Meet us in September at the 11th Modelica Conference You can find us in the exhibition hall, the Vendor Session, the Tutorial Session, and at our nine paper presentations.

It is now less than three weeks left to the 11th International Modelica Conference. This bi-annual main event for the Modelica and FMI communities is this year held in Versailles, France on September 21-23. Registration is still open at the conference webpage, where you also find the preliminary program.

Conference highlights from Modelon

In addition to being Silver Sponsor for the event, Modelon will contribute with:

- A Vendor Session on Modelica and FMI Products from Modelon

  • Modelica libraries for several physical domains, including automotive, aerospace and energy, is highlighted and demonstrated for selected applications.
  • New features of Modelon’s FMI connectivity products, including the FMI toolbox for MATLAB and the FMI Add-in for Excel, are presented, as well as new optimization capabilities in and the OPTIMICA Compiler Toolkit.
  • A novel Modelica and FMI cross test platform, the OPTIMICA Testing Toolkit is introduced.

- A Tutorial on Optimal control and state estimation with Modelica and Optimica. This tutorial demonstrates how Modelica and Optimica are used to formulate and solve optimization problems targeting control industrial processes. A key focus of the tutorial is usability of numerical algorithms when solving dynamic optimization problems arising in control applications. Participants are offered hands on experiences with effective tools for achieving convergence in industrial optimization problems.

The tutorial offers on open source tool track which is based on and one commercial tool track which is based on the OPTIMICA Compiler Toolkit. The latter track also offers hands on experiences with optimization of industrial power generation systems.

- A generous space in the conference exhibition.

Take the opportunity to meet our sales team, developers, experts and management, and learn all about Modelon solutions and our offerings!

Seven articles and two poster presentations:

A dynamic model of a Central Receiver type CSP plant (CRS) was implemented in Modelica. The model consists of a set of CRS specific components, along with a Rankine cycle to form a complete system. Main components include models of a sun, heliostat field, receiver, storage tank and a Rankine cycle including a steam generator. The system uses a molten nitrate solution, called Solar Salt, as heat transfer fluid.

The components were modelled and configured after a reference system – the Solar Two test facility in CA, USA, operational in the late 1990’s – but are generic and rescalable.

The components and the full system were tested in a series of simulations – both dynamically and during steady state conditions – and the results were compared to data from the reference system. The dynamic behavior of the models aligned with expectations, although time constants could not be evaluated due to lack of dynamic reference data. The steady state characteristics were adequate for most models, although some complementary work needs to be done on the Receiver model.

This work is the result of a Master Thesis project at Lund University in collaboration with Modelon AB, Lund, Sweden (Edman, 2014). The models developed are largely based on the various model libraries in Modelons portfolio, especially the ThermalPower library, the LiquidCooling library and the Modelon Base library.

Production planning for district heating networks aims at finding the most profitable scheduling of the production units of the system. This task is typically handled as an optimization problem. The standard approach for solving this problem is to create a highly simplified model of the system, so that the optimization problem can be solved using linear methods. In this paper an alternative method, previously implemented in (Velut et al, 2013), is presented, in the context of distributed networks.

The production planning problem is solved in two steps by integrating physics-based models into the standard approach.

- The first optimization step solves for the discrete variables of the unit commitment problem (UCP) using mixed integer linear models and standard mixed-integer solvers.

- The second step, the economic dispatch problem (EDP), considers dynamic optimization using physics-based non-linear models that utilize the unit statuses from the first step.

For this purpose the nonlinear optimization features of (Modelon AB, 2015) is used. All optimizations aim at maximizing production profit using fuel, electricity and heat prices as well as maintenance and start-up/stop costs as variables.

The physics-based modeling in the EDP means that important physical variables such as supply temperature, supply flow rate, pump speeds and condenser pressures are included in the formulation. This makes it possible to formulate constraints on these variables corresponding to the limitations of the physical system, which will be utilized in the optimization.

The modeling has focused on distributed consumption and production. The goal has been to represent the most important production units and network distribution of the Uppsala district heating network in Sweden.

The district heating network has been modelled using physics-based pipes, including mass flow dependent delays and temperature dependent (district heating water and outdoor temperature) heat losses. The total heat demand is divided between several customers.

Comparisons between optimizations with and without distribution network models have been performed, showing that more detailed modeling of the net impacts the production planning in several ways. Most notable is the reduction of costly production peaks which is achieved by considering the different transportation times to different customers. Experiments show that costly unit start-ups can be delayed when this effect is considered.

Other results of the distribution model include production compensation for heat losses and time delays and usage of the net for heat storage (accumulation). The optimizations also result in production plans where supply temperature and flow rate is minimized and maximized, respectively, and there is a balance between heat production and heat consumption.

This article presents a way of implementing different control strategies for power electronic converters in the Modelica modeling language. The different control modes were fitted into flexible models that could be interconnected in various power grid topologies.

The power grid examples were controlled and kept stable during various load scenarios, using the developed controlled converter models. The work was performed using the Modelica tool Dymola. Modelica is an equation-based object-oriented modeling language. Electrical components in the Electric power library (EPL) from Modelon were used to model power electronic units, grids and other electrical infrastructures.

The outcome of this effort was simulation results which clearly demonstrate that the developed controllers enable scalable and controllable DC power grids.

Keywords: HVDC, smart grids, converter control.

This paper describes the development and requirement specification of an open-source framework for multi-phase multi-component thermo properties in Modelica. The goal is to have a standardized interface to multi-component multi-phase fluids with access to external property packages in Modelica. This will make it easier to develop models for e.g. the process industry. The library uses a model based interface and implications of such a design are analyzed and compared with the traditional function based interface.

The availability of properties for steam and flue gases initiated the use of Modelica in the power industry, where it today is a well-established technology with several commercial and open source libraries available. High quality fluid properties are laborious to produce and their non-availability is therefore a typical blocking argument for the use of a certain tool or technology.

In this project a Modelica library for multi-phase multi-component fluids has been developed together with an external C/C++ Modelica property interface with back ends to CAPE-OPEN, RefProp and FluidProp. The framework also contains a Modelica library for distillation processes for verification and testing of the media interface design. 

In the past was successfully applied for generating optimal trajectories. Using it for Nonlinear Model Predictive Control (NMPC) is the natural next step and sets high requirements on calculation time.

To improve real time capabilities warmstarting of the optimization and elimination of algebraic variables based on Block Lower Triangular (BLT) form were implemented.

In performance comparisons, using the example of steam temperature control, a speed-up of the optimization time by a factor of five and of two respectively was measured. The increased efficiency allows application of NMPC to faster systems than before.

Keywords: NMPC, BLT, IPOPT,

With the Functional Mock-Up Interface standalone sub-systems can be modelled to be part of larger systems that needs a framework for coupled integration. This paper suggest one way of solving the issue by aggregating sub-systems to one unified system that internally handles sub-system communication by coupling. The aggregated system can then be solved by applying a single solver with the benefit of using an aggregated Jacobian and the ability to monitor all sub-system events.

In a proof of concept two FMUs, each modelling a pendulum with an external force acting on the pivot, are coupled together with a spring and simulated as an aggregated system using the CVode solver in Assimulo. As reference a monolithic model of the coupled system was made as an FMU and integrated using CVode.

The framework is not limited to coupling of FMUs but can be used to couple Python based problem classes defined by Assimulo. It can also add events to the aggregated system externally. The latter is demonstrated in a test where walls are added to the aggregated system of two pendulums coupled with a spring to block each pendulums motion. 

The need for regression testing increases as the size and complexity of software project grows. This is no different from a Modelica library or tool. Large Modelica projects often involves several Modelica tools and libraries which are under development. In those situations, with several orthogonal code bases, the need for systematic regression testing is needed. To address this, Optimica Testing Toolkit (OTT) was developed. OTT is a framework for performing automatic testing on Modelica models.

OTT also provides a testing pipeline that is tool agnostic, meaning it provides the same testing pipeline regardless of what compiler and simulator performs the actual model compilation and simulation. Tool agnosticism is provided by means of an abstraction layer between OTT and the actual tools. Each tool is hooked into the abstraction layer via a plugin tailored specifically to that tool.

As part of the development cycle a Graphical User Interface (GUI) was developed. The GUI can be used for test authoring, test configuration and test execution. One important aspect considered during development was to ensure that the GUI had good usability. We used a number of different user studies together with the users in order to discover usability problems, and then used iterative development to address and fix those issues.

Nonlinear Model Predictive Control (NMPC) is a control strategy based on repeatedly solving an optimal control problem. In this paper we present a new framework for the platform, developed specifically for use in NMPC schemes. is an open-source software for simulation, optimization and analysis of complex dynamic systems described by Modelica models.

The new framework, the MPC framework, utilizes the fact that the structure of the optimal control problem to be solved does not change between solutions, thus decreasing the computation time needed to solve it. The new framework is compared to the old optimization framework, the open-loop framework, in in regards to computation time and solution obtained through a benchmark on a combined cycle power plant. The results show that the MPC framework obtains the same solution as the open-loop framework, but in less than half the time.

For the benchmark system presented in this article, the total computation time for each sample was decreased by an average of 70 %. The MPC framework also includes some features which makes easier to use for NMPC applications.

This paper describes a method for automated deployment of Modelica models as simulators in Microsoft Excel using Functional Mockup Interface (FMI) and FMI Add-in for Excel. 

Using existing interfaces, integration with modeFRONTIER, a process automation and design optimization tool widely used in industry, is demonstrated and illustrated with several different example models in different physical domains to highlight the range of applications and types of analyses that can be covered with the automated toolchain. 

This toolchain can be applied to any FMU and streamlined with automation enabled by the supporting annotations. The sample applications include model correlation with an HIV virus dynamics model, hydraulic crane optimization, heat exchanger robust design, and hybrid vehicle electric range fleet population estimation. 

by Adina Tunér