The functions important for the safety of the operator must be identified. This can be fairly easy for machines which have an obvious main risk. It will be harder to identify all dangerous situations for machines which contain multiple risks. Information of the risk analysis is presumed to be available as input to the assessment method discussed in this report.
Before an assessment can be made of any control system, the machinery has to be identified and delimited. The functionality and the safety principles must be understood by the engineers who are to perform the assessment. Important questions are;
The first step of the assessment will then be to check if the required safety-related functions exist. All machinery are affected by requirements from directives and standards. The assessment must answer the questions;
A design fault may adversely affect the required function. Even if the intention is to have the function implemented correctly in the machinery, it will have to be checked.
Requirements for behaviour at fault will be made. Fault simulation, analysis of redundancy, hardware analysis and test and software analysis and test will answer the question
Figure 14. Questions during the assessment.
The validation of a machine control system based on distributed control requires the evaluation of aspects not present in a conventional control system. That will raise new questions of what has to be evaluated to verify adequate safety of the machinery. The objective is to develop an assessment method which is capable of answering these specific questions.
This study has grouped the errors into node errors, bus errors, timing errors, data consistency errors, initialisation or restart errors, 'babbling idiot' errors and configura-tion errors. An assessment method has to address all these errors.
Node errors have been subject for many scientific papers and research. The operation of the control system is depending on the correct operation of all nodes. Examples of questions to address are:
Bus errors
Timing errors are perhaps most commonly addressed when discussing errors in distributed systems. An evaluation will have to answer at least following questions:
Data consistency errors occur when co-operating nodes use data of different age. Useful checks include;
Initialisation or restart errors occur at the start up sequence of the control system. To verify the restart procedure, the following questions may be asked:
'Babbling idiot' is the phrase used to describe when a node is constantly transmitting and occupying the bus.
Configuration errors are the result of user faults when connecting and configuring the nodes on the bus. Examples of questions of configuration errors are:
The assessment method will probably consist of several validation methods which can be used to answer one or several of the questions above. These validation methods should be possible to use as tools for assessment of many different types of machinery.
The validation methods will all require documentation of the control system. Suggestions for what information the different documents must include will be required.
Back to Contents Continue to read the report; Chapter 8
© Kvaser AB - All rights reserved | Contact us |
Webmaster | SitemapDirect link to this page: