Using supply chain optimization audits for carbon footprint minimization in an energy supply chain

1
Using supply chain optimization audits for carbon
footprint minimization in an energy supply chain
Decision Systems Lab University of Wollongong

• Professor Aditya K. Ghose
• Director
• Decision Systems Lab
• School of Computer Science and Software
  Engineering
• University of Wollongong
• www.uow.edu.au/aditya
• aditya_at_uow.edu.au

• Saugato Mukerji
• Senior Automation Engineer and Supply Chain
  Thought Leader
• Bluescope Steel
• saugato.mukerji_at_bluescopesteel.com

2
Keywords in the soup

• Supply chain optimization
• Optimization technology
• Decision triage
• Audits/methodologies
• Cost/benefit analyses
• Energy supply chains
• Manufacturing supply chains
• But also
• Business process management/improvement
• Enterprise process architectures
• Strategic alignment

3
Decision Systems Lab_at_UoW

• 7 academics 30 research students
• 3 million in current research funding
• Key themes Supply chain management, requirements
  engineering, BPM, service-oriented computing).
  Also, constraint programming, agent systems (with
  significant supply chain applications)

• Funding agencies
• Australian Research Council
• Canadian NSERC, Carnegie-Bosch Foundation
• Japanese Institute for Advanced IT
• Cooperative Research Center for Smart Services
  (140 miliion industry/government/academia-funded
  entity)
• Research contracts from organizations in a range
  of industry sectors

4
Where the war stories come from

• SAP/Infosys and several others in the CRC for
  Smart Services
• IBM Research
• BlueScope Steel
• CSC
• Pillar Administration
• NSW State Emergency Services
• Actenum Corporation (Vancouver, Canada)
• Several SMEs in the IT space

5
The war stories

• Constraint-based production scheduling (steel
  sector)
• Integrated constraint-based planning and
  scheduling (steel sector)
• Constraint and market-oriented programming in
  scheduling (steel sector)
• Optimal truck dispatch systems (mining sector)
• Optimal manpower scheduling (a major airline, an
  employment agency)
• Optimal design of blasting rounds in open pit
  mines

6
And more war stories

• EBAR enterprise architecture being deployed at
  several Australian Federal agencies
• Large-scale enterprise process architecture
  project at a emergency services agency
• Business process integration at a major hosted
  financial services provider
• Next generation model management systems (with a
  major modeling tool vendor)

7
Basic propositions

• Operational efficiency is fundamental to
  C-footprint minimization
• Optimization is fundamental to achieving
  operational efficiencies
• Awareness of (and capability to deploy)
  optimization solutions is often missing
• Organizations often face challenges when
  deploying optimization solutions
• Decision triage

8
Basic propositions

• Optimization decision triage
• What to optimize?
• What optimization scope?
• How to optimize?
• Plus others strategic alignment, organizational
  impact etc.
• Need principled basis for this
• This has intrinsic business value
• Butspecially important in a business environment
  where the impact of optimization is magnified

9
The Univ. of Wollongong Carbon-Centric Computing
Initiative

• Change of discourse IT as a polluter ? IT as a
  source of climate change solutions
• High impact IT for C-footprint reduction
  optimization technologies, business process
  management systems, grid computing/virtualization,
  telepresence technologies
• UoW project on the Optimizing Web
• Industry consortium seeded at UoW (see similar US
  initiative at www.gesi.org)
• Upcoming events
• Release of the UoW CCCI report (a week from now)
• A National Research Summit (October)
• Major academia/industry conference (in tandem
  with European IT for Climate Change Conference)

10
What is SCOA?

• An optimization landscape mapping methodology
• A requirements engineering methodology for supply
  network-wide optimization
• A design methodology for network-wide
  optimization architectures
• An audit methodology for optimization across the
  supply network
• What are the optimization requirements across the
  network?
• Are these requirements being met?
• The audit methodology thus subsumes all of the
  above
• If the answer to the last is question is no, that
  serves as a trigger for additional requirements
  acquisition and re-design

11
Why SCOA (1/7)

• Decisions on optimization loci are often ad-hoc
• In designing the deployment of optimization
  systems within a supply chain, a key question is
  what to optimize and where.
• Each answer to these questions determines a locus
  of optimization.
• These decisions are often opportunistic and
  ad-hoc, ignoring the broader context of the
  supply chain and the organizations involved.

12
Why SCOA (2/7)

• Decisions on optimization scope lack a principled
  basis
• Key challenge in designing optimization systems
  is deciding on the interplay between local and
  global optimization.
• Local optimization example A large manufacturer
  might, for instance, adopt a local optimization
  approach in which stand-alone optimization
  systems are implemented for each separate
  business unit
• Benefits problem scope is usually well-defined,
  the constraints and objectives are easily
  obtained and the system can be easily
  implemented.
• Drawbacks local optimization can lead to very
  limited efficiency gains since a collection of
  locally optimal solutions can be significantly
  sub-optimal from a global perspective.

13
Why SCOA (3/7)

• Decisions on optimization scope (contd.)
• Global optimization can be more difficult
• optimization system would need to cut across
  organizational (or at least business unit)
  boundaries.
• The entities involved may not be willing to
  reveal all of the required constraints and
  objectives, because of the potentially
  business-sensitive nature of the information (or
  even for privacy reasons).
• The objectives of the participating entities
  might be conflicting, in addition to the
  collected set of constraints being
  over-constrained.
• Benefits global optimization guarantees that the
  solutions generated are truly optimal, form a
  global (enterprise-wide or network-wide)
  perspective.
• Local and global optimization represent two ends
  of a spectrum.
• the configuration that is most appropriate might
  represent some intermediate point on that
  spectrum.

14
Why SCOA (4/7)

• Decisions on technological platforms are often
  ad-hoc
• the landscape of optimization technologies is
  fairly large.
• the best decision might require a mix-and-match
  approach.
• There are no principled methodologies to guide
  such decisions.
• Cost-benefit analyses of optimization system
  implementations often ignore the enterprise- and
  network-wide perspective

15
Why SCOA (5/7)

• The organizational impact of the implementation
  of optimization systems is often ignored
• In many instances, the optimization system would
  replace a largely human-mediated business process
  that would have evolved organically over a long
  period of time
• the collateral deliverables of these processes
  are never explicitly identified, yet are critical
  to the organization in question.
• When these processes are supplanted by new
  optimization systems, these undocumented
  functionalities are often irretrievably lost (but
  this might not become immediately apparent,
  making remedial action difficult).

16
Why SCOA (6/7)

• The impact of optimization systems on supply
  chain alliances and relationships is frequently
  ignored
• legacy inter-organizational coordination
  processes are often supplanted by the
  introduction of optimization systems.
• These processes often also deliver value (e.g.,
  by reinforcing inter-organizational trust) in
  ways that are not made explicit.
• Such value can also be lost as a consequence of
  optimization system deployment.

17
Why SCOA (7/7)

• The extent to which optimization systems align
  with network-level and organization-level goals
  is often not analyzed
• An optimization system that seeks to generate
  production schedules that minimize makespan (and
  thus cost) without paying heed to the priority of
  the customers associated with specific jobs may
  be badly misaligned with a business strategy that
  seeks to deliver fast lead times on a core set of
  high priority customers.
• The business objectives of organizations within a
  supply network may pull in contradictory
  directions
• Inadequate attention is paid to the integration
  of optimization systems with B2B/market
  interfaces

18
Key elements of SCOA

• Optimization decision triage
• What to optimize?
• What optimization scope?
• How to optimize?
• Plus others strategic alignment, organizational
  impact etc.
• Measurement
• Judiciously chosen measure points and dimensions
• Helps establish baselines
• Intra-enterprise
• Cross-enterprise/sector-specific
• Provides basis for causal model discovery (this
  can be lightweight!)

19
Objective alignment

• Designing local objectives that enable us to do
  the right thing by the global objective
• Ensuring that a network of optimizers pull in
  the right direction
• Ensuring that a network of optimizers comply with
  global optimization requirements

20
Optimization architectures (1/8)

• An optimization architecture defines what
  optimization systems are used within a supply
  network, what specific problems they address, how
  these systems interface with each other (and with
  B2B interfaces across the supply network) and how
  these systems relate to the architecture of the
  supply network (defined in terms of the key
  organizational entities and their
  inter-relationships).

21
Optimization architectures (2/8)

• Network architecture
• defined in terms of the key organizational actors
  that participate in a supply network, together
  with a high-level, abstract specification of
  their inter-relationships and dependencies.
• best represented using diagrammatic enterprise
  modeling notations.

22
Optimization architectures (3/8)

• Optimization loci
• defined via reference to the network
  architecture, by identifying specific network
  entities, or groups of entities, that specific
  (atomic) optimization systems deal with.
• Diagrammatically, these are best described as
  overlays on top of a network architecture model.

23
Optimization architectures (4/8)

• Optimization scope
• defined by a set of optimization loci (and hence
  their respective optimization systems) that
  inter-operate to conjointly generate optimization
  solutions
• individual optimization systems within a given
  scope generate their locally optimal solutions,
  but also coordinate, in either a loosely- or
  tightly-coupled fashion, to generate more
  globally optimal solutions.
• It is usually possible to identify, within a
  given supply network, several such disconnected
  optimization clusters.
• Diagrammatically, these are also easily
  represented as overlays on top of a network
  architecture model.

24
Optimization architectures (5/8)

• Optimization interfaces Within the scope of a
  given optimization cluster, multiple optimization
  systems need to inter-operate. Optimization
  interfaces determine the mode of inter-operation.
  
• On the one extreme, optimization interfaces can
  be tightly-coupled, with the participating
  optimization systems revealing all of their
  constraints and objectives to each other.
• At the other extreme, optimization interfaces can
  be loosely-coupled, with the participating
  systems coordinating via a minimal repertoire of
  messages under privacy-preserving protocols.
• A range of interfaces between these two extremes
  are also of interest.

25
Optimization architectures (6/8)

• B2B interfaces and e-market architectures
• B2B interfaces might be as simple as traditional
  catalog-based ordering systems, or as complex as
  online combinatorial auctions.
• These too can be represented on top of a
  diagrammatic model of the network architecture.

26
Optimization architectures (7/8)

• Key components (contd.)
• Connections between optimization loci and market
  interfaces
• Legacy process models An optimization
  architecture must make explicit the legacy
  processes that are supplanted. A variety of
  process modelling notations such as BPMN are
  suitable for this purpose.
• Desired process models An alternative view of an
  optimization architecture can be provided via a
  collection of process models of the target
  processes.

27
Optimization architectures (8/8)

• Connections between optimization processes and
  network- and enterprise-level objectives
• A key component of an optimization architecture
  is explicit goal models and the establishment of
  traceability links between goal models and other
  elements of the architecture.
• A goal model specifies the organizational and
  enterprise-level objectives that we week to
  satisfy (or optimize), and their hierarchical
  decompositions to sub-goals etc.
• Goals can be of two types achievement goals
  (where the intent is to achieve a state of
  affairs that makes some conditions true) and
  optimization objectives (these can be described
  qualitatively, or as mathematical objective
  functions).
• Several techniques exist for relating goals to
  process models which are directly applicable in
  this instance (Koliadis 2006).

28
SCOA Outline

• Define the network architecture
• Define the optimization architecture
• Define optimization architecture benchmarks
• Perform optimization architecture analytics
• Use benchmarks and analysis outcomes to drive
  optimization architecture re-design.

29
Actor Dependency Maps
30
Actor Goal Maps
31
The actor eco-systems analysis tool
32
The end

• More information available at
• www.dsl.uow.edu.au
• www.uow.edu.au/aditya