Whitecliffe by Ambisense

Shortlisted for Brownfield Awards Category 1 - Best Project Preparatory Work

A robust assessment combining best practice and innovative data science analysis methods characterises the work at Whitecliffe. The conclusions from this analysis showed that gas membranes were not required in the buildings which resulted in significant cost savings of over £1m and increased sustainability, while still providing a safe development. The preparatory work also helped to rationalise the post earthworks monitoring for future stages, leading to additional savings and reduced programme risk. 

Stages of Project Preparatory Works 

1) Presenting options and demonstrating value
There was a proven methodology for gas investigation and assessment at the site from Stage 1. Rather than continuing with the status quo during Stage 2, which would have likely resulted in gas membranes being required across the development, the opportunity to innovate and add value was identified. Arup presented multiple fully costed options to the client to demonstrate the value of an intensive scheme of investigation for the first phase of development. The goal was to establish the low risk from ground gases at the site, remove the need for membranes, and enable a streamlined scope of investigation across future phases.  


2) Preliminary Conceptual Site Model and design of the site investigation
The Stage 1 data on Ambilytics was reviewed by EPG, providing deeper insight into the likely gas generation mechanisms causing the elevated CO₂. Concentration gradients of CO₂ increasing with depth (Figure 1) showed that the flow of gas through the ground would be driven by diffusion.  


Figure 1: Carbon dioxide concentrations with depth in Stage 1

Modelling the diffusive flow provided an indication of likely surface emissions. Results from flux boxes verified the model (Figure 2).  

Stockpiles Soils with Hydrocarbon Impact Prior to Treatment


Figure 2 Summary of modelling of gas flow into flux boxes and buildings

Analysis of the results also demonstrated that the stack effect was driving flow rates, as these increased with well depth. The deeper wells acted as chimneys.  


This reassessment demonstrated best practice through understanding the likely processes that could be generating gas, considering the geology, and developing a preliminary conceptual site model for gas to support the design of the Stage 2 investigation.


The Ambilytics platform was used for the initial assessment of the historical data; the multi-source, contextual data was analysed, while machine learning rapidly identified patterns and relationships, and generated statistical models. This suggested that the three most significant variables influencing CO₂ concentrations in the wells were differential pressure, atmospheric pressure, and temperature.

This reassessment demonstrated best practice through understanding the likely processes that could be generating gas, considering the geology, and developing a preliminary conceptual site model for gas to support the design of the Stage 2 investigation. 


However, a key limitation of the continuous monitoring data collected for Stage 1 was the absence of flow rates.

Consequently, the post earthworks investigation in Stage 2 was designed to allow a more robust gas risk assessment. 

The investigation included:

  • Installing shallow and deep wells to assess the gas regime at different depths and in different strata.

  • Simultaneous continuous gas monitoring, including flow rates, in all wells. The concurrent monitoring in all 32 wells was a crucial element to understanding the drivers for gas generation and migration.

  • On Ambilytics, linking meteorological data from a local weather station to each gas monitoring result. This allowed consideration of the relationships between rainfall, atmospheric pressure etc and the gas parameters.

  • Surface emissions surveys and flux chamber tests to better understand the relationship between the flow of gas from boreholes and the emissions from the site surface and to allow verification of models.

  • Falling head permeability tests to confirm the low permeability of the ground.


Arup engaged in proactive stakeholder engagement in presenting the analysis of the Stage 1 data, the updated conceptual site model, and the resulting investigation strategy to the NHBC, for agreement ahead of starting the works. 

3) Lines of evidence considered
Best practice requires consideration of multiple lines of evidence. To support the gas risk assessments, further testing in addition to the gas monitoring included:


  • Testing of the fill materials before and during earthworks being carried out, ensuring they would meet the engineering specification and be compacted appropriately. There was full-time site attendance during the works by Arup. Total organic carbon (TOC) testing was conducted on the fill materials to understand the potential source of gas production below the site. Very low TOC results and the high performance of the earthworks fill materials were key lines of evidence in demonstrating the low risk.

  • Intelligent data analysis and modelling to understand variations in gas flow from the different strata in response to changes in atmospheric pressure and other influences such as rainfall. The data was used to assess likely generation rates and the intelligent models showed potential gas flow from the ground into buildings.

  • Analysis of the surface emissions measurements.

This approach is consistent with that described in BS8485: 2015 + A1: 2019.  


4) Conceptual site model

Understanding the geology and potential processes that can cause elevated gas concentrations is crucial.  The former quarry has been backfilled with natural reworked overburden materials comprised of:

  1. Lambeth Group: with low levels of organic carbon (plant debris concentrated in thin lignites), carbonate (shelly clays and glauconitic, calcareous pebbly sandstone of the Upnor Formation), calcite nodules, and pyrite.

  2. Thanet Sand: containing low levels of organic carbon derived from ancient biological deposition of matter (kerogen) and carbonate (shells). Glauconite is known to cause oxygen deficiency in the soil atmosphere.

  3. Chalk is present below the fill and contains carbonate and low levels of organic material.


There are several potential contributors to the CO₂ found on this site:


  • Carbonate dissolution from acid rainfall or humic acid from plants.

  • Carbonate dissolution due to pyrite (sulphide) oxidation producing sulphuric acid.Oxidation of organic material within the soils.

  • Oxidation of ferrous carbonate, a component of siderite.

None of these processes are likely to cause large surface emissions. The investigation also identified methane in the groundwater in the chalk layer at depth. 


Figure 3 Conceptual Site Model for ground gas

5) Analysis and modelling
A tiered approach to analysing the gas monitoring data was adopted starting with the screening of hazardous gas flow rates (HGFRs) using the NHBC Green limit for methane and carbon dioxide. Shallow wells in the engineered fill were indicative of a low-risk gas regime with low HGFRs and occasional methane concentrations below the limit of 30% for “whole gas produced by anaerobic degradation” as defined by Ekland (2011).  The methane is most likely caused by reduction of carbon dioxide from other sources. 

The gauge pressure never exceeded 500Pa in any of the wells, and diffusive flow dominates as described in ASTM E2993-16. Following best practice, the effect of covering the ground with areas of hardcover was considered. It was deemed unlikely that small footprint buildings with ventilated voids below, or disconnected areas of parking and roads would change the gas regime.

The maximum HGFR for NHBC Green for methane and carbon dioxide was exceeded in the deeper wells, but was variable.  The more significant rainfall events were found to coincide with falls in barometric pressure (Figure 4).  Thus, any changes could not automatically be attributed to changes in barometric pressure. 


Figure 4 Barometric pressure and rainfall variations

Intelligent analytics and machine learning generated statistical modelling identified an interdependence between HGFR and both rainfall and barometric pressure (Figure 5). HGFRs were dependent on:

  • Mainly rainfall in some wells, with little influence from barometric pressure.

  • Occasional changes in barometric pressure when there was less rainfall.


Figure 5 Example 2D dependency of methane HGFR on rainfall and barometric pressure 

Rainfall beyond approximately 5mm caused increased moisture at depth in the soil, increasing biological respiration. The rainfall sealed the ground surface and the combined effects resulted in elevated CO₂ levels in the wells. The flux box tests showed increased emission due to higher rainfall, likely due to dissolved oxygen being carried into the soil and, along with the moisture, was causing carbon dioxide generation by biological oxidation of organic material. 

Ternary plots were used to interpret gas concentrations (Figure 6). The plot shows that the gas concentrations are not indicative of anaerobic degradation of organic material in the ground, or landfill gas migration from an off-site source (the blue dotted line shows the composition line for landfill gas/gas migration). 


Figure 6 Example ternary plot

Results of flux box tests were used to calibrate a model of diffusive gas flow into a ventilated void, as in Stage 1. The model used the gas concentrations recorded in monitoring wells close to the flux box locations. The air-filled porosity of the soil in the pathway was adjusted until the surface-emission rate of CO₂ into the void equalled that measured by the flux boxes.

The assessment showed that in all cases the concentration of hazardous gases in the underfloor void was very low and the predicted concentrations in the occupied space were negligible, more than an order of magnitude below already conservative minimal risk limits.

Methane present in deeper wells was shown to be partitioning from groundwater and was not migrating through the fill material to the ground surface.


6) Future changes that could influence the gas regime

The risk assessment considered the impact of future changes in conditions. Large changes in groundwater level had the greatest potential to increase gas emissions. Over the ten years since quarry dewatering ceased, the groundwater had recovered to its original level. It was then lowered again for the earthworks operations. The measured rate of recovery from site-specific long-term monitoring data was shown to not pose a risk of gas emissions.

Climate change may cause greater and faster drops in atmospheric pressure or changes in temperature and rainfall. However, there is no significant source of gas in the ground. Both the volume of gas is low, and the permeability of the ground is low. Neither of these factors could be affected by climate change to such an extent that the gas risk would become unacceptable. 



7) Stakeholder acceptance

The success of the approach on this site has relied upon close co-operation with the NHBC Land Quality team. Both the investigation method and approach to the interpretation of the data were agreed upon before implementation. Interim face-to-face meetings between Arup, EPG and NHBC were held to communicate the results and discuss the approach and lines of evidence in the assessment prior to the final report being issued. This led to approval by the NHBC after a rigorous review process.


Summary of Results

A simple assessment of the site considering only the maximum gas concentrations and flow rates would have concluded that the site was high risk. An innovative and comprehensive assessment using advanced analytics and modelling, considering additional contextual data and thorough characterisation of the sources and drivers for the ground gases showed that there was minimal risk of gas emissions to buildings. This provided significant value to the developer, in cost savings on mitigation measures, increased land value, and a more sustainable solution. Verification gas monitoring in future phases will now also be rationalised. Engagement with the NHBC provided confidence to allow agreement with the proposals.


The cost of the site investigation, continuous monitoring with Ambisense GasfluX units and the detailed assessment by EPG including the use of the Ambilytics software was minimal compared to the savings of over £1m achieved on the project.