HomeFeaturesExamplesTutorialsHelpTheoryVerificationPricingContact
BoreTherm / TutorialsTutorial 03 / 19
Workflow and theory · Design basics

Ground layers and groundwater

How a layered ground enters the design, why two temperature models exist, how groundwater flow is modelled with the moving finite line source, and what it does to the fluid temperature, the borehole length and the thermal plume. The numbers come from two example projects of the same office building (heating 200 MWh, hot water 8 MWh and cooling 45 MWh a year, through a heat pump) on a field of 16 boreholes of 150 m (4 × 4, 6 m apart), over a design period of 25 years: one example project on five ground layers, and one on homogeneous ground with groundwater flow.

01 / The one idea

The one idea: the ground is not one block

The analytic model of BoreTherm's design calculation builds the g-function, the dimensionless step response of the borehole field, from finite line sources (FLS: each borehole a line of finite length releasing heat into the ground), and adds the loads by superposition. It needs one thermal conductivity λ and one volumetric heat capacity C. Real ground is a stack of layers, and it may carry flowing groundwater. BoreTherm handles the layers by weighting them over the active length of the boreholes, the depth range [D, D + H] from the buried depth D (top of the borehole) to D plus the borehole length H (or by giving every layer its own response), and the groundwater by letting the heat source move with the water: the moving finite line source (MFLS).

Layers: properties change with depthClay/silt λ = 1.7Sand λ = 2.4Claystone/siltstone λ = 2.2Limestone λ = 2.8Granite λ = 3.4qgeoradial heat flow: layers side by side (parallel)geothermal flux: layers in seriesGroundwater: the heat drifts downstreamplan, no flowplan, flow towards +xcooled ground carried awayextraction: isotherms of the cooling around one borehole
Left: the layers of the layered example project to scale. Around the borehole the heat flows horizontally through all layers side by side; the geothermal heat flows vertically through them one after the other. Right: in flowing groundwater the cooled ground is carried downstream, so the borehole always sees fresh ground upstream.
Effective properties over [D, D + H]λeff = Σ hi λi / H, Ceff = Σ hi Ci / H
Undisturbed temperature (geotherm)Tu(z) = Ts + qgeo ∫0z dz′/λ(z′)
Heat-transport velocity and Péclet numbervT = u (ρc)w / C, Pe = vT H / α
MFLS, receiver x downstream (c = vT/4α)F = evTx/2α ∫ s⁻² e−d²s² − c²/s² I(s) ds

hi: length of layer i inside [D, D + H]; Ts: ground surface temperature; qgeo: geothermal heat flux; u: Darcy velocity (specific discharge) of the groundwater; (ρc)w: volumetric heat capacity of the water; vT: velocity at which the heat moves; α = λ/C: thermal diffusivity; Pe: Péclet number, the ratio of heat carried by the flow to heat conducted over the borehole length; x, d: distance downstream and horizontal distance between source and receiver; s: integration variable from 1/√(4αt); I(s): the vertical part of the finite line source.

Everything after the g-function stays as for a homogeneous ground: the superposition (the load history as a sum of load steps, each answered by the g-function) uses λeff, the undisturbed ground temperature Tm averaged over the borehole and the effective borehole thermal resistance Rb* between the fluid and the borehole wall. Groundwater changes only the g-function; the grout is not flowed through.

02 / The workflow

The workflow

Two questions decide the ground model: how much is known about the layers, and does groundwater flow through them? Both are answered on the Ground page (Design › Ground). The design then runs as usual with Calculation › Calculate, which gives the mean fluid temperature Tf (the average of inlet and outlet), and Calculation › Size borehole length.

Ground pagewhat is known about the ground?Several layers withknown λ and C?noHomogeneousone λ and C, e.g. from a TRTyesStrong contrast, thin layersor sections along the depth?noLayered – effectiveweighted λ, C (default)yesLayered – segmentedeach layer its own responseGroundwater flow?Pe = vT H / α (Ground page)Pe < 1.2negligible: conductiondominates1.2 ≤ Pe ≤ 10groundwater matters: movingfinite line source neededPe > 10advection dominates(axial effects negligible)Then Calculate, Size borehole length and Results › Thermal plume. With flow > 0 the MFLS is always used; Pe only grades it.
The example's groundwater case has Pe = 24.1 (highlighted). The layered case uses the effective model, the default for layered ground.
Design › Groundmodel, layer table, Tm model;groundwater u, direction,vT and Pe shownCalculationCalculateSize borehole length(properties pertrial length)ResultsSummary,Temperature,SizingThermal plumeisotherms, lengthdownstream, width,clearancesOptional: Advanced analysis › 3D analysisFEM with the layers as they are and advectionmain pathcheck
All inputs are on one page. Sizing evaluates the layer-weighted properties again at every trial length.
  • Ground model: Homogeneous, Layered – effective properties, Layered – segmented. Layers are contiguous from the surface; the last layer extends to infinity.
  • Undisturbed temperature: Mid-depth (Ts + qgeo (H/2)/λeff, the rule of the classic monthly design method, the default) or Active-length average (the mean of the geotherm over [D, D + H], used by the layered example project).
  • Groundwater: a uniform aquifer (Darcy velocity u in m/yr, flow direction from +x counter-clockwise, water heat capacity), or a Darcy velocity per layer in the layer table. The page shows vT and Pe.
03 / How the calculation works

How the calculation works

1 Reference valuesover [D, D + H] of the field:λeff, Ceff, αeff, Tm2 Ground modeleffective: one layer;segmented: every layer3 Segmentssplit at layer interfacesand section boundaries4 Pair kernelFLS with the pair’s λ, α;flow: MFLS, × exp(vTx/2α)5 g-functionsegment heat rates (UBWT);pairs: down-, upstream, across6 Superpositionλeff, Tm, Rb* without flow→ mean fluid temperatureSize borehole length repeats steps 1–6 at every trial length H: λeff, Tm and Pe change with H.
Steps 1–6 for one length. The effective model skips the split at layer interfaces (one homogenised layer); without flow the pair kernel is the plain finite line source.
  1. Reference values at the field's mean D and H: λeff, Ceff, αeff and Tm by the chosen temperature model. They define the units of the g-function (ts = H²/9α).
  2. Ground model. Effective: one layer with λeff, Ceff and the thickness-weighted mean Darcy velocity. Segmented: every layer with its own λ, C and vT.
  3. Segments. Segmented: the boreholes are split at every layer interface (and borehole section boundary), so each segment lies in one layer. A thin layer gets its own thin segment.
  4. Pair kernel. Two segments in one layer respond with that layer’s properties; in two layers with the mean λ and C of the pair. With flow, the kernel is the MFLS: the same integral plus a direction weight evTx/2α (x = how far the receiver lies downstream).
  5. g-function. The segment heat rates under a uniform borehole wall temperature (UBWT) give g. With flow, the pairs are grouped by distance and by position: downstream, upstream or across the flow.
  6. Superposition of the monthly loads as for a homogeneous ground.
04 / Theory

Theory: layers

The layers of the layered example project and their weight in the effective properties (the length of each layer inside the active length of the boreholes):

LayerDepth [m]λ [W/(m K)]C [MJ/(m³ K)]Inside [4, 154] mWeight
Clay / silt, water saturated0 – 81.72.443 %
Sand, water saturated8 – 252.42.51711 %
Claystone / siltstone25 – 802.22.35537 %
Limestone80 – 1252.82.24530 %
Granite125 – ∞3.42.42919 %

Why the arithmetic mean. Around a borehole the heat flows horizontally, through all layers at the same time. The layers act as conductances in parallel, so their thickness-weighted mean is the right average. Layers above D and below D + H do not count, so λeff, Ceff and Tm change with the borehole length.

Undisturbed temperature with depth910111213050100150temperature [°C]depth [m]active-length average 11.07mid-depth 10.72Tu(z)Conductivity per layer and its mean01234050100150λ [W/(m K)]SandClaystone/siltstoneLimestoneGraniteboreholeλeff = 2.62
The layered example project. Left: the geotherm bends at every interface; it is steepest in the poorly conducting clay. Right: λ per layer and the thickness-weighted λeff over the active length (orange bar).
Why the two temperature models differ. The geothermal heat crosses the layers in series. A poor conductor near the surface (here 8 m of clay with λ = 1.7) acts like an insulating blanket and lifts the temperature of everything below it. The mid-depth rule uses only λeff and ignores where the layers lie (and D), so it gives 10.72 °C. The active-length average follows the real profile and gives 11.07 °C: 0.35 K warmer, which shortens the heating-dominated field from 143.1 m to 140.3 m.

Effective or segmented? The effective model assumes every layer takes heat in proportion to its thickness. In reality a conductive layer takes more. The segmented model resolves that; for the example (contrast 1.7–3.4) it gives Tf,min −2.00 °C against −1.92 °C and a length of 140.9 m against 140.3 m. Use it for strong contrasts, thin poor layers or sections along the depth.

05 / Theory

Theory: groundwater

g-function of the 4×4 field1 h10 h100 h1000 h10⁴ h10⁵ h0510152025no flowwith flowtimegLowest fluid temperature per year, H = 150 m1510152025−3−2−10123groundwaterlayeredhomogeneouslimityear°C
Left: without flow g keeps growing; with flow it reaches a steady state, because the water removes the cold as fast as the field makes it. Right: the lowest fluid temperature per year at 150 m; with groundwater it stops falling after a few years.

The moving point source of Carslaw and Jaeger, integrated over two vertical segments, gives the MFLS kernel of section 1. Two things change against the conductive finite line source: a factor evTx/2α makes a receiver downstream feel the source more than one upstream, and the term e−c²/s² cuts off the long times, so the response becomes steady. The example's u = 1.0·10⁻⁷ m/s (3.2 m/a) gives vT = 6.0 m/a: the heat moves faster than the water's specific discharge u because the water carries heat into a ground that stores only C, not (ρc)w, per volume.

The counter-intuitive part. A Darcy velocity of only 3.2 m/a, a few metres a year, cuts the required length from 143.7 m to 120.8 m (−16 %) and lifts Tf,min at 150 m from −2.41 °C to −0.19 °C. The Péclet number over the borehole length is 24: advection dominates. The design depends strongly on a value that is often poorly known — vary u in a parametric study (Tools › Parametric study) before relying on it.
No flowGroundwater towards +x−3 K−2 K−1 K−3 K−2 K−1 K108 m downstream of the field (−1 K)width 64 m16 boreholes, 25 years; isotherms of the cooling
Isotherms −1, −2, −3 K of the ground temperature change after 25 years, averaged over the active length (plume command; the no-flow shapes are offsets of the field by the reported reach; with flow the downstream length and the width are the reported ones, the upstream side is schematic).

The price of the flow is a plume: the cooled ground ends up downstream. Without flow the −1 K isotherm reaches 35 m around the field; with flow it reaches 108 m downstream, over a width of 64 m. Check it against property lines and neighbouring boreholes on Results › Thermal plume.

06 / How each result is derived

How each result is derived

ResultHow BoreTherm derives itExample project
Effective conductivityλeff = Σ hi λi / H over [D, D + H] (parallel layers)4·1.7 + 17·2.4 + 55·2.2 + 45·2.8 + 29·3.4 over 150 m = 2.621 W/(m K)
Effective heat capacityCeff = Σ hi Ci / H; αeff = λeff / Ceff2.315 MJ/(m³ K); α = 1.13·10⁻⁶ m²/s
Undisturbed temperatureactive-length average of Ts + qgeo ∫dz/λ over [D, D + H]; mid-depth: Ts + qgeo (H/2)/λeff11.07 °C (example); mid-depth 10.72 °C
Time scalets = H² / (9 αeff), the time axis of the g-function2.21·10⁹ s = 70 years
Heat-transport velocityvT = u (ρc)w / C (u = Darcy velocity)1.0·10⁻⁷ × 4.18 / 2.2 = 1.90·10⁻⁷ m/s = 6.0 m/a
Péclet numberPe = vT H / α (Ground page; Molina-Giraldo et al. 2011)1.90·10⁻⁷ × 150 / 1.18·10⁻⁶ = 24.1: advection dominates (axial effects negligible)
g-function, end of the gridMFLS kernel; flow carries heat to or away from the field and g levels off14.06 with flow, 23.88 without
Lowest fluid temperaturesuperposition of the monthly loads with Tm, λeff, Rb* unchanged by the flow−0.19 °C with flow, −2.41 °C without (H = 150 m)
Plume, −1 Kisotherm of the monthly base loads at the end of the period (plume command, analytic)108 m downstream, 64 m wide; no flow: 35 m around the field
07 / The example in numbers

The example in numbers

GroundλeffTmTf,min at 150 mTf,max at 150 mSized length
Homogeneous, no flow2.6010.73 °C−2.41 °C16.03 °C143.7 m
Layered – effective, active-length average (example)2.6211.07 °C−1.92 °C16.34 °C140.3 m
Layered – effective, mid-depth2.6210.72 °C−2.27 °C15.99 °C143.1 m
Layered – segmented, active-length average2.6211.07 °C−2.00 °C16.34 °C140.9 m
Homogeneous, groundwater 3.2 m/a (example)2.6010.73 °C−0.19 °C16.04 °C120.8 m

Design limits −3.0 °C / 20.0 °C; the lowest temperature limits every length. Rows 1, 3 and 4 are the example projects with one input changed (no flow; mid-depth; segmented).

  1. Open the layered example project. Ground page: five layers, Layered – effective properties, Active-length average.
  2. Calculate: Tf,min −1.92 °C, Tf,max 16.34 °C. Size borehole length: 140.3 m.
  3. Open the groundwater example project. Ground page: groundwater 3.2 m/yr towards +x, Pe 24.1. Calculate: Tf,min −0.19 °C, Tf,max 16.04 °C.
  4. Results › Thermal plume: compare the extent downstream with the case without flow.
08 / Limits to keep in mind

Limits to keep in mind

  • Effective layers ignore vertical heat flow between layers and give every layer its thickness share of the heat. The segmented model uses the mean properties of two layers for a pair of segments; against a finite-difference reference it deviates 0.6–2.2 % at 50 years (strong contrast, thin poor layer).
  • Mid-depth is the default because it reproduces the classic monthly design method. With known layers and buried depth the active-length average is the better estimate; the difference is typically a few tenths of a kelvin.
  • Groundwater is uniform, horizontal and constant in time; no vertical flow, no thermal dispersion, the aquifer layers are horizontal and infinite. Tm and Rb* do not change with the flow.
  • Darcy, not pore velocity. The input is the specific discharge u = K · i (hydraulic conductivity × gradient). A pore (seepage) velocity is u / n and would overstate the flow several times.
  • Strong flow can expose near-field limits. The 3D analysis (a finite element model, FEM, of the ground with advection) is an independent check; the classic monthly method has no groundwater model, and the Equivalent g-function method falls back to the Similarities method with flow.
  • The thermal plume uses the monthly base loads (no peaks) in an effective homogeneous ground.

Numbers calculated with BoreTherm. Theory: BoreTherm theory manual.