Polfin, a subsidiary company of Sasol Chemical Industries and a leading manufacturer of low-density polyethylene plant. Production on this extremely high-pressure plant has been increased by approximately 50%. This was achieved without increasing the size of major vessels.
Problem identification
In particular, the higher flow velocities through the separators had resulted in these vessels becoming highly susceptible to wild level fluctuations. These separator vessels are used to disengage or separate unconverted Ethylene gas from the molten liquid polyethylene product. They operate at a pressure of around 300 bar. The previous capacitance probe-based polyethylene level control systems worked in conjunction with radioisotope-based high level trips. These simply could not cope with the new demands being placed on them.
Since the upgrading, the vessels frequently suffered liquid polyethylene carry-over into the gas exit lines. This results in liquid solidifying in the gas lines. This overflow condition resulted in unacceptably high losses in production in the past year.
Solution adopted
Polifin's plant management decided to install a reliable interface measurement system: one that could be trusted to indicate the exact position of the interface between the relatively high-density gas and the relatively low-density liquid. The gas is ethylene at a density of about 0,2 g/cm3. The liquid is polyethylene at about 0,5 g/cm3.
After a thorough investigation (including overseas visits) and detailed evaluation, Polifin's plant management selected the Tracerco interface gauges.
The reasons for the radioisotope-based Tracerco gauges being selected were as follows:
p Safety: Disastrous fires had occurred on the plant in the past. Tracerco supplies the only non-invasive level or interface gauge that is intrinsically safe for zones 0, 1 and 2 use. It is therefore guaranteed not to initiate any fires or cause the highly flammable ethylene gas to explode.
p System reliability: High system reliability had to be assured. The Tracerco gauges have an independently assessed Mean Time Between Failures of 14 years.
p Accuracy and speed: Due to the random nature of radioactive decay, long time constants are desirable for high accuracy when using nucleonic systems. However, the system had to be both accurate and fast. Fast response requires a short time constant. The Tracerco gauges have dual dynamically selectable time constants.
The Tracerco control units automatically choose a long time constant under steady state conditions and a short time constant under fluctuating conditions. This ensures no compromise between the two key performance areas of accuracy and response time.
p Pressure effects: The Tracerco systems have pressure compensation built in. The density of the ethylene gas can fluctuate dramatically due to system pressure swings. A sudden drop in pressure could cause dissolved gas in the melt to begin escaping. This causes a foam to begin to rise from the level of the melt. However, the immediate effect is for more radiation to pass through the lower density gas. This results in the anomalous situation that competing nucleonic gauges would initially indicate the level moving in the opposite direction ie falling as a result of increased radiation at the detector. Competing systems would therefore begin to drive control systems in the wrong direction. This sort of foaming-problem would be better dealt with by using the Tracerco high-resolution density profiler. It was, however, considered unnecessary in this particular application. This decision was based on experience gained in solving identical problems at other global sites: ICI's Botany Bay site in Sydney, Australia; at ICI's Wilton site in the U.K.; at Repsol in Spain and at a site in Texas, US.
p Accuracy of the measurements: Tracerco level or interface gauge detectors use Geiger Muller tube technology. These do not operate too close to background radiation levels as do some competing scintillator detectors. They are therefore less susceptible to minor disturbances to the radiation field due to eg process inflows. They will allow for a build-up of a polyethylene 'skin' on the walls to be 'calibrated out'.
p System design: Each interface gauge had the same design, which was as follows:
Four radioactive sources were strung together using a stainless steel cable. As with most interfaces gauges, these were inserted into a pocket or well or dip-pipe that had been inserted into the bottom of the vessel. The reasons for this are twofold:
1. It avoids the need to radiate through two thick vessel walls and through two steam jacket walls.
2. It ensures that the radiation actually passes through even the densest phase and arrives at the detector. This is different to a normal level system where the bulk process contents of a full vessel will usually cut the radiation at the detector to background levels.
On a simple application such as indicating the interface level on an oil/water separator one would firstly obtain purely oil between the sources and the detector. The pulses received by the control unit under such a condition would be made to represent 4 mA. The next step would be to ensure that there was only water between the sources and the detector. The input to the control unit under this condition is set to represent 20 mA.
The control unit will then give a 4-20 mA output that represents exactly where the level of the interface between the oil and the water is. Polifin's plant management however, did not want the vessels filled to the top of the measuring range with polyethylene. The management team was very concerned about the possibility of carry-over or overflows. Therefore, not only system design but also commissioning had to be completed partly by calculation.
Radioisotope selection
A competing supplier who had previously installed the nuclear high-level switch had opted for Cobalt 60 as its source of radiation. This is presumably part of the reason why the switch had failed to protect the vessel against high-level excursions.
The Tracerco team selected Caesium 137. The Cs-137 has a much longer half-life at 30,0 years than Co-60 at 5,26 years (refer BS 4094: Part 1 of 1966). This was not the most significant reason for choosing the Cs-137 as the Tracerco control unit automatically compensates for source decay whether it is Cs-137 or Co-60 or indeed Americium 241. The most compelling reason for choosing the Cs-137 was its lower Gamma energy at 662 keV as opposed to the Co-60 Gamma energy being as high as 1330 keV.
This would make the system far more sensitive to the presence of the low-density gas or liquid. Whereas the Co-60 gamma ray would tend to 'blast' through the process medium with very little attenuation the Cs-137 gamma ray would fluctuate more readily with changes in the density of the process medium.
Design detail
Dip pipe:
Firstly, the dip pipe or well pocket had to be designed to withstand the 32 MPa (g) external vessel pressure. Two failure modes exist and both had to be checked. The dip pipe would either fail by 'Euler Crippling' ie buckling or by straightforward compressive failure.
The formula used to check the crippling mode is
formula 1
Where 't' is the wall thickness, 'E' is Young's Modulus, 'd' is the diameter and '' is the Poisson Ratio.
Compressive strength was checked by realising that the stress in the wall equals the compressive force in the wall divided by its cross-sectional area. The vessel pressure acting on the elevational area of the pipe generates the compressive force.
Formula 2
Source activity sizing
The required source activity was calculated by using the specific gamma ray constant for Cs-137 which is dose rate = 0,33 mR per hour/mCi at 1 metre.
Firstly, the inverse square law had to be applied to take into account the actual distance between the sources and the detector which was not exactly 1 metre. Next, the attenuation due to the process medium, steel capsule holder and dip pipe, vessel wall and jacket had to be taken into account. These calculations resulted in a dose rate per mCi at the detector due to a single source. The dose rate contribution from adjacent sources also had to be considered. Finally this figure was divided into the required dose rate for the detector to operate properly. This quotient gave the source size.
The dose rate had to comply with the Hazardous Substances Act. Tracerco equipment is designed to conform to UK legislation, which is among the strictest in the world. Occupational exposure would be relatively low as the vessel is mounted high above walkways and the general area is not frequented by personnel. The vessel could be made safe for maintenance by withdrawing the sources into an elongated lead-lined safety container.
Commissioning
The various methods of commissioning that ICI had used on similar plants around the world were reviewed in conjunction with Polifin plant management. The calculation method that most suited Polifin's situation was adopted.
The radiation arriving at the detector causes square wave (digital) pulses to be fed into the control unit. As the density of the product between the sources and the detector increases, the dose rate at the detector reduces. This results in a concomitant reduction in the pulse rate. The equation that governs this reduction is:
I = I00e-µrx
Where 'I' is the attenuated radiation due to the ethylene gas or polyethylene product. 'µ' is a gamma ray constant for Caesium. 'r' is the density of either the gas or the melt and 'x' is the path length of the product that the radiation travels through. I0 is the initial maximum pulse rate when the vessel is brought to operating temperature but kept empty and depressurised. The 'I' value for the ethylene gas was calculated. Kallie Holtzhausen the plant Instrumentation Specialist then arranged for the vessel to be charged with ethylene gas.
It was noticed that pulse rate fell to a similar level to that which had been calculated. The polyethylene melt pulse rate was also calculated and the two values were then entered into the control unit.
Operational experience
The plant has operated under various extreme conditions (such as maximum output and sudden shutdown) for a period of nine months with the Tracerco equipment. Not a single moment of production has been lost during this period due to separator vessel overflow. The investment in the Tracerco equipment has been paid back many times over already.
Conclusion
Interface level gauging can be reliably and accurately performed using appropriately designed radioisotope-based systems even when the application is particularly arduous.
Tracerco
(011) 706 2592
© Technews Publishing (Pty) Ltd | All Rights Reserved
printer friendly version