How lagging impacts on temperature measurement

Paul Boughton

Ali Niazi and Dr Sarah Kimpton examine the impact of thermal lagging on temperature measurement for fiscal/custody transfer metering systems.

Large custody transfer/fiscal metering systems in the United Kingdom are regularly audited to ensure compliance with best industry practice and British and international standards. One recurrent finding from these audits has concerned the lack of thermal lagging on both the upstream and downstream lengths.

Typical findings from these audits are:

- The orifice fittings, meter tube upstream and downstream straight lengths and temperature fittings are not thermally insulated and are open to the elements in an exposed location.

- In an exposed site location there may be a big difference between the flowing gas temperature and the ambient temperature depending on the time of the year.

- The exposed position of the meter run and the lack of lagging on the metering system means that the temperature used to calculate gas density is unlikely to reflect the true flowing gas temperature.

Metering standards recommend that the temperature sensor is placed in a thermowell which is then inserted into the gas stream of the flowing gas. The auditors claim without lagging, the temperature measured in a thermowell may not be the same as the gas temperature. A temperature difference of 0.5degC will impact on volume measurement by as much as 0.3 per cent.

For fiscal/custody transfer metering systems, the temperature is always measured downstream of the primary device, regardless of the type of the meter, unless it can be demonstrated that alternative techniques can provide similar performance. Fig. 1 shows a typical orifice plate metering system used for custody transfer/fiscal flow measurement.

The majority of natural-gas metering-systems in the United Kingdom were built in the 1970s and 1980s with no thermal insulation to protect the temperature measurement from ambient variations. Historical work carried out in this area demonstrated that the impact of the ambient temperature on the temperature of the flowing gas was so small that it could be neglected. Historically the belief was that if the meters were larger than 8-in (200 mm), then lagging was unnecessary.

Advantica was requested as an independent organisation to investigate the impact of lagging on temperature measurement by investigating the historical work carried out in this area and developing a computational fluid dynamics mathematical model to simulate the flow measurement for these systems.

The theoretical calculations of Pallan in the early 1970s indicated that the difference in the temperature of the pipe wall and the temperature of the gas were likely to be small even for unlagged pipe subjected to severe conditions. The largest source of temperature error was predicted to be from thermal conduction. Pallan's experimental measurements showed that an unlagged thermowell determined the gas temperature to within +/-1.7°C when the ambient temperature varied between 39degC below and 28degC above the gas temperature. The tests also demonstrated that the temperature readings were within +/-0.5degC when the gas temperature was between 39degC below and 11degC above ambient.

Theoretical calculations by Fenwick on pipe-surface temperatures show that for high gas pressures and velocities the difference between the gas temperature and the pipe surface temperature is less than 0.6degC. The wind speed, however, had a significant impact. Experimental measurements at two compressor stations indicated that the difference between the gas temperature measured in a thermowell and the pipe surface temperature was only 0.2degC. These measurements were made in an enclosed environment so the impact of wind speed was not an issue.

The experimental measurements by Ingram confirmed that the agreement between the gas temperature measured in a lagged thermowell and that measured with a lagged surface temperature sensor was within +/-0.1degC. Ingram also looked at the effect of removing insulation from surface temperature sensors - the temperature difference between the thermowell and the surface sensor with lagging was 0.17degC, When the lagging was removed, the temperature difference increased to 1.2degC.

Other points raised by Ingram were:

- The effectiveness of lagging is considerably reduced if it is not waterproofed.

- Surface-mounted temperature measurement is preferable to very short thermowells in which the sensing element is in the thermowell neck flange.

Nisbert and Robertson's measurements compared surface-mounted temperature measurements with temperature measurements in a thermowell at various depths. The maximum difference between the two sets of measurements was 0.15degC - the greatest difference was for the sensor located within the thermowell stabbing.

The computational fluid dynamics (CFD) study looked at temperature measurements in thermowells in 8-in and 24-in pipes transporting natural gas at 37.7degC and 15degC. Five different ambient conditions - 0degC, -5degC, -10degC, 30degC and 35degC - were considered. Under winter conditions, the simulations show that the temperatures around the top of the thermowell and the flange area were influenced by ambient temperature. The degree of influence increased with the decrease in ambient temperature. Typical winter conditions for a 24-in metering system is shown in Fig. 2. For summer conditions, the simulations show a considerable increase in temperature around the flange area and the top part of the thermowell. This increase is greater for the higher ambient temperature case. Further simulations were carried out with insulation around the thermowell and the pipe surfaces. It was clear that the thermowell and the flange were adequately protected by the insulation and only a small temperature change can be seen around the top of the thermowell and flange. The edge effects at the end of the insulation section did not penetrate into the insulated section.

A simulation with low velocity gas flow in an insulated 8-in pipe was also carried out. The results were similar to the other insulated cases. No significant drop in temperature in the thermowell is seen. However a drop in wall temperature at the end of the insulated section can be seen for the low velocity case, but it does not affect the thermowell temperature in any significant manner. Insulating a short section of the pipe would be adequate to obtain accurate measurements of the gas temperature.

In all cases the bottom of the thermowell was seen to be at the gas temperature and was not affected by the changes in ambient conditions. A temperature measurement taken at the bottom of the thermowell therefore represented the temperature of the gas inside the pipe.

The CFD modelling was carried out at the UK's Loughborough University and was based on the following conditions:

- Pipe sizes between 8-in and 24-in.

- Five ambient temperatures from -10 to 35degC.

- Gas temperatures of 15 and 37.7degC.

- Gas velocities down to 1.1m/s

It should be noted that outside these conditions thermal lagging around the thermowell only is advisable. Also, only one design of thermowell was used in the CFD modelling and other thermowell designs may have different thermal characteristics.

The experimental measurements and the CFD Modelling show remarkable similarities.

The CFD models all indicate that the temperature at the bottom of the thermowell is the same as the gas temperature both with and without insulation. The experimental measurements of Pallan confirm that the temperature in an unlagged thermowell was relatively insensitive to the most extreme ambient conditions.

The CFD model shows that the temperature within the thermowell stabbing is influenced by ambient conditions but the temperature of the gas stream within the pipe diameter is unaffected. This is confirmed by the experimental measurements of Nisbert and Robertson.

The CFD model shows that the gas velocity within the thermowell stabbing is very low. In support of this, Ingram's temperature measurements within the thermowell stabbing showed that there was a very long response time to temperature changes.

The CFD model shows that the temperature of the pipe wall is always very similar to the gas temperature. However, with insulation, the difference between the temperature of the pipe wall and the gas temperature is not detectable. Ingram confirmed that an insulated surface temperature measurement was within 0.1degC of the temperature in the thermowell.

The CFD model shows that local insulation around a thermowell or a surface-mounted sensor is sufficient. There is no need to insulate the entire meter run. Ingram also confirmed that local insulation is sufficient for accurate measurements of gas temperature.o

Enter 61 or XX at www.engineerlive.com/iog

Ali Niazi, Senior Consultant and Dr Sarah Kimpton, Consultant, are with Advantica, Loughborough, Leicestershire, UK. www.advanticagroup.com

Recent Issues