Thermal Hazard Testing of Processes
Prof. Attilio Citterio
Dipartimento CMIC “Giulio Natta”
http://iscamap.chem.polimi.it/citterio/dottorato//
PhD
IN INDUSTRIAL CHEMISTRY AND CHEMICAL ENGINEERING (CII)
Causes of Accidents
In Chemical Sector approximately:
• 25 % lack of chemical knowledge
• 25 % lack of heat transfer capacity
• 25 % other design faults
• 25 % manual operating errors
44
22
12 11
5 5
1
Accidents (%)
Process Design
Nitration 33 Polymerisation phenolic resins 12 non-phenolic resins 16 other than resins 13
Sulfonation 12
Hydrogenation 7
Chlorination 6
Hydrolysis 6
Condensation 4
Oxidation 4
Alcoholysis 3
Amination 3
Cyclization 3
Bromination 3
Diazotization 2
Esterification 2
Isomerization 2
Methylation 2
Electrolysis 1
P F Nolan and J A Barton, J. Hazardous Materials, 1987, 14, 233
Analysis by Type of Process
BASF, Oppau/Ludwigshafen, September 21, 1921
Crater: 80 m diam., 16 m deep 450 dead
AZF, Toulouse,
September 21, 2001 Crater: 50 m diam., 10 m deep
29 dead
Survey of Incidents in the Chemical Industry
1. Basic lack of knowledge of chemistry/thermochemistry 34 2. Mischarging - omission of reactant 2
- undercharging of reactant 7
- overcharging 14
- charging too quickly 7
- charging at wrong temperature 3 - not following precise instructions 5
TOTAL 38
3. Accumulation of reactants 7
4. Impurities - raw materia1 changes 9
- water 10
- other 14
TOTAL 33
5. Delays in processing 9 6. Inadequate cooling - due to incorrect scale-up 9
- due to faulty instrumentation,
supply 12
TOTAL 21
7. Agitation failure/stoppage - inadequate agitation 4
- breakdown 6
- failure to start agitator 10 - intentional stoppage 4
TOTAL 24
Survey of Incidents in the Chemical Industry
8. Inadequate temperature control
- due to method of operation 14 - due to instrumentation 8
TOTAL 22
9. Problems with valves, leaks and blockages 31 10. Opening wrong manhole or reactor 2 11. Failure of auxiliary equipment 10 12. Case histories with insufficient detail 32
GRAND TOTAL 24
Survey of Incidents in the Chemical Industry
Process Changes Which May Be Hazardous
Quality of raw materials, solvents, intermediates
Material of construction of vessel (heat transfer, metal contamination)
Change of scale (agitation, mass transfer, cooling capacity)
Vessel shape, size, configuration (agitation, heat transfer)
Temperature (°C)
Heat flow (mW)
0 100 200 300 400 500
0 10 20 30 40 50 60
Crude DMU
pure DMU
Cl
Cl NH
N O CH3
OCH3 3-(3,4-dichlorophenyl)-1-methoxy- 1-methylurea
(weedkiller)
DSC Data for DMU
Temperature (°C)
Heat flow (mW)
0 100 200 300 400 500
0 10 20 30 40 50 60
Profile B (purity 94 % w/w) J. Thermal Anal., 37, 1991, 10
Profile A (purity 98 % w/w)
70 80
DSC Data for β-Lactam Precursor
Principles of Hazard Testing
Identify hazards
What if?
Can hazards be controlled?
Consequences of equipment failure
Evaluate for thermal safety
- Raw materials and solvents - Intermediates and products - By-products
- Distillation residues
Effect of Changing Circumstances
Order of addition of reagents
Omitting one component
Under- or overcharging reagents
Incorrect pH
Overheating the batch
Extended time
Process Definition
Define all parameters
- Temp, conc., rates of addition, times
Define limits within which NO corrective action will be taken
- Re normal operating variables - e.g. temp 50°C +/- 5°C
Define possible failure situations
Safety test to each limit
Only then can safety assessment give assurance that the process is safe
HEAT
REMOVAL
dQ/dt
Rate of heat production rQR = V ∆HR(-rA)
Rate of heat loss rQW = UA (T-TA)
A
C k
B
1
2
Ta Taz Ta1 TC
Likelihood of loss of control increases with scale
Temperature gradients increase too
∝
∝
TOTAL HEAT GENERATED
TOTAL VOLUME
(AND CONCENTRATION) (i.e. CUBE DIAMETER)
SURFACE AREA
AVAILABLE FOR COOLING (i.e. SQUARE OF DIAMETER)
Thermal Hazard
Changes in Raw Material Spec
Rate of reaction may be affected by
Amount of sodium in Lithium
- (D. Service, Chem. Brit., 1987, 27)
Particle size / grade of aluminium chloride
Grade of catalyst (esp. Raney Nickel)
Water content of solvents
- e.g. Grignard reactions - water delays initiation Too much reagent then added
When exotherm starts, goes out of control
Assessment Strategy Flow Diagram
Possible explosive compound
or mixture
Yes Specialist testing Explosive?
Recommend route/process
changes Yes
No Thermal Stability
Does thermal runaway occur?
Recommend max. safe operating
temp
Is max safe operating temp above reflux temp or max heat supply temp?
Recommend modifications
or control No
Yes No
Yes
Reaction Exotherm Can reflux or max safe operating temp be reached?
No
Recommend control
Can accumulation occur such that reflux or max safe operating temp be reached?
Recommend modifications No
Yes Yes
Recommend minimum operating Gas Evolution temp
Can max rate be
vented/scrubbed safely?
No
Recommend emergency venting
or modifications Yes
Gas Evolution Can max rate be
vented/scrubbed safely?
Assessment complete
Check subsequent changes to
assessed process No
Desk Screening
Presence of unstable groups
- Acetylene, nitro, azide, perchlorate, diazonium, peroxide and similar groups
Comparison with known explosives or potentially explosive materials
Calculation of oxygen balance
Thermodynamic data calculation (CHETAH)
- From molecular formula calculate heats of formation and reaction - Accurate within an order of magnitude
- Positive or small negative heats of formation indicate instability - Large negative heats of reaction indicate high output
oxygen Balance = 1600 [ 2x + (y/2) - z ] / M
M = molecular weight x = N° of carbon atoms y = N° hydrogen atoms z = N° oxygen atoms
(other heteroatoms are ignored)
Hazard potential oxygen balance
High da - 80 a + 120
Medium da + 240 a + 120
da - 160 a - 80
Low > + 240
< - 160
Oxygen Balance
OXYGEN BALANCE 128 (Medium/high Hazard Potential)
DSC-TGA Exotherm at 190-200°C
SIKAREX/ARC Exotherm at 110°C Destruction at 130°C
Heat of reaction - 800 kJ·mol-1 Pressure rise to 2000 psi
THEREFORE SAFETY LIMIT OF 60°C PLACED ON DRYING
Oxmetidine Intermediate
O
N N+ O O- O
O O
O
N S N
N
O O
O
Ranetidine Intermediate
CS2 +
CH3NO2
2 (CH3)2SO4 C C
H NO2 KS
KS
C C H NO2 H3CS
H3CS H
H3C S O
NO2 H3CS
KOH
H2O2
DSC Experiments
Sealed pan
- Sharp melting endotherm at 147.9°C - Exotherm with onset at 160°C
- Sample carbonized
- Pan seal ruptured and pan displace from sensor
Unsealed pan
- Sharp melting endotherm at 147.3°C - Exotherm with onset at 160°C
- Generates 954 J·g-1 - Sample carbonized
Conclusion
- In sealed pan material slowly self-heats before melting
Temperature (°C)
50 90 130 170 210 250
18
Heat flow (mW)
22 26
30 34
38 42
160 J·g-1 146.3°C
175.6°C
147.3°C
202.8°C 954 J·g-1
DSC for SF&F Intermediate
H H3C S
O
NO2
H3CS
ARC Experiments
Sample heated in 10° increments from 50 °C
Searching for self-heat rate of 0.02 °C·min-1 during 10 min
After 239 minutes, melting endotherm sensed at 139.5 °C
Step-heating continued to 150 °C
After 3 minutes, sample detonated and bomb destroyed
Last recorded temperature: 321°C
Self-heat rate – 54 °C·min-1
Examples: (a) AIBN; (b) styrene monomer; (c) gunpowder Showing: onset of exotherm, simple or complex reactions,
rate at any temperature (kinetics), total heat
output (thermodynamics), maximum self heat rate
40 140 240
0.01 0.1 1 10 100
(a)
(b)
(c)
Temperature (°C)
Heat rate (deg/min)
Self Heat Rate
Pressure transducer Heaters Top zone Bomb thermocouple
Jacket Bomb Radiant Heater Bottom zone Heater
Accelerating Rate Calorimeter
Time (min)
Temperature (°C)
search
search wait
wait
Adiabatic conditions
begin
Initial heating
ARC and the Heat-Wait-Search Operating Mode
Adiabatic Heating exotherm potential:
equipment assembly
TO POWER SUPPLY CONTROLLER LOW WATTAGE HEATER
BUNG
1 LITRE OR 500 ml DEWAR FLASK STIRRER GUIDE
STIRRER RECORDER
THERMOCOUPLE
Dewar Calorimeter
Advantages
- cheap and simple - Adiabatic
- Sensitive
Disadvantages
- Sample size
- Time consuming to use
Reaction Calorimeter (RC1)
Measure
- Overall heat of reaction - Heat of reaction with time
- Start of reaction (delay, accumulation) - Heat transfer during reaction
Maximum cooling capacity required
Disadvantages
- Size of sample
- Cannot allow runaway to proceed
Advantages
- Close to batch conditions - Shows when heat released - Cost
Tc(in) Mc
Cpc TR
Tj(in)
Tj(out) M•j
Cpj
MR CpR
Condenser
Reactor Jacket
Vacuum insulator
Tc(out)
Heating/Cooling System
Circulating pump
•
qR = qex + qaccu + qreflux + qdos + qloss Heat balance
TR = temp in the reaction mass TJ in= temp of the cooling inlet TJ out= temp of the cooling outlet Tc in= temp of the cooling inlet Tc out= temp of the cooling outlet
Principle of RC1 Calorimeter
Addition (g)
0 -50 100 200 300
530 570 610 650 690 730 770 810
Time (min.)
530 570 610 650 690 730 770 810 0
40 20 60 80
Time (min.) Heat release rate (W/Kg)
20 100 140 60 180
Integrated Heat kJ/Kg)
Standard Results - Curves from Reaction Calorimeter
Disadvantages of Adiabatic Calorimeters
Expense € 100,000 +
Not truly adiabatic
Agitation
Not entirely suitable for reaction mixtures
Conclusions
No single piece of equipment or technique will give all the answers
Dewar calorimeters give adequate results and are cheap
Combined with DSC, should give information for many applications
ARC and similar instruments give more detailed data and can detect sensitive exotherms
Reaction calorimeters measure heat evolved at each point and more accurately mimic plant conditions
Importance of understanding reaction
Thermal hazard, exotherms, runaway scenario
Rates of reaction, kinetics
Equipment choice
Dyestuff Intermediate - Ciba Geigy
6 H2, catalyst, 55-60°C NO2
NO2 RO
RO
NH2
NH2 RO
RO
Mechanism and Thermochemistry
RNO2 + H2 RNO + H2O ∆H = -134 kJ·mol-1 RNO2 + H2 RNHOH ∆H = -155 kJ·mol-1 RNHOH + H2 RNH2 + H2O ∆H = -259 kJ·mol-1
Overall - 548 kJ·mol-1
Different rates for each step
Different catalyst requirements for each stage Must compromise
RNO2 RNO RNHOH RNH2
RN=NR O
RN=NR
RNH-NHR RNH2
H2 H2 H2
H2
H2
RNO
Reaction Pathways
Reduction of Nitro Compounds
Intermediates
Nitroso Extremely reactive
Short-lived under normal conditions Hydroxylamines More stable
Reduce more slowly Accumulate in reaction
Particularly if catalyst poisoned by sulphides, nitrates, metals
Problems if : Temperature is too low Poor agitation
Catalyst is of wrong type
Catalyst contaminated by aqueous nitrate
Du Pont Explosion 1976
• For reduction to occur, species must be absorbed on the catalyst ArNO > ArNO2 > ArNHOH > ArNH2
• Thus ArNHOH is displaced by the other intermediates and reaction may stop at ArNHOH
Cl Cl
NO2 NH2
Cl Cl
Cl Cl
NHOH Raney Nickel
At Du Pont, when reaction slower
- operator increased temperature by 10°C
- thermal reaction set in - no control - Explosion after 3 - 4 min
At 260°C, the nitro compound itself
decomposes, ∆H = 1000 - 2000 kJ·mol-1
2 RNHOH 2 RNO + RNH2 + H2O ∆H = -55 kJ·mol-1 H2 Overall ∆H = -259 kJ·mol-1
100 200 300
1 2 3 4 5
minutes Temp. (°C)
Disproportionation
Dinitro Reduction - Differences
•
Intramolecular disproportionation•
Cyclization of dinitroso compounds to furoxans (- 257 kJ·mol-1)•
Dinitro compounds less soluble•
Dinitro compounds less stable (215?)•
Bis-hydroxylamino compound decomposes at 65 °C, which is the temperature of the planned processM
Preparation of reaction mixture charging of reactor autoclave with raw materials feed pump
Preparation of reaction mixture the top reactor desired level by
recirculation through
Catalyst addition to reaction through a suspension of raw materials or solvent
Recicl. pump Recicl. pump
1 2 3
Process Sequences
(hydrogenation reaction 1-7, and catalyst separation 10-12)
M
Preparation of reaction mixture charging of reactor autoclave with raw materials feed pump
Preparation of reaction mixture the top reactor desired level by recirculation through reaction heat exchange
Process Sequences
Key Reference:
Solvent Recovery Handbook I. Smallwood, Arnold, London
Solvent Recovery
•
Design work up so that all solvent cab be easily recovered•
Quality of solvent and impurities may vary from batch to batch•
Solvent may need second purification•
Hazards of solvent recoveryBoth caused by lack of attention to detail in solvent recovery process
Two Typical Incidents
•
Shell Stanlow explosion, UK•
Hickson/Angus explosion, Ireland•
Possibility of nitromethane anion adding to give polynitro compounds. Product isolated by extraction•
During solvent recovery, still bottoms contained polynitro impurities. When distilled to low volume, these exploded•
Reactions can occur during distillationAngus
C C H NO2 KS
KS
C C H NO2 H3CS
H3CS CS2 KOH
+
CH3NO2 2 (CH3)2SO4
Shell Fluoroaromatics Explosion
•
On 20th March 1990 the halogen exchange reactor at StanlowFluoroaromatics plant was ruptured by the pressure generated by a runaway reaction
•
Production of 2,4-DFNB from 2,4-DCNB during manufacture of 2,4- DFA•
Blast enhanced by formation of fireball when reactor contents ignited within the plant structure•
Six operators injured, one died three weeks later from post-operation complications following lower-limb surgery•
Plant partially demolished•
Blast and missile damage up to 500 m awayHALEX REACTION
CENTRIFUGATION
BATCH
DISTILLATION
HYDROGENATION &
CENTRIFUGATION
FINAL WORK UP OF 2,4-DFA
Pt/C catalyst Hydrogen Methanol
Recycle DMAc
Potassium fluoride 2,4-DCNB
TMAC catalyst DMAc solvent
Reacted KCl
Intercuts for distillation
Residue for incineration
Spent catalyst for recovery
Difluoroaniline Block Diagram
Cl Cl
NO2
NH2 Cl
Cl Cl
NHOH H2
H2
Plant - Halex Reaction
•
15,000 litre reactor•
Multipurpose plant•
160°C for 14 hours- Pressure held at 0.2 bar maximum
- Pressure build-up had never been observed during the 14 years operation of the Halex plant
•
Relief valve set at 5 barThe Plant - DFNB Distillation
•
Crude DFNB centrifuged to remove KF and KCl•
Cake washed with toluene, toluene added to DMAc•
Product (in DMAc) and toluene in holding tanks•
Batch distillation on small scale through Sulzer Mellapak column•
200-225 mbar used to fractionate into 8 fractionsCUT Pressure Temp (mbar) (deg °C) 1. A toluene/water azeotrope to remove water 200 76
2. “Dry” toluene for re-use - -
3. A toluene/DMAc intercut for redistillation - - 4. DMAc for re-use in the Halex reaction 100 110 5. A DMAc/DFNB intercut for redistillation - -
6. DFNB for hydrogenation 50 125
7. Monofluorinated chloronitrobenzene
(CFNB) for recycle - -
8. Heavy residues for third party incineration - -
Distillation Fractions
Events Leading to the Explosion
Campaign 4 months old - 39 batches run
Weeks prior to incident, water entered distillation pot - water carne from washing centrifuge to remove salts
Contents of distillation vessel were separated, some water removed
Batch 40 proceeded OK and DMAc recovered and appeared to be in spec
Batches 41 and 42 used DMAc recycled before the water incursion and were normal
Batch 43 used some DMAc recycled after water incursion and was normal
Batch 44 used DMAc, all of which was recycled after the water incursion
Batch 44 - Abnormal Reaction
Heated to 165°C but temperature continued to rise
Pressure was rising, but pressure not shown on VDU used by operators
Another operator alerted colleagues to the abnormally high pressure but, before they could take action, the relief valve lifted and the
reactor exploded
Reactor torn in 3 pieces
One piece travelled 200 meters
Plant buckled beyond recognition
Reactor contents formed a fireball which caused secondary fires
6 Operators injured - one died later in hospital
Original Process Development (to 1982)
Recycling of DMAc
- No impurities after 10 recycles - No acetic acid present !!!!
- Use tests with recycled DMAc normal
Stability of reaction mixture - SIKAREX studies
- Storage or reaction mixture at 150°C gave no change, with or without iron present
- At 180°C for 20 hours there was some decomposition but no
exotherm (DFNB and DMAc decompose under these conditions - Distillation residues decomposed slowly at 200°C with a small
exotherm which would not runaway
Development Work 1989
No DFA was made between 1982 and 1989
Reaction was optimised at 165°C compared to the original 145°C to increase throughput
Because thermal stability testing had previously been carried out at 180°C no further work was considered necessary
Chemical Investigation
Presence of acetic acid in Halex reaction caused temperature to rise from 160 to 240°C
At 240°C a second exotherm starts (decomposition of DFNB, DCNB)
Temperature profile of runaway batch mimicked in laboratory by addition of acetic acid
Acetic acid formed from DMAc by hydrolysis, but only in the presence of DFNB, which scavenges dimethylamine
DMAc-acetic acid-water azeotropes at the same
temperature as DMAc, thus water cannot be separated out once acetic acid has formed
DMAc charge vessel contained acetic acid
Lessons
Need to rigorously check for build-up of impurities in recycle streams
Analytical methods must be adequate for contro1 of the specification !
Changes in reaction conditions should be re-evaluated in the light of newer techniques
Old processes should be re-evaluated in the light of changing standards
0 50 100 150 200 1
2 3
0 20 40 60 80 100 120 140 160 180 200
time (min)
temp (°C)
2,4-DFNB Batch 44
Stanlow 20 March 1990 Coil temperature
reaction temperature
pressure
pressure (bar)
Process Conditions - Runaway Batch
CH2=C=O chetene
(b.p. -56 °C)
Role of Acetic Acid
+ CH2=C=O
X
X NO2
X
OCOCH3 NO2
∆
X
OH NO2
X
O
X NO2 NO2 HOAc
KF
Further reaction
etc.