PROCESS DIGITAL TWINS ARTIFICIAL INTELLIGENCE HOLISTIC APPROACH FAST IMPACT

Make your Manufacturing

more Visible and Transparent,
Efficient and Autonomous
CONTACT: Hervé Camus-Haessler | SVP & CRO | +33 6 16 87 92 54 | +44 75 34 54 15 87 | [email protected] | https://www.linkedin.com/in/hervecamus/
Contents
Core Presentation Appendix

• Video • Technical Architecture

• Executive summary • FAQ: expertise, algo & models, twins, SCADA, IoT
• Pharma – Chem – Food | Golden Production Run
• Introduction: About Neewee Analytics • Aerospace | Quality Gate Behavior
• Problem & Solution: Bodhee® • Petrochemicals | Dehydration run
• Process Digital Twins • Petrochemicals | Steam Turbine
• Applicability & ROI • Steel | Raw Material composition
• Industrial AI: Neewee's Bodhee® • Automotive | Engine failures
• Differentiation • Watchmaking | Process outcomes
• Engagement Model • Electrical | First Pass Yield
• AgriTech | Moisture Control
2
PREFER A VIDEO TO SLIDES?

MEET BODHEE: Logistics Industrial Assets

Integrated and Simplified Analytics
delivered on the Shop Floor

Quality Supply Chain

Manufacturing Inventory

Demand Planning

Here it is on LinkedIn: https://tinyurl.com/bodhee-video

3 © 2021 NEEWEE. All Rights Reserved.

Executive Summary
We make Manufacturing
more Visible & Transparent,
Efficient & Autonomous

4 © 2021 NEEWEE. All Rights Reserved.

1-slide Exec Summary: Industrial AI apps, a Must-Have for sustainable profitable growth
• CHALLENGES: 1) lack of integrated analytics: data locked in silos 2) not on the shop-floor 3) not industrialized
• VISION: to provide a holistic understanding of processes and thus creating visibility across the value chain to impact manufacturing outcomes

• SOLUTION PORTFOLIO: leading-edge Industrial Analytics Software: platform + Apps (supported by enabling Devices and Services)
• 6 APPS: Golden Production Run | Integrated Planning | Work-in-Progress Reduction | Improved Quality| Predictive Maintenance | Production Performance

• HOW: smart combination of Process Digital Twins | Sensors & Industrial Internet of Things | Artificial Intelligence | delivering Predictions & Recommendations
• WHERE: in addition to the Control Tower, also empowering the plant operators on the ground (via industrial HMI interfaces, phone, tablet, PC)
• BENEFITS: improvements on production (>10%); cycle time (>10%); costs (>8%); quality (>10%); yield (>3%)
• DIFFERENTIATION: FAST (6-10 Weeks to Deploy); PRACTICAL (Analytics on the shop-floor); UNIQUE (integrating IIOT, Ops, SCM); HOLISTIC (external factors)

• WHO: proven and well-funded startup; 50+ people, including Industry veterans from Accenture, Capgemini, Igate, IBM, Infosys, Nokia, Microsoft, SAP
• WHEN: Founded in 2014; raised 4 million USD Series A capital in 2019; backed by leading financial firm: IIFL (25 billion USD)
• REFERENCES: Pharmaceuticals (Sanofi) – PetroChemicals (IGPL) – Agro (ITC) – Aero (Airbus) – Auto (Mahindra) – Steel (Tata) – Watches (Titan), …
• ENDORSEMENTS: nominated by Forrester as a Top 5 vendor in Machine Learning / Artificial Intelligence; business value: HIGH; Advice: INVEST

• PILOT: 1) Acquire Data 2) Connect Processes & Machines 3) A.I. correlations, predictions, recommendations 4) M.L. continuous impact (output, quality, cost)
• INVESTMENT: around 50K€ for a pilot; around 15K€ per process unit for the annual software license. Fast payback period. Strong ROI. High NPV.
• A MUST-HAVE for Thought Leaders accelerating their digital transformation!

5 © 2021 NEEWEE. All Rights Reserved.

Value Proposition | double-digit impact
• For: • It provides a key problem-solving capability: going beyond human
• Companies operating in discrete or process / batch / continuous capabilities by using Artificial Intelligence to detect patterns, make
manufacturing, correlations, predict, recommend and control.
• Their decision makers, change agents and various stakeholders • And it enables multiple benefits for customers: analytics that are
• Including their users not only in control tower functions but integrated, industrialized and available even on the shop-floor,
also on the ground: on the shop-floor itself delivering double-digit impact on:

• Who are dissatisfied: with the current lack of integrated Industrial • costs (>8); yield (>2%);
Analytics Software alternatives • quality (>10%); production (>15%);
• Either developed in-house • rejected batches(2-4%); cycle times (1-2%).
• Or sourced from established software houses or startups
• And additional outcomes in terms of business continuity and
• Our solution: Bodhee® by Neewee is establishing a new breed: sustainability.
Industrial AI
• Its core being a leading-edge industrial analytics software
platform: Bodhee® and a set of AI Apps (Golden Run etc.) • Unlike direct and indirect alternatives, it provides a Must-Have
solution that is not only unique and holistic in its approach, but
• Complemented when needed by an Industrial Internet of also practical down to the shop floor, and fast to deploy at scale.
Things aggregator and gateway: Rishee
• And a set of Services: Consulting, Implementation and Support • Thanks to unparalleled features such as a Process Digital Twin able
by Neewee and its partners to capture all parameters, such as process, product, asset, work-
center related, and even external factors such as the supply chain.

6
Introduction
We make Manufacturing
more Visible & Transparent,
Efficient & Autonomous

7 © 2021 NEEWEE. All Rights Reserved.

50+ TEAM | EXPERIENCE + TALENT | DIVERSITY | + STRONG NETWORK

CPO: SUYOG CEO: HARSIMRAT CRO: HERVÉ CTO: NATARAJ CPE: GUCCHU
[email protected]
+33 6 16 87 92 54
https://www.linkedin.com/in/hervecamus/

8 © 2021 NEEWEE. All Rights Reserved.

WE ARE PROUDLY ENDORSED BY

CUSTOMERS ANALYSTS INVESTORS

Industry Leading Customers Top 5 vendor Selected Neewee

Deployed across 4 countries in Machine Learning & AI in category: For 4 million USD
Optimizing 70+ production processes across INVEST with business value: HIGH Investment
Discrete & Process Manufacturing https://www.forrester.com/report/The+Forrester+Tech
+Tide+Smart+Manufacturing+Q2+2020/-/E-RES155295

FORRESTER IIFL

9 © 2021 NEEWEE. All Rights Reserved.

OUR PRESTIGIOUS STARTUP MEMBERSHIPS

10 © 2021 NEEWEE. All Rights Reserved.

 Germany
Start of EU Operations

PRODUCT DEVELOPMENT PILOT WITH HAL AIRBUS AS CUSTOMER

2015-2016 2017 2018

OUR JOURNEY SO FAR

2021+ 2020 2019

SCALE UP GROWTH PHASE (Team size 50) NEW LARGE CUSTOMERS

Broughton
France
› Significant Traction across Discrete Series A Funding of $4m from IIFL
and Process Manufacturing
› EU Team expansion in progress
› North American Foray planned in
next 8-10 months
› Series B planned in next 8 months

11 © 2021 NEEWEE. All Rights Reserved.

WHAT IS OUR APPROACH

Understand the impact of upstream

Use the understanding
01 Digitally Connect the Processes 02 and in-process parameters on 03
downstream outcomes for outcome predictions

Process 1 Process 2 …..n Process n+1

( E.g., Supply Chain, Inventory ) ( Reactors and Lines ) (E.g., Quality, Packing, Dispatch)

We do this by creating process replicas (Process Digital Twins) and We use Machine Learning Algorithms, AI such as Neural Nets,
connect them together to map the interactions between process Reinforcement Learning and Optimization techniques to achieve
parameters and end outcomes this goal

12 © 2021 NEEWEE. All Rights Reserved.

 BODHEE® AI APPS

Work-In-Progress (WIP) REDUCTION GOLDEN PRODUCTION RUN

Achieve consistent Production for your campaigns/
Reduced Buffer Stocks across the production process
batches
Working Capital Reduction 8-10%
Yield Improvement of 5-8%

PRODUCTION PERFORMANCE
MONITORING HIGH IMPACT INTEGRATED PLANNING
USE CASES & Synchronized and Live Production micro

OVERALL
Monitor your Production as well as the key planning for Optimized Manufacturing
Performance Metrics such as OEE with one Improved Capacity Utilization 5-7%

BENEFITS
click
Production Improvement 3-5%

PREDICTIVE MAINTENANCE IMPROVED QUALITY

Reduce the impact on your Production schedules as
well as Product quality by proactively reducing the Get improved quality parameters due to visibility of
Asset Downtimes patterns created by Digital Thread
Maintenance Cost Reduction 15-20% Quality Improvement by 10-12%

13 © 2021 NEEWEE. All Rights Reserved.

HOW APPS IMPACT STRATEGIC GOALS

PLANNING MANUFACTURING ASSET MAINTENANCE

Production Performance
PRODUCTIVITY Monitoring
Integrated Micro Planning Predictive Maintenance
IMPROVEMENT
Golden Production Run

YIELD Golden Production Run

IMPROVEMENT

QUALITY Predictive Quality Predictive Maintenance

COST WIP Reduction Predictive Maintenance

REDUCTION

14 © 2021 NEEWEE. All Rights Reserved.

HOW ARE WE POSITIONED?
Bodhee® AI Apps

Bodhee®
ThingWorx Azure IoT MindSphere AWS IoT SAP CP
Core

RECOMMENDATIONS & CONTROL

CORPORATE
REAL TIME INFORMATION

ERP, PLM, PDM, SCM, CRM

DATA LAKE ( OPTIONAL)

PROCESS, LINES MIDDLEWARE

SCADA DCS MES
& FACTORY
MONITORING, SUPERVISORY CONTROLS, MES, HISTORIANS, APS ETC.

MACHINES

PRODUCTION PROCESSES, PLCs, RFIDs, MACHINE VISION ETC.

We work with your existing platforms & support all standard protocols
15 © 2021 NEEWEE. All Rights Reserved.
OVERALL BENEFITS

>15%
>10% >10%
>8%
>2%

IMPROVED IMPROVED REDUCED IMPROVED INCREASED

PRODUCTION CYCLE TIME COST QUALITY YIELD

16 © 2021 NEEWEE. All Rights Reserved.

SNAPSHOT OF NEEWEE USE CASES / REFERENCES
Why the need What was done What outcomes were delivered How was it measured

INDUSTRY CUSTOMER CONTEXT / ISSUES BY NEEWEE'S BODHEE IMPACT

DISCRETE
Prediction of +20% in work closure;
Aerospace AIRBUS Process bottlenecks, lead times Yield + Quality
quality gate behaviour +15% in quality gate pass

Automotive MAHINDRA Field Failures, Rework Prediction engine Failures Quality +18% quality

Electrical ABB Bottlenecks Improvement of first pass yield First pass yield improvement 25%

Watches TITAN Deviations, bottlenecks, rework Prediction of Process Outcomes Rework reduction 10% cost reduction

PROCESSS
Impact on outcomes due to Predict impact of variations in Raw
Steel TATA STEEL Cost Reduction Being measured
raw material variability material composition

Recommendation of optimized
Steel TATA STEEL Low Yield Productivity Improvement Being measured
process parametres

Recommend a system 91% reduction

Agro ITC Moisture deviation, Yield, Quality Yield + Quality
for moisture control in standard deviation

Shorter production run, Increased Recommendations for extending Production Improvement + 2 Cycles per Month
PetroChem IGPL
energy cost, low productivity dehydration run Cost Reduction Reduction

Unplanned downtime, Failure prediction Production Improvement +

PetroChem IGPL +10%
Increased cost for steam turbine Cost Reduction

17 © 2021 NEEWEE. All Rights Reserved.

Problem & Solution

18 © 2021 NEEWEE. All Rights Reserved.

Our view on Digitalization

Not integrated Not Industrialized & Not Leveraged

analytics Not on the Shop Floor

› Analytics silos still exist in the › The last mile of delivering analytics must › Data Exists through Platforms but the
Manufacturing value chain such as be the shop floor and the data should be ROI is not yet generated
Supply Chain, Asset Management etc. leveraged from both IT and OT systems.
› To realize the value of Digitalization - › The analytics outcomes should be simple
positive impact on Yield, Quality etc. enough to be understood and acted
we must integrate these analytics silos. upon by the manufacturing line worker.

19 © 2021 NEEWEE. All Rights Reserved.

What is the right way to overcome these challenges?

PC

Tablet

Smartphone

Machine HMI

20 © 2021 NEEWEE. All Rights Reserved.

Process Digital Twins

21 © 2021 NEEWEE. All Rights Reserved.

How do we view the Manufacturing Value Chain

Entire Manufacturing Value Chain as a set of connected Processes

Digital Twin 1

Digital Twin 2

Digital Twin 6 Digital Twin 7

Digital Twin 3

Digital Twin 5

Digital Twin 4

22 © 2021 NEEWEE. All Rights Reserved.

Create the Process Digital Twin to understand Interactions

We create a Process Digital Twin by integrating:

Predictions based on current state parameters
parameters include reaction
Process
time, steps, no. of stations on
the line etc.
Process / Line (Reactor Unit, Dehydration Plant, Assembly Line etc..)

parameters: the specifications

Product
of the product such as Raw
Material (RM), Work orders
tasks and sequences.
= Process
Step 1/Stn 1 Step 1/Stn 1 ….. Step n/Stn n
Digital Twin parameters are related to real
Asset
time performance parameters
of units such as temperature,
Mapping of Interactions between Steps/Stations
steam pressure, moisture, etc.

parameters are related to the

Work Center
skill set of people on the
stations, no. of shifts, etc.

Correlations between input and output such as RM Quality, Supply

External Factors
based on non-linear interactions of parameters Chain, Available Inventory, etc.

23 © 2021 NEEWEE. All Rights Reserved.

Recommend Clear Actions to Improve the Outcomes

Process 1 Process 2 Process n Outcomes

Actionable Predictions

Connected Process Twins helps identify the patterns and bottlenecks in the Operations
for Improved Productivity, Reduced Quality Issues and improvement in Overall Yield Improved Outcomes

24 © 2021 NEEWEE. All Rights Reserved.

How do we apply ML & AI

Predictions Output Simulations

– Regression and Classification Recommendation – Regression class of Algorithms
class built on the following Key
Algorithms

› XGBoost › Process Memory based Models › Reinforcement Learning

› Neural Networks › Multi-variate Similarity based › XGBoost
› SVMs modelling › Neural Networks
› Time Series › Reinforcement Learning › Random Forest
› Random Forest

25 © 2021 NEEWEE. All Rights Reserved.

Applicability & ROI

26 © 2021 NEEWEE. All Rights Reserved.

Golden Production
Problem Solution Approach

Fluctuating Suboptimal Quality Issues Rejected Feature Identify the Generate a Validate the
Yield across Production batches Generation Golden Thread Parameter Golden Thread
batches Scorecard

Impact

Production Yield Reduced Rejected Reduced Batch Quality

Improvement Improvement Batches 2-4% Cycle Times 1-2% Improvement
more than 10% 3-5% 3-5%

27 © 2021 NEEWEE. All Rights Reserved.

Industrial AI: Bodhee®

28 © 2021 NEEWEE. All Rights Reserved.

We do this very quickly using Bodhee®
Plant A

Plant D

Plant B

8 Plant E
Weeks

Plant C

29 © 2021 NEEWEE. All Rights Reserved.

How do we do Rapid Deployment

Map Data
Create Node Attach KPIs Attach Model Configure Alerts
to Node

Map the Define Data Catalog Define KPIs and Attach AI and ML Configure Alerts on
Manufacturing and Map Data to Thresholds at Node Models to Nodes to Critical Metrics,
Process in Bodhee each Node. Ingest Type level. Build generate the Digital KPIs, Model
by creating Nodes Data from IT/OT business logic for Twin. Predictions and
(Assets, Machines, systems. each KPIs. Forecasts
Lines) and link Enable Predictions,
them to create Forecasts,
Process Maps Simulations and
Optimizations

Process Digital Twin created

30 © 2021 NEEWEE. All Rights Reserved.

Bodhee is a modular and scalable solution

ERP Systems Other Systems

PROCESS METADATA ADVANCED ANALYTICS MODELS

HARDWARE & SOFTWARE DATA INGESTION LAYER

Micro Controller Boards, MQTT/Kafka Servers
DATA INGESTION CONFIGURATION
REPORTING

DATA STORAGE LAYER

IT
MICRO SERVICES USING REST
PLC CNC Machines
SECURITY

CONSTRAINTS & DEPENDENCIES

TENANT & USER MANAGEMENT

WORKFLOWS, MODEL
OT CONFIGURATION
Gateway
Various Sensors
Edge Computing ADMIN CONSOLE
ALERT CONFIGURATIONS

JAVA PLATFORM

31 © 2021 NEEWEE. All Rights Reserved.

How Bodhee® integrates with IoT/Analytics Platforms
Platform Environment

Platform DB and Services

Other Services Repository

Metadata Analytics Back-End User

Models Enablers Interface

APIs, Data Pipes

Single Source
of Data

Data from Factory Systems Leverage Seamless Micro-Services No Data Communication

existing Integration with allow pick and Duplication via APIs, native
deployments any IoT Platform choose feature Data Pipes

32 © 2021 NEEWEE. All Rights Reserved.

We connect Legacy as well as New Age machines
Connect Communicate Concentrate Convey

Machines Machine Protocol Execution Systems Gateway L2/L3

Interface

› MT Connect
› OPC-UA (Open62541)
› OPC Classic
PLC SCADA
› FOCAS via MT Connect
› Proprietary Protocols
Ethernet › Haas CNC Protocol MES Cloud
› Siemens 840D
› DNC 2 / LVDS (Legacy
Special Robot Machine Support)
Purpose M/c › PLC Omron
Non- › Beckhoff
Integrated › Modbus
Serial Custom Systems ERP Systems Other Systems
Overlay › Siemens STEP 7
Sensors

Recommend Control Analyze Monitor

The foundation of Remote Operations begins The data flow should be seamless and Important considerations include data
with the connectivity of different types of support bi-directional communication latency and security.
assets and manufacturing systems. between Machines and L2/L3 systems for
control & recommendations.
33 © 2021 NEEWEE. All Rights Reserved.
How is Bodhee® different

34 © 2021 NEEWEE. All Rights Reserved.

Why Bodhee® Core is a Must-Have solution

Fast Deployment Practical approach Layered approach of Model connectivity

time of 6-8 Weeks of delivering the building a Process enabled to solve
for a medium outcomes on Digital Twin real-life business
complexity Process the Shop Floor problems

35 © 2021 NEEWEE. All Rights Reserved.

Why Bodhee® Apps are Must-Have solutions

Complement Integrate Score Deploy Holistic approach

Engineering with existing over the existing Fast: 6-8 Weeks Apps built on the
Models platforms Statistical Software foundation of a
and Excels Process Digital
Twin

36 © 2021 NEEWEE. All Rights Reserved.

Engagement Model

37 © 2021 NEEWEE. All Rights Reserved.

 STEPS TO SUCCESSFUL IMPACT
5
RUN the solution
and impact KPIs
4
Implementation
ADJUST if needed
3
Implementation AI App Software
TRY for FREE
License (annual)
2
Testing the solution To add a remaining
BUILD the PILOT
process component
1 IF not already in
SITE Implementation the scope of the
IDENTIFICATION pilot
Understand the
context of Use Case
Application in
detail

DURATION 8 Hours 6 to 8 weeks 2 to 4 weeks

38 © 2021 NEEWEE. All Rights Reserved.

TYPICAL RESPONSIBILITY MATRIX

Process Understanding & PoC

Business Value Discovery

Data Analysis

Bodhee Configuration

Infrastructure Set-up

User Trials & Training

Customer Neewee
39 © 2021 NEEWEE. All Rights Reserved.
FEW BASIC DETAILS AND WE ARE READY TO ROCK..

What is the Which High Level Do you want the Will it be a On-
selected use case Processes/Plants Process Overview Use Case to be Prem/Private
from Bodhee® AI you want to will be great! integrated with Cloud/Neewee
Apps and what implement the your Existing IoT Cloud
benefit you use case? platform? If yes, Deployment?
expect to derive? which one would
that be?

…And your proposal is on its way

40 © 2021 NEEWEE. All Rights Reserved.

HOW DO WE KICK START THE ENGAGEMENT

Audience Expectations

Planned Discussion on Plant Manager

Overall Objective of the use case from Plant Manager’s perspective and what
are his expectations as outcomes
4-6 Hours spread over Production Manager Day to Day issues and how outcomes will help
2 Sessions. This can be
Quality Teams Quality challenges and how it is measured currently
either on-site or virtual
Performance Improvement
meeting. In addition to Group
Key areas of concerns and what is being done currently

understanding the Planning Team Current Process Understanding, bottlenecks and areas of Improvements

processes, we would Maintenance/

Data Availability and any blind spots in terms of data. Issues related to Assets
Instrumentation Team
like to talk to following
Industry 4.0 Team Overall Roadmap of Digitalization and where this use case fits in
groups
IT Teams IT Landscape and compliance guidelines

Details questions and expectations will be sent 7 days prior to these meetings

41 © 2021 NEEWEE. All Rights Reserved.

THANK YOU

PARIS EINDHOVEN BANGALORE

+33 (0)6 16 87 92 54 +31 (0) 40 851 7524 +91 (80) 4202 4519

[email protected]

Neewee Confidential
Bodhee Golden Production Run
Architecture

43 © 2021 NEEWEE. All Rights Reserved.

ARCHITECTURE | TECHNICAL
Web Application HTML+AngularJS+D3JS+Bootstrap

API
Getaway Play-Java
Docker Repository Cluster Management

Play-Java

Scheduler
Reverse Proxy Server Nginx
Process UI Configurator Security KPI Manager Admin Service Azkaban
Configurator

ERP Systems Other Systems Calculation Engine

MQTT Broker Configuration Data

(Postgres)

PLC CNC Machines Connectors Data Ingestion

Service
Gateway MQTT
Edge
Computing
HTTP

Various Sensors FTP Kafka Broker Business Data

(Postgres)
COAP
Apache Spark Cluster
Akka + Scala Akka + Scala Spark + Scala

44 © 2021 NEEWEE. All Rights Reserved.

 ARCHITECTURE | DEPLOYMENT
Business
Deployment Support
Team
Team Team
CLI + Web App

Bodhee End Point Floating IP

Reverse Proxy & SSL MQTT Broker Reverse Proxy & SSL MQTT Broker Reverse Proxy & SSL MQTT Broker

Monitoring Agent Monitoring Agent Monitoring Agent

Cluster Manager Cluster Manager Cluster Manager

Alert Calculation
Web App APIs Security
Engine Engine
Bodhee API Analytics - Analytics -
Services Ingestion Training
Gateway Prediction

VM1 VM2 VM3 VMs

45 © 2021 NEEWEE. All Rights Reserved.

ARCHITECTURE | SECURITY OVERVIEW

Data Transfer Data Storage Application Cloud

Role Based Gateway

SSL Encryption Authentication + Reverse Proxy IP Whitelisting Data Isolation Bastion Host
Authorization Firewall

Bodhee Security Management

46 © 2021 NEEWEE. All Rights Reserved.

ARCHITECTURE | DATA TRANSFER / STORAGE MECHANISM

HTTPS

bodhee HTTPS VPC

Security HTTPS

Authentication & Reverse Proxy

Authorization Server
bodhee API
HTTPS Gateway

HTTPS bodhee
HTTPS HTTPS Service-N

bodhee bodhee bodhee Encryption

Service-1 Service-2 Service-3

Business Data
TCP:SSL (Postgres)
With Authentication

MQTT:SSL
With SSL with Authentication
Authentication Configuration
MQTT Broker Kafka Broker Data
(Postgres)
Gateway Edge
Computing

47 © 2021 NEEWEE. All Rights Reserved.

ARCHITECTURE | APPLICATION ACCESS
Username/Password based
Authentication + SSO

Administration &
Monitoring app access Application Access from
only from Neewee pre-approved IP address
Environment

Role based
SSL Encryption Authorization using Data
Filters

48 © 2021 NEEWEE. All Rights Reserved.

ARCHITECTURE | DEPLOYMENT ENVIRONMENT ACCESS

Neewee Environment VPC

TCP:SSL Business Data
With Authentication (Postgres)
Encryption

TCP:SSL
With Authentication

SSH BASTION HOST Configuration Data

(Certificate) + SSH Firewall (Progress)
FIREWALL bodhee vm-1
Firewall
bodhee vm-2
Firewall
bodhee Admin bodhee vm-N
Private Instances

Bastion Host for VM access Access from Neewee No direct access to VM

via SSH + Certificate Environment outside VPC

49 © 2021 NEEWEE. All Rights Reserved.

FAQ

Frequently Asked
Questions

50 © 2021 NEEWEE. All Rights Reserved.

HOW DO WE NARROW THE DOMAIN EXPERTISE GAP?

DOMAIN EXPERTISE PROCESS DISCOVERY

Customer Neewee
Data
Discovery
› Engineering Principles › Mapping Interactions
› Working Experience between Process Steps, Sub-
processes
› Associations

51 © 2021 NEEWEE. All Rights Reserved.

WHAT IS THE DIFFERENCE BETWEEN AN ALGORITHM & A MODEL?

ALGORITHM MODEL

Algorithm 1 Algorithm 2
Input
Output
f(x….y….) Parameters Algorithm 4 Algorithm 5
of Model
(Configurable)
Algorithm 3

› Function acting
on input
System represented Multi-steps achieved Majority of systems in
› Usually Single
in Mathematical form by stitching Manufacturing need
Step
algorithms together multi-step approach

52 © 2021 NEEWEE. All Rights Reserved.

HOW ARE BODHEE® MODELS DESIGNED?

Open Source Open Source

Algorithm 1 Algorithm 2
Input
Open Source Open Source Output
Parameters
Algorithm 4 Algorithm 5 of Model
(Configurable)
Proprietary
Algorithm 3

TYPICAL MULTI-STEP MODEL

Configurable Bodhee
Features of Proprietary Algorithm Models/Algorithms

Encompasses Industry Readily available as compared Transparent and Configurable

01 best practices
02 to being built from scratch

53 © 2021 NEEWEE. All Rights Reserved.

HOW ARE BODHEE® MODELS RESILIENT?

CONFIGURABLE CHANGES

Addition of new Change in Work

Station/Step Configurations
Self
Adjusting
Model
Changes in
Takt Changes
Compositions/Grades

54 © 2021 NEEWEE. All Rights Reserved.

Connecting Twins

55 © 2020 NEEWEE. All Rights Reserved.

WHAT ARE THE Aspects Challenge Description
CHALLENGES IN
CONNECTING For e.g. Data Model to Support
DIGITAL Comprehensive Data Model
› Processes
› Hierarchies
TWINS? › Sequences of executions

› Propagation mechanism for forecasting

impacts
Connectivity of Models › Flexible inputs (immediate prior as well as
hierarchical up steams)

› Alarm handling mechanisms in case of

Alarms on Process Deviations process deviation alerts

› Strategy on handling errors coming in due

Managing Butterfly Effect to model inaccuracies

56 © 2021 NEEWEE. All Rights Reserved.

WE KNOW HOW TO CONNECT PROCESSES
This is a problem most organizations struggle with Plant - X

Product State 0 Process Digital Twin Product State 1 Finished Product

1. Base Raw Material 1. Process Parameter

2. Sub Assembly
2. Product Parameter
Characteristics:
3. Asset Parameter
1. Quality
2. Chemical Composition 4. Work Center
3. Structural Parameter
4. Thermal
5. Supplier Behaviour 5. RM Quality / Supply
(3rd Party vendor or Chain Parameters
different plant)

Time Dimension as Reference

for Process Manufacturing

57 © 2021 NEEWEE. All Rights Reserved.

CONNECTING TWINS – ADDITIVE PROCESS TWINS

Additive Digital Twins

Critical Step: Identification of
Additive Parameters

Ex: Process of a major

component which includes
multiple sub processes from
different workstations.

58 © 2021 NEEWEE. All Rights Reserved.

EXAMPLE OF MULTI-STEP CONNECTIVITY

Multi-step Digital Twins

Critical Step: Identification of
additional Parameters as well as
defining the outcomes

Ex: Process of a major component

which includes multiple sub-processes
from different workstations. Also,
need to define if the outcome to be
tracked is at task level, sub-process
level or at the end of the line

59 © 2021 NEEWEE. All Rights Reserved.

How Bodhee complements
IoT/analytics platforms,
and SCADA

60 © 2021 NEEWEE. All Rights Reserved.

WE CAN WORK ALONGSIDE OTHER IOT/ANALYTICS PLATFORMS
How does Bodhee® work along
IoT Platforms Bodhee®
with IoT/Analytics Platforms
Bodhee® has a ready-made
Do-It-Yourself model with burden on the
manufacturing specific model
Build Vs. Ready-made customer to get the required resources and
repository for end-to-end outcome › Can reuse/leverage/get
time to build
driven use cases deployed on top of any
Connecting Processes is a technically
IoT/Analytics platform
Connecting Processes challenging task and requires significant Enabled by default already available
technical expertise as well as effort
› Bodhee® Models can easily
Majority of the models which are available are integrate with existing IoT
Single Step Vs. Multi-Step suitable for a single-step approach (e.g. Time Easy Single and Multi-steps model platforms
Models Series, Anomaly Detection) and cannot be used configuration based on the use case
for Process twins as is › Micro-services-based
Bodhee® has evolved with Process Architecture allows pick and
IoT platforms have evolved with Asset-centric choose for Bodhee Modules
centric vision and hence hierarchies
vision. Horizontal scaling is easy, but scaling
Vertical Scaling vertically across processes (e.g. sub-process to
can be defined and built and connected
at analytics level thus enabling the › No duplication of data:
process to line to plant hierarchy) is challenging Bodhee® can connect to a
scaling
single source of truth via APIs
For making manufacturing analytics enterprise
wide, additional modules such as
provided by the Platform
Bodhee® has built-in features for
Peripheral Modules Configurable/Personalized UI, Robust and
enterprise level deployments › Cloud Provider Agnostic
scalable data model supporting these features
are time consuming and challenging to build

61 © 2021 NEEWEE. All Rights Reserved.

 PROCESS DIGITAL TWINS ARTIFICIAL INTELLIGENCE HOLISTIC APPROACH FAST IMPACT

GOLDEN PRODUCTION
Approach Note
THE NEED FOR A NEW APPROACH
TO GOLDEN BATCH
The R&D Golden Batch does not consider process variabilities
for e.g., raw material variations, asset performance “I already have a
deterioration, ambient conditions, etc.
Golden Batch, why do
Multi-step and often complex processes require multi-
variable analysis over long and variable batch cycle times
I need another one? “
(BCT). This evaluation is a difficult one and requires
significant manual effort.

Most of this analysis is offline and dated and may not be

most relevant to current production runs.

In terms of alerts and recommendations, real-time guidance

is not given to operators.

Subject Matter Experts understand areas where processes

deviate, but in terms of variations in parameters that cause
them, they are not sure of the exact root cause.

63 © 2021 NEEWEE. All Rights Reserved.

IMPACT OF INADEQUACIES IN CURRENT APPROACH
Complex Interactions between Processes
Raw Material Variabilities
Long batch Timing

Variant
1 Fluctuating Yield across
batches

Suboptimal Production
Variant
2
Quality Issues
RM Loader Blender Emulsifier Reactor
Rejected batches
Variant
Thousands of Parameters interacting and 3
varying simultaneously

Fig: Illustrative Chemical Process

64 © 2021 NEEWEE. All Rights Reserved.

OUR GOLDEN
PRODUCTION SOLUTION

65 © 2021 NEEWEE. All Rights Reserved.

 HOW DO WE GET TO GOLDEN PRODUCTION

FEATURE GENERATION IDENTIFY THE GENERATE PARAMETER VALIDATE THE GOLDEN ENJOY THE GOLDEN
GOLDEN THREAD SCORECARDS THREAD PRODUCTION
› Feature Engineering by
› Statistical Pattern › Identification of Prime › Validation with Domain › Get Consistent Production
consideration of long BCT
Identification and correlations Contributors of Variations in Experts and Model fine-tuning Runs from your processes
(Batch Cycle Time)
between varying parameters output metrics
› Integration of a) Compositions
with the output produced For
b) In-Process Parameters c)
e.g., identifying the of
Yield d) Production Quantity e)
variation of parameters such
Quality Results f) Asset Down
as pH of product, ammonia
times and Maintenance
volume, adjusting pH to make,
› Algorithmic quality checks to
oxygen, temperature etc. with
ensure we process good and
the yield of the antibiotic
relevant data
› We handle the complexities
arising out of Time dimensions
and complex process
interactions in terms of data
preparation

66 © 2021 NEEWEE. All Rights Reserved.

STEP 1.1 – GENERATE FEATURES (PROCESS VIEW)
Illustrative process

Upstream ERPs, eLog

Reactor Blender Distillation
Systems Books

Data Ingestion Data Ingestion Data Ingestion

Raw Material Features Quality Reports for Raw Material (RM), In-
pH value, Temperature, Pressure
Medium Attributes Process, SFGs (Semi-Finished Goods), FGs
Reflux Quantity, Agitation Speed
Medium Quantity (Finished Goods), Asset Down Times,
Raw Material Loading Speed, Batch Cycle Times, Volume, Density, Viscosity
Strain Parameters Maintenance schedules

Feature Generation Process

Automated Data Quality Checks

Harmonized and Integrated Data

67 © 2021 NEEWEE. All Rights Reserved.
STEP 1.2 – GENERATE FEATURES (SYSTEM VIEW)

PLC SCADA

Non-Integrated
Overlay Sensors

Operational DCS ERP Lab Quality eLogBooks Other Source

Systems / Yield Data Systems
Data Ingestion

Feature Generation Process

Automated Data Quality Checks

Harmonized and Integrated Data

68 © 2021 NEEWEE. All Rights Reserved.
STEP 2 – IDENTIFY THE GOLDEN THREAD

Combination of
› Setting up of goal in terms Compositions,
of Yield/Production/Quality Parameters
contributing to
› Identification of good bad batch
signatures
Example of a Pattern

› Signature comparison
(Generation of similarity
scores)

› Ranking of recurring
patterns

› Correlations with
Parameters Good Pattern Bad Pattern
STEP 1
INTEGRATED DATA
PROCESS MEMORY BASED ALGORITHMS

Illustrative Non-Linear Pattern Signatures for Good and Bad Run

69 © 2021 NEEWEE. All Rights Reserved.

STEP 3 – GENERATE PARAMETER SCORECARDS

List of
Parameters and
their impact on
the output
› Identify the combination of
maximum impact
parameters in terms of
significance scores and
magnitude of variations
Example of a Significant scores
when the output of the run
is not up to the mark.

STEP 2
PATTERNS
e.g. Identification of Contributors and scores

70 © 2021 NEEWEE. All Rights Reserved.

STEP 4 – VALIDATING THE GOLDEN THREAD

Golden
Production
Parameters

› Validation with Subject

Matter Experts on
Parameter relevance and
associated importance with
output Metrics
For e.g.

STEP 3
PARAMETERS
Complementing the Subject
Matter Experts with the Data

71 © 2021 NEEWEE. All Rights Reserved.

STEP 5 – ENJOY GOLDEN PRODUCTION ...EVERY TIME!

Golden Run Real Time Guidance Alerts for Production

Every Time System for Operators Leadership Teams

Consistent Quality Online Available Simulations

Outcomes System

72 © 2021 NEEWEE. All Rights Reserved.

WHAT IS THE GOLDEN PRODUCTION ROI

8-10%
3-5% 3-5%

2-4%

1-2%

Production Improvement Yield Improvement Reduced Rejected Reduced Batch Quality Parameter
Batches Cycle Times Improvement

73 © 2021 NEEWEE. All Rights Reserved.

HOW DOES IT COMPLEMENT CHEMISTRY MODELS

Characteristics of Chemical Model

Chemical / Mathematical Model › Based on principles of Chemistry and
Process + Asset complemented by Thermodynamics principles
(Based on First Principles)
› Simplify complex situations
› R&D Golden Batches are generally based on
these models
PLANT

Business Impact
Feature Golden Production
Process + Asset
Engineering Model

Characteristics of Golden Production Model

› Works on real-time and actual production data › Considers various system variabilities like wear & tear
› Time Dimension (Batch Cycle Time) considered during the analysis › Can integrate output of various Engineering / Mathematical Model
› Multilayered – Accommodates process complexities

74 © 2021 NEEWEE. All Rights Reserved.

HOW WE Statistical Software / Golden Production
Excel Macros Solution
SCORE OVER
STATISTICAL Build vs. Do-It-Yourself model with burden on the
customer to get the required resources
Ready-made Model specifically designed
to have consistent Golden Production
Ready-made
SOFTWARE and time to build the statistical Models every time

AND EXCEL Connecting Connecting Processes is a technically

challenging task and requires significant Enabled by default
MACROS? Processes technical expertise as well as effort

Majority of the models which are available

Single Step vs. Easy Single and Multi-steps model
are suitable for a single-step approach and
Multi-Step Models configuration based on the use case
cannot be used for Process twins as is

Scaling is a challenge as we need to rebuild Easy Process Modeling for the new process
Scaling it for each process taking time with one-click reusability

Ready-Made Process/Batch-centric feature

Feature
Feature generation modules must be built generation library for easy reuse and
Generation configuration

75 © 2021 NEEWEE. All Rights Reserved.

WHAT IS OUR
DIFFERENTIATION Impact on the
outcomes in 6-8 Weeks

Run Time guidance to

Specialized
operators on parameters
Algorithm specific
settings to produce
for Golden Batch
optimal outcomes

Ability to handle End-to-end solution —

Variabilities creeping Hardware Integration to
in Run time Golden Batch actionable

76 © 2021 NEEWEE. All Rights Reserved.

 01 Golden Parameters across Process

Optimization/Recommendations for current batch parameters

02 with reference to Golden Batch and Current Variability

Variant
1
WHAT IS
THE END Consistent,
Optimal Capacity
STATE Variant
2 Yield with
desired Quality
RM Loader Blender Emulsifier Reactor

Variant
3

03 Predictions of Current Batch for Yield, Production and Quality

Fig: Illustrative Chemical Process

77 © 2021 NEEWEE. All Rights Reserved.
THANK YOU

PARIS EINDHOVEN BANGALORE

+33 (0)6 16 87 92 54 +31 (0) 40 851 7524 +91 (80) 4202 4519

[email protected]

Neewee Confidential
Available resources
• Other Use Cases: https://app.box.com/v/neewee-case-studies
• Aerospace | Quality Gate Behavior
• Petrochemicals | Dehydration run
• Petrochemicals | Steam Turbine
• Steel | Raw Material composition
• Automotive | Engine failures
• Watchmaking | Process outcomes
• Electrical | First Pass Yield
• AgriTech | Moisture Control

• Special on Remote/Autonomous Operations: https://app.box.com/v/neewee-remote-operations-short

• Introduction in High-Resolution: https://app.box.com/v/neewee-intro-full

• Golden Run Production: https://app.box.com/v/neewee-Golden-Run

79 © 2021 NEEWEE. All Rights Reserved.