Identification of the three dimensional distribution of temperature and composition in the lowermost mantle is necessary to diagnose the pattern of mantle convection and identify controls the extraction of heat from the outer core. In principle a range of geophysical parameters are sensitive to the temperature and composition of this region but trade-offs between competing effects combined with data and model uncertainties makes the creation of a unique model difficult. Here we describe a suite of models where we calculate the seismic, geodynamic, and thermal properties of the core and mantle from an assumed pattern of temperature and mineralogy by making use of a self-consistent parameterisation of the thermo-elastic properties of mantle minerals. For a given pattern, we predict the long-wavelength surface geoid, core-mantle boundary topography, inner-core radius, total surface heat-flux, and mantle seismic velocities. These predictions can be compared with geophysical observations or models. Key features of this approach are that we do not assume a particular scaling between seismic velocities or density and temperature, that we account for the tomographic resolution when comparing seismic velocities with tomography, and that phase transitions in the lowermost mantle emerge from the same mineralogical model as the bulk properties of the mantle. As well as being useful to compare realisations of different conceptual models of the structure and dynamics of the lowermost mantle, our approach is fast enough to permit a very large number of model cases to be evaluated in a reasonable timeframe. This permits the model to be used to infer the properties of the lowermost mantle in a Bayesian approach, or to allow best fitting models to be found using a local non-linear inversion scheme.Invited presentation at the AGU Fall Meeting in 2018, abstract DI51A-08