Instem: The Cable Section Monitoring Project
Instem Computer Systems, Walton Industrial Park, Stone, Staffordshire, UK
By 1990, the National Grid company of the UK had developed a computerised mathematical model of the thermal behaviour of it's electricity transmission lines. Instem was awarded a contract to provide a system to collect data from sections of transmission cables, feed the data into the mathematical model and provide graphical displays of the results. This was to be the second project I would work on at Instem.
The CSM project team consisted of members of the Littlebrook team plus some recent hires. As well as myself the team included Simon Broad, Chris Podmore, Steve Gough, John Gibbs, Iain Jones, Duncan Reade, Dale Kemp, Andy Wale, Mark Rudolph and Mike Kasarov.
Data Collection
My team, the data collection team, were responsible for collecting data from Instems' DMACS Outstations and populating the Commander system's realtime databaseI designed and implemented a layer of software which asynchronously polled upto thirty-two remotely located industrial computers for changes in analogue and digital inputs. The software populated the Instem Commander System's database with the values and alarm conditions received. The software was written in VAX Fortran and runs under VAX VMS. The software was written to replace an obsolete data concentrator normally produced by Instem for this job. Instead of this custom piece of hardware the VAX was linked by DECNet to a DEC terminal server. Each port of the DEC Server was connected by an RS-232 serial link to a modem and from there across leased telecom lines to the industrial computers sitting under electricity pylons somewhere in the UK.
Many of National Grid's modems were old and ran at 70 baud (yes, seventy!). This was a lower speed than Instem's DMAC's equipment had been run at for a long, long time. It was soon discovered that some of the timeout values in the DMAC's software were now too low and were causing communications dropouts. Instead of the DMAC's software providing a trusted baseline against which to test my new comms software, my new software became the test tool for Mike Kasarov as he tracked down each timeout problem and fixed it.
User Interface
The user interface team were responsible for porting the X-Windows user interface developed on an earlier National Grid Remote Substation Controller project to Commander.While I had been working on the Littlebrook project. Steve Gough had been leading a software team to develop a UNIX based system with an X-Windows user interface to Remotely Control electricity Substations for National Grid. Steve's foresight and vision for this project was a dramatic step forward for Instem in the user interface area. It meant that Instem would in the future benefit from standard display components and no longer be reliant on specialised high resolution terminals. For the CSM project we took the RSC system and instead of connecting it to a single modem, connected it via Ethernet to a pair of Digital MicroVAX computers running Instem's Commander system. It was here I learnt my first lesson in User Interface/Business Logic separation the hard way. Much of the old Commander VDU code had embedded business logic and separating it out to feed the Unix workstations instead was a tedious process.
 |
The main characteristics of the RSC user interface were a small three
or four line window at the top of the screen showing the last three events
that had occurred, the large diagram area with dynamically updating graphics
and text and the row of graphic buttons used to navigate the system at the
bottom of the screen. These buttons were mapped to the function keys of
a standard PC keyboard to avoid the necessity of using a mouse or other
pointing device for day to day tasks. Unlike the eight-colours of the Littlebrook project VDU screens, the Unix Workstations boasted sixteen colours and we could benefit from using a light-grey background instead of the black background used at Littlebrook that made thin blue lines in graphs hard to see. In general, the colour coding used on the two projects was consistent with measurements that rose above set levels first shown in yellow and then in red and measurements that sank below set levels being shown in cyan and then dark blue. Magenta was used to show unknown or failed inputs and switch settings were shown in green if they were in their safe settings and red if they were in their dangerous (on) setting. The CSM project used its additional colours to add dark green to show values that had been overridden by an operator and could use a different shade of grey to 'grey' out unused values. |
 |
Commander and the Math Model
The mathematical model team were responsible for integrating the mathematical model software developed by National Grid with the Instem Commander system and enhancing the Commander system to provide a different style of event and alarm reporting.Being half a mathematics graduate, I would have been interested in the details of the mathematical model but I was thankful in the end that John Gibbs took on this area. John made the model his own. He had to struggle with poorly documented FORTRAN code littered with single letter variable names and remarkably few comments. To John's credit he subdued the beast and the results were very impressive.