Failure classification
To construct the relationship between failure modes and causes, each repair order was reviewed. As shown in Fig. 2, all failure modes were divided into 36 types, and all the failure causes were divided into 39 types. According to the location in the system, the failure modes were further clustered into 6 categories, namely, the control panel, probe, appearance, display unit, host system and “other”. The control panel category included failures related to the control panel, such as touch screen, trackball and control key issues. The failures of the display unit mainly manifested as unclear images, striped black shadows and black screens. The failures of the host system mainly included not powering on, shutting down and failing to charge. The probe failures were characterized by bubbles on the probe surface, a broken shell or cable and unrecognized probe errors. The appearance failures included roller malfunction, inability to move up/down, casing/cable damage and stent fracture. Other failure modes included being unable to transfer images, being unable to burn, print, or export data and having unheated couplants. The failure causes were further categorized into 12 major cause types, including panel failure, environmental reasons, software failure, hardware processing unit failure, power supply failure, poor contact, human error, probe failure, monitor failure, appearance or mechanical failure, data transmission failure and peripheral fault. Each major cause category included one or more cause types.
Through the relationship between each failure mode and the corresponding cause, the relationship between the location of failure modes and types of failure causes was established. The number of failures in the Sankey diagram is represented by the thickness of the lines. The location of a failure is not necessarily related to the cause of the failure. The phenomenon of being unable to power on the device was related to 13 types of causes, indicating that the failure could not be assessed very easily.
For the failure modes, the most common modes were control key malfunction (17.55%), unable to power on (11.69%) and unclear image (10.37%). As shown in Fig. 3a, the problems mainly occurred in the host system, display unit and control panel, with 32.5%, 26.9% and 26.7% of failures associated with these components, respectively. Regarding the failure causes, the most frequent ones were control key failures (16.18%), software system failures (10.13%), board failures (6.64%) and probe crystal failures (6.40%). As shown in Fig. 3b, of the major cause categories, panel failures (24.78%), probe failures (16.75%) and hardware processing unit failures (14.66%) were the three most frequent failure types. Additionally, 5.83% of failures were caused by human error, and 5.72% of failures were related to environmental factors.
According to the correspondence between every failure mode and cause, visual correlation analysis was conducted for the six kinds of failure mode locations and twelve major types of failure causes, as shown in Fig. 4. The main reason for the failure of the control panel was the panel itself. The main faults in the host system were related to the hardware processing unit, software and power supply. The specific causes of display unit system failure were complex and included probe failure, hardware processing unit failure and environmental conditions. In addition to probe failure, other probe issues were caused by poor contact, software failure and hardware processing unit failure. The failures generated in the appearance system were all mechanical faults. Other failures were mainly related to data transmission problems and human operator error.
RPN of the failure mode
To quantitatively evaluate the risk of failure modes, the RPNs of different failure causes were determined first. The severity and detectability of the failure causes were scored using the Delphi expert consultation method. To assess the occurrence of failure, the proportion of failures in each category was used to represent the frequency. On the basis of severity, occurrence and detectability, the RPNs were calculated for the failure causes. As shown in Fig. 5, RPNs of 39 failure causes categorized in 12 types were obtained. Different failure cause types were distinguished by different colors. It could be seen that the RPNs of probe failure and hardware processing unit failure were the highest. For specific failure cause, probe circuit faults, board failure and probe crystal failure were the failure causes with the top three priority risk scores.
According to the calculation mode in the Methods section, the RPNs for 36 categories of failure modes were determined. As depicted in Fig. 6, unclear images, inability to power on and dark shadows on an image had the highest RPNs, which meant that such issues could potentially had the greatest impact on ultrasonic diagnostic procedures. Based on the failure mode severity calculation results, dark shadows on an image were the most crucial issue, with lack of information and abnormal noise also receiving high scores. Couplant heating and roller malfunction issues had the least severe impacts. With regard to detection difficulty, an unrecognized probe and lack of image information were the least common issues. Control key malfunction was the most frequently occurring failure type, which suggests that this kind of failure occurs frequently. However, the severity and detection difficulty of such failures were very low, which caused scant attention given to such issues.
Then, according to the correspondence diagram of the failure mode and failure cause, the RPNs of failure mode and causes were connected by the failure classification. As shown in Fig. 7, the RPN of the failure mode and failure cause were not exactly one-to-one correspondence, which was because that one failure mode might be caused by multiple failure causes. Only when the failure mode corresponded to one failure cause, their RPNs were the same. When a failure of ultrasound device occurred, this diagram could not only help physicians evaluate the risk of this failure mode, but also find out the most likely failure causes according to its path tracking, so as to take corresponding measures to solve the failure.