The one idea: the water moves the heat, and the heat moves the water
In a Simple ground, heat moves only by conduction (plus a prescribed, constant groundwater velocity, if one is entered). In the fully coupled TH ground BoreTherm also computes the groundwater itself: the pressure (head), the Darcy flow q (the volume of water per second through one square metre of ground) from the head gradient, and the heat the water carries. The coupling works both ways. The borehole cools the water near it; cold water is heavier and more viscous, so the flow changes; the changed flow carries the heat differently.
q = −K(T) (∇h − Δρr(T) eg)K(T) = k ρ0 g / μ(T)Ss ∂h/∂t − nβ ∂T/∂t + ∇·q = QwC ∂T/∂t + (ρc)w q·∇T − ∇·(Λ∇T) = borehole heath is the hydraulic head (the water pressure as the height of a water column), k the permeability, Δρr = (ρ(T) − ρ0)/ρ0 the buoyancy, Ss the specific storage, n the porosity, β the thermal expansion of the water, Λ the conductivity of the saturated ground (plus dispersion). A temperature front drifts with the heat transport velocity vT = q (ρc)w/C, about 1.9 times the Darcy velocity here.
The workflow
The coupled ground is the right choice when the flow comes from the hydraulic situation (heads, wells, layers of different permeability) or when the temperature changes the water enough to matter. A single known Darcy velocity is handled by the Simple ground, in 2D and 3D. The 2D design is BoreTherm’s standard calculation (Calculation › Calculate and Calculation › Size borehole length), with the ground response from g-functions, the dimensionless temperature response of the field to a constant heat rate.
- Materials: per layer the porosity, K horizontal / vertical (stored as the permeability k) and the specific storage; the dispersivities; the pore water: λw, (ρc)w, the density law (Linear = buoyancy) and the viscosity law (Vogel). Values in italics are assumed (pre-filled).
- Groundwater flow: top, bottom and sides as No flow, Prescribed head or Prescribed flux; the regional head gradient on the head faces and its direction (counter-clockwise from east); wells; the initial state (at rest, or the steady flow). Compute groundwater flow shows the head or the Darcy velocity in a plan map.
- Physics & run › Coupling: iterations per step (25), relaxation (1 = none), temperature and pressure tolerances (10⁻⁴ K, 1 Pa). The defaults are converged.
How a coupled time step works
The time stepping is that of every Full FEM run (backward Euler, steps of 60 s × 2k, step doubling against the step tolerance). What the coupled ground adds is an iteration inside each step between the flow and the heat (a staggered Picard iteration):
- Water properties per element at the mean temperature of its nodes: density (buoyancy) and viscosity.
- Hydraulic step: the head at every node, with storage, the buoyancy load and the thermal expansion of the stored water. Head faces keep their head: the reference head plus the regional gradient, hydrostatic in depth.
- Water fluxes between neighbouring nodes. They balance the water of every node exactly, so the heat they carry is conserved by construction.
- Thermal step with the new flow: advection, conduction, dispersion and a streamline diffusion that acts only along the flow (it keeps the solution free of wiggles; it adds or removes no heat). The transient boreholes (a borehole model whose fluid, pipes and grout store heat) and the loop control are part of this step.
- Convergence when the temperatures change by less than 10⁻⁴ K, the pressures by less than 1 Pa, and the hydraulic equations hold for the new temperatures. Then the step is accepted (or rejected by the step doubling).
The theory: the domain, the plume and the steady state
The model block has hydraulic boundaries as well as thermal ones. In the example the top and the sides are Prescribed head with a gradient of 0.00044, the bottom is No flow, and the run starts from the steady flow. The mesh is extended downstream so that the plume stays inside the block.
How each result is derived
All values for the example, Full FEM with the coupled TH ground and full transient boreholes.
| Result | How BoreTherm derives it | Example |
|---|---|---|
| Water viscosity μ(T) | Vogel: 2.414·10⁻⁵ Pa·s · 10^(247.8 K / (T + 133.15 K)) | 1.300 mPa·s at 10 °C; 1.319 at the start (9.46 °C) |
| Hydraulic conductivity K | k ρ₀ g / μ(T); the page shows K at the reference temperature and stores k | 3·10−11 m² × 999.7 × 9.81 / μ = 2.26·10−4 m/s |
| Regional Darcy velocity q | K · i on the Head faces; computed by Compute groundwater flow (the steady flow) | 2.23·10−4 × 4.44·10−4 = 9.9·10−8 m/s; reported 9.9·10−8 m/s = 3.1 m/a |
| Heat transport velocity vT | q (ρc)w / C: how fast a temperature front drifts | 9.9·10−8 × 4.18 / 2.2 = 1.9·10−7 m/s ≈ 5.9 m/a |
| Block extension downstream | the mesh follows the drift of the heat over the run (estimated before the flow is computed, from K and the gradient) | 9.8 m (K × 10: 97.8 m) |
| Coupling iterations | per accepted step until the increments and the hydraulic defect are converged | 651 / 300 steps = 2.2, at most 3 |
| Water balance | water imbalance of all nodes / water throughput of the run | 1·10−13 |
| Energy balance | (heat in − stored − heat through the faces) / throughput | 5·10−15 |
| Tf,min, Tf,max of the year | monthly loads: the lowest and highest month-end mean fluid temperature, as in the 2D design | −5.13 / −2.75 °C |
| Effect of flow and storage | Tf,min (3D) − Tf,min (2D) | −5.13 − (−5.25) = +0.12 K |
| Required length, Size with 3D | start at the 2D size; steps of 15 %; Brent’s method to max(tolerance, 0.5 m); 4–16 full runs | 75.95 m (2D) → 75.13 m in 4 runs |
Limits to keep in mind
- Assumed inputs: here the porosity, specific storage, water density, reference temperature of the water laws are pre-filled values. The results page and the report list them; replace them with site data before relying on a coupled result.
- Mesh: Coarse in the example, for speed. With K × 10 the element Péclet numbers (advection over conduction) reach 25.6 (91.7 % of the volume above 2): the solution stays stable, but the plume is smeared. Repeat a design with Medium (benchmark B8: Coarse and Medium differ by 0.055 K).
- Physics: saturated, single-phase flow; no freezing of the ground or the brine; no unsaturated zone; constant brine properties. The water balance is a volume balance with the reference density (Boussinesq form).
- 2D design: Calculate, Size borehole length, optimisation and studies stay 2D. They see a groundwater flow only as the Darcy velocity of the Ground page; the coupled ground refuses that velocity (convert it to a head gradient).
- Domain: when more than 5 % of the borehole heat crosses the sides and the bottom, a note asks for a larger block. Near the borehole the radial sub-grid (a fine grid at the wall) is conductive; a note warns when the local flow is fast for it.
- Run time: every step iterates flow and heat (here 2.2 iterations) and Size with 3D needs several full runs. Physics & run estimates the time before you start.
Numbers calculated with BoreTherm. Theory: BoreTherm theory manual; benchmarks B6 / B7 (flow, transport, buoyancy), B9 (coupled TH + transient boreholes), B11 (3D sizing).