Professional Documents
Culture Documents
Farm DESIGN Manual
Farm DESIGN Manual
1. Introduction ..................................................................................................... 4
2. Installation and first use ................................................................................. 5
3. Concept ............................................................................................................. 6
4. Overview graphical user interface ................................................................. 8
4.1 Describe window ............................................................................................................ 8
4.2 Explain window ............................................................................................................ 11
4.3 Evaluate window .......................................................................................................... 13
4.4 Explore window ............................................................................................................ 14
4.5 Repository ..................................................................................................................... 15
4.6 Data from Excel and text files ...................................................................................... 17
4.7 Overview of keys and buttons ..................................................................................... 19
5. Parameter inputs windows (‘Magic windows’) .......................................... 20
5.1 Crop nitrogen fixation .................................................................................................. 20
5.2 Crop effective organic matter ...................................................................................... 20
5.3 Crop nitrogen fixation .................................................................................................. 20
5.4 Crop greenhouse gas emissions .................................................................................. 20
5.5 Animal whereabouts .................................................................................................... 20
5.6 Labour profiles.............................................................................................................. 20
6. Farm model calculations ............................................................................... 21
6.1 Module: Crop areas....................................................................................................... 21
6.2 Module: Product destinations ..................................................................................... 22
6.3 Module: Feed balance ................................................................................................... 22
6.4 Module: Organic matter balance ................................................................................. 28
6.5 Module: Manure production and breakdown ............................................................ 29
6.6 Module: nutrient flows and cycles .............................................................................. 36
6.6.1 Estimation of nitrogen fixation ............................................................................ 38
6.6.2 Estimation of nitrogen intake from pastures ...................................................... 38
6.7 Module: Labour ............................................................................................................. 39
6.7.1 Farm labour............................................................................................................ 39
6.7.2 Family labour ......................................................................................................... 40
6.8 Module: Greenhouse gases .......................................................................................... 41
6.9 Module: Bio-energy production .................................................................................. 42
6.10 Module: Water balance .............................................................................................. 45
6.11 Module: Human nutrition .......................................................................................... 45
6.11.1 Food diversity and sufficiency ........................................................................... 45
6.11.2 Nutritional Functional Diversity ........................................................................ 48
6.12 Module: Farm economics ........................................................................................... 50
6.13 Module: Household budget ........................................................................................ 53
6.14 Module: Flow metrics................................................................................................. 53
6.15 Checking the calculations .......................................................................................... 53
7. Multi-objective optimization ........................................................................ 55
7.1 Pareto-based optimization .......................................................................................... 55
7.2 Differential Evolution (DE) .......................................................................................... 56
7.3 Use of DE in Farm DESIGN ........................................................................................... 58
7.4 Setting decision variables in the Describe window ................................................... 59
7.5 Setting constraints in the Explain window................................................................. 60
If you would find points for improvement, please suggest them to us. We are always
looking for ways to improve our products. For questions concerning the Farm DESIGN
model or other feedback, please contact:
Groot JCJ, Oomen GJM, Rossing WAH 2012 Multi-objective optimization and design of
farming systems. Agricultural Systems 110, 63-77.
http://dx.doi.org/10.1016/j.agsy.2012.03.012
Cortez-Arriola J, Groot JCJ, Améndola Massiotti RD, Scholberg JMS, Mariscal Aguayo VD,
Tittonell P, Rossing WAH, 2014. Resource use efficiency and farm productivity
gaps of smallholder dairy farming in North-west Michoacán, Mexico. Agric Syst
126 15-24. http://dx.doi.org/10.1016/j.agsy.2013.11.001
Cortez-Arriola J, Groot JCJ, Rossing WAH, Scholberg JMS, Améndola RD, Tittonell P 2016
Alternative options for sustainable intensification of smallholder dairy farms in
North-West Michoacán , Mexico. Agric Syst 144 22–32.
http://dx.doi.org/10.1016/j.agsy.2016.02.001
Flores-Sánchez D, Groot JCJ, Lantinga EA, Kropff MJ, Rossing WAH, 2014. Options to
improve family income, labour input and soil organic matter balances by soil
management and maize-livestock interactions. Exploration of farm-specific
options for a region in Southwest Mexico. Renew Agric Food Syst 30(4), 373-391.
http://dx.doi.org/10.1017/S1742170514000106
Flores-Sanchez D, Kleine Koerkamp-Rabelista J, Navarro-Garza H, Lantinga EA, Groot JCJ,
Kropff MJ, Rossing WAH 2011 Diagnosis for ecological intensification of maize-
based smallholder farming systems in the Costa Chica, Mexico. Nutrient Cycling in
Agroecosystems, 91(2), 185–205. http://doi.org/10.1007/s10705-011-9455-z
Groot JCJ, Cortez-Arriola J, Rossing WAH, Améndola Massiotti R, Tittonell P 2016
Capturing Agroecosystem Vulnerability and Resilience. Sustainability, 8(11),
1206. http://doi.org/10.3390/su8111206
Mandryk M, Reidsma P, Kanellopoulos A, Groot JCJ, van Ittersum MK, 2014. The role of
farmers’ objectives in current farm practices and adaptation preferences: a case
study in Flevoland, the Netherlands. Reg Environ Change 14(4), 1463-1478.
http://dx.doi.org/10.1007/s10113-014-0589-9
In the Bin directory you can find the application FarmDESIGN.exe. For your convenience,
create a shortcut (click with right side mouse button on FarmDESIGN.exe; select Send To;
select Desktop (Create Shortcut)).
In the Docs directory you can find documentation about the model, such as the manual
and the release notes, which specify which changes were made to the model in each
version.
You can start the model by clicking on the shortcut. When you use the model for the first
time, you will be asked to select an input and output directory using a folder browser.
You are advised to select the Input and Output folders that have been created when you
extracted the ZIP file. These folders contain the necessary files.
If you select other input and output directories in a later stage, then make sure that the
input directory contains files named ‘FarmDB.mdb’ and ‘DE.mdb’.
Figure 3.1. Adapted learning cycle from Kolb (1983), extended with the DEED concept
presented by Tittonell (2007) and Giller et al. (2008).
The learning and adaptation cycle of the user of Farm DESIGN is supported by the user
interface, which has been designed to support the consecutive steps in the iterative cycle
(Figure 3.2).
Describe
Explain
Design
Explore Evaluate
Window 4 Window 3
Figure 3.2. The windows of the Farm DESIGN application.
The main actions that can be executed by the user from the four windows are listed in
Figure 3.3. More explanation of these actions is provided in the next chapter.
Describe/Design: Explain:
Explore: Evaluate:
Figure 3.3. The actions of the user in the windows of the Farm DESIGN application.
After selecting entities in the tree panel the details are displayed in the edit panel of the
window. These details can be edited. New entities can be added by typing Control-N, and
entities can be removed using Control-D. Please note that all editing is instantly
processed for direct updating of calculations in the ‘Explain’ and ‘Evaluate’ windows,
and therefore cannot be reversed (for deleting after confirmation). Moreover, deleting
For most of the entities the window appears as shown in Figure 4.1. There are
exceptions for crop fertilizers and pesticides, rotations, animal replacements (Figure
4.2). In these cases the edit panel consists of a list of entities that can be selected (on the
left, in grey) and a list of already selected entities (on the right, in yellow). Entities can
be selected and deselected by clicking on the name in either one of the two sub-panels.
After selecting a crop into a rotation (Figure 4.2a) you can specify the crop area. After
specifying animal replacements (Figure 4.2b) the number of the animals for replacement
(e.g. calves) will be recalculated on the basis of the animal number of the main animal
type (for instance milk cow) and its replacement rate; note that the actual farm data
will be overwritten in this case!!!
Figure 4.2. The ‘Describe’ window with data overview in the tree panel on the left-hand
side, and the edit panel for animal replacements on the right-hand side.
The names of the farm entities can be edited directly in the tree in the left-hand side tree
panel. Other data can be edited in the edit panel. Most data items have a white
background, only items that can serve as decision variables in the optimization are
displayed with a light blue background. After activating a candidate decision variable
after double clicking on the data item field and specifying the upper and lower limits for
the allowed values range (Figure 4.3), the background is changed to green. Thus,
candidate (blue) and active (green) decision variables can be readily identified from the
edit panel. Moreover, a complete overview is provided in the ‘Evaluate’ window.
The values in the ‘Describe’ window will be blocked for adjustments by the user after the
start of the explore step.
Figure 4.4. The ‘Explain’ window. Data items or variables that are candidates to act as
objectives or constraints have a light blue background; activated variables have a green
background.
Figure 4.5. Window to edit objectives and constraint variables. It is displayed after
double clicking on a light blue data item on a tab page of the ‘Explain’ window.
Most data items in the Explain window have a white background, only items that can
serve as objective or that can be constrained in the optimization are displayed with a
light blue background. After activating a candidate variable by double clicking on the
data item field and specifying the direction of optimization fore objectives and/or the
minimum and maximum allowed values for constraint variables (Figure 4.6), the
background is changed to green. Thus, candidate (blue) and active (green) constraints
and objectives can be readily identified from the edit panel. Moreover, a complete
overview is provided in the ‘Evaluate’ window.
Figure 4.6. Window to edit objectives and constraint variables. It is displayed after
double clicking on a light blue data item on a tab page of the ‘Explain’ window.
Initially, this window contains two tab pages, the first named ‘Variables’ can be used to
verify whether the value of the decision variables within the current farm set-up are
within the allowed range. The second tab page named ‘Outcomes’ lists the objectives
that have been specified in the ‘Explain’ window, and the actual value and allowed range
of the outcomes that are indicated to serve as constraints. If the value of decision
variables or constraint variables is outside the indicated range colour of the actual value
will change to green, to indicate to the user that adjustments in the input data are
needed. As long as not all values are within the allowed ranges, the user cannot start
with the exploration: after clicking on the ‘Explore’ button on the ‘Describe’ window, the
‘Explore’ window will not open and an information message will be shown to the user.
After running an optimization in the Explore window, the individual solutions can be
reviewed in detail. After clicking the solutions in the graph of the Explore window, two
new tabs will appear, named ‘View variables’ and ‘View outcomes’. In these tabs the
values of the decision variables and the outcomes (objectives/constraints) will be listed,
to be compared. The maximum number of solutions that can be compared is 10. The tabs
can be emptied by pressing the ‘Clean’ button ( ). If these tabs are active, their contents
will also be written to the Excel spreadsheet [Farm name].Evaluate.xls or comma
separated file (.csv) in the output directory.
Figure 4.7. The ‘Evaluate’ window. Values that violate constraints are highlighted in
green.
After activating the ‘Explore’ window the data in the ‘Description’ window are frozen to
block further changes to the data. The original ‘Description’ window is renamed to
‘Design’. It can be enabled only after saving a farm configuration (see below) or selecting
a different farm (via menu options Farm/Select in the ‘Describe’ window).
The optimization can be started by clicking on the ‘Run’ button and will proceed for the
number of iterations indicated in the adjacent text box. The progress of the optimization,
i.e. the number of iterations elapsed, is indicated in the progress bar at the bottom of the
window.
Before you start the optimization you can adjust the parameters of the evolutionary
algorithm (Differential Evolution, button: ‘DE parameters’), for explanation of this
algorithm and its parameters see Section 6. Please note that these parameters cannot be
adjusted anymore after the start of the first optimization round, i.e. the first click on the
‘Run’ button.
Before and after each optimization round you can adjust the X and Y axes to be displayed
(‘X axis’ and ‘Y axis’ buttons), and the naming and scaling of each of the axes (‘Edit axis’
button).
After each optimization round the symbols representing the generated alternatives are
transformed into buttons, the symbols will change color to green (■; Figure 4.9). These
symbols can be clicked one-by-one, and the associated input and output data will be
shown in the ‘Design’ and ‘Explain’ windows. Thus, the generated alternatives can be
evaluated for their configurations (crop areas, product destinations, animal numbers,
etc.), and the resulting outcomes. Selected new configurations can be saved to the
database using the menu option Farm/Save as… in the ‘Design’ window. Subsequently,
after selecting this farm from the database using the menu option Farm/Select,
additional adjustments can be made to the new farm configuration and a new round of
explorative design can commence.
4.5 Repository
Data can be transferred to a repository for later use for other farms. This applies to
component types: crops, animals, animal products, manures, fertilizers, buildings and
machines. Crop products can be added to the repository only as part of a crop. A
component can be added by selecting ‘Copy to repository’ (Control-C) in the context
menu of the tree view of the Describe window. The window displayed in Figure 4.10 will
be opened. Information about the location, context and other details should be entered,
these are added to the note that will be associated to the component in the repository.
The repository can be opened from the main menu, using the icon, or from the
context menu of the tree view of the Describe window by selecting ‘Add from repository’
(Control-Z). The window shown in Figure 4.11 will be opened, and components can be
searched and selected by type. After pushing the ‘Add’ button the component is instantly
added to the farm. Components can also be removed from the repository (also photos
and notes will be deleted, irreversible after confirmation has been given.
Repository data can be exchanged between databases through data files. The repository
data can be written to such a file from the main menu – File – Export – Repository option
(Control-Shift-R). The file ‘Repository.Data.dat’ containing the data is then written to the
output directory. This file can be imported into another database again from the main
menu – File – Import – Repository option (Control-Shift-Z).
Figure 4.10. Window for entering text data copied from Excel.
Following Oyarzun et al. (2013), the Margalef index (M) is computed as an indicator of
on-farm ecological “richness” species, based on the following formula:
𝑆−1
𝑀=
ln(𝑁)
Where S is the number of crop species on the farm (with crop area > 0 ha) and N is the
farm area expressed in m2. The Margalef index increases when there are more species in
a determined farm area or when the same amount is maintained but the area of the farm
decreases (Oyarzun et al., 2013).
𝐻 = − ∑(𝑝𝑖 ln(𝑝𝑖 ))
𝑖=1
In this equation, pi is the area proportion of crop i. H = 0 if there is only one species in
the farm and maximum when each species occupies the same area in the farm. Thus, a
monoculture or situations where a few crops occupy large areas in relation to the total
size of a farm result in a low value for the Shannon index (Oyarzun et al., 2013).
The total farm or rotation cropping area, the crop group frequency and the species
richness/evenness indicators can be selected to serve as objectives or constraints.
Figure 6.2. Screenshot of the Explain window showing an overview of the balance
between production and use of crop (kg dry matter) and animal (kg fresh matter)
products, and the bedding balance (at the bottom of the screen).
Figure 6.2. Window to specify the various aspects of the feed evaluation system used for
a particular farm.
The evaluation system and units used can be specified in the ‘Feed evaluation system’
window that can be opened from the menu through the Settings – Calculations – Feed
evaluation option in the Describe window (Figure 6.2). For each category a choice
should be made, and it is the responsibility of the user to consistently fill in the
associated data in the correct units. When data from the repository defined in another
feed evaluation system are imported into the farm, a warning message for the user is
issued, but also in this case the user has the responsibility to correct the data for energy,
protein and dry matter intake.
The table below shows the options, the related equation to calculate dry matter
digestibility of feeds, and indicates the default system (*). The abbreviations are
displayed for the farm, and the. This information will be displayed in the ‘Feed
evaluation’ tab of crop and animal products and in the ‘Requirements’ tab of animals in
one composite line, e.g. ‘SV; NE (VEM); DVE’.
For protein:
DVE (g)* Intestinally degradable
protein
CP (g) Crude protein
The length of the period (days) that animals are present and are being fed and produce
manure on the farm (‘Managed by the farmer’) can be specified per animal type using
the ‘Whereabouts’ window (click the pink textbox to open). This period can be shorter
than 365 days for instance when animals are only kept for fattening during the growing
season or when animals spend part of the year off-farm on communal lands. Within this
period that the animals spend on the farm, two sub-periods are distinguished for animal
management: the grazing period and the stable (non-grazing) period. The duration of
these periods can be adjusted by the parameter DurationGrazingPeriod, which can be
specified in the ‘Animal grazing period’ window (double-click the pink textbox; see
Figure 6.3 in Section 6.5). In arid regions a distinction between wet and dry seasons is
more relevant and can be implemented using this approach. The length of the non-
grazing period is calculated as 365 – DurationGrazingPeriod.
In the feed balance calculations can be based on any feed evaluation system. In
FarmDESIGN the deviation between the requirements of the whole animal herd and the
feed available from crop production or purchase of feeds is determined for:
• Energy, requirement for animal maintenance, milk production and growth is
expressed in energy units per animal per day, and the availability in feed is
expressed in energy units per kg dry matter.
• Protein, requirement is expressed in protein amount per day. Protein is required
for animal maintenance, milk production and growth (g/animal/day) and the
availability in feed is expressed in g protein/kg dry matter.
• Structural material needed to stimulate the rumen function. Mostly abstract units
related to a given standard are used, representing a relative value set to
requirement for a ‘standard animal’. For other animal types corrections are
needed. Each feed has a different contribution to the structure value, and a
Energy and protein is needed for body maintenance, growth, pregnancy and milk
production. The requirements are calculated with generic equations for cows, sheep and
goats but the proportionality factors in the equations differ between the animal types
(see Table 4.1). Farm DESIGN using calculations specific for the Dutch VEM/DVE
system if these two feed evaluation systems are selected in the feed evaluation
selection system (Figure 6.2). Otherwise, generic calculations are used. Both ways
of calculating animal requirements are explained below.
Animal feed requirements are calculated following the Dutch feed evaluation system,
consisting of the variables:
- VEM = voedereenheid melk, a feed unit milk (FUM) is equivalent to 6.9 kJ net energy
for lactation
- DVE = darmverteerbaar eiwit, protein digestible in the intestines (IDP)
For background information on VEM, see: Van Es (1975; 1978). Energy evaluation
systems for dairy cattle are compared in Vermorel and Coulon (1998) and Yan et al.
(2003). A detailed description of the Dutch DVE/OEB protein evaluation system can be
found in Tamminga et al (1994), and a comparison of protein evaluation systems in
Tuori et al. (1998).
The energy requirement for maintenance and growth is proportional to the weight of
metabolically active tissue in the body, which is calculated as:
MW LW0.75
Where:
MW metabolic weight (kg)
LW live weight (kg)
The energy requirement for maintenance is also dependent on the activity of the animal.
Grazing animals require 15-20% more net energy for maintenance (CVB, 2008, p. 10, 13,
28, 32). We assume a value of 20% for the time spent grazing, which is averaged over
the year. The proportion of time spent grazing is dependent on the number of grazing
days per year and the grazing system (determined by LengthGrazingPeriod and
TimeInPasture parameters), e.g. zero grazing, day grazing or day-and-night grazing.
Total MAINTENANCE energy requirements (ERM for generic calculations and VRM for
the VEM system) are calculated as:
ER M E M MW A (energy generic)
Where:
EM energy needed per kg MW (energy units/kg), parameter EnergyMaint
A activity factor (dimensionless), 0.2 (20%) extra energy needed for activity during
grazing. The value of the factor ranges from 1.0-1.2, and is calculated as: 1 +
0.2*[proportion of time spent grazing].
C correction factor for milk production, see below.
For all animal types the protein requirement for maintenance (PRM for generic
calculations and DRM for the DVE system) is calculated with the following equation (CVB,
2008, p. 9):
PR M PM MW (protein generic)
The amounts of energy and protein needed for GROWTH are calculated using the
equations below, where ERG and PRG are the requirements according to generic
calculations. In the VEM/DVE system, (VRG and DRG) are dependent on the growth rate
and the MW of the animal (NRC, 2001, p. 237; CVB, 2008, p. 13):
ER G E G LG
(energy generic)
VR G 0.05 MW LG
(VEM system)
PR G PG LG
(protein generic)
DR G 0.005 MW LG
(DVE system)
Where:
EG energy needed for growth (VEM/g), parameter EnergyGrowth
PG protein needed for growth (DVE/g), parameter ProteinGrowth
LG live weight gain (g/day)
Both pregnancy and milk production requirements are considered on an annual basis.
As a consequence, the energy and protein needs are averaged over a whole year of 365
days, and not specified per gestation or lactation period. For pregnancy, the body
growth is calculated and energy and protein requirements are attributed to this growth.
The weight of calves is between 35 and 45 kg (on average 43 kg for dairy cattle HF)
(Handboek Melkveehouderij, 2009, p. 6-2). Thus, we can assume an average growth of
Animal requirements for MILK PRODUCTION are related to total milk production in the
generic system, and to the amount of fat and protein corrected milk (FPCM) and milk
protein (MP) produced in the VEM/DVE system (CVB, 2008, p. 8):
ER P M E P
(energy generic)
VR P FPCM E P C
(VEM system)
PR P M PP
(protein generic)
DR P 1.396 MP 0.000196 MP 2
(DVE system)
Where:
mf milk fat content (%), parameter MilkFat
mp milk protein content (%), parameter MilkProtein
M actual milk production (kg/d), parameter MilkProduction
EP energy needed for milk production (VEM/kg FPCM), parameter EnergyMilkProd
PP protein needed for milk production (g DVE/kg FPCM), parameter
ProteinMilkProd
C correction factor for milk production = (1 + (FPCM-15)*0.00165)
MP milk protein production = M * mp * 10
Table 6.1. Parameter values for tshe generic equations to calculate feed requirements of
cows, sheep and goats. The names of the parameters (CamelBoldItalics) are the same as
used in the Farm DESIGN model.
Parameter Name in the model Cow Sheep Goat
EM (VEM/kg MW) EnergyMaintenance 42.4 30.0 36.4
Figure XXX. Screenshot of the Explain window showing the organic matter balance of the
farm.
The balance is calculated as the difference between organic matter (OM) accumulation
and loss. Organic matter accumulated on the farm is partly degraded. The rate of
degradation is affected by the environment variables of soil texture (TextureFactor),
moisture availability (MoistPeriodPf35) and the average temperature
(MeanTemperature). On the basis of these variables a correction factor f is calculated:
The organic matter input from crop residues (OMCR) depends on the effective organic
matter input per crop (EOMc) and crop area (Ac), and is summed for all crops c:
n
OMCR A c * EOMc * f
c 1
The organic matter input from green manures (OMGM) is calculated from the dry matter
of crop products with destination ToSoil. and the organic matter content, derived from
the AshContent in dry matter. It is assumed that 80% of organic matter is degraded
during the year after application, so that 20% (fraction 0.2) remains.
n
OMGM ToSoilp * (100 AshContentp ) / 100 * 0.2 * f
p 1
The residual amount of organic matter from manure produced on the farm (OMMN) is
calculated as described in Section 6.5. Organic matter in imported manure (OMIM) is
derived from the Amount of manure imported and the OMContent of the manure.
OM balance OM CR OM GM OM FL OM MN OM IM OM SD OM ER
n
DMToManure ToAnimalsp *(1 FeedLossp / 100) * (1 DMDp / 1000)
p 1
Similarly, the amounts of OM and C in manure are calculated using the AshContent of
the crop and animal products and assuming a C content in organic matter of 45%
(fraction 0.45). The total amounts of N, P and K in manure (in both urine and feces) is
derived from the difference between intake and products of the animals, as
demonstrated only for N below.
The manure produced on the farm can be deposited at three sites: in the pasture (P), in
the stable (S) and in the yard (Y). A yard or feedlot is an outdoors confined area, often
close to the stable, were animals are kept during a limited time of the day, for instance
before and/or after milking. The excretion at the different sites is assumed to be
proportional to the time spent at each site, which is defined per animal type with the
parameter Whereabouts that defines the time spent in stable, pasture, yard and off-
farm, the total should be 365 x 24 = 8760 hours. For instance, the string
“4000|42000|560|0” for Whereabouts implies that the animals spend 4000 hours in
stable, 4200 in pasture, 560 hours in the yard and no time off-farm. You can specify
these time allocations using the ‘magic window’ that opens when you double-click on the
pink text-box of the Whereabouts parameter, or manually directly in the pink
parameter textbox. The time spent in the barn is calculated as: 24 – time in pasture –
time in yard – time off-farm, per day. The proportions of the time spent at each site
during the year are used to estimate the amount of manure produced at the different
sites.
The site were the manure is produced is indicated the field ProducedAtSite with a letter
P, Y or S. Per site one or more types of manure can be produced, depending on storage
and treatment of the manure. The most apparent example is the production of both
slurry and farm-yard-manure in a stable. The proportions will differ depending on
housing type etc., and can be specified in the variable ProportionAtSite of manures. Per
site the cumulative value of ProportionAtSite for the manures should be 1.
The total amounts of faeces N and urine N produced are calculated with the following
equations. The degradability of crude protein (DCP, expressed in g/100 g, crude protein
content CP = 6.25 N in %) in the ration is derived from an empirical regression equation
derived by Holter and Reid (1959, p. 1345, Fig. 1).
Note that these calculations take the animal intake of DM and N of the whole herd, and
no distinction is made per animal type, production level or age group!
Using the equation below mineral N is lost with a given fraction, as defined by the value
of ExcretionLoss of the manure.
The mineral N amount remaining after excretion can then be derived by integration:
During grazing, about 5 -10 % of urine does not reach the soil surface, but adheres to
grass leaves. All nitrogen in this part is lost by volatilisation. Most urine (90 – 95%) is
absorbed by soil and after microbial decomposition of urea into ammonium the latter is
adsorbed by soil and slowly transformed to nitrate by bacteria. If excreted in early
spring nearly all urine-N will be re-used in the same year and if excreted in autumn
nearly all urine-N will be lost after nitrification during winter, by leaching and
denitrification. In the model these losses by leaching plus denitrification from soil are
included in the balancing item (as soil losses, flow 13 in Figure 6.2 and Table 6.1). The
high nitrogen availability in the form of nitrate in urine affected patches will result in
reduced nitrogen fixation by clover in these patches. Only a minor part of all nitrogen in
faeces excreted during grazing is lost by volatilisation (about 5%). The remaining
organic nitrogen is mineralised slowly and efficiently used by grasses.
In the stable, in deep litter stables total manure is usually separated in a liquid (slurry:
urine with faeces) and a solid (straw with faeces and urine) fraction. In the current deep
litter stables with a closed or slatted concrete floor behind the feeding rack, 30-40% of
faeces and urine are excreted on the concrete floor (ProportionAtSite=0.3-0.4) and are
collected as slurry while 60-70% is deposited in the deep litter part and collected as
deep litter manure (also called farm yard manure; ProportionAtSite=0.6-0.7) (Oenema
et al., 2000). In a similar way the contents of organic matter, P, K, organic N and
inorganic N (urea) are calculated. In tying stables and stables with cubicles where straw
is added before the liquid and solid parts are separated, the percentage of urine
absorbed by straw depends on urine production per animal (litres), straw application
per animal and how much urine is drained before it can be absorbed by straw. We do
not go into details about how to estimate and to check these values. During the first day
after excretion about 5% of urea-N is lost from FYM (ExcretionLoss=0.05) and about
10% from slurry (ExcretionLoss=0.10). In the deep litter part most of the urine is
absorbed by the underlying FYM, but a small parts adheres to the straw at the surface.
Storage - During storage of manure part of the organic matter is degraded, nitrogen can
be immobilized or mineralized and partly lost. The loss of organic matter depends on
temperature and duration of storage. As most of slurry is stored during the cold season
(in temperate regions), loss of organic matter is usually moderate: about 10%. The
composition of slurry hardly changes during storage. A part of organic matter is lost as
CO2 and another part as CH4, a strong greenhouse gas. Slurry can also be used to
produce methane in digesters (higher temperature, stirring and mixing) and in that case
more organic matter is lost (about 40%) and nearly all nitrogen is preserved.
Nitrogen losses remain low, if slurry is stored in a really closed tank like a digester
(about 1%) In practical agriculture, slurry is stored for about 4 months in a nearly
closed tank. During winter, loss of nitrogen is a 1 to 15% of inorganic nitrogen in slurry:
0.2 % per month in a closely sealed tank, 1% per month in a rather well closed tank, 2%
in case of a pit under a slatted floor and 4% per month in a open pit, where wind has
access to. In spring, summer and autumn losses are higher, but during that period less
manure is stored. In the model calculations you have to give one figure (as %) for
inorganic/mineral nitrogen that will be lost during storage and you can start with 10%
of mineral nitrogen in case of a rather well closed tank, with 20% in case of a pit under a
slatted floor and with 40% in case of an open pit (manure covered by a crust). If
measurements are available you can improve your estimate.
The loss of organic matter during storage of farm yard manure (FYM) depends on the
amount of straw used for bedding, compaction / aeration of manure and duration of
storage. As a first step you have to estimate the percentage of manure that is
decomposed aerobically (in a small heap nearly 100%, FractionOxic=1.0) and/or
anaerobically (in a compacted big heap about 80%, FractionOxic=0.2). During
decomposition temperature rises (it peaks at 75 ºC in aerobic decomposing deep litter
manure) and most of produced heat is lost by evaporation of water (about 7 kg water
per kg of lost organic matter). If this loss of water exceeds the initial water content per
kg organic matter manure will dry out and become mouldy (as often happens in a loose
heap of horse manure). A part of the inorganic nitrogen in FYM is lost by volatilisation of
NH3 and another part by denitrification. Loss by NH3 volatilisation occurs during the
first weeks of storage and denitrification during the remaining period. Total loss of
nitrogen depends on the size and form of the heap, its exposition to wind and rain and
initial C to N ratio of manure (including inorganic forms).
The degradation and mineralization processes take place in aerobic (oxic, OX) or
anaerobic (anoxic, AN) conditions, and the proportional division between these
conditions is specified per manure by the value of FractionOxic of the manure. For both
conditions four process parameters should be specified:
- Organic matter degradation rate (fraction per whole storage period), i.e.
DegradationRateOxic and DegradationRateAnoxic. These values are the apparent
degradation rate, because from simultaneously organic matter in microbial biomass
is produced with a given efficiency (MicrobialEffOxic and MicrobialEffAnoxic), so
that a correction with 1/(1-MicrobialEff) is needed to derive the true organic matter
The last equation represents the balance between nitrogen release from degradation
and the incorporation of N into organic matter of microbial biomass. The resulting net N
release can take negative (mineralization) and positive (immobilization) values.
This last equation can be understood as follows: micro-organisms break down the
organic matter with a certain ('observable' or apparent) rate k. However, because micro-
organisms grow due to this degradation process with growth efficiency, this results in
an addition to the organic matter, so the observed degradation rate should be corrected
for their growth efficiency, and the true degradation rate is 0.5*OMDegr /(1-
MicrobialEfficiency).
The degradation of the organic matter is associated with release of nitrogen, determined
by the C:N ratio of the substrate (CManure/NManure). The micro-organisms will
incorporate part or all of this nitrogen, dependent on their C:N ratio (MicrobialCN). In
fact, when the C:N ratio of the substrate is high, the release of nitrogen from organic
matter is lower than the incorporation into microbial biomass, and as a consequence
mineral N from the soil solution may be incorporated. This results in negative values for
N release in the equation above, indicating net immobilization. This applies to both the
two processes of aerobic and anaerobic breakdown of manures, whereas different
parameter values can be used.
From manure both in aerobic and anaerobic conditions N can be lost, governed by
MineralNLossOxic and MineralNLossAnoxic:
The amounts of DM, OM, total N, mineral N and organic N, P and K remaining after
storage are calculated by integration:
Application - When manure is applied to the field again a fraction of mineral N can be
lost as determined by the ApplicationLoss:
Again, the mineral N amount remaining after application and thus actually available for
plant uptake can be derived by integration:
The following table is based on Mulder & Huijsmans (1994) and gives estimates for
losses during application without loss-reducing measures (Table 6.2).
Table 6.2. Loss of ammonia in manure during above-ground application without loss-
reducing measures.
Manure type Grassland Arable land
Slurry 76 % (43 % - 100 %) 71 % (37 % - 100 %)
Urine 60 % 60 %
Farm Yard Manure 70 % 71 %
Volatilisation of ammonia during the first day after above-ground application (without
loss reducing measures) depends very much on the weather (no wind, cool
temperatures, rainy conditions reduce losses; strong wind, high temperatures and dry
conditions increase losses). The quicker and the more completely manure is worked into
the soil, the lower application losses. This phenomenon is accounted for by using the
reduction factors as shown in Table 6.3.
Table 6.3. Fraction of loss compared to aboveground application without loss reducing
manures (here used as reference).
Loss reducing measure Fraction loss compared to reference
Slit injection 0.05
Ploughing <10 min after application 0.1
Harrowing or cultivating (weather and soil 0.1-0.5
structure dependent)
Dilution with water (at least 1:3) or by rain / 0.3-0.5
irrigation
Total losses and degradation - The accumulated losses of mineral N can be derived
from adding the losses during excretion, storage and application for each manure type:
After application to the field the manure organic matter will be further degraded. The
fraction of OM remaining after one year is given by the HumificationCoeff of the manure,
which is affected by environmental conditions (factor f, see Section 6.3). The total
degradation of organic matter during storage and after application can thus be
calculated for each manure type as follows:
a. b.
Figure 6.5. Overview of nutrient flows quantified for C (a.) and nutrients N, P and K (b.).
Explanation of the numbered flows in Table 6.1.
Table 6.2. Nutrient flows for carbon (C) and nutrients nitrogen, phosphorus and potassium
(NPK) in the model. The flow numbers refer to Figure 6.2. The existence of the flows is
indicated (x=yes, o=no) for individual nutrients in the CNPK column.
Flow CNPK Description
1 x x x x Inputs of carbon and nutrients in crop products from outside the farm,
calculated from the difference of use and on-farm production
2 x x x x Net assimilation (harvested or in effective organic matter) for C and
nutrient uptake in harvested crop products for NPK
3 o x o o Symbiotic nitrogen fixation from NFixation of crops
4 o x x x Atmospheric deposition of nutrients from NDeposition, PDeposition
and KDeposition in environment
5 o x o o Non-symbiotic nitrogen fixation by free living soil microorganisms from
The (farm) balance is calculated as: Input – Output, and the efficiencies are calculated as:
Output/Input. Below the flows that are included in the calculations of input and output
for the whole farm/household and per component are listed.
Farm/household
Input: crop product import + animal product import for HH + fertilizer import +
symbiotic fixation + non-symbiotic fixation + deposition
Output: crop product export + animal product export + manure export + household
manure export
Crops
Input: manure and fertilizer application + symbiotic fixation + non-symbiotic fixation +
deposition
Output: crop products export + crop products to household + crop products to animals +
crop products to manure – crop products import
Manure
Input: animal manure production + fertilizer import + crop products to manure
Output: manure applied to soil + manure export
In grass clover mixtures actual nitrogen fixation by clover is close to potential N fixation
(i.e. 6 kg N per 100 kg DM of harvested clover) if manure application is moderate. In
urine patches clover will not fix nitrogen. The clover dry matter yield can be estimated
visually based on the estimated soil coverage: (% clover in dry matter = % clover visual
* 0.7 ) and
The dry matter yield of a sward can be estimated using the feed balance. Based on the
estimated feeding quality of the grass-clover mixture and taking into account the
amount and quality of additionally supplemented feed, the feed balance in terms of VEM
should be 0. The only item missing then is the amount of grass-clover taken up in the
pasture.
Urea production of the lactating herd (kg day-1) can be calculated from total milk
production and urea content of milk.
The time available to provide (regular or casual) family labour can be specified for each
household member. In addition, contract work can be performed, which is calculated at
crop level (ContractWorkCosts) and included in the calculations of the gross margin.
Contract work is not included in the labour balance (see Section 6.12 Farm economics).
At the farm level, for both regular and casual labour, we consider that the labour
requirement (L) should be provided by family (F) and hired (H) labour, i.e. the
relationship L = F + H. The family labour input should be smaller or equal to their total
available working hours (T). The combination of these relationships can be formalized
as:
𝐿 = 𝐹 + 𝐻
{
𝑇≥𝐹
→ T ≥ (L − H)
→ T– (L − H) ≥ 0
L is calculated as the sum of farm labour (LF), crop labour (LC) and animal labour (LA).
A fixed amount of labour LF is needed for general management of the farm
(LabourFarmConstant, in economics variables). This labour requirement will be only
partly dependent on the farm set-up, but cannot be attributed directly to work on crops
or animals.
n
LCregular A c * Re gularLabourc
c 1
Labour requirement related to animal husbandry (LA) can also comprise both regular
and casual labour. A fixed amount of regular labour is required for herd and stable
management (LabourHerdConstant, in economics variables) and a part of the labour is
dependent on the animal type and the number of animals, using the animal variable
RegularLabour.
n
LAregular = LabourHerdConstant + å Numbera * Re gularLaboura
a=1
n
LAcasual = å Numbera *CasualLaboura
a=1
Total required family labour input is calculated as the difference between total required
labour and hired labour:
n
Tregular = å Re gularLabourInput p
p=1
n
Tcasual = åCasualLabourInput p
p=1
The values of the emission factors for fixation, deposition and residue application
(EFNFixation, EFNDeposition and EFResidues) can be set from the main menu in the Describe
window, under Settings / Calculations / Greenhouse gases. For the crops, animals,
manures and fertilizers the emission factors can be found with each of the components
in the Describe window.
- Crop management (from soils under anaerobic conditions, in particular paddy rice)
o Crop CH4 emission: Area * EFCH4Enteric (for each crop type)
- Animals
o Enteric CH4 emission: Number * EFCH4Enteric (for each animal type)
o Manure CH4 emission: Number * EFCH4Manure (for each animal type)
- Manures
o Direct (prod): N produced x N2O emission factor * 44/28 (for each manure)
o Direct (appl): N applied x N2O emission factor * 44/28 (for each manure)
o Volatilization: N volatilized x N2O emission factor * 44/28 (for each manure)
- Fertilizers
o Direct: N applied x N2O emission factor * 44/28 (for each fertilizer)
Figure XXX. Screenshot of the Explain window of FarmDESIGN showing greenhouse gas
emissions.
Figure 6.6. Crop residues and livestock manures on farm. Components in blue boxes are
The on-farm produced energy from anaerobic digestion can be calculated as:
Where:
EPf = Energy produced on farm from the biogas yield (MJ year-1)
MYy = Annual CH4 yield from fermentation (m3 CH4 year-1)
LHVCH4 = Lower heating value of CH4 (MJ m-3)
To calculate methane yield, digestate production and its chemical characteristics, the
mass balance approach is used. Biogas production (CH4+CO2) is calculated from organic
matter degradation, relative to total carbon in dry matter. The amount of CH4 obtained
can be derived using equation 2d.
Where:
MYy = Annual CH4 yield (m3 CH4 year-1)
SubstrateC = Total mass of carbon in substrate (kg)
ResidueC = Mass of carbon in digestate (kg)
DegradedC = Degraded carbon from substrate (kg)
pCH4 = Proportion of CH4 in the biogas (kg/kg).
16/12 = Molecular weight of CH4 (g mol-1)
22.4/16 = Conversion factor: as 1 mol gas at Standard Temperature and Pressure
(STP), i.e. 22.4 L and 16 g CH4 = 22.4 L. MYy divided 16 will yield the moles
of CH4 produced annually which can be multiplied by 22.4 to obtain the
MYy in litres and then converted in m3.
Where:
ReleaseN = Release of nitrogen, either positive (mineralization) or negative
(immobilization) (kg)
SubstrateInorgN = Inorganic nitrogen in the substrate (kg)
DegradedN = Nitrogen in degraded substrate (kg)
MicrobialEff = Efficiency of microbial biomass synthesis (kg/kg)
MicrobialCN = C:N ratio of microbial biomass (kg/kg)
DigestionLossN = Loss of inorganic nitrogen during anaerobic digestion (kg)
NLossDigestion = Fraction of inorganic nitrogen lost during the digestion process
(kg/kg)
ApplicationLossN = Loss of inorganic nitrogen after application of the digestate to the
field (kg)
NLossApplication = Fraction of inorganic nitrogen lost from the digestate after
application to the field (kg/kg)
Equation 3 represents the balance between nitrogen release from degradation and the
incorporation of N into organic matter of microbial biomass. The resulting net N release
can take negative (mineralization) and positive (immobilization) values.
This last equation can be understood as follows: microorganisms break down the
organic carbon with a certain ('observable' or apparent) rate (DegradationRateEP)
leading to breakdown DegradedC. However, because micro-organisms grow due to this
degradation process with a given growth efficiency (MicrobialEff), this results in an
addition to the organic matter, so the observed degradation rate should be corrected for
their growth efficiency, and the true degradation rate is DegradedC /(1-
MicrobialEfficiency).
The degradation of the organic matter is associated with release of nitrogen, determined
by the C:N ratio (DegradedC/DegradedN). The micro-organisms will incorporate part or
all of this nitrogen, dependent on their C:N ratio (MicrobialCN). In fact, when the C:N
ratio of the substrate is high, the release of nitrogen from organic matter is lower than
the incorporation into microbial biomass, and as a consequence mineral N from the soil
solution may be incorporated. This results in negative values for N release in the
equation above, indicating net immobilization.
The nitrogen in crop residues is assumed to be completely in organic form and not
prone to losses during digestion. The nitrogen in manures is partly in inorganic form,
dependent on the ratio of nitrogen excreted in urine (inorganic) and in dung (organic).
During the fermentation process and after application of the digestate to the field, a part
of the inorganic nitrogen will be lost, as defined by parameters NLossDigestion and
NLossFermentation.
We assume that no conversions or losses of other nutrients than nitrogen occur during
anaerobic fermentation. The flows of nitrogen, phosphorus and potassium are
quantified in accordance with nutrient flow calculations (see Section 6.6).
Nutrition indicators are displayed in the ‘Nutrition’ tab page of the ‘Explain’ window.
Figure XXX. Screenshot of the Explain window of FarmDESIGN showing the food group
diversity and nutritional functional diversity of produced and consumed foods.
Dietary diversity scores are proxy indicators that provide qualitative measures of food
consumption reflecting micronutrient adequacy of the diet (Kennedy et al., 2010;
Kennedy et al. in this volume). Individual dietary diversity indicators often focus
particularly on women and young children as these groups are among the most
vulnerable to malnutrition. The minimum dietary diversity score for women (MDD-W;
FANTA, 2014) is the most recent indicator (Figure XXX). This indicator classifies foods
into ten food groups of which at least five should be consumed by women to increase the
likelihood of meeting their micronutrient needs compared to women consuming foods
from fewer food groups (FANTA, 2014). Food groups listed in Table 6.3 have been
defined to allow calculations for dietary diversity indicators. The group indicated with
‘None’ is used for crop or animal products that do not serve as food, such as animal feeds,
crop residues or wool.
Table 6.3. Food groups defined within the dietary diversity questionnaire and aggregation
of groups for HDDS and WDDS calculations. Copied from tables on pages 8 (FAO food
groups) and 24 of Kennedy et al. (2010)’; FAO+1 indicates splitting the original ‘Legumes,
nuts and seeds’ group. Examples for food groups can be found in the Annex 2, page 37, of
Ibid.
Figure 6.7. Window for entering dietary demands in terms of food groups (a.) or
nutrients (b.). The window can be opened from the menu in the Describe window, by
choosing ‘Settings/Calculations/Human nutrition’. Only the nutrients that are ticked are
considered and displayed in the ‘Nutrition’ tab of the Explain window.
Nutritional Functional Diversity (Remans et al., 2011 Petchey & Gaston, 2002) can be
determined at the level of the farm and household (for produced and consumed foods,
respectively), whereas a similar landscape-related indicator has been implemented in
the Landscape IMAGES model.
The nutritional functional diversity (NFD) indicator is calculated as the fraction of the
available diversity that is actually either produced on the farm or consumed by the
household. This fraction is derived from a dendrogram that represents the diversity
present in the ‘landscape’ in terms of relevant traits, which are the nutrient
concentrations of foods for the calculation of NFD, see Figure 6.7a for an example. The
landscape can include the farm itself, the surrounding landscape and the markets from
which the foods are sourced. The dendrogram has to be mapped to the crop and animal
products that have been defined in FarmDESIGN. The algorithm requires three input
files that should be located in the input folder:
[farm name].FD.products contains a list of products that are included in the dendrogram.
Mote that the order of the species should be the same as in the dendrogram file.
a. b.
Figure 6.7. (a.) An example dendrogram for calculation of the Nutritional Functional
Diversity. (b.) Window for entering the mapping between foods in the dendrogam and
the crop and animal products that have been defined for the farm. The window can be
opened from the menu in the Describe window, by choosing
‘Settings/Calculations/Functional Diversity’.
Nutritional system yield (NSY; Figure YYY) is an adjustment of the ‘nutritional yield’
metric proposed by Defries et al. (2015) and uses system productivity in terms of
balanced nutrient supply (for human consumption based on nutrient requirements)
rather than food item yield expressed in dry matter amount. NSY quantifies the number
of consumer units that can obtain their complete daily dietary reference intake (DRI),
particularly the recommended dietary allowance that accounts for the quantities
required by 97-98% of healthy people (Otten et al., 2006) of different micro and macro
nutrients for a year per unit of area of a production system. The production system can
be a field, a farm or a broader landscape where one or more crops are cultivated,
animals are kept and/or ‘wild’ foods grow. The consumer unit can be a reference adult
female or male. The inverse of NSY is the area required of a production system to feed a
consumer unit with energy or individual nutrients during a year.
Figure YYY. Screenshot of the Explain window of FarmDESIGN showing the nutritional
system yield for various nutrients.
Figure ZZZ. Screenshot of the Explain window of FarmDESIGN showing the demand and
supply of nutrients in food, and the deviations (%) between demand and supply.
Gross margin crops – The gross margin of the crops (MC) depends on the revenue from
crop products as affected by their production (FreshYield) and price (FreshPrice), and
the costs for cultivation and subsidies of the crop (Costs, ContractWork and Subsidy).
nc np
MC Area c * ( FreshYieldp * Fresh Pr ice p Costsc ContractWork c Subsidyc )
c 1 p 1
Gross margin animals – The amount of animal products is determined by the Number
of animals and their production of milk (MilkProduction) and meat (resulting from
Growth and Carcass percentage). Produced milk and meat are thus calculated as follows:
na
Pr oducedmilk 365 * Numbera * Milk Pr oductiona
a 1
na
Pr oducedmeat 365 * Numbera * Growtha * Carcassa
a 1
ncp nap
CF (ToAnimalscp /(DMContentcp / 100)) * Fresh Pr ice cp ToAnimalsap * Fresh Pr ice ap
cp 1 ap 1
ncp
CB (ToBeddingcp /(DMContentcp / 100)) * Fresh Pr ice cp
cp 1
The animal herd kept on the farm represents capital, for which interest should be
calculated:
n
CI Numbera * Weight a * Carcassa * Fresh Pr ice meat * InterestRate / 100
a 1
From the results of the equations above the gross margin for animal husbandry (MA)
can be derived as:
n
MA Pr oducedp * Fresh Pr ice p CF CB CI OtherAnimalCosts
p 1
Costs – The additional costs can be variable and related to inputs such as manure and
labour, and fixed for land, buildings, machinery and some general costs. The costs for
manures (CM) depend on the Amount and Price of externally purchased manures, with
ProducedAtSite = ‘E’.
n
CM Amountm * Pr ice m
m 1
The costs for assets (CA) such as buildings and machines (equipment) are resulting from
depreciation, operation costs and interest over the investment. The Depreciation and
OperationalCosts of buildings and machines are expressed as a percentage of the
capital invested. A correction factor of f=0.5 and f=0.6 for the InterestRate is used for
interest costs for buildings and machines, respectively:
n
CA Numbera * Pr ice a * Depreciation a OperationalCostsa f * InterestRate / 100
a 1
The costs for regular and casual labour (CR and CC) depend on the RegularLabourPrice
and the CasualLabourPrice are calculated separately. The amount of labour spent can
be derived from the equations in Section 6.6.
The general costs (CG) are represented by a fixed variable from the input
(GeneralCosts), whereas the land costs (CL) are calculated by multiplying the farm area
with the LandCostsPerHa. The operating profit (OP) is calculated as:
OP MC MA CM CA CR CC CL CG
The household budget can be analysed in the ‘Budget’ tab of the Explain window.
Income:
BO: off-farm income (summed for all household members)
BF: income from farming
Expenditures:
BN: expenditures on nutrition
BR: re-investment in farm
Flow metrics are being calculated for the C, N, P and K cycles on the farm. The algorithms
include the Ulanowicz and Finn indicator calculations.
Note that these are only conducted in the ‘Explain’ mode of the model and cannot be
included in the optimization due to the demanding calculations.
The main checkpoints for correct calculations in the Explain window are:
✓ In the Feed balance tab: is the feed balanced and are deviations within the
predefined ranges?
✓ In the nutrient cycles (N, P and K) tabs: the sums of inputs and outputs for each
component should be exactly balanced.
✓ In the Nutrients and N cycle tabs: animal N efficiency should be between
approximately 10% and 25%, the lower range (10-15%) for meat-producing
ruminants, the higher range (15-25%) for high-productive dairy cattle and
animal stocks containing mono-gastric animals with high feed conversion
efficiency.
Where ri denotes a uniformly distributed random value within the range [0,1] and L and
U are the lower and upper values of the allowed range. A new generation t+1 is created
by applying mutation and selection operators on the individuals in the population P of
the current generation t. The first step of the reproduction process is generation of a
trial population P’ that contains a counterpart for each individual in P, produced by
parameterized uniform crossover of a target vector and a mutation vector. The mutation
vector is derived from three mutually different competitors c1, c2 and c3 that are
randomly selected from the population P in the current generation t. The allele values
are taken from the mutation vector with probability CR:
A trial genotype p’t+1 replaces pt if it outperforms the parent genotype. Here, better
performance is interpreted as a better Pareto ranking or located in a less crowded area
of the search space than the parent genotype. These performance criteria are explained
below. Population size N is determined by the number of alleles in the genotype z and a
multiplication factor M. The last parameter in the DE algorithm is the number of
generations G, which serves as the stopping criterion. The default parameter values for F,
CR, M and G as employed in this study were derived from factorial analysis in
preliminary optimization runs, where G can be chosen such that the volume of the
solution space no longer expands (see also Section 7.3).
The concept of Pareto optimality was used to assess the performance of solutions, since
it avoids the need of normalization and a priori weighing of objectives as is the case with
common multi-criteria methods. As such it fits very well the aim of supporting
negotiation by providing insight in objectives without undue a priori restriction of the
search space.
Figure 7.1. Pareto Optimality: symbol ■ indicates solutions which are not dominated by
any other alternative solutions: setting level f1 for objective function F1 as a minimum
level, there will be no alternative solution in the dataset with a higher value for objective
F2. The arrows indicate the selection pressure exerted by Pareto ranking (solid) and the
The first criterion for the performance of a solution is its Pareto rank as proposed by
Goldberg (1989). Individuals in the population are Pareto-optimal when they do not
perform worse than any other individual for all the objectives, i.e. when they perform
equal to or better than any other individual in at least one objective. In such case, there
is no objective basis to discard the individual. These individuals are called non-
dominated and receive rank 1 (Figure 6.1). This set of solutions is called the trade-off
frontier. The next step in Pareto-ranking the entire population of solutions is to remove
the individuals of rank 1 from the population and identify a new set of non-dominated
individuals, which is assigned rank 2. This process is continued until all individuals in
the population are assigned a Pareto rank.
If two solutions have the same rank, a second selection criterion, the crowding distance,
is taken into account. The metric Θ represents the within-rank solution density and is
calculated from the normalized distance from solution p to the nearest solution in the
search space, as follows (Deb et al. 2002):
k di d
j 1 Bj
where Bj is the range of objective j, which is calculated as the difference between the
minimum and maximum values of objective j. Variable di denotes the Euclidian distance
between genotype p and the nearest neighbouring solution within the Pareto front of a
given rank and the parameter d is the average of these distances. An individual is
replaced by a trial solution of the same rank and efficiency if the latter is located in a less
densely populated part of the solution space. Pareto ranking exerts a pressure
orthogonal to the surface of the trade-off frontier, whereas the crowding metric
stimulates spread of solutions over the surface and within the solution space (see
arrows in Figure 6.1). Together these metrics ensure progress of the entire frontier to
better solutions.
File name: Console.AllFarms.txt (output file, under option 1). The values of the field
indicated in green can be used for calls for farms under options 2 and 3.
Field Description Data type
ID Identification of the farm Integer
Name Name of the farm String
File name: Console.DecisionVariables.txt (output file, under option 2). The values
indicated in the green field can be used in the input file for option 3.
Field Description Data type
Name Name of the farm component String
Parameter Parameter String
Identifier string Internal identifier of the decision variable String
Minimum Lowest allowed value Double
Maximum Highest allowed value Double
Current value Current parameter value Double
In succeeding graphs you will find an indication of the loss of organic under aerobical
and anaerobical conditions in dependence of duration of storage and of the amount of
straw used in a deep litter stable conditions, where 60% of manure is collected as FYM.
These indications are derived from Kolenbrander G.J. & L.C.N. De La Lande Cremer
(1967).
50
% of total nitrogen
40
30
20
10
0
0 5 10 15 20
months of storage
initial C/N manure = 16 initial C/N manure = 24
initial C/N manure = 33 initial C/N manure = 41
Based on the estimated loss of organic matter mineralisation of initial organic nitrogen
and assimilation (=immobilisation) of initial inorganic nitrogen is calculated according
to the same equation as presented above.
A part of inorganic nitrogen in FYM is lost by volatilisation of NH3 and another part by
denitrification. Loss by NH3 volatilisation occurs during the first weeks of storage and
denitrification during the remaining period. Total loss of nitrogen depends on the size
and form of the heap, its exposition to wind and rain and initial C to N ratio of manure
(including inorganic forms). The figures below are based on Kirchmann (1985) and can
help you to estimate the percentage of initial nitrogen that is lost during storage.
60 40
35
50
30
%OM decomposted
%OM decomposted
15 6 kg /day
20
10 9 kg/day
10 15 kg/day
5
0
0
0 2 4 6 8 10 0 5 10
Months of storage Months of storage
Figure. The proportion of organic matter lost during storage by aerobic (a) or anaerobic
(b) degradation in dependence of the amount of straw added.
Note that a high initial C to N ratio reduces loss of nitrogen only during a few months,
but cannot prevent nitrogen will be lost if storage is continued for a long period. If
measurements are available of the amount of manure, organic matter, organic nitrogen
and inorganic nitrogen you can adapt the chosen percentages to come to a better fit
between measured and calculated composition of manure.