dispose: large scale experiments for void fraction ... · pdf filefor void fraction...

$: DISPOSE: Large scale experiments for void fraction ... · PDF filefor void fraction measurement during venting ... Large scale experiments for void fraction measurement ... Commission$
Health and Safety Executive

DISPOSE: Large scale experiments for void fraction measurement during venting Prepared by the Health and Safety Laboratory for the Health and Safety Executive 2007

RR587 Research Report


DISPOSE: Large scale experiments for void fraction measurement during venting T J Snee, J Bosch, L Cusco, J A Hare D C Kerr, M Royle and A J Wilday Health and Safety Laboratory Harpur Hill Buxton SK17 9JN

The AWARD (Advanced Warning and Runaway Disposal) Project addressed the needs to detect runaway initiation in advance so that appropriate countermeasures can be taken and to design emergency relief systems for chemical reactors. The missing step in the design of runaway reactor relief systems was the availability of reliable methods for predicting level swell in the reactor during venting and hence the quantity of liquid requiring to be dealt with by a disposal system (quench tank, catch tank, etc.).

This report and the work it describes were funded by the Health and Safety Executive (HSE) together with the European Commission under the Competitive and Sustainable Growth Programme (project G1RD200100499), Astra Zeneca plc, Syngenta plc, Yule Catto plc and BS&B Safety Systems. Its contents, including any opinions and/or conclusions expressed, are those of the authors alone and do not necessarily reflect HSE policy.

HSE Books

© Crown copyright 2007

First published 2007

All rights reserved. No part of this publication may bereproduced, stored in a retrieval system, or transmitted inany form or by any means (electronic, mechanical,photocopying, recording or otherwise) without the priorwritten permission of the copyright owner.

Applications for reproduction should be made in writing to:Licensing Division, Her Majesty’s Stationery Office,St Clements House, 216 Colegate, Norwich NR3 1BQor by email to hmsolicensing@cabinetoffice.x.gsi.gov.uk

ii

EXECUTIVE SUMMARY

Background

The AWARD (Advanced Warning and Runaway Disposal) Project addressed the needs to detect runaway initiation in advance so that appropriate countermeasures can be taken and to design emergency relief systems for chemical reactors. The missing step in the design of runaway reactor relief systems was the availability of reliable methods for predicting level swell in the reactor during venting and hence the quantity of liquid requiring to be dealt with by a disposal system (quench tank, catch tank, etc.).

Objectives

The primary objective of the DISPOSE part of AWARD was: “To produce a methodology for the design of disposal systems to protect the workers and the environment from the effects of pressure relief of runaway chemical reactions. The methodology needs to be capable of producing a disposal system, which is adequate but not significantly oversized.”

HSL’s objectives were to provide the technical coordination of this part of the project; develop and build a large-scale experimental facility; carry out large scale experiments to obtain the axial void fraction profile in the reactor during venting; compare the results of simple hand-calculation methods with the large-scale experimental results; and produce guidance suitable for SMEs.

Main Findings

A large-scale experimental facility (2.2 m3 reactor; 13 m3 dump tank) was designed and built and eight large-scale experiments were performed. Void fraction was measured by means of both differential pressure and a novel technique of scanning gamma densitometry. The experiments were successful in providing datasets which can be analysed to obtain the axial void fraction profile in the reactor during pressure relief. A methodology has been developed and initial results of this analysis have been presented.

The mechanism of pressure turnover was derived by analysis of the experimental data for each experiment, and has been shown to be different from that assumed in commonly used vent sizing calculation methods. Comparisons were made between the experimental results and sizing methods given in the HSE Workbook (Etchells & Wilday, 1998). These methods were found to be conservative but this conservatism may be fortuitous given that the mechanism of pressure turnover was not as assumed by the vent sizing methods for most of the experiments.

Recommendations

Further work should be done to investigate the mechanism of pressure turnover for vented runaway reactions under conditions with larger vent sizes than those in the AWARD experiments. This would give greater confidence in the general validity of common vent sizing methods.

Recommendations for the sizing of vent disposal systems have been made and included in guidance for SMEs on the AWARD project website.

iii

iv

CONTENTS

1. Introduction ........................................................................................................................... 11.1 CEC AWARD Project................................................................................................... 11.2 Objectives...................................................................................................................... 21.3 This Report.................................................................................................................... 2

2. Main Tasks ............................................................................................................................ 53. Reaction System.................................................................................................................... 94. Previous Work..................................................................................................................... 11

4.1 Calorimetry ................................................................................................................. 114.2 Laboratory Scale Experiments .................................................................................... 124.3 Pilot Scale Experiments .............................................................................................. 12

5. Large Scale Experiments..................................................................................................... 145.1 Experimental facility and Procedure ........................................................................... 145.2 Test Matrix .................................................................................................................. 145.3 Results ......................................................................................................................... 15 5.4 Main features of Vented Runaway results .................................................................. 15

5.4.1 Maximum pressure .............................................................................................. 155.4.2 Behaviour with time ............................................................................................ 15

5.5 Derivation of Void Fractions....................................................................................... 175.6 Further Discussion of Experimental results ................................................................ 19

5.6.1 Mass remaining after venting .............................................................................. 195.6.2 Densitometer results ............................................................................................ 205.6.3 Temperature records............................................................................................ 205.6.4 Mechanism of pressure turnover ......................................................................... 23

6. Comparison of Experimental Results with Vent Sizing Predictions................................... 256.1 Introduction ................................................................................................................. 256.2 Methods which treat the reactor as homogeneous....................................................... 266.3 Methods which account for level swell ....................................................................... 27

6.3.1 Estimation of the void fraction at disengagement (for relief sizing purposes).... 276.3.2 Vent sizing using methods which account for level swell .................................. 27

7. Comparison of Experimental Results with methods for disposal system sizing ................. 297.1 Level swell .................................................................................................................. 29

7.1.1 Methodology ....................................................................................................... 297.1.2 Void fraction ....................................................................................................... 297.1.3 Mass vented......................................................................................................... 30

7.2 Vent Flowrate to Disposal System .............................................................................. 308. Conclusions ......................................................................................................................... 339. Recommendations ............................................................................................................... 3510. References ....................................................................................................................... 3711. APPENDIX A: Phi-Tec Adiabatic Calorimetry.............................................................. 39

11.1 Phi-Tec adiabatic calorimeter...................................................................................... 3911.2 Experimental procedure .............................................................................................. 40

12. APPENDIX B: Design of the Large-Scale Facility ........................................................ 5312.1 Design Considerations................................................................................................. 5312.2 Instrumentation............................................................................................................ 56

12.2.1 Void fraction ....................................................................................................... 5612.2.2 Other instrumentation.......................................................................................... 58

13. APPENDIX C: Large-Scale Experimental Results......................................................... 5913.1 Reactor Facility ........................................................................................................... 5913.2 Experimental Procedure .............................................................................................. 59

v

13.2.1 Hydrolysis of Acetic Anhydride.......................................................................... 5913.2.2 Acetic Acid Blowdown Tests.............................................................................. 59

13.3 Experimental Results................................................................................................... 6314. APPENDIX D: Derivation of Void Fractions ................................................................ 92

14.1 Gamma tomography .................................................................................................... 9214.2 Differential pressure cells............................................................................................ 9414.3 Density ........................................................................................................................ 9414.4 Conversion .................................................................................................................. 94

15. APPENDIX E: LITERATURE PAPER......................................................................... 97

vi

1. INTRODUCTION

1.1 CEC AWARD PROJECT

The maintenance of safe operating conditions for chemical reactors is of paramount importance to avoid accidents with potentially fatal consequences for personnel, major damage to installations and large-scale environmental pollution. The design of relief disposal systems for exothermic, runaway reactors is an important problem faced by the chemical industry across the European Union.

The AWARD (Advanced Warning and Runaway Disposal) Project has addressed the needs to detect runaway initiation in advance so that appropriate countermeasures can be taken and to design emergency relief systems for chemical reactors. It has done this by developing a device capable to detect runaway initiation together with engineering design tools to protect the environment using relief disposal systems such as catch and quench tanks. An innovative Early Warning Detection System (EWDS) has been developed and tested in calorimetric reactors, pilot plants and in a series of industrial reactors. The device is based on the application of non-linear dynamical systems theory.

The missing step in the design of runaway reactor relief systems was the availability of reliable methods for predicting level swell in the reactor during venting and hence the quantity of liquid requiring to be dealt with by a disposal system (quench tank, catch tank, etc.). The project has considerably advanced the understanding of level swell by means of a unique set of large-scale (2.2m3) experiments and by the development of a number of new improved models. A large-scale reactor facility was developed and instrumented to measure level swell by means of differential pressure measurements and a novel technique of scanning gamma ray tomography. Improved level swell modelling was developed, both using a drift flux approach within the reactor relief software code (RELIEF), which was developed by the European Joint Research Centre (JRC); and via a new dynamic model, incorporating non-equilibrium effects and foaminess. Simplified design guidelines on runaway reactor vent disposal system sizing/design and aimed at the needs of SMEs have been produced.

The project involved fourteen partners and an Associate Contractor from eight countries plus the European Commission Joint Research Centre. The partnership comprises the University of Manchester Institute of Science and Technology, UK (UMIST); the Health and Safety Laboratory of the Health and Safety Executive, UK (HSL); the European Commission (EC) Joint Research Centre, Italy (JRC) – Institute for Environment and Sustainability (IES) and Institute for the Protection and Security of the Citizen (IPSC)-; the Institut Químic de Sarrià (IQS), Spain; the Università Carlo Cattaneo (LIUC), Italy; the Università degli Studi di Messina (UM), Italy; Sanofi Chimie, France (Sanofi); Arran Chemical Company, Ireland (Arran); the Rohm & Haas Italia (R&H), Italy; the Esteve Química S. A. (EQ), Spain; the Segibo Srl, Italy (SEGIBO); Investigacao e Desenvolvimento em Enegnharia e Ambiente Lda, Portugal (IrRADIARE); and Warsaw University of Technology, Poland (WTU). Inburex, Germany, joined the consortium as an Associate Contractor to take over some of the responsibilities that were originally with JRC-IPSC.

The work plan comprised twelve work packages (WP’s):

WP1: Project management: administration, financial and technical WP2: Extension and theoretical development of the EWDS WP3: Experimental validation of the EWDS in small scale reactors

1

WP4: Small-scale and pilot plant venting experiments WP5: Industrial plant experiments for the EWDS WP6: Large-scale venting experiments WP7: Fundamental phenomena, Non-equilibrium effects WP8: Level swell WP9: Criteria for foaminess WP10: EWDS prototype design and development WP11: Sizing methods and guidelines on disposal system design WP12. Project exploitation and dissemination

Although some input to various other WP’s was provided, the contribution of HSL mainly focused on work packages 4, 5, 6, & 11 (highlighted above).

1.2 OBJECTIVES

The objectives of the project were:

1. Theoretical development and improvements to the early warning detection system (EWDS), aiming at reducing the number of temperature measurements required inside the reactor and extending the range of applicability of the detection criteria.

2. Application of CFD techniques for supporting the development of the early detection system by predicting temperature profiles and indicating a number and location of temperature sensors inside the reactor.

3. Experimental validation of the EWDS in small-scale reactors with reactions of industrial interest.

4. Experimental validation of the EWDS in industrial sites under normal and abnormal operating conditions to assess its robustness and final verification using an industrial process by experimentally simulating plant malfunctions, i.e. loss of coolant, stirrer failure, etc. leading to a runaway incident.

5. Design of the EWDS and improvements of the prototype after each experimental validation stage.

6. Development of reliable methods to predict the level swell in a venting chemical reactor during exothermic runaway and hence the flow rate and quantity of liquid which needs to be retained by the disposal system.

7. Interfacing of these level swell methods with existing calculation methods for sizing of the disposal system (both software models and simplified calculations).

8. Testing both the level swell methods and the design/sizing methods against large-scale experiments.

9. Development of a better design of reactors to promote effective mixing and so prevent runaway

10. Development of simplified guidelines for the design/sizing of disposal systems, suitable for use by SMEs.

11. Dissemination of and exploitation of the research results so that they are available for use by European industry, including SMEs.

1.3 THIS REPORT

This report describes the work performed by HSL on the runaway disposal aspects of the AWARD project. The report includes the results from adiabatic calorimetry and laboratory and pilot scale runaway reaction experiments on acetic anhydride hydrolysis, with and

2

without surfactant. Prediction of level swell, flow rate and calculation methods for sizing of disposal systems are discussed.

A description of the work performed by HSL on the early warning aspects of the AWARD project is provided in HSL report no: PS/05/02 (Snee, 2005). This includes results from the experimental validation of the early warning detection system (EWDS) at pilot-scale and large-scale.

3

4

2. MAIN TASKS

Various tasks were identified within the work packages described above. Those performed by HSL were as follows.

Project Management

Task 1.2: Scientific and technical coordination

HSL were responsible for the scientific and technical coordination of the work programme with respect to runaway relief disposal systems.

Small-scale and Pilot Plant Venting Experiments

Task 4.1: Summary of existing experimental results and physical properties data for three chemical systems.

HSL had already performed experiments at calorimeter, 1.5 litre vented reactor and 340 litre pilot reactor scales for three of the experimental systems to be used in this project. The following data were made available to the partners in the form of a CD-ROM: - System S1: reaction of acetic anhydride with water in stoichiometric quantities without surfactant to produce acetic acid. The results of 5 pilot-scale experiments at a range of fill levels and relief system set pressures, together with supporting calorimetric and 1.5 litre vented reactor experiments. - System S2: reaction of acetic anhydride with water in stoichiometric quantities to produce acetic acid, with a silicon-based surfactant. The results of 4 pilot-scale experiments at a range of fill levels and with a relief system set pressure equal to that in 3 of the system S1 experiments. Supporting 1.5 litre vented reactor experiments were also available. - System S3: decomposition of tert-butyl peroxy-2-ethylhexanoate in a high boiling solvent (Shellsol T), catalysed by cobalt octoate. Results of 8 pilot-scale experiments at a range of fill levels and catalyst concentrations, together with supporting calorimeter and 1.5 litre vented reactor experiments.

Large-scale Venting Experiments

Task 6.1: Design of experiments.

A new large-scale experimental facility was designed, including suitable instrumentation to measure the void fraction profile during pressure relief. Small-scale results were used in planning the experimental conditions. Costs were shared with another project for which a smaller reactor was required. The reactor was designed in three parts, such that a middle section could be installed for the AWARD experiments.

Task 6.2: Modifications to the experimental facility.

HSL’s new 2.2 m3 experimental facility was modified to carry out the large scale venting experiments required for AWARD. The modifications included: • Installation of the middle section of the reactor. • Provision of a suitable experimental relief system to connect the reactor to a quench tank. • Provision of an agitation system for the reactor.

5

• Provision of the means to measure local void fraction at axial points within the reactor, via gamma ray densitometry and differential pressure measurment.

• Provision of suitable instrumentation to measure temperature and pressure in the reactor, vent line and quench tank.

• Provision of load cell measurements for the reactor.

Development of improved level swell models in WP8 required the axial level swell profile, e.g. hydrostatic pressure with distance up the height of the reactor, for validation of level swell models.

Task 6.3: Commissioning of experimental facility.

The new test facility was commissioned and the measuring devices checked for their ability to give the required accuracy of results.

Task 6.4: Large-scale experiments.

Eight experiments were performed (six runaway reaction experiments and two blowdown experiments using the reaction products). Extensive measurements as a function of time were made during the experiments. These included pressure, temperature, void fraction (via hydrostatic pressure and density) and mass of fluid vented to the quench tank.

Task 6.5: Summary of results.

Results from large-scale experiments were made available to the partners as the experiments were completed. This report comprises the formal summary of results.

Sizing methods and guidelines on disposal system design

Task 11.1: Identify existing sizing/design methods

All the contractors provided HSL with details of existing sizing and design methods for determining the quantity and flow rate of material from the reactor to the disposal system and for sizing the different possible types of disposal system given this information. HSL also carried out a literature search. The description of the available methods and their underlying assumptions were collated.

Task 11.3: Evaluate sizing/design methods.

A spreadsheet was developed to facilitate calculations. Calculations were made, using a range of sizing methods, for the large-scale vented runaway experiments. The calculation methods used were those that could be evaluated without the use of a proprietary computer code. The calculations made use of data from Work Package 4.

Task 11.4: Compare sizing/design method predictions with experimental results.

The predictions of the different sizing methods were compared with the results of the large-scale experiments from Work Package 6. The sizing methods were evaluated in the light of these comparisons.

6

Task 11.5: Prepare design guidelines on disposal system sizing/design suitable for use by SMEs

Simple design recommendations for the sizing of disposal systems were produced in a form intended to be suitable for the needs of SMEs. The resulting document (Hare, 2005) has been made publicly available via the AWARD project web-site.

Task 11.6: Prepare report

The formal report on WP11 is incorporated within this report.

7

8

3. REACTION SYSTEM

Acetic anhydride reacts with water to produce acetic acid:

(CH3CO)2 + H2O d 2CH3COOH

The reaction is moderately exothermic and the kinetics are such that a runaway reaction can be initiated in either the pilot or large-scale reactors at temperatures well within the operating range of their respective heat transfer systems. The reaction was responsible for a severe explosion that occurred at an acetic anhydride plant in Australia (Leigh, 1992).

Small quantities of surfactant were added to the reagents and products of the reaction in order to produce foaming behaviour. Earlier work had indicated that Silwet L-7622, a silicon-based surfactant, was the most effective in producing a persistent foam.

Adiabatic and isothermal calorimetry were used to determine the temperature and concentration dependence of the rate of heat generation both with and without surfactant. The calorimetric data were used for vent-sizing calculations and also to determine the conditions for laboratory, pilot and large-scale reaction venting experiments.

9

10

4. PREVIOUS WORK

Previous work carried out by HSL on the reaction chosen for the AWARD large-scale experiments was made available to the AWARD partners as part of Task 4.1. This work is briefly summarised here.

4.1 CALORIMETRY

The experimental method and results of adiabatic calorimetry are summarised in Appendix A. Figure 1 shows the results of two calorimetric experiments designed to investigate whether the reaction kinetics are influenced by the addition of small quantities of surfactant. The figure shows that the addition of surfactant has no significant influence on the rate of self-heating. The comparison of the pressure temperature relations from the two experiments shown in Figure 2 indicates that the addition of surfactant has no strong influence on the vapour pressure of the reacting system.

The results shown in Figures 1 and 2 were used for vent sizing calculations. The similarity in the data sets means that calculations based on the closed system adiabatic data give the same recommended vent diameter whether or not surfactant is present.

0

2

4

6

wi

-1)

no surfactant th surfactant

Log

(dT/

dt) (

K m

in

-3.2 -3.0 -2.8 -2.6 -2.4 -2.2 -2.0 -1.8

-1000/T (K)

Figure 1. Adiabatic (Phi-Tec) self-heat rate data for the hydrolysis of acetic anhydride with and without surfactant.

11

14

12

10

8

6

4

2

0 50 100 150 200 250

(

wi

Pres

sure

bara

) no surfactant th surfactant

Temperature (°C)

Figure 2. Adiabatic (Phi-Tec) pressure-temperature relationships for the Hydrolysis of acetic anhydride with and without surfactant

4.2 LABORATORY SCALE EXPERIMENTS

The exothermic reaction between acetic anhydride and water had previously been studied in a 1.5 litre laboratory-scale facility (Snee, 1999). Experiments were performed with batch volumes of 0.5 and 0.75 litres over a range of relief set pressures both with and without surfactant.

Tests with and without surfactant, but with the same batch volume and relief set pressure, gave similar temperature-time profiles up to the point of vent opening. These findings are consistent with the calorimetric evidence that the addition of surfactant does not affect the reaction kinetics.

After vent opening, significantly higher reactor pressures were recorded for the tests when surfactant was present. This is consistent with the supposition that the addition of surfactant increases the proportion of liquid entering the vent-line causing a reduction in the volumetric discharge rate of vapour and a corresponding reduction in the degree of tempering.

The increase in maximum reactor pressure, due to surfactant, particularly for the larger batch volume and relief set pressure, indicated that considerable care would be required to establish safe conditions for the pilot and large-scale tests.

4.3 PILOT SCALE EXPERIMENTS

Nine experiments were performed in a 340 litre pilot plant reactor; 5 with no surfactant and 4 with surfactant (Snee, 1999). All used a vent diameter of 75 mm with no restriction orifice. Most used a set pressure of 200 kPa. and a range of fill levels were investigated.

12

The following qualitative features were evident from an overall examination of the video records and the results:

(a) The video records showed that two-phase flow was obtained in all the experiments. This could be seen as the reaction mixture level reached the top, and liquid could be seen in the vent line and entering the catch tank.

(b) The experimental results varied between conditions which gave full tempering with no overpressure to conditions which gave very high mass discharge rates and maximum pressures close to the safe working pressure of the reactor.

(c) Relatively small increases in the relief set pressure, batch volume or addition of a small quantity of surfactant resulted in large increase in the maximum reactor pressure.

(d) The addition of surfactant resulted in large increases in the maximum pressure and mass discharge rates.

(e) The pressure records indicated that, for the experiments which gave large overpressures, critical flow was obtained with the choke at the vent line exit in the catch tank.

The experimental results were compared with calculation for a range of simple methods summarized in the HSE` Wotkbook (HSE, 1998). These vent sizing methods undersized slightly for those experimental conditions with the highest batch volumes, both with and without surfactant. This undersizing was negligible given that designers will round up to the next available diameter. Further investigation found:

Detailed analysis of the experimental data indicated that vapour-liquid disengagement in the reactor was significantly reduced when surfactant was present. However, homogeneous venting, assumed in the some of the hand calculation methods, was not observed.

There was reasonable agreement between the mass discharge rates in the experiments with surfactant and values calculated using the Omega model and the homogeneous vessel assumption, although the void fraction was higher than for the homogeneous vessel assumption.

Self heat rates measured as a function of temperature in an adiabatic calorimeter gave good reproducibility and good agreement with those measured in the pilot-scale venting experiments.

Attempts to deduce the mechanism of pressure turnover (e.g. tempering due to sufficient vapour venting removing latent heat; emptying of the reactor; or consumption of the reactants) were inconclusive. However, for those vent sizing methods which undersized slightly, the mechanism of pressure turnaround was not as assumed by the method used.

Use of the same reaction system for the large-scale experiments was therefore of interest to further check any potential for the simple vent sizing methods to undersize, and to further investigate the mechanism of pressure turnover.

13

5. LARGE SCALE EXPERIMENTS

5.1 EXPERIMENTAL FACILITY AND PROCEDURE

The design of the large-scale experimental facility is discussed in Appendix B. The initial version of the experimental facility, which was used for other purposes, was modified to carry out the large scale venting experiments required for this project. Details of the modified facility, and the experimental procedure used, are given in Appendix C.

5.2 TEST MATRIX

A 100 mm diameter orifice plate was used in the experimental vent line to give a suitably long period of venting and facilitate capture of the process by the scanning gamma tomography system. All experiments were performed with the dump tank closed so as to prevent any emissions to the environment. This was required both as a result of HSL’s environmental risk assessment for the experiments and to maximise the collection of reaction products for subsequent blowdown experiments. The blowdown experiments were requested by the partners who were carrying out model validation as giving a simpler yet relevant case to model as a first stage in their validation procedure, before using the vented runaway results.

Six vented runaway reaction experiments were performed under conditions summarised in Table 1.

Table 1. Large scale vented runaway experimental conditions

Parameter Experiment number HP1 HP2 HP3 HP4 HP5 HP6

Acetic anhydride (kg) 1076.3 1510.5 1292.5 1066.9 1508.4 1282.5 Water (kg) 189.8 267.1 228.1 188.5 266.4 226.4 Surfactant (%) 0 0 0 0.25 0.25 0.25 Surfactant (kg) 0 0 0 3.14 4.44 3.77

In addition to these, two acetic acid blowdown tests were carried out using reaction products from a previous experiment. If required, surfactant was added to the reactor prior to heating. Both experiments were carried out with the dump tank unvented. Experimental conditions are summarised in Table 2.

Table 2. Large scale ‘Blowdown’ experimental conditions

Parameter Experiment number blow1 blow2

Acetic acid (kg) ~1560 1520.6 Surfactant (%) 0 0.25 Surfactant (kg) 0 3.8

14

5.3 RESULTS

The results of each of the large-scale runaway reaction experiments are summarised in Appendix C. Temperature and pressure records for the reactor, vent line and dump tank are plotted for each large-scale runaway reaction experiment. The responses from the gamma ray densitometers and the reactor load cells are also given.

Graphs indicating calculated void fractions during the blow down tests are shown.

5.4 MAIN FEATURES OF VENTED RUNAWAY RESULTS

5.4.1 Maximum pressure

The maximum pressures in each of the vented runaway reaction experiments are shown in Table 3 as a function of the initial fill level and whether surfactant was used.

Table 3. Maximum pressures for vented runaway experiments

Initial fill (%) Maximum pressure (bara) No surfactant With surfactant

50 3.60 5.76 60 6.00 7.19 70 6.78 7.71

It can be seen that there is a large increase in pressure from the experiment with 50% fill and no surfactant, if either surfactant is added or if the fill level is increased to 60%. At fill levels of 60 or 70%, adding surfactant has a more modest effect. For the experiments with surfactant, increasing the fill level from 50% to 60% gave a larger increase in maximum pressure than increasing it from 60 to 70%.

5.4.2 Behaviour with time

The main features of the experimental results will be discussed with reference to the experiment with 50% fill and with surfactant added (experiment HP4). Figure 3 gives experimental results as a function of time for both pressure and differential pressure. Figure 4 gives raw gamma ray densitometer and temperature results.

Both figures show pressure in the reactor (pink), vent line (yellow) and dump tank (dark blue). The reactor pressure continues to rise after vent opening and reaches a maximum before dropping to the dump tank pressure. The vent line pressure is taken just after the orifice and shows a venturi effect in that it is lower than the dump tank pressure.

15

0

1

2

3

4

5

6

7

0

2

4

6

8

10

12

14

16 Di

(

Diff

eren

tial p

ress

ure

(kP

a)

fferential pressure top - bottom mid top - mid bottom top - halfway

vent open

Pressure reactor vent line dump tank

Pre

ssur

e ba

ra)

2040 2060 2080 2100 2120

Time (sec) Figure 3. Differential pressure results for experiment HP4

(50% fill with surfactant)

0

80

0

2

4

6

8

10

)

i

top

160

Atte

nuat

ion

/ Pre

ssur

e (b

ara

Attenuat on upper source lower source


vent open

Reactor temperature

bottom T

empe

ratu

re (°

C)

2020 2040 2060 2080 2100 2120 2140

Time (sec)

Figure 4. Densitometer and temperature results for experiment HP4

16

The black differential pressure trace in Figure 3 is for the whole vessel and so can be used to give the mass in the reactor and differentiated to obtain mass flowrates (an otherwise difficult measurement to make, particularly for two-phase flow). The rise in the green differential pressure trace after vent opening is indicative of level swell such that liquid has moved from the lower half of the reactor into the upper half. The red differential pressure trace decreases after vent opening as material swells out of the bottom half of the reactor; the slight increase after approximately 2085 seconds indicates the end of two-phase venting and collapse of level swell back into the lower half of the vessel.

In Figure 4, each cycle of the gamma tomography system is seen as an increase to a peak followed by a decrease. The black trace is for the top gamma ray source and the lowest attenuation corresponds to the beam being horizontal and travelling through vapour. The increase in attenuation before vent opening is caused by the beam passing through the main vessel flange. After vent opening, the further increase in attenuation is due to level swell such that the beam is passing through a two-phase mixture in the reactor. The onset of two-phase venting immediately after vent opening is also shown by the temperature traces: the light blue trace (temperature at the top of the reactor) rapidly joins the green trace (temperature in the bottom of the reactor), indicating that liquid is present throughout the reactor.

5.5 DERIVATION OF VOID FRACTIONS

The gamma tomography and differential pressure results can be used to derive an axial void fraction profile for the reactor. The methodology used for this is given in Appendix D.

Within the time available within the AWARD project, analysis has concentrated on two of the experiments, HP1 and HP4, which were the experiments at 50% fill fraction with and without surfactant. The present analysis demonstrates features which can be derived from the experimental dataset.

An initial analysis of the tomography results used the signals from the densitometers in the top and bottom of the reactor at the points in the scanning cycles when the beams are horizontal. This allows void fraction profiles to be derived at points towards the top of the reactor (“top”) and towards the middle of the reactor (“bottom”). Results are shown in Figure 5 for experiment HP1 without surfactant and in Figure 6 for experiment HP4, with surfactant. In Figure 6 (with surfactant) the void fraction curves for the top and middle of the reactor meet, signifying that the reactor contents had become homogeneous. This did not occur in Figure 5 (without surfactant).

The time at which the maximum pressure occurred is also shown in these graphs. The ability to measure the void fraction at the top of the vessel at the maximum pressure is important in understanding the mechanism of pressure turnover (see 5.6 below). The void fraction profile was obtained by combined interpretation of the response from the scanning gamma ray system and the output of the differential pressure sensors. The differential pressure data was needed to derive the void fraction at the inlet to the vent line because this was above the top position covered by the scanning gamma tomography system. An example of the results of this analysis, for experiment HP1, is shown in Figure 7.

17

void

frac

tion

1

0.8

0.6

0.4

0.2

0 -20 0 20 40 60 80

Time to Pmax

Top bottom

time (s)

Figure 5: Void fraction at two levels in the reactor for experiment HP1 (50% fill, no surfactant)

void

frac

tion

1

0.8

0.6

0.4

0.2

0 -20 0 20 40 60 80 100 120

Ti

bottom top

me to Pmax

time(s)

Figure 6: Void fraction at two levels in the reactor for experiment HP4 (50% fill, with surfactant)

18

0.0

0.2

0.4

0.6

0.8

1.0

1.2

id

vent inlet top upper mid lower m bottom

Voi

d fra

ctio

n

vent open

3150 3160 3170 3180 3190 3200

Time (sec)

Figure 7: Axial void fraction profile for experiment HP1 (50% fill, no surfactant)

5.6 FURTHER DISCUSSION OF EXPERIMENTAL RESULTS

As discussed in 5.4.1, relatively large increases in the maximum pressure were obtained in changing conditions from those in experiment HP1 (50% fill and no surfactant), either by:

• adding surfactant (experiment HP6), or • increasing the fill level to 60% (experiment HP3).

The experimental results are further discussed here, in terms of possible reasons for these increases.

5.6.1 Mass remaining after venting

Table 4 gives the mass remaining in the reactor at the end of experiments HP1, 3 and 6. There is not a large difference in these results. Hoever, for both the experiments which gave rise to higher maximum pressures, there was a larger percentage of the mass discharged before the maximum pressure occurred, i.e. a greater amount of two-phase discharge resulted in higher maximum pressures. Level swell behaviour may therefore have influenced the maximum pressure attained, and it is therefore instructive to examine the experimental records in more detail.

19

Table 4. Mass remaining at the maximum pressure

Experiment Fill level Surfactant % of mass discharged Maximum pressure at maximum pressure (kPa)

HP1 50 No 27.3 3.60 HP3 60 No 33.3 6.0 HP6 50 Yes 29.6 5.76

5.6.2 Densitometer results

Densitometer output for experiments HP1 and HP3 are shown in Figures 8 and 9 and that for experiment HP4 was shown in Figure 4.

Comparing Figures 8 and 9, it can be seen that for Figure 9 (at 60% fill), there were significant bubbles in the lower half of the reactor (see lower densitometer trace in red) while this was not the case in Figure 8 (at 50% fill). Figure 4 (50% fill with surfactant) shows a response from the lower densitometer even sooner in the venting process.

5.6.3 Temperature records

Figures 10 shows temperature records at different levels in the reactor for experiment HP1 with 50% fill and no surfactant. A temperature profile can be seen in which the bottom temperature is significantly higher than the temperature at higher levels in the reactor. This is consistent with the densitometer measurements in Figure 8. However, Figure 11 (50% fill with surfactant) shows a very uniform temperature throughout the reactor during venting. This is consistent with the densitometer results, for example as seen in Figures 5 and 6, where the reactor became homogenous for the experiment with surfactant but not for that without surfactant.

Although the densitometer results show bubbles in the lower part of the reactor (Figure 9) for the experiment at 60% fill without surfactant. Temperature results indicate that there was still a profile in temperature and that all the experiments without surfactant were significantly less homogenous than those with surfactant.

The temperature and densitometer results therefore indicate that, although both adding surfactant and increasing the fill level with no surfactant gave similar increases in maximum pressure, the level swell behaviour in these cases was different.

20

0

25

50

75

0

2

4

6

8

10

i

)

top bottom

100

125

150

175

200

Attenuation upper source lower source

Pressure reactor vent l ne dump tank

Atte

nuat

ion

/ pre

ssur

e (b

ara

Tem

pera

ture

(°C

)

Reactor temperature

3120 3140 3160 3180 3200 3220 3240

Time (sec.)

Figure 8: Scanning gamma densitometer measurements for experiment HP1 (50% fill, no surfactant)

0

20

40

60

80

100

120

140

160

180

200

0

2

4

6

8

10

)

top

Attenuation upper source lower source


Atte

nuat

ion

/ pre

ssur

e (b

ara

vent open

Tem

pera

ture

(°C

)

Reactor temperature

bottom

1940 1960 1980 2000 2020 2040 2060

Time (sec.)

Figure 9: Scanning gamma densitometer measurements for experiment HP3 (60% fill, no surfactant)

21

top

)

1.5

2.0

2.5

3.0

3.5

4.0

4.5

5.0

130

140

150

160

170 Te

mpe

ratu

re (°

C)

reactor temperatures

bottom

vent open

Pre

ssur

e (b

ara

reactor pressure

3140 3150 3160 3170 3180 3190 3200

Time (sec)

Figure 10: Temperature measurements for experiment HP1 (50% fill, no surfactant)

2

3

4

5

6

120

130

140

150

160

170

180

vent open

reactor temperatures top

bottom

Tem

pera

ture

(°C

)

Pre

ssur

e (b

ara)

reactor pressure

2050 2060 2070 2080

Time (sec)

Figure 11: Temperature measurements for experiment HP4 (50% fill, with surfactant)

22

5.6.4 Mechanism of pressure turnover

For experiments HP1 and HP4, the void fraction versus time profile entering the vent line was used to estimate the energy removal rate via venting. In Figure 12, this energy release rate per unit mass has been plotted as a function of temperature, along with the heat release rate per unit mass due to the reaction (measured by calorimetry).

10000

8000

6000

4000

2000

0

400 410 420 430 440 450 460 470 480

)

l

i

) )

openHea

t rat

e (W

/kg

heat generation adiabatic data

heat remova no surfactant with surfactant

tempering, heat removal exceeds heat generation

maximum adiabt c rate

maximum temperature (tempered

turnaround due to reactant consumption

maximum temperature (untempered

vent

Temperature (K)

Figure 12. Energy balance for the reactor during venting for experiments HP1 and HP4 (50% fill, no surfactant and with surfactant)

For experiment HP1 with no surfactant, there is a point during the venting when the heat removal (green trace) exceeds the heat generation (red trace). This indicates that tempering occurred at the relevant temperature. Because of the tempering, the final (maximum) temperature is less than that for experiment HP4 with surfactant (blue trace). It can also be seen that at no point during the runaway did the heat removal (blue trace) exceed the heat generation (red trace) for the experiment with surfactant. For this experiment, tempering did not occur. It can also be seen that the turnaround in pressure for this experiment was caused by reactant consumption, since the maximum temperature for the blue trace roughly coincides with the maximum rate of heat generation by the reaction (red trace).

Analysis of the other experiments showed that tempering occurred only in experiment HP1. In all the other experiments, the level swell have sufficient two-phase venting that the heat removal rate was reduced such that tempering did not occur. As discussed above, the flow regime in the reactor for this two-phase venting was different between the experiments with surfactant and those without. The reactor contents became more homogenous for the experiments with surfactant.

In none of the experiments did sufficient emptying occur for the pressure turnaround to be due to emptying (or to the flow becoming single phase vapour).

The mechanism of pressure turnover was therefore concluded to be:

23

• Tempering for experiment HP1 (50% fill and no surfactant). • Reactant consumption for all the other experiments.

The detailed analysis to determine the mechanism of pressure turnaround in the large-scale experiments is described in Appendix E. Appendix E also includes a comparison with pilot-scale results on the hydrolysis of acetic anhydride and review of pilot-scale data for other reaction systems.

24

6. COMPARISON OF EXPERIMENTAL RESULTS WITH VENT SIZING PREDICTIONS

6.1 INTRODUCTION

The large-scale experiments were designed primarily to measure level swell and this meant that the vent area was smaller and the overpressure higher than in most pressure relief system designs. However, a comparison with available simple relief system sizing methods was carried out and is described below.

The calculation methods used for the comparison were those given in the HSE Workbook (Etchells, 1998). These are hand-calculation methods, i.e. methods which can be evaluated using a pocket calculation. However, for convenience, a spreadsheet was developed to carry out the calculations.

The calculation of the vent area (A) is a two-stage process. The required relief rate (W) is first calculated. This is the mass flowrate, which must be removed from the reaction vessel in order to prevent overpressurisation. Secondly the relief system capacity (G) is calculated. This is the mass flowrate per unit area through the pressure relief system. The vent area is then calculated as:

A = W / G (1)

The large-scale AWARD experiments were on a vapour pressure system (for which the pressure rise during runaway is due to the vapour pressure of the components). The required relief rate (W) for vapour pressure systems can be calculated by the following methods which are described in Etchells (1998).

The methods which assume that the reactor contents are a homogenous mixture are: • Leung’s method (Workbook section 6.3.2), • Huff’s method (Workbook A5.2), and • Fauske’s method (Workbook A5.3.2).

The methods which take some account of level swell in the reactor are: • Fauske’s method with disengagement (Workbook A5.3.4), and • Wilday’s method with disengagement (Workbook A5.5).

These methods require a level swell calculation to be made using the methods in Annex 3 of the Workbook, and using equation A3.1 for the superficial vapour velocity. This assumes that disengagement (the end of two-phase venting) occurs at the maximum pressure with the vapour being produced by the runaway reaction.

Leung’s method originally used an arithmetic mean for the average heat release rate per unit mass of reactants. Subsequently, Leung proposed an alternative average, utilising the Boyle time and this is more appropriate for the high overpressures in the experiments. Both methods have been used for both Leung’s method and Wilday’s method with disengagement, which also requires an average value of the heat release rate per unit mass.

The flow capacity per unit area (G), which is required by most of the methods, has been calculated using Leung’s Omega method, given in Annex 8 of the Workbook. In calculating G it has been assumed that the two-phase mixture entering the vent results from homogeneous mixing in the reactor. This is the assumption suggested by the Workbook.

25

All vent sizing calculations have been performed so as to use the experimental set pressure and maximum measured pressure to back-calculate a predicted required vent diameter. This is then compared with the experimental vent diameter of 100 mm.

6.2 METHODS WHICH TREAT THE REACTOR AS HOMOGENEOUS

Table 5 gives the results of the predicted required vent diameters. For Leung’s method, Figure 13 shows the relationship between the predicted vent diameter and the maximum pressure.

Table 5. Vent sizing results for methods which treat the reactor as homogeneous

Surfactant 50 60 70 50 60 70

Experiment 100 100 100 100 100 100 318 193 188 181 157 158

time mean) 321 214 214 200 186 190

Fauske’s method 210 144 144 135 131 126

315 166 169 159 140 139

No surfactant Fill (%)

Vent diameter (mm)

Leung’s Method (using arithmetic mean) Leung’s Method (using Boyle

Huff’s method

450

400

350

300

250

200

150

100

50

0

)Ve

nt D

iam

eter

(mm

50% fill Homog Calc 60% fill Homog Calc 70% fill Homog Calc 50% fill No Surf Expt 60% fill No Surf Expt 70% fill No Surf Expt 50% fill Surf Expt 60% fill Surf Expt 70% fill Surf Expt

0.00 200.00 400.00 600.00 800.00 1000.00 1200.00

Maximum Pressure (kPa)

Figure 13. Vent diameter as a function of maximum pressure for Leung’s method (with arithmetic mean heat release per unit mass)

26

All of the methods proved to be conservative for these experiments in that they overestimated the required vent diameter. Fauske’s method was outside its range of applicability of 10-30% overpressure. Leung’s method was outside its original applicability range of 050% overpressure, but met the alternative criterion given in CCPS (1998).

6.3 METHODS WHICH ACCOUNT FOR LEVEL SWELL

6.3.1 Estimation of the void fraction at disengagement (for relief sizing purposes)

Churn Turbulent void fractions were therefore calculated with correlation parameter Co values of 1 and 1.5. Calculations for the Bubbly and Droplet flow regimes did not generate realistic solutions. The experimental and calculated disengagement void fractions are compared in Table 6. Experimental void fraction measurements were available from two sources: from a gamma ray densitometer and from differential pressure measurement. From the separate measurements, void fractions were estimated at different heights in the vessel. Disengagement was identified as when lack of homogeneity began to develop in the void fraction measurements, toward the end of the reactor venting.

The calculated void fractions for C0 of 1 were conservative for the experiments without surfactant and good estimates of the final void fraction for those with surfactant (although the churn turbulent flow regime would not be expected for a system including surfactant). The calculated void fractions for a C0 of 1.5 were potentially non-conservative in that they predicted disenagagement before (i.e. at a higher level in the reactor) it occurred in the experiments.

Table 6. Void fractions at disengagement for the purpose of vent sizing

No surfactant Surfactant Fill (%) 50 60 70 50 60 70

Void fraction Experiment 0.545 0.586 0.664 0.813 0.811 0.806 (from Differential Pressure) Experiment (from 0.645 0.665 0.695 0.825 Not 0.815 Densitometer) available Calculated 0.936 0.908 0.905 0.901 0.870 0.867 Churn Turbulent (Co = 1) Calculated 0.638 0.625 0.623 0.621 0.607 0.605 Churn Turbulent (Co = 1.5)

6.3.2 Vent sizing using methods which account for level swell

Table 7 gives the results of the predicted vent diameters. Figure 14 shows the variation in predicted vent diameter with maximum pressure for Wilday’s method with disengagement.

27

Table 7. Vent sizing results for methods which take some account of level swell

Surfactant 50 60 70 50 60 70

100 100 100 100 100 100

o=1) 199 135 136 124 115 120

o

252 155 158 132 121 128

o

253 180 172 145 155 144


Vent diameter (mm) Experiment (mm)

Fauske’s Disengagement Method (CT CWilday’s Disengagement Method (CT C = 1.5) (using arithmetic mean) Wilday’s Disengagement Method (CT C = 1.5) (using Boyle time mean)

50

0

350

300

250

200

150

100

Vent

Dia

met

er (m

m)

50% fill Churn Turb Calc 60% fill Churn Turb Calc 70% fill Churn Turb Calc 50% fill No Surf Expt 60% fill No Surf Expt 70% fill No Surf Expt 50% fill Surf Expt 60% fill Surf Expt 70% fill Surf Expt

0.00 200.00 400.00 600.00 800.00 1000.00 1200.00

Maximum Pressure (kPa)

Figure 14. Vent diameter as a function of maximum pressure for Wilday’s method (with arithmetic mean heat release per unit mass)

Fauske’s method with disengagement was used outside its stated applicability range of 1030% overpressure. Churn Turbulent vent sizing calculations with Co=1 normally give larger vent sizes than those with Co=1.5. The opposite effect is apparent in Table 7 because Fauske’s method was valid when Co=1 and Wilday’s method when Co=1.5.

The methods were all found to be conservative in that they predicted required vent diameters which were larger than the actual vent diameter. The methods were less conservative for the experiments with surfactant (for which the level swell models are not really applicable).

28

7. COMPARISON OF EXPERIMENTAL RESULTS WITH METHODS FOR DISPOSAL SYSTEM SIZING

7.1 LEVEL SWELL

7.1.1 Methodology

For relief disposal system sizing, the end of two-phase venting needs to be predicted. This will often occur after the pressure has peaked, when the reactor is undergoing depressurisation back to atmospheric pressure, or to the set pressure of any safety valve. Depressurisation may cause flashing and/or dissolved gas to come out of solution.

Disengagement void fractions required for the Churn Turbulent flow regime were obtained using the method outlined in Annex 3 of the HSE Workbook. (Calculations for the Bubbly and Droplet flow regimes did not generate realistic solutions.) Void fractions at the end of two-phase relief for disposal system sizing (Workbook Eqn A3.2 at the maximum pressure) were used. (Note that g in the equation is a subscript to the ρ).

Provided that vapour flow from the reactor to the disposal system would be choked, the method is roughly independent of the reactor pressure assumed. The maximum reactor pressure was used for convenience as physical properties had already been developed for the relief sizing calculations discussed in section 6.

7.1.2 Void fraction

The experimental and calculated disengagement void fractions in the reactor are compared in Table 8. The experimental data are identical that to those in Table 6. The Churn Turbulent void fractions were calculated using correlation parameter Co values of 1 and 1.5.

Table 8. Void fractions at the end of two-phase venting

No surfactant Surfactant Fill (%) 50 60 70 50 60 70 Void fraction Experiment (from 0.545 0.586 0.664 0.813 0.811 0.806 Differential Pressure) Experiment (from 0.645 0.665 0.695 0.825 Not 0.815 Densitometer) available Calculated 0.756 0.752 0.752 0.76 0.76 0.761 Churn Turbulent (Co = 1) Calculated 0.549 0.546 0.546 0.551 0.551 0.551 Churn Turbulent (Co = 1.5)

Without surfactant, the calculated disengagement void fractions were slight over-predictions (for Co=1) and slight under predictions (for Co=1.5). With surfactant, the calculated disengagement void fractions were slight under-predictions (for Co=1) and gross under predictions (for Co=1.5). However, the flow regime for the experiments with surfactant would not be expected to be churn-turbulent as was assumed in the calculations.

29

7.1.3 Mass vented

The mass vented can be calculated assuming homogeneous venting (the entire batch mass is vented). It can also be calculated using the disengagement void fractions from Table 8. The calculation method is explained in the Workbook A3.3.5 – End of Two-Phase Relief. The experimental and calculated values of the mass vented are compared in Table 9. The experimental data were obtained from differential pressure measurements.

Table 9. Mass vented from reactor to disposal system

No surfactant Surfactant Fill (%) 50 60 70 50 60 70

Mass vented (kg) Experiment (from 382.9 745.7 1152.7 907.5 1163.4 1421.8 Differential Pressure) Calculated 1266.2 1520.6 1777.6 1266.2 1520.6 1777.6 Homogeneous Calculated 793.5 1056.5 1316.3 815.3 1077.7 1337.4 Churn Turbulent (Co = 1) Calculated 391.9 670.6 933.3 422.7 691.4 951.1 Churn Turbulent (Co = 1.5)

For the experiments without surfactant: • The calculated mass vented was over predicted by the churn turbulent method (for Co=1)

and under-predicted (for Co=1.5). • The calculated mass vented was grossly over predicted assuming homogenous venting..

For the experiments with surfactant: • The calculated mass vented was under-predicted by the churn turbulent method (for

Co=1) and grossly under-predicted (for Co=1.5). • The calculated mass vented was significantly over predicted assuming homogenous

venting.

7.2 VENT FLOWRATE TO DISPOSAL SYSTEM

This is a necessary input for the sizing of disposal systems which act as separators.

The mass flow rate per unit area was calculated using Leung’s Omega method, given in Annex 8 of the HSE Workbook. Calculations assumed that the mixture entering the vent was for homogeneous mixing in the reactor.

The experimental and calculated mass fluxes (flow rate per unit area) are compared in Table 10. The calculated values were also used in the vent sizing calculations reported in section 6. Two experimental mass fluxes were available: An average mass flux calculated from the longer overall venting duration (includes single phase flow periods) and a two phase mass flux calculated from the shorter two phase duration. Both times are given in Table 10.

30

Table 10. Comparison of vent mass flux predictions with experiments

Surfactant 50 60 70 50 60 70

Mass flux (kg/m2.s) Experiment (Average) Experiment (Two Phase)

1806

method)


635.1 1865.3 2297.5 3357.5 3118.5 3200.3

2826.6 298.6 3814.4 3245.2 4949.2

Calculated (Omega 1253.1 2104.2 2249.2 2026.9 2312.6 2426.5

For the experiments without surfactant, the experimental flow rates were close to calculated values.

For the experiments with surfactant, the experimental flow rates exceeded calculated values. This is not conservative for disposal system sizing. However, it is also uncertain to what extent separator disposal systems will work for foamy systems. For these foamy systems, both the average and maximum flow rates were underestimated. A safety factor should therefore be applied to the calculated flow rate and the comparisons show that the factor of 2 suggested in the HSE Workbook was sufficient in most cases. For the experiment with surfactant and with the highest fill ratio, a factor of 2.5 was needed to estimate the maximum flow rate.

31

32

8. CONCLUSIONS

The main conclusions are as follows.

1. A large-scale experimental facility (2.2 m3 reactor; 13 m3 dump tank) has been designed and built to allow investigation of vented runaway reactions, and in particular the void fraction distribution up the reactor due to level swell.

2. Eight large-scale experiments have been performed: six vented runaways and two blowdowns of reaction products. The experiments used the reaction of acetic anhydride with water and half the experiments were performed with a surfactant added to increase the foaminess of the mixture. Void fraction has been measured by means of both differential pressure and a novel technique of scanning gamma densitometry.

3. The experiments have been successful in providing datasets which can be analysed to obtain the axial void fraction profile in the reactor during pressure relief. A methodology has been developed and initial results of this analysis have been presented.

4. Analysis of the experimental results is able to demonstrate the different level swell behaviour between experiments performed with and without surfactant.

5. The mechanism of pressure turnover has been derived by analysis of the experimental data for each experiment. Only for one experiment was tempering shown to have significantly influenced the pressure turnover. For the other experiments, reactant consumption was the dominant effect. The mechanism of pressure turnover has therefore been shown to be different from that assumed in commonly used vent sizing calculation methods.

6. A comparison has been made between the experimental results and required vent sizes predicted using methods from the HSE Workbook (Etchells, 1998). These methods were found to be conservative (over-estimating the required vent size) for the experiments performed. However, this conservatism may be fortuitous given that the mechanism of pressure turnover was not as assumed by the vent sizing methods for most of the experiments.

7. A comparison was also made between the experimental results and level swell and flow rate calculations required as an input to disposal system sizing. The methods were found to be conservative (for disposal system sizing purposes) for those experiments without surfactant, but were not always conservative for the experiments with surfactant.

33

34

9. RECOMMENDATIONS

The following recommendations are made.

1. Further work should be done to investigate the mechanism of pressure turnover for vented runaway reactions under conditions with larger vent sizes than those in the AWARD experiments. The results have shown that a small change in batch volume or the addition of surfactant can produce a large increase in the maximum reactor pressure during venting. Because mechanism of pressure turnaround observed in the experiments differed from that assumed in most vent-sizing methods, it is not possible to determine whether these methods are generally conservative without such further experiments. Previous comparisons with pilot-scale vented experiments showed that the relief sizing methods for this reaction system ranged from conservative to just adequate.

2. In using the simple methods in the HSE Workbook for disposal system sizing:

• For systems which are not surface-actively foamy, an approximate estimate of the quantity of liquid vented to the disposal system can be obtained using the level swell methods detailed in Appendix 3 of the HSE Workbook (for estimating the end of two-phase flow). For systems which are foamy, it can be conservatively assumed that all the contents of the reactor vent to the disposal system, although significantly less than this was vented in the HSL large-scale experiments with surfactant.

• For disposal systems which act as separators, the maximum two-phase flow rate into the disposal system is also required for sizing. The omega method given in Appendix 8 of the HSE Workbook can be used to estimate this flow rate, using inlet conditions based on assuming a homogenous two-phase mixture in the reactor. For systems which are not surface-actively foamy, this gave a conservative overestimate of the average flow rate but underestimated the maximum flow rate. For foamy systems, both the average and maximum flow rates were underestimated. A safety factor should therefore be applied to the calculated flow rate and the comparisons show that the factor of 2 suggested in the HSE Workbook was sufficient in most cases. For the experiment with surfactant and with the highest fill ratio, a factor of 2.5 was needed to estimate the maximum flow rate.

35

36

10. REFERENCES

Etchells, J C and Wilday, A J (1998), "Workbook for chemical reactor relief system sizing", http://www.hse.gov.uk/research/crr_htm/1998/crr98136.htm, HSE Contract Research Report 136/1998, HSE Books

Hare, J A and Wilday, A J, (2005), “The Sizing of Disposal Systems for Runaway Reaction Emergency Relief Systems: Simplified Guidance form the AWARD European Project”, AWARD Project website: http://www.arpconsortium.org/AWARD.htm

Snee, T J, Bosch J, Cusco L and Kerr DC, (2005), “AWARE: Investigation of the Early Warning Detection System through Pilot and Large Scale Tests”, HSL Report No PS/05/02

Snee, T J, Butler, C, Hare, J A, Kerr, D C, Royle, M and Wilday, A J, (1999), “Venting studies of the hydrolysis of acetic anhydride with and without surfactant (Vapour System 3)”, HSL Report No PS/99/13

37

38

11. APPENDIX A: PHI-TEC ADIABATIC CALORIMETRY

11.1 PHI-TEC ADIABATIC CALORIMETER

The Phi-Tec adiabatic calorimeter consists of a small thin-walled test cell, of around 100 ml capacity, suspended in the centre of a set of electrically powered heaters within a stainless steel pressure vessel. The sample is placed in the test cell and heated until a reaction is detected. The reaction temperatures and pressures are monitored and adiabatic conditions are maintained by controlling the heaters such that their temperature tracks the sample temperature. The lack of strength of the thin walled test cells is compensated for by automatically applying an external nitrogen pressure. It is also possible to apply heating directly to the test cell by means of the calibration heater. The cell contents may be stirred magnetically or directly depending upon the type of test cell used. A schematic diagram of the apparatus is given in Figure A1.

Figure A1. Schematic diagram of Phi-tec apparatus

39

11.2 EXPERIMENTAL PROCEDURE

A magnetically stirred, PHI-TEC test cell (type 1a) was assembled into the apparatus and evacuated. 62.9 g of acetic anhydride containing the required amount of surfactant was then drawn into the test cell and the vacuum subsequently restored. The guard heaters were held at 55°C and the calibration heater was used to heat the acetic anhydride to a temperature sufficient to provide the required reaction starting temperature, taking into account the cooling produced by the addition of cold water and endothermic mixing. Upon reaching this preheat temperature, the calibration heater was switched off and cold distilled water (11.1 g) was drawn into to the test cell. When the post mixing temperature had stabilised, both the test cell and the outer vessel were opened momentarily to atmosphere to provide a common reaction start pressure of approximately 100 kPa for all tests. The guard heaters were then switched on and allowed to track the exotherm to completion. The phi factor value for the tests depends on the test cell type, sample mass and specific heat capacity. A low value of 1.07 was calculated and so no correction was made to the test results.

This procedure was initially carried out with the guard heaters held at 50°C. Two further experiments were carried out with lower starting temperatures.

Table A1. Experimental conditions

Run number after water

addition (°C)

Surfactant concentration

(%wt/wt) PA81 49.6 0 PA82 49.2 1 PA83 30.0 1 PA91 25.0 0.3

A3.

in Figure A3.

Table A2. Results

Run no.

Nominal start temp. (°C)

surfactant concentration

(% wt/wt)

Adiabatic temperature

rise (K)

temperature rate

(Ks-1)

Time to

rate (s) (kPa)

PA81 50 0 4.3 946 1349 PA82 50 1 3.2 996 1250 PA83 30 1 1.7 2765 1015 PA91 25 0.3 2.4 3153 909

Temperature

PHI-TEC RESULTS

Results from PHI-TEC experiments are summarised in Table A2 and limited data sets from each experiment are presented in Tables A3 to A6. Plots of temperature against time for each test are presented in Figure A2. The corresponding plots for pressure against time are given

Maximum Maximum

Maximum pressure

186.5 183.0 190.9 189.1

40

250

225

200

175

150

125

100

75

50

25

0

Pre

ssur

e (b

ara)

Te

mpe

ratu

re (°

C)

PA81 PA82 PA83 PA91

0 1000 2000 3000 4000 5000

Time (s)

Figure A2. Temperature v Time

16

14

12

10

8

6

4

2

0

PA81 PA82 PA83 PA91

0 1000 2000 3000 4000 5000

Time (s)

Figure A3. Pressure v Time

41

Table A3. PHI-TEC data for hydrolysis of acetic anhydride without surfactant with a starting temperature of 323K (50°C), run No PA81.

Time (s) (°C) (kPa)

dT/dt (°C/s)

dP/dt (kPa/s)

0 49.6 112 0 51.6 113 52.3 113 -0.013 53 113 0.02

53.7 114 54.5 115 0.02 -0.005 55.3 116 0.02 56.2 117 57.1 117 -0.008 58 123

59.2 128 60.1 129 61.1 129 0.04 62.1 130 63.2 132 64.5 133 0.05 65.8 135 67.3 137 68.9 139 0.07 70.8 141 72.9 145 74.3 147 -0.375 75.9 148 0.11 77.6 151 79.5 155 81.7 159 84.2 163 0.48 86 167 -0.046

88.1 169 90.5 174 93.2 180

885 96.5 186 98.4 191 0.38

195 0.44 0.83 200 207 214 1.7 218 1.58 221 221 -1.382 225 235 3

120 241 3.4 248 1.27 256 1.45 265

Temperature Pressure

0.082 125.1 0.017 0.008 165.3 0.018 205.4 0.019 246.4 0.019 0.049 287.3 328.3 0.032 369.2 0.021 0.001 410.2 0.022 450.4 0.025 0.325 491.4 0.034 0.036 516.5 0.037 0.006 541.8 0.042 566.9 0.043 0.051 592.2 0.047 0.042 617.4 0.052 642.6 0.056 0.084 667.8 0.063 0.115 692.9 0.046 718.2 0.079 0.132 743.4 0.089 0.177 758.4 0.104 773.5 0.089 788.7 0.121 0.358 803.9 0.129 0.007 819.1 0.154 0.662 834.2 0.178 844.4 0.193 854.5 0.218 0.292 864.7 0.245 0.287 874.9 0.285 0.683

0.338 1.205 890.2 0.748 895.4 100.5 900.6 102.9 0.483 0.753 905.7 105.6 0.577 1.145 910.9 108.7 0.673 913.1 110.3 0.735 915.2 111.9 0.775 1.272 917.4 113.7 0.852 919.6 115.6 0.925 6.083 921.7 117.7 1.023 923.9 1.127 926.1 122.6 3.133 928.2 125.5 3.733 930.4 128.8 1.622 5.183

42


dT/dt (°C/s)

dP/dt (kPa/s)

277 1.9 281 285 2.1 290 294 2.3 7.2 299 305

143 312 320 2.75 330 340 353 3.25

153 368 384 25 401 3.55 422 32 442 464 3.6 487 512 39.5 539 4.3 42.5

176 568 43.5 597 4.1 45 627 3.9 658 689 721 753 784 45 814 845 3 873 42.5 901 929 954 2.3 978 2.2 35

1001 2 209 1022 1.75

1042 1099 23 1144 0.98 1182

219 1212 1265 0.33 1322 4.1

224 1349 0.23


932.6 132.6 5.717 933.2 133.9 1.983 5.917 933.9 135.2 6.567 934.6 136.7 2.217 6.883 935.2 138.1 935.9 139.7 2.417 8.067 936.6 141.4 2.533 9.183 937.2 2.633 11.133 937.9 144.8 14.05 938.6 146.7 2.833 15.783 939.2 148.6 3.117 17.833 939.9 150.7 21.167 940.6 3.317 22.833 941.2 155.2 3.583 941.9 157.6 28.167 942.6 160.1 3.867 943.2 162.5 3.933 32.833 943.9 165.2 32.667 944.6 167.7 3.933 35.833 945.2 170.3 4.117 945.9 173.1 946.6 4.267 947.2 178.7 947.9 181.5 46.667 948.6 184.2 4.133 46.667 949.2 186.7 3.683 46.333 949.9 189.1 3.233 47.833 950.6 191.3 3.333 47.667 951.2 193.5 3.267 951.9 195.6 3.133 45.833 952.6 197.7 44.667 953.2 199.6 2.833 953.9 201.5 2.617 41.333 954.6 203.2 2.433 38.833 955.2 204.8 37.833 955.9 206.3 956.6 207.7 32.167 957.2 30.333 957.9 210.1 1.638 29.667 960.1 213.3 1.288 962.2 215.7 18.833 964.4 217.6 0.758 15.517 966.6 0.573 12.383 971.7 221.3 8.217 981.9 223.3 0.102 997.1 0.007

43

Table A4.


dT/dt (°C/s)

dP/dt (kPa/s)

0 49.5 120 51.6 50.3 122 92.4 51 122

51.7 123 52.3 124 53 124

53.7 120 -0.287 54.4 116 55.2 116 56 117 0.02

56.9 119 57.8 119 0 59 120

60.4 121 61.3 122 62.3 124

620 63.4 125 64.6 127 0.05 65.9 128 0.04 67.4 130 0.06 69 132

746 70.7 134 72.8 137 0.11 74.1 139 0.08 75.6 142 77.2 144 79 147 81 150

83.2 155 -0.044 85.8 158 0.18 87.8 162 90 166

92.4 171 95.2 177 98.5 183 0.35

188 193 197 203 211 1.64

114 220 225 229 234 0.94 240 246 1.06 3

PHI-TEC data for hydrolysis of acetic anhydride with a starting temperature of 323K (50°C) and 1% surfactant, run No PA82.


0.017 0.177 0.016 0.018 0.016 0.016

135.1 0.016 0.018 175.8 0.016 0.011 217.2 0.016 0.005 258.1 0.017 298.7 0.018 0.039 339.6 0.021 0.023 380.6 0.016 421.5 0.022 0.066 462.5 0.025 503.4 0.031 0.024 544.4 0.036 0.046 569.6 0.039 0.051 594.8 0.043 0.045

0.045 0.054 645.2 0.095 670.4 0.055 695.6 0.093 720.8 0.067 0.065

0.076 0.129 771.2 0.086 786.9 0.091 802.1 0.091 0.088 817.2 0.115 0.053 832.4 0.118 0.407 847.6 0.142 0.039 862.8 0.157 877.9 0.407 888.1 0.175 0.222 898.3 0.215 0.347 908.4 0.245 0.507 918.6 0.297 0.905 928.8 1.043 933.9 100.4 0.387 0.532 939.1 102.5 0.433 1.005 944.2 104.8 0.475 0.785 949.4 107.5 0.547 1.417 954.6 110.5 0.622 959.8 0.735 2.267 961.9 115.7 0.797 2.017 964.1 117.5 0.857 2.133 966.2 119.4 2.733 968.4 121.5 1.022 2.667 970.6 123.8

44


dT/dt (°C/s)

dP/dt (kPa/s)

253 1.22 129 262 4.4

272 283 4.8 295 6.65 300 1.9 305 310 316 324 332 341 351 361 372 17.5 385 19.5 398 2.8 412 2.75

161 427 442 3 459 477 496 3.05 510 29.5 529 3.1 550 31 571 3.2 593 33.5 616 3.05 34 639 34.5 661 2.8 684 705 30 726 751 775 792 2.35 812 2.3 832 851 2.15 28.5 869 888 1.85 27 907 1.7 924 25.5 974

1019 1056 1087 1144 0.36


972.7 126.2 3.533 974.9 1.352 977.1 132.1 1.495 5.267 979.2 135.5 1.663 981.4 139.3 1.833 982.1 140.4 6.967 982.7 141.7 1.983 8.133 983.4 143.1 2.033 9.033 984.1 144.4 2.117 10.017 984.8 145.9 2.167 11.183 985.4 147.4 2.283 13.667 986.1 148.9 2.367 14.383 986.8 150.5 2.383 15.25 987.4 152.1 2.433 16.383 988.1 153.7 2.533 988.7 155.5 2.683 989.4 157.3 21.167 990.1 159.1 21.667 990.7 2.833 23.167 991.4 162.9 24.333 992.1 164.9 3.033 26.333 992.8 166.9 2.933 27.167 993.4 168.9 28.167 993.9 170.5 3.167 994.6 172.5 30.167 995.2 174.6 3.083 995.9 176.7 32.667 996.6 178.7 3.133 997.3 180.9 997.9 182.9 2.867 998.6 184.7 35.333 999.3 186.7 3.033 30.167 999.9 188.7 2.983

1000.6 190.6 2.733 39.167 1001.3 192.4 2.617 36.333 1001.9 194.1 2.483 33.833 1002.4 195.3 32.167 1003.1 196.8 30.167 1003.8 198.4 2.267 28.333 1004.4 199.9 1005.1 201.2 2.017 28.333 1005.8 202.6 1006.4 203.8 27.167 1007.1 204.9 1.575 1009.3 208.1 1.348 22.333 1011.4 210.6 1.038 18.333 1013.6 212.6 0.805 15.633 1015.8 214.2 0.643 13.883 1020.9 216.9 9.617

45


dT/dt (°C/s)

dP/dt (kPa/s)

1183 5.95 1227 3.05 1251 -0.009


1026.1 218.3 0.208 1036.3 219.6 0.115 1051.4 219.8 0.005

46

Table A5.


dT/dt (°C/s)

dP/dt (kPa/s)

0 28.5 116 0 0 41.5 28.7 115 -0.022

29.2 115 -0.018 29.6 115 -0.002 30.1 114 -0.001 30.6 114 31.1 114 -0.004 31.6 114 -0.002 32.2 114 32.7 114 33.3 114 0 33.8 114 34.4 115 35 115

35.6 115 36.3 116 0 36.6 115 37 116

37.3 116 37.7 116

1324 38.1 117 38.5 117 38.8 117 0.01 39.2 118 0.01 39.7 118 0.01 40.1 118 40.5 119 41 119

41.5 119 42.1 120 42.7 121 43.4 122 44.1 122 44.8 123 45.6 123 0.02 0.01 46.4 124 47.3 124 48.2 125 49.2 126

2105 50.3 127 2146 51.5 128

52.8 130 54.1 131 55 132 56 133

PHI-TEC data for hydrolysis of acetic anhydride with a starting temperature of 303K (30°C) and 1% surfactant, run No PA83.


0.006 119.6 0.006 196.4 0.006 273.4 0.006 350.3 0.006 0.002 427.2 0.007 504.1 0.007 580.9 0.007 0.004 657.9 0.007 0.001 734.9 0.007 811.7 0.008 0.004 888.4 0.008 0.005 965.6 0.008 0.005

1042.4 0.008 0.002 1119.4 0.008 1160.4 0.009 0.002 1201.4 0.009 0.009 1242.4 0.009 0.009 1283.4 0.009 0.008

0.009 0.006 1364.6 0.009 0.005 1405.6 0.012 1446.5 0.012 1487.5 0.011 1528.6 0.011 0.009 1569.6 0.011 0.004 1610.5 0.012 0.006 1651.4 0.013 0.012 1692.3 0.015 0.021 1733.3 0.015 0.021 1777.7 0.016 0.021 1818.7 0.017 0.009 1859.3 0.019 0.008 1900.3 1941.2 0.021 0.011 1982.2 0.022 0.017 2023.1 0.024 0.008 2064.1 0.025 0.033

0.027 0.011 0.029 0.039

2186.9 0.032 0.021 2227.9 0.035 0.029 2253.1 0.037 0.033 2278.3 0.038 0.046

47


dT/dt (°C/s)

dP/dt (kPa/s)

57 134 58.1 135 59.3 136 60.5 138 61.9 139 63.4 141 65 143

66.7 145 68.7 147 70.8 150 72.3 153 73.8 155 -0.002 75.5 157 77.4 160 0.13 79.5 163 81.8 166 84.4 171 86.4 174 -0.07 88.5 177 0.27 91 182 0.19

93.6 187 96.7 193 98.4 197

202 206 0.52 210

107 217 224 231 240 244 0.76 248 253 259 265 0.95 270 1

2742 277 1.04 3.95 286 294 1.17 303 313 324 336 6.85 351 366 383 8.95 403 11.2 426 433 1.7


2303.5 0.043 0.066 2328.7 0.045 0.057 2353.9 0.048 0.026 2379.1 0.051 0.057 2404.3 0.056 0.052 2429.5 0.061 0.062 2454.7 0.067 0.075 2479.9 0.073 0.078 2505.1 0.082 0.114 2530.3 0.089 0.082 2545.3 0.101 0.225 2560.5 0.104 2575.7 0.118 0.532 2590.9 0.248 2606.1 0.151 0.117 2621.3 0.161 0.492 2636.4 0.183 0.395 2646.6 0.197 2656.7 0.228 2666.8 0.505 2676.9 0.275 0.362 2687.1 0.337 1.213 2692.1 0.343 1.022 2697.2 100.3 0.383 1.137 2702.3 102.1 1.023 2707.3 104.5 0.462 0.542 2712.4 0.505 1.638 2717.4 109.7 0.567 1.482 2722.5 112.7 0.615 1.598 2727.6 116.1 0.713 2.417 2729.6 117.6 1.967 2731.7 119.2 0.795 2.083 2733.8 120.9 0.843 2.633 2735.8 122.7 0.897 2.917 2737.9 124.6 2.733 2739.9 126.6 2.633

128.7 2744.1 130.9 1.095 4.133 2746.1 133.2 4.133 2748.2 135.7 1.258 4.467 2750.3 138.4 1.352 5.183 2752.3 141.2 1.397 5.633 2754.4 144.2 1.448 2756.4 147.2 1.478 7.183 2758.5 150.4 1.633 7.367 2760.6 153.8 1.663 2762.6 157.2 1.637 2764.7 160.6 1.667 11.383 2765.3 161.5 11.45

48


dT/dt (°C/s)

dP/dt (kPa/s)

439 1.7 446 1.7 454 462 470 499 530 562 15.1 594 626 656 687 14.5

189 716 744 0.86 12.7 769 0.76

194 792 0.68 813

198 859 888 942 0.18 3.3 972 2.5 997

1012 1015 -0.067

206 1008 -0.003 -0.067


2765.8 162.5 12.317 2766.4 163.4 13.317 2766.9 164.4 1.683 13.433 2767.5 165.3 1.667 14.033 2768.1 166.3 1.657 14.517 2770.1 169.7 1.638 14.633 2772.2 172.9 1.565 15.167 2774.3 176.1 1.502 2776.3 179.2 1.395 15.983 2778.4 181.9 1.305 14.95 2780.4 184.5 1.192 14.817 2782.5 186.9 1.078 2784.6 0.945 14.133 2786.6 190.8 2788.7 192.5 12.25 2790.8 9.733 2792.8 195.3 0.597 10.417 2797.9 0.427 7.817 2802.9 199.9 0.322 10.767 2813.1 202.4 2823.2 203.8 0.096 2838.4 204.9 0.052 1.023 2863.5 205.7 0.012 0.018 2904.2 206.1 0.006 2996.9

49

Table A6.


dT/dt (°C/s)

dP/dt (kPa/s)

0 25.4 102 0 0 18.3 25.5 102 -0.037 60.9 25.9 101 -0.004

26.5 100 -0.016 27 99 0

27.4 99 -0.004 27.9 98 -0.006 28.4 98 -0.003

520 28.8 98 -0.002 29.2 98 -0.001 29.7 98 0 30.2 98 -0.001 30.7 98

907 31.2 98 31.7 98 32.2 98 -0.002 32.7 98 -0.001 33.3 99 33.8 99 34.4 99 35 99

35.6 100 36.2 101 36.9 101 37.3 102

1762 37.7 102 38 102

1847 38.4 102 0.01 -0.002 38.9 102 0.01 39.3 103 39.8 103 40.3 103 40.9 104

2091 41.5 104 42.1 105 -0.001 42.8 105 43.5 106

2255 44.2 106 2296 45 107 0.03

45.8 108 46.6 108 -0.004 47.6 109 48.5 110 49.6 110 0.03 50.7 111

PHI-TEC data for hydrolysis of acetic anhydride with a starting temperature of 298K (25°C) and 0.3% surfactant, run No PA91.


0.009 0.008

137.6 0.007 214.8 0.006 290.6 0.006 366.7 0.006 443.7 0.006

0.006 596.4 0.006 675.3 0.006 751.4 0.006 830.4 0.006 0.001

0.006 0.006 985.4 0.007 0.005

1062.7 0.007 1137.9 0.007 1214.6 0.007 0.004 1294.6 0.007 0.005 1373.2 0.008 0.001 1449.7 0.008 0.005 1524.9 0.008 0.014 1600.3 0.008 0.009 1676.8 0.009 0.001 1720.2 0.009 0.005

0.009 0.008 1802.7 0.009 0.009

1888.1 0.003 1928.3 0.011 0.007 1968.5 0.013 0.004 2008.9 0.014 0.018 2050.3 0.014 0.019

0.016 0.004 2133.1 0.015 2174.1 0.017 0.016 2214.4 0.018 0.022

0.018 0.001 0.019

2336.9 0.021 0.021 2377.2 0.022 2417.6 0.023 0.026 2457.7 0.025 0.007 2498.7 0.027 2539.7 0.029 0.002

50


dT/dt (°C/s)

dP/dt (kPa/s)

51.9 112 0.04 53.3 113 54.2 114 55.1 116 56.2 117 57.2 118 58.4 118 0.04 59.6 120 0.05 61 121

62.4 122 64 124

65.7 125 -0.016 67.6 126 69.7 130 0.25 71.1 131 0.07 72.6 134

2976 74.2 135 0.12 -0.42 2991 76 136

78 139 80.2 143 0.44 82.7 145 0.37 84.6 148 0.21 86.7 152 -0.092 89 155

91.8 159 0.3 0.6 94.8 164 96.6 167 98.6 170

3102 173 0.43 179 184 0.58 1.7 191 0.65 198 202 207 211 213 222

3135 229 3.6 244 250 256 259 5.1 262 266 1.85 6.25 270 6.6

140 274 1.95 7.75 279 1.9 284


2580.7 0.032 2621.7 0.034 0.043 2647.7 0.036 0.045 2673.7 0.038 0.054 2699.7 0.041 0.028 2725.3 0.041 0.031 2751.3 0.046 2777.3 0.019 2803.3 0.054 0.052 2829.2 0.059 0.085 2854.3 0.065 0.053 2880.3 0.071 2905.7 0.078 0.222 2930.7 0.087 2945.8 0.093 2960.9 0.103 0.437

0.112 0.373 3006.1 0.132 0.378 3021.2 0.147 3036.3 0.178 3046.4 0.093 3056.5 0.218 3066.5 0.258 0.385 3076.6 3086.7 0.347 0.808 3091.8 0.352 0.723 3096.9 0.408 0.722

100.7 0.917 3107.1 103.2 0.507 1.015 3112.2 105.9 3117.3 109.1 1.045 3122.4 112.8 0.785 1.867 3124.5 114.4 0.827 2.017 3126.6 116.3 0.903 2.583 3128.7 118.2 0.972 1.867 3130.8 120.4 1.053 3.817 3132.9 122.7 1.172 3.383

125.3 1.272 3137.1 128.1 1.395 14.283 3139.2 131.2 1.545 0.685 3141.3 134.5 1.683 4.633 3141.9 135.5 1.733 3142.5 136.6 1.783 5.583 3143.1 137.7 3143.7 138.8 1.917 3144.3 3144.9 141.1 8.067 3145.5 142.3 2.067 8.467

51


dT/dt (°C/s)

dP/dt (kPa/s)

3146 288 294 2.1 300 2.15 307 314

3149 322 2.2 13.4 330 338 348 2.35 357

3152 367 2.35 17.5 377 387 2.4

161 399 2.35 410

3155 421 19 165 431 2.35

443 2.3 19.5 454 466 2.25 20.5 479 2.2 493 2.2 22.5 505 517 2.1 19.5 528 19 540 20

178 552 564 1.8 575 1.75 585 1.7 596

3166 633 666 695 1 721 743 764 802 6.15 834 873 896


143.3 2.083 9.517 3146.6 144.6 10.067 3147.2 145.9 11.583 3147.8 147.2 2.217 11.633 3148.4 148.5 2.217 12.967

149.8 3149.6 151.1 2.267 13.217 3150.2 152.5 2.317 14.483 3150.8 153.9 16.417 3151.4 155.3 2.367 14.967

156.7 3152.6 158.2 2.383 16.367 3153.2 159.6 19.833 3153.8 18.833 3154.4 162.4 2.367 18.667

163.8 2.367 3155.5 20.167 3156.1 166.4 3156.7 167.8 2.283 19.667 3157.3 169.2 3157.9 170.5 24.667 3158.5 171.8 3159.1 173.1 2.133 17.333 3159.7 174.4 3160.3 175.6 2.017 3160.9 176.8 1.933 3161.5 1.917 19.667 3162.1 179.1 19.167 3162.7 180.2 17.167 3163.3 181.2 17.833 3163.9 182.2 1.647 17.667

185.5 1.435 17.167 3168.1 188.2 1.207 14.15 3170.2 190.5 14.067 3172.3 192.5 0.835 11.217 3174.4 194.1 0.713 9.917 3176.5 195.5 0.573 9.433 3181.6 197.9 0.385 3186.7 199.5 0.275 5.333 3196.8 201.2 0.123 2.067 3212.2 202.1 0.008 0.937

52

12. APPENDIX B: DESIGN OF THE LARGE-SCALE FACILITY

12.1 DESIGN CONSIDERATIONS

An imperative for the validation of the models developed in WP7, 8 and 11 was the measurement of the axial variation of void fraction in the venting reactor. All efforts were made to make the design of the large-scale facility as flexible as possible in terms of its ability to achieve the requirements of the AWARD partners.

It was originally intended to measure void fraction entirely by the use of differential pressure (DP) measurement. JRC-ISIS carried out an analysis of experimental results from the EC CHEERS project and proposed a minimum height difference between adjacent DP measurements. It was concluded that this could be achieved and led to an increase in the proposed height of the reactor to the maximum possible within the constraints of the existing building which was to house it.

It was arranged that the reactor could be modified to increase its height by means of a middle section which could be bolted between the top and bottom sections. The height of the reactor (just over 3 metres) is similar to that of the reactor at CMR (which was used for CHEERS). The diameter (1 metre) is smaller so as to limit the volume to approximately 2.2 m3, for which it was feasible to provide a total containment system for the relief system. This was required to meet environmental protection standards during the experiments. The relief system was designed to vent to a 13m2 catch tank. A second-hand vessel with a suitable design pressure was procured for this purpose.

A dimensioned sketch of the reactor is shown in Figure B1 and a process and instrumentation diagram is shown in Figure C1 in Appendix C. Photographs of the reactor and dump tank are shown in Figures B2 and B3.

A 200mm diameter experimental relief system was provided. This can be opened using an actuated ball valve to simulate the operation of a relief device in a reproducible way. Provision was made for installation of an orifice plate in the experimental vent line.

The reactor can be heated to the temperature required to initiate a runaway reaction by circulating fluid through an external steam-heated heat exchanger. The use of a heating jacket was precluded by the need for access by instrumentation to the whole surface of the reactor.

Because the reactor is relatively tall and thin compared with the majority of industrial reactors, conventional agitation was not the best option. A specialist agitator manufacturer recommended the use of jet mixing, using the returned fluid from the circulation loop. Modelling obtained via this manufacturer indicated that mixing would be satisfactory. This was to be checked during commissioning and provision was made to install side-mounted or bottom-mounted agitation if necessary. However the jet mixing agitation was found to be satisfactory.

53

54

N1

N3

N4

N6

N2

VESSEL ELEVATION

N9 N7

N5

N8N10

TYP

TYP

TYP

5

5

3 TYP

2 TYP

TYP

TYP

DISHED HEAD: -1000mm INSIDE DIAMETER800mm SPHERICAL RADIUS150mm KNUCKLE RADIUS25mm STRAIGHT FLANGE 15mm THICK11.5mm MIN. THK A.F.

DISHED HEAD: -1000mm INSIDE DIAMETER800mm SPHERICAL RADIUS150mm KNUCKLE RADIUS25mm STRAIGHT FLANGE 10mm THICK8mm MIN. THK A.F.

Figure B1 Dimensioned sketch of reactor

Figure B2. Large-scale reactor

Figure B3. Dump tank (also showing vent lines and middle section of reactor before its installation)

55

12.2 INSTRUMENTATION

12.2.1 Void fraction

Measurement of the axial variation of void fraction in the venting reactor was an imperative for the project. The reactor was designed to measure this by means of differential pressure. However, initial commissioning showed that the measurements were not accurate enough for the purposes of the project. Very sensitive measurement of differential pressure was required to derive estimates of the void fraction, particularly towards the top of the reactor where high void fraction and low differential pressure were expected. Because of the need for high sensitivity, differential pressure sensors using oil-filled lines sealed with diaphragms had been procured. The alternative without diaphragms would risk contamination of the reacting mixture with oil and contamination of the oil with reacting mixture; both of these could adversely affect results. The differential pressure transducers obtained were the most sensitive that could be sourced for this application. Serious problems were identified during commissioning. Firstly the level of noise was found to be very high in comparison with the signal, even when commissioning using much larger heights than would be necessary during the large-scale experiments. Secondly, and perhaps more seriously, it became apparent that corrections would need to be made to the results to account for the temperature differences between the reactor contents and the oil in the oil-filled lines. It would be possible to develop empirical corrections for this in a static system, but it would be much more challenging for a reacting system whose temperature is changing significantly with time. These problems led to the conclusion that differential pressure was not going to be a viable option for obtaining the accuracy of measurement necessary for the experiments.

The possibility of developing a probe based on capacitance or impedance measurement for measurement of void fraction was investigated together with UMIST. This proved not to be a practicable option. Materials which had previously been used by UMIST for such probes were unsuitable for the temperatures expected in the experiments. There was also considerable technical uncertainty in developing this approach, which could not readily be solved within the cost constraints of the project.

It was decided to develop and use the novel technique of scanning gamma ray tomography as the primary measure of void fraction during the large-scale experiments. Two gamma ray densitometers were employed in the top and bottom sections of the reactor respectively. Mechanisms were developed to move the source holders so that each will scan through half the vessel height. Both sources are aligned with the centre of the reactor, but shifted to avoid interference between them. The sources are mounted on vertical rotating frames actuated with a pneumatic system which enables them either to scan the vessel or to be set in a fixed position. The alignment between the gamma sources and their detectors is achieved by the geometry of the detectors. Figure B4 is a sketch showing the general arrangement and a photograph showing the top source holder and its tilting mechanism is shown in Figure B5.

The gamma ray sources produce columnated beams, which are received by rod detectors at the far side of the reactor. The tilting mechanisms were synchronised such that both were in their top positions at the same time, and both in their bottom positions at the same time. The system has been designed for remote operation, such that people are excluded from the building when the system is in use. This allowed a practicable system to be developed as otherwise the amount of shielding required would have precluded use of a tilting mechanism.

The scanning gamma ray tomography system was calibrated during commissioning by means of adding known quantities of water to the reactor and comparing the resulting known level with output from the densitometers.

56

D1S2

S1 S1

S2

D1

D2

D2

Figure B4. Configuration of scanning gamma ray densitometer system showing sources (S) and detectors (D)

Figure B5. Upper source holder of scanning gamma ray densitometry system showing tilting mechanism and radiation shielding

57

DP 101

DP 102

2.95

1m

1.72

0m

1.51

3m

DP 103

Figure B6. Configuration of differential pressure cells

The scanning gamma densitometry system was supplemented by a small number of differential pressure measurements, using sensors with oil-filled lines and diaphragms. These were made over substantial heights so as to obtain meaningful results. Figure B6 provides a sketch of the general arrangement of the differential pressure measurement.

12.2.2 Other instrumentation

Other instrumentation included axial temperature measurements, pressure at the top of the reactor, and weight of the reactor and its contents on load cells positioned between reactor support brackets and the support frame.

58

13. APPENDIX C: LARGE-SCALE EXPERIMENTAL RESULTS

13.1 REACTOR FACILITY

The pilot scale chemical plant is used to investigate chemical reaction hazards, reactor venting etc. The High Pressure Reactor is located at the far end of the main process building. A schematic diagram of the system is shown in Figure C1. The reactor is equipped with two, glass, feed vessels. The stainless steel reactor vessel is designed to BS5500 with 12 barg maximum allowable working pressure and is fitted with an automatic valve and bursting discs which vent to a dump tank outside the building. The valve is linked to a pressure controller and is opened automatically when the pressure reaches a pre-selected value. The dump tank is fitted with a pressure relief valve operating at 7 barg. The vent line between the reactor and dump tank is fitted with a restricting orifice plate. The installation includes a pump and heat exchanger system for circulating and heating the vessel contents. Interconnecting chemical transfer pipe work is fitted with remotely operated actuated valves. The vessel is equipped with sight glasses and a range of transducers for measuring temperature, pressure, differential pressure and mass. Density of the reactor contents is indicated by two scanning densitometers, located towards the top and bottom of the vessel, respectively. The gamma ray densitometers may be synchronised, or move independently, at various scan rates. The positions of all transducers are shown in Table C1. The pilot plant can be controlled and monitored remotely from a control room 100 m from the reactor building. Also indicated on the figure are manual valves (MV), automatic valves (AV), temperature control transducers (TE), temperature controller outputs (TY), and temperature controllers (TC).

13.2 EXPERIMENTAL PROCEDURE

13.2.1 Hydrolysis of Acetic Anhydride

The acetic anhydride was charged to the reactor from a series of drums installed on a weigh scale. If required, surfactant solution was charged to the reactor via a small feed vessel (FV5). Water was charged to the 300L feed vessel (FV6) from a drum installed on the weigh scale. The circulation pump was started and the acetic anhydride pre-heated by means of the steam/hot water and heat exchanger systems. Once the required initial temperature of 50°C was reached, the water was remotely charged to the reactor. The temperature and pressure in the reactor were monitored during the course of the runaway reaction. When a pre-selected relief set pressure was achieved, the relief valve between the reactor and dump tank was automatically opened. Video recording was used to observe the two-phase discharge from the reactor to the dump tank.

The relief valve set pressure was 200 kPa (2 bara) for all experiments. The conditions of the experiments and a summary of the main results are given in Table C2. Table C3 gives the mass transfer results. Graphs showing transducer reading versus time histories for key transducers in the reactor, dump tank and vent line are given for each experiment as Figures C2 to C37.

13.2.2 Acetic Acid Blowdown Tests

Vented reaction product (acetic acid) from hydrolysis experiment HP3 was transferred back into the reactor vessel. If required, surfactant solution was charged to the reactor via a small feed vessel (FV5). The circulation pump was started and the acetic acid was heated by means

59

of the steam and heat exchanger systems. When a pre-selected relief set pressure was achieved, the relief valve between the reactor and dump tank was automatically opened. Video recording was used to observe the discharge from the reactor to the dump tank.

The reactor relief valve set pressure was 480 kPa (4.8 bara) for both experiments and the dump tank was unvented. Graphs showing calculated void fractions versus time are given for each experiment as Figures C38 and C39.

60

Figure C1. Large scale reactor instrumentation diagram

61

Table C1. Key to high-pressure reactor instrumentation diagram

SENSOR DESCRIPTION LOCATION TE101 Reactor TE102 Reactor TE103 Reactor TE104 Reactor TE105 Reactor TE106 Reactor TE109 TE110 TE111 TE112 TE114 Vent line (inside)

TE115 Vent line (before bend)

TE116 Vent line (after bend)

TE117 Vent line

TE118 TE119 TE120 TE121 HX utility top TE122 HX utility bottom PT102 Manifold N9 PT103 PT104 After HX PT105 Vent line (inside)

PT106 Vent line (before bend)

PT107 Vent line (after bend)

PT108 Vent line

PT109 PT110 Before HX WT101 Load cells FT101 Flow Meter After HX DT101 DT102 DT103 Pos. Conf. 1 DT104 Pos. Conf. 2 DT105 Detector 1 DT106 Detector 2 AV122 Valve State 8" Reactor valve AV121 Valve State

Thermocouple Thermocouple Thermocouple Thermocouple Thermocouple Thermocouple

Resistance Temperature Device HX process inlet Resistance Temperature Device HX process outlet Resistance Temperature Device Water feed vessel

Thermocouple Feed Vessel 6 Thermocouple

Thermocouple

Thermocouple

Thermocouple (above dump tank) Thermocouple Dump tank top Thermocouple Dump tank bottom

Resistance Temperature Device Pump inlet Resistance Temperature Device Resistance Temperature Device

Pressure Transducer Pressure Transducer Before pump Pressure Transducer Pressure Transducer

Pressure Transducer

Pressure Transducer

Pressure Transducer (above dump tank) Pressure Transducer Dump tank Pressure Transducer

Vessel load cell

Densitometer Position reqst. 1 Densitometer Position reqst. 2 Densitometer Densitometer Densitometer Densitometer

Feed Vessel valve

62

13.3 EXPERIMENTAL RESULTS

Table C2. Large scale experimental conditions and main results

Experiment Number HP1 HP2 HP3 HP4 HP5 HP6

50 70 60 50 70 60 0 0 0 0 0 0

)

PT102 )

-1)

-1)

-1) )

TE106 (K) )

-1)

-1)

(K s-1)

)

) (s)

Critical Variables Post-mixing Temp. (K) 311.3 312.8 309.5 312.0 311.3 313.1 Fill Level (%) Surfactant mass (kg) 3.14 4.44 3.77 Surfactant (% wt/wt) 0.25 0.25 0.25 Set Press. (kPa 200 200 200 200 200 200

Reactor Max. Press. (kPa) 360.6 677.9 604.9 576.4 771.6 719.2

Overpress. (kPa 160.6 477.9 404.9 376.4 571.6 519.2 Overpress. (abs) % 44.5 70.5 66.9 65.3 74.1 72.2 Max. Press Rate before vent op. (kPa s 6.94 7.71 11.07 8.12 11.54 11.79

Max. Press Rate after vent op. (kPa s 12.89 33.0 37.03 32.83 48.47 48.04

Press Rate at vent op. (kPa s 11.11 12.84 15.87 12.96 12.95 12.41

Max Temp. TE106 (K 439.1 466.6 458.5 455.6 470.3 466.6 Temp. at vent op 403.2 401.8 401.4 402.9 397.7 401.7

Overtemp. TE106 (K 35.9 64.8 57.1 52.7 72.6 64.9 Max temp. rate before vent op. (K s 1.86 1.81 2.96 2.40 1.62 2.53

Max temp. rate after vent op. (K s 1.98 4.37 4.50 3.67 7.16 5.62

Temp. rate at vent op. 1.71 2.60 1.46 1.95 2.99 1.8

Times from initiation Vent op. (s 1594.8 1604.2 1627.6 1509.1 1627.4 1443.0 Max press. (s) 1637.0 1628.2 1648.9 1528.1 1647.2 1462.5 Max temp. (TE106 1645.7 1629.7 1649.9 1532.4 1648.9 1463.5

63

Table C2. (continued)


50 70 60 50 70 60 0 0 0 0 0 0

)

op.

) (s)

Vent Line )

()

)

) (

(K)

)

)


Times relative to vent

Max press. (s) 42.2 24.0 21.3 19.0 19.8 19.5 Max temp. (TE106 51.9 25.5 22.3 23.3 21.5 20.5

Max press. rate (s) 4.8 9.9 7.8 7.7 11.4 9.1 Max temp. rate (s) 1.7 9.0 8.0 7.4 10.4 9.6

Max. press (kPaPT108) 359.0 301.8 479.6 412.9 301.8 306.6

Max. temp. (K(TE114 436.2 431.1 449.0 442.1 431.1 431.1

Dump Tank Max. press (kPaPT109) 357.4 301.1 478.3 411.2 299.7 306.4

Max. temp. of material entering.TE117 435.5 430.7 449.0 441.7 430.7 431.1

Max. temp. of liquid (K) (TE119 412.4 399.9 428.9 420.4 406.5 403.6

Max. vapour space temp.TE118 (K 410.2 398.8 427.8 419.3 404.7 402.5

64

Table C3. Pilot scale mass transfer results


50 70 60 50 70 60 0 0 0 0 0 0

)

reactor (kg)

) (1)

) (2)

((1)/(2))

(s)

)


Charge Mass (kg) 1266.1 1777.6 1520.6 1258.5 1779.2 1512.7 Mass remaining in 883.2 624.9 774.9 351.0 357.4 349.3

Mass remaining in dump tank (kg 382.9 1152.7 745.7 907.5 1421.8 1163.4

Duration of reactor venting (s 74.5 62.0 49.4 33.4 54.9 46.1

Av. mass discharge rate (kg s-1) 5.1 18.6 15.1 27.2 25.9 25.2

Time from vent opening to level swell 1.8 0.3 1.6 2.1 1.4 1.9

Duration of two-phase flow (s 26.2 47.7 32.6 29.4 35.5 44.3

65

Experiment HP1

CRITICAL VARIABLES

Relief set pressure 200 (2) kPa (bara)

Orifice diameter 100 mm

Fill level 50 %

Surfactant concentration 0 % (wt/wt)

Experiment No: HP1 Date: 10 February 2005 File: HP1ev.opj 450

400

350

300

)Te

mpe

ratu

re (K

Heat exchanger bottom (TE122)

Heat exchanger top (TE121)

Liquid (TE106)

Vapour (TE101)

-2000 -1000 0 1000 2000 3000

Time (s)

Figure C2. Overall reactor temperature records

66

Experiment No: HP1 Date: 10 February 2005 File: HP1ev.opj

100

200

300

400

250

300

350

400

450

)

(

500 1000 1500 2500

Vent open

Pressure (PT102)

Liquid (TE106)

Tem

pera

ture

(K

Vapour (TE101)

Pre

ssur

e kP

a)

2000

Time (s)

Figure C3. Temperature and pressure profiles in the reactor during venting


100

200

300

400

10

12

14

16

)

(

Dump tank pressure (PT109)

Vent open

Reactor pressure (PT102)

Reactor top to bottom differential pressure (DPT101)

Diff

eren

tial p

ress

ure

(kPa

Pre

ssur

e kP

a)

1600 1700

Time (s)

Figure C4. Pressure profile in the reactor and dump tank during venting

67


0

20000

40000

60000

80000

Vent open Gamma ray detector 1 (DT105)

Den

sito

met

er (c

ps)

1500 1550 1600 1650 1700

Time (s)

Figure C5. Density profile in the reactor during venting


0

100

200

300

400

300

350

400

)

i l ifi

)

Outlet temp. (TE117)

Inlet pressure (PT105)

Vent open

Outlet pressure (PT108)

Inlet temp. (TE114)

Tem

pera

ture

(K

Note: Inlet nstrumentation ocated downstream of or ce plate.

Pre

ssur

e (k

Pa

1000 1500 2000 2500

Time (s)

Figure C6. Vent line temperature and pressure profiles during venting

68


100

200

300

400

250

300

350

400

450

Di

)

(

Liquid temp. (TE119)

scharge from vent line (TE117)

Pressure (PT109)

Vent open

Vapour temp. (TE118)

Tem

pera

ture

(K

Pre

ssur

e kP

a)

1000 1500 2000 2500

Time (s)

Figure C7. Dump tank temperature and pressure profiles during venting

69

70

Experiment HP2

CRITICAL VARIABLES



Fill level 70 %



0 1000 2000 3000

300

360

420

480

Heat exchangerbottom (TE122)

Heat ex. top (TE121)

Vapour (TE101)


Tem

pera

ture

(K)

Time (s)

Liquid (TE106)


275

300

325

350

375

400

425

450

475

0

100

200

300

400

500

600

700

(

(

)

)

1000 2000

Vent open

Pressure PT102)

Vapour TE101)

Tem

pera

ture

(K

Liquid (TE106) Pre

ssur

e (k

Pa

1500

Time (s)



6

12

18

24

0

100

200

300

400

500

600

700

((

)

(


Vent open

Reactor pressure PT102)

Reactor top to bottom differential pressure DPT101)

Diff

eren

tial P

ress

ure

(kPa

Pre

ssur

e kP

a)

1550 1600 1650 1700

Time (s)


71


0

20000

40000

60000

80000

( Gamma ray detector 1 DT105)

Vent open

Den

sito

met

er (c

ps)

1500 1550 1600 1650 1700 1750

Time (s)



275

300

325

350

375

400

425

450

50

100

150

200

250

300 (

)

l ifi late.

)


Vent open


Inlet temp. TE114)

Tem

pera

ture

(K

Outlet temp. (TE117)

Note: Inlet instrumentation ocated downstream of or ce p

Pre

ssur

e (k

Pa

1000 1500 2000

Time (s)


72


275

300

325

350

375

400

425

450

100

150

200

250

300

(

(

(

)

Di li

)

Vapour temp. TE118)

Pressure PT109)

Vent open

Liquid temp. TE119)

Tem

pera

ture

(K

scharge from vent ne (TE117)

Pre

ssur

e (k

Pa

1000 1500 2000

Time (s)


73

74

Experiment HP3

CRITICAL VARIABLES



Fill level 60 %



0 1000 2000 3000

300

350

400

450

Liquid (TE105)

Vapour(TE101)

Heat exchangertop (TE121)


Experiment No: HP3 Date: 03 March 2005 File: HP3ev.opj

Tem

pera

ture

(K)

Time (s)


100

200

300

400

500

600

)

300

350

400

450

Tem

pera

ture

(K)

Vent open

Pressure (PT102)

Vapour (TE101)

Liquid (TE105)

Pre

ssur

e (k

Pa

1000 1500 2000 2500

Time (s)



8

10

12

14

16

18

20

Di

i (

)

)

)

ffere

nt a

l pre

ssur

e kP

a

200

400

600

Dump tank pressure (PT109

Vent open

Reactor pressure (PT102)

Reactor top to bottom differential pressure (DPT101)

Pre

ssur

e (k

Pa

1600 1625 1650 1675 1700

Time (s)


75


0

20000

40000

60000

80000

Gamma ray detector 1 (DT105)

Vent open

Den

sito

met

er (c

ps)

1500 1550 1600 1650 1700 1750

Time (s)



100

200

300

400

500

ifi l( )

)

250

300

350

400

450

Tem

pera

ture

(K)

Note: Inlet instrumentation located downstream of or ce p ate.


Vent open


Outlet temp. TE117

Inlet temp. (TE114 P

ress

ure

(kP

a)

1000 1500 2000 2500

Time (s)


76


100

200

300

400

500

Li

Di )

)

250

300

350

400

450

Tem

pera

ture

(K) quid temp.

(TE119)

Pressure (PT109)

Vent open


scharge from vent line (TE117

Pre

ssur

e (k

Pa

1000 1500 2000 2500

Time (s)


77

78

Experiment HP4

CRITICAL VARIABLES



Fill level 50 %

Surfactant concentration 0.25 % (wt/wt)


-500 0 500 1000 1500 2000 2500

300

350

400

450

Vapour (TE101)

Liquid (TE106)

Heat ex. top(TE121)



Tem

pera

ture

(K)

Time (s)


)

)

280

300

320

340

360

380

400

420

440

460

Pressure (PT102

Vent open

Vapour (TE101)

Liquid (TE106)

Tem

pera

ture

(K

100

200

300

400

500

600

Pre

ssur

e (k

Pa)

1000 1500 2000

Time (s)



5

10

15

di

i )

100

200

300

400

500

600

)


Reactor pressure (PT102) Vent open

Reactor top to bottom fferential pressure

(DPT101)

Diff

eren

t al P

ress

ure

(kP

a

Pre

ssur

e (k

Pa

1500 1530 1560

Time (s)


79


0

20000

40000

60000

80000

(

Vent open

Gamma ray detector 1 DT105)

Den

sito

met

er (c

ps)

1400 1450 1500 1550 1600

Time (s)



Outlet

l )

)

)

)

0 ifi280

300

320

340

360

380

400

420

440

460

pressure (PT108)

In et pressure (PT105

Vent open Inlet temp. (TE114

Outlet temp. (TE117

Tem

pera

ture

(K

100

200

300

400

Note: Inlet instrumentation located downstream of or ce plate.

Pre

ssur

e (k

Pa)

1000 1500 2000

Time (s)


80


Di li )

)

)

)

ne

280

300

320

340

360

380

400

420

440

460

scharge from vent (TE117

Pressure (PT109

Vent open


Liquid temp. (TE119

Tem

pera

ture

(K

100

200

300

400

500

Pre

ssur

e (k

Pa)

1250 1500 1750 2000 2250

Time (s)


81

Experiment HP5

CRITICAL VARIABLES



Fill level 70 %



)

480

460

440

420

400

380

360

340

320

300

280

)

(

((

Tem

pera

ture

(K

Liquid (TE106

Vapour TE101)

Heat ex. top TE121)

Heat exchanger bottom TE122)

0 500 1000 1500 2000 2500

Time (s)


82


)

0

(

280

300

320

340

360

380

400

420

440

460

480

Tem

pera

ture

(K

100

200

300

400

500

600

700

800 Vent open

Pressure (PT102)

Vapour TE101)

Liquid (TE106)

Pre

ssur

e (k

Pa)

1000 1500 2000

Time (s)



5

10

15

20

0

300

600

)

)

di

)

Diff

eren

tial p

ress

ure

(kP

a)

Dump tank pressure (PT109

Vent open

Reactor pressure (PT102

Reactor top to bottom fferential pressure

(DPT101)

Pre

ssur

e (k

Pa

1580 1600 1620 1640 1660 1680 1700 1720

Time (s)


83


0

20000

40000

60000

80000


Vent open

Den

sito

met

er (c

ps)

1550 1600 1650 1700 1750

Time (s)



)

50

l ifi

(

)

(

260

280

300

320

340

360

380

400

420

440

Tem

pera

ture

(K

100

150

200

250

300

350

Note: Inlet instrumentation ocated downstream of or ce plate.

Outlet pressure PT108)

Vent open

Inlet pressure (PT105 Outlet temp. TE117)

Inlet temp. (TE114)

Pre

ssur

e (k

Pa)

1000 1500 2000

Time (s)


84


)

Di ( )

)

)

)

260

280

300

320

340

360

380

400

420

440

Tem

pera

ture

(K

100

150

200

250

300

scharge from vent line TE117

Vent open Pressure (PT109

Vapour temp. (TE118

Liquid (TE119

Pre

ssur

e (k

Pa)

1500 2000

Time (s)


85

86

Experiment HP6

CRITICAL VARIABLES



Fill level 60 %



-1000 -500 0 500 1000 1500 2000 2500 3000

300

350

400

450

Vapour (TE101)

Liquid (TE106)


Experiment No: HP6 Date: 19 May 2005 File: HP6ev.opj

Tem

pera

ture

(K)

Time (s)

Heat exchanger top (TE121)


(

Vapour (TE101)

Liquid (TE106)

)

250

300

350

400

450

Vent open

Pressure PT102)

Tem

pera

ture

(K

200

400

600

800

Pre

ssur

e (k

Pa)

500 1000 1500 2000 2500

Time (s)



2

4

6

8

10

12

14

16

18

20

(DPT101)

Vent open

(

)

Reactor top to bottom differential pressure


Reactor pressure PT102)

Diff

eren

tial p

ress

ure

(kP

a)

200

400

600

800 P

ress

ure

(kP

a

1450 1500 1550

Time (s)


87


0

20000

40000

60000

80000

( )

Vent open

Den

sito

met

er c

ps


1400 1450 1500 1550

Time (s)



i l

(PT108)

Inl )

Vent open

Inlet

(PT105) (TE117)

)

300

350

400

450

Note: Inlet instrumentat on ocated downstream of orifice plate.

Outlet pressure

et temp. (TE114

pressure Outlet temp. Tem

pera

ture

(K

100

200

300

400 P

ress

ure

(kP

a)

1000 1250 1500 1750 2000 2250

Time (s)


88


Pressure )

Li

(TE117)

(TE118)

)

300

350

400

450

(PT109

quid temp. (TE119)

Vent open Discharge from vent line

Vapour temp.

Tem

pera

ture

(K

100

200

300

400

Pre

ssur

e (k

Pa)

1000 1250 1500 1750 2000 2250

Time (s)


89

Experiment blow1

CRITICAL VARIABLES

Relief set pressure

Orifice diameter


480 (4.8) kPa (bara)

100 mm

0 % (wt/wt)

0.0

0.2

0.4

0.6

0.8

1.0

Voi

Di l

lower source

d fra

ctio

n

fferentia pressure trans. top - halfway mid top - mid bottom top - bottom halfway - bottom

upper source

-80 -60 -40 -20 0 20 40 60 80 100 120

Time from vent opening (s)

Figure C38. Void fraction - acetic acid blow down

90

Experiment blow2

CRITICAL VARIABLES

Relief set pressure

Orifice diameter


480 (4.8) kPa (bara)

100 mm

0.25 % (wt/wt)

Di

Voi

d fra

ctio

n

1.0

0.8

0.6

0.4

0.2

0.0

upper source fferential pressure trans.

top - halfway top - bottom mid top - mid bottom bottom - halfway

lower source

-80 -60 -40 -20 0 20 40 60 80 100 120

Time from vent opening (s)

Figure C39. Void fraction - acetic acid blow down with surfactant

91

14. APPENDIX D: DERIVATION OF VOID FRACTIONS

14.1 GAMMA TOMOGRAPHY

The high-pressure reactor is equipped with two gamma sources with their respective detectors. One source is placed at the top of the reactor and the second mid-lower zone of the reactor. Both sources are aligned with the centre of the reactor, but shifted to avoid interferences amongst them. The sources are mounted on vertical rotating frames actuated with a pneumatic system which enables them either to scan the vessel or to be set in a fixed position. The alignment amongst the gamma sources and their detectors is achieved by the geometry of the detectors. A schematic diagram of the experimental set-up is given in figure B4 in Appendix B.

The attenuation of a monochromatic beam that passes through a section of path length L of a material of density ρ is described by the Lambert-Beer law equation:

ln⎜⎜⎛ I ⎞

⎟⎟ = −µ ⋅ L = −⎛ µ

⎟⎟⎞

⋅ ρ ⋅ L (1) ⎝ I 0 ⎠ ⎝

⎜⎜ ρ ⎠

where I and I are the measured and initial intensities, and µ and ⎛ µ ⎞

are the material’s0 ⎜⎜ ⎟⎟⎝ ρ ⎠

linear and mass attenuation coefficients, respectively. The mass attenuation coefficient is normally preferred to the linear coefficient because is independent of the temperature, pressure and state of the material.

In our system, the gamma radiation beam passes through three different types of materials: the air outside the reactor, the stainless steel of the reactor walls and the material inside of the reactor which may be vapour, liquid or a mixture of both.

For either mixtures or compounds, the linear attenuation coefficients may be calculated as a linear combination of the attenuation coefficients of the components:

µ mix = ∑ v ⋅ µ i (2)i i

⎛ ⎞ ⎞µ µ⎜ ⎟⎜ ⎟ = ∑ wi ⋅ ⎝

⎛

ρ ⎠⎟ i ⎜ ⎜ (3)⎝ ρ ⎠

⎟ mix i

where vi and wi are the volume and the mass fraction of the ith compound, respectively.

Introducing the assumed forms of the attenuation coefficients for mixtures in Equation (1) and regrouping the terms outside the reactor vessel as a background equivalent term, B, we obtain the following expression:

⎤⎛ I ⎞ ⎡ ⎛⎛ µ ⎞ ⎛ µ ⎞ln⎜⎜ ⎟⎟ = −⎢

⎢⎣

⎜⎜⎝⎝ ρ ⎟⎟

⎠ v

⋅ ρ ⋅ α + ⎜⎜ ρ ⎟⎟ ⋅ ρ f ⋅ ( 1 −α )⎟⎟

⎞⋅ LR +

⎛ µ ⎞⋅ ρ ⋅ LB ⎥ (4)

⎝ IO ⎠⎜⎜ v

⎝ ⎠ f ⎠ ⎝⎜⎜ ρ ⎠⎟

⎟B

B ⎥⎦

92

where α is the void fraction in the reaction vessel and LR and LB are the path lengths inside the vessel and through the background, respectively. The mass attenuation coefficients for the

vapour, ⎛ µ ⎞

, and liquid, ⎛ µ ⎞

, may be calculated according to their composition using⎜⎜ ⎟⎟ ⎜⎜⎝ ρ ⎟

⎟⎠ f⎝ ρ ⎠ V

Equation (3).

Analogous equations may be written for the vessel empty, Equation (5), and full of acetic acid, Equation (6):

⎛ I E ⎞ ⎛ µ ⎞ ⎛ µ ⎞Eln ⎝⎜⎜ I

⎟⎟ = −⎜⎜ ρ ⎟⎟ ⋅ ρ air ⋅ LR − ⎜⎜ ρ ⎟⎟ ⋅ ρ B ⋅ LB (5) o ⎠ ⎝ ⎠ air ⎝ ⎠ B

⎛ µ ⎞ ⎛ µ ⎞ln

⎛⎜⎜

I F ⎟⎟⎞

= −⎜⎜ ρ ⎟⎟ ⋅ ρ F ⋅ LR − ⎜⎜ ρ ⎟⎟ ⋅ ρ B ⋅ LB (6) ⎝ I o ⎠ ⎝ ⎠ acd

acd ⎝ ⎠ B

E Fwhere ρair is the density of air when the reactor is scanned empty and ρacd is the density of acetic acid when the reactor is scanned full.

Inserting equations (5) and (6) into (4) and rearranging the terms, we can relate the radiation intensity attenuation to the void fraction as:

⎛ I ⎞ln⎜⎜ ⎟⎟ ⎛ µ ⎞ ⎞

⎛ I E ⎟⎟

⋅⎝

⎜⎝ ⎠ air

⋅ ρ air − ⎜⎜ ρ ⎟⎟ ⋅ ρ acd ⎝ I F ⎠ ⎛ ⎛ µ ⎞ E ⎛ µ ⎞ F

⎞⎟ −

⎛⎜⎛ µ ⎞

⋅ ρ f − ⎜⎜ ρ ⎟⎟ ⋅ ρ F ⎟ ⎟⎞ ⎜⎜⎜ ρ ⎟⎟

⎝ ⎠ acd ⎠ ⎜⎝ ⎜⎜⎝ ρ ⎠

⎟⎟f ⎝ ⎠ acd

acd ⎟⎠ln⎜⎜

⎝ I F ⎠α = (7)⎛⎛ µ ⎞ ⎛ µ ⎞ ⎞⎜ ⎜⎜⎜ ρ ⎟⎟ ⋅ ρ − ⎜⎜ ρ ⎟⎟ ⋅ ρ f ⎟

⎟ v

⎝⎝ ⎠ v ⎝ ⎠ f ⎠

Assuming that the differences between densities and attenuation coefficients due to composition and temperature of liquids and gases are not significant, we obtain the following simplified equation:

⎛ I ⎞

α ln⎜⎜ ⎟⎟

sim ⎝ IF ⎠= (8)

⎞ln⎜⎜

⎛ IE ⎟⎟⎝ IF ⎠

93

14.2 DIFFERENTIAL PRESSURE CELLS

The high-pressure reactor is equipped with three sets of differential pressure cells designated as dP101, dP102 and dP103. A schematic diagram of the differential pressure cells configuration is displayed in Figure B6 in Appendix B.

The mean void fraction between the differential pressure cells is obtained as:

∆ Pρ f −

ρ⋅ α = h g

(9) f − ρv

where ∆ P denotes the differential pressure, g is the gravitational acceleration and h the height between the differential pressure cells.

With the configuration of the differential pressure cells in the high-pressure vessels is possible to obtain the mean void fraction at four different zones of the reactor. These zones are: - Overall void fraction of the reactor from dP101. - Top part of the reactor from dP103. - Middle section of the reactor from dP102. - Lower part of the reactor when dp103 is subtracted from dP101.

14.3 DENSITY

Liquid density is assumed to be a linear combination of the different compounds densities according to Equation (10) and to be dependent on temperature according to Equation (11):

ρ mix = ∑ wi ⋅ρ fi (10)i

1ρ =

⋅( −1

⋅( C T B )+ D )A

(11)

where wi is the weight fraction of component i; T is the temperature (in Kelvin); and A,B and C are constants in the correlation of liquid density with temperature.

The vapour phase is assumed to behave as an ideal gas, and therefore the vapour density may be obtained as:

⋅ M P vρ = ⋅ 1000 (12)v ⋅T R

where P denotes pressure in kPa, Mv the molecular weight and R the is the universal gas constant.

14.4 CONVERSION

Assuming that conversion is completed when the reactor temperature reaches its maximum and that conversion proceeds linearly as temperature increases, conversion is obtained as follows:

94

⎧ 0 tadd > t

Z = ⎪⎪ T − Tadd tadd ≤ t ≤ t (13)⎨ max ⎪ Tmax − Tadd ⎪⎩ 1 t > tmax

where Z is the fractional conversion; T is temperature and t is time; subscript ‘add’ refers to the addition of water to the acetic anhydride, i.e the start of the reaction and subscript ‘max’ to the maximum temperature or conversion.

95

96

15. APPENDIX E: LITERATURE PAPER

The following paper was accepted for publication at the IChemE Hazards XIX Symposium, Process Safety and Environmental Protection – What do we know? Where are we going?, Manchester, 28-30 March 2006.

97

LARGE-SCALE EVALUATION OF VENT-SIZING METHODOLOGY FOR VAPOUR-PRESSURE SYSTEMS

T J Snee*, J Bosch Pagans, L Cusco, F Gallice**, C Hoff**, D C Kerr, A Rovetta** and M Royle Health and Safety Laboratory, Buxton, SK17 9JN, UK *Corresponding author: [email protected] ** Sanofi - Aventis

Crown Copyright 2005. This article is published with permission of the Controller of HMSO and the Queen’s Printer for Scotland

A large-scale facility for investigating the performance of emergency pressure relief and disposal systems for chemical reactors has been constructed at the Health and Safety Laboratory as part of the EU AWARD Project. The facility has been used to investigate venting of the runaway reaction between water and acetic anhydride. Results are reported for a series of experiments in the 2,500 litre reactor over a range of batch volumes, along with data obtained previously using a 350 litre vessel. In some of these experiments, a small quantity of surfactant was added to the reaction mixture in order to change the void fraction distribution and investigate how this affects the maximum pressure in the reactor. Temperature, pressure, differential pressure and void-fraction measurements are used to establish the mechanisms which determine the maximum pressure and temperature in the reactor during pressure relief. This analysis is compared with the assumptions made in the most widely used design calculation methods for vapour pressure systems. Differences between the experimental results and current assumptions are identified. The safety implications of these discrepancies are explored.

INTRODUCTION A wide range of techniques are available for designing emergency vents for runaway reactions where the overpressure is due to elevated vapour pressure of reagents and products1. Many of the methods for vapour pressure systems were developed as part of the DIERS project2. Some companies rely on dynamic computer modelling to determine the size and operating conditions for the relief system but most vents are designed using simplified equations. Both approaches require chemical kinetic and physical property data for the reaction system and models to predict level swell in the reactor and the two-phase flow regimes in the reactor and vent line. The relationship between pressure, temperature and the rate of heat production for a reaction system can be determined reliably using adiabatic calorimetry. Prediction of the flow regimes and the onset of two-phase flow is more problematical. The simplified equations and the dynamic computer models both contain assumptions about level swell, the void fraction distribution in the reactor and the ratio of liquid to vapour entering the vent line. The validity of these assumptions determines whether the mechanism of pressure turnaround, implicit in a particular calculation method, is correct. For example, whether pressure turnaround is due fundamentally to tempering, vessel emptying or reactant consumption. Limits of applicability are often specified in connection with a

98

particular calculation method but these limits have no significance if the fundamental assumptions are not correct.

MECHANISM OF PRESSURE TURNAROUND The simplest and most conservative approach to vent sizing is to design for no overpressure. If the rate of heat generation at vent opening is equal to the rate of heat removal due to vapour generation, the reaction will be fully tempered and the pressure and temperature will gradually decline. However, large vent areas are required in order to achieve pressure turnaround at vent opening, particularly if the required volumetric flow rate of vapour is subsequently reduced due to two-phase flow. Substantial reductions in the required vent area can be achieved, if the maximum pressure (Pmax) can be allowed to exceed the relief set pressure. However, this may affect the mechanism of pressure turnaround, producing, for example, a transition from tempering to vessel emptying or reaction completion.

The mechanism of pressure turnaround, during venting with overpressure, can be distinguished by the following characteristic features:

Tempering: With a relatively high proportion of vapour in the discharge stream, the increase in pressure above the relief set pressure increases the mass discharge rate of vapour until the rate of heat removal becomes equal to the rate of heat generation. At this point the temperature and pressure reach their maximum values and then gradually decrease.

Emptying: Two-phase flow, with a high proportion of liquid, causes the mass in the reactor to decrease rapidly. The rate of temperature rise is not strongly affected until the point is reached where the vessel is virtually empty and the rate of heat removal per unit mass becomes significant. Under these conditions, the temperature and pressure reach maximum values and then decrease rapidly at the point when the entire contents of the reactor have been discharged.

Reactant consumption: The ratio of liquid to vapour in the two-phase discharge from the reactor is not sufficient to empty the vessel rapidly but the proportion of vapour is insufficient to cool the contents and prevent the reaction from accelerating rapidly to completion. The temperature increases at an accelerating rate with no significant reduction in the rate of temperature rise at vent opening. The temperature and pressure go through defined maxima as the reaction nears completion. Two-phase flow continues after Pmax and the reaction products cool due to vapour production. The final mass in the reactor is determined by the disengagement void fraction for the chemically inert system undergoing depressurisation.

The likely mechanism of pressure turnaround can be established by examining the relationship between pressure, temperature and reaction mass during a vented runaway reaction. In the present investigation, the characteristics of each mechanism of pressure turnaround are compared with the results of pilot and large-scale runaway reaction experiments performed at the Health and Safety Laboratory.

99

SIMPLIFIED EQUATIONS The simplified vent sizing equations recommended by DIERS rely mainly on tempering and the emptying time principle. The most widely used equation was proposed by Leung 3:

oA = m ⋅ q

(1)2

⎞1/ 2 ⎤

/ 2 ⎥G ⋅ ⎢⎡ ⎜⎛ V ⋅ h fg ⎟

⎟ + (C f ⋅ ∆T )1⎢⎜ mo ⋅ v fg ⎠ ⎥

⎦⎣⎝

Where: A = vent area, mo = initial reaction mass, q = heat production rate (averaged between the relief set pressure (Pset) and the allowable overpressure), G = mass flux, V = reactor volume, hfg = latent heat, vfg = difference in specific volume between vapour and liquid, Cf = liquid specific heat capacity and ∆T is the temperature rise between Pset and the allowable overpressure.

The rate of heat production in Equation 1 is assumed to have a constant (average value) between Pset and the allowable overpressure and the void fraction distribution in the reactor is assumed to be homogeneous. The homogeneous vessel assumption is also used in the calculation of G.

Equation 1 can be rearranged to give the emptying time directly:

⎛⎛⎜⎜ V ⋅ h fg ⎞1/ 2 ⎞

2

/ 2 ⎟ ⎜⎜ m ⋅ v fg

⎟⎟ + (C ⋅ ∆ T )1f ⎟

otemptying = Gm ⋅ A

= ⎝⎝ o ⎠

q ⎠ (2)

If the runaway reaction is not mitigated by the cooling effect of vapour production, the available emptying time (Cf.∆T / q) is obtained by setting hfg to zero in Equation 2. This corresponds to the time, under adiabatic conditions, for the temperature to increase by ∆T above the temperature at Pset. The rate of temperature rise decreases and the available emptying time increases when some of the reaction energy is used to produce vapour. Under the homogeneous vessel assumption, the increase in available emptying time due to tempering, represented by (V.hfg /mo.vfg)/q in Equation 2, is relatively small. As the allowable overpressure is increased, the emptying time is governed by increases in Cf.∆T while V.hfg /mo.vfg remains small and independent of overpressure.

In practice, there will be some degree of vapour-liquid disengagement and the emptying time will increase as the proportion of liquid in the discharge stream decreases. Increased disengagement increases the cooling effect due to vapour production and this may lead to tempering before the reactor is empty.

Because of disengagement, the homogeneous assumption leads to an overestimate of the mass discharge rate and, if pressure turnaround were due primarily to emptying,

100

the maximum reactor pressure could exceed the calculated value. If the reaction is tempered, the homogeneous vessel assumption gives an underestimate of the cooling effect and the maximum pressure is likely to be less than the calculated value. If pressure turnaround is due to reactant consumption, the homogeneous assumption does not guarantee a conservative prediction of the maximum pressure.

PILOT-SCALE EXPERIMENTS Maximum reactor pressures calculated using Equation 1 have been compared with the results of a large number of venting experiments using the 350 litre pilot-scale facility at the Health and Safety Laboratory. Various reaction systems have been investigated with a range vent areas, batch volumes and relief set pressures. In general, Equation 1 was conservative, yielding a calculated maximum pressure which exceeded the experimentally observed value. However, the degree to which the calculations were conservative depended on experimental conditions. There was no strong correlation between the calculated values and the observed variation in the maximum pressure. When the hydrolysis of acetic anhydride was investigated, in some cases, the experimental maximum pressures exceeded the calculated values. The cases included experiments in which surfactant was added to the reaction mixture. The variability in the degree to which to which some calculations were conservative and the non-conservative values for the hydrolysis reaction may arise because the pressure turnaround was not due to vessel emptying.

The experiments using the 350 litre facility were designed to establish directly whether a particular sizing method was conservative under the selected experimental conditions. However, where possible, additional instrumentation was provided to assess separate elements of the vent-sizing method such as the estimation of two-phase flow capacity, interpretation of the calorimetric data and level swell calculations to predict the onset of two-phase flow. The mechanism of pressure turnaround can be established from measurements of the mass of the reactor contents, the mass discharge rate and the void fraction at vent-line input. These data are difficult to obtain during rapid discharge from a relatively small glass-lined reactor. However, with detailed interpretation of the response from a gamma-ray densitometer mounted on the vent line and indications from load cells mounted under the catch tank, the results suggested that, in most cases, pressure turnaround was not primarily due to mass depletion.

Observed temperature and pressure variations at vent opening for a series of pilot-scale experiments can also be used to distinguish between possible mechanisms of pressure turnaround. A progressive reduction in vent area or an increase in batch volume is expected to produce a transition from full tempering to conditions under which turnaround is either due to emptying or reactant consumption. A transition to vessel emptying would give a gradual increase in the maximum pressure, as conditions are made more severe. By contrast, because of the exponential dependence of reaction rate on temperature, a transition from tempering to rapid reactant consumption is evident when small reductions in vent area or small increases in batch volume produce an abrupt increase in the maximum temperature and pressure.

101

METHANOL -ACETIC ANHYDRIDE REACTION Results from three pilot-scale experiments on venting of the reaction between methanol and acetic anhydride are summarised in Table 1. The batch volume and relief set pressure were held constant while the vent area was reduced progressively. The table shows no significant increase in Pmax when the vent diameter was reduced from 35 to 25 mm but a further reduction to 15 mm caused an increase in Pmax from 200 to 439 kPa. The mass remaining in the reactor increased when the vent diameter was reduced from 35 to 25 mm but the further reduction in vent diameter produced no substantial change in the final mass.

Table 1 Summary of results from pilot-scale experiments on reaction between methanol and acetic anhydride with batch volume of 250 litres and relief set pressure of 200 kPa.

(mm) (kPa) (K) (kg) 35 200.2 360.8 109.6

25 200.8 364.2 145.1

17.5 436.8 394.7 147.5

Vent diameter Maximum pressure Maximum temperature

Final reaction mass

Tem

pera

ture

(K)

Methanol/acetic anhydride 250 litres, R.S.P. 200 kPa, 25 mm vent380

)

Liqui

Pressure (K

Pa

d temperature

Vapour temperature

vent open Pressure 200

370

180

360

160

350

140

340

120

330

100

320 1400 1500 1600 1700 1800 1900 2000 2100

Time (s)

Figure 1 Temperature and pressure variations during pilot-scale venting of runaway reaction between methanol and acetic anhydride with a 25 mm vent diameter (batch volume 250 l: relief set pressure 200kPa).

102

Methanol/acetic anhydride 250 litres, R.S.P. 200 kPa, 17.5 mm vent P

ressure (KP

a) Tem

pera

ture

(K)

400

vent open

pressure

vapour temperature

liquid temperature

500

400380

300 360

200

340

100

320 1600 1800 2000 2200 2400

Time (s)

Figure 2 Temperature and pressure variations during pilot-scale venting of runaway reaction between methanol and acetic anhydride with a 17.5 mm vent diameter (batch volume 250 l: relief set pressure 200kPa).

Figures 1 and 2 show the temperature and pressure records from the experiments with 25 and 17.5 mm vent diameters, respectively. Both figures show that, after vent opening, the temperature in the vapour space at the top of the reactor becomes equal to the liquid temperature, indicating the onset of two-phase flow. On vent opening, the temperature records in Figure 1 show a progressive reduction in the rate of rise until, after about 70 sec., a maximum is reached, indicating full tempering with equality between the rate of heat production and heat removal.

Figure 2 shows no significant reduction in the rate of temperature rise at vent opening. Pressure and temperature continue to increase rapidly, reaching a maximum after 240 sec. The temperature records show that the reduction in vent diameter has reduced the rate of cooling due to vapour removal but the increases in the mass discharge rate as the pressure rises has not been sufficient to empty the vessel before the reaction proceeds to completion. For the experiment with a 17.5 mm vent diameter, the total mass discharged was 38% of the initial mass. The proportion discharged up to Pmax would have been substantially lower and the corresponding reduction in the total rate of heat generation would have been small compared with the increase in rate due to the temperature rise between vent opening and the maximum temperature (Tmax). With rapidly increasing temperature, equality between the rate of heat production and heat removal would have occurred only when the reaction neared completion.

103

HYDROLYSIS OF ACETIC ANHYDRIDE Pilot-scale experiments on the reaction between water and acetic anhydride were

Pressure (kP

a) Tem

pera

ture

(K)

performed over a range of relief set pressures with batch volumes from 75 to 200 litres. For some experiments a small quantity of surfactant was added in order to investigate how the maximum reactor pressure is influenced by level swell and the degree of vapour-liquid disengagement in the reactor.

The temperature and pressure records from two pilot scale experiments on venting of the runaway hydrolysis with and without surfactant are shown in Figure 3. At vent opening, with no surfactant, the reaction is fully tempered and the temperature and pressure begin to decline as the reaction proceeds to completion. Under the same conditions except for the addition of surfactant, the pressure increases rapidly after vent opening and reaches a value 80 % above the relief set pressure.

Acetic anhydride hydrolysis: Batch Volume 100 litres Set P. 200 kPa Comparison of results with and without surfactant

Pressure with surfactant no surfactant

Temperature with surfactant no susrfacatnt

vent open 400440

350420

300400

250380

200360

340 150

320 100

300 1570 1580 1590 1600 1610

Time (s)

Figure 3 Temperature and pressure variations during pilot-scale venting of hydrolysis of acetic anhydride with and without surfactant (batch volume 100 l: relief set pressure 200 kPa)

Results of adiabatic calorimetry on the hydrolysis reaction with and without surfactant are shown in Figure 4. The addition of surfactant was found to have no effect on the adiabatic rates of temperature rise. Similarly, the presence of surfactant did not affect the pressure–temperature relationship. This means that calculations based on the closed system adiabatic data and the homogeneous vessel assumption give the same recommended vent diameter whether or not surfactant is present.

104

0

2

4

6

wi

in-1

)

no surfactant th surfactant

Log

(dT/

dt) (

K m

-3.2 -3.0 -2.8 -2.6 -2.4 -2.2 -2.0 -1.8

-1000/T (K)

Figure 4 Adiabatic self-heat rate data for the hydrolysis of acetic anhydride.

Maximum pressures calculated using Equation 1 were found to be less than the experimental values in five out of nine cases. With surfactant, the greatest difference was 1.0 bar and, with no surfactant and the highest batch volume, a difference of 1.2 bar was observed. However, the required vent diameters (without safety factors and calculated using a different procedure) were only slightly smaller than the actual diameter in three cases out of nine5. Detailed analysis indicated that the addition of surfactant or the increase in batch volume increased the proportion of liquid in the discharge stream and reduced the volumetric flow capacity for vapour. The corresponding reduction in heat removal due to vapour generation caused a transition from full tempering to conditions where the reaction proceeded rapidly to completion. Increased mass discharge rates and correspondingly reduced emptying times caused no significant reduction in maximum pressure. However, with relatively low batch volumes, integrated heat losses due to vapour generation between Pset and Pmax appeared to cause some reduction in the maximum pressure and temperature.

LARGE-SCALE EXPERIMENTS The pilot-scale experiments on the hydrolysis of acetic anhydride have provided evidence for likely mechanisms of pressure turnaround and demonstrated that the maximum pressure cannot be predicted reliably unless the degree of vapour-liquid disengagement is known. Detailed analysis of the experimental data was required in order to establish the relative importance of tempering, emptying and reactant consumption and it was not clear how far conclusions from pilot scale experiments could be extrapolated to larger, industrial-scale, vessels.

105

Figure 5 Large-scale facility for investigating runaway reactions.

LARGE-SCALE FACILITY As part of the EU funded AWARD project, a large-scale facility was constructed at the Health and Safety Laboratory in order to investigate the reliability of emergency relief system (ERS) design methodology using a reactor similar in size to industrial vessels.

The facility, shown in Figure 5, comprises a 2,500 litre reactor connected via a 200 mm diameter vent line to a 13,000 litre catch tank. The height of the reactor is 3 m and was chosen to give a hydrostatic head similar to many industrial vessels. A diameter of 1m was selected in order to facilitate the installation of specialised instrumentation to determine the axial variation in void fraction in the reactor during venting and to limit the total volume so that the experiments could be performed safely. The contents of the reactor are heated and mixed by pumped circulation through an external heat exchanger. A large number of temperature and pressure transducers are installed at various points in the reactor, vent line and catch tank. Level swell and the density distribution in the reactor are determined using differential pressure and a specially designed scanning gamma-ray system.

The configuration of the gamma system is shown in Figure 6. Collimated gamma-ray beams emerge from source holders on one side of the reactor and rod-shaped detectors on the other side are used to monitor changes in attenuation caused by variations in the two-phase density in the vessel during pressure relief. Two systems have been installed to monitor density variations in the upper and lower halves of vessel. The sources are rotated in order record the variation in attenuation as the angle of the beam changes. Under these conditions, the response from the detectors can be related to the axial variation of void fraction in the vessel.

106

rotating source holders

detectors

Figure 6 Configuration of scanning gamma-ray system

The differential pressure sensors comprise a differential pressure cell connected by oil-filled lines to diaphragms which are bolted to the side of the reactor. The configuration of the sensors is shown in Figure.7. One of the transducers (DP101) records the pressure difference between diaphragms at the top and bottom of the reactor. The response is directly proportional to the mass of the contents and is not affected by the density distribution or absolute pressure in the reactor. The average density in the central section of the reactor can be calculated from the response of DP102. Bubbling in the bottom of the vessel and level swell above the upper diaphragm causes mass transfer in and out of the central zone. The net effect of the onset of two-phase flow can be to leave the average density unchanged. More gradual changes occur at the end of two-phase flow, which is indicated by an increase in differential pressure from DP102 as the level drops back toward the upper diaphragm followed by a decrease as bubbling subsides in the zone below the lower diaphragm. The third sensor (DP103) shows bubbling in the lower half of the vessel, causing level swell above the central diaphragm.

The runaway reactions were initiated by first heating acetic anhydride in the reactor to a temperature of 50°C and then charging an equimolar quantity of water. An actuated valve in the vent line was opened automatically at the relief set pressure (2 bara) to allow discharge from the reactor to the catch tank. A 100 mm diameter orifice plate was installed downstream from the actuated valve and the vent valve from the catch tank to atmosphere was kept closed during the tests.

107

DP 101

DP 103

DP 102

2956

mm

668

mm

800m

m80

0mm

Figure 7 Configuration of differential pressure sensors.

RESULTS FOR HYDROLYSIS OF ACETIC ANHYDRIDE The large-scale experiments were designed to determine the axial variation of void fraction in the reactor in order to assess the validity of various level swell models and assumptions regarding the flow regimes in the reactor and vent line. The analysis of these aspects has been reported elsewhere4. In the present discussion, the records of pressure, temperature and reaction mass and the evaluated void fraction distributions are used to determine the mechanism of pressure turnaround in the large-scale experiments. The enhanced instrumentation installed on the large-scale facility and increased reactor volume provide data on mass discharge rates and void fraction distributions which were not available from the pilot-scale tests and allow a more rigorous assessment of vent-sizing methodology.

Table 2 Summary of results of large-scale experiments on venting of the runaway hydrolysis of acetic anhydride with a relief set pressure of 2 bara.

Initial fill Maximum Pressure (bara)

No surfactant With surfactant 50 % 3.60 5.76

60 % 6.00 7.19

70 % 6.78 7.71

108

0

2

4

6

8

10

12

14

16 ia

l pre

ssur

e (k

Pa)

)

di

mi

(top)

100

110

120

130

140

150

160

170

180

Tenm

pera

ture

(°C

)

Diff

eren

tP

ress

ure

(bar

a , A

ttenu

atio

n

end of two-phase flow

temperature bottom top

fferential pressure top - bottom

d top - mid bottom

pressure

attenuation

vent open

3140 3160 3180 3200 3220 3240

Time (sec.)

Figure 8 Variations in temperature, pressure and gamma-ray attenuation during large-scale venting of hydrolysis of acetic anhydride with 50% fill and no surfactant.

0

80

0

2

4

6

8

10

12

14

16

i )

ion

i (

i

mi i

low )

160

Diff

eren

t al p

ress

ure

(kP

a) P

ress

ure

(bar

a , A

ttenu

at


vent open

pressure attenuat on top)

differant al pressure top - bottom

d top - m d bottom

end of two-phase f T

empe

ratu

re (°

C

2020 2040 2060 2080 2100 2120

Time (sec)

Figure 9 Variations in temperature, pressure and gamma-ray attenuation during large-scale venting of hydrolysis of acetic anhydride with surfactant and 50% fill.

109

Experiments, with and without surfactant, were performed over a range of batch volumes. The results are summarised in Table 2. The experiment with 50% fill and no surfactant gave a relatively small overpressure and Table 2 shows that the addition of surfactant or a small increase in batch volume produces a large increase in the maximum pressure. This parametric sensitivity was evident in the pilot-scale experiments and is likely to be associated with the exponential dependence of reaction rate and temperature. Further increases in batch volume produced relatively small increases in pressure.

Results from the experiments with an initial fill of 50% and no surfactant are shown in Figure 8. After vent opening at 2 bara, the pressure rises gradually and reaches a maximum at 3.60 bara and then remains approximately constant. At vent opening, the temperature in the vapour space at the top of the vessel increases rapidly to become equal to the liquid temperature at the bottom, indicating the onset of two-phase flow. At the same time, level swell to the top causes a sharp increase in attenuation of the beam scanning the upper part of the vessel. The end of two-phase flow is evident from the sharp reduction in attenuation approximately 25 sec after vent opening. The reduction in level causes an increase in the differential pressure in the central zone. The changes in attenuation beyond the period of two-phase flow are due to the gamma ray beam periodically passing across the flange in the upper part of the reactor. Mass discharge rates recorded by the top to bottom differential pressure sensor show a gradual decrease during the period of two-phase flow. The transition to vapour only flow results in slow changes in the reaction mass as the temperature and pressure remain approximately constant, indicting that equality between rates of heat generation and removal has tempered the reaction.

Figure 9 shows results from the experiment with surfactant and an initial fill of 50%. In contrast to the experiment without surfactant, the temperatures and pressure show accelerating rates of increase after vent opening and reach defined maxima after 19 sec. Two-phase flow is evident from the changes in vapour-space temperature, attenuation and the differential pressure in the central zone of the reactor. Two-phase flow continues beyond Pmax and the mass discharge rates and total mass discharged are much higher than those observed without surfactant. The addition of surfactant is expected to increase level swell, producing a more rapid onset of two-phase flow and increasing the proportion of liquid entering the vent line. Comparison of Figures 8 and 9 shows that the increases in vapour-space temperature and attenuation are more rapid when surfactant is present. The attenuation, between vent opening at Pmax is higher for the experiment with surfactant. Because of the exponential relationship between attenuation and density, this represents a large reduction in the void fraction at the top of the vessel. The increase in two-phase density in the vent line causes a marked reduction in the in the flow of vapour and the rate of heat removal from the reactor. With surfactant, the rates of heat removal are not sufficient to temper the reaction and the mechanism of pressure turnaround changes. The form of the pressure and temperature variations and, particularly, the records of reaction mass, strongly indicate the maximum pressure is determined by reaction completion rather than emptying.

The effect on the experimental results of increasing the batch volume from 50 to 60% is, in some ways, similar to the effect of adding surfactant. With a 60% initial fill

110

0

20

40

60

80

0

2

4

6

8

10

12

14

16

18

20 /

( )

(

l

mi i

open

(

100

120

140

160

180

200

220 At

tenu

atio

n pr

essu

re k

Pa

attenuation top)

pressure


differentia pressure top - bottom

d top - m d bottom

vent

Tem

pera

ture

°C

)

1940 1960 1980 2000 2020

Time (sec.)

Figure 10 Variations in temperature, pressure and gamma-ray attenuation during large-scale venting of hydrolysis of acetic anhydride with 60% fill and no surfactant.

(Figure 10), accelerating rates of pressure and temperature rise and increased mass discharge rates are accompanied by a large increase in the maximum pressure. The vapour space temperature and attenuation variation, shown in Figure 10, indicate a more rapid onset of two-phase flow that that observed due to the addition of surfactant (Figure 9). Earlier onset of two-phase flow will have increased the proportion of liquid entering the vent line and correspondingly reduced the rates of cooling due to vapour production. The results indicate that the increase in batch volume has caused a transition from tempering to a pressure turnaround caused by reactant consumption.

The preceding qualitative and semi-quantitative assessment of records from three of the large-scale experiments provides evidence for the mechanisms which determine the maximum temperature and pressure in the reactor. A quantitative indication of the relative importance of the underlying mechanisms is given in Table 3. The reduction in reaction mass at Pmax is around 30% for each experiment and the total rate of heat production would be reduced proportionately. By contrast, the temperature increase between Pset and Pmax produces increases in the rate of heat production of 130% for the tempered case and 160% for the other two experiments. This implies that, if gradual increases in the rate of vapour removal do not produce steady state tempering, equality between the rate of heat production and removal is unlikely to occur unless the reaction subsides due to reactant consumption. The untempered experiments (Figures 9 and 10) show sustained periods of two-phase flow and mass discharge after Pmax, also implying that the pressure turnaround is not due to emptying.

111

Table 3 Relationship between mass discharge, rates of heat production and duration of two-phase for three of the large-scale venting experiments on venting of the hydrolysis of acetic anhydride.

between Pset and Pdischarged

Duration of two phase vent open toMass

discharged

(%) (%) (%) (sec) (sec)

27.0 129.7 30.2 26.2 42.2 no

29.6 161.7 72.1 29.4 19.0 with

33.2 162.0 49.0 32.6 21.3 no

max Total mass Time from Increase in heat rate/kg flow Pmax

50% fill

surfactant 50% fill

surfactant 60% fill

surfactant

VOID FRACTION DISTRIBUTIONS AND COOLING RATES The changes in void fraction in the reactor during venting have been determined by combined interpretation of the data from the gamma system and the differential pressure cells4. In general, the addition of surfactant was found to reduce the void fraction entering the vent line, increase the void fraction in the lower part of the reactor and give an extended period of two-phase flow. Once the void fraction distribution has been established, vapour production and corresponding cooling rates can be calculated using the experimentally observed mass discharge rates and the void fraction at the top of the reactor. Results of these calculations for the experiments with and without surfactant and an initial fill of 50% are shown in Figure 11; along with heat production rates calculated using the adiabatic data. There are large fluctuations in the calculated cooling rates, particularly for the experiment without surfactant. Video records from a camera mounted on a sight glass at the top of the reactor indicate that the fluctuations are due to oscillations between two-phase and vapour-only flow before the complete transition to vapour only flow. With no surfactant, the increase in pressure after vent opening gives average cooling rates which are comparable with the rate of heat production. This causes a reduction in the rate of temperature rise leading to steady state tempered condition at 439 K, with equality between the rate of heat production and removal (Figure 11). With surfactant, the rate of cooling remains below the adiabatic rates of heat production and, as the temperature increases after vent opening, steady state conditions become impossible. Under these conditions, maximum temperature coincides approximately with the temperature at which the adiabatic heat production rate reaches a maximum. The final temperature increase is determined by the adiabatic temperature rise reduced by an amount related to the integrated rates of heat removal due to heat transfer during the induction period and vapour generation after vent opening.

112

Hea

t rat

e (W

/kg)

10000

8000

6000

4000

2000

0

400 410 420 430 440 450 460 470 480

l

l

i i

maxi ( maxiopen

heat generation adiabatic data

heat remova no surfactant with surfactant

tempering, heat remova exceeds heat generation

maximum ad abt c rate

mum temperature tempered)

turnaround due to reactant consumption

mum temperature (untempered)

vent

Temperature (K)

Figure 11 Adiabatic self-heat rate data compared with cooling rates calculated for large-scale experiments on venting of the hydrolysis of acetic anhydride with and without surfactant.

IMPLICATIONS OF ASSUMPTIONS ABOUT PRESSURE TURNAROUND If the maximum temperature is determined by the total reaction enthalpy or adiabatic temperature rise, the maximum pressure may differ significantly from values calculated assuming tempering or vessel emptying at Pmax. The simplified DIERS equations include self-heat rates averaged between Pset and Pmax, without reference to the total exothermicity. If pressure turnaround is due to reactant consumption, the maximum pressure can be lower or higher than the calculated values, depending on the total heat of reaction. This can be seen from Figure 12, which shows adiabatic data for the hydrolysis reaction along with theoretical values for reactions with similar initial rates of heat production but differing heats of reaction. Each of these adiabatic data sets would give the same recommended vent area and, in principle, the reactions could be fully tempered at the same moderate overpressure. However, if the allowable overpressure was increased, relying increasingly on the emptying time principle, the corresponding reduction in vent area could cause a transition from tempering to reaction completion at Pmax. The maximum pressures for the theoretical reaction with the highest adiabatic temperature rise would be higher than those observed for the hydrolysis reaction. Alternatively, with a lower adiabatic temperature rise, the vent area could be lower than that obtained using the emptying time principle.

113

3

2

1

0

-1

-2

-3

-4

-5 -3.2 -3.0 -2.8 -2.6 -2.4 -2.2 -2.0 -1.8 -1.6

ln(d

T/dt

) (K

min

-1 )

acetic anhydride hydolysis

∆ Tad1

∆ T ∆ Tad3

∆ Tad2

temperatures at P max(allowable) at P set

-1000/T (K-1)

Figure 12 Adiabatic self-heat rate data for the hydrolysis of acetic anhydride compared with curves for theoretical reactions with similar initial heat rates but with heats of reaction and corresponding adiabatic temperature rises (∆ Tad) which are above and below the value for the hydrolysis.

VENT-SIZING FOR REACTION COMPLETION If pressure turnaround occurs due to reactant consumption, the following expression can be used to relate the recommended vent area to the maximum allowable temperature and pressure5.

⎛ ⎞⎜ ⎟⎜ 1 ⎟mo

GtA = ⎜ 1 − ⎟ (3)

ad ⎜ exp⎜

⎜⎛ C f ( T + ∆ Tad − Tmax )⎟

⎞ ⎟o

⎜⎜ x⎝ ⎝ h fg ⎠⎟

⎠⎟⎟

Where: tad = adiabatic time between Pset and maximum temperature rate, To = initial temperature, Tmax = maximum allowable temperature, ∆Tad = adiabatic temperature rise, x = vapour mass fraction at vent line input.

Equation 3 is obtained by relating the rate of cooling due to vapour generation to the average mass discharge rate and the rate of heat production and integrating over the period between vent opening and the maximum adiabatic rate of temperature rise.

114

Equation 3 gives larger vent areas as the heat of reaction increases and gives conservative predictions for the maximum pressures in the pilot-scale experiments, when the experimentally determined value of the inlet quality is used in the calculations. In common with the DIERS simplified equations and dynamic computer models, accurate implementation of Equation 3 is dependent on a reliable model for the flow regime in the reactor. When the homogeneous vessel assumption, implicit in Equation 1, is used to calculate the inlet quality and flow capacity in Equation 3, conservative vent areas are obtained for all of the nine pilot-scale experiments on the hydrolysis of acetic anhydride.

FURTHER INVESTIGATION Large-scale experiments have been performed with and without surfactant over a limited range of batch volumes keeping the relief set pressure and vent area constant. The vent area was chosen to give sustained periods of two-phase flow, in order that changes in the axial variation of void fraction could be measured during the venting period. In the pilot-scale experiments, relatively larger vent areas were chosen in order to give a direct assessment of the reliability of simplified vent-sizing equations. The large-scale experiments provided data on mass discharge rates and void fraction distributions, which were not available from the pilot-scale experiments. Further large-scale tests using larger vent areas should be performed in order to develop criteria for predicting whether tempering will occur. With larger vent areas, it may be possible to investigate conditions where emptying produces pressure turnaround before the reaction proceeds to completion.

CONCLUSIONS A large number of pilot-scale venting experiments have been performed using a range of reaction systems, vent areas, batch volumes and relief set pressure. In general, the DIERS simplified vent-sizing equations gave conservative predictions of the maximum pressure. However, for the reaction between acetic anhydride and water with and without surfactant, in some cases, the observed pressure exceeded the calculated value. The difference between experimental and calculated values depended on the detailed calculation method. Some procedures gave recommended vent areas (without the use of safety factors) which were smaller than the actual diameter in only a few cases and, in these cases, the difference in vent area was small.

Detailed analysis of the pilot-scale results for the range of reaction systems indicates that either tempering occurred shortly after vent opening or the reaction proceeded rapidly to completion. Pressure turnaround due to emptying of the reactor was not observed in the pilot-scale tests.

Large-scale experiments on the hydrolysis of acetic anhydride were performed with instrumentation specially designed to determine the reaction mass and the axial variation in void fraction in the reactor during venting. The addition of surfactant, or a small increase in batch volume, produced large increases in the maximum pressure. This effect has been related to changes in the void fraction distribution and the mass discharge rate. Cooling rates, calculated from the mass discharge rates and the void fraction distribution, have been compared with rates of heat production derived from adiabatic data. The comparison shows how changes in the degree of vapour liquid

115

disengagement in the reactor produces a transition from full tempering to conditions where the reaction accelerates to completion.

If the reaction proceeds rapidly to completion, the final temperature and pressure are determined by the thermochemistry and not the kinetics. Under these conditions, vent-sizing equations that do not contain the heat of reaction or adiabatic temperature rise cannot reliably predict the maximum pressure and temperature. Vent sizing equations based on the emptying time principle have the potential to under predict the maximum pressure, if the heat of reaction is relatively large. However, in other cases, with low heats of reaction, the vent areas calculated to empty the vessel rapidly could be adequate to protect the vessel. The pilot and large-scale results indicate that vent-sizing calculations, using current methodology, should include safety factors6, unless the mechanism of pressure turnaround can be predicted with confidence. Further work is needed in order to establish whether the maximum pressure is generally determined by emptying, reactant consumption or tempering.

ACKNOWLEDGEMENTS The support of the European Commission under the Competitive and Sustainable Growth Programme (project G1RD-2001-00499), the Health and Safety Executive, Sanofi-Aventis, Astra Zeneca plc, Syngenta plc, Yule Catto plc and BS&B Safety Systems is gratefully acknowledged.

DISCLAIMER The opinions expressed in this paper are those of the authors and do not necessarily represent those of the sponsoring organizations.

REFERENCES

1. Etchells, J C and Wilday, A J (1998), "Workbook for chemical reactor relief system sizing", http://www.hse.gov.uk/research/crr_htm/1998/crr98136.htm, HSE Contract Research Report 136/1998, HSE Books

2. H G Fisher et al., "Emergency Relief System Design Using DIERS Technology", DIERS/AIChE, 1992, ISBN 0-8169-0568-1

3. J C Leung, "Simplified Vent Sizing Equations for Emergency Relief Requirements in Reactors and Storage Vessels", AIChE Journal, 32, (10), 1622-1634, 1986

4. Snee T J, Bosch J, Cusco L, Hare J A, Royle M and Wilday A J, 2005, DISPOSE: “Large scale experiments for void fraction measurement during venting,” HSL Internal Report PS/05/03

5. Snee, T J, Butler, C, Hare, J A, Kerr, D C, Royle, M and Wilday, A J, (1999), “Venting studies of the hydrolysis of acetic anhydride with and without surfactant (Vapour System 3)”, HSL Report No PS/99/13

6. Hare, J A, Wilday, A J and Owens, A, (2005), “Simplified methods for vent disposal system sizing for runaway chemical reactors: EC AWARD project guidance for SMEs”, IChemE Hazards XIX International Symposium, Manchester, March 2006

Published by the Health and Safety Executive 09/07


DISPOSE: Large scale experiments for void fraction measurement during venting The AWARD (Advanced Warning and Runaway Disposal) Project addressed the needs to detect runaway initiation in advance so that appropriate countermeasures can be taken and to design emergency relief systems for chemical reactors. The missing step in the design of runaway reactor relief systems was the availability of reliable methods for predicting level swell in the reactor during venting and hence the quantity of liquid requiring to be dealt with by a disposal system (quench tank, catch tank, etc.).

This report and the work it describes were funded by the Health and Safety Executive (HSE) together with the European Commission under the Competitive and Sustainable Growth Programme (project G1RD200100499), Astra Zeneca plc, Syngenta plc, Yule Catto plc and BS&B Safety Systems. Its contents, including any opinions and/or conclusions expressed, are those of the authors alone and do not necessarily reflect HSE policy.

RR587

www.hse.gov.uk

dispose: large scale experiments for void fraction ... · pdf filefor void fraction...

Documents