Page 1 Tanvir Madan M.S. Technology and Policy Program, December 9 th, 2011. ESD. 71 Application Portfolio Evaluating flexibility – Grid Scale Solar PV

Page 1

Tanvir Madan

M.S. Technology and Policy Program, December 9th, 2011.

ESD. 71 Application PortfolioEvaluating flexibility – Grid Scale Solar PV Plants

Overview > Description> Design> Implementation> Results> Lessons

Page 2

Application Portfolio - Overview

• System Description

• System Design

• Method of implementation• Uncertainty• Flexibility• Simulation

• Results• Fixed Design• Flexible Design• Target curves and other criteria

• Lessons Learned


Abstract:

This project aims to evaluate the introduction of flexibility into designing grid-scale solar power plants. There exist certain forms of tracking PV panels that can be manually tracked/ moved in order to increase the output of the plant. At the same time though, certain costs are incurred to introduce flexibility by employing manual labor and hence through this project, taking into account certain core uncertainties in this system, we hope to address whether employing manual labor will help the overall financials of the PV project.

Page 3

Application Portfolio - Description


• Grid connected 5MW PV project, operational for 25 years

• Based on financial model built while interning this summer at an Independent Power Producer

• Technical and Financial inputs; Financial outputs

• Economic Evaluation metric of choice – Equity Internal Rate of Return (IRR); ease of comparison to large cost of capital (interest rate)

Page 4

Application Portfolio – System Design


-10.

00-8

.00

-6.0

0-4

.00

-2.0

00.

002.

004.

006.

008.

0010

.00

0.00%

2.00%

4.00%

6.00%

8.00%

10.00%

12.00%

14.00%

16.00%

18.00%

Senstitivity Analysis

PLF Sensitivity

EPC Sensitivity

Interest Rate Sensitivity

% change in underlying paramter

Equity IRR (IRR)

• Three underlying variables evaluated• Plant load factor

(PLF)• Engineering,

Procurement and Construction costs (EPC)

• Interest rates

• PLF evaluated to be most sensitive – hence PLF to be modeled as an uncertain distribution

Page 5

Application Portfolio - Implementation


Pre-existing Model •Model for a 5MW solar project with certain fixed assumptions as inputs and an IRR that meets the developers threshold

Base model •Solar project evaluation model (pre-existing model) with uncertainty introduced for the PLF

Flexible Model •Solar project evaluation model with uncertainties (Base model) and flexibility for a particular design choice

• Step 1: Adding uncertainty (in PLF) to pre-existing model

• Step 2: Adding flexible design choice to model

• Step 3: Evaluating flexible model using economic evolution metrics

Page 6

Application Portfolio – Implementation Step 1


• Using existing data to model distribution of uncertainty

• Modeled as normal distribution with mean as predicted and variance of 1%

• Utilizing other existing input and output parameters from a benchmarked case

Variable Type Number

EPC Cost Input Rs. 103 Million/MW

PLF Input 19.6%

Interest rate Input 10% 1 yr after Commission

Equity IRR Output 13%

Time Period Base PLF without Uncertainty

Base with PLF Uncertainty

Yrs 1-5 19.6% S.D. with mean = 19.6%, Variance = 1%





Page 7



There exist certain PV panels that are manually tracked i.e. they can be moved by manual labor in order to increase the output they can capture. Hence, project developers have the ability to make a decision on whether they will employ manual labor with an expectation of extracting a greater amount of energy from such panels. It is not atypical for panels to be manually moved as seasons change in order to allow the panels to face as large as possible amount of incident rays from the sun.

• Flexibility in question: employing labor to manually track panels and increase output

• Divided into 5 steps of 5 years each

• Cost of flexibility: 3.5% of total project cost

• Gain from flexibility: additional output i.e. potentially increased PLF

Page 8



• Implemented choices• Decision tree (discrete

distributions)• Lattice Model

(stationary uncertainty) • Simulation

• Decision – if the expected value of using flexibility > expected vale of base case, then exercise flexible choice

(a) Total Size of plant: 5MW

(b) Size of each panel: 200W

(c) No. of panels: ~25000 (a)/(b)

(d) Total hrs/shift: ~6250 hrs (assuming 15 mins per panel shift) (c)/6

(e) No. of days/shift: 10 days (assume each tracking operation lasts for two weeks)

(f) No. of hrs/day ~625 (d)/(e)

(g) No. of labor ~78 (assuming 8 hr working days) (f)/8

(h) No. of shifts a yr: 4 (once a quarter)

(i) Cost of Labor/Hr: Rs 30/hr

(j) Total Cost: ~Rs 750,000 (e)*(f)*(h)*(i)

(k) Cost/MW: ~Rs 150,000 (j)/(a)

Page 9

Application Portfolio – Results Fixed Design


• Fixed Design• IRR of 13%

• Flexible Design• Expected value always

less than 13%!• Hand picked cases to

evaluate target curves and other criteria (four cases)

5.0% 90.0% 5.0%

11.772% 14.381%

10

.5%1

1.0

%11

.5%1

2.0

%12

.5%1

3.0

%13

.5%1

4.0

%14

.5%1

5.0

%15

.5%1

6.0

%

0

10

20

30

40

50

60

Equity IRR(%)

Equity IRR(%)

Minimum 10.84%Maximum15.63%Mean 13.02%Std Dev 0.803%Values 500

@RISK Student VersionFor Academic Use Only

Page 10

Application Portfolio - Results Flexible Design


5.0% 90.0% 5.0%

9.98% 16.32%

8%10

%12

%14

%16

%18

%20

%22

%24

%26

%28

%

0

5

10

15

20

25

Equity IRR(%)

Equity IRR(%)

Minimum 8.45%Maximum 27.3%Mean 12.5%Std Dev 2.16%Values 500


Flexibility exercised in all periods

5.0% 90.0% 5.0%

9.82% 16.43%

8%

10

%1

2%

14

%1

6%

18

%2

0%

22

%2

4%

26

%

0

5

10

15

20

25

Equity IRR(%)

Equity IRR(%)



Flexibility exercised in years 1-5 (time block 1 only)

Page 11

Application Portfolio – Results Target Curve


5.0% 90.0% 5.0%41.6% 42.0% 16.4%

11.75% 14.45%

8%

10

%1

2%

14

%1

6%

18

%2

0%

22

%2

4%

26

%2

8%

0.0

0.2

0.4

0.6

0.8

1.0

Case A Equity IRR(%)

Equity IRR Case A(%)


Equity IRR Case D(%)


Equity IRR Case B(%)


Equity IRR Case C(%)



• Fixed design – highest expected IRR

• Flexible design – better about 25% of the time• Risk nature of

developer• Ability to extract and

utilize information

• Cases – all flexible cases similar – low sensitivity to timing of flexibility

Page 12

Application Portfolio – Results Other Evaluation Criteria


Mean Min. Max. Std. Dev. P5 P95Base Case (Case A) 13.02% 10.90% 15.42% 0.825% 11.05% 14.45%Flex in years 1-5 (Case B) 12.5% 8.45% 23.1% 2.16% 10.03% 16.15%Flex in years 1-10 (Case C) 12.5% 9.05% 25.4% 2.04% 9.89% 16.07%Flex in years 1-25 (Case D) 12.5% 8.70% 27.5% 2.12% 9.89% 16.34%

• The fixed design always provides a greater expected value.

• Based on the target curves, there is a 60% probability for the flexible cases to return an IRR less than the mean of the base case (13%)

• As expected, the standard deviation is higher for the flexible projects.

• The P5 and P95 numbers demonstrate the variation b/w the fixed and flexible designs. A risk taking developer looking to capture some of the low probability upside would find this information exciting!

Page 13

Application Portfolio – Main Lessons


Recognize •Appreciating uncertainty in a system•Analyzing the sensitivity of various underlying input parameters

Design •Modeling and creating distributions to classify uncertainty•Analyzing and utilizing methods to classify and incorporate uncertainty

Evaluate •Understanding the value of flexibility in design and incorporating flexibility into existing systemic models•Evaluating designs based on economic and mathematical indicators (NPV, IRR, P5, P95, etc.)

Page 14

Application Portfolio – Questions?


Source: http://37signals.com/Source: http://www.coachwithjeremy.com/

Documents

Page 1 Tanvir Madan M.S. Technology and Policy Program, December 9 th, 2011. ESD. 71 Application Portfolio Evaluating flexibility – Grid Scale Solar PV