The ground model in a geotechnical assessment

The ground model in a geotechnical assessment is the interpreted picture of what lies beneath the site: the sequence and geometry of the strata, the groundwater regime, and the parameters that will be used in design. It is an interpretation, built from a desk study, a walkover and a set of exploratory holes, and it is the part of the report that a reviewer should read first. Everything downstream, from foundation type to the cost of the substructure, follows from it. A ground model that is really a borehole log with a cover sheet is the most expensive document on a scheme, because the money it costs arrives later.

What the ground model has to establish

Five things, and a planning-stage report should be explicit about each.

The stratigraphic sequence and how it varies across the site. The depth and nature of any competent bearing stratum. The groundwater regime, including perched water and seasonal variation. Any geohazards present, whether that is made ground, shrinkable clay, running sand, peat, solution features, mining legacy or slope instability. And the parameters that design will use.

Note that this is a different exercise from the conceptual site model built for contamination. The two share boreholes and often share a report, but they answer different questions and can reach different conclusions about the same ground. Keep them separately headed, and cross-refer rather than merge. Contamination assessment to BS 10175 must be carried out by a suitably qualified and experienced person, and the contaminated land assessment is a legally distinct piece of work from the geotechnical one.

Building stratigraphy from exploratory holes

Exploratory holes give you points. The model is what joins them, and joining them is an act of judgement rather than arithmetic.

Start with what is already known. Published geological mapping and archived borehole records from the British Geological Survey give the regional picture and often reveal a feature the site investigation will not reach. Historical mapping tells you where the made ground is likely to be, because it shows the pits, ponds and buildings that were filled.

Then look at spacing and depth. The relevant guidance in BS 5930 on the layout of exploratory holes is a starting point, and the right answer depends on the variability of the ground and the size of the structure. Two boreholes on a 1.5 hectare site tell you very little about a buried channel running diagonally beneath it.

Made ground deserves particular suspicion. Its thickness and composition can change over a few metres, and a model that shows a uniform 1.2 metre layer across a former industrial site is far more likely an artefact of the hole spacing than a finding.

Where the interpretation is uncertain, draw it that way. Dashed boundaries, alternative sections and a note saying which interpretation governs design are all more useful than a confident single line.

Groundwater and why one dip is not a regime

A water strike recorded during drilling is a snapshot taken while the ground is disturbed. It is not the groundwater level.

Standpipes or piezometers installed and monitored over time give you a regime. Monitor across enough of the year to capture seasonal variation, and record rainfall alongside the readings so that anomalies can be explained. Where multiple aquifers or perched water are possible, install responsive zones at different depths rather than one long slotted section that mixes them.

The planning consequences are real. Groundwater level determines whether a basement is viable, whether infiltration drainage will work, whether dewatering will be needed during construction, and whether there is a risk of settlement affecting neighbours. Each of those becomes a condition or an objection if the report cannot answer it.

Selecting characteristic values

Eurocode 7 asks for a characteristic value defined as a cautious estimate of the value affecting the occurrence of the limit state. That phrasing does the work, and it is worth being precise about what it excludes.

It is not the arithmetic mean of the test results. It is not the single worst result either. It is a cautious estimate, informed by the scatter in the data, the volume of ground involved in the failure mechanism and the degree to which the structure can redistribute load.

Show the derivation. A table of characteristic values with no route back to the test data cannot be checked, and a reviewer will either take it on trust or reject it. Neither is a good outcome.

Where the uncertainty stays

A planning-stage ground model is normally based on limited intrusive work, and saying so plainly is a strength rather than a weakness.

State what has not been investigated, what assumptions have been carried, and what further work is needed at detailed design. Where a feature could change the development's feasibility rather than its detailing, say that too. A mineshaft in a coalfield referral area is a different category of unknown from a slightly deeper layer of soft alluvium.

The honest uncertainty in most of these reports is lateral variability. You have interrogated perhaps a thousandth of the volume of ground the building will sit on, and the interpretation between holes carries more risk than any single test result.

Writing it for a planning audience

The case officer is not a geotechnical engineer, and the report is not being read for its parameter tables. Four questions decide whether it does its job at planning stage.

1. Is the site developable for the use proposed, and on what basis.
2. Does the ground condition drive anything that appears on the drawings: levels, retaining structures, basement extent, building footprint, tree retention.
3. Are there off-site effects, such as settlement, dewatering, vibration or groundwater change, that neighbours or consultees will raise.
4. What conditions are needed, and what should be reserved for detailed design.

Put those answers near the front, in plain language, with the technical detail behind them. Where trees are being retained near foundations, the relationship between the ground model and the tree survey needs to be explicit, because foundation depths on shrinkable clay near mature trees are a routine source of late redesign.

An illustrative example. A 60 unit scheme on a former brickworks has four boreholes showing 1 to 3 metres of made ground over firm clay. A fifth hole, added after an anomaly in the historical mapping, finds 7 metres of fill in the footprint of a former clay pit. The layout moves two blocks by 15 metres, converting a piled solution to shallow foundations across most of the site. The extra hole cost a fraction of the piling it avoided. Figures here are illustrative.

Feeding the model into foundation options

The model should end by setting out the foundation options it supports and the ones it rules out, with the reasoning attached.

That is also where the interface with the structural design gets tested. A Geotechnical Assessment that stops at parameters and leaves the engineer to infer the rest tends to produce a conservative design, because the engineer has no choice. One that sets out bearing options, likely settlements and the sensitivity of each to the uncertainty in the model gives the design team something to optimise against.

Where an existing building is being retained or underpinned, read the ground model alongside the condition grading in a structural survey, because existing movement and ground conditions usually explain each other.

Geotechnical work is something councils expect a suitably qualified author to prepare, and professional review is strongly advised before submission. An AI-assisted draft can build the report structure, set out the standards and policy framing and mark exactly where your factual data goes, then flag the values it does not hold rather than inventing them. The interpretation stays with the engineer.

Enter your postcode on PlanScope to see which technical reports a site is likely to need before you commission the ground investigation.

This is general guidance, not legal advice; check the position with your local planning authority or a qualified professional before relying on it. PlanScope provides AI-assisted draft reports for planning applications. Reports should be reviewed before submission. We do not guarantee planning approval. For specialist technical reports, qualified professional input may be required.