Use cases for process data mining –

– Metsä Board view

Matti Ketonen

QPR Conference 2018 – Sept 18th 2018

2 18/09/2018 QPR Conference 2018

 Contents

Metsä Group 4
Metsä Board supply chain 5
Journey from 2011 6
1 Setting base case 7-8
2 Continuous improvement of process performance 9-10
3 Identifying bottlenecks for process automation 11-12
Success ? 13
Mindset 14
Future 15

3 18/09/2018 QPR Conference 2018

 METSÄ GROUP | Sales EUR 5.0 billion | Personnel 9,800

METSÄLIITTO COOPERATIVE | Group’s parent company | Owned by 120,000 Finnish forest owners

METSÄ FOREST METSÄ WOOD METSÄ FIBRE METSÄ BOARD METSÄ TISSUE

Wood supply and Tissue and

forest services Wood products Pulp Paperboard cooking papers
Sales: Sales: Sales: Sales: Sales:
EUR 1.6 billion EUR 0.9 billion EUR 1.3 billion EUR 2.0 billion EUR 1.0 billion
Personnel: Personnel: Personnel: Personnel: Personnel:
900 2,150 850 2,650 2,800

4 18/09/2018 QPR Conference 2018

Metsä Board supply chain
Americas 500 000 tons

• Personnel: ~ 150 including Customer Service

• Order lines: ~ 220 000 per annum
• Mills (paperboard only): 6
• Delivery countries: ~ 100
• Delivery points: ~ 1000
 Process mining journey
2011 2012 2014 2017

Analysis
What – How – Why

Process re-engineering
From “anything goes” to set processes

Continuous Monitoring
Keep processes running as they are designed

Analysis
New opportunities
Changing Environment

18 September 2018 Use cases for process data mining / Matti Ketonen
 Fact finding 2011
Setting base case 1/2
• ERP backbone roughly 10 years old
– Totally different business scope and company structure
– Previous core businesses sold or closed – supply chain tools design obsolete for current core
– Missing levels of planning and poor integration of the existing functions resulting in reactive
and manual order level processing

Original assumption:
” Customers change the product specifications on their orders at the last
minute”
Process data mining based fact:
” We make a lot of delivery time changes on our orders and mostly the
changes are not because of the customer in question”
7 18/09/2018 QPR Conference 2018
 Investment decision 2012
Setting base case 2/2
Customer offering
not clearly defined
”We can do
anything!” Effective
& processes for
managing
”Let’s play safe!” deviations
Process missing
performance ”We’ll work hard
measurement to manage
missing changes.”
”Is my best good
enough?”
8 18/09/2018 QPR Conference 2018
 Process Implementation 2013 - 2014
Continuous improvement of process performance 1/2

Process data mining facilitates

• Standard processes for 5 supply models
– Mill Direct Order Definition of
– Customer Dedicated Stock the supply
– Common Stock models
– Convert to Order
– Vendor Managed Inventory
• Main deviations identified and recovery processes defined

• Ability to drill into deviations from several angles Keeping the

– Mill, product, customer, customer hierarchy, geography, processes
organization …. running as
• Intervention based on planned
– Facts
– Definite knowledge of the scope of any issue

9 18/09/2018 QPR Conference 2018

 Continuous since 2014
Continuous improvement of process performance 2/2
Example: One supply model & One geography

10 18/09/2018 QPR Conference 2018

 New opportunities 2017-
Identifying bottlenecks for process automation 1/2
Automate what ?
This ?

Or this ?

11 18/09/2018 QPR Conference 2018

 New opportunities 2017 -
Identifying bottlenecks for process automation 2/2

Design better Can I split the

processes for process so that
deviations or get parts can be
rid of them? automated?
What does that
mean for How stable is the
customer performance?
experience? When does it
Which work?
customers? Or not?

12 18/09/2018 QPR Conference 2018

 Success?
• Conformant order lines from 40% to 80 %
• Successful mitigation of geographical handicap
• Order lines and volume: + 60%
• Headcount in supply chain remains the same

13 18/09/2018 QPR Conference 2018

 It only gets better since 2011-
Positive mindset change creates
exponential improvements
Management Everyone

Making “I did my best”

decisions

Improving “Thank you for

processes telling me”
14 18/09/2018 QPR Conference 2018
 Future: From physical supply chain to value cycle

Design
Product idea to
package design

Core supply chain

Product from Metsä
Board mill to converter

Converting
• Optimize time from product idea to
From board to package
consumer
• Utilize novel technologies to capture
value cycle knowledge
Distribution
• Improve collaboration capabilities
• Close to real-time operations Production to consumer

18 September 2018 Use cases for process data mining / Matti Ketonen
What was it again?
Thank you !