Performance Modelling with TRNSYS

Solar Thermal Power plants have components with thermal time constants that can be minutes or even significant fractions of an hour. As a consequence, a reliable prediction of a real plant’s annual performance, requires an accurate transient system model.

Simulation of dish based solar thermal plants using steam is currently being investigated using the transient simulation package TRNSYS [1]. TRNSYS together with the user interface IISiBat, offers many features; including the availability of libraries of existing simulation components, the capability of interconnecting system components in any desired manner to accomplish a specified task and especially user friendly tools to facilitate the creation of custom components.

A Solar Thermal Electric Component (STEC) model library for TRNSYS has been created under the SolarPACES umbrella by DLR Köln and Sandia.[2]. The STEC library includes models for Rankine cycle, Brayton cycle, solar thermal receivers, parabolic trough, condensers amongst others. Previous work at ANU [2] has developed new components application to ammonia based thermochemical energy storage systems using dish concentrators. This has included a paraboloidal dish component (Type 251), a reactor component and a thermochemical receiver component.

The current work is looking at the simulation of Direct Steam Generation (DSG) dish receivers in the context of multiple dish, central generation Rankine cycle power plants. The type 251 dish component is being used, it produces the aperture area and incident solar power for a receiver, using rim angle, slope error and dish collector area, provided by a user supplied data table. To date the component has been used to simulate the ANU 20m2 and 400m2 dish systems. Components to model a DSG receiver and an element of steamline are currently under development. The receiver component models the true transient behaviour of a DSG receiver. It solves the energy balance taking into account the thermal mass of the receiver tubes and the movement of the boiling zone and subsequent variations in exit mass-flow in response to insolation variations.

Figure 2 shows some preliminary results obtained for a dish and receiver geometry the same as the ANU SG3 400m2 dish system. Typical meteorological year (TMY) data for Canberra [4], has been used for a particular day chosen randomly in the summer time of Canberra. One minute time steps have been used with TRNSYS. At this stage model parameters such as receiver mass have not been exactly matched with the SG3 prototype.

Figure 2 Daily result of TRNSYS simulation on SG3 Dish system.

Numerical Study of Convection Loss from Cavity Receivers

The transient model discussed in the previous section requires specification of the various thermal losses by the user in the form of heat transfer coefficients and effective emissivities. To provide these parameters accurately and to develop more efficient receiver designs, experimentally verified models are required.

Convection losses are currently being examined in detail. For a cavity receiver, convection losses depend on receiver geometry, inclination angle, temperature distribution and wind-speed. During normal operation, convection losses can be of the order of 10% of solar power input [5,6]. There have been a small number of experimental studies into convection losses from cavity receivers. In the work by Koenig and Marvin as summarized in [7], an empirical correlation that takes into account the inclination angle and aperture/cavity radius ratio was proposed. Counterpart numerical studies are relatively scant. Most work [8] addresses the open cavity of the central receiver which generally has a fixed zero angle of inclination (f=0). This motivated a numerical study of convection in cavity receivers, including inclination as a variable.

The work employs the widely used CFD Software Package, Fluent (version 5.5), as the simulation tool. Fluent utilizes the finite volume algorithm in the discretization of the set of the governing equations, and so provides the applicability of unstructured mesh [9]. Practically, the validity of its results strongly depends on the physical model and boundary conditions applied.

A cavity receiver of the same geometry as the ammonia dissociation receiver reactor discussed in the following section, has been investigated. To deal with the ambient temperature surrounding air boundary condition, the receiver, as shown in Figure 3, is placed in the centre of a large imaginary cylinder with wall temperature equal to ambient temperature. The size of this cylinder was increased until, with diameter and height of 12m, the flow conditions predicted at the receiver were found to be independent of it.

In initial investigations of natural convection, an isothermal boundary condition was applied to the cavity wall with Tw=650oC, while the other outer walls of the receiver were assumed to be adiabatic. The wall temperature of the cylindrical box was set to the ambient temperature Ta=27oC. The Rayleigh number of the system was estimated to be around 6×106 which falls in the turbulent flow regime. The present work employed the steady momentum-energy coupled solver. The Spalart-Allmaras turbulent model was applied because it was simple and had shown satisfactory agreement with other more complicated turbulent models in preliminary calculations. All thermal-dependent properties of air such as density were handled by using fit equations derived from thermodynamic tables. Before starting the actual simulation, the grid dependency was also investigated. The final grid consisted of 192695 hexahedral cells. A typical computational time was approximately 30 hours on an ALFA-1GHz machine.

Figure 3 Cross sectional diagram of the cavity receiver, and typical profile of mass flux through the aperture at angle =45° and wall temperature Tw = 650 °C.

Figure 3 also shows the typical profile of mass flux through the aperture at f=45o. As expected, there is bulk movement of air entering and exiting the cavity at the lower and upper halves of the aperture, respectively. A comparison of the average cavity wall heat flux obtained from the present work and the Koenig & Marvin (K&M) correlation is illustrated in Fig. 4.2. There is good agreement at a receiver inclination angle of approximately 40o, however below this the present result is obviously lower than that obtained by the K&M correlation and above it, the opposite is the case. Note that the K&M correlation predicts zero wall heat flux at f=90, which is not essentially true. Further numerical investigation and experimental work will be carried out to investigate this discrepancy.

Figure 4. Cavity wall heat fluxversus angle for wall temperature of 650 degrees celsius.

Exergy Analysis

The use of closed-loop thermochemical energy storage systems for the storage of solar energy places fundamental limits on the amount of work that can be extracted from the recovered energy. These arise because of thermodynamic irreversibilities associated with the storage system itself and because of the need to degrade collected solar energy to the characteristic temperature of the reaction system chosen.

The assumption that the irreversibility is entirely due to the reaction processes has allowed useful general expressions for exergetic and work recovery efficiencies to be derived. The exergetic efficiency of such systems is limited only by how close reaction paths can be made to follow the equilibrium line. The work recovery efficiency on the other hand has a maximum value which is characteristic of the reaction chosen and reflects the loss in exergy associated with storing energy at the characteristic temperature of that reaction. The spontaneous separation attribute of the ammonia-based thermochemical system results in thermodynamically reversible paths having constant work recovery efficiencies irrespective of their reaction extent endpoint. This is in contrast to the case of systems where all reactants remain in the gas phase. The transition between the two situations can be examined by calculating constant efficiency contours for the ammonia-based system for various assumed sink temperatures.

The analyses undertaken represent an essential prerequisite for the theoretical development of reactor configurations which optimise efficiencies.

Further reading, see [10]

Reactor Modelling

Steady state modelling of receiver performance for example involves detailed analysis of the energy balance of finite elements of receivers. Taking into account incident flux, re-radiation and reflection to and from all other receiver elements, convective heat loss plus the action of the working fluid. Successful modelling of receivers in this way gives predictions of temperature distributions and overall receiver efficiency. It thus allows alternative receiver geometries to be investigated for performance improvement.

Another line of investigation concerns the optimisation of the heat recovery part of the ammonia based thermochemical system. As part of this investigation, performance modelling of the lab scale synthesis reactor represents a necessary step towards an improved performance of the reactor for further tests with the 1 kWsol lab scale system. Variation of the outer reactor wall temperature profile, gas inlet temperature and reaction extent inlet leads to optimum conditions for certain massflows and pressures. It was found that the heat recovery is quite sensitive to the reaction extent inlet and rather insensitive to the inlet temperature of the synthesis gas. While the choice of an appropriate average outer reactor wall temperature is important, the influence of its slope is negligible. The optimum average outer reactor wall temperature was obtained for massflows between 0.1 and 1.3 g/s and pressures 10, 20 and 30 MPa and will be used for reference in future laboratory experiments. Further modelling will be carried out for the design of a new reactor, as the system will be scaled up to accept the full 15 kWsol power input from the ANU’s 20 m2 dish.

Further reading, see [11].

Control Strategies

Conventional endothermic reactions are controlled by variation of heat input. In solar-driven operation however, control has to be developed around mass flow variation. Solar irradiation input patterns are stochastic in nature and, at times of fast cloud transients, can be very extreme with time constants well below one second.

Control of the reaction extent of ammonia dissociation via mass flow variation has been trialed using a prototype solar reactor. The assessment of solar transient experiments has shown that the dynamics of the reaction extent is very much congruent with the peak operating temperature of the reactants. Control disturbances arising from solar irradiation, system pressure and peak reactant temperature were found to be insignificant due to the large thermal inertia of the thick-walled pressure vessel design necessary to undertake high-pressure solar ammonia dissociation. If part of a closed-loop system, the solar ammonia dissociation reaction can therefore be operated via conventional temperature control (PI or PID), which negates the need for gas analysis.

Economic Analysis

Thanks to funding from the Swiss Office of Energy (BFE/OFEN), an international study group combining partners from industry and academia was formed to examine the techno-economic viability of a hypothetical 10 MWe solar thermochemical base-load power plant for Alice Springs in Central Australia.

The main project partners were Ammonia Casale S.A. (Lugano, Switzerland), FC Consulting (Rickenbach, Switzerland), L.&C. Steinmüller GmbH (Gummersbach, Germany), Siemens Power Generation (Kuala Lumpur, Malaysia) and the Energy Research Centre as well as the Centre for Sustainable Energy Systems of the Australian National University (ANU).

The study group was able to formulate a first-pass power plant design that could be constructed by dominantly using proven and standard materials, components and technologies. The design considered the application of two solar technologies. These were ANU’s 400 m2 paraboloidal dish collector technology and L.&C. Steinmüller’s volumetric- air “power tower” technology. The latter technology was assessed to be less effective for this thermochemical storage concept than the dish-based technology due to a less direct energy transfer from the sun to the ammonia system.

Analysis of the possibilities for sub-system improvements, that would come from a thorough pre-construction design study, revealed that the overall system performance of such a first demonstration power plant could be significantly augmented. Whilst still using standard components and techniques, a net solar-to-electric conversion efficiency of the order of 16% – 18% was predicted for such an optimised design, delivering an LEC of AUD 0.20 – 0.25 per kWhe. These results were encouraging for what would be a first pre-commercial demonstration system.

Considering the inherent potential for cost reduction from progressing through development stages and from economy- of-scale, it was concluded that this technology could well become one of the earliest and potentially cost-effective solution to the challenge of producing solar-only electricity on a continuous 24-hour basis. Provided the necessary project progress and experiences can be accomplished, an LEC of as low as AUD 0.12 – 0.15 can be expected in the future.

For further reading see [12]

References

[1] Solar Energy Laboratory, University of Wiscosin-Madison. TRNSYS – A transient simulation program. Version 15, Volume I. Madison, WI, USA (2000).

[2] Pitz-Paal R. and Jones S. A model library for solar thermal electric components (STEC). Technical Report; DLR Köln: Köln, Germany (1999).

[3] Kreetz H. Heat Recovery in a solar thermochemical power system. Ph.D. Thesis, Department of Engineering, ANU, Australia (2001).

[4] Morrison G. and Litvak A. Condensed solar radiation data base for Australia. Technical Report, Solar Thermal Energy Laboratory, University of New South Wales, Sydney, Australia (1988).

[5] Javam A. and Armfield S. Stability and transition of stratified natural convection flow in open cavities. J. Fluid. Mech. 445, 285-303 (2001).

[6] Harris J. and Lenz T., Thermal performance of solar concentrator / cavity receiver systems. J. Sol. Energy 34, 135-142 (1985).

[7] Anderson R. and Kreith F. Natural convection in active and passive solar thermal systems. Adv. Heat Transfer 18, 1-86 (1987).

[8] Chen Y. and Tien C. A numerical study of two-dimensional natural convection in square open cavities. Numer. Heat Transfer 4, 249-283 (1981).

[9] Fluent Inc., Fluent 5 User Guide, (1998).

[10] “Thermodynamic Limits on the Performance of a Solar Thermochemical Energy Storage System”, K. Lovegrove, Int. Journal of Energy Research, 17, 817 (1993).

[11] Kreetz H. and Lovegrove K. (1998). Performance Modelling of a Synthesis Reactor for a Solar Thermochemical Energy Storage System. In proceedings of Solar’98- ANZSES Annual Conference, Christchurch, New Zealand.

[12] Luzzi A., Lovegrove K., Filippi E., Fricker H., Schmitz-Goeb M., Chandapillai M. and Kaneff S. (1998) Base- load solar power using the ‘Haber-Bosch’ process. Final Report, Swiss Office of Energy (BFE/OFEN), Bern 3003, Switzerland.