Advanced Building Simulation

Ali Malkawi, Godfried Augenbroe, Ali Malkawi, Godfried Augenbroe

• Publisher: Routledge
• Year: 2004
• ISBN:
  9781134357529
• Edition: 1
• Topic: Architecture
• Subtopic: Architecture General

This book introduces recent advances in building simulation and outlines its historic development. Two important topics are described: uncertainty in simulation and coupled simulations, which are both closely linked to attempts to improve control and accuracy. This is followed by coverage of wind simulations and predictions, and then by an introduction to current systems and phenomenological modelling. Written by leading experts in the field both in the US and Europe, Advanced Building Simulation is an excellent graduate-level student textbook as well as a practical guide for architects, engineers and other construction professionals.

Chapter 1
Trends in building simulation

Godfried Augenbroe

1.1 Introduction

The total spectrum of “building simulation” is very wide as it spans energy and mass flow, structural durability, aging, egress and even construction site simulation. This chapter, and indeed the book, will deal with building performance simulation in the narrower sense, that is, limited to the field of physical transport processes. This area of building performance simulation has its origin in early studies of energy and mass flow processes in the built environment. Meanwhile, the role of simulation tools in the design and engineering of buildings has been firmly established. The early groundwork was done in the 1960s and 1970s, mainly in the energy performance field followed by an expansion into other fields such as lighting, Heating Ventilation and Air-Conditioning (HVAC), air flow, and others. More recent additions relate to combined moisture and heat transfer, acoustics, control systems, and various combinations with urban and micro climate simulations. As tools matured, their proliferation into the consultant’s offices across the world accelerated. A new set of challenges presents itself for the next decade. They relate to achieving an increased level of quality control and attaining broad integration of simulation expertise and tools in all stages of the building process.

Simulation is credited with speeding up the design process, increasing efficiency, and enabling the comparison of a broader range of design variants. Simulation provides a better understanding of the consequences of design decisions, which increases the effectiveness of the engineering design process as a whole. But the relevance of simulation in the design process is not always recognized by design teams, and if recognized, simulation tools cannot always deliver effective answers. This is particularly true in the early design stages as many early research efforts to embed “simplified” of “designer-friendly” simulation instruments in design environments have not accomplished their objectives. One of the reasons is the fact that the “designer”and the “design process” are moving targets. The Internet has played an important role in this. The ubiquitous and “instant” accessibility of domain experts and their specialized analysis tools through the Internet has de-emphasized the need to import “designer-friendly” tools into the nucleus of the design team. Instead of migrating tools to the center of the team, the opposite migration may now become the dominant trend, that is, delegating a growing number of analysis tasks to (remote) domain experts. The latter trend recognizes that the irreplaceable knowledge of domain experts and their advanced tool sets is very hard to be matched by designer-friendly variants. With this recognition, sustaining complete, coherent and expressive communications between remote simulation experts and other design team members has surfaced as the real challenge. After an overview of the maturation of the building simulation toolset in Section 1.2, we will discuss the changing team context of simulation in Section 1.3.

Simulation is also becoming increasingly relevant in other stages of a project, that is, after the design is completed. Main application opportunities for simulation are expected during the commissioning and operational facility management phases. Meanwhile, the “appearance” of simulation is changing constantly, not in the least as a result of the Internet revolution. This is exemplified by new forms of ubiquitous, remote, collaborative and pervasive simulation, enabling the discipline to become a daily instrument in the design and operation of buildings. The traditional consultancy-driven role of simulation in design analysis is also about to change. Design analysis does not exist in isolation. The whole analysis process, from initial design analysis request to model preparation, simulation deployment and interpretation needs to be managed in the context of a pending design, commissioning or maintenance decision. This entails that associations between decisions over the service life of a building and the deployment of building simulation must be managed and enforced explicitly across all members of the design, engineering and facility management team. A new category of web-enabled groupware is emerging for that purpose. This development may have a big impact on the simulation profession once the opportunities to embed simulation facilities in this type of groupware are fully recognized. Section 1.4 will look at the new roles that building simulation could assume over the next decade in these settings. It will also look at the developments from the perspective of performance based design, where simulation is indispensable to quantify the new “metrics” of design quality. Finally in Section 1.5, emerging research topics ranging from new forms of calibration and mold simulation to processes with embedded user behavior are briefly discussed.

1.2 The maturation of the building simulation toolset

Simulation involves the “creation” of behavioral models of a building for a given stage of its development. The development stage can range from “as-designed” to “as-built” to “as-operated”. The distinction is important as correctness, depth, completeness and certainty of the available building information varies over different life cycle stages. The actual simulation involves executing a model that is deduced form the available information on a computer. The purpose of the simulation is to generate observable output states for analysis, and their mapping to suitable quantifications of “performance indicators”, for example, by suitable post-processing of the outputs of the simulation runs. The post-processing typically involves some type of time and space aggregation, possibly augmented by a sensitivity or uncertainty analysis.

Models are developed by reducing real world physical entities and phenomena to an idealized form at some level of abstraction. From this abstraction, a mathematical model is constructed by applying physical conservation laws. A classic overview of modeling tasks in the building physics domain can be found in Clarke (2001). Comparing simulation to the design and execution of a virtual experiment (Figure 1.1) is not merely an academic thought experiment. The distinction between computer simulation and different means to interact with the behavior of a building can become rather blurred indeed. Interesting new interaction paradigms with simulation have emerged through combinations of real and virtual environments. This subject will resurface in later chapter in this book.

Figure 1.1 Simulation viewed as a (virtual) experiment.

The modeling and simulation of complex systems requires the development of a hierarchy of models, or a multimodel, which represent the real system at differing levels of abstraction (Fishwick and Zeigler 1992). The selection of a particular modeling approach is based on a number of (possibly conflicting) criteria, including the level of detail needed, the objective of the simulation, available knowledge resources, etc. The earliest attempts to apply computer applications to the simulation of building behavior (“calculation” is the proper word for these early tries) date from the late 1960s. At that time “building simulation” codes dealt with heat flow simulation using semi-numerical approaches such as the heat transfer factor and electric network approach (both now virtually extinct). Continued maturation and functional extensions of software applications occurred through the 1970s. The resulting new generation of tools started applying approximation techniques to the partial differential equations directly, using finite difference and finite element methods (Augenbroe 1986) that had gained popularity in other engineering domains. The resulting system is a set of differential algebraic equations (DAE) derived through space-averaged treatment of the laws of thermodynamics as shown in Figure 1.2

Since these early days, the finite element method and special hybrid variants such as finite volume methods have gained a lot of ground and a dedicated body of knowledge has come into existence for these numerical approximation techniques. Due to inertia effects, the computational kernels of most of the leading computer codes for energy simulation have not profited much from these advancements.

In the late 1970s, and continued through the 1980s, substantial programming and experimental testing efforts were invested to expand the building simulation codes into versatile, validated and user-friendly tools. Consolidation set in soon as only a handful tools were able to guarantee an adequate level of maintenance, updation and addition of desired features to a growing user base. As major software vendors continued to show little interest in the building simulation area, the developer community started to combine forces in order to stop duplication of efforts. The launch of EnergyPlus (Crawley et al. 1999) is another more recent indication of this. Until the mid-1990s the landscape of tools was dominated by the large simulation codes that were generated with research funding, for example, DOE-2, ESP-r ad TRNSYS. As new simulation domains came along, these tools tried to expand into these domains and outgrow their traditional energy origin. However, since the late 1990s, domains other than energy are increasingly covered by specialized tools, for example, in air flow simulation, moisture and mold simulation, and others. Specialized tools do generally a better job in these specialized fields. Another new trend was the entry of commercial packages, some of which were offered as shells around the existing computation kernels mentioned earlier, and some of which were new offerings. These and all major tools are listed on (DOE 2003).

Figure 1.2 Standard approach to simulation

As to computational elegance, it cannot escape closer inspection that computational kernels of the energy simulation tools (still the largest and most pronounced category of building simulation tools) date back more than 15 years. Rather primitive computing principles have remained untouched as the bulk of the development resources have gone into functional extensions, user interfaces and coverage of new transport phenomena. But thanks to the fact that Moore’s law (in 1965, Gordon Moore promised that silicon device densities would double every 18 months) has held over the last 25 years, current building energy simulation codes run efficiently on the latest generation of Personal Computers.

The landscape of simulation tools for the consulting building performance engineer is currently quite diverse, as a result of the hundreds of man-years that have been invested. A skilled guild of tool users has emerged through proper training and education, whereas the validation of tools has made considerable progress. As a result, the design profession appears to have acquired enough confidence in the accuracy of the tools to call on their expert use whenever needed. In spite of the growing specialization and sophistication of tools, many challenges still remain to be met though before the building performance discipline reaches the level of maturity that its vital and expanding role in design decisions demands. Many of these challenges have been on the wish list of desired tool characteristics for many years. They relate to improvements in learning curve, GUI, documentation, output presentation, animation, interactivity, modularity, extensibility, error diagnostics, usability for “intermittent” users, and others. The user community at large has also begun to identify a number of additional challenges. They relate to the value that the tool offers to the design process as a whole. This value is determined mostly by application characteristics. Among them, the following are worth mentioning: (1) the tool’s capability to inspect and explicitly “validate” the application assumptions in a particular problem case; (2) the tool’s capability to perform sensitivity, uncertainty and risk analyses; (3) methods to assert preconditions (on the input data) for correct tool application; (4) support of incremental simulation cycles; and (5) standard post-processing of output data to generate performance indicators quantified in their pre-defined and possibly standardized measures. Some of these challenges will be revisited later in this section.

One “development issue” not mentioned above deserves special attention. It concerns the modularity and extensibility of (large) computer codes. In the late 1980s many came to realize that the lack of modularity in current “monolithic” programs would make them increasingly hard to maintain and expand in the future. Object-oriented programming (OOP) languages such as C

were regarded as the solution and “all it would take” was to regenerate existing codes in an OOP language. The significant advantage of this approach is the encapsulation and inheritance concepts supported by object-oriented languages. Projects attempting to use the object-oriented principles to regenerate existing programs and add new functionality were started. EKS (Tang and Clarke 1993), SPARK (Sowell and Haves 1999), and IDA (Sahlin 1996a) are the best-known efforts of that period. They started a new wave of software applications that were intended to be modular and reusable. Ten years later, only IDA (Björsell et al 1999) has evolved to an industry strength application, in part due to its pragmatic approach to the object-oriented paradigm. An important outcome of these attempts are the lessons that have been learned from them, for example, that (1) reverse engineering of existing codes is hard and time consuming (hardly a surprise), (2) it is very difficult to realize the promises of OOP in real life on an object as complex as a building, and (3) the class hierarchy in an OOP application is not a “one fit all” outcome but embodies only a particular semantic view of a building. This view necessarily reflects many assumptions with respect to the building’s composition structure and behavior classification. This particular developer’s view may not be suitable or even acceptable to other developers, thus making the original objectives of code reusability a speculative issue. Another important lesson that was learned is that building an object-oriented simulation kernel consumes exorbitant efforts and should not be attempted as part of a domain-specific effort. Instead, generic simulation platforms, underway in efforts such as MODELICA (Elmqvist et al. 1999) should be adopted. An important step in the development is the creation of a building performance class hierarchy that is identical to, or can easily be mapped to a widely accepted “external” building model. The model proposed by the International Alliance for Interoperability (IAI) seems the best candidate at this time to adopt for this purpose. This is only practical however, if a “semantic nearness” between the object-oriented class hierarchy and the IAI model can be achieved. Whether the similarity in the models would also guarantee the seamless transition between design information and building performance analysis tools (a belief held by many IAI advocates, see for instance Bazjanac and Crawley (1999)) is a matter that needs more study. It can be argued that such seamless transition can in general not be automated as every translation between design and analysis requires intervention of human judgment and expert modeling skills, strongly influenced by design context and analysis purpose.

Figure 1.3 Trends in technical building performance simulation tools.

In an attempt to put the observations of this section in a broad historic perspective, Figure 1.3 identifies building simulation trends between 1970 and 2010.

The foundation for building simulation as a distinct class of software applications came with the advent of first-principles-based formulation of transport phenomena in buildings, leading to DAE formulations that were amenable to standard computational methods. The next step was towards broader coverage of other aspects of technical building behavior. This movement towards function complete tools led to large software applications that are being used today by a growing user base, albeit that this user base is still composed of a relatively small expert guild. The next two major movements started in parallel in the 1990s and had similar goals in mind on different levels of granularity. Interoperability targets data sharing among (legacy) applications whereas code sharing targets reuse and inter-application exchange of program modules. Whereas the first tries to remove inefficiencies in data exchange, the latter is aiming for functionally transparent kits of parts to support the rapid building (or rather configuration) of simulation models and their rapid deployment.

Design integration adds an additional set of process coordination issues to its predecessor movements. Ongoing trials in this category approach different slices of a very complex picture. It is as yet unclear what approach may eventually gain acceptance as the best framework for integration.

The two most recent trends in Figure 1.3 have in common that they are Internet driven. The Web enables a new breed of simulation services that is offered at an increasing pace, ...

Table of contents

• Cover Page
• Title Page
• Copyright
• List of Figures
• List of Tables
• List of Contributors
• Acknowledgement
• Prologue Introduction and Overview of Field
• Chapter 1: Trends in Building Simulation
• Chapter 2: Uncertainty in Building Simulation
• Chapter 3: Simulation and Uncertainty Weather Predictions
• Chapter 4: Integrated Building Airflow Simulation
• Chapter 5: The Use of Computational Fluid Dynamics Tools for Indoor Environmental Design
• Chapter 6: New Perspectives on Computational Fluid Dynamics Simulation
• Chapter 7: Self-Organizing Models for Sentient Buildings
• Chapter 8: Developments in Interoperability
• Chapter 9: Immersive Building Simulation
• Epilogue