Biomass fast pyrolysis energy balance of a 1kg/h test rig

The present paper offers a methodological approach towards the estimation and definition of enthalpies constituting an energy balance around a fast pyrolysis experiment conducted in a laboratory scale fluid bed with a capacity of 1 kg/ h. Pure N2 was used as fluidization medium at atmospheric pressure and the operating temperature (∼500°C) was adjusted with electrical resistors. The biomass feedstock type that was used was beech wood. An effort was made to achieve a satisfying 92.5% retrieval of products (dry basis mass balance) with the differences mainly attributed to loss of some bio-oil constituents into the quenching medium, ISOPAR™. The chemical enthalpy recovery for bio-oil, char and permanent gases is calculated 64.6%, 14.5% and 7.1%, respectively. All the energy losses from the experimental unit into the environment, namely the pyrolyser, cooling unit etc. are discussed and compared to the heat of fast pyrolysis that was calculated at 1123.5 kJ per kg of beech wood. This only represents 2.4% of the biomass total enthalpy or 6.5% its HHV basis. For the estimation of some important thermo-physical properties such as heat capacity and density, it was found that using data based on the identified compounds from the GC/MS analysis is very close to the reference values despite the small fraction of the bio-oil components detected. The methodology and results can help as a starting point for the proper design of fast pyrolysis experiments, pilot and/or industrial scale plants.

Projected sequential Gaussian processes:a C++ tool for interpolation of large datasets with heterogeneous noise

Heterogeneous datasets arise naturally in most applications due to the use of a variety of sensors and measuring platforms. Such datasets can be heterogeneous in terms of the error characteristics and sensor models. Treating such data is most naturally accomplished using a Bayesian or model-based geostatistical approach; however, such methods generally scale rather badly with the size of dataset, and require computationally expensive Monte Carlo based inference. Recently within the machine learning and spatial statistics communities many papers have explored the potential of reduced rank representations of the covariance matrix, often referred to as projected or fixed rank approaches. In such methods the covariance function of the posterior process is represented by a reduced rank approximation which is chosen such that there is minimal information loss. In this paper a sequential Bayesian framework for inference in such projected processes is presented. The observations are considered one at a time which avoids the need for high dimensional integrals typically required in a Bayesian approach. A C++ library, gptk, which is part of the INTAMAP web service, is introduced which implements projected, sequential estimation and adds several novel features. In particular the library includes the ability to use a generic observation operator, or sensor model, to permit data fusion. It is also possible to cope with a range of observation error characteristics, including non-Gaussian observation errors. Inference for the covariance parameters is explored, including the impact of the projected process approximation on likelihood profiles. We illustrate the projected sequential method in application to synthetic and real datasets. Limitations and extensions are discussed. © 2010 Elsevier Ltd.