THE H2OOL
Basic Model Development
100% Demand and Supply Match
One of the key objectives for the group was to create a sizing tool that could allow for the efficient sizing of the solar hydrogen system to meet 100% of the demand required by the load. In order to do this a working backwards approach was adopted, starting from the district's load through to the PV panels.
Figure 1: Sizing approach starting with load
Toolset Input Data
First of all the H2ool needs input data. A demand has been taken for a community size around 300 houses, since this is considered to be a ‘district electricity’ scheme.
The H2ool has been modelled to search for the highest demand reached throughout the year, and the fuel cell power is set in correspondence with this.
Next, the sizing procedure undergoes in two stages: first a preliminary sizing assessment, followed by an iterative search.
Preliminary Sizing Assessment
Fuel Cell
Since the fuel cell power is set and in terms of sizing other components, the fuel cells’ hydrogen consumption needs to be found. Experimentation was carried out to determine a relationship between the hydrogen input flow rate and fuel cell power output. From that, a linear relationship (y = 12.604x + 1.6068) was obtained and applied into the H2ool.
(More about our experimental work in the Fuel cell laboratory can be found ).
Storage
In order to calculate the storage size, the demand profile was reassessed to find the day of highest energy requirement. This is the amount of hydrogen that has to be stored with an additional 10% for safety.
(More about the initial storage sizing can be found ).
Electrolyser
The electrolyser needs to meet the amount of hydrogen required by the system. A ratio that represents the relationship between DC power supply and hydrogen produced has been obtained from literature 4.5 kWh per Nm^3 (Barbier, 2005).
(More about the electrolyser sizing can be found here ).
Iterative Search
PV Area
Until this sizing stage, each component has been sized from the load but PV sizing presents a challenge. The size of the PV required cannot be found directly due to the issue of varying energy production from renewables.
Therefore, determining the PV areas required needs an iterative search based investigation.
To start the iterative search an initial estimate is needed. For this, the average daily PV output has been calculated from the H2ool and then matched with the highest daily demand. The iterative search procedure then starts with this estimation and tries to match the required power from the system by scaling the PV area up or down.
The impact of the iteration processes is shown in figure 3, displaying one week of the modelled year. Before the iteration process, the first estimation (light blue line on figure 3) represents the hydrogen stored in this week and shows there are dips below zero. This means the demand cannot be met meet in that time period. After the iteration process, the PV panels are scaled up and the hydrogen stored over the week meets the demand at every time (dark blue graph line on Figure 3). (How the PV sizing tool work in detail can be seen ).
Consequently, this intelligent iterative search algorithm allows us to find the required PV area while still meeting 100% of the demand.
Toolset Output
The H2ool gives the following information as output data:
All key sizing and cost information is clearly provided by the H2ool which can be downloaded .
Figure 3: Hydrogen stored before and after iteration
Figure 2: Yearly demand profile of a community
Intelligent Iterative Search Algorithm
The iterative search algorithm has been developed to optimise the required PV area and storage size. After matching the average daily PV output to the day with the highest demand as a first estimate, the tool starts looking for the overall match between daily power required by the system and supplied power by the PV panels (which varies due to scaling the PV area up and down).
The storage size is determined by taking the total of the day with the highest requirement of hydrogen and adding an additional 10% for security; this value can then be scaled up to see the impact on the overall system size. If the storage is scaled up, the PV area declines to a certain point until it levels off.
To display this search procedure, the excel tool was modified with IF-statements and Goal Seek Analysis to ensure that the storage tank never goes below zero.
Model Improvement
Our system meets the load 100% and therefore technically works.
The system components of fuel cell, electrolyser and storage can all be analysed a step further in terms of sizing to reduce PV surplus. Fuel cell component is modular and thus can be sized modularly to accomodate differing demand sizes. This is similar for the electrolyser where a bigger/ smaller/ multiple electrolysers can be interchanged given the demand.
It was decided however that of these components, the storage was not being intelligently incorporated into the system. This analysis will look at an area of high irradiance used for comprehending scale and full system potential. Click for further examples of the tool in different climates.
Figure 4 shows over-production of hydrogen on days whereby the irradiance is much higher than from the original iterative search and for what has been designed for. It therefore needs to be determines how the storage can be utilised in order to avoid oversizing the PV. For the current sizing without any system improvements, the PV area would be equivalent to around 5.5 football pitches with a tank size equivalent to a whiskey lorry tank.
Going back to the and with the use of macros, changed the hydrogen store- in increment sizes of the original size, 142m^3 (2x, 4x etc). The sizing represents a full hydrogen store at the start of the system’s operation. The macro searches for the minimum amount of PV panels (per m2) required whereby the hydrogen tank will never be empty (zero). Figure 5 shows tank size and PV area have an inversely logarithmic relationship with PV array size reducing as the tank size increases.
Taking results from the optimum size of four times storage increase after the intelligent search algorithm, figure 6 shows a greater proportion of the hydrogen store being utilized throughout the year (blue area), and reduces the PV size by two football pitches (14,000 m2). This is what gives improved technical feasibility to the model and system. The next stage of investigation is whether the system is economically feasible ( ).
To find out more details about the iterative search taking PV area and storage into account scroll down).
Figure 4: Hydrogen Overproduction
Figure 6: Hydrogen Overproduction after utilising the storage size
Figure 5: Graph showing PV Area and Tank Size relationship following macro enabled intelligent search algorithm
References:
Barbier, F (2005). PEM electrolysis for production of hydrogen from renewable energy sources. Solar Energy 78, pp. 661–669