Introduction
Smartpour is a non-contact infrared temperature monitoring system for molten metal streams. The detector is a ratio or 2 colour device, powered from an electronics unit which also enhances the noisy signal from the head and computes a representative peak temperature for the pour. For long pour periods, a temperature is computed every 20 seconds. The system was developed to overcome a problem at Hepworth Heating Ltd, Milford, Derbyshire on the company's Kunkel-Wagner moulding line. Hepworth required more temperature information between its melting furnaces and casting line. The company already took regular dip temperatures but was unable to increase the number of dips sufficiently, due to time limitations. Separate infrared systems, on each of the four furnaces, was too costly. A short trial on the pouring line proved successful. Target distance variations of 5 to 10 metres proved a challenge which was overcome largely by the skill and patience of Hepworth staff in aligning the head with the target. The first system was commissioned at Hepworth's in November 1992.

Problems dealt with
1. Transit damage assessment and rectification was the first priority.
2. Various appendages on the ladle framework had been removed since the site assessment survey some months previously, the pillar earmarked for the detector head had been cut off. It was decided, during consultation with their production management, to site the head in a completely different place, around the back of the ladle, away from the operator's working area, this necessitated a substantial extension to the ladle frame to provide a platform for the head.
3. Maximum ambient temperature specified for the head was 55oC. The actual ambient of the installed head was found to be 70oC. A water cooling jacket was flown to site. The head was fitted inside this jacket and an air line connected. This did not give sufficient cooling so a water supply had to be connected instead, giving the required temperature drop. Air was still connected, to feed the air purge unit on the front of the head. This minimised build-up of dirt on the lens. Radiation from the autopour and the moulding line plus conduction through the metal frame of the autopour both contribute to the high ambient temperature at the rear of the autopour. Matsushita's established routine for stream temperature checks is to use a portable Disappearing Filament Pyrometer (DFP) every hour. Temperature was measured with +/- 10oC for each good pour. During commissioning of the system a portable digital pyrometer with a short lightweight dip arm and plug-in type 'R' thermocouples was used in the metal stream to check the infrared temperatures obtained and to tune the infrared, using the emissivity, E slope potentiometer, at the rear of the head. The long reach element of the thermocouple enabled minimal disturbance of the molten stream. The limitations of accuracy of the DFP unit was demonstrated and it was eventually agreed that the infrared system to the existing DFP reading should be tuned. At a meeting in November 1997, a problem with the infrared was discussed. When using the 'inmould' system for nodular iron, a recent development, one was unable to measure stream temperatures into the mould, due to the flaring reaction. It was advised that it was not possible to measure temperature with an infrared system, under these conditions. The company was advised to move its point of measurement back to the metal stream into the tundish. The tundish is interposed between the autopour and the moulding line.

Problems dealt with
1. During commissioning the autopour operator found the apparently unstable reading on his indicator to be a distraction. He was asked to ignore the readout until the commissioning was finished. It was later explained that the program was going through an updating procedure, as it searched for a peak stable reading, which it finally held, until updated by a new reading.
2. Shortly after installation, before handover, it was reported that Potterton was obtaining apparent high readings. Burn-off flame, hot gasses or metal dribble were suspected. Visual observation dismissed the presence of any obtrusive target. Metal dribble caused flashing on the temperature readouts, but no final stable readings. Finally, for the record, photographs of the metal stream were taken, of each stage of the pour, including the push. The prints showed the presence of a flame from the pouring aperture, only during the push. This interference was eliminated by increasing the threshold temperature limit to 1350oC. On the first trial of the autopour it was decided to raise the threshold limit from 1200 to 1250oC to avoid interference from the burn-off flame, looking down the moulding line, away from the moulding machine.
3. A requirement for this installation was downloading of data to an existing computer for use with Lotus 123 spreadsheet software for MS Windows. The first download program for Potterton's computer matched its protocol requirements but some data was lost. Finally, after an inappropriate delay, a windows based program was supplied.
4. A variation in measured temperature with the infrared system had been noted, when changing from one type of iron to another type. To compensate for this change an adjustment is need to the emissivity slope potentiometer, mounted at the back of the head, near the eyepiece. Fortunately, this potentiometer is linear, and a pre-set position can be established for that iron type.