A NOVEL FMI AND TLM-BASED DESKTOP SIMULATOR FOR DETAILED STUDIES OF THERMAL PILOT COMFORT *Robert Hällqvist , **Jörg Schminder , *Magnus Eek , ***Robert Braun , **Roland Gårdhagen , ***Petter Krus *Systems Simulation and Concept Design, Saab Aeronautics, Linköping, Sweden **Applied Thermodynamics and Fluid Mechanics, Linköping University, Linköping, Sweden ***Fluid and Mechatronic Systems, Linköping University, Linköping, Sweden Keywords: OMSimulator, FMI, TLM, Pilot Thermal Comfort, Modelling and Simulation Abstract can be shifted to earlier design phases if detailed simulations of large portions of a complete Modelling and Simulation is key in aircraft sys- aircraft are feasible. However, in order to further tem development. This paper presents a novel, increase the use of M&S, and expand the scope multi-purpose, desktop simulator that can be of analysis using models, simulation of coupled used for detailed studies of the overall perfor- models developed in a wide variety of different mance of coupled sub-systems, preliminary con- tools need to be made available on the engineers’ trol design, and multidisciplinary optimization. and researchers’ desktop computers. Also, Here, interoperability between industrially rel- risks associated with tool-vendor lock-in and evant tools for model development and simu- licensing costs need to be kept at a minimum lation is established via the Functional Mock- if the benefits of M&S are to be further exploited. up Interface (FMI) and System Structure and Parametrization (SSP) standards. Robust and The ITEA3 financed research project Open distributed simulation is enabled via the Trans- Cyber Physical System Model-Driven Certified mission Line element Method (TLM). The ad- Development (OpenCPS) [1] aims to address the vantages of the presented simulator are demon- challenges identified by academia and industry strated via an industrially relevant use-case where in terms of efficient model integration and simulations of pilot thermal comfort are coupled simulation. A first step is to deploy available to Environmental Control System (ECS) steady- standards and techniques for robust and parallel state and transient performance. simulation in current industrial M&S meth- ods and processes. This paper aims to show 1 Introduction how existing standards can be used to develop industrially relevant desktop simulators. One As aircraft sub-systems are becoming increas- specific use of such a multi-purpose simulator ingly complex in modern aircraft, Modelling and is presented via an analysis of pilot comfort Simulation (M&S) is essential for understanding coupled to Environmental Control System (ECS) both steady-state and transient behaviour of performance. Pilot thermal comfort is typically tightly coupled sub-systems. Rather than de- assessed without run-time connections to a signing and analysing sub-systems separately, a detailed ECS model. Conversely, ECS analy- holistic approach is needed. Development testing sis simulations are conducted excluding pilot and Validation and Verification (V&V) activities thermal comfort. Consequently, design errors associated with multiple integrated sub-systems occurring as a result of sub-optimization are 1 HÄLLQVIST R, SCHMINDER J, EEK M, GÅRDHAGEN R, KRUS P more likely if the system cross-coupling effects as Functional Mock-up Units (FMUs) [3]. are omitted during design. The bi-directional Exported FMUs can be imported and simulated dependence between pilot thermal comfort and in any FMI supporting tool. Models can be ECS performance is demonstrated by means exported as FMUs for Model Exchange (ME) of a use-case where the pilot comfort measure or Co-Simulation (CS). With FMI for ME, the Fighter Index of Thermal Stress (FITS) [2] is responsibility of solving the model equations used to control the cockpit temperature during a is passed to the integrating tool whereas each simulated mission. CS FMU contains its own independent solver. The integrating tool is merely responsible for In Section2, the enabling technologies are in- orchestrating the communication between FMUs troduced along with the rational of their use in the latter case. Various optional features en- in M&S. The implementation environment is abling advanced master algorithms are specified briefly over-viewed in Section3. The simula- in the standard. These options, for example the tor implementations are described in Section 4.1. provision of directional derivatives for smooth- The simulator characteristics are introduced in ing of sampled inputs, can be used to increase Section 4.2. A pilot comfort study demonstrat- simulation performance and robustness. ing one use of the simulator implementation is presented in Section 4.3. Finally, the conclusions The SSP standard is an emerging standard for de- are given in Section5. scription and exchange of composite simulation models. This standard is the outcome of a Mod- 2 Techniques for standardized and dis- elica Association Project aimed to, along with tributed co-simulation FMI, provide a complete framework for stan- dardized simulation of multiple connected sub- The Functional Mock-up Interface (FMI) stan- systems models, from here on referred to as com- dard [3], the Systems Structure and Parametriza- posite models. The FMI standard is crucial in the tion (SSP) standard [4], and the Transmission sense that it establishes a standardized format for Line element Method (TLM)[5] are three key model exchange between domain specific tools; technologies enabling improvement on the state- however, it does not address the issue of exchang- of-the art in aircraft systems M&S. These tech- ing parameterized composite models consisting nologies are utilized in the OpenCPS project co- of multiple connected model executables. The simulation framework the OMSimulator. A brief SSP standard aims to bridge this gap. In short, overview of a sub-set of the standardized meth- a standardized xml schema (denoted <compos- ods for model export, and composite model de- iteModel>.ssd) is used to carry information re- scription, is presented in Section 2.1. The High garding composite model connections, proper- Level Architecture (HLA) standard [6] may be a ties, and graphics. The <compositeModel>.ssd feasible alternative to FMI for establishing tool is packaged along with its referenced resources interoperability and it is mentioned for context. in a<compositeModel>.ssp file. Examples of ssd However, the focus is placed on FMI and this de- referenced resources are other ssp/ssd files and limitation is motivated by the available tool sup- FMUs. An overview of the SSP standard is pro- port along with the application domain. Some vided by Köhler et al. in [4]. The SSP standard techniques for provision of a numerically stable exists as a mature draft at the time of writing and co-simulation is provided in Section 2.2. as such the tool implementations are scarce. 2.1 Available standards 2.2 Numerically stable co-simulation The FMI standard specifies a generic format The partitioning of co-simulation entities un- for export of model executables, referred to avoidably results in sampling of inputs and 2 A NOVEL FMI AND TLM-BASED DESKTOP SIMULATOR FOR DETAILED STUDIES OF THERMAL PILOT COMFORT outputs. When sampling a continuous system system bandwidth according to the sampling the- information is lost. The impact of this loss of orem; however, this method may have significant information needs to be managed and minimized impact on the simulation performance. Filtering introduces numerical errors as high frequency TLM is a mature and well documented tech- dynamics are removed to allow for a lower nique for numerically stable partitioning of sim- sampling frequency without introducing aliasing. ulation models [7] Naturally occurring time de- Both traditional co-simulation methods, and lays are utilized to derive an independence be- the inherently parallel approach of TLM, have tween connected components or models, specify- benefits and drawbacks if stability is weighted ing a clear and transparent time window for dis- against simulation performance. The traditional connecting and parallelizing coupled simulation algorithms with an adaptive macro step size models without introducing numerical errors as allow for long step sizes, without large numerical a result of sub-system de-coupling. Aliasing ef- errors, for low frequency operating conditions. fects, see for example [8], are reduced as interpo- Schierz et al. present methods for communi- lated input values are available at the discretion of cation step size control for co-simulation with each individual co-simulation FMU solver. TLM FMI in [10]. In such operating conditions, TLM is thus a way to pass high resolution interpo- comes at the cost of simulation performance lated data while maintaining the sub-system in- as the macro step size need to be limited by dependence during the communication interval. the physical delay even though limiting high A lossless, one dimensional, hydraulic transmis- frequencies are unexcited. Much like in the sion line can be described by case of traditional co-simulation, a larger macro step size can be selected at the cost of altered p1(t) = Zc · q1(t) + c1(t) (1) simulation behaviour. p2(t) = Zc · q2(t) + c2(t) Co-simulation entities, utilizing the TLM where technique, require access to interpolated input c1(t) = p2(t − DtTLM) + Zc · q2(t − DtTLM) variables for all internal solver steps. The delay (2) present across a transmission line means that c2(t) = p1(t − DtTLM) + Zc · q1(t − DtTLM): future inputs are