HomeFeaturesExamplesTutorialsHelpTheoryVerificationPricingContact
BoreTherm / TutorialsTutorial 16 / 19
Workflow and theory · Advanced physics

Full FEM with transient boreholes

How Full FEM, a direct 3D finite element (FEM) simulation, steps the ground and the boreholes together, why the borehole buffers short loads, how the time steps are chosen and when the step tolerance must be tightened. The numbers are those of the example project: one 50 m single-U borehole, 2.0 W/(m·K) ground at 10 °C, an hourly heat-pump load over 72 h (a 6-hour peak of 2.5 kW, a load stop, a second peak, a pump stop), 1 year simulated.

01 / The one idea

The one idea: the borehole is a small heat store

The 2D design (Calculation › Calculate and Calculation › Size borehole length, with the ground response from the g-function) treats the borehole as a resistance: the fluid is always R*b·q colder (or warmer) than the borehole wall (R*b the effective borehole resistance, q the heat rate per metre), and the heat the heat pump takes crosses the wall at the same instant. A real borehole holds fluid, pipe walls and grout. They store heat. When the load starts, the first heat comes from this store; when it stops, the ground refills it. Full transient boreholes model this store, and the fluid that needs time to travel down and up the pipes.

Full FEM with full transient boreholesHeat pumpevaporatorBoreholefluid · pipe walls · groutstore heat: EbGround3D FEM meshloop powerPloop = ṁcp(Tin − Tout)wall powerPwallPloop − Pwall = dEb/dt : the borehole buffers the loadThe 2D design (Calculate) and steady-state boreholesHeat pumpBoreholeresistance R*b only, no storageGroundg-function (2D)Pwall = Ploop at every instant; Tf = Tb + q · R*bThe example, first hour of the peak: the heat pump takes 2.50 kW, only 1.20 kW cross the borehole wall.
Top: Full FEM with full transient boreholes. The heat pump sets the loop power; the wall power follows with a delay. Bottom: the steady model of the 2D design.
Borehole store (loop − wall = storage)Ploop − Pwall = dEb/dt
Fluid loop (the heat pump’s heat rate)Ploop = ṁcp(Tin − Tout)
Borehole time constantτ = max(Cb·R*b, 2 ttransit)
Mean fluid temperatureTf = (Tin + Tout)/2

Heat rates are > 0 into the ground; Cb is the heat capacity per metre, ttransit the fluid’s time down and up. The time constant τ tells where the store matters. In the example, Cb = 37.4 kJ/(m·K) and R*b = 0.109 m·K/W give τ = 68 min. Load changes shorter than a few τ (hourly peaks, stops) look different; monthly loads hardly do.

02 / The workflow

The workflow

Full FEM is a check of a chosen design with the actual loads, not a replacement of the 2D design. Two questions decide the settings: do short load changes matter, and which results will you read?

Start: the 2D design of the fieldCalculate, Size borehole length;monthly or hourly loadsDo load changes of minutesto hours matter?noyesThe 2D design is enoughmonthly design values andSize borehole length; Simple FEMchecks the g-function of anunusual field in 3DFull FEM, full transient boreholesProject: 3D analysis = Full FEMPhysics & run: Boreholes = Full transient,Control = Automatic (fluid loop)e.g. hourly peaks,load and pump stops,inlet-temperature limitsWhich resultsdo you read?monthlyMonthly design valuesstep tolerance 0.01 K (default)hourlyHourly values, lowest Tf after a stopstep tolerance 0.003 K (1.3–1.6 × time)Groundwater flow from heads,wells or layers: Ground = Fullycoupled TH (thermo-hydraulic).
The default step tolerance (0.01 K) is set for monthly design values. Hourly values and the first interval after a load or pump stop need 0.003 K.

In the program, Full FEM adds the group Input › Full FEM and the page Results › Full FEM results:

main pathProjectCalculation › 3D analysis= Full FEMLoadstime series: ground loadand flow (0 = pump off)Full FEM › Physics & runphysics, control, stepping,tolerance · Run full 3DResults › Full FEM resultssummary, balances, chart,monthly table, CSVFull FEM › Materialsλ, C per layer; default= the Ground page valuesFull FEM › Transient boreholescells, sub-steps, near-borehole model, capacities3D analysis pagemesh, ground temperaturefield, 2D / 3D tabResults › Hourly temp.the 2D design (Calculate)for the comparisonground fieldCalculate, Size borehole length, optimisation and studies stay 2D; the 3D run checks the chosen design.
Physics & run holds every setting of the run and starts it. Materials and Transient boreholes feed it; the results appear on Full FEM results and on the 3D analysis page.
  • Physics & run: Ground Simple or Fully coupled TH (with groundwater flow); Boreholes Steady-state or Full transient; Control Automatic (with transient boreholes: the heat rate of the fluid loop, i.e. the heat pump); Time stepping (initial step 1 h, smallest step 1 min, largest step 30 d, tolerance 0.01 K); initial state and thermal boundaries. Run full 3D simulation (also Calculation › Run 3D analysis); the line below estimates the run.
  • Materials: λ and C per layer; by default the Ground page values, so the 3D run and the 2D design use the same ground (Check ✓).
  • Transient boreholes: axial cell length (automatic 1 m), sub-steps per transit (automatic 40), the near-borehole model (radial sub-grid, recommended), the pipe-wall capacity (PE 1.8 MJ/(m³·K)), and the summary: capacities per metre, transit time, time constant.
03 / How the calculation works

How the calculation works

Full FEM computes no g-function. It steps the 3D ground mesh in time with the actual loads (backward Euler, an implicit time step) and, inside every ground step, the borehole columns in short sub-steps (multirate). Both meet at the end of each ground step, in one linear system.

1 Choose the step Δtladder 60 s · 2^k; after a changeof the input: 32 min2 Borehole columnssub-steps (transit / 40), wallheat linear in ring temp., Tin3 Loop control + solveTin so that ṁcp(Tin − Tout) =request; ground mesh + rings4 Step doublingone step of Δt againsttwo steps of Δt / 2error ≤ tolerance?max change of groundand wall temperature5 Acceptthe two half steps;outputs per intervalyesno: reject, Δt / 2, againnext stepError ≤ tolerance / 4: the next step is twice as long (up to the largest step). The lowest and highest value of an intervalcome from the borehole sub-steps of the accepted steps; Tin is constant within each ground step.
The adaptive time stepping. Every distinct (configuration, Δt) pair is factorised once and reused, which is why the steps lie on a ladder of powers of 2.
  1. Step size. Steps are 60 s × 2k, at most 30 days. After every change of the input (load, flow, pump on / off) the step restarts at the largest rung not above the initial step: 1 h → 32 min. Every event is hit exactly.
  2. Borehole columns. One 1D column per string of the network: axial cells of about 1 m with the fluid of each leg and the grout nodes of the cross-section. After a change the sub-steps are transit / 40 for three transits, then they double. The column is linear, so the wall heat of the step is an affine function of the ring temperatures and Tin (unit responses, cached per step size, plus the response of the stored state).
  3. Loop control and solve. With Automatic = loop power, Tin follows in closed form so that the step-mean ṁcp(Tin − Tout) equals the load. Tin is constant within a ground step. The ground mesh and one ring unknown per borehole point are solved together.
  4. Step doubling. One step of Δt is compared with two of Δt/2. The error is the largest difference of the ground temperatures and the mean borehole wall temperature. Above the tolerance: reject and halve. At or below a quarter of it: the next step doubles.
  5. Outputs. Each accepted step adds to every output interval it overlaps (the load interval: an hour here). Means are time means; the lowest and highest values come from the borehole sub-steps.
04 / The theory

The theory: what the borehole model adds

The borehole model has two parts (figure below). Inside the borehole, the column: fluid cells that pass the fluid on, and grout nodes that store heat. Near the wall, the ground mesh is too coarse for minutes and hours: an axis cell of about 0.7 m needs days before its node follows the true wall temperature. The radial sub-grid (30 cells from rb to two axis cells, held at the mesh temperature there) gives the short-time wall temperature. It adds no heat: the energy stays in the mesh. The steady well index of Simple FEM (a fixed resistance between the wall and the mesh node) remains selectable; it is 2–3 K off in the first hours.

Along the depthAcross one axial cell (≈ 1 m)inoutaxial cells≈ 1 m eachfluid movescell to cell(upwind)groutF1F2G1G2WRfluid, leg downfluid, leg upgrout + pipe wallgrout + pipe wallwallradial sub-grid: 30 cellsring Tfrom themesh (2 cells)Each node stores heat (fluid, grout + pipe wall); the links are the resistances of themultipole cross-section (TRCM). The wall heat goes into the ground mesh at the axis.
The column of a single-U borehole. The cross-section is the thermal resistance and capacity model (TRCM) of the dynamic borehole model; the pipe wall capacity sits at the grout node of its leg, so nothing is counted twice.
Mean fluid temperature, first 72 hours (hourly means)peakno load, pump onpeakpump off0612182430364248546066720246810time [h]°C2D design (Calculate)Full FEM, 0.01 KFull FEM, 0.003 K: lowest within the hourhour 7: 3D 5.422D 6.84hour 31: 3D 4.352D 6.42
The example. Peak start: the 3D fluid is warmer, because the store delivers heat first. Load stop and pump stop: the 3D fluid recovers more slowly. The 2D model does not see the pump stop at all (it uses the nominal flow throughout).
Heat rates, first 12 hours0246810120−1−2time [h]kWshaded: heat fromthe store, 1.30 kWhin hour 1after the stop theground refills the storeloop (heat pump)wallLowest Tf within each hour46810120246810time [h]°C2.58 (0.01 K)1.23 (0.003 K)load stop at 6 h0.01 K0.003 Khour mean
Left: the heat pump takes 2.5 kW at once; the wall follows over about two hours. Right: the hour means hardly depend on the tolerance; the lowest value right after the load stop does.
Why the tolerance matters only for short-time results. The step-doubling error measures the ground and the borehole wall temperatures. The fluid reacts faster than both: after the load stop it changes within minutes, while Tin is held constant over each ground step. With 0.01 K the steps after the stop stay long, and the lowest value of hour 7 is 2.58 °C; with 0.003 K the steps are shorter and it is 1.23 °C. The hour means differ by at most 0.05 K, at 1.5 times the run time (370 instead of 237 steps).
05 / How each result is derived

How each result is derived

All values for the example, Full FEM with the default tolerance unless noted.

ResultHow BoreTherm derives itExample
Heat capacity per metre Cbfluid ρcp·Afluid + pipe walls 1.8 MJ/(m³K)·π(ro² − ri²) per leg + grout Cg·Agrout (the grout annulus without the pipes)4.3 + 0.95 + 32.1 = 37.4 kJ/(m·K)
Fluid transit timefluid volume of the legs / flow10.78 cm² × 50 m / 0.4 l/s = 135 s (2.2 min)
Borehole time constant τmax(Cb · R*b, 2 transits); also the control interval of the wall-power control37.4 kJ/(m·K) × 0.109 m·K/W = 68 min
First borehole sub-stepshortest transit / 40 (0.5 … 300 s), for 3 transits after a change; then doubling3.4 s
Loop powerṁcp(Tin − Tout) = the load (control Automatic); Tin in closed form per ground step−2.50 kW in hours 1–6
Wall powerthe heat the borehole columns pass into the ground mesh, step mean−1.20 kW in hour 1, −2.33 kW in hour 6
Heat from the borehole store(loop power − wall power) × time−1.30 kWh in hour 1
Mean fluid temperature Tf(Tin + Tout)/2 per step, time mean over the interval; pump off: the volume mean of the standing fluidhour 1: (4.09 + 5.66)/2 = 4.88 °C
Tout − Tin−Ploop / (ṁcp)2500 W / 1590 W/K = 1.57 K
Lowest / highest within the intervalextremes of Tf at the borehole sub-steps of every accepted stephour 7: 2.58 °C (0.01 K), 1.23 °C (0.003 K)
Loop balanceΣ (Ploop − Pwall) Δt − ΔEb; zero to round-off1.5·10−5 J over the run
Design limits linelowest and highest Tf of the run against the limits, next to the 2D design3D 0.03 °C, 2D 0.04 °C; limit −3 °C
06 / Limits to keep in mind

Limits to keep in mind

  • Monthly design loads: transient boreholes change the design temperatures little (validation benchmark B9b: month-end values within 0.017 K of steady boreholes). Use them for hourly peaks, stops and inlet-temperature limits.
  • Step tolerance: 0.01 K for monthly design values; 0.003 K or tighter for hourly values and the first interval after a load or pump stop. The tolerance controls the ground and wall temperatures, not the fluid directly.
  • Control: keep Automatic (loop power) for heat-pump loads. Ground load at the borehole wall with transient boreholes holds Tin over at least the time constant (here 68 min) to stay stable; load intervals shorter than that are merged.
  • Pump off: the fluid stands; a load requested while the pump is off is not delivered and is reported. After the end of a time series shorter than a year there is no load and the pump is off.
  • Fluid: constant brine properties; freezing of the brine is not modelled (the freezing check reports the lowest inlet temperature, here −0.75 °C against −9.8 °C).
  • Cost: every load change needs a new run, and sizing, Calculate and the studies stay 2D. Dense fields (fewer than 3 axis cells between boreholes) get a warning; use the BTES model (borehole thermal energy storage) there.

Numbers calculated with BoreTherm. Theory: BoreTherm theory manual; validation: benchmark B3 (converged within 0.09 K of the converged dynamic model).