16.2 Hazard Identification and Risk Analysis

Process safety in chemical engineering involves identifying hazards and evaluating risks to prevent catastrophic incidents such as fires, explosions, and toxic releases. A hazard is an inherent physical or chemical characteristic (such as toxicity, flammability, or high pressure) that has the potential to cause harm. Risk, by contrast, is a measure of both the probability (or frequency) of an event occurring and the severity of its consequences:

$\text{Risk} = \text{Frequency} \times \text{Severity}$

To manage risks, chemical engineers utilize a variety of qualitative, semi-quantitative, and quantitative hazard evaluation techniques.

Hazard and Operability (HAZOP) Studies

A HAZOP study is a systematic, qualitative, team-based technique used to identify potential hazards and operability problems in a process facility. The HAZOP team examines a Process Flow Diagram (PFD) or Piping and Instrumentation Diagram (P&ID) node by node. A node is a specific section of the process with defined design intentions (e.g., a feed line, a reactor vessel).

For each node, the team combines standard guide words with process parameters to identify potential deviations from the design intent:

$\text{Guide Word} + \text{Parameter} = \text{Deviation}$

Common deviations include:

• NO FLOW: Caused by pump failure or a closed block valve, potentially leading to reactor starvation or loss of cooling.
• MORE PRESSURE: Caused by control valve failure or external heating, potentially causing vessel rupture.
• LESS TEMPERATURE: Caused by steam trap failure, potentially leading to incomplete reaction or solidification.
• REVERSE FLOW: Caused by check valve failure, potentially resulting in raw materials flowing back into storage containers.

The team analyzes the causes of each deviation, evaluates the consequences, identifies existing safeguards (such as interlocks or relief valves), and recommends additional actions if the risk is unacceptable.

Layer of Protection Analysis (LOPA)

LOPA is a semi-quantitative risk assessment methodology used to evaluate the sufficiency of safeguards protecting against a specific accident scenario. It builds upon qualitative HAZOP findings. LOPA assumes that multiple independent barriers, known as Independent Protection Layers (IPLs), must fail sequentially for an accident to occur.

To qualify as an IPL, a safeguard must be:

• Specific: Designed to prevent the specific consequence.
• Independent: Its operation is completely independent of the initiating event and any other IPLs.
• Dependable: It can be relied upon to perform as designed.
• Auditable: It can be inspected and tested regularly.

The frequency of the mitigated consequence ($f_{\text{mitigated}}$) is calculated using the initiating event frequency ($f_{\text{IE}}$, in events per year) and the Probability of Failure on Demand (PFD) of each IPL:

$f_{\text{mitigated}} = f_{\text{IE}} \times \prod_{j=1}^m PFD_j$

Typical PFD values range from $0.1$ for operator response to an alarm, to $0.01$ to $0.001$ for safety instrumented systems (SIS) with different Safety Integrity Levels (SIL 1 to SIL 3).

Worked Example 1:

An exothermic reactor run has an initiating event of cooling pump failure, which occurs once every 5 years ($f_{\text{IE}} = 0.2 / \text{year}$). The safeguards in place are an operator intervention based on a high-temperature alarm ($PFD = 0.1$) and a safety valve ($PFD = 0.01$). Calculate the mitigated event frequency.

1. Identify variables: $f_{\text{IE}} = 0.2 / \text{year}$ $PFD_1 = 0.1$ $PFD_2 = 0.01$

2. Apply LOPA product rule: $f_{\text{mitigated}} = 0.2 \times 0.1 \times 0.01 = 0.0002 \text{ events/year}$

If the facility's acceptable risk target for this consequence is $10^{-5}$ events/year, this configuration is insufficient, and an additional IPL (such as an automatic interlock system) must be added.

Fault Tree and Event Tree Analysis

• Fault Tree Analysis (FTA): A deductive, top-down quantitative method. It starts with an undesired 'Top Event' (e.g., vessel explosion) and works backward to identify the combinations of basic component failures and human errors that could cause it. FTA uses logic gates:
  • AND Gate: The top output event occurs only if all input events occur. For independent events, the probability is: $P(\text{Output}) = \prod_{i=1}^n P_i$
  • OR Gate: The output event occurs if any of the input events occur. For independent events: $P(\text{Output}) = 1 - \prod_{i=1}^n (1 - P_i) \approx \sum_{i=1}^n P_i \quad (\text{if } P_i \ll 1)$
• Event Tree Analysis (ETA): An inductive, bottom-up method. It begins with an 'Initiating Event' (e.g., pipe rupture) and traces the sequential success or failure of subsequent safety barriers to determine the various possible end-states and their frequencies.

Atmospheric Dispersion Modeling

When a chemical release occurs, engineers use dispersion modeling to estimate the downwind concentration of the toxic or flammable gas. The standard model is the Gaussian Plume Model, which describes the steady-state dispersion of a continuous gas release from a point source:

$C(x,y,z) = \frac{Q}{2\pi u \sigma_y \sigma_z} \exp\left( -\frac{y^2}{2\sigma_y^2} \right) \left[ \exp\left( -\frac{(z-H)^2}{2\sigma_z^2} \right) + \exp\left( -\frac{(z+H)^2}{2\sigma_z^2} \right) \right]$

where:

• $C(x,y,z)$ is the concentration ($\text{g/m}^3$) at downwind distance $x$, crosswind distance $y$, and vertical height $z$.
• $Q$ is the source release rate ($\text{g/s}$).
• $u$ is the wind speed at stack height ($\text{m/s}$).
• $H$ is the effective stack release height ($\text{m}$), which is physical stack height ($h$) plus plume rise ($\Delta h$).
• $\sigma_y, \sigma_z$ are the dispersion coefficients ($\text{m}$), representing the standard deviation of the concentration profile. They grow with downwind distance $x$ and depend on the Pasquill-Gifford atmospheric stability classes (Class A: extremely unstable, Class D: neutral, Class F: moderately stable).

At ground level directly downwind ($y=0$ and $z=0$), the concentration equation simplifies to:

$C(x, 0, 0) = \frac{Q}{\pi u \sigma_y \sigma_z} \exp\left( -\frac{H^2}{2\sigma_z^2} \right)$

Higher wind speeds ($u$) and unstable atmospheres (which increase $\sigma_y, \sigma_z$) lead to rapid dilution and lower downwind concentrations. Conversely, stable atmospheric conditions (Class F, typical of clear nights with low wind) lead to minimal mixing and high toxic gas concentrations at long downwind distances.