A non-history dependent temporal superposition algorithm for the finite line source solution
Alberto Lazzarotto, Marc Basquens, Massimo Cimmino · 2025
Read A non-history dependent temporal superposition algorithm for the finite line source solution by Alberto Lazzarotto, Marc Basquens, Massimo Cimmino (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-16027.txt.
Introduction
Alberto Lazzarotto, Marc Basquens, and Massimo Cimmino speed up ground heat simulations with a memory-light time method, in work from twenty twenty-five. Fields of borehole heat exchangers need hour-by-hour runs over decades. Standard models stack basic heat solutions in space and time to build answers. That stacking is robust but grows costly very fast with size. Large fields can become too slow to run at all.
Their method reworks time stacking so past steps need not be stored in full. It applies to both point and line heat source solutions. Cost grows in a straight line with the count of time steps. Accuracy can reach near machine precision by refining the grids used. Integrals inside the scheme carry the long memory in compact form. The result opens the door to bigger, longer field runs.
Catalog record
| CREATOR | Alberto Lazzarotto, Marc Basquens, Massimo Cimmino |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 53 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-16027
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A non-history dependent temporal superposition algorithm for the finite line source solution Alberto Lazzarotto, Marc Basquens, Massimo Cimmino Abstract Simulations of the operation of fields of borehole heat exchangers involve a wide spectrum of time scales, and hourly simulations for decades are required for the evaluation of the heat transfer in the subsurface due to these systems. Most current models rely on time and space superposition of fundamental analytical solutions of the heat equation to build the solution for complex borehole fields configurations and loading conditions. These procedures are robust and accurate but do not have favorable scaling properties, and the problems can become quickly computationally intractable as the size increases. In this context, acceleration algorithms for temporal superposition are key to overcome this limitation. This paper presents developments on the so-called “non-history dependent” acceleration scheme and its application to the point and line source solutions, which are commonly used as building blocks in borehole field simulations. The results obtained show promising properties as the computational complexity of the proposed algorithm is linear in the number of time steps, and near double precision accuracy can be achieved by refining the discretizations used to compute the integrals arising from the scheme. 1 Introduction Detailed thermal modeling of borehole fields is essential for the accurate design and operation of geothermal systems, ensuring that the expected performance is maintained throughout the lifespan of a shallow geothermal installation. Current modeling techniques primarily use “response factors”, the method of images, and superposition of the effects as foundational tools to build solutions for the heat transfer problem in the subsurface [ 1 , 2 , 3 ] . Response factors represent how ground temperature changes over time due to a step heat source in an infinite volume. The method of images pairs a set of heat sources with a corresponding set of fictitious heat sources called “images” mutually symmetric with respect to a given plane, and enables representing heat transfer in a semi-infinite solid with boundary conditions that are relevant for geothermal applications. Finally, superposition allows to account for variations in heat distribution over time and space. A naïve implementation of time superposition results in a complexity of \mathcal{O}(N_{t}^{2}) , where N_{t} is the number of uniformly spaced time steps in the simulations. This complexity is not suitable for conducting 20-year hourly simulations, which are typically required for designing borehole systems. Several strategies have been proposed in the literature to reduce the computational complexity of the time superposition problem and accelerate simulations. Load aggregation methods [ 4 , 5 , 6 , 7 ] that reduce the complexity by aggregating the effect of past loads in the time superposition evaluation were developed. Later, in [ 8 ] , the use of the Fast Fourier Transform (FFT) algorithm was proposed to compute the discrete convolution. Finally, the so-called “non-history dependent” method was ideated in [ 9 ] . This method is the least studied of the three techniques, but has theoretically good properties in terms of complexity and is a marching scheme, suitable for integration in borehole network simulation [ 10 , 11 ] and building simulations. In this article, we focus on this method. The original “non-history dependent” method studied the particular case of the infinite cylindrical source solution evaluated at the borehole radius and reduced the computational complexity of the time superposition process by leveraging its structure [ 9 ] . An extension of this scheme for generic responses by means of the Inverse Laplace Transform was put forward by Lamarche in [ 12 ] . However, in that work, the focus was put on numerical inversion of the Laplace Transform, which is in general an ill-conditioned problem and, as such, can be unstable and dependent on the tuning of the numerical algorithm parameters. In the present paper we explore the “non-history dependent” scheme using the analytical Inverse Laplace transform. Although this is hard to do in general, it can be done in the particular case of the point source solution [ 13 ] . While the point source solution cannot be applied directly for geothermal calculations, it is the fundamental building block to build other relevant solutions for heat transfer in solids. In particular, this allows us to extend the “non-history dependent” method to responses to a finite line source which is the critical solution for practical applications in borehole field simulations [ 3 , 14 , 15 ] . The paper is structured as follows. In Section 2 , we review the original “non-history dependent” scheme in [ 9 ] and the generalization explained in [ 12 ] . Then, we apply it to derive a “non-history dependent” approach for the temperature field generated by a point source, and use the result to compute the temperature generated by a finite line source. In Section 3 , we develop a numerical scheme to compute the integrals that appear in the expressions found in Section 2 . In Section 4 , we analyze the precision of the algorithm and study its computational complexity. 2 Methodology Lamarche and Beauchamp [ 9 ] derived a scheme that enables to evaluate the response to an infinite cylindrical source with time varying intensity, and advance a step forward in time using only information regarding the current time step. We provide a short outline of that method here. The quantity of interest is the temperature at time t caused by a heat extraction load q^{\prime}(t) during the time interval [0,t] . According to Duhamel’s theorem, this is given by the convolution \displaystyle T(t)-T_{0}=\frac{1}{k_{g}}\int_{0}^{t}\,q^{\prime}(\tau)\frac{d{h}}{dt}(t-\tau)\ \mathrm{d}\tau\ , where k_{g} is the thermal conductivity, q^{\prime}(t) is the load at time t , and \frac{d{h}}{dt} the impulse response of the particular system on which the load is applied. It should be remarked that the units of q^{\prime}(t) depend on what the units of \frac{dh}{dt} are. In the paper we will deal with linear and point sources, hence q^{\prime}(t) will be expressed in W/m and W, respectively. Note that T , q^{\prime} and \frac{d{h}}{dt} depend on spatial coordinates, however for the sake of notational simplicity we first address the case of a generic \frac{d{h}}{dt} dropping the dependence on space, and then recover it when addressing specific impulse response functions, e.g. point source to point target. Following [ 9 ] , we will use the non-dimensional time \tilde{t}=\frac{\alpha}{r_{b}^{2}}t and the non-dimensional distance \tilde{r}=\frac{r}{r_{b}} to rewrite this expression. Here, r_{b} is the borehole radius and \alpha the thermal diffusivity. We reparametrize T with t(\tilde{t}) and change variable \tilde{\tau}=\frac{\alpha}{r_{b}^{2}}\tau to obtain the expression \displaystyle T(\tilde{t})-T_{0}=\frac{1}{k_{g}}\frac{r_{b}^{2}}{\alpha}\int_{0}^{\tilde{t}}\,q^{\prime}\left(\tau(\tilde{\tau})\right)\frac{d{h}}{dt}\left(t(\tilde{t})-\tau(\tilde{\tau})\right)\ \mathrm{d}\tilde{\tau}\ , where now t and \tau are to be understood as functions of \tilde{t} and \tilde{\tau} , respectively. It is also useful to define \frac{d{\tilde{h}}}{d\tilde{t}}(\tilde{t}) as the reparametrized impulse response and \tilde{q}^{\prime}(\tilde{t}) as the reparametrized load with t(\tilde{t}) , so we can write \displaystyle T(\tilde{t})-T_{0}=\frac{1}{k_{g}}\frac{r_{b}^{2}}{\alpha}\int_{0}^{\tilde{t}}\,\tilde{q}^{\prime}(\tilde{\tau})\frac{d{\tilde{h}}}{d\tilde{t}}(\tilde{t}-\tilde{\tau})\ \mathrm{d}\tilde{\tau}\ . (1) The key insight of the scheme was to realize that the impulse response of the infinite cylindrical source \frac{d\tilde{h}_{\text{ICS}}}{d\tilde{t}} [ 12 ] , due to its particular form, could be expressed in the form \displaystyle\frac{d\tilde{h}_{\text{ICS}}}{d\tilde{t}}(\tilde{t})=\int_{0}^{\infty}e^{-\zeta^{2}\tilde{t}}v(\zeta)\ \mathrm{d}\zeta\ , (2) for a certain function v(\zeta) . Substituting Equation ( 2 ) into Equation ( 1 ) and changing the order of integration between the variables \tilde{\tau}