governance committee for the ademes database technical committee database management and quality - 6...
TRANSCRIPT
Governance committee for the ADEME’s database
Technical CommitteeDatabase management and
quality-
6th of June 2011Service Eco-Conception & Consommation Durable (SECCD)
ADEME
Avec RDC Environment
Goal of the meeting
General goal : write a specification document to any provider of LCA datasets in order to supply the database
Goal of the meeting : Define / validate the proposals of guidelines for the ADEME’s database management and quality
These requirements will have to be fulfilled by any provider of LCA data
These requirements will be in annex to all technical specification document in every future calls to tender.
2
General approach
Analysis of the ADEME’s database needs : Recognition at national and international level Tool to help environmental labeling Compliance to the guidelines of the BPX 30-323 referential and the
general principles of LCA
How do the existing databases answer to these needs ? (ILCD, EcoInvent, GaBi, DEAM, CODDE-BV-TEX)
Analysis of the data quality guidelines that the databases have developed
Proposal of quality and management guidelines for the ADEME’s database
3
Agenda
4
Background and goal The ADEME’s database Content of the ADEME’s database Goal for management and quality guidelines
Database building guidelines Format Homogeneity Quality
Database utilization guidelines Update Integration of data Need for « default datasets » Confidentiality
Background and goal
Support environmental labeling for consumers goods Help to make a LCA assessment (based on several impact
categories and on a life cycle approach) by providing : A methodology
Common methodology : BPX 30-323 Sector-specific methodology : one per product category referential
Some generic LCA data
The ADEME’s database will be the recognized generic LCA database in France and at international level :
Quality of the database Co-construction : governance
5
6
Sectorial technical
committees
ADEME –AFNORplatform
Links between the ADEME’s work and the ADEME-AFNOR platform’s work
ADEME-AFNOR methodological referentials
Calculation tools
ADEME’s database
avis
Common ref. BPX 30-323 + Sector-specific ref.
Governance Committee (advisory)
7
Database governance
Sector-specific working groups of ADEME-AFNOR platform Definition of needs from : a preliminary study and some specification in
the specific product category referentials
Technical Committees Needs refinement Confrontation to available data Till the specification document redaction that will explain the data that
are needed (precise type of process) and the management and quality requirements
Governance Committee Administration NGO Enterprises
8
Content of the ADEME’s database
Some processes One reference flow (unit of the process) An elementary flows list An impact categories list Some metadata
Some characterization factors
9
Content of the ADEME’s databaseProcesses
MetadataSources, contacts, external documents
UnitReference
flow, Unitgroups
Characterization factors*
LCIA Method
Result by impact category
X*Common data to any data
provider (to come from JRC)
LCI (inventories)Flows*, flow
property*,Unitgroups
10
Data sources : 3 channels of supply
Existing or adapted data from databases Coming from existing databases developers : PE, Ecoinvent, PWC, BV,
federations (ex WorldSteel, PlasticsEurope) Modality : framework agreement and subsequent contracts (call to tender
to select candidates and then estimation of costs per sectorial technical committee)
Co-produced data To fill the pointed out lacks for some specific sectors The studies will be launched by ADEME (e.g. AgriBalyse for agricultural
productions)
Industrial date (from third part) Help to integrate some data that are still not available in the existing
databases Incentive to develop LCA
11
Planning
Questions that have structured the work
Content of the database (substance) : Which data ?
Content of the database (form) : Which format ? Which building rules ?
IT development Which integration rules ? Which management rule over the time ?
12
Content of the « ADEME’s database management and quality guidelines » document Structure of the database : which criteria?
Format Homogeneity Quality of data
Methodological conformity Inventorities clarity Recognition Transparency
Utilization practice Update Integration of data Need for « default » datasets Confidentiality
13
Agenda
14
Background and goal The ADEME’s database Content of the ADEME’s database Goal for management and quality guidelines
Database building guidelines Format Homogeneity Quality
Database utilization guidelines Update Integration of data Need for « default datasets » Confidentiality
Format – Definition of needs
15
Format must help to : Structure information Exchange data (import/ export) Be internationally recognized Provide impact categories results Provide transparency on methodology and more especially on BPX
30-323 referential
Format – How do the databases answer to the needs ?
16
2 major formats :
XML – ILCD : complete architecture that will be the future European exchange format
Problem : the today format does not provide a mandatory flows list and does not contain many mandatory fields for metadata
GaBi, CODDE-BV-TEX et DEAM are also developing processes in the ILCD format.
XML – Ecospold (EcoInvent) : format with a different structure from ILCD
Problem : the format is not the European exchange format. However some conversion systems to ILCD format exist.
Format – Proposal for ADEME DB
17
The selected format is ILCD format because : It is developed by European commission It ensures international recognition It is easy to export in XML or Excel files The metadata are well and finely structured It provides a tool for industrial (third part) providers : ILCD Editor
A new « Extension » file is added to check compliance with BP X 30-323 referential (Cf. homogeneity)
Requirements for the providers LCI delivery according to the elementary flows list defined by the ADEME Some fields of metadata are mandatory because they need to be visible at
first glance Rules in matter of reference (UUID, URI, links to other files…) Uniqueness of reference flows in the processes Uniqueness of the value for an elementary flow
Can these rules be easily fulfilled ?
Agenda
18
Background and goal The ADEME’s database Content of the ADEM’s database Goal for management and quality guidelines
Database building guidelines Format Homogeneity Quality
Database utilization guidelines Update Integration of data Need for « default datasets » Confidentiality
Homogeneity – definition of needs
How are the data linked together ? The data sources can be different and therefore the choices of
building the data either. The level of aggregation can be different.
The data must respect the same quality rules and methodology.19
Transverse data (upstream data)
Sector-specific data
TransportEnergy Shoes
TextileEEEEnd of life
?
Homogeneity – How do the databases answer to the needs ? Processes data can be divided into elementary processes (e.g. :
EcoInvent) e.g.: electrical mix, infrastructures ….
-> All the data are linked together and the modification of one elementary process generates a modification chain all over the database
20
Elem. process 1
Elem. process. i
« Aggregated » process
(to integrate into ADEME DB)
…
The data convergence is provided.
Homogeneity – Proposal for ADEME DB Constrain for ADEME DB :
No complete dynamic structure that will link all the data.
Proposal : Homogeneity could at least be ensured by industrial sector
Selection of a unique provider by sector (ex : Energy, Textile, Chemistry, Plastics …)
Limit desaggregation of the data to upstream data such as : Electrical mix : all (or only a part of) processes could be delivered as
desaggregated processes to which electrical mix from the database would be added
Infrastructures Fulfillment of common quality guidelines (including methodology)
21
Homogeneity – Questions
Sectorial homogeneityCan only one provider per industrial sector ensure homogeneity?
Limited desaggregationMust these desaggregation rules be fulfilled for all the processes or only for a part ?
Is it possible to make such a desaggregation ?
Must ADEME ask to provider for all processes (1 per country) or only one, that any user of the database could adapt with the correct electrical mix ?
Common quality requirements (cf. quality)
22
Agenda
23
Background and goal The ADEME’s database Content of the ADEM’s database Goal for management and quality guidelines
Database building guidelines Format Homogeneity Quality
Database utilization guidelines Update Integration of data Need for « default datasets » Confidentiality
Quality criteria proposed by ADEME
Inherent quality
Methodology conformity
Inventories clarity
Recognition
Transparency
24
Data quality – definition of needs
Temporal representativeness : typical data for the considered period
Geographical representativeness : representation of the considered zone
Technological representativeness : typical data for the considered technology (market representativeness)
Completeness of environmental impacts covered by the inventory
Precision / reproductibility variability and error on data
Uncertainty
Consistency
(norme ISO 14044)
Data quality – How do the databases answer to the needs ?
26The common criteria : geographical, technological and temporal representativeness.
ISOJRC - « ILCD Compliance »
JRC - « ILCD
data entry
level »
EcoInvent
Gabi DEAMCodde - EIME et
EIME-Tex
Plastics Europe
Worldsteel EAA
CONSIDERED QUALITY CRITERIA
Geographical representativeness
X X X
X
+ compleness
(market representati
veness)
X X X X X X
Technological representativeness
X X X X X X X X X X
Temporal representativeness
X X X X X X X X X X
Elementary flows completeness
XX
(calculated)? X ? ? X X ?
Precision - reproductibility
XX
(calculated)X X ? ?
X
(calculé)X ?
Consistency X X ? ? ? ? ? X ? ?
Data quality – How do the databases answer to the needs ?
27
ISOJRC - « ILCD Compliance »
JRC - « ILCD
data entry
level »
EcoInvent
Gabi DEAMCodde - EIME et
EIME-Tex
Plastics Europe
Worldsteel EAA
EVALUATION OF QUALITY
Qualitative approach XX
(doc)X X X X X X X
Semi-quantitative approach
X
(quantitative on
completeness and
uncertainty)
X
(quantitative on
completeness and
uncertainty)
Rating X
X
(qualitative eval.)
X
2 major evaluation systems : On documentation (qualitative)On ratings
Data quality – How do the databases answer to the needs ?
28
ISOJRC - « ILCD Compliance »
JRC - « ILCD
data entry
level »
EcoInvent
Gabi DEAMCodde - EIME et
EIME-Tex
Plastics Europe
Worldsteel EAA
PROOF MODE OR CHECKMandatory external peer review
X X X X X X
Mandatory internal peer review
X X
Review by the database provider
X X X X
LEVEL OF DEPTH FOR THE REVIEWSmall ? ?
Medium X X ? ? X X
High X X ? ? X
According to the organisations, either the database provider checks by himself or asks for an external peer review (no mandatory accreditation for reviews)
Data quality – Proposal forADEME DB
29
Minimum criteria to take into account : Temporal, geographical, technological representativeness
Must there be other requirements (uncertainty/precision, completeness) ? How can these criteria be proposed and checked?
Quality evaluation Short term : qualitative approach
What is the proof mode?
In long term, must we develop also the quantitative approach ? Who will establish the threshold values : the data provider ? By what kind of review process ?
Quality of data – Proposal for ADEME DB
30
Proof mode for quality, for each channel of data supply :
Data coming from databases : the providers’ review is enough
Can we valid that the ADEME’s database can be satisfied with the providers’review?
Co-produced data : require that a peer review is available
Data coming from industries (third part) : need to get an external peer review to be sure of quality
Can we require a peer review to industries without the risk that this will limit this channel for data supply ?
Must we require « accreditation » reviewers from ADEME?
Quality criteria proposed by ADEME
Inherent quality
Methodology conformity
Inventories clarity
Recognition
Transparency
31
Methodology conformity – Definition of needs
32
The data must fulfill:
The ISO 14040-44 norm Choice for impact categories Choice for methodologies (allocation rules, cut-off criteria…)
The Annex A of BPX 30-323 referential
The requirements of the ADEME AFNOR platform Methodology and Product Category Working groups and of the Technical Committees for the ADEME’s database.
Can these requirements be fulfilled ?
Can the database accept to integrate some data that do not fulfill all the criteria ? For which criteria ?
Methodology conformity – Réf. BPX 30-323
33
Le référentiel BP X 30-323 s données doivent respecter :
Topic Extract of the referentialMandatory or
optionnal?RDC/ADEME Proposal
GHGTake into account all the GHG mentionned in the BP X 30 323 referential
Mandatory
GHGTake into account the timescale for GHG (some emissions are postponed)
When relevant
Relevance evaluation from product category working groupsConsequence on tools linked to the DB as it will be necessary to use a correction factor on LCINo consequence on data in database
Land use changeTake into account impact due to land use chage
When significantWho does assess if the criterion is significant ?
InfrastructuresTake into account impacts of infrastructures for transport and energy
MandatoryFulfill requirement proposed in conclusions of Energy and Transport Committees
ExclusionsMust be excluded : carbon compensation, home-work transports, advertising, customers transports
Mandatory
Cut off criteriaCut-off criteria on mass, environment and energy
Mandatory for mass cut off criteria : 5% (95% of the mass formula has to be known precisely)Experiment environmental cut off criteriaDelete energy cut off criteria
ElectricityTake into account production electrical mixes
Mandatory Production mix
End of lifeTake into account the correct values for energy recovery at incineration, degradation rate at landfill
To be defined by the End of Life Committee
Methodology conformity – Other requirements
34
Common methodological rules for any product (included in the methodological referential)
Impacts allocation rules between co-products (e.g. : allocation leather / meat / milk)
Sectorial rules (included in the product category referential) If they have a big influence on the environmental balance (e.g : allocation
between leather and split leather)
Technical Committees recommendations for upstream data Energy End of life Transport
Must we require that the rules defined in a product category referential should be validated by the ADEME AFNOR Methodological Working Group in order to get more homogeneity?
Methodology conformity – How do the databases answer to the needs ?
35
Follow the ISO 14040-44 norm
The norm principles are followed by any database.
Fulfill the rules of BP X 30-323 referential
Do these requirements need to be adapted in so far as the data are not always able to pass the criteria ?
Methodology conformity – Proposal for AEME DB
36
Follow the ISO 14040-44 norm Choice of impact categories : from the product group and based on impact
categories from JRC (LCIA)
Fulfill the rules of BP X 30-323 referential Proposal to make visible all information linked to the BPX fulfillment :
possibility to have a complete vision of the Extension file
Can we accept non-conform data ?
Follow the common and sector- specific methodological choices
In general, requirements will be defined in specificaions and call to tender
Fictive example of BP X 30-323 fulfillment
37
Quality criteria proposed by ADEME
Inherent quality
Methodology conformity
Inventories clarity
Recognition
Transparency
38
Inventories clarity – Definition of needs
Criteria for inventories quality : Completeness of flows : have all the major contributing flows been taken
into account ?
Reliability of data (order of magnitude)
Background precision : explanation of process specificity in terms of representativeness, yields, sources
39
Inventories clarity – How do the databases answer to the needs?
Completeness of flows: qualitative check (cf. quality)
Reliability of data in terms of order of magnitude : qualitative check (cf. quality)
Background details : metadata well structured with fine level of detail
Metadata fields structuration : ILCD, EcoInvent, GaBi Provided documentation : all databases
40
Inventories clarity – Proposal for ADEME DB
Access to the elementary flow list that can be classified
41
Inventories clarity– Proposal for ADEME DB
Automatic checks in order to check orders of magnitude : if the checks are not fullfilles , then alerts are created
Check links between CO2 emissions and use of fossile fuels burning Check links between combustion processes and NOX and SOX emissions Check links between electricity processes and resource depletion Check links between transport, oil consumption and CO2 emissions Check links between water emissions and water consumption Check values for« Exotic flows » (usually very weak)
42
Inventories clarity – Proposal for ADEME DB
ILCD format used for metadata with some « at first glance » criteria (fields must be filled and visible)
seem easy to reahc ?43
Source
Géo & techno representativeness
Unit
Does the first vision of the Extension file fields seem easy to reach?
44
Temporal representativeness
Unit
Geographical representativeness
Technological representativeness
2nd visibility access
Are the mandatory fields relevant ?( )
45
Sources& Representativeness
Méthodo
Completeness for flows
2nd visibility access
Are the mandatory fields relevant ( )?
46
Type of revision
Les champs obligatoires ( ) sont-ils pertinents?
2nd visibility access
Quality criteria proposed by ADEME
Inherent quality
Methodology conformity
Inventories clarity
Recognition
Transparency
47
Recognition – Definition of needs
The ADEME’s database will be public National recognition
ADEME would like to join the ILCD Network International recognition
What steps to check recognition ? Data supply Data quality requirement Calendar
48
Recognition –How do the dabases answer to the needs ?
Quality and peer acceptance
Documentation Provide a documentation Structure data into the ILCD fields (ILCD, EcoInvent)
49
Recognition – Proposal for ADEME DB
Recognition is insured because of:
Governance (data supply)
Data quality validation
Use of ILCD format
50
Quality criteria proposed by ADEME
Inherent quality
Methodology conformity
Inventories clarity
Recognition
Transparency
51
Data transparency– Definition of needs
Need to have a documentation and an access for the following topics:
Data quality
Data sources
Methodogical assumptions
52
Data transparency – How do the databases answer to the needs ?
Documentation Provide a documentation
Structure data in some fields (ILCD, EcoInvent)
53
Data transparency – Proposal for ADEME DB
Some fields of ILCD format will be mandatory Geographical, temporal and technological repesentativeness Short description of the data set Sources Date for the delivered version
Availability of the documentation data in the supply channels Database providers: internet link to the source Co-production : available report
Must this accessibility be systematic or on demand ? Industrial data (third part): availability before integration in the database
According to the level of confidentiality, is it possible to remain data available ?
54
Agenda
55
Background and goal The ADEME’s database Content of the ADEME’s database Goal for management and quality guidelines
Database building guidelines Format Homogeneity Quality
Database utilization guidelines Update Integration of data Need for « default datasets » Confidentiality
Update – definition of needs
56
The database will evolve according to the principles of continuous improvement
The needs for update are different : Deep update : modification of structure (in link with ILCD) Mid update : general modification due to a methodological
modification Light update : ponctual update due to :
Better knowledge Improvement in data quality and reliability
Update – How do the databases answer to the needs ?
57
Frequence for update depends on refinement of methodologies
What is the frequency of the update for ILCD database ?
What is the frequency of the update for other databases ?
Update – proposal for ADEME DB
58
Structure update mainly linked to evolutions of ILCD format
Frequent revision according to some changes on upstream data (e.g. : electrical mix or new common data) : every 3 to 5 years
Very frequent update for processes modfications
Version nomenclature : XX.YY.ZZ. Z: minor modif, Y= common modif., X : structure modif
Are the frequency proposed appropriate ?
Agenda
59
Background and goal The ADEME’s database Content of the ADEME’s database Goal for management and quality guidelines
Database building guidelines Format Homogeneity Quality
Database utilization guidelines Update Integration of data Need for « default datasets » Confidentiality
Integration of data – Proposal for ADEME DB
60
Archive file saved Miscellaneous checks
Creation of the dataset in the database and storage of the XML file in a dedicated place
Type of check If no fulfillment : rejection If no fulfillment : alert
Format check
XML structure X
Are mandatory data present ? X
Reference checks
Does it exist some references in the import
X
LCA checks X
Agenda
61
Background and goal The ADEME’s database Content of the ADEME’s database Goal for management and quality guidelines
Database building guidelines Format Homogeneity Quality
Database utilization guidelines Update Integration of data Need for « default datasets » Confidentiality
Default dataset – definition of needs
62
The default datasets will more especially help to model for environmental labeling
When the provider of data does not very precisely know the formulation of his product, he can ask for a default dataset that will help him :
To check the mass cut off criteria To calculate the environmental balance
Default dataset – proposal for ADEME DB
2 possibilities for 2 needs :
1 : Definition of the build of materials :« I know what is in my product but I can not find the exact data set in the database »
Model with a « default dataset » to choose
Search for the user : « nearest dataset »: the user will choose the nearest dataset
according to its knowledge And ADEME creates and proposes an average or maximum
default dataset : average of maximum of impacts for the same kind of processes= « precise default dataset »
What choice for non precise default dataset ? average , max or percentile ?
Default dataset – Proposal for ADEME DB
Once the mass cut off criteria has been passed :
2 : Check environmental cut off criteria « How can we assess the environmental impact for the X% not modelled (X<5%) ? »
ADEME creates and proposes a default dataset (maximum a priori). Each default dataset is defined on the basis of a large family of
products (e.g. vegetables, …)= Non precise default dataset
What choice for non precise default dataset ? average , max or percentile ?
Must we leave this default datasets non visible in the database as they are very imprecise ?
Default dataset – Proposal for ADEME DB
Need : the database must contain a conversion factor that will help to use default datasets in the mass balance.
Proposal : Create a field « Reference flow weight in kg » in the « Extension »
file Mandatory field for any component that could introduced in the
mass balance : materials, components, ingredients, semi-finished goods.
Fictive example of the vision of the Extension file
66
Agenda
67
Background and goal The ADEME’s database Content of the ADEME’s database Goal for management and quality guidelines
Database building guidelines Format Homogeneity Quality
Database utilization guidelines Update Integration of data Need for « default datasets » Confidentiality
Confidentiality – definition of needs
68
The 3 data supply channels imply that it will be possible to integrate in the database some private data which could have problems of confidentiality.
The ADEME wants to make some incentive in order that the database is step by step developed.
Do we accept to integrate in the database some « private » data or must we require that the data will be public and therefore that there is a need to make the dataset anonymous ?
Confidentiality – proposal for ADEME DB
69
Proposal for 3 levels of diffusion : Public :
No restrictions at publication Anonymous, public :
Restriction on Source information Private
Not available by another company
AND threshold date : after this day, the status of the process is « public »
OR wait for several LCI and calculation of an average that could become public.
What would be the best proposal ?