GSL NOTIONAL DAIRY FARM EXAMPLE REPORT

TO EXAMINE OPTIMISATION OF A DAIRY FARM SYSTEM, MANAWATU REGION WITH EMPHASIS ON ENVIRONMENTAL AND ECONOMIC IMPACTS.

FARM 4.5.240

March 2011

Goals/Objectives:

  • To investigate ways to reduce N loss while investigating options for retaining profitability and minimising negative change to overall farm performance.
  • To establish at what point any further N loss requirement becomes less economic than the current farm system.

It is hoped to compare these options through Overseer and determine which mitigation scenarios may impact most on N leaching and their economic impact in a systems context.

MODEL:

A resource allocation model (the “GSL” model) which is based on mathematical equations and relationships established from animal production research was used. The integration and relationships required for systems analysis are provided by the technique of linear programming (LP). This is an optimisation routine that allows many different types and levels of input to be assessed through a sequential process of analysis (iterations) to establish the point at which adding further input results in no further improvement (nominally) in profit.

The base farm system was set up within the model to ensure that management events and model events coincided. A number of computer runs were completed with the aim of reducing N loss at least cost while also moving towards the longer term objectives of the farm owners. The results from these runs were then entered into Overseer 6.03 to establish N loss figures as generated from this program.

PROCESS:

After establishing the base farm, a series of runs were completed in. As much as possible, input and system changes were constrained to provide a stable base for comparison of performance.

NOTE:

The base farm run quantifies the base resources that the farm uses. Each subsequent run provides a variation to the way the farm system can be run using the same or very similar resources. This ensures that ANY of the options reported can be implemented on the particular farm. They are runs that are associated in the manner reported to the actual farm being studied.

However:

Although the base farm has been constructed with inputs and outputs that correspond to those provided, this situation is only a “snapshot” of the conditions for the current year. These inputs and outputs may vary in subsequent years but will continue to be relevant to the farm. Of importance are the underlying resources in land, soils, pastures, stock, infrastructure, management objectives and environment for the area into the future.
Management objectives alter and any farm will require extended time frames for any changes to have an effect. The initial data used may have come from a dryer or wetter year. No attempt was made to model for different climate scenarios as this increases the complexity of the reporting but can be completed against defined climate scenarios when specified.

This process can be followed between each run and finally results in answers overriding the best use of resources within the more productive system to one of using resources to achieve less N loss. This introduces inefficiencies in resource use being forced into the analysis rather than optimal resource use.
This report details the point at which this transition is most likely to occur.

Inputs for this farm.

  • Data from the farm was used to establish pasture growth rates and pasture quality for each 2 weekly period for a year.
  • The data input are those that correspond to the total input requirements for the stock carried and the outputs produced.
  • The dairy herd consisted of 750 mainly cross bred cows producing 318kg MS/cow per cow and total 238,434 kgMS.
  • Lactation from 27th July until May.
  • Replacement rate of 25% was used for the base farm and included an average 6 year period for each cow to remain in the herd.
  • A series of Runs were completed to investigate current, interim and intended production scenarios and then several others where N loss was constrained.
  • A price of $6.40/kgMS for milk solids used.
  • An initial FWE cost was derived from Dairy Base records. This was $950 per cow for the costs associated with the base running of the farm and cows. Additional feed and feeding costs were included as part of the model function as was cost of purchase and application of nitrogen applied and any silage made.
  • No depreciation, wages of management, capital or financing costs were added the per cow average FWE
  • Nitrogen was applied to pasture (750Ha) as specified for dates and rates split into autumn and spring dressings.
  • The effluent and non-effluent blocks were treated as one block when running the GSL model
  • 750 ha total pasture growing 11,844 kg DM/ha/year net of Nitrogen.
  • Pasture quality varied with an average of 11.3 – 11.7 MJME/ kg DM.
  • All replacement stock grazed off the milking platform 15 November to May returning as in calf heifers.
  • 750 cows grazed off farm for 4 weeks. All other cows are grazed on the farm.
  • No other options to graze stock off were investigated.
  • These data were used to establish an initial “Base” system that modelled and provided the base for the complete analysis.
  • This system was either constrained or optimised for individual inputs depending on the objective of each scenario run.
  • The amount of N and GHG as (CO2) was calculated and reported.
  • The inputs required for an N leach figure from the Overseer ® program were produced.
  • These data are summarised in spreadsheet form.

See spreadsheet table for the 8 Runs/scenarios report.

Discussion of Table results.

  • The tables contains a large amount of interrelated data that, although specific to each model run, should be used to confirm resource allocation trends and effects as inputs alter.
  • It should be emphasised that there is no “one answer”. What modelling is able to do is to establish farm systems that vary as the inputs are constrained. From this it becomes more obvious as to what management changes will likely lead to higher efficiencies and better economic results.
  • If this was a change from the current farm system, the data and farm system will be proven as each subsequent season evolves.
  • At current levels of production per cow, input costs, product prices and system performance provided, the marginal cost of supplements adds little or no improvement to the final economic surplus.
  • The main use of supplements is to support more cows through the late summer and autumn due to the demands for feed as stock are wintered on.
  • The use of nitrogen in particular periods was shown to improve surplus but this also increases N loss in Overseer. The model could use nitrogen and replace other supplements but only up to a certain point, then all N use had to cease.
  • However there was a very high economic value to use N in autumn as the cheapest way to boost feed supplies into the winter. The model when allowed to Range N use always chose to apply the maximum allowable N in March.
  • No other options than reducing herd number and grazing off were investigated in any detail but it would seem obvious that some progressive herd reduction into autumn may be economic in reducing later winter and calving demand.
  • Nitrogen was still preferred over any supplement in subsequent runs.
  • The model found having the cows grazed off in winter to be preferable to wintering cows on in terms of economics and could profitably increase cow numbers.
  • The farm stocking level (even when winter grazing off was allowed) was higher than the optimised model chose. The model in different scenarios did purchase supplements but also reduced herd numbers.
  • If the farm system were altered to allow more grazing off (it may be physical conditions of contour or walking distance on farm that deter this option due to areas that suit dry cows) the requirement to buy in feed will diminish. There will need to be other adjustments made (drying off/ culling / stock numbers/ return of grazed off stock) to ensure improved systems and economics.
  • With the herd not drying off until mid-May, pressure was on during the later summer and autumn to feed cows (if grazed on the farm) and allow some increase in PGR for calving in spring.
  • This autumn feed requirement was best met (economically) through using nitrogen and in one non reported run, some cows were dried off and culled early to save pasture and minimise bought in feed.
  • The ability to dry off and cull cows later in the lactation would reduce the amount of feed required at that time and be marginally more economic than buying in feed. It would also reduce N loss at that critical autumn period. Certainly if production per cow reduces, supplements become more economically marginal.
  • Overseer however views use of N in autumn as a critical factor in N loss figures depending on the timing of application?
  • The most valuable production is achieved through higher productivity (achieving more product from similar input) by using the potential of the herd to ensure high per cow production.
  • However this can be at an additional cost that is less than the additional return (marginal cost less than marginal return; MC<MR) when supplements that are both more expensive and poorer quality than N boosted pasture are used. In this case it is preferable to substitute N pasture for purchased supplements to maintain production while maintaining or decreasing N loss. But this may impact N leaching more.
  • In most cases the model would rather save feed costs and the costs associated with extra cows by ensuring all pasture is efficiently used by fewer, better producing cows. Supplements would be used only to cover genuine small feed deficits. These do occur on this farm during summer but reducing herd number and some wintering off reduces this requirement and any gaps are filled by adding smaller quantities of strategic silage. As herd number decreases, these gaps are covered by farm silage made from late spring surpluses.
  • This shows that adding supplement (of the required quality and quantity and price) for short periods to overcome underfeeding will increase profit if increased total per cow production results.
  • Also using only pasture with a higher return of MS/kg pasture DM eaten reduces the need for the additional expenses that may occur with high supplements in terms of R&M, increased labour and fuel, removal of waste and higher effluent system costs.

(The report from here on is for example of the discussion only and does not relate to the attached spreadsheet)

Inputs

In the Base Run 1, input feed quality was provided on a fortnightly basis. All feeding levels related to LW, breed, bodyweight gain/loss profiles, two
weekly production figures and activity according to feed energy levels which varied with period of the year and feed used.

Average pasture covers were constrained within fortnightly specified limits for both maximum (2500) and minimum (1500) levels as a means to represent conditions necessary to retain both pasture quality and animal intake. No attempt to model effects of pasture growth and quality nor pasture offered, animal intake and production were made other than those presented within the model. The successful management of these is left to the farm manager on a daily basis. All figures represent the conditions for this particular application but can be varied in addition runs.

Run 1. (Base). This initial run models a similar production system for this farm as for the 2012/13 season. It details the base resources (quantity, timing of use, quality and production from all stock, both LWG and MS production, amount and time) for the current farm. Average herd number used was 568 cows producing 357 kg/MS/cow and 202,693 kg/MS total.

The model required some extra energy in the August -September period to achieve the required MS per cow for the younger animals. It may be that some extra weight loss than has been modelled occurs in practice.

Pasture growth seems lower than expected and it is presumed that the farm has some hilly areas or is an awkward shape. The pastures may have a negative impact on per cow performance as the high “supplementary” feed inputs (37% of total feeds) are constrained due to reducing Nx (n loss) limits.
However, the, base per cow production (357kg/cow) is maintained for all scenarios but some upward movement in subsequent scenarios results from a change in herd structure as the herd total is reduced..

This is a high input system and therefore there are opportunities to decrease Nx in a number of ways.

Nitrogen use was as stated at a response rate of 15kgDM/kg N applied spring and 10:1 autumn. This base run was then used as the comparative farm base for subsequent scenario changes.

The LP was adjusted in terms of pasture growth rates (PGR), quality, utilisation, average pasture cover (APC), animal performance and energy expended walking until an “interim” balance of all inputs and outputs for the 2012/13 season was simulated. The model inputs and outputs are summarised in a number of rows which are all interconnected to form the
overall system. All figures relate to each other and add to form a final total of kg/DM used (this figure is calculated from the energy of the differing feeds used within the model.) Each Run is related to the previous Run and then back to the original farm resources.

The model reports carcass weights of animals culled based on liveweight (LW) profiles for each 2 week period which relates to the dates of culling and the likely numbers from each age group within the herd. Calf sales are related
to the stock reconciliation that is performed based on the Herd Makeup replacement rate. All these figures will vary with herd makeup and date of sale.

The CO2, nitrogen sold in product and nitrogen excreted (Nx) figures are based on a “greenhouse gas” (GHG) layer within the model related to all kg/DM and crude protein per cent (CP %) data within the model and are calculated as per IPPC recommendations.

These figures are very useful in comparing the relative efficiencies of feed use as the use of inputs change. They are relevant as they quantify likely future charges or constraints to be applied to farm systems.

So Run 1 Base answers the question on likely resources the 2012/13 Base system requires to successfully perform to expectations of the production and profit listed. It should be noted that the data supplied apply to a specific
circumstance that it is subject to both change in different years and accuracy of original data and the modellers interpretation of the data received. Each season and year will vary and as change occurs to any system it is how the changes are managed that becomes as important as the current data.

The mix of resources were then fixed or varied as required in subsequent runs as the questions on declining Nx are answered. Examining how the various inputs and outputs change (substitute) with each new option provides an insight into just what factors are important within this farm system when managing for maximum profit within N leach constraints.

Note the GSL model is a linear programming model that allows numerous calculations in pursuit of an optimal result as to which option is most economic within each completed calculation or “Run”. This allows a marginal analysis to
be undertaken for each input on the basis of how the last unit of each input may alter the economic result. This allows LP to “choose” the best mix of inputs rather than just selecting what may appear to be the “best”.

Note that the model eliminates what is the least economic feed input first (grass silage and grain feeds in this case) and the most economical is discarded last (nitrogen) as increasing constraints are applied to the volume of Nx allowed.

NOTE 1: The data used is the best available from the reports provided. Any farms data is “transitory” in its timing and provides a guide only to trends and a base for comparative analysis between years and management options.
Farm systems will change over time as better technology becomes available. Data is not as detailed as required for complete accuracy in this case.

A series of runs to cover options are Tabled. They may not all conform with required limits to N loss but provide a stable basis to understand likely trends in economics as N loss figures are reduced. This is done by using the N
excreted (Nx) figure from the model and reducing this as a means to reduce N loss as subsequently calculated by Overseer.

NOTE 2:

It is always difficult to establish a true “Base year” to compare, as each year will be different in terms of timing and amount of feed grown and used. Later reductions in the N loss requirement forces reductions in the
amount of feed (crude protein) that can be consumed. Such reductions may also contribute to changes in pasture species which may in turn affect both quality and quantity if management of both pasture and stock is not maintained at a high standard.

Run 1 is the Base farm run. The grazing off policy was not altered even though less cows and young stock were present in later runs.
Utilisations varied according to where the supplement was fed.

  • Grass Silage 90%
  • Maize Silage 90%
  • PKE average 90%

8 ha of turnips sown November, finished in March are sown with a 10,000 kg DM/ha yield assumed.

Run 2. As this farm is high input, it has previously been requested that the analysis allow a more gradual decline in herd numbers as a means to reduce Nx through less feed and nitrogen used. This is done by reducing herd numbers by allowing the model a range of herd size and reducing the lower limit by approximately 50 cows each run until the optimised herd is within the allowed range. This was undertaken in approximately 50 cow reductions but with an upper limit of herd number the same as the Base farm.

Runs from Run 3 through to Run 7 resulted in the model choosing the lower limit of herd number range and to respond to the progressive DECREASE in Nx as bought in feed is reduced along with herd numbers; but this contrary to expectations also INCREASED the $surplus over the base farm $surplus.

The crop area was allowed to optimise from Run 4 and was immediately dropped demonstrating the uneconomic benefit of cropping on this farm separate from the constrains of Nx produced from cropping..

This indicates that at these levels of per cow production and input costs, a MS price of $6.50/kgMS is insufficient to justify high cost supplementary inputs (except for Nitrogen). The first feed discarded was silage (due to cost/MJME), then PKE.

Over these initial 7 Runs, herd size decreased from 568 cows to 338 cows; Grass silage was not used from Run 2 and PKE use dropped from 330000 kg DM to 0 kg DM by Run 3.. Nx reduced from 72,275 to 50,593 kg, the 30% Nx reduction as directed.

Nitrogen use was allowed to optimise and initially remained at a relatively high level but by run 7 is 22% of the amount applied to the base run. The $surplus rose with each successive Run to run 3 as feed with a marginal cost greater than the marginal return from the added production gained was removed. Runs 4 to 7 then showed and increasing rate of $Surplus decline. Run 7 shows the minimum number of cows while meeting the target -30% Nx with $surplus above the base run. With 338 cows this run closely equated the Base Run in $Surplus. Meaning all prior runs exceeded the base run $Surplus.

Effectively the current farm is making no additional economic gain from running an extra 230 cows.

The reasons for this may need more analysis and explanation at some time with the farmer.

Run 7 has an Nx that is already 30% below the Base Nx and the farm can still reduce this further as bought feeds are still being imported, but at this stage the feeds are real “supplements” as they are supplementing real deficits
between pasture growth (plus nitrogen) and herd requirements for particular periods of the year..

Run 8. Reduces Nx by 35% of Base and optimises the herd at 312 cows but all supplements have been eliminated. Using maize silage rather than PKE to reduce the CP% and likely Nx (N loss) total while still enabling a reasonable
$surplus is demonstrated in runs 3 to 8. The Linear Programming model is now optimising for both Nx and $surplus and finding the best ways of juggling herd numbers, feeds (and crude proteins) with $cost and income.

Run 8 is the only run where feed is conserved and fed later to the stock.
However it is also likely that per cow performance is likely to suffer as less feeds are added and cows will be required to harvest pasture in greater quantities. In order to maintain per cow production, more feed will need to be
offered and it is likely that utilisation and/or feed quality will suffer which in turn may limit per cow production unless management responds to the new environment.

This assumed change in feed quality was not tested but is an option if further insight is sought, but the “straight line” reduction of Nx between the runs gives a clear demonstration of the underlying issues of profit and Nx reduction.

Runs 9 to 12 shows the accelerating decline of $Surplus as all added feeds and N applications are eliminated from the feeding equation. In these runs the volume of feed discarded rapidly increases reflecting the lowering Nx limit
being imposed and the need to reduce the total crude protein being processed by the animals. These later runs take the dairy farm system into rapid economic decline as the Nx limits reach 40 to 55% of Base farm levels.
These Runs are clearly outside the limits for viable dairying for this farm.

Summary:

This farm has productive cows being supplemented by high levels of bought in feed and nitrogen. This provides options to reduce Nx, but such reductions come at increasing cost once the initial 20% Nx reduction is achieved. In fact, it could be argued that much of the “supplement” is actually just feeding extra cows rather than filling in feed gaps at strategic times.

This makes the use of such extra feed more expensive as the costs involved in running each additional cow (per cow costs such as those of breeding, animal health, labour and interest) when supported fully by this extra feed needs to be costed fully against that “marginal” cow. This makes even

Click here to go back to the main page


Login