http://homepages.which.net/~roger.hill/bougainville.htmlBougainvi...

http://homepages.which.net/~roger.hill/bougainville.htmlBougainville Copper Limited in Papua New Guinea has introduced a cheap but effective on-line computer control system for truck and shovel operations based on a mathematical model of the batch allocation of trucks to shovels. Introduction of the system centralised control of the mine operations and special emphasis was given to simple but effective communications. Two way radios are used in mobile equipment, while computer video displays connect mine operations to other sections. The system encompasses pre-shift equipment and operator allocation, operations during shift, and post-shift reconciled reporting.

Bougainville Copper commenced open pit production in April, 1972. At that time the shift report completed by the pit operations general foremen was an extreme summarisation of shift detail. In 1974, a mine "base" was constructed and placed in a location offering good visibility of the pit. Continuous manual recording from this vantage showed considerable errors. The shift report still did not contain all the information required by the foreman to effectively supervise the pit during his shift. By January 1977, the need for a logical and accurate reporting system had come under discussion. Truck dispatching as a method of obtaining accurate reports was considered, but rejected due to excessive cost.

Alternatives, from programmable calculators to the central computers already on site, were evaluated in April. In May 1977, the mine engineering and computer services staff began a joint project to supply pit operations with an on-line computer system. It was named the mine production (MP) system, and its primary justification was that of recording accuracy.

A total of 24 man-months was spent on system development, on-line testing, documentation, parallel running, and introducing the system into pit operations. Additional computer equipment required included six visual display units and two printers, at a cost of $166,450. This included increasing the speed of the central processors, which are used to run Bougainville Copper's commercial work. However, return on investment is currently estimated at $665,82O to $1,109,700 per year. The COBOL language was used to ensure continued maintenance of the programs.

After a foreman's training course lasting less than a week, the system commenced operation on afternoon shift, January, 1978.

Tool for the Foreman

At the beginning of each shift, the manual system required foremen to tell the drivers which truck to operate, from which shovel to load, and where to dump. This allocation of trucks to shovels in effect poses a queue theory problem. Graff [1] observed that although the truck rejoined its original queue, arrivals were random. When trucks and shovels break down, and as shovels move, the optimum allocation changes in order to maintain maximum production. Essentially the optimum allocation of trucks to shovels is a problem of judging a series of changing probabilities. Beer [2] argued that this problem was intrinsically difficult for people to solve. Furthermore, he stated that a model was an essential tool to control such a system.

It was recognised that the only way to guarantee the accuracy of the recorded information was to make it useful to the recorder. Therefore, the system was designed as a tool for the pit foremen to keep tight control of this section of the mine's operation. A suggested batch allocation of trucks to shovels was the model chosen. The method spreads the trucks across the mine based on an even match factor for each shovel.

Pre-production schedules were established by mine engineering. Pit operations then attempted to meet these schedules. The two sets of figures were compared in a post production reconciliation without the guarantee of a common data base. The MP system overcame the inconsistencies of these techniques by supplying pit operations with the mine engineering data in the form of the suggested batch allocation. The system then supplies mine engineering with reconciled reports automatically. It was designed so that everyone who supplied data received better results with less work on his part.

Other problems in the mine were lack of feed back, and communication difficulties between different branches of the organisation. The system was therefore designed around the operations room concept with distribution of information by computer.

The isolated location of the mine made systems reliability a major design criteria. Systems reliability can be split into four sections: users, operations, software, and hardware. By making the system useful to the foreman and simple to use, user reliability is ensured.

For computer operations, all programs were designed so that on failure (power cut, etc.), the programs are reloaded. They take their own error action and no switches or special operator action is required. In the computer operations area simplicity guarantees reliability.

For software, error restarts were designed into the system. They share common code with shift change procedures. The effect of this was:

To seriously delay testing, since testing could not start until recovery from errors worked
To ensure error recovery worked.
A transient hardware fault broke the system 35 times during the first day of live running, after which the system reloaded properly every time.

To maintain hardware reliability the computer services department supports two ICL 2904's with 48K of store each. Ease of switching between machines guarantees machine availability 24 hours a day, 365 days a year.

Widespread Use of Video

A purpose built Base overlooking the mine was a telephone, two radio systems, and two video terminals. One radio is for contacting individual trucks and the other for supervisory personnel. One of the videos gives details of shovel positions and truck allocation; this replaced the shift report. The other video lists drivers and details of the equipment they are working on.

The method of communication with the video is novel, simple, and highly efficient. ICL 7502 video has an optional badge reader, taking the form of a magnetically coded pen. By allocating a pen for each driver, truck, and shovel, any message is input by inserting the pen (item) into the badge reader and pressing one key (action). For example, truck 22 (item) is broken down (action). The control key is locked down on the video so that most operations consist of pressing one button. A mask is placed over the keyboards allowing access only to the required keys. The functions of the keys, such as break down, sick, etc. are inscribed on the mask.



Consider a message of the form: Driver Joe Blow (item) is driving (action) truck 22 (item). This message consists of two items and one action. The message is split into two parts. For example, driver (Joe Blow) (item) is put on a truck (action). The computer then writes to the video all the trucks without drivers. The tab key is pressed until the cursor is positioned over the required truck. The send key is then pressed. Similar choice actions are used to specify the dumps and benches for shovels, etc. As each change is made the computer produces on the video screen a new set of recommendations for both truck and shovel allocation based on the batch allocation model.

Besides updating the videos in Base, the computer also updates a video in the pit maintenance workshop so that foremen from that department can verify the list of equipment broken down. This additional display has prevented trucks being unaccounted for. Break down times are both accurate and agreed upon by the operations and maintenance departments.

Another video in the pit operations office is used at the start of shift enabling the on-coming shift to prepare its line up while the current shift is still running. A third end-of-shift video is used for punching the number of loads of rock moved per truck for haul records from Base during the shift. This then prints out an end-of-shift report (Figure 1).



The system has proved successful. The schedule is not followed blindly, but serves as a useful guide particularly in the aftermath of equipment break-downs. It tends to guide the foremen towards maximum efficiency as each problem arises. The display in Base gives foremen a much clearer idea of what is going on. The end of shift report tells the foremen both how much output he achieved and the equipment efficiency. Total loads no longer constitute the only performance criteria, and this direct feed back has been very well received. It allows the foreman to judge and evaluate the consequences of his decision. Shift change is faster, and there is less queuing and delay within the mine.

Pit operations information from MP Systems can be displayed on videos throughout the site. However, unauthorised access to the system is prevented.

Part of the MP system consists of data set up and report generation. Schedules are no longer solely the concern of the mining engineer. A common data set is supplied prior to commencement of shift, used by pit operations foremen during the shift, and finally used in report reconciliations. To complete the reporting function of the system only head grade for each shovel has to be supplied to the system.

Conclusions, Recommendations

Much has been written about Bougainville Copper's geographical isolation. This necessitated using standard hardware and software, and the COBOL language in order to guarantee continued support in a commercial services environment.
The ease of use and the value of information supplied resulted in excellent user acceptance.
The accuracy of a fully integrated system has saved engineers otherwise wasted schedule reconciliation time.
Management at all levels now considers on-line computer technology as appropriate to a mining environment.
The system does not replace supervisory personnel. Rather, as foremen use the system fully, it promotes quantitative communication of operating conditions.
Designing systems for use by people of a different culture emphasises the need to include them in the systems design.
The man who makes the most difference to operating efficiency is the foreman. He generates most of the useful data, and it is the feedback to him that is most effective. The key position of the foreman must be the starting point of a production system. The parameter the foreman is most interested in is his productivity, defined as actuality divided by capability, Beer[2].
The authors wish to acknowledge Bougainville Copper Limited's permission to publish this paper. Dr. P. K. Chatterjee is thanked for his comments on an early draft.

By J. Hempenstall and R. Hill

References

[1] Graff, G. 1971 Simple Queuing theory saves unnecessary equipment, Industrial Engineering (Feb 1971).

[2] Beer S. 1975 Platform for Change (John Wiley: London).

Roger Hill's Published Papers