Dolphin (software)

Dolphin 6
Basic data
developer	Bauklimatik Dresden Software GmbH in cooperation with the Institute for Building Climatics of the TU Dresden
Current version	6.0.20
operating system	Microsoft Windows XP / 2000/7/8/10; MacOS 10.10 and newer; Linux: pre-build binaries for Ubuntu 16.04 and newer; Unix derivatives
programming language	C ++
category	Simulation software
License	commercial, academic and student license
German speaking	Yes
	https://www.bauklimatik-dresden.de/delphin

Delphin is the name of a simulation software for the coupled heat, moisture, air and salt transport in porous materials. The simulation program contains the implementation of a complex, extensive and in this respect unique model for the description of these processes.

Significance for teaching, research and economy

The numerical simulation of the coupled heat and material transport has been an essential tool for basic research and engineering practice for several decades. The complexity of the models and balance equations are the reason for a general lack of available simulation tools in this field. The DELPHIN software and the integrated model are characterized by the following aspects:

complex, thermodynamically sound, physical model with easily expandable mathematical formulation,
Program structure allows easy integration of new physical models and components,
Use of a very efficient numerical solution method (time integration with higher order),
implements an accurate description of salt transport coupled with moisture transport,
contains a user-friendly interface (pre- and post-processing) which allows adjustment and evaluation of (almost) all model parameters.

The program is used in teaching to convey the basics of building physics. In the economy it is used for expert reports and in materials research. In research it is used for basic research in a wide variety of fields of application . The source text is available to doctoral students for their own research and thus removes the hurdle of first writing their own simulation program for heat and moisture transport.

application areas

Moisture proofs for critical building structures (compared to simplified procedures such as the Glaser procedure or COND )
Assessment of structural details (thermal bridges, local temperature drop below the dew point)
Monument protection and restoration (salt transport and crystallization, e.g. desalination)
Optimization of building materials (e.g. calcium silicate board )
Spread of moisture in protective clothing
Assessment of air-flow structures (external walls in lightweight construction, rear-ventilated facades, roofs)
Pollutant emissions (VOC) from building materials, release of pollutants from building structures
Analysis of active components, for example heating registers (underfloor heating, geothermal collectors, concrete core activation, thermal analysis of PV panels ...)
Storage modeling, e.g. analysis, design and optimization of earth-ice storage facilities

In connection with the integrated FMI (Functional Mock-Up Interface), DELPHIN can be coupled with other simulation programs. So z. B. implement more complex control mechanisms in Modelica and thus control source and sink models in DELPHIN .

history

The history of the DELPHIN program begins in 1987. As part of the diploma thesis by the author John Grunewald, the predecessor program DIM 1 was developed between 1987 and 1991 at the Cottbus University of Civil Engineering (from 1991 Brandenburg Technical University Cottbus). The aim of the program development was to support the building physics education of the students, which besides research has remained the main task to this day. At the beginning, an analytical procedure for solving the heat conduction equation in semi-infinite space was implemented in the Turbo Pascal programming language. Later, a numerical method ( Crank-Nicolson method ) implemented in Standard C was added to the solution of the differential equation (s) , which enabled greater flexibility with regard to the problem to be solved. In other versions of DIM 1.x, it was already possible to calculate coupled heat and moisture transport processes and the model was expanded so that simulations under real climatic conditions were possible.

The development of the second version of this software, called DIM 2, began in 1991–1993. At the Federal Institute for Materials Research and Testing (BAM) Berlin and at the Materials Research and Testing Institute (MFPA) Weimar, significant extensions were made, such as: B. the step from 1D to 2D simulations, the use of the LSODI solver for a more efficient numerical solution, and the implementation of various boundary conditions, field conditions (source and sink models) and contact conditions at the material boundaries. Using the new solver increased the computing power up to 40 times.

The further development of DIM 2.x then took place again at the Technical University of Dresden from 1994. As part of the European joint project IEA-Annex 24 (HAMTIE), the Institute for Building Climatology participated in the solution of 'Common Exercises', which represented a great challenge but also the motivation for constant further development. In the version DIM 2.4 z. B. an air balance equation integrated to calculate air flow through a roof can. Working in international projects resulted in many contacts with colleagues in the field. Version DIM 2.6 was the first program version used internationally, installed at INSA Toulouse in France and at Helsinki University in Finland. DIM 2.7 was then used at various research institutes for very different tasks.

Extensions of the program in the years 1997–1998 resulted in the program version DIM 3, in which the CVODE solver was used (the successor of the LSODI solver). This resulted in a further increase in computing speed (up to 80 times faster). Programming in C enabled the program to be compiled for the PC (Windows) and the Apple Macintosh System 7-8. Based on the dissertation of the author John Grunewald (1997), the thermodynamic fundamentals of the program were significantly expanded and, for example, the capillary pressure model was implemented as an alternative to the humidity gradient model. The concept of the phase separation function (division of the moisture transport and vapor diffusion and capillary conduction) was introduced. Additional models were integrated which made it possible to work on specific projects, e.g. B. the dimensioning of a leak detection system in the base plate of the SLUB (Saxon State and University Library), which is based on the temperature effect of the groundwater. A simulation-compatible determination of the input parameters, primarily the material data, became the focus of interest (combined with the construction of a building materials laboratory under the direction of Rudolf Plagge, from which today's building physics research and testing laboratory at the Institute for Building Climatology has developed). A first pre-processing tool, called PreDim 1.0, was developed in 1998 for the Windows platform to facilitate the creation of complex input data. A user interface was added to the DIM 3 program by 2002 in order to increase the number of users and to make it easier for researchers and students to work with the software.

From 2001, work was carried out on a version with pre- and post-processing that could run completely under the Windows user interface. Heiko Fechner devoted himself to the post-processing, which allows the evaluation of the various calculation data according to building physics criteria. The result was the DELPHIN4 software package, which has since been widely used in basic research and industrial research. One example of this is the development of a robust, moisture-tolerant interior insulation system in the INSUMAT project, based on the calcium silicate panel and financed from the funds of the 5th Framework Program of the European Union (EU). Funds from the German Research Foundation (DFG) also went into the development of DELPHIN . As part of the priority program SPP 1122 (2001–2007) of the DFG, the focus of research interest was on damage to the building fabric by salts in a subproject of the TU Dresden, the University of Hamburg, the TU Hamburg-Harburg and the Bauhaus University Weimar. In this context, between 2003 and 2007, Andreas Nicolai wrote a completely new simulation program in C ++ called DELPHIN 5 based on the algorithms and concepts of the DELPHIN4 . By using object-oriented programming, the program was given a very modular structure, which greatly simplifies the expansion of the software with new physical models. This enables other researchers and doctoral students to develop and test new models based on a HAM transport model more quickly.

In the new version, meanwhile a joint development of a programming team under the direction of Andreas Nicolai and John Grunewald, the highly efficient solver CVODE from the SUNDIALS library has been used and the calculation scheme has been further optimized. This led to a further increase in speed (1D up to 4 times, with 2D up to 10 times faster than DELPHIN 4 ). In addition, a very extensive salt transport and phase change model was integrated. The pollutant balance equation (VOC) has also been added, as well as an improved and more efficient airflow model.

While the DELPHIN 5 is still being maintained and supported, the successor version DELPHIN 6 has been available since 2018 . This represents a significant change in technology and progress in several respects:

by using the Qt library, the platforms Windows , MacOS and Linux / Unix are now natively supported.
the calculation kernel is a completely newly developed, modular system of different physical modules for balance equations, currents, sources / sinks and corresponding sub-models
the numerical integration is taken over by an equally modular library (also developed at IBK / TU Dresden), which can be flexibly configured with regard to the following components:
- Integrator (CVODE, ImplicitEuler, Runge-Kutta, ADI, ExplictEuler)
- Linear equation system solver (direct: KLU-Sparse, Band, Block-Tridiag, Dense: Iterative: GMRES, BiCGStab)
- Preconditioner for iterative solvers (ILU, tape, block diagonal, ADI)
Completely parallelized computing kernel (OpenMP-based), which enables significant increases in performance, especially in connection with iterative equation system solvers
Support for 3D geometries

The development of the new computing kernel began conceptually in 2011 with a concept presentation at the NSB conference [3]. The prototype implementation of the computing kernel and the underlying integrator library had been completed by 2015. The new program interface was developed in 2016-2018, almost entirely under Linux. In 2019, many of the above-mentioned extensions were integrated.

Due to the large number of users, the distribution and commercial support of the software has been taken over by an external company since 2019. With regard to the further development of the software, testing and tests, there is still a close cooperation with the Institute for Building Climatology at TU Dresden.

Physical basics

A comprehensive model for the coupled heat and material transport in capillary-porous building materials is implemented in the simulation program. It is based on thermodynamic principles and has the following basic principles and properties:

macroscopic model of capillary-porous materials,

describes moisture transport through vapor diffusion and capillary conduction,

the driving potential for vapor diffusion is the vapor pressure gradient,
the driving potential for capillary conduction is the gradient of the capillary pressure,

describes moisture storage through material functions: moisture storage function and sorption isotherm ,

describes heat conduction and consideration of enthalpy transport.

Balance equations

The balance equations used in the model can be expressed by the following equation:

{\ displaystyle {\ frac {\ partial \ rho ^ {<E>}} {\ partial t}} = - {\ frac {\ partial} {\ partial x_ {k}}} \ left (j_ {k, conv } ^ {<E>} + j_ {k, diff} ^ {<E>} \ right) + \ sigma ^ {<E>}}

where is the vector of the conserved quantities. ${\ displaystyle E}$

{\ displaystyle E = \ left \ {U, m_ {w + v}, m_ {VOC}, m_ {s, i}, m_ {p, j} \ right \}}

.

The densities of the primary conservation quantities energy , moisture mass and other conservation quantities taken into account if required ( - pollutants, - dissolved substances / ions, - crystalline substances / salts) are balanced. Transport terms (convective and diffusive) are taken into account by and source and sink terms by . This general formulation allows the balance equation system to be easily expanded to include other variables to be balanced. ${\ displaystyle U}$ ${\ displaystyle m_ {w + v}}$ ${\ displaystyle VOC}$ ${\ displaystyle m_ {s, i}}$ ${\ displaystyle m_ {p, j}}$ ${\ displaystyle j}$ ${\ displaystyle \ sigma}$

Modeling of the moisture transport

In the transport model, a distinction is made between moisture transport in the gas phase (water vapor diffusion) and in the liquid phase (capillary line).

Water vapor diffusion

The driving potential for vapor diffusion is initially the gradient of the chemical potential, which under normal conditions can be replaced by the gradient of the vapor concentration and can finally be mathematically converted into the gradient of the water vapor partial pressure. The vapor diffusion is thus described by the following equation.

{\ displaystyle j_ {k, diff} ^ {m_ {v}} = - {\ frac {D_ {v}} {R_ {v} T}} {\ frac {\ partial p_ {v}} {\ partial x_ {k}}}}

In the notation used here, Einstein's rule of summation is used, the index stands for the directions of transport (in the Cartesian coordinate system ). denotes a mass flow density, and the subscript denotes the component water vapor. is the vapor pressure, is the gas constant for water vapor and is the temperature. The diffusion coefficient depends on the moisture content of the material. ${\ displaystyle k}$ ${\ displaystyle k = x, y, z}$ ${\ displaystyle j ^ {m}}$ ${\ displaystyle v}$ ${\ displaystyle p_ {v}}$ ${\ displaystyle R_ {v} = R / M_ {w}}$ ${\ displaystyle T}$ ${\ displaystyle D_ {v}}$

The use of the water vapor partial pressure has the advantage that effects such as B. the lowering of the vapor pressure via salt solutions need not be considered separately in the vapor diffusion model.

Capillary liquid water transport

The moisture transport in the liquid phase can be described by various models, which are also supported in the simulation program:

Moisture gradient model or diffusivity model
Capillary pressure model or Darcy flow model

Humidity gradient model

The simplest model is the mechanistic model, which describes capillary moisture transport as a function of a moisture content gradient : Moisture moves from areas of high humidity to areas with lower humidity and thus leads to a redistribution of humidity. This approach is very simple and can be used in a very wide variety of applications. However, there are a number of limitations:

The model is only applicable under isobaric conditions , i.e. H. as soon as the air pressure in the pore space reaches significant levels (e.g. in pressure plate tests or absorption tests with sealed surfaces) the moisture content gradient model delivers incorrect results.

The model cannot describe liquid water transport due to hydrostatic pressures in the saturated material. This is relevant for structures that are exposed to pressing groundwater. The moisture content gradient model can only be used for the unsaturated area .

Capillary pressure model

Alternatively, the liquid water transport can be described by a Darcy flow model. The pore system is imagined as a network of cylindrical pores through which the liquid phase moves as a result of a pressure gradient. In every cylindrical pore, Hagen-Poiseuille's law can be used to describe the mass flow due to pressure gradients. The following transport equation describes the liquid water transport due to a gradient of the pressure in the liquid phase . ${\ displaystyle p _ {\ ell}}$

{\ displaystyle j_ {k} ^ {m _ {\ ell}} = - K _ {\ ell} {\ frac {\ partial p _ {\ ell}} {\ partial x_ {k}}}}

The transport coefficient depends on the moisture content of the material, which defines the number and size of the water-filled pores in the model. In the unsaturated area, the liquid water pressure is approximately equal to the capillary pressure (the difference consists in additional hydrostatic pressures and the pressure of the gas phase), which is why this model is also known as the capillary pressure model . ${\ displaystyle K _ {\ ell}}$

The advantage of this model is that all processes that can be described with the moisture content gradient model can also be mapped. However, the effects of hydrostatic pressures (moisture transport in the fully saturated material) and gas pressures are also taken into account in the model. The transport of salt solutions can only be described with this model by adjusting the transport coefficient, while the driving potential remains independent of salt.

The capillary pressure model is therefore the standard model in the Delphin simulation program.

Alternative simulation programs

Simulation program WUFI of the Fraunhofer Institute for Building Physics IBP in Holzkirchen for the coupled heat and moisture transport
COND program from the Institute for Building Climatology at TU Dresden, especially for hygrothermal assessment and proof of moisture protection for interior insulation systems with condensation

Publications

[1] Grunewald, J. (1997), Diffusive and Convective Material and Energy Transport, Dissertation, Technical University of Dresden
[2] Nicolai, A. (2007), Modeling and Numerical Simulation of Salt Transport and Phase Transitions in Porous Building Materials, Dissertation, Syracuse University
[3] Nicolai A. and Grunewald J., Towards a Semi-Generic Simulation Framework for Mass and Energy Transport in Porous Materials , Proceedings of the Nordic Symposium for Building Physics, 2011

A list of other publications on the subject can be found on the website [1] .