CN
Failure Modes and Their Analysis – From Theory to Field Practice
Reliability testing in the laboratory is essential, but it tells only half the story. The real test takes place in the field, where products are exposed over many years to a wide range of stresses that can only be roughly simulated in the laboratory. The systematic analysis of failure modes—both during the testing phase and after delivery—is the key to continuous improvement and the validation of design decisions. It closes the loop between theory, testing, and practice, making it possible to learn from every failure and make future generations more robust.
FMEA: Systematic Risk Analysis in the Early Stages
Failure Mode and Effects Analysis (FMEA) is a proactive method for identifying potential failure modes as early as the design and process planning phase, before a single product is manufactured. It is the opposite of a reactive approach, in which one only responds to problems once they arise. FMEA forces the development team to systematically ask: “What could go wrong, how likely is it, how severe would the consequences be, and how well can we detect it?”
Potential failure modes are defined for every element of the HMI stack—from the adhesive to surface preparation to edge sealing. For example, the failure mode “delamination of the optical adhesive” could be caused by inadequate surface preparation, moisture ingress, or CTE mismatch. The consequence would be a total failure of the optical system. Each failure mode is evaluated based on three criteria: Severity (S) on a scale of 1 to 10, where 10 represents a catastrophic failure; Occurrence (O) from 1 (very unlikely) to 10 (almost certain); and Detection (D) from 1 (easy to detect) to 10 (practically undetectable). The product of these three numbers yields the Risk Priority Number (RPN), which can range from 1 to 1000. Failure modes with a high RPN are prioritized and addressed through design changes, process improvements, or additional controls.
The FMEA is not a one-time document, but a continuously used tool that is updated throughout the entire product lifecycle. It serves as the basis for informed decisions regarding material selection, process parameters, and quality control measures.
The Bathtub Curve: Three Phases of a Product’s Lifespan
The bathtub curve is a classic model in reliability engineering that describes a product’s failure rate over its lifespan. It is so named because its shape resembles the profile of a bathtub: a high failure rate at the beginning, a long phase of low failure rate in the middle, and a rise at the end.
The first phase, infant mortality, is characterized by a decreasing failure rate. In the first weeks to months after commissioning, failures occur that are caused by manufacturing errors, material defects, or process weaknesses. A particle trapped during lamination, an inadequately cleaned surface, or a faulty edge seal often manifest quickly. These failures are annoying, but they also present an opportunity: they show where the process needs to be improved. Through burn-in tests or HASS (Highly Accelerated Stress Screening), these early failures can be induced and the defective units sorted out before they reach the customer.
The second phase, Useful Life, is the goal of every design. Here, the failure rate is constant and low. Failures in this phase are random and usually caused by external events—such as power surge damage, mechanical damage from an accident, or an unforeseen extreme event. A well-designed and manufactured HMI component spends the majority of its lifespan in this phase, which ideally lasts ten years or longer.
The third phase, Wear-Out, is characterized by an increasing failure rate. Materials age, adhesives lose their elasticity, UV damage accumulates, and mechanical fatigue sets in. This phase is inevitable, but its onset can be significantly delayed through material selection and design. The use of silicone-based adhesives is an example of a measure that delays the wear-out phase.
Reliability Metrics: MTBF, MTTF, and FIT
To quantify reliability, standardized metrics are used that allow products to be compared and service life predictions to be made.
The Mean Time Between Failures (MTBF) is the average time between two failures in repairable systems. It is calculated as total operating time divided by the number of failures. An MTBF of 100,000 hours means that, on average, a failure can be expected every 100,000 operating hours. For an HMI that runs 24 hours a day, this corresponds to approximately 11.4 years. However, MTBF is only meaningful if the system is repaired and put back into operation after a failure.
For non-repairable components, the Mean Time To Failure (MTTF) is the relevant metric. It describes the average time until the first (and only) failure. MTTF is calculated in the same way as MTBF, but the interpretation is different: it is the expected service life of the unit.
For highly reliable components with extremely low failure rates, MTBF values in the millions of hours are difficult to grasp. This is where FIT (Failures In Time) comes into play. FIT indicates the number of failures per billion operating hours. An FIT value of 100 corresponds to an MTBF of 10 million hours. FIT values are particularly common in the semiconductor industry.
These metrics are not just theoretical numbers. They form the basis for warranty calculations, spare parts planning, and life cycle cost analyses.
Weibull Analysis: Statistical Life Span Prediction
Weibull analysis is a powerful statistical tool for analyzing life span data. It allows conclusions to be drawn about the entire population from a limited number of test data points and enables the prediction of failure probabilities for any given time point.
The Weibull distribution is characterized by two main parameters: the shape parameter (β) and the scale parameter (η). The shape parameter is particularly revealing, as it indicates the nature of the failure mechanism. A β less than 1 indicates early failures (infant mortality)—the failure rate decreases over time. A β equal to 1 corresponds to a constant failure rate, as expected during the useful life phase. A β greater than 1 indicates an increasing failure rate, typical of wear-out failures. Values of β between 3 and 4 are characteristic of fatigue failures.
The scale parameter η indicates the characteristic lifetime at which 63.2% of the products have failed. By fitting a Weibull distribution to test data, engineers can predict what percentage of products will still be functioning after 5, 10, or 15 years. This is crucial for determining warranty periods and planning for spare parts.
Weibull analysis is also an excellent tool for comparing different designs or materials. For example, if two adhesives are tested and one exhibits a higher η value at the same β, it is clearly more durable.
Root Cause Analysis: From Symptoms to Root Causes
When a failure occurs, the first question is not “What happened?” but “Why did it happen?” Root Cause Analysis is the systematic process of moving from measurable defects to the underlying causes.
A proven method is the 5-Why Technique, in which you ask “Why?” five times in a row. Example: A display shows delamination. Why? Because moisture has penetrated. Why? Because the edge sealing failed. Why? Because the edge sealing was not applied completely. Why? Because the dispenser was incorrectly calibrated. Why? Because regular maintenance was neglected. The fifth answer leads to the root cause: a maintenance deficit.
The Fishbone Diagram (Ishikawa Diagram) is another method that organizes causes into categories such as Material, Method, Machine, Man, Measurement, and Environment. This helps to systematically consider all possible influencing factors.
Sophisticated techniques are used for the physical analysis of failed components. Optical microscopy enables the visual inspection of defects. Scanning Electron Microscopy (SEM) provides high-resolution images of surfaces and fracture surfaces. Energy Dispersive X-ray Spectroscopy (EDS) identifies chemical elements and can detect contamination. Fourier Transform Infrared Spectroscopy (FTIR) analyzes the chemical composition of materials and can detect degradation. Cross-Sectioning (cross-sectional analysis) enables the examination of internal structures and interfaces.
The findings from the root cause analysis are directly incorporated into the FMEA, update process parameters, and lead to design improvements or material and process adjustments. This is the continuous improvement cycle that increases the reliability of products from generation to generation.
Field Failure Statistics: The Ultimate Reality Check
Laboratory tests are controlled and reproducible, but the field is chaotic and unpredictable. Field failure statistics are therefore the ultimate reality check for any design. They show how the product behaves under real-world conditions, which often differs from the assumptions made in the laboratory.
The comparison between air-gap displays and optically bonded displays in field statistics is clear. Air-gap systems exhibit a significantly higher failure rate, with internal condensation being by far the most common failure mode. In humid or variable climates, the failure rate of air-gap displays can reach 10% or more within the first five years. Dust ingress and mechanical breakage due to shock are other common issues.
Optically bonded displays exhibit a drastically lower failure rate. Studies and field data show that the MTTF is 3 to 5 times higher. The primary failure modes are delamination (which is rare with proper material selection and process control) and yellowing (which occurs only with low-quality acrylate adhesives lacking UV stabilizers). Mechanical failures are significantly reduced, as full-surface bonding increases shock resistance.
These field statistics are not just numbers — they form the basis for Total Cost of Ownership (TCO) calculations. An HMI that fails in the field not only incurs replacement costs but also leads to downtime, lost productivity, and reputational damage. A more reliable system justifies a higher purchase price, as it can significantly reduce costs over the entire product lifecycle in the field.
Note on the technical information in this chapter
The technical values, test profiles, and performance data stated in this chapter are provided for technical classification purposes. Actual results depend on the specific display configuration, material system, bonding process, application profile, and scope of qualification. Product-specific values are validated on a project basis using data sheets, measurements, test reports, and customer specifications.