Man-made Environmental Disasters
Can knowledge of Probabilistic methods
• Bhopal Disaster had its 20th Anniversary
• A lot of Man-made disasters are caused by ‘errors’
• Uncertainty analyses/ Error Analysis/ Probabilistic
Methods/Risk Management can all help to reduce the
• We’ll talk about one Man-made Environmental Disasters:
Bhopal Disaster (India)
• Gas Leak on December 3, 1984 – MIC (Methyl
• 8000 people died, 50,000 injured
• Gas leaked from a Union Carbide Plant at night
Main effect: Pulmonary Edema, skin
and eye damage (total blindness)
Secondary effect: Bronchitis, Bronchial
People fled from the poison
A Bhopal Victim
gas in the middle of the
How it Happened
• In the middle of the night. People fast asleep (many on
• 40 tonnes of MIC leaked from Union Carbide Plant
Union Carbide still maintains
‘Sabotage’ as the cause:
‘Someone’ put water in the boiler to
trigger a set of reactions
Here’s One Story
• Bhopal station was filling with panicking passengers
fleeing the fumes
• Station Manager Mr. Dastagir was sensing something
wrong (note: no one yet knew what was going on)
• The next train was scheduled to depart 20 minutes later
• He ordered the train to leave immediately (early)
• Incoming trains were all diverted
• It was Catch 22 situation. Uncontaminated passengers
incoming to Bhopal could not be put at risk.
• Yet, those in Bhopal wanted to flee the fumes boarding
as many trains.
• The station had instead become a scene of misery and
death all around.
• Mr. Dastagir’s action saved many lives. Unfortunately
he’s a forgotten Hero (died a year spending most of his
remaining life in hospitals
Could The Disaster Been Avoided?
• Probably (Bhopal lacked emergency planning – had
never considered of such a scenario )
• Union Carbide had a sloppy management practice (poor
• Such situations could be avoided through Rigorous Risk
• All kinds of possible scenarios are considered with their
• Joint probabilities, independence, dependence, mutually
• Example: Nuclear Waste Regulatory Practice – Yucca
Mountains – Groundwater Contamination.
But How Exactly do We Minimise the
Impact of Disasters?
• First, before even actual implementation of a
project/plant, we consider ‘what if’ scenarios.
• We ask then the most plausible effect of the ‘what if’
• Next we ask ‘can we afford to live with it?’
• Next we try to compute the chance, odds (odds ratio),
probability (using very much the CEE6430 theory we’ve
learnt) of those ‘what ifs’.
• It’s impossible to consider all ‘what ifs’ scenarios
• But good news – Computers are getting powerful to do
the job – to consider error in all sources.
• New numerical techniques emerging, Fuzzy concepts,
Optimization, AI, Error Analysis, Error Propagation
More Specific Example: Hydrologic
• Imagine this: National Weather Service
• Issuing Flood Forecasts
• Considers only upto 100 years of return period in its
• What if – storm were a 500 year return period?
• How do we calculate the flood level (probabilistic)?
Remember, we need to consider uncertainty in both
rainfall and model – random processes
• Next question – can the city afford the damage expected
from a 500 year storm?
• If not – what kind of proactive measures can it take
(emergency measures) to minimise loss of life.