Automation failures

Y. Papadopoulos, D. Parker, A method and tool support for model-based semi-automated failure modes and effect analysis of engineering design, in Brisbrane Conference in Research and Practices in Information Technology, vol. 47, University of Hull Christian Grante Volvo Car Corporation Australian Computer Society Inc., 2004. 

Factors that have been investigated in empirical studies as possible influences in people s vulnerability to automation bias include individual differences among operators [5, 13, 16, 17] people s accountability for their own decisions [17] the levels of automation at which the computer support is provided [12, 18] the location of computer advice/warnings with respect to raw data or other non-automated sources of information [19, 20] people s exposure to automation failures [21]. 

Subjective measures of the trust of human operators in a computer tool have been found to be highly predictive of people s frequency of use of the tool [5, 30]. Use of automation (or reliance in its generic sense) is usually assessed with observations of the proportion of times during which a device is used by operators or by assessing the probability of operators detecting automation failures [19]. 

In this paper, we propose a safety analysis technique, failure propagation and transformation analysis (FPTA), which follows the direction of FPTC [15]. The FPTA method integrates an automated failure analysis algorithm presented in [15], and it also allows the application of model checking technique as provided by the PRISM model checker [7]. The approach is therefore a probabilistic safety analysis technique for component-based system development. 

Insufficient information about the properties, layout pattern of small defects, potential for their growth in time, usually leads either to an unjustified rejection (repair) or to underestimation of the importance of the defect and, as aconsequence, construction failure. Use of automated computerised means of control allows safe service of the old constructions, periodically repeating the UT and monitoring the development of discontinuities in the metal. The main idea of such policy is periodical UT of development of discontinuities or, in a more general form, monitoring of the metal condition. 

The technology involved makes recirculating systems expensive to constmct and operate. Redundancy in the system, ie, providing backups for all critical components, and automation are important considerations. When a pump fails, for example, the failure must be instantly communicated to the culturist and the culturist must have the abiUty to keep the system operating while the problem is being addressed. Loss of a critical component for even a few minutes can result in the loss of all animals within the system. 

For example, the characteristic dimension A on the cover support leg was critical to the success of the automated assembly process, the potential failure mode being a major disruption to the production line. An FMEA Severity Rating (S) = 8 is allocated. See a Process FMEA Severity Ratings table as provided in Chrysler Corporation et al. (1995) for guidance on process orientated failures. The component cost, Pc = 5.93 and the number planned to be produced per annum, N = 50000. 

Example 2.5 Failure of Automated Process System Because Critical Information Was Not Displayed... 

As failure data relating to mechanical components differ widely from source to source, TNO has set up a documentation system in which all relevant information is stored in one, uniform automated code called COMPI, which uses a component description code for the following information system of... 

Once it is determined that data exist, the next step is to begin the collection process. If sufficient thought and training is provided in the development and operation of the maintenance and operating reporting systems, much of the collection process can be automated. Automation assumes that a well-thought-out taxonomy is in place. If this is not the case, then an analyst must collect and review the records manually. In either case, the analyst must collect data from the plant sources previously discussed in order to determine the numerator (number of failures within a unique plant equipment population), and denominator (the operating time or number of demands for the equipment) of the equation to calculate failure rates. 

The second area will be feedwater pumps. It is normal to have a duplicate standby feedwater pump. Sometimes this may be two for each boiler or one duplicate pump to serve any selected boiler. These will usually require manual changeover in the even of failure of the duty pump. It is practical to automate this changeover by using pressure sensors and motorized valves. The same can apply to oil-circulating pumps, gas boosters, water-treatment plant and any other valves and motors. It is possible to do most things, but in the end there is the cost to be considered. An... 

While providing many advantages, simplified data acquisition and analysis also can be a liability. If the database is improperly configured, the automated capabilities of these analyzers will yield faulty diagnostics that can allow catastrophic failure of critical plant machinery. 

Bradley P, Chivian D, Meiler J, Misura KMS, Rohl CA, Schief WR, et al. Rosetta predictions in CASP5 successes, failures, and prospects for complete automation. Proteins 2003 53 457-68. 

For walkways, where 3 foot-candles is an adequate amount of illumination, two 175-watt R-40 mercury vapor lamps placed every 54 ft may be specified. For these and areas that are used continuously but have adequate windows it may be assumed that the lights are on 4,500 hours per year. (This is approximately half the time.) For some interior areas they may never be turned off. At the other extreme, Sarah Lee has an automated warehouse in which the only time lights are needed is when there is an equipment failure. 

Nevertheless, the acquisition of a sufficiently detailed body of physical information can allow a formulator to go far beyond the mere ability to cope with crises when they develop at unexpected times. For a well-understood system, it is theoretically possible to design an automated or semi-automated manufacturing scheme for which the processing variables would be appropriately controlled so as to minimize the possibility of batch failure. Materials passing the hurdles of physical test specifications would be totally predictable in their performance, and they could therefore be blended, granulated, dried, compressed, and delivered into containers without operator intervention. 

However, many recent instruments are still not considered satisfactory, since professional developers in the field of high-throughput screening (HTS) want to use the full performance of the latest generation of robots and computers for automation. This results in new instrumental developments, like the possibility of reading not only 96, but 384 or even 1536 wells plates as well as DNA chips, very rapidly (in a minute or so) and repeatedly without any mechanical failures. Hence, in the eyes of company scientists developing new assays, many present-day instruments still correspond to an intermediate stage of development. For research laboratory scientists, on the other hand, the actual equipment offers excellent performance. 

Simpler plants are friendlier than complex plants because they provide fewer opportunities for error and because they contain less equipment that can cause problems. Often, the reason for complexity in a plant is the need to add equipment and automation to control the hazards. Simplification reduces the opportunities for errors and misoperation. For example, (1) piping systems can be designed to minimize leaks or failures, (2) transfer systems can be designed to minimize the potential for leaks, (3) process steps and units can be separated to prevent the domino effect, (4) fail-safe valves can be added, (5) equipment and controls can be placed in a logical order, and (6) the status of the process can be made visible and clear at all times. 

These authors reported greatly improved sample transfer without pipette failure due to plugging caused by thrombin clot formation when a LEAP HTS PAL autosampler was used for liquid transfer automation. 

Laboratory automation in pharmaceutical analysis attained maturity since robots first appeared in pharmaceutical laboratories more than 20 years ago. While automation offers great promise for improving sample throughput and reducing sample backlog, its implementation has not been without problems. The industry cannot invest heavily in tools that produce little return on investment. Strategies in key aspects of automation such as planning, vendor selection, personnel, and efficient use of systems can determine the success or failure of an automation project. 

In the practical world, results not only need to be reproducible but also transferable. This requirement helps assure that differences in apparatus for the purpose of automation do not interfere with the method and demands a validation to demonstrate equivalency. Designs which diverge from the strict USP and industry convention run the risk of developing a system that cannot be validated at the specific method level. The authors have personally observed cases where extremely subtle changes in apparatus resulted in a failure to demonstrate suitability. 

Advances in information technology and the necessity of improved efficiency have resulted in increasingly automated and interlinked infrastructures, and have created new vulnerabilities due to equipment failure, human error, weather and other natural... 

Architecture. Many common practices negatively affect SCADA security. For example, while it is convenient to use SCADA capabilities for other purposes such as fire and security systems, these practices create single points of failure. Also, the connection of SCADA networks to other automation systems and business networks introduces multiple entry points for potential adversaries. 

---