A bridge load test gives you a spreadsheet of numbers that look reassuringly precise: deflections to a tenth of a millimetre, strains to a few microstrain, movements logged every second. Precision, though, is not the same as meaning. The numbers only become evidence when you compare them with something — a prediction, an earlier load step, or the reading taken before the load arrived.
Most of the interpretation work comes down to one question asked repeatedly: what came back when the load was taken off, and what did not? Everything below is a variation on that theme.
Start with the load–deflection plot, not the peak value
Before you look at any single figure, plot deflection against applied load for every instrumented point, with the loading and unloading branches on the same axes. This one graph tells you more than a table of maxima ever will.
A narrow hysteresis loop — the loading and unloading lines close together — means the structure absorbed very little energy and returned to where it started. A wide loop means something was moving, seating, slipping or cracking. A curve that bends progressively towards the deflection axis means stiffness was falling as load increased, which is the definition of non-linear behaviour.
Accept that the first part of the loading branch is usually soft. Bearings settle onto their seats, packers compress, bedding mortar takes up, and gaps close. On many structures the first ten to twenty per cent of the load does little more than tidy up the connections. What matters is that the response becomes proportional and stays that way through the working range.
Deflection: three numbers, not one
For each measuring point, extract three values:
- Total deflection at the peak load step, relative to the zero reading taken immediately before loading.
- Elastic deflection recovered once the load is removed and the structure has had time to settle.
- Residual deflection, the difference between the two.
Compare the elastic deflection with your predicted value. If the two agree within the tolerance set out in the test specification, the structure is behaving broadly as the analysis assumed. Where they disagree significantly, the cause is usually one of a familiar handful: support conditions stiffer or softer than modelled, unintended composite action, unmodelled parapet or surfacing stiffness, or a model that underestimated shear deformation in a short span.
Read the shape as well as the size
A deflection profile along the span is more informative than its maximum. A smooth, roughly parabolic curve with a midspan peak and near-zero movement at the supports is what you expect from a simply supported span. Asymmetry suggests uneven support stiffness, a bearing that is not working, or load applied off-centre. A quarter-point reading nearly as large as midspan deserves an explanation. Negative readings, where a point rises, are not automatically wrong — they can be genuine for cantilevers or continuous spans — but they should match the predicted deformed shape.
Strain: direction and distribution matter more than magnitude
Strain gauges tell you what the material is doing, not just where the structure went. Check three things.
Sign. A gauge on the soffit of a sagging span should read tension. A gauge reading compression there is either mislabelled, temperature-affected, or telling you the section is behaving in a way you did not expect.
Magnitude, against prediction. With a known modulus, measured strain converts to stress and can be compared with calculated values. Treat the comparison as an order-of-magnitude check, because concrete modulus varies, gauges drift, and the effective section may be stiffer than assumed.
Distribution through the depth. Strains that vary linearly from compression at the top to tension at the bottom indicate that plane sections are staying plane. A kink in that distribution often points to cracking in the tension zone, partial composite action, or a shear connection that is not transferring load as designed.
Remember that strain readings drift. Gauges bonded to concrete are sensitive to temperature, moisture and creep, and the zero you set on a cold morning may not hold at midday. Take a baseline reading set with the structure unloaded, repeat it at intervals, and keep a dummy gauge on unloaded material for comparison.
Residual movement: the number that earns its keep
Not everything that fails to come back is damage. A small, consistent residual is normal and expected. Typical innocent causes include bearings settling onto new seats, packers relaxing, hairline cracks closing imperfectly, and fines working into gaps that were already there. On a first load cycle, a modest permanent set is often simply the structure finding its seating.
What should prompt closer investigation is residual movement that is large in relation to the elastic response, that grows with each successive load cycle, or that is accompanied by visible change. Walk the structure after each cycle. Look for new cracks, widening of existing ones, crushing at bearings, paint flaking over a weld, or local buckling in a steel compression flange. A residual deflection with no visible explanation is a question, not a conclusion.
Warning signs in the data
Certain patterns should stop a review and force a second look:
- Deflection increasing disproportionately with each load increment, so the plot flattens progressively.
- Different behaviour between repeat cycles — the same load producing noticeably more movement on cycle two than cycle one.
- Sudden step changes in a single reading, which are usually instrumentation problems but occasionally are not.
- Strain readings that reverse sign or jump non-linearly at a particular load level.
- Residual deflection that keeps accumulating rather than stabilising.
Temperature and time are moving the bridge too
A concrete deck and a steel girder do not expand at the same rate, and a bridge can move several millimetres over a daily temperature swing. That movement will appear in your displacement readings unless you control for it. It is why many tests run at night or in the early morning, when temperatures are stable.
Long hold periods introduce creep and settlement. A reading that drifts steadily during a two-hour hold may reflect nothing more sinister than thermal change or the supports taking up load. Log temperature alongside displacement, take unloaded readings before and after each cycle, and compare any drift against what the structure was doing before the test started.
Reviewing the data: a short checklist
- Confirm the zero readings were taken with the structure unloaded, at a stable temperature, and after any overnight settlement.
- Plot load against deflection for every point and examine the loop width.
- Split total deflection into elastic and residual for each load step.
- Compare elastic deflection and strain against predicted values and record the differences.
- Check the deformed shape and the strain distribution through the section, not just the peak numbers.
- Compare repeat cycles for consistency.
- Walk the structure and match physical observations to the data.
Interpretation is a judgement call, and the person best placed to make it is the engineer responsible for the assessment, working against the design model, the specification and the relevant standards. If the data does not support a clean verdict, say so, and recommend further monitoring, a repeat test at a different load level, or a more detailed assessment. A test that raises a good question has done its job.
Photo: ChiemSeherin / Pixabay
