Bad data collected by a remote monitoring system can wreak havoc on industrial organizations.
When prevalent in business processes and analysis, bad data creates a fundamental disconnect between what is actually happening in the field and what decision-makers think is happening in the field. Put simply, bad data misrepresents reality. It “lies” about the current – and historic – state of operations to those who need quantitative context and intelligence.
Bad data can fly under the radar, making it one of the scarier data problems that companies face. Bad data often checks all of the right boxes and disguises itself as high-quality intel. It may arrive at the analytics engine or show up in reports on time without missing any information, just as uncompromised data would. It may also refer to assets correctly and tell a believable story.
However, at the end of the day, if data doesn’t describe the real world accurately, it’s bad data. And industrial leaders with bad information are dangerous. They can push their organizations down the wrong paths, invest capital inappropriately, and forego valuable opportunities.
Once the wrong decision is made, industrial businesses can struggle to recover from the repercussions of bad data. This is why it is such an important topic to address as industrial companies engage in digital technology projects, such as asset digitization or IIoT, to improve business outcomes.
What causes bad data?
There are many reasons why industrial companies collect bad data.
Improper calibration, failed sensors, and conversion errors are just a few of the ways that remote monitoring systems fall prey to bad data. These issues can have massive implications on how businesses manage their physical assets.
For example, an incorrectly calibrated pressure sensor could send the wrong message about how a physical asset is performing.
In this case, a sensor calibrated for 2,000 PSI should output 12 mA for a pressure reading of 1,000 PSI. If that sensor, instead, was configured as a 3,000 PSI device, a 12 mA signal would correspond to 1,500 PSI.
On the conversion front, digital project managers need to be able to understand and communicate how data is transformed as it travels from the field to the office. For example, a temperature sensor configured to output data readings in Kelvin may lead to confusion further downstream. Analysts may incorrectly assume that information is already in a familiar standard, such as Celsius or Fahrenheit, when it hasn’t yet been converted.
Or, they may unknowingly duplicate a conversion that already happened at a previous step. Devices that are configured with incorrect multipliers or offsets can compromise final data outputs dramatically.
These are just a few examples of how bad data is born. Overall, there are many ways digital signals can misrepresent physical assets. Developing a sophisticated process for evaluating data integrity is crucial.
Who is affected by bad data?
Bad data affects everyone, from field personnel all the way up to the executive suite.
Office-based analysts are often the most likely to be led astray. They are typically tasked with performing large-scale analyses of operational processes. They have so much data to synthesize and process, which means they are less likely to notice errors associated with individual assets in the remote monitoring system.
For example, a level reading of 25 feet created by a poorly calibrated tank sensor on a 20-foot tall tank would raise a red flag for any field personnel who closely monitor that specific asset and know how tall it is.
Office-based analysts, on the other hand, wouldn’t have the context to recognize such an issue. They may also not have the capacity to sift through large datasets for errors. It’s not uncommon for analysts to aggregate information for hundreds or thousands of assets simultaneously. Bad data for one or a handful of assets can easily get buried and skew results.
When analysts have bad data, they draw incorrect conclusions. Incorrect conclusions can cause leaders to make unnecessary adjustments to field processes and assets. Consequently, digital projects don’t achieve their intended purposes or efficiencies.
Office analysts aren’t the only ones affected by bad data. Executives and board members often make important decisions about company direction and strategy based on a synthesis of operational metrics. If these metrics are skewed just slightly, it may be enough to push a decision over the edge – or prevent one from being made.
Even investors have to avoid bad data. For asset managers or active, short-term investors looking to fund industrial companies, bad data can lead them to misunderstand the operational or financial state of the business and make a poor investment. Consider the rising interest in ESG investing. Sustainability-conscious investors often have difficulty getting the data they need to make investment decisions, and bad data can be a key culprit.
What do we need to consider?
In some cases, it may be hard to identify the exact cause or source of bad data.
Because there are so many devices involved in digital transformation projects, personnel may have to evaluate every component. And, in legacy OT network architectures, each component might be managed by a different person, different team, or different organization. As a result of how system management is often dispersed, finding bad data in SCADA, BMS, or any other remote monitoring system can be especially challenging.
Additionally, many sensors are programmed to send alerts only if readings deviate from the norm. Therefore, devices may be quiet for long periods — up to several months. As an operator, silence can mean one of two things: everything in the system is normal, or a sensor is broken. Discerning between the two is hard.
When it comes to implementing digital technologies, we have to be mindful of who has accountability over certain assets, as well as how various remote monitoring systems fit into existing process controls. Neglecting these considerations can be the difference between good and bad data.
What’s at stake for your business?
Bad data can be a major problem for those who want to improve their businesses with digital technology. Digital analysis built on misinformation can quickly eliminate cost savings or value creation potential.
Below are a few considerations that may help you mitigate bad industrial data problems:
- Who in your business has permission to configure and calibrate sensors?
- Who is responsible for maintaining data integrity?
- How often is your remote monitoring system data validated for accuracy?
- Do you proactively evaluate field sensors?
- Do your office-based analysts perform regular “gut checks” on their datasets?
At WellAware, we understand how detrimental bad data can be for your business. We help companies overcome industrial data challenges and optimize digital projects over the long term through a multi-layered strategy.
Worried you might have a bad data problem?