• No results found

Simulating Fuel Cell Operation

Chapter 4: Simulating Fuel Cell Micro-CHP Systems

4.2. Theoretical Arguments

4.2.2. Simulating Fuel Cell Operation

4.2.2.1. Importance of Domestic Energy Demand

The way in which the fuel cell is operated, rather than how well it can operate is what ultimately determines the impact it will have on domestic energy consumption. The amount of energy displaced from traditional and less efficient means will be greater if the fuel cell is given ideal operating conditions – long running times at high output. The pattern of energy consumption in UK homes will therefore have substantial influence over the benefits the fuel cell can provide as current systems follow the instantaneous or predicted energy demands.40 The importance of the energy profile is compounded by the lack of a quantitative relationship between profiles and the impacts of installing micro-CHP, as explained in Section 3.2. It is therefore crucial that the simulation of the fuel cell’s operation is representative, and accurately mirrors how it would be run in reality.

In previous work, it is typical for a small number of profiles to be used, due to the limited availability of data and computational time required to run the simulation. It is typical to use data from 1-3 houses either measured over a whole year, or just from a selection of days (e.g. winter, summer and shoulder).[109, 130, 132, 205] Such a limited number of profiles will give only a small subset of the results that could be expected. There is no reliable method to categorise profiles and choose a uniformly distributed selection of houses or days, and so there

is no guarantee that this subset will be representative of the whole population, or that a skewed set of results has been avoided. It is therefore imperative to use the greatest possible number of energy profiles; placing an equal importance on the amount of data used as on its quality.

There is however very little data on energy profiles available in the UK, which poses a barrier to researching microgeneration. There is no particular reason for this lack of data, although it is time consuming and laborious to collect and process. It is common for fuel cell manufacturers to collect their own demand profiles for in-house research, however the commercial value of such data prevents it from being made available to the public.[206] The following data sets were found to be publicly available on request41 in the UK:[207]

Electricity demand collected from 217 homes as part of the DTI photovoltaic field trials,[208] which is available as a DVD set from the DBERR.[209]

Electricity and thermal demand data collected by the BRE from 130 homes in the Milton Keynes Energy Park between 1988 and 1991.[210-212] The data set was made available by Alex Summerfield at The Bartlett, University College London.

Data could alternatively be sourced from other countries, however energy consumption habits are different throughout the world, and so demand profiles are not easily transferred from other countries’ building stock.[85] Another option would be to simulate domestic energy profiles; the integrated building simulations introduced in Section 3.2 have been used by many authors [86, 108, 135], and other methods of simulating demand have been demonstrated.[213-216] These are typically based on time-usage surveys of household occupants, which are combined with consumption profiles for individual devices.

It is argued that these simulations add a further level of theoretical abstraction to the model and are no substitute for real data.[126] The time-use surveys they rely on are also limited in number and often outdated; and those methods which rely on building construction and design will not capture the enormous variation in demand that two outwardly similar houses could have.42 Furthermore, there have been no comparisons between simulated and measured profiles (as even the methods of comparison are not well established), and the impact of using simulated rather than measured profiles on micro-CHP models has not been investigated. For

41 In addition to these, researchers at Herriot-Watt University have access to electricity and gas consumption data measured with 1 minute resolution in a set of 30 homes.[123, 205]

42 A Danish study showed that two identical houses on the same street with the same number of occupants could have energy demands that varied by a factor of 10 or more; one household presumably operates on a thrift economy while the other is more hedonistic (or has teenagers).[217]

these reasons simulated energy profiles were avoided in favour of the limited amount of measured data available.

4.2.2.2. Choice of Operating Strategy

The system controller will govern how the fuel cell system operates in the home. As well as making low level decisions (fuel and air inlet rates, coolant pump speeds) it will decide when the unit should switch on and off, and what level of power output should be produced. Two classes of logic have been demonstrated in current micro-CHP products: those which follow the instantaneous energy demand (electrical, thermal, or both); and those which use predictive or learning algorithms. The latter of these attempt to compensate for the limited transient performance of the fuel cell (i.e. long start-up time) by analysing the demand of energy from the home and ensuring that the fuel cell is operational when energy is required. Hawkes however showed that by simply following the maximum of electricity and thermal demands from the house, a fuel cell with constrained ramping ability can approach the optimal minimum operating costs.[132]

While the predictive type of controller is beneficial for maximising the utilisation of the fuel cell, it will increase the overall complexity of the system, and thus increase capital costs and the number of components that are susceptible to failure. Similarly, modelling such a system controller would increase the complexity of the computer simulation, reducing transparency and increasing the time required to perform calculations. It was therefore decided that simple load- following operating strategies would be used, allowing high volumes of profile data to be processed and yet still producing results which approached the optimum for each system.

From the load-following strategies identified in Section 3.2, following the maximum of heat and electricity demand was seen as ideal, enabling the greatest running time for the fuel cell and thus maximising the benefits it provides. Following the electricity demand was used as an alternative in scenarios where the production of excess electricity was forbidden; i.e. when there was no ability to export to the national grid.43 Following only the heat demand or the minimum of heat and electricity would be sub-optimal, as electricity output should be maximised due to its high value and displaced carbon intensity.

These simple load-following strategies were augmented with simple fuzzy logic which attempted to maximise the utilisation of the fuel cell and minimise the number of shut-downs required. These adaptations, and the way in which they were implemented in the fuel cell simulation are discussed later in Section 4.4.