Transcript
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
Health and Safety Executive
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 be
reproduced, stored in a retrieval system, or transmitted in
any form or by any means (electronic, mechanical,
photocopying, recording or otherwise) without the prior
written 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 1BQ
or 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 handcalculation 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 ........................................................................................................................... 1
1.1 CEC AWARD Project................................................................................................... 1
1.2 Objectives...................................................................................................................... 2
1.3 This Report.................................................................................................................... 2
2. Main Tasks ............................................................................................................................ 5
3. Reaction System.................................................................................................................... 9
4. Previous Work..................................................................................................................... 11
4.1 Calorimetry ................................................................................................................. 11
4.2 Laboratory Scale Experiments .................................................................................... 12
4.3 Pilot Scale Experiments .............................................................................................. 12
5. Large Scale Experiments..................................................................................................... 14
5.1 Experimental facility and Procedure ........................................................................... 14
5.2 Test Matrix .................................................................................................................. 14
5.3 Results ......................................................................................................................... 15
5.4 Main features of Vented Runaway results .................................................................. 15
5.4.1 Maximum pressure .............................................................................................. 15
5.4.2 Behaviour with time ............................................................................................ 15
5.5 Derivation of Void Fractions....................................................................................... 17
5.6 Further Discussion of Experimental results ................................................................ 19
5.6.1 Mass remaining after venting .............................................................................. 19
5.6.2 Densitometer results ............................................................................................ 20
5.6.3 Temperature records............................................................................................ 20
5.6.4 Mechanism of pressure turnover ......................................................................... 23
6. Comparison of Experimental Results with Vent Sizing Predictions................................... 25
6.1 Introduction ................................................................................................................. 25
6.2 Methods which treat the reactor as homogeneous....................................................... 26
6.3 Methods which account for level swell ....................................................................... 27
6.3.1 Estimation of the void fraction at disengagement (for relief sizing purposes).... 27
6.3.2 Vent sizing using methods which account for level swell .................................. 27
7. Comparison of Experimental Results with methods for disposal system sizing ................. 29
7.1 Level swell .................................................................................................................. 29
7.1.1 Methodology ....................................................................................................... 29
7.1.2 Void fraction ....................................................................................................... 29
7.1.3 Mass vented......................................................................................................... 30
7.2 Vent Flowrate to Disposal System .............................................................................. 30
8. Conclusions ......................................................................................................................... 33
9. Recommendations ............................................................................................................... 35
10. References ....................................................................................................................... 37
11. APPENDIX A: Phi-Tec Adiabatic Calorimetry.............................................................. 39
11.1 Phi-Tec adiabatic calorimeter...................................................................................... 39
11.2 Experimental procedure .............................................................................................. 40
12. APPENDIX B: Design of the Large-Scale Facility ........................................................ 53
12.1 Design Considerations................................................................................................. 53
12.2 Instrumentation............................................................................................................ 56
12.2.1 Void fraction ....................................................................................................... 56
12.2.2 Other instrumentation.......................................................................................... 58
13. APPENDIX C: Large-Scale Experimental Results......................................................... 59
13.1 Reactor Facility ........................................................................................................... 59
13.2 Experimental Procedure .............................................................................................. 59
v
13.2.1 Hydrolysis of Acetic Anhydride.......................................................................... 59
13.2.2 Acetic Acid Blowdown Tests.............................................................................. 59
13.3 Experimental Results................................................................................................... 63
14. APPENDIX D: Derivation of Void Fractions ................................................................ 92
14.1 Gamma tomography .................................................................................................... 92
14.2 Differential pressure cells............................................................................................ 94
14.3 Density ........................................................................................................................ 94
14.4 Conversion .................................................................................................................. 94
15. APPENDIX E: LITERATURE PAPER......................................................................... 97
vi
1. 1.1
INTRODUCTION
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 nonequilibrium 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 a fter 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 largescale 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:
(CH 3 CO) 2 + H 2 O d 2CH 3 COOH 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 selfheating. 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.
6
-1
Log (dT/dt) (K min )
no surfactant with surfactant 4
2
0
-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
Pressure (bara)
12
no surfactant with surfactant
10 8 6 4 2 0 50
100
150
200
250
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. 5.1
LARGE SCALE EXPERIMENTS
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 Acetic anhydride (kg) Water (kg) Surfactant (%) Surfactant (kg)
HP1 1076.3 189.8 0 0
HP2 1510.5 267.1 0 0
Experiment number HP3 HP4 HP5 1292.5 1066.9 1508.4 228.1 188.5 266.4 0 0.25 0.25 0 3.14 4.44
HP6 1282.5 226.4 0.25 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 Experiment number
Parameter Acetic acid (kg) Surfactant (%)
Surfactant (kg)
blow1 ~1560 0 0
14
blow2 1520.6
0.25
3.8
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 (%) 50 60 70
Maximum pressure (bara) No surfactant With surfactant 3.60 5.76 6.00 7.19 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
7
16 Differential pressure top - bottom
12
6
vent
mid top - mid bottom
open
top - halfway
5
10 4 Pressure
8
reactor
3
vent line
6
dump tank
2
4
Pressure (bara)
Differential pressure (kPa)
14
1
2
0
0 2040
2060
2080
2100
2120
Time (sec) Differential pressure results for experiment HP4 (50% fill with surfactant)
Figure 3.
Reactor temperature
10
top bottom
160
open
8 Attenuation upper source
6
lower source
80
4
Pressure
2
reactor vent line dump tank
0 2020
0 2040
2060
2080
2100
2120
2140
Time (sec)
Figure 4.
Densitometer and temperature results for experiment HP4 16
Temperature (°C)
Attenuation / Pressure (bara)
vent
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
1
void fraction
0.8
0.6
0.4
Time to Pmax
0.2
Top bottom
0 -20
0
20
40
60
80
time (s)
Figure 5: Void fraction at two levels in the reactor for experiment HP1 (50% fill, no surfactant)
1
void fraction
0.8
0.6
0.4
Time to Pmax 0.2
bottom top 0 -20
0
20
40
60
80
100
time(s)
Figure 6: Void fraction at two levels in the reactor for experiment HP4 (50% fill, with surfactant)
18
120
1.2
vent open 1.0
Void fraction
0.8
0.6
0.4
vent inlet top upper mid lower mid bottom
0.2
0.0 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
HP1 HP3 HP6
50 60 50
No No Yes
5.6.2
% of mass discharged at maximum pressure 27.3 33.3 29.6
Maximum pressure (kPa) 3.60 6.0 5.76
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
200 175 150
8
Attenuation upper source
125
lower source
6 100 75
4
50
Pressure
2
Reactor temperature
0 3120
3140
3160
reactor
top
vent line
bottom
dump tank
3180
3200
3220
Temperature (°C)
Attenuation / pressure (bara)
10
25 0
3240
Time (sec.)
Figure 8: Scanning gamma densitometer measurements for experiment HP1 (50% fill, no surfactant) 200 10
180 160
open
8
Attenuation
140
upper source lower source
120
6 100 80
4
60 Pressure
2
Reactor temperature
40
reactor
top
vent line
bottom
dump tank
20
0 1940
1960
1980
2000
2020
Temperature (°C)
Attenuation / pressure (bara)
vent
2040
0 2060
Time (sec.)
Figure 9: Scanning gamma densitometer measurements for experiment HP3 (60% fill, no surfactant)
21
170
5.0
reactor pressure
4.5
vent open 4.0
3.5 150
reactor temperatures top
3.0
2.5
140
Pressure (bara)
Temperature (°C)
160
2.0
bottom 130 3140
3150
3160
3170
3180
1.5 3200
3190
Time (sec)
Figure 10: Temperature measurements for experiment HP1 (50% fill, no surfactant)
reactor pressure
180
6
170 5
160
reactor temperatures top
150
140
4
3
130
bottom
2
120 2050
2060
2070
2080
Time (sec)
Figure 11: Temperature measurements for experiment HP4 (50% fill, with surfactant)
22
Pressure (bara)
Temperature (°C)
vent open
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).
heat generation adiabatic data heat removal no surfactant with surfactant
10000
Heat rate (W/kg)
8000
6000
maximum adiabtic rate
turnaround due to tempering, heat removal reactant consumption exceeds heat generation
4000
maximum temperature maximum temperature (tempered) (untempered)
vent open 2000
0 400
410
420
430
440
450
460
470
480
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 pilotscale results on the hydrolysis of acetic anhydride and review of pilot-scale data for other reaction systems.
24
6.
6.1
COMPARISON OF EXPERIMENTAL RESULTS WITH VENT SIZING PREDICTIONS
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
Fill (%) Experiment Leung’s Method (using arithmetic mean) Leung’s Method (using Boyle time mean) Fauske’s method Huff’s method
No surfactant Surfactant 50 60 70 50 60 70 Vent diameter (mm) 100 100 100 100 100 100 318 193 188 181 157 158 321
214
214
200
186
190
210
144
144
135
131
126
315
166
169
159
140
139
450
50% fill Homog Calc
400
60% fill Homog Calc 70% fill Homog Calc
350
50% fill No Surf Expt 60% fill No Surf Expt
Vent Diameter (mm)
300
70% fill No Surf Expt 50% fill Surf Expt 60% fill Surf Expt
250
70% fill Surf Expt 200
150
100
50
0 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 over estimated 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 0 50% 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
Fill (%) Experiment (from Differential Pressure)
Experiment (from Densitometer) Calculated Churn Turbulent (Co = 1)
Calculated Churn Turbulent (Co = 1.5)
6.3.2
No surfactant 50 60 0.545
0.586
Surfactant 50 60 Void fraction
0.664 0.813 0.811
0.645
0.665
0.695
0.825
0.936
0.908
0.905
0.638
0.625
0.623
70
70
0.806
0.901
Not 0.815
available
0.870 0.867
0.621
0.607
0.605
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
Fill (%)
No surfactant 50 60
Experiment (mm)
100
100
Fauske’s Disengagement Method (CT Co=1) Wilday’s Disengagement Method (CT Co = 1.5) (using arithmetic mean) Wilday’s Disengagement Method (CT Co = 1.5) (using Boyle time mean)
199
135
136
124
115
120
252
155
158
132
121
128
253
180
172
145
155
144
Surfactant 70 50 60 Vent diameter (mm) 100 100 100
70 100
350
50% fill Churn Turb Calc 300
60% fill Churn Turb Calc 70% fill Churn Turb Calc 50% fill No Surf Expt
250
60% fill No Surf Expt
Vent Diameter (mm)
70% fill No Surf Expt 50% fill Surf Expt 200
60% fill Surf Expt 70% fill Surf Expt
150
100
50
0 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 10 30% 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
Fill (%) Experiment (from Differential Pressure)
Experiment (from Densitometer) Calculated Churn Turbulent (Co = 1)
Calculated Churn Turbulent (Co = 1.5)
No surfactant 50 60 0.545
0.586
Surfactant 50 60 Void fraction 0.664 0.813 0.811
0.645
0.665
0.695
0.825
0.756
0.752
0.752
0.549
0.546
0.546
70
70
0.806
0.76
Not 0.815
available
0.76 0.761
0.551
0.551
0.551
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
Fill (%)
Experiment (from Differential Pressure)
Calculated Homogeneous
Calculated
Churn Turbulent
(Co = 1)
Calculated Churn Turbulent
(Co = 1.5)
No surfactant 50 60 382.9
745.7
Surfactant 70 50 60 Mass vented (kg) 1152.7 907.5 1163.4
1266.2
1520.6
1777.6
1266.2
1520.6
1777.6
793.5
1056.5
1316.3
815.3
1077.7
1337.4
391.9
670.6
933.3
422.7
691.4
951.1
70 1421.8
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
Fill (%) Experiment (Average) Experiment (Two Phase) Calculated (Omega method)
635.1
1865.3
Surfactant 70 50 60 Mass flux (kg/m2.s) 2297.5 3357.5 3118.5
1806
2826.6
298.6
3814.4
3245.2
4949.2
1253.1
2104.2
2249.2
2026.9
2312.6
2426.5
No surfactant 50 60
70 3200.3
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 pilotscale 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. 11.1
APPENDIX A: PHI-TEC ADIABATIC CALORIMETRY
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
PA81 PA82 PA83 PA91
A3.
Temperature after water addition (°C) 49.6 49.2 30.0 25.0
Surfactant concentration (%wt/wt) 0 1 1 0.3
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 in Figure A3. Table A2. Results Run no.
PA81 PA82 PA83 PA91
Nominal start temp. (°C) 50 50 30 25
surfactant concentration (% wt/wt) 0 1 1 0.3
Adiabatic temperature rise (K) 186.5 183.0 190.9 189.1
40
Maximum temperature rate (Ks-1) 4.3 3.2 1.7 2.4
Time to Maximum rate (s) 946 996 2765 3153
Maximum pressure (kPa) 1349 1250 1015 909
250
225
200
Temperature (°C)
175
150
125
PA81 PA82 PA83 PA91
100
75
50
25
0
0
1000
2000
3000
4000
5000
Time (s)
Figure A2. Temperature v Time
16
14
Pressure (bara)
12
10
8
PA81 PA82 PA83 PA91
6
4
2
0 0
1000
2000
3000
4000
Time (s)
Figure A3. Pressure v Time
41
5000
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) 0 125.1 165.3 205.4 246.4 287.3 328.3 369.2 410.2 450.4 491.4 516.5 541.8 566.9 592.2 617.4 642.6 667.8 692.9 718.2 743.4 758.4 773.5 788.7 803.9 819.1 834.2 844.4 854.5 864.7 874.9 885 890.2 895.4 900.6 905.7 910.9 913.1 915.2 917.4 919.6 921.7 923.9 926.1 928.2 930.4
Temperature (°C) 49.6 51.6 52.3 53 53.7 54.5 55.3 56.2 57.1 58 59.2 60.1 61.1 62.1 63.2 64.5 65.8 67.3 68.9 70.8 72.9 74.3 75.9 77.6 79.5 81.7 84.2 86 88.1 90.5 93.2 96.5 98.4 100.5 102.9 105.6 108.7 110.3 111.9 113.7 115.6 117.7 120 122.6 125.5 128.8
Pressure (kPa) 112 113 113 113 114 115 116 117 117 123 128 129 129 130 132 133 135 137 139 141 145 147 148 151 155 159 163 167 169 174 180 186 191 195 200 207 214 218 221 221 225 235 241 248 256 265 42
dT/dt (°C/s) 0 0.017 0.018 0.019 0.019 0.02 0.02 0.021 0.022 0.025 0.034 0.037 0.04 0.043 0.047 0.05 0.056 0.063 0.07 0.079 0.089 0.104 0.11 0.121 0.129 0.154 0.178 0.193 0.218 0.245 0.285 0.338 0.38 0.44 0.483 0.577 0.673 0.735 0.775 0.852 0.925 1.023 1.127 1.27 1.45 1.622
dP/dt (kPa/s) 0.082 0.008 -0.013 0.02 0.049 -0.005 0.032 0.001 -0.008 0.325 0.036 0.006 0.042 0.051 0.042 0.052 0.084 0.115 0.046 0.132 0.177 -0.375 0.089 0.358 0.007 0.662 0.48 -0.046 0.292 0.287 0.683 1.205 0.748 0.83 0.753 1.145 1.7 1.58 1.272 -1.382 6.083 3 3.4 3.133 3.733 5.183
Time (s) 932.6 933.2 933.9 934.6 935.2 935.9 936.6 937.2 937.9 938.6 939.2 939.9 940.6 941.2 941.9 942.6 943.2 943.9 944.6 945.2 945.9 946.6 947.2 947.9 948.6 949.2 949.9 950.6 951.2 951.9 952.6 953.2 953.9 954.6 955.2 955.9 956.6 957.2 957.9 960.1 962.2 964.4 966.6 971.7 981.9 997.1
Temperature (°C) 132.6 133.9 135.2 136.7 138.1 139.7 141.4 143 144.8 146.7 148.6 150.7 153 155.2 157.6 160.1 162.5 165.2 167.7 170.3 173.1 176 178.7 181.5 184.2 186.7 189.1 191.3 193.5 195.6 197.7 199.6 201.5 203.2 204.8 206.3 207.7 209 210.1 213.3 215.7 217.6 219 221.3 223.3 224
Pressure (kPa) 277 281 285 290 294 299 305 312 320 330 340 353 368 384 401 422 442 464 487 512 539 568 597 627 658 689 721 753 784 814 845 873 901 929 954 978 1001 1022 1042 1099 1144 1182 1212 1265 1322 1349
43
dT/dt (°C/s) 1.9 1.983 2.1 2.217 2.3 2.417 2.533 2.633 2.75 2.833 3.117 3.25 3.317 3.583 3.55 3.867 3.933 3.6 3.933 4.117 4.3 4.267 4.1 3.9 4.133 3.683 3.233 3.333 3.267 3.133 3 2.833 2.617 2.433 2.3 2.2 2 1.75 1.638 1.288 0.98 0.758 0.573 0.33 0.102 0.007
dP/dt (kPa/s) 5.717 5.917 6.567 6.883 7.2 8.067 9.183 11.133 14.05 15.783 17.833 21.167 22.833 25 28.167 32 32.833 32.667 35.833 39.5 42.5 43.5 45 46.667 46.667 46.333 47.833 47.667 45 45.833 44.667 42.5 41.333 38.833 37.833 35 32.167 30.333 29.667 23 18.833 15.517 12.383 8.217 4.1 0.23
Table A4. PHI-TEC data for hydrolysis of acetic anhydride with a starting temperature of 323K (50°C) and 1% surfactant, run No PA82. Time (s) 0 51.6 92.4 135.1 175.8 217.2 258.1 298.7 339.6 380.6 421.5 462.5 503.4 544.4 569.6 594.8 620 645.2 670.4 695.6 720.8 746 771.2 786.9 802.1 817.2 832.4 847.6 862.8 877.9 888.1 898.3 908.4 918.6 928.8 933.9 939.1 944.2 949.4 954.6 959.8 961.9 964.1 966.2 968.4 970.6
Temperature (°C) 49.5 50.3 51 51.7 52.3 53 53.7 54.4 55.2 56 56.9 57.8 59 60.4 61.3 62.3 63.4 64.6 65.9 67.4 69 70.7 72.8 74.1 75.6 77.2 79 81 83.2 85.8 87.8 90 92.4 95.2 98.5 100.4 102.5 104.8 107.5 110.5 114 115.7 117.5 119.4 121.5 123.8
Pressure (kPa) 120 122 122 123 124 124 120 116 116 117 119 119 120 121 122 124 125 127 128 130 132 134 137 139 142 144 147 150 155 158 162 166 171 177 183 188 193 197 203 211 220 225 229 234 240 246 44
dT/dt (°C/s) 0.017 0.016 0.016 0.016 0.016 0.016 0.017 0.018 0.021 0.02 0.022 0.025 0.031 0.036 0.039 0.043 0.045 0.05 0.055 0.06 0.067 0.076 0.086 0.091 0.091 0.115 0.118 0.142 0.157 0.18 0.175 0.215 0.245 0.297 0.35 0.387 0.433 0.475 0.547 0.622 0.735 0.797 0.857 0.94 1.022 1.06
dP/dt (kPa/s) 0.177 0.018 0.016 0.018 0.011 0.005 -0.287 0.039 0.023 0.016 0.066 0 0.024 0.046 0.051 0.045 0.054 0.095 0.04 0.093 0.065 0.129 0.11 0.08 0.088 0.053 0.407 0.039 -0.044 0.407 0.222 0.347 0.507 0.905 1.043 0.532 1.005 0.785 1.417 1.64 2.267 2.017 2.133 2.733 2.667 3
Time (s) 972.7 974.9 977.1 979.2 981.4 982.1 982.7 983.4 984.1 984.8 985.4 986.1 986.8 987.4 988.1 988.7 989.4 990.1 990.7 991.4 992.1 992.8 993.4 993.9 994.6 995.2 995.9 996.6 997.3 997.9 998.6 999.3 999.9 1000.6 1001.3 1001.9 1002.4 1003.1 1003.8 1004.4 1005.1 1005.8 1006.4 1007.1 1009.3 1011.4 1013.6 1015.8 1020.9
Temperature (°C) 126.2 129 132.1 135.5 139.3 140.4 141.7 143.1 144.4 145.9 147.4 148.9 150.5 152.1 153.7 155.5 157.3 159.1 161 162.9 164.9 166.9 168.9 170.5 172.5 174.6 176.7 178.7 180.9 182.9 184.7 186.7 188.7 190.6 192.4 194.1 195.3 196.8 198.4 199.9 201.2 202.6 203.8 204.9 208.1 210.6 212.6 214.2 216.9
Pressure (kPa) 253 262 272 283 295 300 305 310 316 324 332 341 351 361 372 385 398 412 427 442 459 477 496 510 529 550 571 593 616 639 661 684 705 726 751 775 792 812 832 851 869 888 907 924 974 1019 1056 1087 1144
45
dT/dt (°C/s) 1.22 1.352 1.495 1.663 1.833 1.9 1.983 2.033 2.117 2.167 2.283 2.367 2.383 2.433 2.533 2.683 2.8 2.75 2.833 3 3.033 2.933 3.05 3.167 3.1 3.083 3.2 3.133 3.05 2.867 2.8 3.033 2.983 2.733 2.617 2.483 2.35 2.3 2.267 2.15 2.017 1.85 1.7 1.575 1.348 1.038 0.805 0.643 0.36
dP/dt (kPa/s) 3.533 4.4 5.267 4.8 6.65 6.967 8.133 9.033 10.017 11.183 13.667 14.383 15.25 16.383 17.5 19.5 21.167 21.667 23.167 24.333 26.333 27.167 28.167 29.5 30.167 31 32.667 33.5 34 34.5 35.333 30.167 30 39.167 36.333 33.833 32.167 30.167 28.333 28.5 28.333 27 27.167 25.5 22.333 18.333 15.633 13.883 9.617
Time (s) 1026.1 1036.3 1051.4
Temperature (°C) 218.3 219.6 219.8
Pressure (kPa) 1183 1227 1251
46
dT/dt (°C/s) 0.208 0.115 -0.009
dP/dt (kPa/s) 5.95 3.05 0.005
Table A5. PHI-TEC data for hydrolysis of acetic anhydride with a starting temperature of 303K (30°C) and 1% surfactant, run No PA83.
Time (s) 0 41.5 119.6 196.4 273.4 350.3 427.2 504.1 580.9 657.9 734.9 811.7 888.4 965.6 1042.4 1119.4 1160.4 1201.4 1242.4 1283.4 1324 1364.6 1405.6 1446.5 1487.5 1528.6 1569.6 1610.5 1651.4 1692.3 1733.3 1777.7 1818.7 1859.3 1900.3 1941.2 1982.2 2023.1 2064.1 2105 2146 2186.9 2227.9 2253.1 2278.3
Temperature (°C) 28.5 28.7 29.2 29.6 30.1 30.6 31.1 31.6 32.2 32.7 33.3 33.8 34.4 35 35.6 36.3 36.6 37 37.3 37.7 38.1 38.5 38.8 39.2 39.7 40.1 40.5 41 41.5 42.1 42.7 43.4 44.1 44.8 45.6 46.4 47.3 48.2 49.2 50.3 51.5 52.8 54.1 55 56
Pressure (kPa) 116 115 115 115 114 114 114 114 114 114 114 114 115 115 115 116 115 116 116 116 117 117 117 118 118 118 119 119 119 120 121 122 122 123 123 124 124 125 126 127 128 130 131 132 133
47
dT/dt (°C/s) 0 0.006 0.006 0.006 0.006 0.006 0.007 0.007 0.007 0.007 0.007 0.008 0.008 0.008 0.008 0.008 0.009 0.009 0.009 0.009 0.009 0.009 0.01 0.01 0.01 0.011 0.011 0.012 0.013 0.015 0.015 0.016 0.017 0.019 0.02 0.021 0.022 0.024 0.025 0.027 0.029 0.032 0.035 0.037 0.038
dP/dt (kPa/s) 0 -0.022 -0.018 -0.002 -0.001 0.002 -0.004 -0.002 0.004 0.001 0 0.004 0.005 0.005 0.002 0 0.002 0.009 0.009 0.008 0.006 0.005 0.012 0.012 0.011 0.009 0.004 0.006 0.012 0.021 0.021 0.021 0.009 0.008 0.01 0.011 0.017 0.008 0.033 0.011 0.039 0.021 0.029 0.033 0.046
Time (s) 2303.5 2328.7 2353.9 2379.1 2404.3 2429.5 2454.7 2479.9 2505.1 2530.3 2545.3 2560.5 2575.7 2590.9 2606.1 2621.3 2636.4 2646.6 2656.7 2666.8 2676.9 2687.1 2692.1 2697.2 2702.3 2707.3 2712.4 2717.4 2722.5 2727.6 2729.6 2731.7 2733.8 2735.8 2737.9 2739.9 2742 2744.1 2746.1 2748.2 2750.3 2752.3 2754.4 2756.4 2758.5 2760.6 2762.6 2764.7 2765.3
Temperature (°C) 57 58.1 59.3 60.5 61.9 63.4 65 66.7 68.7 70.8 72.3 73.8 75.5 77.4 79.5 81.8 84.4 86.4 88.5 91 93.6 96.7 98.4 100.3 102.1 104.5 107 109.7 112.7 116.1 117.6 119.2 120.9 122.7 124.6 126.6 128.7 130.9 133.2 135.7 138.4 141.2 144.2 147.2 150.4 153.8 157.2 160.6 161.5
Pressure (kPa) 134 135 136 138 139 141 143 145 147 150 153 155 157 160 163 166 171 174 177 182 187 193 197 202 206 210 217 224 231 240 244 248 253 259 265 270 277 286 294 303 313 324 336 351 366 383 403 426 433
48
dT/dt (°C/s) 0.043 0.045 0.048 0.051 0.056 0.061 0.067 0.073 0.082 0.089 0.101 0.104 0.118 0.13 0.151 0.161 0.183 0.197 0.228 0.19 0.275 0.337 0.343 0.383 0.52 0.462 0.505 0.567 0.615 0.713 0.76 0.795 0.843 0.897 0.95 1 1.04 1.095 1.17 1.258 1.352 1.397 1.448 1.478 1.633 1.663 1.637 1.667 1.7
dP/dt (kPa/s) 0.066 0.057 0.026 0.057 0.052 0.062 0.075 0.078 0.114 0.082 0.225 -0.002 0.532 0.248 0.117 0.492 0.395 -0.07 0.27 0.505 0.362 1.213 1.022 1.137 1.023 0.542 1.638 1.482 1.598 2.417 1.967 2.083 2.633 2.917 2.733 2.633 3.95 4.133 4.133 4.467 5.183 5.633 6.85 7.183 7.367 8.95 11.2 11.383 11.45
Time (s) 2765.8 2766.4 2766.9 2767.5 2768.1 2770.1 2772.2 2774.3 2776.3 2778.4 2780.4 2782.5 2784.6 2786.6 2788.7 2790.8 2792.8 2797.9 2802.9 2813.1 2823.2 2838.4 2863.5 2904.2 2996.9
Temperature (°C) 162.5 163.4 164.4 165.3 166.3 169.7 172.9 176.1 179.2 181.9 184.5 186.9 189 190.8 192.5 194 195.3 198 199.9 202.4 203.8 204.9 205.7 206.1 206
Pressure (kPa) 439 446 454 462 470 499 530 562 594 626 656 687 716 744 769 792 813 859 888 942 972 997 1012 1015 1008
49
dT/dt (°C/s) 1.7 1.7 1.683 1.667 1.657 1.638 1.565 1.502 1.395 1.305 1.192 1.078 0.945 0.86 0.76 0.68 0.597 0.427 0.322 0.18 0.096 0.052 0.012 0.006 -0.003
dP/dt (kPa/s) 12.317 13.317 13.433 14.033 14.517 14.633 15.167 15.1 15.983 14.95 14.817 14.5 14.133 12.7 12.25 9.733 10.417 7.817 10.767 3.3 2.5 1.023 0.018 -0.067 -0.067
Table A6. PHI-TEC data for hydrolysis of acetic anhydride with a starting temperature of 298K (25°C) and 0.3% surfactant, run No PA91. Time (s) 0 18.3 60.9 137.6 214.8 290.6 366.7 443.7 520 596.4 675.3 751.4 830.4 907 985.4 1062.7 1137.9 1214.6 1294.6 1373.2 1449.7 1524.9 1600.3 1676.8 1720.2 1762 1802.7 1847 1888.1 1928.3 1968.5 2008.9 2050.3 2091 2133.1 2174.1 2214.4 2255 2296 2336.9 2377.2 2417.6 2457.7 2498.7 2539.7
Temperature (°C) 25.4 25.5 25.9 26.5 27 27.4 27.9 28.4 28.8 29.2 29.7 30.2 30.7 31.2 31.7 32.2 32.7 33.3 33.8 34.4 35 35.6 36.2 36.9 37.3 37.7 38 38.4 38.9 39.3 39.8 40.3 40.9 41.5 42.1 42.8 43.5 44.2 45 45.8 46.6 47.6 48.5 49.6 50.7
Pressure (kPa) 102 102 101 100 99 99 98 98 98 98 98 98 98 98 98 98 98 99 99 99 99 100 101 101 102 102 102 102 102 103 103 103 104 104 105 105 106 106 107 108 108 109 110 110 111
50
dT/dt (°C/s) 0 0.009 0.008 0.007 0.006 0.006 0.006 0.006 0.006 0.006 0.006 0.006 0.006 0.006 0.007 0.007 0.007 0.007 0.007 0.008 0.008 0.008 0.008 0.009 0.009 0.009 0.009 0.01 0.01 0.011 0.013 0.014 0.014 0.016 0.015 0.017 0.018 0.018 0.019 0.021 0.022 0.023 0.025 0.027 0.029
dP/dt (kPa/s) 0 -0.037 -0.004 -0.016 0 -0.004 -0.006 -0.003 -0.002 -0.001 0 -0.001 0.001 0.006 0.005 -0.002 -0.001 0.004 0.005 0.001 0.005 0.014 0.009 0.001 0.005 0.008 0.009 -0.002 0.003 0.007 0.004 0.018 0.019 0.004 -0.001 0.016 0.022 0.001 0.03 0.021 -0.004 0.026 0.007 0.03 0.002
Time (s) 2580.7 2621.7 2647.7 2673.7 2699.7 2725.3 2751.3 2777.3 2803.3 2829.2 2854.3 2880.3 2905.7 2930.7 2945.8 2960.9 2976 2991 3006.1 3021.2 3036.3 3046.4 3056.5 3066.5 3076.6 3086.7 3091.8 3096.9 3102 3107.1 3112.2 3117.3 3122.4 3124.5 3126.6 3128.7 3130.8 3132.9 3135 3137.1 3139.2 3141.3 3141.9 3142.5 3143.1 3143.7 3144.3 3144.9 3145.5
Temperature (°C) 51.9 53.3 54.2 55.1 56.2 57.2 58.4 59.6 61 62.4 64 65.7 67.6 69.7 71.1 72.6 74.2 76 78 80.2 82.7 84.6 86.7 89 91.8 94.8 96.6 98.6 100.7 103.2 105.9 109.1 112.8 114.4 116.3 118.2 120.4 122.7 125.3 128.1 131.2 134.5 135.5 136.6 137.7 138.8 140 141.1 142.3
Pressure (kPa) 112 113 114 116 117 118 118 120 121 122 124 125 126 130 131 134 135 136 139 143 145 148 152 155 159 164 167 170 173 179 184 191 198 202 207 211 213 222 229 244 250 256 259 262 266 270 274 279 284
51
dT/dt (°C/s) 0.032 0.034 0.036 0.038 0.041 0.041 0.046 0.05 0.054 0.059 0.065 0.071 0.078 0.087 0.093 0.103 0.12 0.112 0.132 0.147 0.178 0.21 0.218 0.258 0.3 0.347 0.352 0.408 0.43 0.507 0.58 0.65 0.785 0.827 0.903 0.972 1.053 1.172 1.272 1.395 1.545 1.683 1.733 1.783 1.85 1.917 1.95 1.9 2.067
dP/dt (kPa/s) 0.04 0.043 0.045 0.054 0.028 0.031 0.04 0.019 0.052 0.085 0.053 -0.016 0.222 0.25 0.07 0.437 -0.42 0.373 0.378 0.44 0.37 0.093 -0.092 0.385 0.6 0.808 0.723 0.722 0.917 1.015 1.7 1.045 1.867 2.017 2.583 1.867 3.817 3.383 3.6 14.283 0.685 4.633 5.1 5.583 6.25 6.6 7.75 8.067 8.467
Time (s) 3146 3146.6 3147.2 3147.8 3148.4 3149 3149.6 3150.2 3150.8 3151.4 3152 3152.6 3153.2 3153.8 3154.4 3155 3155.5 3156.1 3156.7 3157.3 3157.9 3158.5 3159.1 3159.7 3160.3 3160.9 3161.5 3162.1 3162.7 3163.3 3163.9 3166 3168.1 3170.2 3172.3 3174.4 3176.5 3181.6 3186.7 3196.8 3212.2
Temperature (°C) 143.3 144.6 145.9 147.2 148.5 149.8 151.1 152.5 153.9 155.3 156.7 158.2 159.6 161 162.4 163.8 165 166.4 167.8 169.2 170.5 171.8 173.1 174.4 175.6 176.8 178 179.1 180.2 181.2 182.2 185.5 188.2 190.5 192.5 194.1 195.5 197.9 199.5 201.2 202.1
Pressure (kPa) 288 294 300 307 314 322 330 338 348 357 367 377 387 399 410 421 431 443 454 466 479 493 505 517 528 540 552 564 575 585 596 633 666 695 721 743 764 802 834 873 896
52
dT/dt (°C/s) 2.083 2.1 2.15 2.217 2.217 2.2 2.267 2.317 2.35 2.367 2.35 2.383 2.4 2.35 2.367 2.367 2.35 2.3 2.283 2.25 2.2 2.2 2.133 2.1 2.017 1.933 1.917 1.8 1.75 1.7 1.647 1.435 1.207 1 0.835 0.713 0.573 0.385 0.275 0.123 0.008
dP/dt (kPa/s) 9.517 10.067 11.583 11.633 12.967 13.4 13.217 14.483 16.417 14.967 17.5 16.367 19.833 18.833 18.667 19 20.167 19.5 19.667 20.5 24.667 22.5 17.333 19.5 19 20 19.667 19.167 17.167 17.833 17.667 17.167 14.15 14.067 11.217 9.917 9.433 6.15 5.333 2.067 0.937
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
DISHED HEAD: 1000mm INSIDE DIAMETER 800mm SPHERICAL RADIUS 150mm KNUCKLE RADIUS 25mm STRAIGHT FLANGE 15mm THICK 11.5mm MIN. THK A.F.
N10 N8 N1
TYP
N9
N7 3 TYP
5
N3 N5
TYP
TYP
2 TYP TYP
N4 N6
5
DISHED HEAD: 1000mm INSIDE DIAMETER 800mm SPHERICAL RADIUS 150mm KNUCKLE RADIUS 25mm STRAIGHT FLANGE 10mm THICK 8mm MIN. THK A.F.
TYP
N2 VESSEL ELEVATION
Figure B1
Dimensioned sketch of reactor
54
Figure B2.
Figure B3.
Large-scale reactor
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
S1
S1
D1 S2
D2 D2 S2
D1
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 102
1.720m
2.951m
1.513m
DP 103
Figure B6.
DP 101
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 TE101 TE102 TE103 TE104 TE105 TE106 TE109 TE110 TE111 TE112 TE114
DESCRIPTION Thermocouple Thermocouple Thermocouple Thermocouple Thermocouple Thermocouple Resistance Temperature Device Resistance Temperature Device Resistance Temperature Device Thermocouple Thermocouple
TE115
Thermocouple
TE116
Thermocouple
TE117
Thermocouple
TE118 TE119 TE120 TE121 TE122 PT102 PT103 PT104 PT105
Thermocouple Thermocouple Resistance Temperature Device Resistance Temperature Device Resistance Temperature Device Pressure Transducer Pressure Transducer Pressure Transducer Pressure Transducer
PT106
Pressure Transducer
PT107
Pressure Transducer
PT108
Pressure Transducer
PT109 PT110 WT101 FT101 DT101 DT102 DT103 DT104 DT105 DT106 AV122 AV121
Pressure Transducer Pressure Transducer Load cells Flow Meter Densitometer Densitometer Densitometer Densitometer Densitometer Densitometer Valve State Valve State
62
LOCATION Reactor Reactor Reactor Reactor Reactor Reactor HX process inlet HX process outlet Water feed vessel Feed Vessel 6 Vent line (inside) Vent line (before bend) Vent line (after bend) Vent line (above dump tank) Dump tank top Dump tank bottom Pump inlet HX utility top HX utility bottom Manifold N9 Before pump After HX Vent line (inside) Vent line (before bend) Vent line (after bend) Vent line (above dump tank) Dump tank Before HX Vessel load cell After HX Position reqst. 1 Position reqst. 2 Pos. Conf. 1 Pos. Conf. 2 Detector 1 Detector 2 8" Reactor valve Feed Vessel valve
13.3
EXPERIMENTAL RESULTS Table C2. Large scale experimental conditions and main results
Critical Variables Post-mixing Temp. (K) Fill Level (%) Surfactant mass (kg) Surfactant (% wt/wt) Set Press. (kPa)
Experiment Number HP3 HP4 HP5
HP1
HP2
311.3 50 0 0 200
312.8 70 0 0 200
309.5 60 0 0 200
312.0 50 3.14 0.25 200
311.3 70 4.44 0.25 200
313.1 60 3.77 0.25 200
677.9
604.9
576.4
771.6
719.2
477.9 70.5
404.9 66.9
376.4 65.3
571.6 74.1
7.71
11.07
8.12
11.54
519.2 72.2 11.79
33.0
37.03
32.83
48.47
48.04
12.84
15.87
12.96
12.95
12.41
466.6
458.5
455.6
470.3
466.6
401.8
401.4
402.9
397.7
401.7
64.8
57.1
52.7
72.6
64.9
1.81
2.96
2.40
1.62
2.53
4.37
4.50
3.67
7.16
5.62
2.60
1.46
1.95
2.99
1.8
1604.2 1628.2 1629.7
1627.6 1648.9 1649.9
1509.1 1528.1 1532.4
1627.4 1647.2 1648.9
1443.0 1462.5 1463.5
Reactor Max. Press. (kPa) 360.6 PT102 Overpress. (kPa) 160.6 Overpress. (abs) % 44.5 Max. Press Rate before 6.94 vent op. (kPa s-1) Max. Press Rate after 12.89 vent op. (kPa s-1) Press Rate at vent op. 11.11 (kPa s-1) Max Temp. TE106 (K) 439.1 Temp. at vent op 403.2 TE106 (K) Overtemp. TE106 (K) 35.9 Max temp. rate before 1.86 vent op. (K s-1) Max temp. rate after 1.98 vent op. (K s-1) Temp. rate at vent op. 1.71 (K s-1) Times from initiation Vent op. (s) 1594.8 Max press. (s) 1637.0 Max temp. (TE106) (s) 1645.7
63
HP6
Table C2. (continued) Experiment Number HP4 HP5 HP3
HP6
HP1
HP2
Critical Variables Post-mixing Temp. (K) Fill Level (%) Surfactant mass (kg) Surfactant (% wt/wt) Set Press. (kPa)
311.3 50 0 0 200
312.8 70 0 0 200
309.5 60 0 0 200
312.0 50 3.14 0.25 200
311.3 70 4.44 0.25 200
313.1 60 3.77 0.25 200
Times relative to vent op. Max press. (s) Max temp. (TE106) (s) Max press. rate (s) Max temp. rate (s)
42.2 51.9 4.8 1.7
24.0 25.5 9.9 9.0
21.3 22.3 7.8 8.0
19.0 23.3 7.7 7.4
19.8 21.5 11.4 10.4
19.5 20.5 9.1 9.6
359.0
301.8
479.6
412.9
301.8
306.6
436.2
431.1
449.0
442.1
431.1
431.1
357.4
301.1
478.3
411.2
299.7
306.4
435.5
430.7
449.0
441.7
430.7
431.1
412.4
399.9
428.9
420.4
406.5
403.6
410.2
398.8
427.8
419.3
404.7
402.5
Vent Line Max. press (kPa) (PT108) Max. temp. (K) (TE114) Dump Tank Max. press (kPa) (PT109) Max. temp. of material entering.TE117 (K) Max. temp. of liquid (K) (TE119) Max. vapour space temp.TE118 (K)
64
Table C3. Pilot scale mass transfer results Experiment Number HP4 HP5 HP3
HP6
HP1
HP2
311.3 50 0 0 200
312.8 70 0 0 200
309.5 60 0 0 200
312.0 50 3.14 0.25 200
311.3 70 4.44 0.25 200
313.1 60 3.77 0.25 200
Charge Mass (kg) 1266.1 Mass remaining in 883.2 reactor (kg) Mass remaining in 382.9 (1) dump tank (kg) Duration of reactor 74.5 (2) venting (s) Av. mass discharge 5.1 rate (kg s-1) ((1)/(2)) Time from vent 1.8 opening to level swell (s) Duration of two-phase 26.2 flow (s)
1777.6
1520.6
1258.5
1779.2
1512.7
624.9
774.9
351.0
357.4
349.3
1152.7
745.7
907.5
1421.8
1163.4
62.0
49.4
33.4
54.9
46.1
18.6
15.1
27.2
25.9
25.2
0.3
1.6
2.1
1.4
1.9
47.7
32.6
29.4
35.5
44.3
Critical Variables Post-mixing Temp. (K) Fill Level (%) Surfactant mass (kg) Surfactant (% wt/wt) Set Press. (kPa)
65
Experiment HP1 CRITICAL VARIABLES
Relief set pressure
200 (2)
Orifice diameter
100
mm
Fill level
50
%
Surfactant concentration
0
% (wt/wt)
Experiment No: HP1
450
Date: 10 February 2005
kPa (bara)
File: HP1ev.opj
Temperature (K)
400
Heat exchanger top (TE121) Liquid (TE106) 350
Heat exchanger bottom (TE122) Vapour (TE101)
300 -2000
-1000
0
1000
2000
3000
Time (s)
Figure C2. Overall reactor temperature records
66
450
Experiment No: HP1
Date: 10 February 2005
File: HP1ev.opj
300
Pressure (PT102)
350
Liquid (TE106) 200
Pressure (kPa)
Temperature (K)
400
400
Vapour (TE101)
300
100
Vent open 250 500
1000
1500
2000
2500
Time (s)
Figure C3. Temperature and pressure profiles in the reactor during venting
16
Experiment No: HP1
Date: 10 February 2005
File: HP1ev.opj
400
300 14
Dump tank pressure (PT109) Reactor top to bottom differential pressure (DPT101)
200
12
Pressure (kPa)
Differential pressure (kPa)
Reactor pressure (PT102)
100
Vent open 10 1600
1700
Time (s)
Figure C4. Pressure profile in the reactor and dump tank during venting
67
Experiment No: HP1
Date: 10 February 2005
File: HP1ev.opj
80000
Densitometer (cps)
60000
40000
20000
Gamma ray detector 1 (DT105)
Vent open 0
1500
1550
1600
1650
1700
Time (s)
Figure C5. Density profile in the reactor during venting
Experiment No: HP1
Date: 10 February 2005
File: HP1ev.opj
Inlet temp. (TE114)
Vent open 400
300
200
350
Inlet pressure (PT105)
Outlet pressure (PT108)
Outlet temp. (TE117)
300
Pressure (kPa)
Temperature (K)
400
100
Note: Inlet instrumentation located downstream of orifice plate.
0 1000
1500
2000
2500
Time (s)
Figure C6. Vent line temperature and pressure profiles during venting
68
Experiment No: HP1
Date: 10 February 2005
File: HP1ev.opj
Discharge from vent line (TE117)
Vent open 400
Temperature (K)
400
300
Liquid temp. (TE119) 350 200
Vapour temp. (TE118)
300
Pressure (PT109)
Pressure (kPa)
450
100
250 1000
1500
2000
2500
Time (s)
Figure C7. Dump tank temperature and pressure profiles during venting
69
Experiment HP2 CRITICAL VARIABLES
Relief set pressure
200 (2)
Orifice diameter
100
mm
Fill level
70
%
Surfactant concentration
0
% (wt/wt)
Experiment No: HP2
480
Date: 17 February 2005
kPa (bara)
File: HP2ev.opj
Temperature (K)
Liquid (TE106) Vapour (TE101)
420
Heat ex. top (TE121)
360 Heat exchanger bottom (TE122)
300 0
1000
2000
3000
Time (s)
Figure C8. Overall reactor temperature records
70
Experiment No: HP2
Date: 17 February 2005
File: HP2ev.opj
475
700 600
Vent open
425 500 400 Vapour
400
(TE101)
375
300 350
Pressure
Liquid (TE106)
200
(PT102)
325
Pressure (kPa)
Temperature (K)
450
100
300 275
0 1000
1500
2000
Time (s)
Figure C9. Temperature and pressure profiles in the reactor during venting
24
Experiment No: HP2
Date: 17 February 2005
File: HP2ev.opj
600
Reactor top to bottom
Reactor pressure
differential pressure
18
(PT102)
(DPT101)
500 400 300
12
Pressure (kPa)
Differential Pressure (kPa)
700
200
Dump tank pressure (PT109)
100 6 1550
Vent open
1600
1650
0 1700
Time (s)
Figure C10. Pressure profile in the reactor and dump tank during venting
71
Experiment No: HP2
Date: 17 February 2005
File: HP2ev.opj
Densitometer (cps)
80000
60000
40000
20000 Gamma ray detector 1
0
(DT105)
1500
Vent open
1550
1600
1650
1700
1750
Time (s)
Figure C11. Density profile in the reactor during venting
Experiment No: HP2
Date: 17 February 2005
Inlet temp.
Vent open
425
File: HP2ev.opj
(TE114)
400
Temperature (K)
300
250 Outlet temp.
375
(TE117)
200
350 150
325 Outlet pressure (PT108)
Pressure (kPa)
450
Inlet pressure
300
(PT105)
100
Note: Inlet instrumentation located downstream of orifice plate.
275
50 1000
1500
2000
Time (s)
Figure C12. Vent line temperature and pressure profiles during venting
72
Experiment No: HP2
450
Date: 17 February 2005
File: HP2ev.opj
Discharge from vent Vent open
425
line (TE117)
250 Liquid temp.
375
(TE119)
200
350 Pressure
325
(PT109)
150
Pressure (kPa)
Temperature (K)
400
300
Vapour temp.
300
(TE118)
100
275 1000
1500
2000
Time (s)
Figure C13. Dump tank temperature and pressure profiles during venting
73
Experiment HP3 CRITICAL VARIABLES
Relief set pressure
200 (2)
Orifice diameter
100
mm
Fill level
60
%
Surfactant concentration
0
% (wt/wt)
Experiment No: HP3
Date: 03 March 2005
kPa (bara)
File: HP3ev.opj
450
Temperature (K)
Vapour (TE101)
Liquid (TE105)
400 Heat exchanger top (TE121)
350
Heat exchanger bottom (TE122)
300 0
1000
2000
3000
Time (s)
Figure C14. Overall reactor temperature records
74
Experiment No: HP3
Date: 03 March 2005
File: HP3ev.opj
600
Vent open
450 Vapour
500
400
400
Pressure (PT102)
300
Liquid (TE105)
350
Pressure (kPa)
Temperature (K)
(TE101)
200
100
300 1000
1500
2000
2500
Time (s)
Figure C15. Temperature and pressure profiles in the reactor during venting
20
Experiment No: HP3
Date: 03 March 2005
File: HP3ev.opj
Reactor pressure (PT102)
16 400 14
Vent open Reactor top to bottom
12
differential pressure (DPT101)
10
Pressure (kPa)
Differential pressure (kPa)
18
600
200
Dump tank pressure (PT109)
8 1600
1625
1650
1675
1700
Time (s)
Figure C16. Pressure profile in the reactor and dump tank during venting
75
Experiment No: HP3
Date: 03 March 2005
File: HP3ev.opj
Densitometer (cps)
80000
60000
40000 Gamma ray detector 1 (DT105)
20000
Vent open
0
1500
1550
1600
1650
1700
1750
Time (s)
Figure C17. Density profile in the reactor during venting Experiment No: HP3
Date: 03 March 2005
File: HP3ev.opj
Vent open
450
500
400
400
300
350 Outlet pressure (PT108)
Pressure (kPa)
Temperature (K)
Inlet temp. (TE114)
200
300
Inlet pressure (PT105) Note: Inlet instrumentation located downstream of orifice plate.
Outlet temp. (TE117)
100
250 1000
1500
2000
2500
Time (s)
Figure C18. Vent line temperature and pressure profiles during venting
76
Experiment No: HP3
Date: 03 March 2005
Vent open
450
File: HP3ev.opj
Discharge from vent line
500
400
Liquid temp.
400
(TE119)
300
350
300
Pressure
Vapour temp.
(PT109)
(TE118)
Pressure (kPa)
Temperature (K)
(TE117)
200
100 250 1000
1500
2000
2500
Time (s)
Figure C19. Dump tank temperature and pressure profiles during venting
77
Experiment HP4 CRITICAL VARIABLES
Relief set pressure
200 (2)
Orifice diameter
100
mm
Fill level
50
%
Surfactant concentration
0.25
Experiment No: HP4
Date: 21 March 2005
kPa (bara)
% (wt/wt)
File: HP4ev.opj
450
Temperature (K)
Liquid (TE106)
400 Heat ex. top (TE121) Heat exchanger
350
bottom (TE122)
300 Vapour (TE101)
-500
0
500
1000
1500
2000
2500
Time (s)
Figure C20. Overall reactor temperature records
78
Experiment No: HP4
Date: 21 March 2005
File: HP4ev.opj
460
600
Vent open
440
Temperature (K)
400
360
400
Vapour
380
(TE101) Liquid (TE106)
300 Pressure
340
(PT102)
Pressure (kPa)
500
420
200 320 300
100
280 1000
1500
2000
Time (s)
Figure C21. Temperature and pressure profiles in the reactor during venting
Experiment No: HP4
Date: 21 March 2005
File: HP4ev.opj
600 15 500
(PT102)
Vent open
400 10 Reactor top to bottom
300
differential pressure (DPT101)
Pressure (kPa)
Differential Pressure (kPa)
Reactor pressure
200
Dump tank pressure
5
(PT109)
1500
1530
100 1560
Time (s)
Figure C22. Pressure profile in the reactor and dump tank during venting
79
Experiment No: HP4
Date: 21 March 2005
File: HP4ev.opj
Densitometer (cps)
80000
60000
40000 Vent open
20000 Gamma ray detector 1 (DT105)
0
1400
1450
1500
1550
1600
Time (s)
Figure C23. Density profile in the reactor during venting
Experiment No: HP4
Date: 21 March 2005
File: HP4ev.opj
460 Vent open
440
Inlet temp.
400
(TE114)
420 300
380
Outlet pressure (PT108)
360
200
340 320 300
Outlet temp.
Pressure (kPa)
Temperature (K)
400
100
(TE117)
Inlet pressure (PT105)
Note: Inlet instrumentation located downstream of orifice plate.
280
0 1000
1500
2000
Time (s)
Figure C24. Vent line temperature and pressure profiles during venting
80
Experiment No: HP4
Date: 21 March 2005
File: HP4ev.opj
460 Discharge from vent line
Vent open
(TE117)
420
400
Temperature (K)
400
Liquid temp. (TE119)
380
300 Vapour temp. (TE118)
360 340
200
320
Pressure (kPa)
440
500
Pressure (PT109)
300
100 280 1250
1500
1750
2000
2250
Time (s)
Figure C25. Dump tank temperature and pressure profiles during venting
81
Experiment HP5 CRITICAL VARIABLES
Relief set pressure
200 (2)
Orifice diameter
100
mm
Fill level
70
%
Surfactant concentration
0.25
Experiment No: HP5
Date: 31 March 2005
kPa (bara)
% (wt/wt)
File: HP5ev.opj
480 460
Liquid (TE106)
Temperature (K)
440 420 400 380 360
Heat ex. top
Heat exchanger
(TE121)
bottom (TE122)
340 320
Vapour (TE101)
300 280 0
500
1000
1500
2000
2500
Time (s)
Figure C26. Overall reactor temperature records
82
Experiment No: HP5
Date: 31 March 2005
File: HP5ev.opj
480
800 Vent open
460
700 600
Liquid (TE106)
420
500
400 Vapour (TE101)
380
400
360
300
Pressure (PT102)
340
Pressure (kPa)
Temperature (K)
440
200 320 100
300
0
280 1000
1500
2000
Time (s)
Figure C27. Temperature and pressure profiles in the reactor during venting
Experiment No: HP5
File: HP5ev.opj
Reactor top to bottom differential pressure
600
(DPT101)
Reactor pressure (PT102)
15
Vent open
10
300
Pressure (kPa)
Differential pressure (kPa)
20
Date: 31 March 2005
Dump tank
5
pressure (PT109)
0 1580
1600
1620
1640
1660
1680
1700
1720
Time (s)
Figure C28. Pressure profile in the reactor and dump tank during venting
83
Experiment No: HP5
Date: 31 March 2005
File: HP5ev.opj
Densitometer (cps)
80000
60000
40000
Gamma ray detector 1
20000
(DT105)
0
Vent open
1550
1600
1650
1700
1750
Time (s)
Figure C29. Density profile in the reactor during venting
Experiment No: HP5
Date: 31 March 2005
File: HP5ev.opj
350
440 Vent open
Inlet temp. (TE114)
420
300
250
380
Outlet pressure (PT108)
360
200
340 320
150
Inlet pressure (PT105)
Pressure (kPa)
Temperature (K)
400
Outlet temp.
300
(TE117)
280
Note: Inlet instrumentation located downstream of orifice plate.
260 1000
100
50 1500
2000
Time (s)
Figure C30. Vent line temperature and pressure profiles during venting
84
Experiment No: HP5
Date: 31 March 2005
File: HP5ev.opj
440 Vent open
Pressure (PT109)
300
420
Vapour temp. (TE118)
380 360
Liquid
250
200
(TE119)
340 320
150
Discharge from
300
Pressure (kPa)
Temperature (K)
400
vent line (TE117)
280
100
260 1500
2000
Time (s)
Figure C31. Dump tank temperature and pressure profiles during venting
85
Experiment HP6 CRITICAL VARIABLES
Relief set pressure
200 (2)
Orifice diameter
100
mm
Fill level
60
%
Surfactant concentration
0.25
Experiment No: HP6
Date: 19 May 2005
kPa (bara)
% (wt/wt)
File: HP6ev.opj
450
Temperature (K)
Liquid (TE106)
400
Heat exchanger bottom (TE122)
350
Vapour (TE101) 300
Heat exchanger top (TE121)
-1000
-500
0
500
1000
1500
2000
2500
Time (s)
Figure C32. Overall reactor temperature records
86
3000
Experiment No: HP6
Date: 19 May 2005
File: HP6ev.opj
800
450
400
Liquid (TE106)
400
350
Pressure (PT102)
200
Vapour (TE101)
300
Pressure (kPa)
Temperature (K)
600
Vent open 250 500
1000
1500
2000
2500
Time (s)
Figure C33. Temperature and pressure profiles in the reactor during venting
Experiment No: HP6
Date: 19 May 2005
File: HP6ev.opj
Reactor top to bottom differential pressure (DPT101)
18 16
Differential pressure (kPa)
800
600
Reactor pressure (PT102)
14 12
400 10 8
Dump tank pressure (PT109)
6 4
Pressure (kPa)
20
200
Vent open
2 1450
1500
1550
Time (s)
Figure C34. Pressure profile in the reactor and dump tank during venting
87
Experiment No: HP6
Date: 19 May 2005
File: HP6ev.opj
Densitometer (cps)
80000
60000
40000
Vent open
20000
Gamma ray detector 1 (DT105) 0
1400
1450
1500
1550
Time (s)
Figure C35. Density profile in the reactor during venting
450
Experiment No: HP6
Date: 19 May 2005
File: HP6ev.opj
400
Vent open
Temperature (K)
Inlet temp. (TE114) 200
350
Outlet pressure (PT108)
Inlet pressure (PT105)
300
Outlet temp. (TE117)
Pressure (kPa)
300
400
100
Note: Inlet instrumentation located downstream of orifice plate. 1000
1250
1500
1750
2000
2250
Time (s)
Figure C36. Vent line temperature and pressure profiles during venting
88
Experiment No: HP6
Vent open
Date: 19 May 2005
File: HP6ev.opj
400
Discharge from vent line (TE117) 300
Temperature (K)
400
Liquid temp. (TE119) 200 350
Pressure (PT109)
1000
100
Vapour temp. (TE118)
300
1250
1500
1750
Pressure (kPa)
450
2000
2250
Time (s)
Figure C37. Dump tank temperature and pressure profiles during venting
89
Experiment blow1 CRITICAL VARIABLES
Relief set pressure
480 (4.8)
Orifice diameter
100
Surfactant concentration
0
kPa (bara) mm % (wt/wt)
1.0 upper source
Differential pressure trans. top - halfway mid top - mid bottom top - bottom halfway - bottom
Void fraction
0.8
0.6
0.4 lower source
0.2
0.0 -80
-60
-40
-20
0
20
40
60
80
100
Time from vent opening (s)
Figure C38. Void fraction - acetic acid blow down
90
120
Experiment blow2 CRITICAL VARIABLES
Relief set pressure
480 (4.8)
kPa (bara)
Orifice diameter
100
mm
Surfactant concentration
0.25
% (wt/wt)
1.0
upper source
Void fraction
0.8
0.6
Differential pressure trans. top - halfway top - bottom mid top - mid bottom bottom - halfway
0.4
0.2
lower source
0.0 -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. 14.1
APPENDIX D: DERIVATION OF VOID FRACTIONS
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:
⎛ I ⎞ ⎛µ⎞ ln⎜⎜ ⎟⎟ = − µ ⋅ L = −⎜⎜ ⎟⎟ ⋅ ρ ⋅ L ⎝ρ⎠ ⎝ I0 ⎠
(1)
⎛µ⎞
where I and I 0 are the measured and initial intensities, and µ and ⎜⎜ ⎟⎟ are the material’s ⎝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 i ⋅µ i
(2)
i
⎛µ⎞ ⎜ ⎟ = ⎜ρ⎟ ⎝ ⎠ mix
⎛
⎞
⎝
⎠i
∑ w i ⋅⎜⎜ µρ ⎟⎟ i
(3)
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 ln⎜⎜ ⎝ IO
⎡⎛ ⎛ µ ⎞ ⎤ ⎞ ⎞ ⎛µ⎞ ⎛µ⎞ ⎟⎟ = − ⎢⎜ ⎜⎜ ⎟⎟ ⋅ ρ v ⋅ α + ⎜⎜ ⎟⎟ ⋅ ρ f ⋅ (1 − α )⎟ ⋅ LR + ⎜⎜ ⎟⎟ ⋅ ρ B ⋅ LB ⎥ ⎟ ⎢⎣⎜⎝ ⎝ ρ ⎠ v ⎥⎦ ⎝ ρ ⎠f ⎝ ρ ⎠B ⎠ ⎠
92
(4)
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
⎛µ⎞ ⎛µ⎞ ⎟⎟ , and liquid, ⎜⎜ ⎟⎟ , may be calculated according to their composition using ⎝ ρ⎠f ⎝ ρ ⎠V
vapour, ⎜⎜
Equation (3). Analogous equations may be written for the vessel empty, Equation (5), and full of acetic acid, Equation (6):
⎛I ln⎜⎜ E ⎝ Io
⎞ ⎛µ⎞ ⎛µ⎞ E ⎟⎟ = −⎜⎜ ⎟⎟ ⋅ ρ air ⋅ LR − ⎜⎜ ⎟⎟ ⋅ ρ B ⋅ LB ⎝ ρ ⎠B ⎝ ρ ⎠ air ⎠
(5)
⎛I ln⎜⎜ F ⎝ Io
⎞ ⎛µ⎞ ⎛µ⎞ F ⎟⎟ = −⎜⎜ ⎟⎟ ⋅ ρ acd ⋅ LR − ⎜⎜ ⎟⎟ ⋅ ρ B ⋅ LB ⎝ ρ ⎠B ⎝ ρ ⎠ acd ⎠
(6)
E F where ρ air is the density of air when the reactor is scanned empty and ρ acd 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:
α=
⎛ I ln⎜⎜ ⎝ IF ⎛ IE ⎜ ln ⎜
⎝ IF
⎞
⎟⎟
⎠ ⋅ ⎛⎜ ⎛⎜ µ ⎞⎟ ⋅ ρ E − ⎛⎜ µ ⎞⎟ ⋅ ρ F ⎞⎟ − ⎛⎜ ⎜⎛ µ ⎞⎟ ⋅ ρ − ⎛⎜ µ ⎞⎟ ⋅ ρ F ⎞⎟ air acd f acd ⎜ρ⎟ ⎜ρ⎟ ⎟ ⎟ ⎜ ⎜⎝ ρ ⎟⎠ f ⎞ ⎜⎝ ⎜⎝ ρ ⎠⎟ air ⎝ ⎠ ⎝ ⎠ acd acd ⎝ ⎠ ⎠ ⎟⎟ ⎠
⎛⎛ µ ⎞ ⎞ ⎜ ⎜ ⎟ ⋅ ρ − ⎜⎛ µ ⎞⎟ ⋅ ρ ⎟ v f ⎜ρ⎟ ⎜ ⎜⎝ ρ ⎠⎟ ⎟ ⎝ ⎠f v ⎝ ⎠
(7)
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:
α sim
⎛ I ln⎜⎜ I = ⎝ F ⎛I ln⎜⎜ E ⎝ IF
⎞ ⎟⎟ ⎠ ⎞ ⎟⎟
⎠
(8)
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:
∆P g ⋅h α= ρ f − ρv
ρf −
(9)
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):
∑ wi ⋅ρ fi
ρ mix =
(10)
i
ρ = 1 A
1 ⋅ (B ⋅ (T −C )+ D )
(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:
ρv =
P ⋅ Mv ⋅ 1000 R ⋅T
(12)
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 ⎪⎪ T − T add Z=⎨ − T T add
⎪ max ⎪⎩1
t add > t t add ≤ t ≤ t max
(13)
t > t max
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:
A=
mo ⋅ q
⎡⎛ V ⋅ h fg G ⋅ ⎢⎜ ⎜ ⎢⎝ mo ⋅ v fg ⎣
⎞ ⎟ ⎟ ⎠
1/ 2
⎤ 1/ 2 + (C f ⋅ ∆T ) ⎥ ⎥ ⎦
(1)
2
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:
t emptying =
mo G⋅A
⎛⎛ V ⋅ h fg ⎜⎜ ⎜⎜ m ⋅ v ⎝ o fg =⎝
⎞ ⎟ ⎟ ⎠
1/ 2
⎞ 1/ 2 ⎟ + (C f ⋅ ∆T ) ⎟ ⎠ q
2
(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 nonconservative 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 twophase 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 pilotscale 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.
Final reaction mass
(kPa) 200.2
Maximum temperature (K) 360.8
25
200.8
364.2
145.1
17.5
436.8
394.7
147.5
Vent diameter
Maximum pressure
(mm) 35
(kg) 109.6
Methanol/acetic anhydride 250 litres, R.S.P. 200 kPa, 25 mm vent
380
vent open
200
Pressure
370
360 160
Liquid temperature 350
140 340
Pressure (KPa)
Temperature (K)
180
120 330
Vapour temperature 320 1400
1500
1600
1700
100 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 500
400
vent open 400
pressure liquid temperature
300
360
200
Pressure (KPa)
Temperature (K)
380
340
vapour temperature
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 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 440
400
vent open Temperature with surfactant no susrfacatnt
350
400
300
380
250
Pressure with surfactant no surfactant
360
200
340
150
320
100
300 1570
1580
1590
1600
Pressure (kPa)
Temperature (K)
420
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
6
-1
Log (dT/dt) (K min )
no surfactant with surfactant 4
2
0
-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
detectors rotating source holders
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
668 mm 800mm
DP 103
Figure 7
800mm
2956mm
DP 102
DP 101
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
50 %
Maximum Pressure (bara) With surfactant No surfactant 3.60 5.76
60 %
6.00
7.19
70 %
6.78
7.71
108
16
vent open
12
160
attenuation (top)
10
170
differential pressure 150 top - bottom mid top - mid bottom 140
8
130
6
end of twophase flow
4
120
2
110
pressure
0 3140
3160
3180
3200
Tenmperature (°C)
Differential pressure (kPa) Pressure (bara), Attenuation
14
180
temperature bottom top
100 3240
3220
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.
16
end of two-phase flow
12
160
vent open differantial pressure top - bottom mid top - mid bottom
10 8 6
temperature bottom top
80
Temperature (°C)
Differential pressure (kPa) Pressure (bara), Attenuation
14
4
pressure
2
attenuation (top) 0 2020
0 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
20
220
18
200 180
vent open
differential pressure top - bottom mid top - mid bottom
14 12
160 140 120
10 100
pressure
8
80
6
60
4
temperature bottom top
2
40
attenuation (top)
0 1940
1960
1980
Temperature (°C)
Attenuation / pressure (kPa)
16
2000
20 0 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 Pmax Increase in Mass discharged heat rate/kg
50% fill no surfactant 50% fill with surfactant 60% fill no surfactant
Total mass discharged
Duration of two phase flow
Time from vent open to Pmax
(%)
(%)
(%)
(sec)
(sec)
27.0
129.7
30.2
26.2
42.2
29.6
161.7
72.1
29.4
19.0
33.2
162.0
49.0
32.6
21.3
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 vapouronly 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
heat generation adiabatic data heat removal no surfactant with surfactant
10000
Heat rate (W/kg)
8000
6000
maximum adiabtic rate
turnaround due to tempering, heat removal reactant consumption exceeds heat generation
4000
maximum temperature maximum temperature (tempered) (untempered)
vent open 2000
0 400
410
420
430
440
450
460
470
480
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
acetic anhydride hydolysis temperatures at Pmax(allowable)
-1
ln(dT/dt) (K min )
1
at Pset
0 -1 -2
∆Tad1
-3
∆Tad2 ∆Tad3
-4
∆T -5 -3.2
-3.0
-2.8
-2.6
-2.4
-2.2
-2.0
-1.8
-1.6
-1
-1000/T (K )
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.
⎞ ⎛ ⎟
⎜ ⎟
mo ⎜ 1 A= ⎟ ⎜1 − Gt ad ⎜ ⎛ C f (To + ∆Tad − Tmax ) ⎞ ⎟ ⎟⎟ ⎜⎜ exp⎜⎜ ⎟⎟ h x fg ⎠⎠ ⎝ ⎝
(3)
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, ventsizing 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 ventsizing 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.
2. 3.
4.
5.
6.
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 H G Fisher et al., "Emergency Relief System Design Using DIERS Technology", DIERS/AIChE, 1992, ISBN 0-8169-0568-1 J C Leung, "Simplified Vent Sizing Equations for Emergency Relief Requirements in Reactors and Storage Vessels", AIChE Journal, 32, (10), 1622-1634, 1986 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 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 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
Health and Safety Executive
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 G1RD2001 00499), 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
