Qualitative and Semi-Quantitative Risk Methods

Introduction

Fire risk analysis supports decisions about prevention, detection, protection, emergency response and business continuity. In many workplaces, reliable numerical data are limited: the frequency of ignition may be unknown, consequences may depend on occupancy and firefighting response, and safeguards may not have been tested sufficiently to justify precise probabilities. Qualitative and semi-quantitative methods are therefore essential professional tools.

Qualitative methods describe risk using words, categories and structured reasoning. Semi-quantitative methods assign scores or bands to likelihood and consequence so that hazards can be compared and priorities established. Neither method produces an objective “true” risk value. Their value depends on the quality of hazard identification, the competence of the team, the evidence used, and the clarity of assumptions.

A central principle is that a score is a decision aid, not a measurement of reality. A risk rating of 12 is not necessarily twice as dangerous as a rating of 6. Scores should never create false precision or replace professional judgement.

Learning objectives

On completing this chapter, the learner should be able to:

• distinguish qualitative, semi-quantitative and quantitative risk analysis;
• construct and interpret a risk matrix;
• use checklists, what-if analysis, job safety analysis, HAZID and bow-tie analysis;
• combine structured professional judgement with evidence;
• identify weaknesses in scoring systems and avoid false precision;
• apply these methods to fire-risk decisions and communicate findings professionally.

1. Risk concepts and method selection

Risk is commonly understood as the combination of the likelihood of an unwanted event and the severity of its consequences. A broad conceptual representation is:

This expression is a framework rather than a universal physical law. Likelihood may mean frequency, probability over a stated period, or a qualitative category. Consequence may include deaths, injuries, property damage, environmental harm, interruption and reputational loss. The two dimensions must be defined before a rating is assigned.

Qualitative analysis uses descriptions such as low, medium and high. Semi-quantitative analysis uses ordered scales, for example 1–5 for likelihood and 1–5 for consequence. Quantitative analysis uses numerical frequencies, probabilities, loss estimates or models supported by data and explicit assumptions.

Method selection should reflect the decision. A checklist may be adequate for routine inspections; a bow-tie may be preferable for a major fire scenario; a detailed quantitative assessment may be necessary for complex, high-consequence facilities. Legal requirements, corporate standards and competent-person requirements must also be considered.

2. Risk matrices

A risk matrix combines a likelihood scale with a consequence scale. For example, likelihood may be classified as rare, unlikely, possible, likely and almost certain. Consequence may be classified from minor injury or limited damage to multiple fatalities or catastrophic loss.

A numerical matrix often uses:

where R is the risk score, L is the likelihood score and C is the consequence score. The multiplication is a chosen scoring convention, not a claim that the underlying risk is measured on a ratio scale. It is meaningful only when the scale definitions and decision rules are established.

Example matrix

An organisation might define 1–4 as low, 5–9 as moderate, 10–16 as high and 17–25 as very high. These boundaries are policy choices and must be justified. A low score does not mean that no controls are required.

Worked example

A storage room contains combustible packaging near a battery-charging station. The team judges ignition as “possible” (3) and potential consequence as “major” (4), because smoke could spread to an occupied building before evacuation. The initial score is:

After relocating packaging, providing separation, maintaining charging equipment and improving detection, the team judges likelihood as “unlikely” (2). If consequence remains 4, the residual score is:

The analysis does not prove that risk has reduced by one-third. It records a reasoned change in category based on specified controls. The consequence score should not be reduced merely because ignition is less likely; consequence and likelihood represent different questions.

Good practice in matrix use

A defensible assessment records the scenario, exposed people, existing safeguards, scale definitions, evidence, assumptions, assessor competence, additional actions, responsible persons and review date. Separate initial risk from residual risk. Confirm that safeguards are actually available, effective, maintained and independent where independence matters.

3. Checklists and structured inspections

A checklist converts known expectations into observable questions. Fire checklists may cover ignition sources, electrical installations, housekeeping, storage, hot work, smoking control, fire doors, extinguishers, alarms, emergency lighting, escape routes, training and records.

Checklists are valuable because they reduce omission and support consistency between inspections. They are particularly useful for routine workplaces and regulatory compliance. They should be treated as prompts, not as complete risk assessments. A tick in the “compliant” column does not demonstrate that an unusual interaction or credible escalation has been considered.

Effective checklists:

1. identify the area, activity and inspection date;
2. use observable, unambiguous questions;
3. distinguish not applicable from not inspected;
4. record evidence and corrective action;
5. assign priority, owner and completion date;
6. are periodically updated after incidents, changes and new hazards.

Poor checklists encourage mechanical compliance. For example, “Are extinguishers present?” is weaker than “Are the correct extinguishers accessible, identified, unobstructed, inspected and suitable for the credible fire classes?”

4. What-if analysis

What-if analysis uses questions beginning with “What if...?” to explore deviations, failures and consequences. A multidisciplinary team may ask:

• What if a flammable liquid is released during transfer?
• What if an ignition source is introduced?
• What if the alarm is not heard in a noisy area?
• What if the fire door is wedged open?
• What if the emergency access route is blocked?
• What if the automatic suppression system is isolated?

The team records the initiating condition, possible consequence, existing safeguards, recommendations and responsible person. The method is flexible and efficient, but its completeness depends on the experience and imagination of participants. Facilitators should use drawings, operating procedures, incident records and site observations to reduce reliance on memory.

5. Job safety analysis

Job safety analysis (JSA), also called task risk assessment, examines a task step by step. It is especially useful for hot work, fuel transfer, cylinder handling, confined-space work, maintenance and emergency activities.

A JSA normally contains:

The analysis should distinguish normal steps from abnormal conditions, include simultaneous activities, and specify stop-work criteria. Controls should follow the hierarchy of controls: eliminate the hazard where possible, substitute, use engineering controls, establish administrative controls, and use personal protective equipment as the last line of defence. PPE does not prevent a building fire and should not be presented as the primary control for an ignition hazard.

6. Bow-tie diagrams

Bow-tie analysis presents one major unwanted event, or top event, between causes and consequences. Preventive barriers act on the left side; mitigative barriers act on the right side.

Figure: A bow-tie model separates preventive barriers from consequence-reducing barriers.

For a solvent-release fire, threats may include hose failure, incorrect valve operation or vehicle impact. Preventive barriers may include inspection, bunding, isolation and operating procedures. The top event is loss of containment with ignition potential. Mitigative barriers may include gas detection, automatic shutdown, fire detection, suppression, separation, emergency response and evacuation.

A bow-tie is not automatically a fault tree or an event tree. It is a structured communication model. It becomes stronger when each barrier has an owner, performance standard, inspection method, proof-test requirement and impairment procedure. “Training” alone is often a weak barrier unless competence, supervision and verification are specified.

7. HAZID

Hazard identification (HAZID) is a structured, early-stage examination of hazards associated with a facility, process, project or change. It is commonly conducted by a multidisciplinary team including operations, engineering, maintenance, fire safety, emergency response and, where appropriate, contractors or worker representatives.

A HAZID considers materials, energy sources, occupancy, interfaces, abnormal operations, external events, utilities, access, emergency arrangements and credible escalation. Guidewords such as fire, explosion, release, loss of utility, human error, vehicle impact and natural hazard stimulate discussion.

The HAZID register normally records the hazard, cause, consequence, safeguards, risk ranking, recommendation, owner and due date. HAZID is valuable during concept and design stages because hazards can be eliminated or engineered out more easily before construction. It is not a substitute for detailed design verification, fire engineering analysis, permit systems or operational inspections.

8. Structured professional judgement

Professional judgement is unavoidable when evidence is incomplete. Structured professional judgement makes that judgement transparent rather than pretending that it is purely objective. The process should:

• define the scenario and decision question;
• assemble relevant evidence, including incident history and standards;
• involve suitably competent and diverse participants;
• state assumptions and uncertainty;
• consider alternative interpretations;
• document reasons for the conclusion;
• identify actions that would change the decision;
• obtain independent review for high-consequence decisions.

Teams should guard against groupthink, anchoring on an initial score, authority bias and optimism bias. A facilitator can ask each participant to make an initial independent judgement before discussion. Disagreement should be recorded when it reveals genuine uncertainty rather than suppressed to produce consensus.

9. Limitations of scoring and false precision

Semi-quantitative scores have several important limitations.

First, ordinal categories are ranks, not necessarily equal numerical intervals. The difference between “unlikely” and “possible” cannot automatically be treated as equal to the difference between “likely” and “almost certain.” Multiplication can therefore create arithmetic-looking results without a strong measurement basis.

Second, matrices can hide catastrophic low-frequency events. A rare event with extreme consequences may receive the same score as a frequent event with moderate consequences, although the controls and decision implications are different.

Third, different combinations can produce the same score. For example, 2 × 10 and 5 × 4 both equal 20, but prevention and emergency planning may be entirely different.

Fourth, the categories may be sensitive to assessor interpretation. “Possible” may mean once in a year to one assessor and once in ten years to another. Scales should therefore include time frames, examples and evidence.

Fifth, risk scores may be manipulated to meet an acceptance threshold. Reducing the score on paper is not risk reduction. Controls must be linked to credible mechanisms and verified in practice.

Better practice is to report the scenario and narrative alongside the score, show uncertainty, identify dominant hazards, distinguish worst credible consequence from most likely consequence, and use sensitivity analysis. Where consequences are potentially severe, a low likelihood rating should trigger scrutiny rather than reassurance.

10. Integrated application: illustrative case

Background

This illustrative case concerns a two-storey training centre with classrooms, a workshop, a small kitchen and a battery-charging area. A renovation introduces temporary hot work near stored packaging while the fire alarm system is partly isolated.

Problem

A checklist identifies combustible storage and an impaired alarm, but it does not explain how welding sparks could ignite concealed combustibles or how occupants would be warned during isolation.

Analysis

A JSA divides the hot-work task into preparation, equipment setup, welding, breaks and close-out. A what-if workshop asks what happens if sparks enter a ceiling void, the fire watch leaves, or the alarm is not restored. HAZID participants identify interactions between contractors, occupied classrooms and the temporary impairment.

The bow-tie top event is ignition of combustible material during hot work. Preventive barriers are removal of combustibles, a permit, isolation of ignition sources, screens and a competent fire watch. Mitigative barriers are portable detection, communication arrangements, extinguishers, an evacuation plan and a defined post-work watch.

Using a five-by-five matrix, the initial judgement is likelihood 3 (possible) and consequence 4 (major), giving a score of 12 under the organisation’s convention. The team does not lower consequence merely because the work is temporary. Additional controls reduce the likelihood category to 1 (rare) only after the contractor removes combustibles, verifies the void, establishes continuous fire watch, provides temporary detection and limits work to a controlled period. If these controls cannot be verified, the work is postponed rather than receiving a favourable score.

Outcome

The project manager assigns barrier owners and requires a documented alarm-impairment permit, hourly checks and a final inspection. The outcome is not “zero risk”; it is a better-controlled, more observable and more accountable work arrangement.

Learning points

The case demonstrates that methods complement one another. The checklist found conditions, the JSA examined task steps, what-if analysis explored deviations, HAZID identified interfaces, the bow-tie clarified barriers, and the matrix supported prioritisation. The narrative evidence was more important than the numerical score.

11. Governance, competence and technology

Risk assessments should be controlled documents with revision history, approval, review triggers and clear ownership. They should be reviewed after incidents, near misses, process changes, occupancy changes, impairment of protection systems, new materials, changes in legislation and significant organisational changes.

Assessors need knowledge of fire dynamics, human behaviour, building use, process hazards, control reliability and emergency arrangements. Consultation with workers is essential because written procedures may not reflect actual work.

Digital risk registers can improve action tracking, version control, trend analysis and auditability. Artificial intelligence may assist with searching records, identifying recurring themes or generating checklist prompts, but it should not independently determine fire risk. AI outputs may contain omissions, biased training data or fabricated reasoning. A competent professional must verify sources, assumptions, site conditions and recommendations, and sensitive data must be governed appropriately.

Best-practice workflow

1. Define the scope, decision and assessment boundaries.
2. Describe credible fire scenarios, not merely equipment or rooms.
3. Identify people, property, environment and business consequences.
4. Use more than one method where their perspectives complement one another.
5. Record existing barriers and test whether they are available and effective.
6. Apply a matrix consistently, without treating its score as a measured probability.
7. Assign actions according to the hierarchy of controls.
8. Verify completion and inspect performance in the workplace.
9. Communicate residual uncertainty and escalation conditions.
10. Review after change, learning or impairment.

Chapter Summary

Qualitative and semi-quantitative methods provide practical ways to analyse fire risk when precise data are unavailable or unnecessary. Checklists support systematic observation; what-if analysis explores deviations; JSA examines task steps; HAZID identifies hazards early; bow-tie analysis links causes, preventive barriers, top events, mitigative barriers and consequences; and structured professional judgement makes assumptions and uncertainty visible. Risk matrices help prioritise action but do not measure risk with scientific precision. Their categories, arithmetic and acceptance boundaries must be defined and used consistently. The most defensible assessment combines scores with scenario descriptions, evidence, barrier verification, competent participation and continuing review.

Key Terms

Qualitative analysis: Risk analysis using descriptive categories and narrative judgement.

Semi-quantitative analysis: Analysis using ordered scores or bands to support comparison.

Quantitative analysis: Analysis using numerical frequencies, probabilities or loss estimates.

Risk matrix: A table that combines likelihood and consequence categories.

Likelihood: An assessment of how plausibly or frequently an event may occur.

Consequence: The outcome or severity resulting from an event.

Initial risk: Risk before additional proposed controls are implemented.

Residual risk: Risk remaining after controls are implemented and verified.

Checklist: A structured list of questions or conditions used to reduce omission.

What-if analysis: A facilitated method using hypothetical questions to explore deviations.

Job safety analysis: Step-by-step analysis of hazards and controls for a task.

HAZID: Structured hazard identification for a facility, process, project or change.

Bow-tie analysis: A model linking threats, a top event, consequences and barriers.

Top event: The central loss-of-control event in a bow-tie model.

Preventive barrier: A control intended to stop a threat causing the top event.

Mitigative barrier: A control intended to reduce consequences after the top event.

False precision: Unwarranted confidence created by exact-looking numbers or scores.

Structured professional judgement: Transparent expert judgement supported by evidence and explicit assumptions.

Impairment: A condition in which a safety system is unavailable or degraded.

Barrier verification: Confirmation that a control exists, functions and is maintained.

Review Questions

Short Answer Questions

1. Distinguish qualitative, semi-quantitative and quantitative risk analysis.
2. What are the two principal dimensions of a conventional risk matrix?
3. State two limitations of multiplying likelihood and consequence scores.
4. What is the purpose of a top event in bow-tie analysis?
5. Why should a checklist not be treated as a complete risk assessment?

Long Answer Questions

1. Explain how a fire risk matrix should be designed, documented and used responsibly.
2. Describe the process of conducting a job safety analysis for hot work.
3. Compare what-if analysis, HAZID and bow-tie analysis in terms of purpose and output.
4. Explain how structured professional judgement can reduce groupthink and false confidence.
5. Discuss the role of barrier ownership, verification and impairment management in fire risk control.

Analytical / Application Questions

1. A warehouse has combustible packaging beside a charging station. Develop a scenario-based assessment using a checklist, what-if questions, a matrix and a bow-tie model.
2. A contractor proposes welding in an occupied building while the alarm is isolated. Decide what information and controls are required before work begins, and identify stop-work conditions.
3. Two hazards receive the same matrix score: one is frequent with moderate consequences and the other is rare with catastrophic consequences. Explain why identical scores do not justify identical controls.

Multiple Choice Questions with Answers

Question 1

Which statement best describes a semi-quantitative risk method?

A. It uses ordered scores or bands to support comparison and prioritisation.

B. It always calculates an exact annual probability from complete data.

C. It eliminates the need for professional judgement.

D. It applies only to financial loss estimation.

Correct Answer: A. It uses ordered scores or bands to support comparison and prioritisation.

Question 2

In a conventional risk matrix, what do likelihood and consequence represent?

A. The cost of controls and the time required to install them.

B. The plausibility or frequency of an event and the severity of its outcome.

C. The number of workers and the size of the building.

D. The age of equipment and its purchase price.

Correct Answer: B. The plausibility or frequency of an event and the severity of its outcome.

Question 3

A risk matrix uses R = L × C. If L = 3 and C = 4, what is the score?

A. 7

B. 9

C. 12

D. 16

Correct Answer: C. 12

Question 4

Which activity is the primary purpose of a job safety analysis?

A. To estimate national fire statistics.

B. To examine a task step by step and identify controls.

C. To calculate the structural load of a building.

D. To replace emergency drills.

Correct Answer: B. To examine a task step by step and identify controls.

Question 5

In a bow-tie diagram, what is a preventive barrier intended to do?

A. Stop a threat from causing the top event.

B. Increase the severity of the consequence.

C. Record the final inspection date only.

D. Measure employee satisfaction.

Correct Answer: A. Stop a threat from causing the top event.

Question 6

Which is a major limitation of numerical risk scores?

A. They make all hazards impossible to compare.

B. They can create false precision when category intervals are not truly measured quantities.

C. They cannot be recorded in a risk register.

D. They always overestimate consequence.

Correct Answer: B. They can create false precision when category intervals are not truly measured quantities.

Question 7

What-if analysis primarily uses which technique?

A. Hypothetical questions about deviations and failures.

B. Microscopic examination of materials.

C. A legal judgement about liability.

D. A fixed list that cannot be changed.

Correct Answer: A. Hypothetical questions about deviations and failures.

Question 8

Why is HAZID particularly valuable early in a project?

A. Hazards can often be eliminated or engineered out before construction.

B. It proves that operations will never change.

C. It replaces all detailed design verification.

D. It guarantees zero residual risk.

Correct Answer: A. Hazards can often be eliminated or engineered out before construction.

Question 9

Which statement reflects structured professional judgement?

A. The most senior participant’s opinion automatically determines the result.

B. Judgement is documented with evidence, assumptions, uncertainty and reasons.

C. Only numerical data may be considered.

D. Disagreement must be removed from the record.

Correct Answer: B. Judgement is documented with evidence, assumptions, uncertainty and reasons.

Question 10

A consequence score should generally not be reduced merely because ignition likelihood has been reduced because:

A. Consequence and likelihood address different dimensions of the scenario.

B. Consequence scores are always fixed by law.

C. Likelihood cannot be changed by controls.

D. Matrices contain no consequence categories.

Correct Answer: A. Consequence and likelihood address different dimensions of the scenario.