Shifting Paradigms in Greenfield Geothermal Systems

Solution

Our novel approach to resource assessment combines numerical modelling and uncertainty quantification to help address some of the uncertainties in early-stage geothermal developments. The method uses the same data as traditional stored heat calculations but also includes reservoir physics, wellbore physics and realistic energy extraction scenarios to provide more accurate forecasts of a resource’s potential output. This method leverages cloud computing resources and fast simulation software like Waiwera to run thousands of reservoir simulations in parallel. 

The method begins with developing a conceptual model of the geothermal system based on data from 3G surveys. These surveys provide insights into the geological and structural setting, alteration zone and deep upflow locations. The conceptual model informs the development of the numerical model, which can be continuously updated as more data becomes available. We refer to this workflow as integrated reservoir modelling.  

Greenfield systems are inherently uncertain. We represent this by treating the reservoir model parameters (e.g., deep hot upflows, rock permeability and porosity) as uncertain. Each parameter is assigned a statistical distribution that describes the probability of the parameter taking a given value. We generate an ensemble of reservoir models from these distributions, which are run to their natural state. Each sample model has a different set of reservoir parameters that may (or may not) represent the greenfield system. 

We use approximate Bayesian computation (ABC) to ‘condition’ the ensemble of reservoir models on the location and temperature of the base of the clay cap. This results in a ‘filtered’ set of reservoir models, which match the available exploration data and can be used for forecasting.  

We have developed an innovative production algorithm to forecast the potential power output of a greenfield system. This algorithm implements a realistic extraction scenario that maximises power production from each reservoir model by targeting favourable locations for production and injection wells. This results in a distribution of electrical power outputs, which reflects the uncertainty in the potential of the resource. This offers more realistic estimates than traditional methods as it can represent complex reservoir dynamics.