This is the second post in our series on the benefits of using power analytics in the ‘New Electric World.’ In this post, I’ll take a closer look at an important key to any power analytics solution: data quality.
In my last post, I talked about how the increased electrification of commercial buildings and industrial processes is making electrical distribution networks more dynamic and complex. Fortunately, this increasing complexity can be managed with analytics, whether in IoT-enabled devices, edge control systems, or cloud applications.
Data is the fuel that powers the analytics in these three tiers, providing facility teams with deeper insights into electrical operations and boardrooms with a better sense of risks and investment opportunities to optimize performance. However, if there are errors in the data used in the analytics (i.e. if the data quality is poor), the insights will be incorrect. Here is an example that shows the impact of poor data quality:
An airport was planning to expand a terminal. To forecast the energy usage of the added tenants and gates, they fed energy data from existing gates and tenants into a capacity and redundancy planning application. That app indicated that a new transformer would be required to support the increased load of the new space. However, on deeper investigation, they discovered several meters that supplied the energy data were misconfigured, which caused the reported energy to be double the true value. This data quality issue created a misleading insight. Once the misconfiguration was fixed, it was apparent that the new gates and tenants could be served with the existing transformer. A large capital expense was avoided.
The best power analytics solutions will provide capabilities that allow you to detect, track, and correct data quality issues before you start to make business decisions.
Power Analytics Information Is Undermined by Poor Data Quality
We have analyzed thousands of power management systems, in different facility types, of varying ages and across multiple vendors. In that analysis, we found that 98 percent of these systems had some data quality issues. Note, that here we are talking about the data acquisition and control system (i.e. connected devices, gateways, databases, and software), not the actual electrical distribution system itself.
So exactly what are typical data problems that can impact power analytics? Here are a few examples:
- Zero values – when a sensor has come loose or is faulty, or a voltage sensor is reversed
- Unchanging values – when someone has added custom logic to a device, but the logic is in an error state
- Instrument Transformer ratios incorrect – when a device is misconfigured to show a higher or lower measurement than the actual
- Incorrect nominal voltage – when the device thinks the nominal voltage is different from the actual
4 Common Data Quality Problems and Their Origin
The fact that 98 percent of analyzed systems had data quality problem is a staggering statistic. But there are many reasons this can happen. Here are four common sources of these problems:
1. Device Wiring
Device wiring errors are a common issue causing poor data quality. A contractor may install a current transformer (CT) or power transformer (PT) backwards so that polarity is reversed when connected to an electrical meter. In this case, a load like a backup fan might seem to be generating power rather than consuming it. If it is one load of thousands in a complex system, the problem may go undetected, until the fan is used for backup. Imagine an emergency when the fan is activated, and it appears the fan is feeding power into the network. This can add confusion and perceived risk at a critical moment, which may impact safety and time-to-recover operations.
2. Device Configuration
This problem occurs during commissioning or maintenance when the configurations of multiple devices are typically changed. Contractors might accidentally reset device configurations (e.g. CT/PT ratios, voltage type, nominal voltage), disable alarms without re-enabling them later, change trip settings temporarily but forget to set them back, or fail to transfer a configuration from an older device to its newer replacement. In these situations, the device could generate incorrect data or might not generate data (i.e. an alarm) during a critical condition. These issues can cause a power analytics application to misdiagnose maintenance needs or to misinterpret process or equipment performance.
3. Edge System Functionality
The ‘edge’ system comprises any onsite gateways, software, and databases used to collect and store data from connected devices. Data quality problems can occur at start-up or during maintenance if, for example, a service provider defines a connected device using a ‘generic’ device type instead of the proper ‘advanced’ meter type. In this case, many of the important values from that device are not made available to the power analytics application. Data can also disappear if some software services are not working or if a computer or gateway runs out of storage space. The latter can happen if high-speed logging is temporarily turned on to diagnose a problem but is not disabled afterwards.
4. Enterprise Cloud Platform Functionality
Data from your digitized electrical network will often be aggregated at the enterprise level in the cloud. Here, you will typically build a digital twin of the network. Data quality issues from lower in the architecture will also aggregate, intersect, and multiply in this tier. One example is mismatched logging intervals. This is when some devices are logging a measurement at one internal (e.g. once every 10 minutes) and another device is logging the same measurement at a different interval (e.g. once every 15 minutes). A demand forecasting application would find it difficult to process that data.
Returning to the airport example, imagine if, after the expansion, no meters were added to the new gates and tenants. Now the power analytics has a blind spot and cannot include the new space in its algorithms, causing potentially misleading insights. Problems can also occur if devices are not updated with the most recent firmware versions, meaning that measurements may be calculated differently from one device to another, or that data is vulnerable to tampering due to a device not having the latest cybersecurity updates.
How to Ensure Good Data Quality
The good news is that data quality problems can be detected and corrected with a power analytics solution that includes data quality capabilities. If your team does not have the time or expertise to use these tools, you can call on expert support, such as EcoStruxureTM Power Advisor services.
Data quality validation should be run immediately after commissioning and then on a continuous or periodic basis. It is important to monitor progress over time because situations are dynamic. You may have personnel changes (sometimes without full transfer of knowledge), process changes, maintenance contractors performing adjustments, a facility expansion, or new firmware installed in devices. These activities can affect data quality.
When problems are detected, they must be prioritized for correction. You will typically maintain a backlog of issues and resolve them in order of importance. Best-in-class tools give you an aggregate “data quality index” to track total data quality health over time.
The bottom line: Rather than assuming that your system is providing perfectly trustworthy data, assume that you have data quality issues and work defensively. Remember that 98 percent of installed systems have quality issues. Once you can trust your data, your power analytics system can help you to better manage operations and to identify risks and opportunities that can improve the resilience of your electrical network.
For More Information
The IoT-enabled, future-proof EcoStruxure Power architecture from Schneider Electric supports digitized power distribution, enabling enhanced connectivity, real-time operations, and smart analytics. EcoStruxure Power Advisor is a proactive, analytics-based service for your power management system, delivering optimized energy performance, power reliability, and resilience.