Skip to main content

Thermal Exchange Modelling and Power Optimization

Final Report Summary - TEMPO (Thermal Exchange Modelling and Power Optimization)

Executive Summary:
The objective of the TEMPO project was the development of a comprehensive model library for aircraft environmental control and cooling system simulation. The library covers multiple physical and functional domains with focus on thermo-fluid and power modelling. In the thermo-fluid domain the library covers air systems, liquid cooling systems and vapour cycle systems, allowing the simulation of comprehensive power system architectures on aircraft level. The library was implemented using the object-oriented modelling language Modelica. The library is designed to allow scalable modelling. Scalable modelling means that the model equations representing the system model can be easily exchanged to adapt a system model to the simulation requirements at a specific phase in the system design cycle. Instead of rebuilding a system model for a new simulation task as the system design cycle progresses, the system topology information already available in existing system models can be reused by automatically switching the equations underneath. During the course of TEMPO, a library structure that supports this feature has been developed and implemented. Currently, the library contains four "layers", each designed for the application in a specific phase of the design cycle. The first layer consist of low complexity, static models. This layer is intended for use in early conceptual design, where different architectures are under investigation and component performance requirements need to be deducted from high level system performance requirements. The second implemented layer consists of more detailed static models. Thermodynamic behaviour of the components is governed by performance
maps and allow to feed data that becomes available from detailed component design into the system model. This allows early predictions and verification of system performance. The third set of models include dynamic models. The models include one-dimensionally discretized components and first principle modelling approaches. The physical and modular approach followed in modelling these components result in highly adaptable models which can be parametrized over a broad range for fast simulation or high thermodynamic accuracy. The fourth layer of models is suitable for fixed-step solver application and designed for fast simulation speed while capturing the system dynamics. The library is complemented with a cabin model for a single-aisle type commercial passenger
aircraft. The cabin model includes three independent temperature zones including the flight deck and the air distribution system. The dynamic thermal behaviour of the fuselage structure is modelled in detail with multi-layered wall, floor panel and internal structure models. Radiation effects are modelled internally as well as externally.

Project Context and Objectives:
In the frame of the Clean Sky Systems for Green Operations (SGO) technology demonstrator, an electrically driven environmental control system (ECS) is developed. This system is both capable for air conditioning as well as for thermal load management. The development of the real system is accompanied by a comprehensive simulation study of the complete system in order to optimize and
validate the performance of such architectures. Multiple members are involved in the model development process. The complete system model is created from the models supplied by the different members. Modelica was chosen as a modelling language and common base for model exchange. The work carried out in the frame of TEMPO provides the models for the air system (Pack), vapour cycle system
(VCS) and liquid cooling system (LCS) to the responsible SGO member. In addition to providing the models for the SGO technology demonstrator, the library addresses as a second aspect the management of system models along the progress of the system design cycle. In order to efficiently manage the numerous tasks and individuals involved in the development of complex technical systems
and to ensure a successful completion of the development process, the steps and actions taken follow a predefined formalized process. This process is typically broken down into characteristic phases. During all phases, modelling and simulation can be used to support the design process. In each phase, the simulation tasks at hand require models that are designed to specifically answer to the simulation problem to be solved at the current stage of the design cycle. However, the simulation tasks in each phase may impose different requirements on the models regarding e.g. the physical phenomena of interest, the thermodynamic accuracy, or the simulation speed. Specialized tools can be used to meet the requirements for each simulation task. The library developed in TEMPO aims at
using the object-oriented features of the Modelica language to integrate multiple layers of models. Each layer corresponds to a set of models designed for a specific simulation task. Different layers are integrated such that a single system model that has been assembled from available library component models, can be “scaled”, i.e. reused, for a simulation activity of a different phase of the system design cycle.

Project Results:
The object-oriented modelling language Modelica allows to organize models in a package tree structure and thus to create model libraries. The library created in TEMPO has the models organized in functional groups. I.e. there is a package "Machines" for components that act on the thumb-fluid domain such as pumps, compressors or turbines, a package and "HeatExchangers" for heat exchanger components.

Following the general design principle of a flat hierarchy, the components are designed with a low number of inheritance levels and without multiple inheritance. This serves to achieve better readability of the implemented model equations, easier localization of model equations by the model developer at the cost of some duplicate code.

The inheritance structure is limited to three inheritance layers. All usable component models inherit from a common base class which provides an interface to system wide data. This bottom level base class does not contain any equations that describe component behaviour, so the additional inheritance layer does not affect the readability of model equations. The second inheritance layer is the base class for a specific component such as e.g. an air domain centrifugal compressor, an air to air heat exchanger or a liquid domain valve. The base class of a component provides the component interfaces like fluid connectors, the graphical icon information and initial value parameters. The base class does not contain the model equations that describe the components behaviour. The third and last inheritance layer then contains the model equations. The library developer does not have to browse multiple classes to get a picture of the overall equation set that makes up the model.

The developed library covers models for three different fluid domains employed in aircraft power systems. The air domain for the air conditioning pack and cabin system, the incompressible liquid domain for single phase fluid cooling systems and the refrigerant domain for phase changing vapour compression cycle systems. Although a common approach in Modelica modelling, the component model equations are not entirely decoupled from the fluid model. In the developed library, all thermo-fluid components are by design associated to fixed fluid domains. Again, this allows to reduce code overhead within the individual components at the cost of duplicate component models. The domain to which a component model belongs is colour coded into the graphical layer of the models.

The library is interfaced to the SGO member in house tools. The Modelica language's feature for external function calls are used to enable access to a C++ library. The C++ library is compatible to the Modelica external function interface and has been implemented to act as a wrapper library around the SGO member internal tools. SGO member tool interfaces have been developed for fluid property calculation and for performance map data evaluation.

The library contains models for multiple "Detail levels". A "Detail level" (DL) is a full set of component models that are modelled towards a specific simulation task. Depending on the component, an individual Modelica class exists for each Detail Level. Some classes are used for multiple Detail Level; this depends on the requirements for a particular Detail Level. Within TEMPO, four Detail Levels were implemented.

The first Detail Level is intended for the initial development, the conceptual phase of the design cycle of an environmental control and/or cooling system.
System simulation is used at this stage to investigate possible technical solutions and architectures which fulfil the customer's system performance requirements. The simulation problem at this stage is an inverse problem, the question is 'What must the system parameters be, how must the subsystems or components perform' in order to meet a system performance requirement as opposed to a forward problem, where the system performance is computed from given system parameters. The components for this Detail Level are statically modelled. The model equations do not aim for detailed or accurate thermodynamic behaviour. The user can use parameters to specify component behaviour. Local over- and underdetermination options allow to inverse causality where it is needed.

The second Detail Level is designed for application in the pre-design and detailed design phase of the system design cycle. The components for this Detail Level are also modelled statically. Thermodynamically the equations are more accurate and are able to predict off-design point performance. Typically, component models are map-based so that data which has been generated or otherwise become available with the progress of the design cycle can be fed back into the system model in order to verify compliance of the system performance with the original high-level requirements.

The third Detail Level focusses on the system dynamics. Parallel to the development of the actual system, a control strategy and the controller itself is developed. The modelling approach in this Detail Level is mainly physical.
Core components such as heat exchangers are parametrized geometrically.
One-dimensional discretization and first principle modelling with strict implementation of mass and energy balances enables accurate prediction of dynamics and thermodynamic performance.

A fourth Detail Level consists of models that support simulation with fixed-step solvers such as Euler and Runge-Kutta. Fixed step solvers are necessary for Real-time and Hardware-in-the-loop (HIL) simulations. HIL simulations are typically required later in the system design cycle, when controller hardware has been manufactured or acquired. Within the scope of TEMPO, no export to real time hardware or actual HIL simulations were carried out. For the fixed-step solver application, models were developed that fulfil the particular stability criteria which apply for this category of solvers as well as the constraint on computational complexity that allows real time computation of the full ECS system.

The library models for the different Detail Levels are integrated into a scaling mechanism that was developed in TEMPO using the object-oriented features of the Modelica language. The scaling mechanism allows to efficiently change the model equations of components that have been assembled and connected to form a system model. Instead of having to re-build a system with models of another Detail Level and to parameterize the entirety of the components in the system as the system design cycle progresses, one central parameter flag can be changed to switch the mathematical representation of the system model to any one of the available Detail Levels. The developed scaling mechanism uses the container-approach, where a scalable component model acts as a container model.
The container model contains objects of all available detail levels, only that the Detail Level specific sub-models are only instantiated conditionally. When the Modelica model is compiled into an executable simulation program, only the equations of the selected Detail Level will have an effect on the components simulated behaviour. The integration extends to the graphical layer of the Modelica models, so that the different parameter sets that may be required for different Detail Level specific representations of a component are available to the model when necessary. The functionality of the scaling mechanism has been demonstrated with the successful simulation of an example multi-domain environmental control system including the air system, a liquid cooling loop and a vapour cycle system.

The library has been complemented with a dynamic cabin model developed through an subcontract. The cabin model replicates a single-aisle sized commercial airliner and models three independently controllable temperature zones, the flight deck and forward and aft passenger cabin temperature zones. Each passenger cabin temperature zone is represented by two discrete air volumes, one representing the upper section of the passenger compartment and the second representing the lower section of the passenger compartment. Additionally represented by fluid control volumes are the crown area of the cabin and the triangle/underfloor area. The different layers of the cabin walls and floor are included in the model as well as the distribution of transparent and opaque surfaces and their orientation towards external and internal radiation sources.

Potential Impact:
The developed Modelica library has the potential to push the efficiency of modelling and simulation activities of the SGO member by integrating models for all phases of the system design cycle into the library and by combining them with a model scaling mechanism. A possibility is the unification of the tool chain for the different modelling and simulation activities. Where currently independent tools are used for different activities which each require a complete set-up of the system model, the scaling feature of the Modelica library allows to re-use existing system architecture information and parametrization present in a model.

The library will be continued and maintained at the Institute of Thermo-Fluid Dynamics and is available for application in future research projects and student training activities.

The work progress and results were presented at international conferences of the aviation industry and the modelling and simulation community, involving both academia and industry. The cumulated results of the TEMPO project are subject to an ongoing PhD-Thesis work at the Institute of Thermo-Fluid Dynamics.

List of Websites: