The MAGLEV Project Evaluation Method - Web Publishing · PDF fileThe MAGLEV Project Evaluation Method: ... –MS in Project Management at Northwestern University ... expected value

The MAGLEV Project Evaluation Method: Based on Stochastic Life Cycle Cost Approach

Ph.D. Hyunchul, Choi

July 9, 2013

1

Chair Committee: Lawrence Chiarelli

Fletcher (Bud) Griffis

Andrew Bates

John Falcocchio

Bharat Rao8/24/2016

Researcher’s Background• Academic History

– BS in Mechanical Engineering at Yonsei University (2005)

– MS in Project Management at Northwestern University (2009)

• Publication

– Design of Public Projects: Outsource or In-House? (Journal of

Management in Engineering, ASCE)

– Large Scale Storage of Electrical Energy Using Maglev

(Maglev 2011 International Conference)

– Adaptation of Existing Railroad Trackage for Levitated Maglev

Vehicle Operation (Maglev 2011 International Conference)

– The Maglev America Project: A 28,800 mile national Maglev

network (Maglev 2011 International Conference)

8/24/2016 2

Table of Contents

• The Motivation

• The Problem Statement

• Literature Review

– Magnetic Levitation

– Life Cycle Cost Analysis

– Stochastic Method

• Importance of the Research

8/24/2016 3

Table of Contents

• Research Questions

• Research Methodology

– Markov Chain Model for the Maglev

– Data collection and analysis

a) Maglev initial capital cost

b) Maglev operation and maintenance cost

• Conclusions and Further research

8/24/2016 4

Motivation

• The Maglev System are becoming one of

alternative for the future transportation mode

– Environmental Benefits

– Social Benefits

– Economical Benefits

• The system is shifting from a technical research

stage to a commercial operation stage

– China (Shanghai Transrapid), Korea (UTM-03)

– Japan (Yamanashi, Linimo)

8/24/2016 5

The Problem Statements

• The purpose of this study is to provide the frame

work for predicting the Maglev’s Life Cycle Cost

based on the commercial operation and

maintenance data by using the stochastic

methodology.

8/24/2016 6

Literature Review

• The Fundament in the Magnetic Levitation

– The mechanism illustration

8/24/2016 7

• Propulsion

• Levitation

• Stabilization

• Quadrupole

Magnet

Magnetic Levitation

• State of the art in Magnetic Levitation

– There are two basic systems

• EMS, Electromagnetic attractive force system

• EDS, Superconducting repulsive force system

– 1st Gen. Dipole magnet

» Low strength, High magnetic field in the vehicle

– 2nd Gen. Quadrupole magnet

» High strength, Low magnetic field in the vehicle

» No prototype as yet developed

8/24/2016 8

Magnetic Levitation

• Electromagnetic Suspension (EMS)

– The technical characteristics

8/24/2016 9

Characteristics

Advantages• Commercial Operation History

• The lower magnetic field strength

• No wheel assistant system for the levitation

Disadvantages• Complex servomechanism to control a 1 cm gap

• Extreme structure tolerance and extreme high

cost in guideway construction

Magnetic Levitation

• Electromagnetic Suspension (EMS)

– The current operation model

8/24/2016 10

Linimo in JapanShanghai

TranspidUTM-02 in Korea

Magnetic Levitation

• Electrodynamic Suspension (EDS)

– The technical characteristics

8/24/2016 11

Characteristics

Advantages

• Above 10cm gap is allowed

• This system has the highest speed record

• Liquid nitrogen level superconducting magnet

has been introduced

• Quadrupole system operates on existing RR

Disadvantages

• Strong magnet field (Dipole Magnet)

• Wheel Assistant System for levitation

(Dipole Magnet)

• Quadrupole system has not been fully tested

Magnetic Levitation

• Electrodynamic Suspension (EDS)

– The current developing model

8/24/2016 12

Yamanashi line in Japan Maglev 2000 in U.S

Magnetic Levitation

• Comparison of guideway designs (EMS & EDS)

8/24/2016 13

• German Maglev: Uses Attractive Force. High Energy Consuming

Conventional Electromagnets. Ultra-Engineering Tolerances and

Complex Control Required to Maintain Narrow Air Gap of 3/8th

inch. Difficult Weather and Seismic Issues.

• Japanese Maglev: Uses Repulsive Force. Low Energy Consuming

Superconducting Di-Pole Magnets. Large Air Gap (4-6 inches) U-

Shape Creates Weather Issues. High magnetic field

• U.S. Maglev: Uses Repulsive Force. Low Energy Consuming

Superconducting Quadrupole Magnets. Fringe field approximate

the ambient Earth field. Large Air Gap (4-6 inches.) Inherently

strongly stable. No weather or seismic issues. Operation on the

existing railroad

LCC Analysis

• Life Cycle Cost (LCC) Analysis

– LCC Analysis is defined as “an economic evaluation

technique that determines the total cost of owning and

operating a facility over period of time”.

– The effectiveness of using LCC Analysis is listed as

followed:

a. Evaluating alternatives in the economic feasibility

b. Recognizing of the true total cost

c. Supporting for preparing the future cost plan

d. Providing the resources for the decision-making

8/24/2016 14

LCC Analysis

• Life Cycle Cost (LCC) Analysis

– Fundamental cost elements in LCC:

𝐋𝐂𝐂 = 𝑪𝒄𝒂𝒑𝒊𝒕𝒂𝒍 + 𝑪𝒐𝒘𝒏𝒆𝒓𝒔𝒉𝒊𝒑 + 𝑪𝒅𝒊𝒔𝒑𝒐𝒔𝒂𝒍

– In the LCC, the relationship between the value and time

𝑷𝑽 =𝑭𝑽

𝟏 + 𝒊 𝒏, (𝒘𝒉𝒆𝒓𝒆 𝒊 = 𝒅𝒊𝒔𝒄𝒐𝒖𝒏𝒕 𝒓𝒂𝒕𝒆 𝒂𝒏𝒅 𝒏 = 𝒏𝒖𝒎𝒃𝒆𝒓 𝒐𝒇 𝒑𝒆𝒓𝒊𝒐𝒅)

– The Equivalent Annual Cost (EAC)

𝑬𝑨𝑪 =𝑵𝑷𝑽

𝑨𝒏,𝒊, 𝑨𝒏,𝒊 =

𝟏 − (𝟏 + 𝒊)−𝒏

𝒊

8/24/2016 15

Stochastic Methods

• Monte Carlo Method

– The Monte Carlo Methods provide approximate solution

to a variety of mathematical problems by performing

statistical sampling experiments.

• Markov Chain Analysis

– The Markov Chain analysis is a stochastic process

characterized as memoryless which is a markov property.

– It has the conditional probability distribution of a future

state given by the present states function alone

8/24/2016 16

The Importance of this Research

• It provides the frame work for predicting

Maglev’s Life Cycle Capital, Operation, and

Maintenance Cost considering uncertainty.

• It provides the method for calculating the

commercial viability of Maglev system such as

revenue (ticket price) vs. all cost.

• Finally, it also provides the basis for comparison

analysis among the different Maglev technologies.

8/24/2016 17

Research Questions

• The Maglev systems’ initial capital cost and risk

associated with that cost

• The Maglev systems’ operation and maintenance

behavior, cost and risk associated with that cost

8/24/2016 18

Research Questions

– The Maglev systems’(EMS and EDS) future initial

capital cost behaviors

a. The initial capital cost has a large deviation due to

the uncertainty in the cost elements

b. The EMS Maglev construction cost will be higher

than EDS Maglev due to EMS’s technical

characteristics

8/24/2016 19

Research Questions

– The Maglev systems’ operation and maintenance cost

behaviors

a. The maglev operation condition deterioration

during the life period is a limited due to its

technical characteristics.

b. The Maglev’s operation and maintenance behavior

will have the Markov property characteristics.

c. Due to the EMS Maglev’s 1 cm gap requirement, its

O&M cost will higher than other Maglev system.

8/24/2016 20

Research Methodology

8/24/2016 21

• Markov Chain Model for the Maglev

– The “Markov Chain Model” is name after Andrey Markov

– It also is characterized as memoryless property or Markov

property: the next system state depends only on the current

system state and not on the sequence of state transition that

preceded it.

– This can be expressed as:

𝑷𝒓 𝑿𝒕+𝟏 = 𝒋 𝑿𝟎 = 𝒌𝟎,𝑿𝟏 = 𝒌𝟏,⋯ ,𝑿𝒕 = 𝒊 = 𝑷𝒓{𝑿𝒕+𝟏= 𝒋 𝑿𝒕 = 𝒊}

– The daily operation & maintenance cost is the function of

the system state


• Markov Chain Model for the Maglev

– Terms in the Markov Chain Model

a. Time: in a Markov chain, the system is observed at

discrete points in time.

b. State: The state describes the situation at a point.

c. Transition probability: The set of possible events

and their probabilities depend on the state.

d. Transition Matrix: It is the matrix which is

consisted with the set of transition probability in the

system

8/24/2016 22


• Data Collection and Analysis (Capital Cost)

– Maglev System initial capital cost characteristics from

the Maglev cost estimation data from Incheon Airport

project.

a. Construction cost estimation and Probability data

b. Each cost items is stochastically independent

c. The Maglev initial construction cost has been

simulated using the Monte Carlo techniques.

– The probability data are obtained from the Canadian

railway cost data.

8/24/2016 23



8/24/2016 24

Observed Maglev Cost Estimates

Observed Maglev Cost Estimates

Canadian Risk Information in the

Railway

Canadian Risk Information in the

Railway

PDFPDF

j = nj = n

j = j + 1j = j + 1

NONO

YESYES

ResultResult



8/24/2016 25

Earth Work 3,846,000

Structure Work 62,761,000

Water Management Facility & Equip. 1,033,000

Temporary Facility 5,467,000

Relocation Construction 1,557,000

Main Material Cost for Construction 15,781,000

Guideway 68,733,000

Architecture 20,910,000

Mechanical Equip Purchasing 27,194,000

Overhead 23,643,000

Other Cost 15,509,000

Construction Cost Sub Total 246,434,000

Design Cost 21,576,000

Total Cost 268,010,000

MAGLEV project construction cost estimate (USD)



8/24/2016 26

Subsystem Cost Range in the Railway Construction from

Ministry of transportation in Ontario Canada



– The EMS Maglev initial construction cost

8/24/2016 27

• Mean:

$ 251 million

• Standard Deviation:

$ 4.7 million

• 80% Chance Value:

$ 255 million


• Data Collection and Analysis (O&M Cost)

– Maglev Operation & Maintenance (O&M) Cost

a. The Markov Chain Model provides the tool for

predicting the daily Maglev operation behavior.

b. The Maglev O&M costs are estimated by the

expected value of the operation behavior

probability and the observation of Maglev O&M

cost data.

c. The discount rate also is taken from the Korean

commerce department statistics.

8/24/2016 28




a. The Markov Chain probability matrix and O&M

cost data are obtained from 2 years operation and

maintenance data of the UTM-02 Maglev system

b. This Maglev system is divided into 6 sub-systems:

Vehicle, Guideway, Power supply, Power control,

Signal and Communication system

8/24/2016 29



– The case for evaluating the proposed model

a. Feb. 2006 ~ Aug. 2012

b. App. $ 371 million

c. Evaluation of Maglev

Commercial Operation

d. Incheon International

Airport Complex

e. App. 6.1 km distance

8/24/2016 30



8/24/2016 31

State 1Clean

State 2 Normal

State 3Failure

Observed Daily Cost-Condition DataObserved Daily Cost-Condition Data

Observed Daily Operation Report

Observed Daily Operation Report

Develop the transition

probabilities from RFH

Develop the transition

probabilities from RFH

ResultResult

𝑷𝒏, 𝒏 = 𝟏,⋯𝟗𝟎MCM

𝑷𝒏, 𝒏 = 𝟏,⋯𝟗𝟎MCM

𝑷𝒎𝒏

𝒏 = 𝟏,⋯𝟗𝟎𝒎 = 𝟏,⋯𝟓𝟎

MCS

𝑷𝒎𝒏

𝒏 = 𝟏,⋯𝟗𝟎𝒎 = 𝟏,⋯𝟓𝟎

MCS




8/24/2016 32




a. Developing the transition probability matrix

8/24/2016 33

State 1Clean

State 2 Normal

State 3Failure

• 3 states

• 9 path between the states

• Frequency Histogram by

counting the states

transition activities in the

daily operation report




a. Developing the transition probability matrix

8/24/2016 34

• 9 probabilities in the

transition matrix

• Each probabilities stands

for transition activities

between the states

• 7 transition probability

matrices




b. Developing the frequency histograms

8/24/2016 35

366

12 8 7

102

8 12 4 00

20406080

100120140160180200220240260280300320340360380400420440460480500520

11 12 13 21 22 23 31 32 33

Number of transition in the vehicle condition





8/24/2016 36

312

13 0 9

180

3 2 0 00

20406080

100120140160180200220240260280300320340360380400420440460480500520

11 12 13 21 22 23 31 32 33

Number of transition in the guideway condition





8/24/2016 37

485

8 2 7 15 0 2 0 00

20406080

100120140160180200220240260280300320340360380400420440460480500520

11 12 13 21 22 23 31 32 33

Number of transition in the power supply system condition





8/24/2016 38

505

5 2 5 0 0 2 0 00

20406080

100120140160180200220240260280300320340360380400420440460480500520

11 12 13 21 22 23 31 32 33

Number of transition in the power control system condition





8/24/2016 39

297

2 3 2 0 0 4 0

211

020406080

100120140160180200220240260280300320340360380400420440460480500520

11 12 13 21 22 23 31 32 33

Number of transition in the signal system condition





8/24/2016 40

495

1 8 1 3 0 8 0 30

20406080

100120140160180200220240260280300320340360380400420440460480500520

11 12 13 21 22 23 31 32 33

Number of transition in the communication system condition





8/24/2016 41

232

234

21

223

6 5 5 00

20406080

100120140160180200220240260280300320340360380400420440460480500520

11 12 13 21 22 23 31 32 33

Number of transition in the operation condition




c. Developing the transit probability matrix

8/24/2016 42

Vehicle

P 1 2 3 Total

1 0.948 0.031 0.021 1.000

2 0.060 0.872 0.068 1.000

3 0.749 0.250 0.001 1.000

Guideway

P 1 2 3 Total

1 0.959 0.040 0.001 1.000

2 0.047 0.938 0.015 1.000

3 0.998 0.001 0.001 1.000

Power Supply

P 1 2 3 Total

1 0.980 0.016 0.004 1.000

2 0.318 0.681 0.001 1.000

3 0.998 0.001 0.001 1.000

Power Control

P 1 2 3 Total

1 0.986 0.010 0.004 1.000

2 0.998 0.001 0.001 1.000

3 0.998 0.001 0.001 1.000




c. Developing the transit probability matrix

8/24/2016 43

Signal

P 1 2 3 Total

1 0.983 0.007 0.010 1.000

2 0.998 0.001 0.001 1.000

3 0.019 0.001 0.980 1.000

Communication

P 1 2 3 Total

1 0.982 0.002 0.016 1.000

2 0.250 0.749 0.001 1.000

3 0.726 0.001 0.273 1.000

Operation

P 1 2 3 Total

1 0.896 0.089 0.015 1.000

2 0.084 0.892 0.024 1.000

3 0.500 0.499 0.001 1.000


8/24/2016 44




8/24/2016 45




8/24/2016 46




8/24/2016 47




8/24/2016 48




8/24/2016 49




8/24/2016 50




8/24/2016 51




8/24/2016 52

State 1 daily cost State 2 daily cost State 3 daily cost

Operation 940,175 (Korean Won) 940,175 (Korean Won) 0 (Korean Won)

Vehicle 195,695 (Korean Won) 195,695 (Korean Won) 813,258 (Korean Won)

Guideway 195,695 (Korean Won) 457,871 (Korean Won) 24,013,028 (Korean Won)

Power supply 195,695 (Korean Won) 195,695 (Korean Won) 645,695 (Korean won)

Power control 195,695 (Korean Won) 195,695 (Korean Won) 195,695 (Korean Won)

Signal 195,695 (Korean Won) 195,695 (Korean Won) 195,695 (Korean Won)

Communication 195,695 (Korean Won) 195,695 (Korean Won) 195,695 (Korean Won)




e. Monte Carlo Simulation (MCS) with Markov Chain

Probability information (50 times trial)

f. The expected Maglev’s O&M cost is estimated

following the function:

𝑬[𝒇(𝑿𝒏)] = 𝑷𝒏 ∙ 𝑪𝒔

• 𝑓 𝑋𝑛 = the system cost at the state 𝑋𝑛

• 𝑃𝑛 = the transition matrix

• 𝐶𝑠 = the state cost

8/24/2016 53



– The annual EMS Maglev operation cost

8/24/2016 54

• Mean:

$ 300,500


$ 605.85


$ 300,900



– The annual EMS Maglev maintenance cost

8/24/2016 55

• Mean:

$ 527,120


$ 823.72


$ 527,966



– Comparison of observation data and prediction data

– The difference between the observation data and model

prediction in the maintenance cost is 4.7%

8/24/2016 56

Annual MAGLEV operation and maintenance data from the model USD 827,620

Annual MAGLEV operation data from the model USD 300,500

Annual MAGLEV maintenance data from the model USD 527,120

Observed annual MAGLEV maintenance data USD 503,283


• Data Collection and Analysis (LCC Analysis)

– The Maglev EAC of the life cycle 20 years

8/24/2016 57

• Mean:

$ 24.5 million


$ 0.42 million


$ 24.9 million


• Data Collection and Analysis (Comparison

Analysis)

– The cost comparison between EMS and EDS Maglev

systems. The EDS system data was obtained from the

recent NYSERDA study.

8/24/2016 58

EMS EDS

Guideway Construction Cost per mile $35.06 mil. $32.14 mil.

O&M Cost per passenger mile (USD) $0.73 $0.11

Conclusions and Further research

8/24/2016 59

• Findings

– This research presents that the expected construction cost

of EMS MAGLEV $251 million and the expected annual

operation and maintenance cost of EMS MAGLEV for

$0.8 million.

– The expected values of EAC in EMS Maglev is $24.5

million.

– In the cost comparison analysis between EMS and EDS,

the EDS system is 9% cheaper in the initial cost. In the

O&M cost , the EMS system is 6 times as expensive.


8/24/2016 60

• Conclusion

– The combination of Markov Chain model and Monte

Carlo method provides basis for predicting the MAGLEV

operation and maintenance behavior over the certain

period.

– This combination also provides a basis for LCC analysis

for systems with incomplete or uncertain data.

– In an addition, this combination can be used in the

comparing the various MAGLEV technologies.


8/24/2016 61

• Further research

– Comparison Analysis with other transportation modes

(High Speed Railway)

– Maglev Subway system LCC analysis with the proposed

model

– EDS Maglev system LCC analysis with the proposed

model

Question and Answer

8/24/2016 62

Documents

The MAGLEV Project Evaluation Method - Web Publishing · PDF fileThe MAGLEV Project Evaluation Method: ... –MS in Project Management at Northwestern University ... expected value