Hydrates inLPG CargoesA Technological Review

SIGTTO

HYDRATES INLPG CARGOES

A Technological Review of the Presence of Water inLPG Cargoes, the Consequent Formation of Hydrates

and Proposals for their Elimination

Prepared by Mr. R. C. Gray of the Technology Department of

British Shipbuilders

for

The Society of International Gas Tanker and Terminal Operators Ltd

Comments, experience and other data relevant to this Review should be addressed toThe Society of International Gas Tanker and Terminal Operators Ltd.,

London Liaison Office, Staple Hall, 87/90, Houndsditch, London, EC3A 7AXTele. 01-621-1422 Telex 894525 SIGTTO - G

First Edition 2008

Printed & bound in Great Britain by Bell & Bain Ltd. Glasgow

Published byWitherby Seamanship International

4 Dunlop Square, Deans EstateLivingston, EH54 8SB

United Kingdom

Tel No: +44(0)1506 463 227Fax No: +44(0)1506 468 99

Email: info@emailws.comwww.witherbyseamanship.com

Copyright of SIGTTO, Bermuda

ISBN 978 1 905331 277

Notice of Terms of UseWhile the information given has been gathered from what is believed to be the bestsources available and the deductions made and recommendations put forward areconsidered to be soundly based, the Review is intended purely as helpful guidance

and as a stimulation to the development of more data and experience on thesubject. No responsibility is accepted by the Society of International Gas Tankerand Terminal Operators Ltd or by any person, firm, corporation or organisationwho or which has been in any way concerned with the compilation, publication,supply or sale of this Review, for the accuracy of any information or soundness of

any advice given herein or for any omission herefrom or for any consequencewhatsoever resulting directly or indirectly from the adoption of the guidance

contained herein.

CONTENTS

Page No.

PREFACE V

EXECUTIVE SUMMARY VI

SECTION 1 - INTRODUCTION OF WATER INTO LPG 11.1 AT PRODUCTION PLANTS 31.2 DURING STORAGE 31.3 ON BOARD SHIP 3

1.3.1 Inerting/Ventilation 31.3.2 Moisture in rust on steel surfaces 31.3.3 Leakage in water cooled condensers 4

1.4 HYDROSTATIC TESTING 4

SECTION 2 - THE FORMATION OF HYDRATES 52.1 GENERAL 72.2 PROPANE HYDRATES 72.3 BUTANE HYDRATES 72.4 PROCESSES IN WHICH HYDRATE FORMATION CAN BE EXPECTED 72.5 MODIFICATION OF HYDRATE FORMATION BY METHANOL ADDITION 82.6 SOLUBILITY OF WATER IN LPG LIQUIDS AND THEIR VAPOURS 9

SECTION 3 - AVOIDANCE OF ICE OR HYDRATE FORMATION 113.1 GENERAL 133.2 REDUCTION IN WATER DEWPOINT ON SHIPS 133.3 SOLID DESICCANTS 133.4 HYDRATE FORMATION TEMPERATURE DEPRESSANTS 13

3.4.1 Commissioning Procedures 133.4.2 Reliquefaction Plants 143.4.3 Injection Quantities 143.4.3 Methanol contamination of LPG 14

3.5 HYDRATE CONTROL WITHOUT METHANOL INJECTION 16

3.5.1 Defreezing of LPG condensers 163.5.2 Promotion and removal of hydrates 16

SECTION 4 - SHIPBOARD TEST METHODS 174.1 SAMPLING METHODS 194.2 WATER CONTENT 19

4.2.1 Dewpoint Meters 194.2.2 Valve Freeze Test Method 19

4.3 METHAN L DETECTION 20

SECTION 5 - HAZARDS ASSOCIATED WITH ICE/HYDRATES 215.1 RELIQUEFACTION UNITS AND CARGO PIPING SYSTEMS 235.2 CARGO PUMPS 23

SECTION 6 - SUMMARY AND RECOMMENDATIONS 256.1 FULLY PRESSURISED LPG 276.2 FULLY AND SEMI-REFRIGERATED SHORE SYSTEMS 276.3 FORMATION OF HYDRATES ON SHIPS 276.4 SHIPBOARD COUNTER MEASURES 286.5 SHIPBOARD TEST METHODS 286.6 VALIDATION OF DATA ASSUMPTIONS 28

III

Page No.

REFERENCES

29

APPENDIX I - EXPLANATION OF THE PROPANE-WATER SYSTEM 31

1. Why LPG boil-off vapour contains more water than the liquid 33

2. The vapour, liquid and solid phases of the propane-water system during 35reliquefaction

2(i) Direct cascade reliquefaction 35

2(ii) Direct two stage reliquefaction 35

3. Formation of hydrates during vaporisation of propane liquid contaminated 35with water

4. Ethane, iso- and n-butane 36

APPENDIX II - LPG FREEZE VALVE 37

APPENDIX III - ESTIMATION OF METHANOL TO ADD TO PREVENT ICE/HYDRATE 45FORMATION

A. Methanol to add to a tank containing propane or n-butane saturated with 47water at 38°C to prevent ice/hydrate formation on cooling down tankcontents from SVP at 38°C to atmospheric pressure

B. Methanol to add to the reliquefaction condenser in propane and n-butane 50duty to prevent hydrate formation in the condenser or in the condensate

return

IV

A Technological Review Of The Presence Of WaterIn LPG Cargoes,

The Consequent Formation Of HydratesAnd Proposals For Their Elimination

PREFACE

The Society of International Gas Tanker and Terminal Operators Ltd (SIGTTO) is dedicated to the promotion of the mutualinterests of its Members in all matters relating to safety and reliability in the marine transportation and terminalling ofliquefied gases.

The Society, through its Members, was made aware of the occasional and seemingly unpredictable appearance of hydratesin the cargo systems of refrigerated LPG carriers resulting, at times, in damage to pumps and other plant or, at least, ininterference with expeditious cargo handling. While the production of LPG and its handling both on ship and shore includeprocedures aimed at minimising the water content of the product, there is little collated information which can be foundbearing specifically upon the formation of hydrates in commercially dry refrigerated LPG. The Society, on behalf of itsMembers, therefore required a review of the processes and procedures commonly practised in production, storage andtransportation of LPG with a view to identifying by what routes water can enter or remain in cargo and lead to hydrateformation.

This Review was conducted for SIGTTO by Mr. R.C. Gray of the Technology Department of British Shipbuilders and itsfindings together with recommendations aimed at eliminating hydrate formation are reported in the following pages.

The Society is indebted to British Shipbuilders and to Mr. R.C. Gray for the comprehensive technological nature of theReview which, it is believed, provides a unique collation of established data on the subject and some sound deductions andimaginative recommendations based upon these data. Produced primarily for the benefit of the SIGTTO Membership, theReview, because of its envisaged usefulness to the liquefied gas industry, is being made generally available.

In seeking to provide practical quantitative guidance in the control of hydrate formation in the low temperature processesof refrigerated LPG, it has been necessary to make various assumptions and extrapolations outside published experimentaldata which are presently largely only available for the ambient temperature range. Operational experience or experimentaldata relevant to these extrapolations or to any aspect of this subject would be welcomed by the Society with a view to up-dating or expanding the Review as may become desirable.

V

EXECUTIVE SUMMARY

LPG is peculiar in that, certainly at temperatures above 0°C, the boil-off vapour will contain a higher proportion of water byweight than the liquid with which it is in equilibrium. Semi-refrigerated propane at +6°C, for example, containing 40 ppmwater by weight in solution, will produce boil-off vapour containing 580 ppm water w/w. The solubility of water in theliquid phase of LPG decreases markedly with temperature.

All refrigerated LPG delivered by a terminal to a ship, while it may be free from ice or hydrates, will contain a small amountof water dissolved in the cargo. While the dissolved water content in the cargo liquid will be low (a few ppm at most), theboil-off vapour will contain a substantially higher proportion of ppm water vapour by weight than does the liquid. In viewof the very much higher vapour pressure of LPG compared with the vapour pressure of water, this may seem strange butindications are strongly that this is so. As the hot boil-off gases containing this higher water content are condensed, themajority of the water vapour in the cargo vapour will separate out as particles of solid hydrate. The LPG condensate at thetypical condenser temperatures and pressures will contain the remaining concentration of water from the boil-off vapour insolution. However, as the condensate is flashed off via the regulating valve to tank pressure, the resultant further cooling ofthe condensate will render it unable to hold this higher water concentration as dissolved water, free water will separate outand this free water will form hydrates.

Whilst there is evidence that fully refrigerated propane contains less than 5 ppm water in solution, little information isavailable concerning the water content in boil-off vapours at temperatures below 0°C. Despite this lack of supportive data,however, there is reason to believe that the boil-off vapour at these temperatures similarly will contain a higher ppm w/wconcentration than that in the liquid phase.

Propane may form hydrates at temperatures below +6°C and pressures up to 35 bar and commercial butanes attemperatures below + 3°C.

The data available on the water solubilities in liquid and vapour above 0°C lead to a suggested mechanism for theformation of hydrates in vapour recovery systems operating in this temperature range. Operational experience has shownhowever that such formations can and do occur in vapour recovery systems associated with fully refrigerated propane anddespite the present lack of water solubility data available at such temperature, it is assumed that a similar mechanism ofhydrate formation applies.

Section 1 of this Study describes how water may be introduced into LPG cargoes during production, storage andtransportation. Section 2 addresses the formation of hydrates. It describes how, once hydrate formation is initiated, thereis a significant "seeding" effect which can block valves and pipelines and gives detailed information on the water contentin LPG and vapour. Section 3 proposes how hydrate formation may be avoided or, at least, controlled. Section 4 outlinesvarious test methods for use on board ships. Section 5 describes the potential hazards to the cargo handling system of iceor hydrate formation. Section 6 summarises the results of this Review and outlines recommendations for further research.

V I

INTRODUCTION OF WATER INTO LPG

Section 1

1.1. At Production Plants

Most hydrocarbon gases become saturated with water at some stage in their production and must be dried. Williams andLom (Ref. 2, page 43) state that Molecular Sieve Absorption will typically reduce the water content of warm LPG to 10 ppmmaximum. This is a generally accepted specification for pressurised LPG marketed in even cold climates and avoids wateror ice deposit on ambient cooling of the liquid or on vapour expansion. Refrigerated LPG needs to be similarly dried butRef. 2 also points out that "even so refrigerated propane will invariably deposit some water as ice in tankage due to itsvery low solubility for water at the storage temperature (-35°C, 2 to 3 ppm)." Any lack of efficiency in the drying processbefore refrigeration or leakage in water cooled compressors or condensers would be responsible for a higher water contentand greater hydrate deposits.

LPG plants are normally designed to reduce water in process gas to a dewpoint somewhat lower than the lowest operatingtemperature in the process. Thus plants for LNG production, with LPG as a by-product, may dehydrate the LPG to less than1 ppm and avoid the hydrate deposit mentioned above.

1.2. During Storage

Some LPGs are stored underground in salt cavities under semi-pressurised conditions and may thus become fully saturatedwith brine at cavity storage temperature. Some LPG is stored in rock caverns either fully or semi-refrigerated. Thereforethese products may be contaminated with brine or fresh water.

1.3. On Board Ship

1.3.1. Inerting/Ventilation

Ships are vulnerable to water ingress whilst changing grades that require inerting and ventilation of cold cargo tanks withinert gas or air which may have a dewpoint above that of the tank temperature.

For example, inert gas with a dewpoint of +5°C contains .007 kg water vapour per cubic metre which, should it condenseon tank steelwork having an average temperature of -20°C, would deposit .0067 kg water vapour/cbm as a coating ofwhite frost. Thus a 10,000 cbm internally-stiffened ship's cargo tank could acquire a significant amount of water per tankvolume of inert gas.

Ventilation air too can introduce water into cargo tanks. Air at Bahrain, for example, typically has a dewpoint of 1 3°C inwinter rising to 20°C in summer. Such air in summer used to ventilate a tank with surfaces at, say, +5°C, has a potentialfor depositing condensation up to 0.01 kg of water per cubic metre of air introduced. Bearing in mind the size of tanks andthe number of air changes required, this could amount to a copious water input.

This is pointed out in the ICS Tanker Safety Guide (Liquefied Gas) (Ref. 3) in Section 1.41: "Reaction with Water - HydrateFormation".

1.3.2. Moisture in rust on steel surfaces

Rust surfaces of internal cargo tank structures, pipelines and process vessels will absorb moisture. Experiments have shown(Ref 4) that steelwork may absorb up to 30 gm/m 2 of water in its rust surface following water washing after ammonia ordue to condensation during inerting or ventilation. Thus a 10,000 cbm Type A cargo tank which typically has an internalsteelwork surface area of 10,000 m 2 may absorb up to 300 kg of water.

Tests on an LPG ship in 1966 indicated that the dewpoint of air in the cargo tank could be reduced reasonably quicklyto-18°C by circulation through and condensation in a refrigerated drier, but that the dewpoint continued at this temperaturethrough 2 days of further drying at a refrigerated drier temperature of -40°C (using propane) before further reduction. Anexplanation of this pause might be that moisture removal in the drier was at the same rate as moisture evaporation fromthe tank rust surfaces.

3

Section 1

Figure 1. Rate of Drying Out Water from Steel Rust Surfaces as a Function of Tank Temperature and Time.

The rate of drying out water from steel rust surfaces as a function of tank temperature against time is shown in Figure 1(taken from Ref. 4). Thus, if the cargo tank is water washed and free water is manually mopped-up, the steel surfaces willcontain about 30 gm/m 2 which may be reduced to about 1.15 gm/m 2 after ventilation for 12 hours at a tank steelworktemperature of +5°C. A 30,000 cbm ship with Type A cargo tanks would then have a total of about 35 kg water in thetank rust surfaces.

1.3.3. Leakage in water cooled condensers

Any seal failure in a water cooled condenser or compressor between the water side and the cargo side will result in waterleakage into the cargo side when the reliquefaction compressor is shut down.

1.4. Hydrostatic Testing

An occasional additional source of water into LPG may stem from the commissioning phase of new plants or start-up after maintenance periods/refits. Pipelines and equipment are often hydraulically tested during these periods andthen drained as far as practical. Pockets of water may, however, persist in pipeline dead legs, valve body cavities,corrugated expansion pieces, pumps, sumps, etc. Most LPG plants incorporate a refrigeration process, and the presenceof such water may manifest itself by hydrate or ice blockage which in some cases may prevent operation of the plantcompletely. A thorough flushing of the plant with methanol following a hydraulic test has been found to avoid suchproblems (see Section 3.4).

4

THE FORMATION OF HYDRATES

Section 2

2.1. General

Hydrates are a solid water lattice with hydrocarbon within the lattice, the bonding forces being of the loose physical,rather than firm chemical, type. If conditions of pressure and temperature are suitable and if enough water is present, theformation of hydrates is a continuing process which may lead to the plugging of valves, pipelines and regulators and todamage of cargo pumps. The pressure/temperature conditions necessary for the formation of hydrates can be most easilyillustrated by means of a diagram such as Figure 2 for propane contaminated with water.

Free water must be present in LPG vapour for the hydrate formation to take place with the vapour. Ref. 5 states that, "thecomposition of the hydrates is such that free water must be present for their formation from vapour because the vapourphase is simply incapable of supplying the amount of water vapour required at any localised point". This applies to theboil-off reliquefaction process. Similarly, LPG liquid must contain more than its saturation concentration of water beforehydrates can form in the liquid.

Once hydrate has begun to form, a seeding effect promotes rapid crystal growth.

2.2. Propane Hydrates

As will be seen from Figure 2, hydrates may form in propane vapour or liquid at temperatures below +6°C and at pressuresfrom the saturated vapour pressure of 5.61 bar absolute up to as high as 35 bar absolute.

As the pressure is reduced below 5.61 bar absolute:

Propane vapour and liquid water are found to form hydrates at pressures and temperatures to the left of the boundary lineB, C, C.

Propane liquid and liquid water are found to form hydrates at pressures and temperatures to the left of the saturated vapourpressure line for propane, line B B ' A.

2.3. Butane Hydrates

While hydrate formation is particularly relevant to the production, storage and transportation of propane, commercialbutane containing propane or iso-butane will also produce stable hydrates. A phase diagram, similar to Figure 2, may bedrawn for such butane. The diagram would have the general configuration of Figure 2 but with point B, at about +3°Cand 1.16 bar absolute.

2.4. Processes in Which Hydrate Formation can be Expected

Hydrates may form in a variety of handling and processing situations with propane where the necessary temperature/pressure conditions are encountered. For example:-

in propane vapour:

a. in condensation of propane and water vapour at a temperature between 0°C to +6°C so that free liquid water may bein contact with propane vapour at pressures over 1.62 to 5.61 bar absolute respectively;

b. systems where "wet" LPG is being used as a refrigerant gas at the hydrate forming conditions of pressure andtemperature;

In propane liquid:c. in cooling condensate to below +6°C and at pressures above the hydrate formation pressure at that temperature. (For

example, in inter-stage coolers or suction separators);d. in cooling condensate by flashing to tank storage conditions at a temperature below +6°C and with a significant

pressure drop in the condensate return line due to two phase flow of liquid and vapour;e. in taking LPG from deck storage tanks for spray cooling;f. in feeding LPG vaporisers with cool liquid in winter conditions where liquid temperatures before or after the inlet

regulating valve may be less than +6°C; andg. in the natural cooling in winter conditions of pressurised stored LPG.

7

Section 2

Figure 2. Diagram Showing the Onset of Hydrate Formation in Propane Vapour and in Propane Liquid Contaminatedwith Water.

2.5. Modification of Hydrate Formation by Methanol AdditionFigure 2 demonstrates how the addition of methanol to a propane/water system modifies the boundary line for the onsetof hydrates.

Figure 2 shows the pressure and temperature boundaries for the onset of hydrate formation in both propane vapour andin propane liquid and the effect of adding methanol to depress hydrate (and ice) formation.

The line A-A' is a plot of the boiling point of liquid propane contaminated with water. Above this line the propane is a liquidand below the line it is a vapour.

The line A-B1' -B1-B2 is an experimentally derived line showing the onset of hydrate formation in propane liquid. The shadedarea represents the pressure/temperature conditions under which hydrates may form. (See Appendix I for a prediction ofthe phases which may be present at a given pressure and temperature for a propane-water mixture.)

The line C-C, -B, is an experimentally derived line showing the onset of hydrate formation in propane gas when free wateris present. The broken shaded area represents the pressure/temperature conditions under which hydrates may form in thegas phase.

The line -B2 ' indicates the effects of the addition of an appropriate quantity of methanol in lowering the temperatureboundary limits under which hydrate (or ice) can occur.

8

Section 2

2.6. Solubility of Water in LPG Liquids and their VapoursEssential to the understanding of the formation of hydrates, particularly in respect of the availability of free water, is anawareness of the solubilities of water in liquid light hydrocarbons and in their equilibrium vapours at various temperatures.

It has not been possible to discover any published data on water solubility in hydrocarbons below 0°C. Figure 3 plots andextrapolates (dotted lines) and Table I tabulates data derived from References 1, 8, 9, 13 and 14 and the Figure and theTable together provide the best information found available for our present purpose.

Figure 3. Solubility of Water in LPG Liquid and in Equilibrium Boil-Off Vapour at Temperatures from –45°C to +45°C.

In general, the amount of water dissolved at saturation in light hydrocarbon liquids depends upon the hydrocarbon compositionand the temperature. Solubility reduces markedly with fall in temperature and at all temperatures the water vapour content inthe boil-off vapour is proportionately by weight in excess of that in the liquid from which the vapour derives.

9

Section 2

Table I — Solubility of Water in LPGs

Gas Temp °C Liquid Phase (Ref. 13)ppm w/w

Vapour Phaseppm w/w

Ratio V/L

Propane - 40 About 3 About 60-70 (App I) 20- 30 About 5 About 93 19- 15 About 13 About 260 20+ 6 44 (Ref 13) 580 (Ref 13) 13+15 75 870 12+ 30 173 1530 9+ 45 360 2300 6

Propylene - 45 About 20 About 26 (App I) 1.3- 30 About 40 About 76 1.9- 15 About 90 About 196 2.2

0 About 190 (Ref 14) 420 (Ref 14) 2.2+ 15 370 770 2.1+ 30 660 1300 2.0+ 45 1080 2100 1.9

N-Butane + 3 20 (Ref 8) 1720 86 (Ref 1)+15 44 3040 69+ 30 108 4640 43+ 45 238 6430 27

Iso Butane - 5 14 (Ref 8) About 930 (App I) About 2010 66+ 15 51 About 3040 39+ 30 120 About 4800 25+ 45 270 18

Butadiene 1.3 - 5 270 ± 50 (Private) About 1330 (App I) About 2800 4.9+ 15 560 (Ref 8) About 4250 5+ 30 820 About 6250 5.2+ 45 1020 6.1

Iso-Butene - 5+ 15 325 (Ref 8)+ 30 600+ 45 1080

This fact is most pertinent to icing and hydrate formation in LPG production, storage and transportation practice. The ratiobetween the ppm w/w water content of the vapour and that of the liquid is given in the right hand column of Table I for thevarious temperatures and hydrocarbons listed. In view of the great disparity in the boiling point temperature of LPG andwater, this fact may be surprising to some but it is well established and proven. A more detailed explanation of LPG-waterequilibria is given in Appendix I. Examination of these data and appreciation of the LPG-water equilibria mechanism leadto a number of useful conclusions in the consideration of practical situations as in the following examples:-

a. At the refinery run down to storage, the liquid temperature of LPG may be as high as 40° to 45°C, at which temperaturesthe solubility of water in propane, for instance, is over 300 ppm w/w. If the liquid subsequently stands and cools, thesolubility rapidly reduces, e.g. 75 ppm w/w at + 15°C or 44 ppm w/w at +6°C, and free water will separate out inthe bottom of the tank. If this water is not periodically run off, it may find its way into customers' deliveries. If standingtime is long enough in wintry conditions, hydrates may form (at less than +6°C) or the separated water may freezeleading to blockage of discharge piping (Ref. 2, pp. 57, 58).

b. Pressurised LPG passing through a pressure reducing valve will cool by flashing and, at temperatures below +6°C, mayresult in hydrate formation.

c. Semi-refrigerated propane at +6°C containing 44 ppm water by weight in solution will produce boil-off vapourcontaining 580 ppm w/w water. In reliquefying this vapour the excess ppm of water in the vapour over the water inthe final condensate will become available as free water and may form hydrates in the reliquefaction process.

d. There is evidence that fully-refrigerated propane at -40°C may contain about 3 ppm of dissolved water in solution.Although no published data are available, an examination of Table I and Figure 3 would appear to indicate thatboil-off vapour at this temperature will contain a higher water concentration than that in the liquid phase. The VapourPhase figures which are qualified by "about" have been calculated using the method described in Appendix I. By thesame mechanism as described in (c) above, this free water would be likely to form hydrates during the reliquefactionprocess. (Operational experience has shown that propane carried under fully refrigerated conditions can, and frequentlydoes, result in ice or hydrate deposits in the condenser and in condensate return systems).

10

AVOIDANCE OF ICE OR HYDRATE FORMATION

Section 3

3.1. General

Since the presence of water is a prerequisite for hydrate formation, its removal is clearly desirable. Provided sufficient wateris removed to bring the gas to its dewpoint with respect to water to below the lowest temperature encountered in the system,total removal is not necessary.

This drying process may be achieved in the production plant by using a liquid desiccant (di or triethylene glycol), a soliddesiccant (Molecular Sieve/silica/alumina), by expansion refrigeration or by the promotion of hydrate formation andits subsequent removal using suitable filters. Additionally ice or hydrate formation may be inhibited by the injection oftemperature depressants (methanol, ethanol, iso-propyl alcohol, glycol, etc).

LPG ships typically use solid desiccants, expansion refrigeration and methanol injection techniques and, occasionally, thedeliberate promotion of hydrate formation, its collection and removal, may be practised.

3.2. Reduction in Water Dewpoint on Ships

LPG ships are not usually able to maintain the water dewpoint of gases in the cargo tanks below the lowest tank temperaturebecause the water absorbed in the rust surfaces of carbon steel tanks is only reduced very slowly at typical operatingtemperatures (see Section 1.3.2 and Figure 1). The production of large quantities of inert gas from an oil-fired generator isalso costly if a dewpoint of –50°C is required. Many LPG carriers use a +5°C to –18°C dewpoint inert gas and purge outthe water vapour with dry propane or butane vapour.

In gas purging it must be remembered that gas from the top of an LPG shore storage tank will contain more water ppm thanwill the liquid, whereas if a purge vapour is obtained by passing liquid LPG through a vaporiser then the water content ofthe vapour and liquid will be the same.

3.3. Solid DesiccantsThese are typically used in oil-fired inert gas generators to achieve a dewpoint of –40°C for inerting interbarrier holdspaces for the carriage of fully refrigerated propane at –40°C to ensure that water vapour does not condense out on thecargo tank external steel surfaces and damage the tank insulation on warm up. They may also be used for drying inert gasproduced from a refrigerated drier outlet at about +2°C to a dewpoint of about –18°C for the inerting of cargo tanks.

Solid desiccants are not normally used aboard ships for drying cargo liquid or vapour. Unsaturated gases such as propyleneare, in fact, polymerised by some desiccants which contain a large proportion of silica although suitable molecular sievesare available which avoid this problem.

3.4. Hydrate Formation Temperature Depressants

This is the normally adopted method of hydrate control on board ships. The most widely used depressant is methanol (seeSection 3.4.3).

3.4.1. Commissioning procedures

Methanol may be used to inhibit water left in the cargo handling system after hydraulic testing and repair periods or thecondensation after inerting or ventilating with a water dewpoint higher than the cargo containment system.

While cargo systems are normally designed with draining facilities in pipeline and process vessel low points, water willnonetheless accumulate in the body cavity of ball valves, in the corrugations of horizontally mounted expansion bellowsand in pump shaft bearings, pump sumps, etc.

Flushing of these parts with methanol, which is then removed via the draining facilities, will minimise the possibility of iceor hydrate formation due to residual water. Methanol is flammable and should be used after inerting. (See Ref. 12 forprevention of hydrates in pipelines).

13

Section 3

3.4.2. Reliquef action plants

Injection of methanol into the hot gas stream entering the condenser can prevent hydrate formation if it is in sufficientquantity and is fully mixed with the gas. A typical "cascade" reliquefaction system in which propane is condensed onevaporation of R22 at –15°C would require methanol to be sprayed into the hot gas stream entering the condenser toprevent hydrate formation in the condenser. At this temperature about 13 ppm of the, perhaps, 60 ppm water which mayinitially be present in the boil-off gas will remain in solution in the propane condensate. Thus some 45-50 ppm free watermay be made available for hydrate formation within the liquid condensate droplets unless inhibited from doing so. In atypical "direct" reliquefaction system, however; where the condenser is operated with all the condensing surface above+6°C, hydrate formation cannot occur in the condenser and methanol may usefully be injected into the condensate outletline upstream of any subcooling of the condensate below +6°C or the level regulating valve. Methanol injection is onlyrequired if the condensate temperature is likely to fall below +6°C before re-entering the bulk liquid in the tank. There is,however, a danger that if condensate is sprayed into the compressor inter-stage cooler the methanol may contaminatethe lubricating oil via the high pressure oil separator and might eventually so impair the lubricating process as to lead tocompressor seizure.

3.4.3. Injection quantities

Temperature depressants are commonly injected on board ship into reliquefaction units, condensate return line filtersand cargo pumps. Ref. 7 describes experimental work to determine the quantity of alcohols necessary to prevent hydrateformation in propane and n-butane liquid, assumed initially water saturated at +38°C, for different temperatures downto –45°C. From these experiments it was concluded that methanol was preferable to ethanol since LPG absorbed a muchhigher proportion of ethanol (6 times more for propane, 12 times more for butane) (Ref. 7) and thus required substantiallymore ethanol to depress the freezing point of the water which separated out of the LPG as it was cooled. As it is, up to 98%of the methanol goes into solution in the propane liquid phase because, whilst methanol is almost "immiscible" in LPG, thereis a relatively large quantity of LPG in comparison with the small amount of water.

The rate of methanol injection required to prevent hydrate formation in reliquefaction plants is dependent on the amount ofwater present and the type of reliquefaction plant in use. The Valve Freeze Method, for testing whether the LPG will causevapour regulating valve freeze-ups, should indicate if hydrate formation is likely. It may also indicate whether sufficientmethanol has been added to prevent such freeze-ups. (See Section 4.2.2.).

Figure 4 indicates the quantity of methanol required to be added to LPG in a cargo tank to prevent ice/hydrate formation ifthe LPG is to be cooled in the tank and it is not possible to drain off any free water that may "drop out" during the coolingprocess. Such would be the case, for example, when LPG is loaded into a ship under semi-pressurised conditions and isto be discharged fully refrigerated. Figure 5 indicates the quantity of methanol to be added to the hot gases entering thecargo condenser of a cascade R22 refrigeration plant or to the condensate return of a two-stage sea water cooled plant.The method and calculation principles from which Figures 4 and 5 are derived are as given in Appendix III. The reductionin guideline methanol addition for butanes and propane (below –15°C in Figure 4) require to be validated in practice.

When considering the lowest condensate temperature, it is necessary to take into account the substantially higher proportionof the lighter hydrocarbon gases in boil-off vapour compared to the liquid mixture analysis. This will reduce the condensatetemperature to below the liquid temperature in the cargo tank.

3.4.4. Methanol contamination of LPG

Some LPG cargo quality specifications limit the maximum methanol content. For example, propylene feedstock to be usedfor the manufacture of polypropylene is limited to a maximum methanol content of 5 ppm and for commercial butanefeedstock, to be isomerised to iso-butane for the manufacture of gasolene blend stocks, the limitation is 150 ppm maximum.LPG dosed with methanol and stored over water in underground caverns will contaminate the water with methanol andmay cause water pollution problems. As discussed in Section 1.4, methanol may also be used during commissioning orfollowing maintenance refits of plant equipment. These considerations may limit the amount of methanol which may beadded during shipping. (Alternative means of controlling hydrate formation without the addition of methanol are describedin Section 3.5.).

14

Section 3

Cargo pumps, particularly the submerged electric motor type, in some LPG ships are dosed routinely with methanol inrefrigerated propane service with, say, 10 gallons prior to tank cooldown (to inhibit any moisture in the cargo pump fromfreezing), 5 gallons during the loaded voyage and a further 10 gallons the day before cargo discharge. Deepwell pumps,however, may be rotated by hand to ensure that they are free before start-up and methanol dosing undertaken only whennecessity demands.

Figure 5. Methanol to Add to Prevent Formation of Hydrates in Reliquefaction Condensers and Condensate Return Systems.

Figure 4. Methanol Required to Prevent Hydrates in LPG in Cargo Tanks During Cooling Cargo Liquid.

15

Section 3

Provided that the quantity of methanol injected can be carefully measured, the added methanol contamination of thecargo may be simply calculated. For example, the addition of 25 UK gallons of methanol, density 796.5 kg/cbm at+ 15°C, per cargo pump into 30,000 cbm fully refrigerated propane cargo in a ship with 8 cargo pumps would give acontamination of

8 x 25 x 796.5 x 1,000,000 – 41.6 ppm methanol by weight220 x 30,000 x 580

The addition of 1800 ppm by weight of methanol dosed to a propane reliquefaction plant handling 750 cbm/hr of ' vapour(density 2.4 kg/cbm) and operating continuously for 15 days on 30,000 cbm of fully refrigerated propane cargo wouldresult in added contamination of

1800 x 750 x 2.4 x 24 x 15 x 1,000,000– 67.0 ppm methanol, by weight

1,000,000 x 30,000 x 580

3.5. Hydrate Control Without Methanol Injection3,5.1. Defreezing of LPG condensers

Section 3.4.2 describes how excess water in vapour boil-off gas must come out of the propane condensate at –15°C asparticles of solid hydrate. For example, if boil-off gas from propane at –42°C and containing, say, 60 ppm water formscondensate containing about 13 ppm water at –1 5°C (see Table I), then the rate of hydrate formation in a condenserhandling 750 cbm per hour of vapour of density 2.4 kg/cbm would be

750 x 2.4 x (60-13) = 0.0846 kg per hour or 2.03 kg/day1,000,000

An alternative means of control, instead of methanol injection, could be a de-freezing operation of the reliquefactionsystem on a regular basis dependent upon the rate of hydrate accumulation. This could involve shutting down the R22 plantmaintaining LPG flow if possible until the condenser temperature had risen to 0°C, at which point the condensate returnvalve would be closed and the LPG compressor shut down. During subsequent hours the condenser temperature would riseto above +6°C and the hydrates would become readily-drained free water. Cargo Officers report that the condensers may"ice" up regularly on a fully-refrigerated propane cargo, indicated by the rise in condenser pressure with a reduction in unitefficiency perhaps due to particles of hydrate accumulating on condenser surfaces protected from hot gas impingement.They also report regular blockage of the condensate return regulating valve which manifests itself by an immediate rise incondensate level in the condenser and subsequent rise in condenser pressure.

33.2. Promotion and removal of hydrates

Ref. 10 states: "Fine mesh filters (say ASTM Sieve No. 100 - aperture 0.0059 inches, wire diameter 0.0040 inches) forma natural surface on which hydrate/ice crystals grow and, after sufficient build up within the pressure drop of the filterbasket, they can be removed."

Two such filters could be installed downstream of the condensate return regulating valves with a further filter (Sieve No.30 aperture 0.0232 inches, wire diameter 0.0130) at the condensate return dome connection as a final protection oftank sprays, cargo pumps etc. About 10 ppm (or 0.43 kg/day) of free water is estimated to form hydrates downstreamof the regulating valve for the example 30,000 cbm propane cargo.

16

SHIPBOARD TEST METHODS

Section 4

4.1. Sampling Methods

To obtain a sample of vapour from the vapour space of a fully refrigerated cargo tank presents few problems and theliquefied gas tanker normally carries instruments for vapour analysis such as dewpoint meters, explosimeters, chemicalsampling tubes and oxygen meters.

To obtain a representative liquid sample, however, is more difficult and it may be necessary to start a cargo pump in orderto tap-off a sample from a deck connection. Unless good pump recirculation is achieved prior to taking the sample, it maycontain contaminants in proportions unrepresentative of the bulk liquid.

Gas tankers do not normally carry liquid sample analysis instrumentation although some are provided with sample "bombs"so that a liquid sample, as captured, may be taken ashore for analysis. Ref. 2, page 120, stresses the value of liquidsampling in the assessment of cargo quality. Thus:

"Most analytical methods for LPG are based on gaseous samples, although the bulk of industrial users are more concernedabout the quality of the liquid LPG which is supplied to them. Therefore, the authors consider it essential to carry outevery analysis on a totally vaporised, liquid sample. This ensures that no high boiling components of the liquid are leftunaccounted for. It also prevents any bias due to preferential concentration of volatile impurities in the vapour standingabove the liquid ...".

However, most sampling methods are unlikely to pick up any solids which may be present in the bottom of shore or shipscargo tanks and which may be transferred during loading or discharging such as, for instance, ice or hydrate crystals.

4.2. Water Content

4.2.1. Dewpoint meters

Some LPG tankers carry an intrinsically safe, battery driven, portable dewpoint tester such as the Shaw Moisture MetersLtd, Model SADP with a dewpoint measuring range of –60°C to + 10°C. This meter may be used for water dewpoint of air,inert gas or LPG vapour. Dewpoint meters with a wider range, appropriate for colder cargoes, also are available. Thesewider range meters generally are mains driven and must be mounted in a safe area with an intrinsically safe extension tothe sampling point.

At levels of only a few ppm of water, sample piping impervious to water vapour must be used, eg. PTFE, copper or stainlesssteel. Rubber or plastic pipes should never be used. Their porosity to water vapour is likely to lead to misleading results.Such meters require careful, on-site calibration.

4.2.2. Valve freeze test method

The dewpoint test will indicate the quantity of water but will not indicate directly whether there is a risk of freeze-up. Such atest does exist and is described in the American Society for Testing and Materials (ASTM) D2713-81 "Standard Test Methodfor Dryness of Propane (Valve Freeze Method)". Details of the Freeze Valve considered , as suitable by ASTM for this TestMethod are attached as Appendix II. The author is indebted to Seismograph Service Corporation for their kind permissionto reproduce this information.

The method is used extensively on shore in the field testing for moisture content of pressurised LPG. There seems no reasonwhy the method should not be a useful means on shipboard to detect whether the moisture content of condensate was likelyto lead to freeze-up of the condensate return regulating valve and also to check the adequacy of methanol injection.

The liquid to be tested is first flowed through the fully open Test Valve to cool the valve body. Once so cooled, the Test Valve ispartially closed to a pre-set opening and the time for the valve to freeze-up and interrupt the flow is recorded. The averageobserved time for several successive and consistent tests is taken as the observed freeze time.

In shore usage on pressurised propane, an observed freeze time of more than 3 minutes or more is generally accepted asindication that freeze-up of system regulating valves will not occur. However, since moisture levels resulting in regulatingvalve freeze-up will vary with the operating conditions (liquid pressure and temperature) and on the size of the regulatingvalve, observed freeze-up time as an indication of potential freeze-up in the shipboard reliquefaction condensate situationwill be a matter for individual experience. The method may be used when the test liquid is dosed with hydrate temperature

19

Section 4

depressant. Thus, once a pass level of observed freeze time has been established, the test method may be found also toprovide an indicator of adequate methanol injection. However, the optimum procedure for the Valve Freeze Test Method inthe shipboard situation needs to be developed and its usefulness as proposed above requires validating.

4.3. Methanol Detection

The addition of methanol is a quite normal procedure in LPG production plants if a flow restriction is attributed to a freeze-up or on commissioning or recommissioning of plant. Thus, LPG in the loading terminal storage may already containmethanol in the 0-100 ppm range or even higher following hydraulic testing or a serious plant blockage.

Methanol in LPG can be detected in the laboratory by Gas Chromatography but only with difficulty. The Draeger MethanolTube will not read in a background of hydrocarbon gas. No suitable test method for methanol content of LPG appearspresently to be available for use on board ship.

20

HAZARDS ASSOCIATED WITH ICE/HYDRATES

Section 5

5.1. Reliquefaction Units and Cargo Piping Systems

There is a risk that condenser outlet regulating valves, condensate return lines/filters, cargo tank sprays and cargo tankpuddle heating coils may become choked by ice or hydrate crystals as a result of the free water remaining on commissioningor the likely water content of LPG boil-off vapours and the drop in temperature consequent upon the reduction in pressurethrough the associated control valve, spray orifice or heating coil. Condensation may also occur when there is a relativelylow weather-deck temperature compared to the stored bulk liquid temperature.

Hydrate temperature depressants, if injected into reliquefaction plant condensate, may contaminate lubricating oil if thecondensate is sprayed into the compressor inter-stage cooler and contaminated oil is returned to the compressor crankcasefrom the high pressure oil separator.

5.2. Cargo Pumps

Pumps handling LPG, irrespective of whether they are of the submerged or deepwell type, are especially vulnerable tofailure of product flow because starvation of flow through shaft bearings and impeller/casing wear rings will result inoverheating, vaporisation of LPG coolant, and possible seizure. Such overheating may be caused by friction from particlesof ice/hydrate scoring the running surfaces and accumulation of ice/hydrate could block the product flow.

23

SUMMARY AND RECOMMENDATIONS

Section 6

6.1. Fully Pressurised LPGPressurised LPG systems operating at above +6°C for propane and +3°C for butane present no hydrate problem inproduction, storage or transportation. Provided the LPG is dried in production to the commonly accepted level of 10 ppmw/w, there should be no water drop out in storage. However, storage in underground caverns/salt domes may lead towater saturation of the LPG at the storage temperature.

Problems may be experienced by the users of such LPG due to the higher water content of boil-off vapour and therefrigeration effect on pressure reduction resulting in hydrate formation and possible blockage of regulating valves anddownstream lines. The addition of 1950 ppm w/w methanol to propane when used as fuel is a recognised method ofavoidance. (Sections 2 and 3 and Appendix III).

6.2. Fully and Semi-Refrigerated Shore SystemsWith propane in such systems operating below +6°C, there is always a potential risk of hydrate formation from waterdropping out of solution as the propane is cooled. This is particularly so if the production process does not sufficiently drythe propane gas or if a failure of the drying system occurs. Whilst most producers dry their gas to less than the saturationconcentration at the minimum operating temperatures in their plant, this is difficult to achieve for the extremely low saturationconcentration of about 3 ppm w/w of water in fully refrigerated propane. (Sections 2 and 3).

Reliquefaction of the water-rich boil-off vapour in shore vapour recovery plants is bound to produce hydrates in thecondensate return unless the storage tank contents or the condensate are dosed adequately with methanol or the hydratesolids are filtered out of the condensate before its return. (Section 3).

The standard methods of sampling and measurement of propane dryness are unlikely to detect other than dissolved watersince ice/hydrate solids would not normally enter the sampling system. (Section 4.1).

Thus there are various avenues by which ice and hydrates could, on occasions, find their way into the loading pipeline withrefrigerated LPG. Final filtration ashore, therefore, appears to be a responsible practice to guard against inadvertentdelivery of ice and hydrate solids to ships. The cost of such filters should not be prohibitive and their presence would be anassurance to ships that any solids subsequently found aboard were unlikely to have come from the shore. (Section 3.5.2.).

6.3. Formation of Hydrates on ShipsDue to the LPG-water equilibria and solubility characteristics whereby the water content of boil-off vapour is likely to bemuch higher than the saturation solubility of water in the parent liquid, ships can, and undoubtedly will, produce hydratesin their reliquefaction plants. This reduces plant efficiency and may temporarily block condensate return systems. Solidsproduction is most noticeable in the cooldown, in loading and in the early stages of the voyage. Thereafter, the rate ofproduction generally has been found to decrease and this appears most probably to be due to the consequent gradualdehydration of the cargo liquid. (Section 2.6.).

Rusted surfaces of cargo tank steel are also a significant potential source of water vapour which may be taken up by theboil-off vapour especially in the ballast voyage and add to the hydrate problem in reliquefaction. During water washing, upto a tonne of water can be absorbed into the tank surfaces of a typical 30,000 m 3 Type A LPG carrier. This water quantitymay be rapidly reduced by thorough ventilation over a 6 to 12 hour period to leave a total residual water content of,perhaps, 35 kg which will evaporate at a decreasing rate into the tank vapour space depending upon the steel temperatureand the amount of steel surface above the liquid level. (Section 1.3.2.).

A further source of water will be from the inert gas and propane purging operations. If a 95% purge of inert gas bypropane gas is achieved, the 5% inert gas remaining will contain 10.5 kg water vapour (1.3.1). Boil-off vapour from fullyrefrigerated propane storage containing around 60 ppm by weight of water vapour used to purge the inert gas and top upduring tank cooldown will add another 1.7 kg water.

27

Section 6

Finally, there is likely to be up to 35 kg of water in solution in the cargo received. Thus, a well operated 30,000 m 3 shipmay have initially up to 82 kg total water in its cargo system. However, with normal operation it is unlikely that more thanperhaps half this total will be involved in the production of hydrates during cooldown, loading and the delivery voyage.(Section 1.3. and 3).

6.4. Shipboard Counter MeasuresIn all cases, methanol injection, if permissible, can prevent ice/hydrate formation but an operating procedure of systematicchange-over of reliquefaction units and the use of suitable filters in the condensate returns would appear to offer a methanol-free alternative procedure at a similar cost. (Section 3.5.). It is therefore recommended that the practicability and efficiencyof such a procedure and the availability and cost of suitable condensate filtration equipment be investigated as atechnically more acceptable solution for cargoes with a maximum methanol limitation.

Where the methanol injection method is used, it is recommended that methanol is injected into the hot gas stream beforeit enters the condenser so as to prevent hydrate formation where the condenser operates at temperatures below + 6°C.In following this recommendation, however, the warning noted in Section 5.1 on possible damage to lubricated cylindercompressors due to excessive contamination of the compressor lubricating oil should be borne in mind and proceduralmethods of avoiding such contamination be developed. Methanol injection into the condensate liquid, upstream of thecondensate return regulating valve, however, is acceptable for condenser temperatures above +6°C.

6.5. Shipboard Test MethodsThe Valve Freeze Test Method is used extensively and successfully in the distribution of pressurised LPG ashore in orderto detect the likelihood of regulating valve freeze-ups. It is recommended that use of the available Valve Freeze Testinstrument is developed for similar detection for ship's reliquefaction condensate return and for control of methanolinjection rates on ships carrying LPG at below 0°C and especially for fully-refrigerated propane. (Section 4.2.2).

In conducting the present Review, no test method was found for assessing methanol content of LPG and/or its vapour andsuitable for use on board ship. (Section 4.3.). It is recommended that thought be given in appropriate quarters to thedevelopment of such a method or methods.

6.6. Validation of Data AssumptionsThe data on water solubilities presented in Fig.3 and Table I contain a number of extrapolations from already publishedfindings. It would be most helpful if this data could be confirmed by a suitably equipped laboratory facility. An analysisof vapour and liquid (LPG/water/methanol) equilibria characteristics at appropriate temperature and pressures wouldalso be advantageous. Such studies would lead to a more detailed understanding of the water content of petroleum gasesin the various stages of processing and perhaps would indicate more clearly how methanol added to LPG liquid mayprevent freeze-ups when handling boil-off vapour.

The accuracy of the predicted methanol additions derived as outlined in Appendix III and illustrated in Figures 4 and 5,are dependent upon the reliability of the methanol distribution coefficient data extrapolated from the meagre informationpresently available. Figure 4, for instance, indicates that the quantity of methanol to add to prevent formation of hydratesin LPG at –45 °C is half that required at –10°C. The inference from Appendix III calculations is that between –10°C and–50°C more methanol comes out of solution with propane than is required to prevent freezing of the water which alsocomes out of solution with the same drop in temperature. If true in practice, this would halve the guideline methanol addition,provided that the LPG liquid was not required to be handled by pumps until reduced to –45°C. Additional research in thisarea on LPG-methanol-water systems (including the unsaturated gases) would be beneficial.

Lastly, an investigation of the rates at which ice/hydrates go into solution with LPG below its water saturationconcentration, would indicate the extent to which LPG can be dried through reliquefaction of its water-rich boil-offvapour despite the presence of ice or hydrate in the LPG which is available to go into solution in the liquid propane.

28

References

1. Poettman and Dean, Petroleum Refiner Vol. 25, No. 12, December 1946.

2. Williams and Lom, Liquefied Petroleum Gases, Guide to Properties, Applications and Uses. 2nd Revised Edition 1982.John Wiley & Sons Ltd., Baffins Lane, Chichester, West Sussex, UK. £45.00.

3. Tanker Safety Guide (Liquefied Gas), International Chamber of Shipping 1978. Witherby & Co. Ltd., 32-36 AylesburyStreet, London EC1 R OET, UK. £85.00.

4. Backhurst, Gray, and Harker, Problems from Contamination of LPG Cargoes with Ammonia, Marichem '82 Proceedings.Gastech Ltd., 2 Station Road, Rickmansworth, Herts, WD3 1QP, UK. £40.00.

5. Campbell, Gas Conditioning and Processing, Campbell Petroleum Services, 121 Collier Drive, Norman, Oklahoma73069, USA.

6. Katz, Handbook of Natural Gas Engineering. McGraw-Hill.

7. Dean and Poettman. Solubility of Alcohol in Liquid Propane and Normal Butane. (Phillips Petroleum Co. Research Dept.memo FHP-1-46).

8. Gas Processors Suppliers Association, Engineering Data Book (1980 Edition, Page 15-9).

9. AP Technical Data Book, July 1968, Figure 9A1.1.

10. Private Report submitted to SIGTTO by Member terminal operator, September 1982.

11.Miller & Carpenter, Solid-liquid phase diagram of the system methanol-water. Journal of Chemical & EngineeringData, July 1964.

12. Cooper, Dewpoint control and prevention of hydrate formation in the British Gas high pressure transmission system.Natural gas processing and utilisation conference, Institution of Chemical Engineers, Dublin 1976.

13. Kobayashi and Katz, Vapour-Liquid Equilibria for Binary Hydrocarbon Water Systems, Industrial and EngineeringChemistry, February 1953.

14. Gentry and Gunther, Water Solubility in Liquid Propane, Oil and Gas Journal, 1955.

29

EXPLANATION OF THE PROPANE-WATER SYSTEM

Appendix I

This Appendix explains why LPG boil-off vapour contains proportionately more water than is present in solution in theliquid and describes a phase diagram which shows the vapour, liquid and solid phases of propane and water which maybe present over the range of temperatures and pressures generally experienced in ship cargo tanks and their reliquefactionsystems (or in terminal storage tanks and their vapour recovery systems).

1. Why LPG Boil-off Vapour Contains More Water than the LiquidThe Handbook of Natural Gas Engineering (Ref. 6, Page 192) gives a phase diagram for the range of Propane-Watermixtures at a pressure of around 21 bar absolute and part of this diagram, slightly modified for our purpose, is reproducedas Figure Al . The left hand side of this diagram shows the phases which may be present in boil-off vapour reliquefaction.This diagram shows that the boiling point of pure propane at 21 bar absolute at 60°C, Point C, is depressed by about 1°Cwhen saturated with about 750 ppm by weight of water in solution in the propane liquid to Point B. L 2 is liquid propanecontaining up to its saturation quantity of liquid water in solution. Please note that since the solubilities of water andpropane in each other are very low, no attempt has been made to draw Figure Al to scale.

Figure Al. Phase Diagram for Propane – Water Mixtures at About 21 Bar Absolute.

33

Appendix I

The line CB is a plot of the Bubble Point temperature at which liquid L2 begins to boil on an increase in temperature. The line CG is a plotof the Dew Point temperature at which liquid L 2 begins to condense on a decrease in temperature. Thus a gas mixture X of propane andwater vapour on cooling to Point 1 deposits a dew of liquid L2 of composition shown by Point 2; conversely, a bulk liquid of compositionY will commence to boil at Point 2 and give off a vapour of composition shown by Point 1.

This is the explanation for why LPG boil-off vapour contains more water than is present in the bulk liquid solution. This higher watercontent in the boil-off vapour will rapidly dehydrate a small liquid sample so that the water content of boil-off vapour should be sampledfrom bulk liquid. This phenomenon is similar to the compositional change of boil-off vapour above an LPG mixture which is describedin Ref. 3 on Page 184.

At the top right hand side of the diagram the boiling point of pure water at 21 bar absolute, shown as 215°C at Point F, is depressedby the addition of liquid propane in solution in the water. L, is liquid water containing up to its saturation quantity of liquid propane insolution.

Ref. 13 describes the experimental determination of the Propane-Water composition at Points B, G and E, where the three phasesL, + L2 + Gas exist together for a range of temperatures and pressures. Thus, for example, at 21 bar absolute and +59°C, Point B is liquidpropane containing 750 ppm w/w water in solution, Point G is propane vapour mixed with 3340 ppm w/w water vapour and Point Eis liquid water containing 490 ppm w/w liquid propane in solution. Ref. 13 gives the temperature, pressure and composition of PointsB, G and E over a range of temperatures down to the hydrate formation temperature of +6°C.

Analysis of liquid propane samples at —40° C indicates a very low water content (less than 5 ppm) which is also indicated by extrapolatingthe straight line showing the solubility of water in propane at temperatures over +6°C in Figure 3 of the Study. Analysis of the boil-offvapour from the same bulk propane liquid at —40°C shows a substantially higher water content which confirms that the character of theleft hand side of Figure Al is similar at this lower pressure and temperature.

The mechanism can be approached numerically by recourse to Dalton's Law of Partial Pressures.

Where:-

Y is the mole fraction of water in the vapour phaseP W is the total vapour pressure of the systemPw is the partial pressure of the water in the vapour phasePHis the partial pressure of the hydrocarbon in the vapour phase

P°w is the vapour pressure of pure waterP° H is the vapour pressure of pure hydrocarbon

From Dalton's Law:-

P = + PH

Pw = Yw P

Pwor, Yw =

or, Y = w W Pw + Pw

At the saturation limit Pw will be approximately equal to P°wand PH will be approximately equal to P°H

P°P° w Ywsat approx. = „P w P°H

To obtain Yw, in terms of ppm w/w in propane vapour at saturation:-

Ywsat x 18.02 x 106ppm w/w —

Ywsat x 18.02 + (1— Ywsat) x 44.09

34

Appendix I

Calculating Yws., and ppm w/w water in boil-off vapour for a range of temperatures we find:-

Temp Water in P°w P°HYwsat Water in Ratio

°C Liquid*ppm w/w

bar a bar a Vapour ppmw/w

ppm in vapourppm in liquid

+ 40 280 0.07358 13.85 .0052845 2167 7.7

+ 20 100 0.02332 8.38 .002775 1136 11.40 About 32 0.00610 4.67 .0013045 534 16.7

– 20 About 10 0.00125 2.42 .000516 211 21.1– 40 About 3 0.00019 1.13 .000168 69 23.0

*From Figure 3 of Study

2. The Vapour, Liquid and Solid Phases of The Propane-Water System During Reliquefaction

The water content of boil-off vapour at various storage temperatures of bulk liquefied propane will therefore lie in therange between zero (Point C of Figure Al) for completely dry propane and Point G for fully water saturated propane liquid.Assuming that the boil-off vapour is always condensed at a temperature and pressure equal to or higher than its bulkstorage conditions, then the phases present in the reliquefaction system will be Gas (mixture of propane vapour with a smallquantity of water vapour), L2 and L1 or Hydrate.

Figure A2 has been constructed from information and data given in References 6 and 13. It is a more detailed version ofFigure 2 in the Study.

2(i) Direct cascade reliquefaction

The heavy dotted line in Figure A2 shows the cooling and condensation of propane boil-off gas at –15°C and 3 barabsolute pressure from fully refrigerated propane storage. If the bulk liquid is fully saturated with about 3 ppm w/w water,the boil-off vapour is expected to contain around 60-70 ppm w/w water vapour. Gas hydrate will not be formed as thepropane-water vapour mixture crosses line C, –B, since no free liquid water L, is present. Gas hydrates may be formedwhen the condensing surfaces are between 0°C and +6°C. As the gas mixture is condensed at –15°C the condensate L2

cannot carry more than about 13 ppm w/w water in solution at that temperature (from Figure 3 in the Study). Thereforesome 45-55 ppm by weight of water is freed to form hydrate with the liquid propane (one molecule of propane is associatedwith 17 molecules of water to form propane hydrate). This hydrate would seed out of the condensate as solid particles in theliquid. These solid particles may grow sufficiently numerous to choke the condensate return regulating/ expansion valve,depending on the size of the aperture through the valve when it opens. As the regulating valve opens to allow condensatepressure and temperature to drop to, say, 1.03 bar absolute and –40°C, a further 10 ppm water is freed to form additionalparticles of hydrate.

2(ii) Direct two stage reliquefaction

Propane boil-off gas from fully refrigerated storage is compressed in two stages and condensed in a sea water cooledcondenser at temperatures above +10°C and 7 bar absolute. The propane condensate formed will carry the maximum ofaround 60 ppm w/w water in solution L2 under these conditions. Hydrate will form when the condensate is cooled below +6°Cif the ppm water in the boil-off gas is above the maximum solubility L 2 at the condensate temperature in any part of the system(for example in inter-stage coolers or suction separator vessels) and it will certainly form downstream of the condensate returnregulating valve. The hydrate particle size may be sufficiently small to pass through filters above ASTM Sieve No. 100 mesh.

3. Formation Of Hydrates During Vaporisation Of Propane Liquid Contaminated With Water

As described in 2 above, if liquid propane contaminated with water is vaporised by passing through a regulating valve(as, for example, in producing vapour by drawing liquid from a deck storage tank), hydrates may form downstream of thevalve and may be in sufficient quantity to completely choke the vapour line.

35

Appendix I

Figure A2. Pressure — Temperature Phase Diagram of Propane — Water System in Region of Interest to PropaneReliquefaction Showing Boundary Conditions for Hydrate Formation in Propane Liquid and Vapour.

4. Ethane, Iso- And N-Butane

Ethane forms hydrates at temperatures below + 14.5°C with its upper quadruple point at 33 bar absolute. Iso-Butaneforms hydrates at tern peratures below +3°C with its upper quadruple point at 1.7 bar absolute. Liquid n-butane will form

hydrates in the presence of other hydrate forming LP Gases.

36

LPG FREEZE VALUE

Appendix II

OPERATIONAND MAINTENANCE

INSTRUCTIONS AT-1000L.P.G. FREEZE VALVE

Figure 1. Model At-1000 L.P.G. Freeze Valve.

General InformationThe Seiscor Model At-1000 LPG Freeze Valve was developed and calibrated to establish the presence or absence ofmoisture in propane type liquified petroleum gases by the Valve Freeze Method. The Valve Freeze Method was publishedfor information only by the American Society for Testing Materials in the "ASTM 1963 Standards-Volume 18". The useof this valve and test method provides a relative measure of the tendency of propane type liquified petroleum gasescontaining moisture to freeze in pressure reducing regulators and thereby interrupt the normal flow of gas. The test method is

applicable to all propane type LP gases including those that contain certain anti-freeze agents.

A liquid-phase sample is allowed to flow through the valve under the full flow condition to chill the valve housing by thecooling effect of the change from a liquid to the gaseous state. After the housing has been chilled, the opening of the valve

is switched to the testing condition and the time required for the freezing moisture to close the valve opening is measured.This time is recorded as the freeze time of the sample. If the freeze time indicates that the product may cause freezing ofpressure reducing regulators, an anti-freeze agent may then be added, if desired.

The valve and test method are particularly suited for use outside the laboratory. Therefore, they may be used by comparatively

unskilled labor and under existing conditions at commercial terminals with sufficient accuracy to determine if the moisture

content of the product meets specifications.

39

Appendix II

Figure 2. At-1000 L.P.G. Freeze Valve Parts Location.

Functional description

1. The freeze valve has two operating positions; a full flow position for chilling and/or purging, and a restricted flowposition for testing.

2. A spring loaded valve stem and cam action actuator provide instantaneous switching from the full flow conditionto the testing condition.

3. The valve features include four (4) independent restricted flow passages acting as flow smoothers to maintain apressure drop which will result in increased expansion of the fluids and increased cooling effect when enteringthe testing zone.

4. The valve incorporates a filter to exclude foreign particles from the testing zone.5. The outlet of the valve has internal threads to assist the operator in determining when the flow has been interrupted

or shut off as a result of ice having formed in the aperture of the valve. The instant the LP gas stops flowing over thethreads, a frost line rolls over the lip of the valve outlet.

Installation

The sensitivity of the moisture test measurements is such that all tests should be performed with the freeze valve apparatusconnected to the bulk supply source. If this is not possible, a sample may be taken into a sample cylinder having a minimumcapacity of three gallons. The sample should be collected in accordance with directions given in the ASTM Method D1265,Sampling Liquified Petroleum (LP) Gases or in NGPA Publication 2140. In no case shall the sample pressure be over 100pounds per square inch in excess of the vapor pressure of the product at the sampling temperature.

For proper assembly, the valve should be connected to the bulk supply source in a horizontal position with the valve outletin the most advantageous position for visual observation to assist the operator during the measurements.

NOTE: Connect the freeze valve to the sample source with a clean pipe or metal tubing.DO NOT USE RUBBER HOSE OR PLASTIC LINED HOSE.

40

Appendix II

The freeze valve has no full off condition. Therefore, a valve to shut off the gas flow should be installed in the systembetween the bulk supply and the freeze valve.

NOTE: Remove the protective set screw (Item No. 11, Figure No. 2) which protects the outlet of the valve, before anymeasurements are attempted.

Operating instructions

1. After the valve apparatus has been connected to the LP gas source, open the valve between the main source and thefreeze valve and operate the valve actuator to the purge position (actuator parallel to the valve body).

2. Purge the sample line and valve for 30 seconds to 1 minute, then switch the valve actuator to the test position (actuatorperpendicular to the valve body) for 2 or 3 seconds.

3. Continue the intermittent purging until the valve housing around the valve outlet is covered with frost. Snap the valveactuator to the test position and simultaneously start a stop watch. Stop the watch the instant the liquid ceases to flow.Flow cessation is indicated when a frost line rolls over the lip of the valve outlet.

4. Disregard the time of the first test. Immediately wipe the threads of the valve out let with a clean dry cloth and operatethe valve actuator to the purge position to remove the ice from the testing orifice. Repeat the test to ob tain the freeze-offtime until three (3) succes sive tests give consistent freeze-off times to within ± 10%.

NOTE: Failure to purge the valve 15 to 20 seconds between tests to remove all ice will result in inaccurate freeze timemeasurements.

5. When the freeze valve is operated under calibration conditions on a product containing no anti-freeze agent, a threetest average freeze time of 24 seconds is approximately equivalent to the initial color change of the indicator in theNGPA Cobalt Bromide Test method. (See NGPA Publication 2140). When the product being tested contains an anti-freeze agent, freeze time is no measure of the moisture content of the product.

Maintenance

The screen (Item No. 8, Figure 2) covering the inlet of the connection nipple (Item No. 9) and the filter element (Item No. 3)should be cleaned at regular intervals determined by the service hours of the valve. It is recommended that they be washed

Figure 3. L.P.G. Freeze valve CURVE.

41

Revised November, 1966

Appendix II

in petroleum ether or mineral spirits and dried with dry air or nitrogen. Avoid contamination with dirt or grease while

handling the filter element.

If the valve is dropped or damaged in any manner, it should be returned to Seiscor for reconditioning and recalibration.

The freeze valve is packed in a poly-urethane-lined case. The case is of high-impact material to resist damage to the

valve.

Parts list

Table 1 names and describes the maintenance parts of the Model AT-1000 LPG Freeze Valve. The part number column

indicates

Seiscor's part number and the description column names the part. All other parts are listed by item number and commercial

description only. The quantity column indicates total quantity per valve.

An asterisk notes a recommended spare part.

Table 1. Parts List.

ITEM NO. PART NO. NAME AND DESCRIPTION QTY.

At-1000 L.P.G. Freeze Valve 1

1 At-1100 Assembly 1

2 At-1017 Gasket 1

*3 At-1016 Filter Element 1

4 At-1015 Compression Spring 1

*5 At-1014 Gasket 1

6 At-1013 Plug 1

7 At-1012 Connection Nut 1

*8 At-1011 Screen 1

9 At-1010 Connection Nipple 1

*10 At-1019 Waldes Tru-Arc N-5000-31 Internal Retaining Ring 1

11 At-1020 1/4-20 x 1/4 Lg. Soc. Hd. Set Screw 1

42

Appendix II

The Model-AT-1000 LPG Freeze Valve meets the requirements of the American Society for Testing and Materials "StandardTest Method for Dryness of Propane (Valve Freeze Method)" and may be obtained from the following sources:-

Seismograph Service Corporation Stanhope Seta LtdPO Box 1590 London StreetTulsa ChertseyOklahoma Surrey KT16 8APU.S.A. England

Yashima Export & Import Co Ltd Thermal Scientific Inc9-1 Akasaka, 3-Chome PO Box 1669Minato-ku OdessaTokyo Texas 79760Japan U.S.A.

Seisco Scientific Ltd3320 9th StreetCalgaryAlbertaCanada

43

ESTIMATION OF METHANOL TO ADD TOPREVENT ICE/HYDRATE FORMATION

Appendix III

A. Methanol to add to a tank containing propane or n-butane saturated with water at 38°C to prevent ice/hydrateformation on cooling down tank contents from SVP at 38°C to atmospheric pressure.

1. Ref. 11 has been used to determine the quantity of methanol to add to "free" water in a propane-water and butane-water mixture to prevent ice/hydrate formation below +6°C and below +3 °C respectively. The water solubility levelsshown in Ref. 8 have been used as these are higher than in Ref. 13 and should result in safer methanol additionquantities.

2. Ref. 7 has been used to determine the distribution of the added methanol between water and LPG phases of the mixture.The data in Tables I and II of Ref. 7 have been re-worked to give grams methanol per gram water as a proportion ofgrams methanol per gram propane (or n-butane) and the results have been plotted in Figure A3 as a function of theliquid LPG-water-methanol mixture temperature. In this form, three of the n-butane results lie on a straight line between—46°C and —3°C on a log scale for the distribution coefficient and a linear temperature scale. This line is exactlyparallel to two results for the hexane-water-methanol system given in a paper by Kalervo Heinonen and Eero Tommilain Helsinki in 1969.

In constructing the distribution coefficient/temperature relationship for the propane system in Figure A3 therefore, aline parallel to the n-butane and hexane lines has been drawn through the Ref. 7 results for propane. Figure A3 hasthen been used to derive Figures 4 and 5 (given in Section 3.4.3) by the method which follows, albeit with differentresults for the methanol additions required than those given in Ref. 7.

3. Figure 1 of Ref. 11 provides information on methanol additions required to depress the freezing point of water. It hasbeen assumed that methanol is equally effective, concentration for concentration, in depressing the onset of hydrateformation. Required methanol additions to prevent hydrate formation have therefore been taken from Ref. 11 andconcentrations assumed to have the same depression range for hydrates as for ice but bearing in mind, of course, thatpropane hydrates form at +6°C and butane hydrates at +3°C.

4. Calculation Method

At +38°C, water saturated propane contains around 360 ppm w/w water; at - 20°C, water saturated propanecontains 9 ppm w/w water. Thus "free" water released during cooldown is 351 ppm w/w water. Figure 1 of Ref.11 indicates that 20% mole methanol, 80% mole water, will not quite freeze at —20°C — (+6°C) = —26°C. (See A.3above).

20% mole methanol in water is

20 x 32.03 x 100

64060 = 30.77% methanol w/w in mixture2082.2

Therefore, 351 ppm w/w free water requires 351 x(100— 30.77)

= 156ppm w/w methanol in the water

Figure A3 suggests that the methanol partition coefficient at —20°C is 360 "grams methanol per gram water per grammethanol per gram propane".

The methanol required to add to the LPG liquid phase is therefore

(106 —351) x 156/351360 = 1234 ppm w/w methanol

Total methanol added to mixture should be 156 + 1234 = 1390 ppm w/w. A similar calculation using values from Figure 1of Ref. 11, Ref. 8 and Figure A3 may be performed for other temperatures as in Table II.

18.0281x80+32.0332x% methanol by wt in mixture

20

30.77

47

Appendix III

Temperature (°C) of LPG-Water-Methanol liquid mixture

Figure A3 Methanol Distribution Coefficient in Propane-Water-Methanol and N-Butane-Water-Methanol

48

Appendix III

Table li — Added Methanol to Prevent Hydrate/Ice Formation in Propane and N-Butane ina Storage Tank on Cooling Tank Contents

Temperature °C Methanol

LPG LiquidPhase

Addition in ppm by

Free WaterPhase

Weight

Total

For propane initially saturated w ith water at 38°C

-46 526 359 885-37 761 299 1060-30 967 234 1201-20 1234 156 1390-15 1374 125 1499-10 1564 97 1661- 7.5 1587 84 1671- 5 1441 64 1505

0 1135 35 1170+5+7

250 5 255nil required

For n-butane initially saturated with water at 38°C

-11 510 37 547- 2.5 340 14 354

5. As can be seen from Table I of Section 2, propylene carries more water in solution in the liquid phase than propaneand, in lowering liquid temperature from 30°C to 0°C, the reduction in dissolved water is 470 ppm for propylenecompared with 140 ppm for propane and thus propylene-water mixtures will require a higher dosage rate. Noinformation has been obtained on the methanol partition coefficients between propylene and water. It has been advisedalso that propylene forms a hydrate at higher temperatures than propane.

Similarly, iso-butene "frees" 4.3 times as much water and iso-butane 1.9 times as much water as n-butane and willprobably require a higher dosage rate than n-butane. No information has been obtained on methanol partitioncoefficients between these butanes and water.

6. If propane is saturated with water at, say, + 15°C then it will contain only 75 ppm by weight of water instead of the360 ppm at +38°C assumed in the calculations of para. A.4. While less methanol is required for the much reduced"free" water on cooling from + 15°C to - 20°C (for example 75 -9 = 66 ppm instead of 360 - 9 = 351 ppm), the samequantity will go into solution in the propane liquid phase, i.e. 1234 ppm in the propane plus 156 x 66/351 = 29 ppmin the free water = 1263 ppm total. This compares with the not very dissimilar figure of 1390 ppm total calculated forpropane initially at +38°C.

7. Figure 4 plots the results of these calculations as given in Table II. Allowing a 15% margin on the minimum calculatedquantities, guidelines for methanol addition to propane and n-butane have been added. It is interesting to note thatthe 1950 ppm guideline figure of propane for cooling from +10°C to -15 °C is that as recommended in BritishSpecification BS.4250/1975 for this temperature range. (BS.4250 recommendation of 1 vol. methanol to 800 volumespropane = 1250 ppm v/v = 1950 ppm w/w). Figure 4 suggests, however, that if the propane is already below -15°Cthe methanol addition may be reduced.

49

Appendix III

B. Methanol to add to the reliquefaction condenser in propane and n-butane duty to prevent hydrate formation in the

condenser or in the condensate return.

1. It has been suggested that the boil-off vapour above LPG dosed with methanol will contain about twice the quantity ofmethanol in the vapour as in the LPG. If this is true then an inspection of Table II and of Table III indicates that this boil-off vapour will have sufficient methanol vapour to prevent ice/hydrate formation and no additional methanol need beadded at the cargo condenser.

2. Calculations have been made, therefore, on the assumption that no methanol has been added to the LPG in thecargo tank. The calculation method is similar to that detailed in A.4. and estimates the minimum required injectionof methanol into the boil-off vapour entering the condenser from LPG saturated with water at carriage temperaturesbelow +6°C for propane and +3°C for butane. The results are given in Table III and plotted on Figure 5.

3. For identical reasons described in A.6., the larger quantity of methanol required in solution in the liquid hydrocarboncondensate phase will not be reduced even if the water content in the tank liquid is significantly below saturationlevel.

4. For propane carried at atmospheric pressure and at a temperature of less than —40°C, similar methanol additionsare required as for propane carried at say 0°C, because propane at condensation temperatures of — 20°C or highercarry a significantly higher proportion of the added methanol in solution. Thus, at a condensate temperature of—7.5°C, 1587 ppm w/w methanol goes into solution in the propane. However, in view of what has been saidin B.1. above, and because the consequences of a freeze-up in the condensate system can be easily rectified, asmaller margin has been assumed in the guideline addition of 1800 ppm.

Table III — Added Methanol to Prevent Hydrates/Ice Formation in the Reliquefaction Condenser and CondensateReturned in Propane and N-Butane Reliquefaction.

Water ppm byInLiquid

WeightInVapour

Temp °C Methanol AdditionLPG LiquidPhase

in ppm by Weight

Free WaterPhase

Total

For propane saturated with water at temperatures below +6°C:

1 37 -45 540 36 576

1.4 50 -40 658 43 7014 110 -30 966 70 10366.5 150 -25 1122 81 1203

10 220 -20 1235 93 132816 284 -15 1373 98 147124 420 -10 1564 115 167930 500 - 7.5 1587 119 170636 580 - 5 1436 107 154352 780 0 1134 83 1217

For n-butane saturated with water at temperatures below +3°C:

9 1000 -10 488 218 705

13 1300 - 5 412 172 58417 1600 0 229 87 316

50

4 Dunlop Square,Deans Estate,Livingston,West Lothian,EH54 8SB

Tel: +44 (0) 1506 463 227Fax: +44 (0) 1506 468 999

E-mail: info@emailws.comweb: www.witherbyseamanship.com