Vibration: What you think you know and why it might not matter
Vibration monitoring is one of the most effective ways to detect and prevent premature and even catastrophic equipment failure. Vibration often indicates faults such as; imbalance, misalignment, looseness and late stage bearing wear – all early warning signs of impending failure. Internet of Things (“IoT”) platforms are an effective way to perform vibration analysis. As such IoT platforms are increasingly deployed at the center of Condition Based Monitoring maintenance programs. IoT’s ability to enable plant technicians to remotely view the operating condition of critical assets is a major plus. Below, we discuss some important characteristics that set IoT vibration monitoring apart from more traditional methods.
IoT vibration monitoring begins with sensors attached to equipment. Information from the sensors is transmitted wirelessly or via a wired connection to a server. The server analyzes the collected information and transmits the data to authorized users. Vibration information is collected in either the time domain (vibration energy magnitude) or the frequency domain (vibration spectral energy). For vibration magnitude, sensor information is ideally measured in X/Y/Z axes allowing operators to understand the dominant vibration direction. For vibration spectral energy, a more complex understanding of the assets rotational speed, vibration frequency, and harmonics are required. Fast Fourier Transforms (FFT) are often used to separate noise from the data while highlighting peaks in the energy occuring at a specific frequency or frequencies.
A key differentiator between IoT vibration monitoring and specialized vibration analysis is that IoT sensors are predominantly used to capture changes in the vibration energy magnitude while specialized vibration analysis is used to analyze vibration spectral energy. The translation of vibration energy magnitude into a high-level assessment of “good/warning/bad” is straight-forward. It can be performed using either vibration acceleration or velocity- velocity being the dominant metric that is explicitly defined in many ISO standards for industrial equipment. Conversely, evaluating vibration spectral energy is not an easy task. It involves taking multiple vibration measurements at several different locations, using extremely sensitive/expensive equipment. Measurements are then post-processed using specialized software at the hands of experienced technicians. In short, vibration magnitude-based analysis is ideal for widely deployed, low cost IoT sensors that are typically static in position, while vibration spectral analysis is best suited for skilled technicians using highly specialized and expensive equipment, and then only when it is needed.
The magnitude of vibration is a great leading indicator delivered to plant personnel from IoT condition-based monitoring sensors. However what is straightforward in concept, is not always easily implemented in practice. Competitive performance testing (see table) demonstrates that the absolute magnitude of vibration measurement can vary significantly from vendor to vendor- even as much as a 24% difference measuring the exact same vibration in the exact same location, compared against a known reference. Considering an industry standard such as ISO10816, a 24% deviation from actual is not only significant, but can be the difference between normal operations and impending failure. Factors such as sensor weight, accelerometer accuracy, sampling frequency and how the raw data is processed all play a role in the measured accuracy.
To complicate matters, many users are unaware of significant factors that can impact accuracy (shown in the list to the right). When combined, these installation factors can add up and bias the measurement by as much as 2X! For this reason, comparing sensor accuracy between a spec sheet and a real world result can be challenging. Deploying systems with marginal accuracy can send a user down one or two paths- 1) they are led to believe that everything is “normal” when in fact there is a significant underlying issue (false negative), or 2) they are alerted to an “urgent issue” when in fact operations are normal (false positive).
When sensor accuracy is poor, maintenance people are often forced to spend time focused on the perceived bad actors- while they miss actual impending failures.
So, what should you do? First and foremost, realize that in condition monitoring specific to vibration, the absolute vibration magnitude usually only matters to a vibration expert. Rather than focusing on an absolute number, users keeping track of relative changes from normal are better able to focus their more sophisticated vibration analysis routines on equipment that needs it the most. This helps to reduce cost and maximize the ROI of vibration analysis services.
Fortunately, for maintenance professionals, IoT condition monitoring devices, even if not highly accurate, are low cost, very repeatable and linear. This makes these products perfect candidates for trend monitoring, acting as 24/7 equipment watch dogs. In this capacity, the job of the IoT vibration condition monitor, just like the job of the maintenance engineer who previously did spot checks, is to identify when an expert is needed for more advanced root cause analysis. The key difference though, is that IoT monitors can watch over every piece of equipment simultaneously, 24 hours a day, 7 days a week, 365 days a year.
At Preddio, we make sure our OEMs and Service Professionals understand the clear distinction between detection versus analysis, and never mix the two. We do not overcomplicate our condition monitoring sensors and instead, only focus on presenting the key metrics that are most important for our general users.
In condition-based monitoring applications– simplicity is most important. In our experience, IoT products and systems that are simple to install, simple to use, but also provide powerful cost savings always result in customer satisfaction and broad usage. Want to learn more? Contact us at firstname.lastname@example.org