HomeFeaturesExamplesTutorialsHelpTheoryVerificationPricingContact
Theory manual

15. Short-time borehole model, TRCM parameters and load aggregation

The steady model of Chapters 5 and 8 treats the inside of the borehole as a resistance: Tf=Tb+qRb* at every time. The fluid, the pipes and the grout store no heat, so a load step reaches the fluid temperature at once and in full. The borehole model is an option:

Table 21. Borehole models.
ModelWhat it addsStatus
Steady resistance (default)nothing; the model of the classic monthly methodthe classical results
Short-time capacitythe heat capacity of fluid, pipes and grout at short times, as a combined response gcavailable
Dynamictransient fluid and grout along the depth, transit delay, flow schedule (TRCM network)a separate transient model, not treated here

The classic monthly method has neither of the new models; the classic-method preset resets the model to the steady resistance.

15.1 Equivalent single pipe

The n legs of a U-tube borehole (n = 2, 4, 6) are replaced by one concentric pipe (Xu & Spitler 2006) that keeps the fluid volume and the pipe-wall volume (ri,eq=nri, ro,eq=nro), the film and wall resistances of the n pipes in parallel, and the effective borehole resistance: the grout annulus ro,eq…rb gets the conductivity that makes

Rfilm,eq+Rwall,eq+Rgrout,eq=Rb*,Rfilm,eq=Rp−Rwalln,Rwall,eq=1n[ln(ro/ri)2πλp+Rcontact]
(15.1)

Xu & Spitler keep Rb; keeping Rb* makes the long-time end of the model equal to the steady model, including the short-circuit term, and lets the correction start at 0. Coaxial: the outer pipe is already concentric, and the whole fluid is one well-mixed node (the inner channel's capacity is slightly overstated). Sections enter as length-weighted equivalents (mean radius, sectioned Rb*, mean grout capacity). Heat capacities: fluid ρcp at the property temperature, pipe walls 1.8 MJ/(m³ K), grout as given in the borehole design (default 2.0 MJ/(m³ K)), ground C of the ground reference.

15.2 Radial model and the combined response

A one-dimensional radial finite-volume model (a well-mixed fluid node, 3 pipe-wall cells, 12 grout cells and 90 log-spaced ground cells out to max(10 m, 4αtblend, 20rb)) receives a unit step heat rate in the fluid node and is integrated with variable-step BDF2 (about 320 steps to 100 h). It is run twice, with and without the capacities of fluid, pipes and grout; the difference is the capacity correction

δg(t)=2πλ[Tf(t; capacities)−Tf(t; none)]w(t)
(15.2)

with a blend weight w=1 for t≤tblend/4 and a smoothstep in lnt to 0 at tblend = 100 h. δg starts near −2πλRb* (the fluid has not warmed yet) and tends to 0; at 100 h the unblended capacity effect is below 0.3 % of the response. The combined response

gc(t)=g(t)+2πλRb*+δg(t),Tf(t)−Tm=qgc(t)2πλ
(15.3)

is tabulated at 40 nodes per decade from 1 min to tblend and interpolated in lnt; below 1 min it is linear in t from 0, from tblend on it is exactly g+2πλRb*. In the first minutes the radial wall warms faster than the line-source g, so the table is kept ≥0, non-decreasing and ≤g+2πλRb*. With all capacities 0 the correction is exactly 0. Two radial runs take under 1 ms and are repeated at every simulation and sizing iteration (Rb* depends on H).

15.3 Use in the simulation

Everything the steady model computes stays; the capacity correction is added as a short convolution:

Tf(n)=Tf,steady(n)+12πλ∑j:(j+1)h<tblendΔqn−j[gc−g−2πλRb*]((j+1)h)
(15.4)
Tf,peak=Tf+(qpeak−qn)gp(tp)+2πλRb*+(gc−g−2πλRb*)(tp)2πλ
(15.5)

with h = 1 h and fewer than 100 terms, the same for the FFT and the load-aggregation method, so the steady part is untouched. The monthly base temperatures are unchanged (the correction is 0 after 100 h); the peak pulses of (8.4) use gc(tp), also with the classic asymptote (8.5).

Table 22. Effect of the short-time model on the examples (simulated and sized with the steady and with the short-time borehole model).
ExampleTf,min steady → short-time [°C]Tf,max steady → short-time [°C]max |ΔTf| hourlySized H [m]
3 × 3 field, hourly loads−0.130 → −0.01022.416 → 22.2880.73 K–
Sizing case, hourly−4.342 → −4.17021.883 → 21.7421.00 K117.26 → 116.02
Office example (4 × 4)−2.413 → −2.24416.026 → 15.925–143.72 → 141.89
Sizing case, monthly−5.098 → −4.94322.524 → 22.397–122.70 → 121.59
Linden worked example−15.575 → −15.29921.438 → 21.001––
Ahmadfard & Bernier (2019), case 3, 6 h peaks−2.084 → −1.80329.072 → 28.691–110.93 → 108.34

Short peaks of a few hours are 0.1–0.45 K less extreme and sizing is 1–2.5 % shorter where they bind; hourly Tf moves by up to 1 K in the first hours after a load change (Figure 13). A single borehole with 10 kW monthly peaks of 2 h needs 219.0 m with the steady model and 210.4 m with the short-time model.

Hourly mean fluid temperature of the 3 × 3 hourly example with the steady and the short-time borehole model: (a) four days around the largest difference; (b) the difference over the first two weeks.
Figure 13. Hourly mean fluid temperature of the 3 × 3 hourly example with the steady and the short-time borehole model: (a) four days around the largest difference; (b) the difference over the first two weeks.

15.4 TRCM parameters from the multipole matrix

Transient borehole models (a dynamic model along the depth, a finite-element borehole element) use the thermal resistance and capacity model (TRCM) of Bauer et al. (2011): every leg i has a fluid node fi and a grout node gi; fi is connected to gi through Rfg,i, gi to the borehole wall through Rgb,i, and the grout nodes to each other through Rgg,ij, with

Rfg,i=Rp,i+xiRg,i,Rgb,i=(1−xi)Rg,i
(15.6)

Bauer et al. give Rg and the grout–grout resistance from two-pipe line-source formulas. BoreTherm derives them from the multipole matrix of Chapter 5 (with sleeves and casing) so that the steady network reproduces it exactly. With 𝐊=𝐑−1, 𝐀=diag(1/Rfg,i), 𝐁=diag(1/Rgb,i) and the grout Laplacian 𝐋, eliminating the grout nodes gives 𝐊=𝐀−𝐀(𝐀+𝐁+𝐋)−1𝐀, hence

𝐌=𝐀(𝐀−𝐊)−1𝐀,𝐁=diag(𝐌1)−𝐀,𝐋=𝐌−𝐀−𝐁
(15.7)
Rg,i=1∑jKij−Rp,i
(15.8)

For a single U-tube this is Bauer's Rg=2Rb−Rp with his grout–grout resistance. The grout-node position is Bauer's

x=ln(Db2+ndo2/(2ndo))ln(Db/(ndo))
(15.9)

(Db the borehole and do the pipe diameter; n = 2 and 4 are Bauer's single- and double-U formulas, triple-U uses 6). Second law: 𝐀−𝐊 and 𝐁+𝐋 must be positive definite and Rgb>0; otherwise x←2x/3 (Diersch et al. 2011). The steady TRCM discretised along the depth at a uniform wall temperature reproduces Hellström's leg temperatures and (5.22).

15.5 Incremental load aggregation

The Claesson–Javed cells of (8.10)–(8.11) are also available one step at a time for time-stepping models and co-simulation, with four operations: commit a step with its load; the value, the response at the end of the last committed step (the batch computation is a loop over commit and value and bit-identical to the incremental result); the history, the response at the end of the next step to the committed loads only; and the partial step with q¯ and τ, the response a fraction τ into the next step with the running mean load q¯ of its elapsed part:

S(τ)=∑m≥1q¯m′[g(τm−1+τ)−g(τm−1−1+τ)]+q¯g(τ),0<τ≤1
(15.10)

with q¯′ the cells as the next commit would shift them; the partial step with load q and τ=1 equals the value after committing q exactly. Every commit conserves the energy of the loads, and the state can be saved and restored (rollback of a co-simulation step).