Hazard assessments and the hierarchy of controls

Every safety system on a construction site rests on one skill: looking at a job before it starts and seeing what could hurt someone. The site safety plan from lesson 3 identifies hazards at the level of the whole project and each phase. This lesson works at the level of the task: installing the level 5 wall panels, cutting core holes in the podium slab, setting the roof-top units. It covers the two kinds of task hazard assessment used on Larkspur Commons, a practical way to rate risk, and the hierarchy of controls that tells a supervisor which fixes to reach for first. HSE-103, Hazard Recognition and Risk Assessment, covers the general method in more depth; here the focus is on how a superintendent and foremen use it on a construction site.

Hazard, risk and control

A hazard is anything with the potential to cause harm: an unprotected floor edge, a suspended load, a trench wall, a silica-generating saw. Risk combines how likely harm is and how bad it would be. A control is anything that removes the hazard or reduces the risk. The distinctions matter because they change the conversation. A crew that says 'the edge is a hazard, so we will be careful' has identified a hazard but done nothing about the risk. A crew that says 'the edge is a hazard; with guardrails installed before the floor is loaded, the chance of a fall is very low' has applied a control and judged the remaining risk.

Two levels of task hazard assessment

Larkspur Commons uses two levels of task assessment, matching the way most construction firms work.

The JHA is planning; the FLHA is checking. A JHA written in the office cannot know that today the wind is gusting, a new laborer has joined the crew, the scaffold beside the work area was altered overnight or the electrician has opened a section of floor. The FLHA exists to catch exactly those things. HSE-230, Field-Level Hazard Assessment, in the Safety stream covers the FLHA in detail.

Building a job hazard analysis

A JHA has three columns: the steps of the task in order, the hazards in each step, and the controls for each hazard. The skill is in the first column. If the steps are too broad ('install wall panels'), the hazards will be too general to be useful. If they are too fine, the JHA becomes unreadable. Most construction tasks break into eight to fifteen steps.

A good JHA uses the crew. The framers who will set the panels know where the real problems are: the panel that always catches on the scaffold brace, the corner where the anchor is out of reach, the gust that comes around the building at about two in the afternoon. Thornbury's rule is that at least two workers who will do the task review the JHA before it is used, and that their names are on it.

Rating risk

Thornbury uses a simple five-by-five risk matrix: likelihood from 1 (rare) to 5 (almost certain) and severity from 1 (first aid) to 5 (fatality or permanent disability). The product is the risk score. Scores of 15 or more are high and the task cannot proceed until controls bring the score down; 8 to 14 are medium and need controls and supervision; 7 or less are low.

Two cautions apply to any risk matrix. First, the numbers are judgments, not measurements; two foremen may score the same hazard differently, which is why scoring is done with the crew and reviewed. Second, a matrix can hide serious hazards behind low likelihood. A suspended load falling on someone is unlikely on any given lift, but the severity is fatal, so the score can come out as medium. Thornbury therefore flags any hazard with a severity of 5 as having serious injury and fatality (SIF) potential, whatever its score, and requires at least one control from the top three levels of the hierarchy below. Lesson 11 returns to SIF prevention.

# JHA risk scoring before and after controls, with SIF flag (illustrative scores)
hazards = [
    # (hazard, likelihood, severity, controls, likelihood after, severity after, top-three control?)
    ("fall at open edge while setting panel", 3, 5, "engineered anchors, tie-off, guardrails elsewhere", 1, 5, True),
    ("struck by swinging panel",             3, 4, "tag lines, exclusion zone, signal person", 1, 4, False),
    ("rigging failure",                      2, 5, "inspected slings, spreader bar, qualified rigger", 1, 5, True),
    ("wind gust during lift",                3, 4, "anemometer limit, lift plan", 1, 4, False),
    ("hand caught between panel and plate",  3, 3, "push sticks, gloves, briefing", 2, 2, False),
]
def band(s): return "HIGH" if s >= 15 else "medium" if s >= 8 else "low"
for h, l, s, c, l2, s2, top in hazards:
    before, after = l * s, l2 * s2
    sif = " SIF" if s == 5 else ""
    flag = "" if (s < 5 or top) else "  <- SIF needs a top-three control"
    print(f"{h:38s} {before:2d} {band(before):6s} -> {after:2d} {band(after):6s}{sif}{flag}")

fall at open edge while setting panel  15 HIGH   ->  5 low    SIF
struck by swinging panel               12 medium ->  4 low   
rigging failure                        10 medium ->  5 low    SIF
wind gust during lift                  12 medium ->  4 low   
hand caught between panel and plate     9 medium ->  4 low

All five hazards come down to low or medium after controls. The two SIF hazards, the fall at the open edge and rigging failure, both have controls from the top of the hierarchy: engineered anchors and guardrails for the fall, and engineered rigging for the lift. Note that severity rarely changes: controls usually reduce the likelihood of harm, not how bad it would be if it happened. That is why a fatal-severity hazard with a low score still deserves attention.

The hierarchy of controls

The hierarchy of controls ranks the ways of dealing with a hazard from most to least effective. The higher controls work whether or not anyone remembers to use them; the lower ones depend on people doing the right thing every time.

Many of the most effective controls are decided long before a foreman writes a JHA. Prefabricating the exterior wall panels, choosing a building design with a parapet that doubles as a guardrail, specifying permanent roof anchors, and sequencing the stairs to be installed early so workers do not climb ladders between floors are all decisions made in design and preconstruction. CON-315's constructability review is a safety review as much as a quality one, and Dana now asks the design team at every constructability meeting a simple question: how will someone build this, and maintain it, without falling? The same thinking is known in the industry as prevention through design.

On site, the hierarchy tells a foreman which fix to reach for first. When the drywall crew needed to work near the level 3 floor opening for the mechanical shaft, the first suggestion was harnesses. Omar asked whether the opening could be covered instead. It could, with a plywood cover screwed down and marked, for the two weeks before the shaft walls closed it in. The cover protected everyone, including the electrician who came through the next day and had not been at the huddle. Personal fall protection would have protected only the people who remembered to use it.

When conditions change

The most dangerous moment in many tasks is when something changes and the crew carries on with the old plan. The wind picks up, a different crane is used, a new worker joins, a delivery arrives early and blocks the exclusion zone, the scaffold beside the work is altered, or the task turns out to be different from what was planned. Thornbury's rule is simple: if the conditions change, stop and re-assess. The FLHA form has a section for changes during the shift, and foremen are expected to use it. A crew that stops for five minutes to re-plan when the wind rises is showing exactly the judgment the system is meant to build.

Keeping hazard assessments honest

The biggest weakness of hazard assessment forms is that they can become a signature exercise. A crew that signs the same pre-filled form every morning, with the same hazards ticked, has not assessed anything. Priya calls this pencil-whipping. Thornbury's foremen are coached to keep FLHAs short, specific and different each day, and the superintendent and safety coordinator check a sample during their site walks by asking a worker at the location what the hazards of their task are and comparing the answer with the form. If the worker cannot describe the hazards on the form they signed, the form is not doing its job.