System supervisor

The outermost structure for our KNODE is illustrated in Figure 1, which shows the major functional components and flow of data. Scientific data, in the form of numbers specifications of chemistry, physics, or geometry algorithms code or remote host output, generally flows clockwise around the diagram as indicated by solid arrows. Control flow is generally outward from the System Supervisor module. Data in the form of knowledge may flow in various directions as indicated by shaded arrows. 

The System Supervisor. The system supervisor is the administration component of KNODE. It must define and control access to the other modules of the system as well as manage most inter-module communication. A command interface will serve as the user s primary access to KNODE. The supervisor will translate user requests into appropriate instructions to be passed on to the other modules. The software modules of the system will be independent programs which communicate with each other by passing information through the supervisor and through information stored in intermediate files. 

The System Supervisor must provide the Job Controller with information about the program, including the computational host ID, the number and names of all the input and output files used by the program, and the names of the libraries or pools called by the program. The Job Controller will also have access to all the information necessary to operate the scientist/user s accounts on each computational host. Using the information obtained from the System Supervisor, the Job Controller will be able to decide which job commands are necessary to attach specified input and/or output files and to attach specified software libraries. The Job Controller will also be able to choose host parameters that are machine specific. 

Eventually, the Job Controller will be able to package and submit programs from any host to any computational host, transfer all output files back to the return host, and notify the System Supervisor of the termination of the process. With these capabilities the Job Controller can be used not only with the KNODE project, but with other knowledge based systems with similar implementiation or as a stand-alone asynchronous Job Controller. 

There are a few notable communication problems that could be encountered between the computational host and the Job Controller. One problem that occurs regularly is a faulty computational host, that could abort the batch job abnormally. If this occurs, the System Supervisor must be notified, given the corresponding error code, in order to take the proper action. Another problem can be encountered when a communications link between the computational host and the return host is down. This will cause the output files to be LOST until the link is back up and running properly. 

We must also mention some problems that may occur when the output files are being transferred back to the return host. If, for instance, the return host has a full file system, the output files cannot be saved immediately, but the output files cannot return to the computational host either. The output files must reside on some type of scratch space until the System Supervisor can be notified. 

Under the new system, supervisors could monitor the relative progress of different lanes and adjust allocation of workers to keep the lanes coordinated (to reduce sorting downstream). Bucket brigades simplified the shifting of workers from a fast lane to a slow one because there is no need to redefine zones. The research measured a 34% increase in pick rates at Revco when they shifted to a bucket-brigade-based task allocation. 

Management at all levels requires safety education and training. In the most effective safety systems, supervisors are expert in their knowledge of the hazards and safe procedures required in their departments. Super-... 

Early Activities. Historically, quahty was entmsted to the artisan, who was solely responsible for the products made. Thus financial success often rested on product quahty. The industrial revolution replaced this system with one in which product quahty was the result of the combined efforts of a group of factory workers. Quahty was ensured through the combination of worker skills and production supervisor monitoring. 

Forty-hour training was required for personnel entering the exclusion zone, and additional supervisory training was required for site supervisors. Site control procedures described in the Site C contractor s SSAHP included maintenance of site control logs at each access point, use of red tape or chainlink fencing to demarcate hot zones, and use of the buddy system in all exclusion and contamination reduction zone areas. Site communications relied almost exclusively on visual sighting of employees the plan did not describe the use of two-way radios. This suggested that all employees in hot zones can be observed continuously from the support zones. 

A common solution, which satisfies both the inventory control and quality control, is to institute a stock requisition system. Authorization of requisitions may be given by a person s supervisor or can be provided via a work order. If someone has been authorized to carry out a particular job, this should authorize the person to requisition the items needed. Again for inventory control reasons, you may wish to impose a limit on such authority requiring the person to seek higher authority for items above a certain value. 

Sometimes managers and supervisors lose interest, and unknown to them, operators stop carrying out procedures. However, shutting the flare system down for repair and taking the refrigeration system out of use were not decisions operators would make on their own. Managers must have made these decisions and thus showed a lack of understanding and/or commitment. 

In all of these cases, the extent and nature of supervision can vary greatly. Like management systems themselves, supervision can be results-oiiented or procedurally focused, depending on the preferences of the company and the individual. One boss may tell her staff, "Fix that problem we re having with the product specs in the ag-chem division," and expect to hear no more about it until the problem is resolved. Another supervisor with the same problem may expect a detailed situation analysis, formal recommendations, and con-... 

Corporate MOC team Develop management of change system for all sites W. Hiller, Plant Maintenance Supervisor... 

An opportimity for error recovery would have been to implement a checking stage by a supervisor or independent worker, since this was a critical maintenance operation. However, this had not been done. Another aspect of the unforgiving environment was the vulnerability of the system to a single human error. The fact that the critical water jacket flow was dependent upon a single pump was a poor design that would have been detected if a hazard identification technique such as a hazard and operability study (HAZOP) had been used to assess the design. 

Disadvantages may arise because the behavior observed may not be fully realistic. A static simulation, for instance, may not reveal the true nature of operators dynamic interaction with the system. There is also the possible disadvantage of behavior in a simulator not fully replicating that found in the real situation. This can happen because of the absence of real stressors found in the actual task, for example, risk to life, criticality of the process, and presence of other workers and supervisors. 

This phase of the program is designed to ensure that the information produced by the system is fed back to all levels of the workforce, including process operators, managers, supervisors, engineers, and senior policy makers. 

Started. Since pump A and its associated pipework was off-line, the supervisor took the opportunity to carry out scheduled maintenance on the pressure relief valve (PRV) downstream of pump A. The valve had been malfunctioning, and although the work was not scheduled to be done for some weeks, the specialist contractor team who maintain the PRVs had a team available to carry out the work immediately. The supervisor therefore now had two teams working on the pump A systems the shift maintenance team working on the pump itself, and a two-man contractor team working on the PRV and its associated pipework. The PRV for pump A is not located immediately adjacent to the pump, and is above floor level, close to a number of other pipe runs. The following description represents a hypothetical sequence of events based on the inquiry findings, but embellished for the purposes of the case study. 

At shift handover at 18 00, the incoming operations supervisor was briefed by the day supervisor. The conversation centered on the vibration fault and subsequent repair work carried out. However, no mention was made of the work on the PRV, so consequently none of the incoming shift were aware of it. The night shift supervisor, wanting to return pump A to standby as soon as possible, asked the plant operator to check the status of the pump, and together with the shift electrician, to reset it and put it back on stand-by. The operator, unaware of the work being done on the PRV, did not check this part of the system and, following inspection of the pump, returned it to stand-by. 

The root causes for this critical event both concern the operations department and the service department who ran the contractor maintenance team. The operations department (i.e., the day shift operations supervisor) failed to provide adequate supervision and instructions to the contractor team. Explanations of the nature of the permit-to-work systems (i.e., the need to report back at end of shift) should have been given, and the possibility and implications of work not being completed before the end of the shift should have been considered by both parties. 

On the part of the contractor team, two root causes were identified, root cause 1 being insufficient training of the contractor team leader. He was uncertain of permit systems, specifically whether they should report in at the end of shift, and, if so, who should do it. The second root cause relates to the procedure used at the end of the shift for supervisors to sign back permits. Although according to procedure all workers should hand back permits to supervisors in person, in practice this did not occur. If no one is present, or they are busy, it had become common practice to either leave the permits on the supervisor s desk, or to sign them back in the morning. 

Fear of Fraud It is illuminating that there are companies that lock the input to a computer system (specifically a so-called LIMS, a Laboratory Information Management System) in such a manner that it takes the supervisor s password and signature to correct a missed keystroke, even if this is noticed before the (Enter) key is pressed. If the unwitnessed use of the (Backspace) key betrays malicious intent to falsify raw data, then every scientist should be in jail. 

---