Allocating intra-industry material and energy flows using physico-chemical allocation matrices. Application to the Australian Dairy Industry

Transcription

1 Allocating intra-industry material and energy flows using physico-chemical allocation matrices Application to the Australian Dairy Industry Andrew Feitz, Sven Lundie, Gary cheese guru Dennien, Marc Morain and Michael Jones UNSW 1

2 Introduction Australian Dairy industry one of the largest agricultural industries ~13 billion litres of milk ~$2.8b contribution to Aust. Economy (0.5%) High water use and eutrophication impacts Greatest challenge during the LCA - allocating milk, water, energy etc to different products (cheese, milk, butter, etc) 2

3 The problem Many different products manufactured using the same resource (milk, fuel, electricity, etc) SMP - skim milk powder WP - whey powder WMP - wholemilk powder BMP - butter milk powder 3

4 ISO recommended allocation 1. Avoid - expand system boundaries, don t differentiate between co-products 2. Use physico-chemical allocation wherever possible 3. Use a causal relationship for allocation, e.g. metabolisable energy from feedstuffs to allocate grain, fodder to milk/meat/calf Economic a proxy causal relationship ISO do not recommended allocation on a mass, volume or energy basis - can introduce large errors 4

5 Obtaining allocation data Ideally use metered flows for different products (e.g. electricity, water, etc) - rarely available Metering likely to become more prevalent, e.g. wastewater from cheese, butter, powder departments (competition within plant) This study: Iteratively built up co-product input data based on 19 dairy manufacturing plants surveys 5

6 Iterative data (e.g. energy use) Energy for milk powder only plant (cream sent off site) Use to determine butter energy from a powder/butter plant Use to determine AMF energy from a powder/butter/amf plant Etc, etc Validate with literature energy values (usually old though), industry standards, industry know-how, suppliers Ok for plants with the same level of technology 6

7 Raw milk allocated using milk solids % Milk solids % Protein % Fat kg milk solids per kl of raw Milk solids concentration factor milk Raw milk Pastuerised milk UHT milk Cheese (Cheddar) WPC (65%) WPC (35%) Whey powder e.g. 1000L milk 105kg cheese & 53kg WP 7

8 Average resource use and w/water for Australian Dairy plants Raw milk Water use W/water Electr. use Fuel use Alk. use Acid use Units t/t t/t t/t GJ/t GJ/t kg/t kg/t Milk powders Yoghurt Market milk Cream Butter AMF/Ghee Cheese Whey powder UHT milk Ice Cream Conc. Milk WPC/lactose 8

9 Average resource use and w/water for Australian Dairy plants - relative Raw milk Water use W/water Electr. use Fuel use Alk. use Acid use Milk powders Yoghurt Market milk Cream Butter AMF/Ghee Cheese Whey powder UHT milk Ice Cream WPC/lactose

10 Other relationships Fuel use air emissions from burning fuel (per product) Raw milk raw milk transport (per product) Wastewater milk solid losses, nutrient emissions (per product) Not perfect = dairy plant wastewater composition mostly uniform except for cheese w/w Sweet whey treated as separate flow Sanitisers not included at this stage 10

11 Example multiproduct dairy plant Annual Inputs Raw milk 750 ML (774 kt) Utlilties 140,000 GJ Electricity 720,000 GJ Natural gas 1100 ML water 800 tonne alkali 200 tonne acid Annual Outputs Products 100,000 tonnes/yr of Market milk 30,000 tonnes/yr of Skim milk powder 20,000 tonnes/yr of Whole milk powder 5,000 tonnes/yr of Butter milk powder 5,000 tonnes/yr Whey powder 20,000 tonnes/yr of Cheddar 15,000 tonnes/yr of Butter Waste 1100 ML of wastewater treated using facultative ponds and then irrigated 11

12 Example allocation Raw m ilk alloc at ion f act or * product %alloc at ion kt t m ilk/t Marke t milk , , SMP , , WMP , , BMP , , Whey Po wder , , Butte r , , cheese , , total 100, ,000 Elect rici t y alloc at ion f act or * product %alloc at ion GJ GJ/t Marke t milk , , SMP , , WMP , , BMP , , Whey Po wder , , Butte r , , cheese , , total 97, ,000 12

13 Problems with standard allocation for raw milk (i.e. farm) to products Matrix (milk solids) Economic Jul-01 Economic Jun-02 Mass Energy Fat Protein Market milk 13.1% 31.4% 34.5% 51.3% 6.2% 13.2% 12.1% Skim milk powder 30.2% 20.8% 17.4% 15.4% 42.6% 0.8% 40.5% Whole milk powder 20.1% 11.1% 10.5% 10.3% 28.4% 18.4% 19.9% Butter milk powder 5.0% 3.5% 2.9% 2.6% 7.1% 1.0% 6.2% Whey Powder 4.9% 1.0% 1.1% 2.6% 9.4% 0.2% 2.4% Butter 13.1% 9.5% 10.9% 7.7% 1.9% 42.8% 0.3% Cheese 13.5% 22.7% 22.6% 10.3% 4.5% 23.5% 18.5% Total 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% Enormous errors if non-causal relationship used 13

14 Economic allocation of raw milk Economic allocation 34.5% x 774,000t milk = 267,700t for 100,000t of raw milk! Nonsense where has the other 167,000t gone? Farm impacts (e.g. the 99% of total water impacts) are therefore 2.7x higher than they should be 14

15 Gets worse Fuel use Cheese Butter Whey Powder Technical Protein Mass Economic Physicochemical allocation Butter milk powder Whole milk powder Skim milk powder Market milk 0% 10% 20% 30% 40% 50% 60% Allocation 15

16 Comparison of errors Introducing order of magnitude errors in some cases! Compared with: <1% error associated with dairy input data (raw milk, electricity, fuel, etc) ~10% error associated with wastewater volumes ~ 50% truncation error (Input/Output) Getting allocation right is critical 16

17 Allocation examples Economic allocation widely used in food LCAs One recent Dairy LCA used economic allocation for upstream and downstream processes for a multiproduct cheese plant ( 67% ), for farm contribution, fuel, electricity, water, wastewater, transportation.. Propagation of errors 17

18 Could allocation matrices be used in Input/Output analysis? Input/output - very high sectoral level Sectors can be linked using economic tools Matrices could allow allocation of inter-sector flows within a sector, to enable tracing of a single product Dairy farming sector Raw milk ($) Dairy products ($) Raw milk (CO 2 -e) Raw milk ($/product) Milk (CO 2 -e/product) Dairy Manufacturing sector Dairy products (CO 2 -e) Cheese, milk, butter, etc Supermarket sector I/O I/O+ 18

19 Conclusions Economic tools fine for input/output analysis but not suitable for complex multiproduct LCAs (cost a poor indicator of resources used per product) Physico-chemical allocation crucial for multiproduct dairy plants Despite good primary data quality, poor allocation can lead to >10x errors Matrices could be developed for other multiproduct industry sectors, e.g. cropping, petrochemicals, mining, etc 19

20 Acknowledgements 20