes of data to be managed. A billing manager on the other hand may only want a cheap and simple meter to collect only one reading every quarter that fits within a tight operating budget. These are two very different sets of requirements from different stakeholders that must be considered in the formation of a solution.
Those who pay the largest share for a solution tend to have the loudest voice in these discussions but are not always the ones whose goals result in the greatest total benefit for all. The success of digital metering at scale is not achieved when a single party gets their idea of a gold-plated solution while other parties are left empty handed. The wants, and more importantly the needs, of each of the stakeholder groups must be weighed up to form a solution that benefits all parties as much as possible while meeting the overarching requirements.
The full lifecycle of the project must be assessed in order to identify key stakeholders at all stages from planning through to implementation and on-going operation. These stakeholders must develop productive partnerships and be willing to work together to find mutually beneficial solutions. Keep in mind that there is also new stakeholders and value multipliers on the horizon through the sharing of data with proper governance to support the ever increasingly connected world.
By understanding these ‘digital metering killers’ and how to avoid them, digital metering is able to fulfil its purpose; to deliver social, sustainability, and economic value by improving water management.
Original article by Rian Sullings