digital modelling for logistics decision making

© The AnyLogic Company | the.anylogic.company 1

digital modelling for logistics decision making


• 10.00 – 10.20 The AnyLogic Company: Introduction• 10.20 – 11.30 anyLogistix: introduction• 11.30 – 12.00 Coffee break• 12.00 – 13.00 anyLogistix: demo• 13.00 – 13.30 Prof. Dr. Dmitry Ivanov, Berlin School of Economics and

Law: Managing supply chain resilience with simulation-based digital twins

• 13.30 – 14.30 Lunch• 14.30 – 15.30 AnyLogic: Introduction• 15.30 – 16.00 Coffee break• 16.00 – 17.00 AnyLogic: demo• 17.00 – 17.30 Dr. Harry Kestenbaum, SimPlan: Business simulation case

studies

agenda

© The AnyLogic Company | the.anylogic.company

The AnyLogic Company

MunichFebruary 13, 2019

Timofey Popkov


quick facts

• Founded in 1998• 100% software, no services• B2B, key industries:

Consulting Distribution, wholesale, and retail Manufacturing Mining Transportation and storage Healthcare Defense and security Information and communication

• 20,000+ users, 900+ commercial organizations, 1,600+ universities• Worldwide market leader in dynamic simulation


technologypharmahealthcare

railroadsglobal consulting

energy

automotiveaerospace

defensefundamental researchsupply chains

logistics

miningoil & gas

finance consumer goods

selected clients


The AnyLogic geography

Genoa

TECHSIM

Blue Stallion Technologies

DSE Consulting

SIMPLAN

Fair Dynamics

IBNLDM

AtWorth

CarilaTech TechSupport MgmtPitotech

Simlogy

TSG Consulting

Advisian

MaxSoft

Sela Digital

Techenware

DecisionesLogisticas

Zecctron

St.Petersburg

ChicagoParis


our products

General purpose dynamicsimulation software

Areas of application:• Material handling• Pedestrian flows• Road traffic• Processes• Rail logistics• Mining• Supply Chains

Platform for model execution Supply chain optimization andsimulation solution

Areas of application:• Greenfield analysis• Network design & optimization• Master planning • Transport optimization• Inventory optimization• Risk analysis• Digital twin


the team

© The AnyLogic Company | www.anylogistix.com

Design Your Supply Chain.Run with Digital Twin.

DesignOptimize Experiment Innovate

Timofey Popkov

10© The AnyLogic Company | www.anylogistix.com 10

Your Supply Chain is Constantly Being EvaluatedWhat should be

the design of our supply chain?

How should we deploytransportation policies?

Which resourcesdo we need?

Which business processes should we implement

to reach operation excellence

What if we introducenew products?

How can I improve my existingsupply chain?What if I change inventory,

transportation or sourcingpolicies?

Where is my SupplyChain susceptible

to risk?

What if our workersgo on strike?

What capacity do we need?What service

level we willbe able to provide

our customers? What manufacturingcapacity do we need?

Is there a risk ofbullwhip effect inmy supply chain?

What if we lose a key supplier?

Should we Make or should we Buy?

How much to produce and

where to store?


Number of supply chainProblems by level of detail

Pyramid Supply Chain Analytics Problems

Low AbstractionMore Detail

Dynamic (time)

High AbstractionLess Detail

Static

LocationsFlowsPeriodsLinear DependenciesParameter Aggregation

Where to Build DCsWhere to Stock ProductsMaster Planning

Inside 4 walls ProcessesInside 4 walls ResourcesInside 4 walls Logic

Dynamics (time)RandomnessParameters DetailingNetwork ProcessesNetwork ResourcesNetwork Logic

How “Inside” influences “Outside”“Inside” Resources OptimizationProduction Planning“Inside” Bottlenecks Identification Risk Analysis

Transportation PlanningInventory & Sourcing Policy PlanningFleet Size OptimizationService Level & Capacity Estimation Bullwhip AnalysisRisk AnalysisResources Planning & Optimization

Level of Detail Problems Addressed


Common Methods Used to Solve the Problems

Spreadsheets

Pros: Everybody knows Excel

Cons: Application area is limited

Dynamic Simulation

Pros: Detailed SC modeling. Considers dynamics, specific logic and randomness

Cons: • By nature simulation-based optimization

may take significant time• Difficult to use w/o a specialized UI

Analytical Optimization

Pros:Used to address SC problems which can be described with set of linear equations

Cons:• Generalization• Represent the supply chain as a

continuous flow model• Difficult to use w/o a specialized UI


Analytical Methods: Problems That Can be Addressed

• Supply Chain Design & GFA Where to locate our facilities? Identify throughput of the facilities Determine product flows

• Master Planning by Period: Where to produce or stock products? Determine optimal quantities to produce

and order Throughput requirements

• Transportation Fleet Size Estimations

Answers the question;“What is the plan?”


Analytical ModelsSupply Chain - ACTUAL

Time (dynamic)Products

OrdersTransportation

ResourcesFacilities

RandomnessNetwork Level Logic/ProcessesInside 4 Walls Logic/Processes

Time = PeriodProducts = Flows per PeriodOrders = Demand Aggregated per PeriodResources = Flows per PeriodTransportation = Cost per PeriodFacilities = Throughput per PeriodRandomnessNetwork Level Logic/ProcessesInside 4 Walls Logic/Processes

Labor/period Goods/period

ModelingProcess

Consultant

Supply Chain - MODEL

OPTIMIZATION


Demo: Network Optimization


Analytical Methods: Key Points

• Analytical Optimization is well equipped for large-scale data intensive problems

• Analytical Models are not able to properly describe the “reality” of the Supply Chain. The approach forces you to make simplifications and assumptions. Ex: Assuming all the events are uniformly distributed within a period

• Attributes not considered: Dynamics (Time) Randomness Supply chain operating Logic

• The approach is like a “BLACK BOX” You do not know how the solver works or why a problem is infeasible


• A Simulation model is described as a set of LOGICAL RULES:

• Simulation is the process of executing the LOGICAL RULES over time

• The output of a simulation is the behavior of a system over time

What is Dynamic Simulation Modeling

If factory has enough ordersfor 3000 units then production

is started

If supplier workersare on strike it will

not process the orders

Customer places orders for 300 units

per day 4 days a week

Initial inventory is 3000.If inventory is below 2000

the DC orders a batchof 3000 units If raw materials stock drops

below 100m3 order 300m3

ModelingtimeDay 1

order for 300 units

Day 5

Stock < 2000order 3000

Day 6

Startsproduction

Day 7

Stock < 100m3order 300m3

Day 8

Strike, do not process the order

Day 2

order for 300 units

Day 3

order for 300 units

Day 4

order for 300 units


Dynamic Simulation: Problems That Can be Addressed

• Examination\implementation of the Supply Chain What if the answer suggested by network

optimization is not feasible How to implement the solution suggested by

network optimization

• Understanding of how the Supply Chain operates You need to understand how your Supply Chain

functions at detailed level to manage it effectively and efficiently

• Experimenting with\Improving the Supply Chain You already performed a high-level design of the

Supply Chain and want to know how improve operational performance

You have an idea of how to improve the supply chain efficiency and want to test it prior to implementation

• Risk assessment What are the risks related to SC design?

A Simulation Model is a DIGITAL TWIN of your supply chain!

Answering the question:“How to reach operationalexcellence in yourSupply Chain?”


Dynamic Simulation Model

Supply Chain - ACTUAL

ModelingProcess

Consultant Time = Actual TimeProducts = Products & ParametersOrders = Orders & ParametersResources = Any Resource, Any LogicTransportation = Any RulesFacilities = Location, CapacityRandomness = anyNetwork Level Logic/ProcessesInside 4 Walls Logic/Processes

ProcessesResourcesLogic…

Time (dynamic)Products

OrdersTransportation

ResourcesFacilities

RandomnessNetwork Level Logic/ProcessesInside 4 Walls Logic/Processes

Supply Chain - MODEL

SIMULATION


Demo: Dynamic Simulation


Dynamic Simulation: Key Points

• Dynamic simulation methods allow you to model a Supply Chain with limitless detail, including “Inside the 4 walls”

• Simulation-based optimization is fundamentally different from mathematical-optimization The optimization engine is a separate program working in conjunction

with the Simulation Model – measuring the model output and generating a new set of input parameters

• You must be careful when deciding on the level of abstraction to build your Dynamic Simulation model - too many details may result in slower performance


Number of supply chainProblems by level of detail



Dynamic (time)


Static

LocationsFlowsPeriodsLinear DependenciesParameter Aggregation

Where to Build DCs (GFA)Where to Stock Products (Net Opt)Master Planning

AnalyticalMethods

DynamicSimulation Methods



How “Inside” influences “Outside”“Inside” Resources OptimizationProduction Planning“Inside” Bottlenecks Identification Risk Analysis

Transportation PlanningInventory & Sourcing Policy PlanningFleet Size OptimizationService Level & Capacity Estimation Bullwhip AnalysisRisk AnalysisResources Planning & Optimization

Level of Detail Problems Addressed




Dynamic (time)


Static

LocationsFlowsPeriodsLinear DependenciesParameter Aggregation Analytical

Methods




Opp

ortu

nitie

s fo

r Inn

ovat

ion

The MORE LEAN you are trying to be, the more need for DYNAMIC SIMULATION you will have

Less Lean

More Lean

Level of Detail


Example 1: Supply Chain Design

+ Aggregated demand+ Transportation cost (m3 per km)

Analytical Methods

(GFA)Areas to place the DCs

Details MethodResult

+ Possible facilities locations+ Real roads+ Sites throughput

Configurations of supplychain ranked by cost

Analytical Methods(CPLEX)

+ Sites capacity (not throughput!)+ Inventory policies+ Sourcing policies

Analytical Methods+

Dynamic Simulation

Configurations of supplychain ranked by

service level


Example 2: Omnichannel Supply Chain


+ Processes inside 4 walls+ Sites capacity+ Working hours+ Resource sharing+ Refunds/returns+ Inventory policies+ Transportation policies

How should the omnichannel supply

chain operate? (service level, resourcescapacity and utilization,

reliability…)

DynamicSimulation

+ Aggregated demand / month+ Production throughput / month+ Transportation cost (m3 / km)

Analytical Methods(GFA, NO)

Supply Chain Configurationsranked by cost


Example 3: Risk Analysis

+ Operational risks+ Demand changes+ Lead time changes+ Supplier reliability+ Strikes

DynamicSimulation

Supply chain robustnessService levelCostsUtilization


+ Disruption risks+ Natural disasters+ Terrorism+ Political and financial crisis

Supply chain robustnessService levelCosts

DynamicSimulation


• Definition by Gartner: A digital twin is a digital representation of a real-world entity or system

• Some definitions from Wikipedia: A Digital Twin is an integrated multiphysics, multiscale, probabilistic

simulation of an as-built vehicle or system that uses the best available physical models, sensor updates, fleet history, etc., to mirror the life of its corresponding flying twin

Coupled model of the real machine that operates in the cloud platformand simulates the health condition with an integrated knowledge from both data driven analytical algorithms as well as other available physical knowledge

A dynamic virtual representation of a physical object or system across its lifecycle, using real-time data to enable understanding, learning and reasoning

…

• The words most often used to describe a Digital Twin Simulation, real-time data, dynamic, understanding, learning, reasoning

What is a Digital Twin


• A supply chain digital twin is a detailed simulation model of the actual supply chain which predicts the behavior and dynamics of the supply chain to make tactical or operational decisions: Tactical decisions are mostly related to how to organize processes in your supply

chain. It may require to do the simulation of few months or years Operational decisions are mostly related to identification of the problems and

analysis of the corrections. It may require to do the simulation of few days or weeks.

• The data you need to run supply chain digital twin depends on the answer you are looking for

A Digital Twin in Supply Chain

Detailed Supply ChainSimulation Model

Supply Chain DynamicsForecast

Understanding

Learning

ReasoningReal Time Data

\Snapshot

Digital Twin


• Optimization, analytics, AI can be a part of digital twin

• Simulation is used to forecast the dynamics of a supply chain

• Optimization, analytics, AI are used by a simulation model to make decisions in certain situations

Digital Twin: Is There a Space for Optimization

Supply Chain Digital Twin

Simulation modelEarthquake destroyed

DCs network

Day 1 Day 2 Day 3 Day 4 Day 5 Day N

Optimization

Adjusted SCstructure


Control Tower

Digital Twin and Control Tower

Data

Visualization (BI, alerts, dashboard…)

Supply ChainDigital Twin

ManufacturingDigital Twin

Analytics

Decision Support/Forecast

…

Current stateof the supply chain

Current stateof the facilities

Finances

Current State/History

…

Tasks

Cases

Actions

Task and CaseManagement

…

Collaboration


• A digital twin is a detailed simulation model of a supply chain A digital twin is used to forecast the behavior of the supply chain, predict abnormal

situation and work out action plan

• Real Time Data/Snapshot A digital twin uses real time data to do the forecast

• Integration with Control Tower/BI In many cases a digital twin is part of a “bigger thing” e.g. control tower and should be

able to be integrated with this

• Notifications/alarms about abnormal situations A digital twin should allow to define what abnormal behavior is and send notifications

about critical/abnormal situations which may happen Example:

If the inventory level drops below 3000 units send notification “critical level of inventory at DC1”

• Triggers A digital twin should allow the definition of automatic actions for some events Example:

If a supplier is on the strike serve only high priority customers

• Test an action plan A digital twin should allow you to test actions to efficiently manage your supply chain

What Makes a Simulation Model a Digital Twin


Example 4: Budgeting & Cash Flow

+ Products flows+ Transportation cost (m3 per km)+ Flows processing cost+ Labor and facilities cost

ExcelBudget,Cost-to-Serve


+ Time-based costs e.g.:+ Resources cost+ Operations cost+ Handling\carrying cost+ Idle cost

Activity-Based Budget

Analytical Methods(linear, flows)

DynamicSimulation

+ Payment terms+ Randomness+ Production processes+ Production schedule

DynamicSimulation

Cash-to-ServeWhen and how much

cash do you needto operate


Example 5: Supply Chain Behavior Forecast

+ Supply chain structure+ Supply chain logic+ Supply chain resources+ Factory processes/logic/resources+ DCs processes/logic/resources+ Customers behavior+ Transportation rules/resources

SupplyChainDigitalTwin

Short term forecastof the supply chain behavior to identify unusual/dangerous situation and find out

the ways to manage them





Dynamic (time)


Static

LocationsFlowsLinear DependenciesContinuousParameter Aggregation Analytical

Methods


Level of Detail

strategic,tacticaldecisions

tactical, operationaldecisions

Digital Twin

Dynamic Simulation






Dynamic (time)


Static

LocationsFlowsLinear DependenciesContinuousParameter Aggregation Analytical

Methods


The majority ofSC tools

Level of Detail

Digital Twin

Dynamic Simulation



strategic,tacticaldecisions

tactical, operationaldecisions


anyLogistix Simulation Modeling Capabilities

Customers

Factory

Warehouses

Distribution Centers

Suppliers

Factory

AnyLogic


Demo: Inside 4 Walls


anyLogistix Key Differentiators

• Integration of Analytical and Dynamic simulation methods for precise end-to-end supply chain design and analysis Powered by the leading analytical solver (CPLEX) and leading dynamic simulation

engine (AnyLogic)

• Digital twin of your Supply Chain Create a detailed dynamic simulation model, providing a true representation of the

reality for better decision support

• Modeling “Inside the 4 walls” Go deeper in your analysis to improve your Supply Chain efficiency and

effectiveness

• Extensibility Represent true business dynamics in your Supply Chain by extending ALX

functionality

• Measure your Supply Chain With Dynamic Simulation you can measure everything in your model Use standard statistics or create your own

• Visualization Observe how your Supply Chain works, validate the model and make

improvements while verifying assumptions


Master Planning

• Demand is defined separately for each period

• The output of one period is the input for the next

• A site can be closed/opened during a period

• Inventory can be planned for the beginning and end of a period

IQ IIQ IIIQ IVQ


• Problem USAGE BARRIER: How to switch

between different types of scenario?

• Solution: anyLogistix enables you to convert

results or scenarios into other types of scenario

Scenario Conversion

GreenfieldAnalysis

(GFA)

Network Optimization

(NO)

TransportationOptimization

(TO)

DynamicSimulation

(SIM)


Cash to Serve

Paymentterms

AccountreceivablesAccount

payable

Bank

Paymentterms

SC cost

Cash account


• Performance Impact Service level Revenue Total cost Profit Demand fulfillment Lead time

• Time to recover

Risk Analysis

Operational risks(Bullwhip effect)

Disruption risks(Ripple effect)Fr

eque

ncy

Performance Impact

Low

High

Low High


• Goal: To deliver the desired end customer service levels with minimum inventory

investment among the various echelons

• Requirements for true multi-echelon inventory optimization Avoid multiple independent demand forecasts in each echelon Account for all lead times and lead time variations Monitor and manage the bullwhip (variability) effect Enable visibility throughout the supply chain Different service level targets for various DCs/Customers Account for various replenishment strategies

• To satisfy all of the above requirements ALX uses simulation modeling to perform inventory optimization

Multi-echelon Inventory Management

demand

lead time

demand demand

lead time lead time


Safety Stock Estimation

Simulation modeling time

Quantity

“Reorder up to”quantity

Reorder point

Redundant safety stock

Ideal inventorydynamics

Actual inventorybehavior

Safety stock to provide 100% service level


Statistics: Bullwhip effect

• Bullwhip effect shows demand variability amplification during the supply chain simulation: from the point of actual (final) order demand to the point of origin (DC, Factory) BWE > 1 means that outgoing variability prevails over the incoming 0 =< BWE < 1 means that incoming variability prevails over the outgoing BWE = 1 means there is NO bullwhip effect BWE = -1 means that incoming variability is 0

• Bullwhip effect is calculated per product per site

DC

𝜎𝜎𝑖𝑖𝑖𝑖2

𝜇𝜇𝑖𝑖𝑖𝑖𝜎𝜎𝑜𝑜𝑜𝑜𝑜𝑜2

𝜇𝜇𝑜𝑜𝑜𝑜𝑜𝑜 𝐵𝐵𝐵𝐵𝐵𝐵 =�𝜎𝜎𝑜𝑜𝑜𝑜𝑜𝑜2𝜇𝜇𝑜𝑜𝑜𝑜𝑜𝑜�𝜎𝜎𝑖𝑖𝑖𝑖2 𝜇𝜇𝑖𝑖𝑖𝑖

Variance In Variance Out

digital modelling for logistics decision making

Documents