A comparison of economic evaluation models as applied to geothermal energy technology

Eneqy Vol. 8. 4’0. IO. pp. :‘%Sll. Pnnted I” Great Bnlain.

033SW83 Pergmon

1983

sm3- .w Press Lib

ACOMPARISONOFECONOMICEVALUATIONMODELS ASAPPLIEDTOGEOTHERMALENERGYTECHNOLOGY? G. MICHAELZIMANand LEIGH S. ROSESBERG Jet Propulsion Laboratory. California Institute of Technology, 4800 Oak Grove Drive, Pasadena. C.4 91103, U.S.A. (Received 4 October 1982) Abstract-Several cost estimation and financial cash flow models have been applied to a series of geothermal case studies. In order to draw conclusions about relative performance and applicability of these models to geothermal projects, the consistency of results was assessed. The model outputs of principal interest in this study were net present value, internal rate of return. or levelized breakeven price. The models used were VENVAL, a DuPont, Inc. venture analysis model; the Geothermal Probabilistic Cost Model (GPC Model) and the Alternative Power Systems Economic Analysis Model (APSE.GI), which were developed at the Jet Propulsion Laboratory (JPL); the MITRE Corporation’s Geothermal Loan Guarantee Cash FlowModel(GCFM); andtheGEOCOST andGEOCITY geothermal models developed by Battelle Pacific Northwest Laboratories. The case studies to which the models were applied include a geothermal reservoir at Heber.CA; a geothermal electricpower plant to be located at the Heber site; an alcohol fuels production facility to be built at Raft River, ID; and a direct-use, district heating system in Susanville. CA.

I. INTRODUCTION

Relative merits of different cost estimation computer models are often unknown because sufficient comparisons between the models have not been made. When previously uncompared models are run individually, it is difficult to assess the validity of the results. Reliance on results from a particular model used in isolation may lead to conclusions that deviate from those predicted by other models. In order to help alleviate this problem for geothermal energy technology, this study was initiated by the Division of Geothermal Energy (DGE) of the U.S. Department of Energy (DOE). Several financial cash flow and cost-estimation models were applied to case studies representing four different geothermal end-uses; judgments about their relative performance and applicability to different types of geothermal projects were made. The models used in this study range from general financial cash flow models which can be applied to any type of energy project, to models developed for assessing only one type of geothermal end-use. A secondary result of the project was an analysis of the economic viability of the case studies evaluated. In the following sections we give the study methodology, a description of the models evaluated, a description of the case studies, the results of the analysis, and overall conclusions. 2. STUDY METHODOLOGY

The models used in this study were applied to a set of case studies which reflected a wide range of geothermal end-use applications. By applying several models to a particular case study, it was possible to draw conclusions about their relative strengths and weaknesses. This was done primarily by observing whether each model would yield the same aggregate figures of merit. By focusing upon aggregate measures rather than a line-by-line accounting analysis, we were able to expand considerably the number of models and case studies evaluated. Likewise, space limitations have prevented us from presenting a detailed discussion of the more theoretical aspects of each model’s economic approach. If the reader desires additional information, he can consult the references or contact the authors. The aggregate figures of merit that were used were net present value, internal rate of return, or levelized breakeven price. The net present value (NPV) is defined as +The research described in this paper was carried out at the Jet Propulsion Laboratory, California Institute of Technology, and was sponsored by the U.S. Department of Energy through an agreement with the National Aeronautics and Space Administration. EGY Vol. 8. No. I&D*

79-l

798

G. !A ZIMXS and L. S. RGEXBERG

NPV = 5 (R, - C,)/( I + d)‘. c=o

(1)

where R, = revenue (price x quantity) obtained from the project in year t, C, = cost of the project in year t, IV = lifetime of the project (construction plus operation), and d = discount rate. When the NPV is set equal to zero and Eq. (1) is solved for d, the solution is defined as the internal rate of return. Finally. the per unit price of output that exactly pays back all project expenses, including a return to capital over the lifetime of the project, is the levelized breakeven price.: The models used in this study were DuPont’s VENVAL;’ the Geothermal Probabilistic Cost Model (GPC Model)’ and Alternative Power Systems Economic Analysis Model (APSEAM) developed at JPL; the MITRE Corporation’s Geothermal Loan Guarantee Cash Flow Model (GCFM),4 which is currently in use at the DOE San Francisco Operations office (DOE/SAN); and the GEOCOST’ and GEOCITY6 models developed at Battelle Pacific Northwest Laboratories (PNL). The case studies are based on the Heber Reservoir in the Imperial Valley of California, the Heber Electric Plant to be built at the site of the Heber Reservoir, the proposed Raft River Ethanol Facility in Idaho, and the Susanville, CA District Heating System. For each end-use application (reservoir, electric plant, alcohol fuels facility. and district heating system), a master data set (consisting of cost, financial, physical, and engineering data) was compiled. From this master set, an individualized data subset was formulated for each model. These data subsets varied considerably because the purposes and capabilities of each model were different, which resulted in differing input requirements. Furthermore, because the models were designed for varying purposes, the application of some of them to ail the cases was precluded. The application of the models to each case is discussed in Section 5. 3. DESCRIPTION

OF THE

MODELS

(a) VENVAL VENVAL is a financial cash flow tool developed at Du Pont, Inc. to aid in the evaluation of proposed business investments. The model is quite flexible and can be applied to a diverse range of ventures provided that a complete set of cost and financial data is available. VENVAL can consider a large number of cost accounts in various accounting categories. Many depreciation schemes and borrowing arrangements can be implemented, and several economic figures of merit can be generated. At the user’s discretion, specific relationships and options can be employed through the use of input coefficients. The program can project interrelated values of installed capacity, production, sales revenue, operating costs, depreciation, investment, debt, earnings, taxes, depletion, return on investment, and other cash flow measures.’ For a given price, VENVAL gives the net present value and can solve for the internal rate of return for the project, or it can be used to solve for the breakeven price required to yield a specified rate of return. It can also compute related public sector and other external costs and revenues. Each case computed may have up to 50 years of venture life. The user has a choice of several hundred relationships (equations) between the different items for the computations. Each case may use these equations any number of times, and values can be automatically escalated from year to year. VENVAL can also perform risk analyses. Cases may be run using Monte Carlo techniques to produce statistics on selected economic parameters if the appropriate inputs have been entered in the form of probability distributions. A flow chart for VENVAL is shown in Fig. 1. (b) Geothermal Probabilistic Cost Model (GPC Model) The GPC model was developed at JPL to analyze projects whose costs were affected by uncertainties in project duration as well as physical parameters. It can be applied to any

+Breakeven prices can be levelized in real or nominal terms. Real-levelized prices grow in nominal value at the rate of general price escalation and thus stay constant in real value, while nominal levelized prices stay constant in nominal terms and thus decrease in real value as price escalation occurs. In this study, when breakeven prices are real-levelized, they will be identified as such and the associated escalation rate will be given.

A comparison of economic evaluation models

799

Cost Accounts INPUTS Schedule Expenditure

FinancialPamneten

Probabilistic Assumptions of UncertainCost Variables Fig. I. Flowchart for VENVAL.

venture, energy-related or otherwise, for which a set of input cost accounts and probability density functions are available.* Most typical projects can be divided into stages. Often, many of the project cost accounts are dependent upon the length of the stage in which they occur: because the greatest uncertainty for the project is often related to the duration of the stages, these costs will also be highly uncertain. The GPC model handles this uncertainty by individually evaluating each possible permutation of stage length and aggregating the economic results of each permutation. or “scenario”, into probability distributions. Along with the stage lengths, permutations of physical parameters are also considered in the scenarios. Thus, because one does not merely give a range of costs, but rather considers a range of durations for the stages (and a range of values for the uncertain physical parameters), it is possible to specify the factors to which any uncertainty can be attributed. The model then calculates the impact of the uncertain variables. through previously chosen scaling relationships, on the individual cost accounts. To avoid the lengthy and difficult task of acquiring the cost accounts for each scenario, the model makes use of a Reference (or baseline) Scenario. Cost accounts are input for only this Reference Scenario. For all other scenarios, only the stage times, physical parameter values, and associated probabilities are input: their cost accounts are derived within the program by modifying the appropriate Reference Scenario cost accounts for any differences in stage lengths or in physical parameters. For each scenario, a required-revenue financial subroutine is employed to derive the project cost and profit which are then aggregated on a net present value basis.’ (Required revenue is that revenue which must be obtained from the system over the lifetime of the project so that all costs, including specified rates of return, can be paid.) For a given price, the GPC model gives net present value and can be used to solve for the internal rate of return for the project. The model can also solve for the breakeven price required to yield a specified rate of return. Once the conditional probabilities for the stage durations and the physical variables are entered in the model, the probability of a given scenario occurring can be calculated. The costs and profits for each scenario along with their probabilities are then aggregated into overall probability distributions for all scenarios within a single run. A flow chart for the GPC model in shown in Fig. 2. The most distinguishing feature of the model is that it allows the outcome of one variable to be dependent upon the outcomes of the other variables. Conditional probability distributions can thus be used. For example, the probability distribution for the length of a development stage may be dependent upon the lengths of the stages that precede it or upon the depths of the wells that have to be drilled, none of which may be known at the beginning of the project. In this way, any statistical correlation between characteristics can be considered explicitly, and a joint probability distribution that has all existing dependency relationships factored into it can be constructed. (c) Alternative Power Systems Economic Analysis Model (APSEAM) APSEAM is a interactive financial cost flow model developed at JPL. It is particularly adaptable to various energy technologies and its financial routines can account for various tax incentives and tax treatments typically available to advanced alternative energy technologies. It can be applied to any investment project for which there is a complete set of costs, a schedule of expenditures, and a set of financial parameters applicable to the investing parties.

INPUTS Coat Accounts for Reference Scenar lo Expenditure Schedule Financial Parameters Probability Distributions for Stage Lengths and Uncertain Variables

Creates Alternative Scenarios

INPUTS CosZiGunt 3 Expenditure Schedule Financial Assumptions

-

SCENARIO ALGORITHM

. -

+ Rate-of-Return

CASH FLOW MODEL Discounting

COST ADJUSTMENT SUBHOUTZNE Costs All Scenarios

__

.

-

1

OUTPUTS Probability Distribution for Net Present Value Profit of the Project

OUTPUTS Annual Cdl FlOWS Figures of Economic Merit

CASH FLOW MODEL Required Revenue Net Present Value

1

I

.A ampariwn

of tc‘onomic

evaluation

model>

801

Detailed cash flow information is projected for each investment alternative for each year in the investment timeframe.: These data are then aggregated to produce various measures of the life-cycle cost of each investment alternative. APSEAM can solve for net present value, internal rate of return. and breakeven price. The model can be used to quantify the impact on annual cash flows of variations in technology cost. general economic conditions. investorspecific financial conditions. the method of financing, resource characteristics, technology performance over time. supply and demand matching, incremental plant start-up, and component replacement scenarios. A flowchart for APSEAM is shown in Fig. 3. (d) Geothrrmnl Loan Guarantrr Cash Now Modrl (GCFM) GCFM a-as developed at the MITRE Corporation for use by DOE in evaluating applications for loans in the Federal Geothermal Loan Guarantee (FGLG) Program. FGLG requirements stipulate that qualifying ventures demonstrate that they can meet several criteria beyond project profitability (e.g.. ability to meet minimum equity fractions in each year of the project and generate a sufficient annual cash flow for the service of debt) in order to be eligible for government loans. The model has the capabilities to check that these requirements will be met. GCFM can be applied to geothermal reservoirs, geothermal electric generation plants, or the combination of the tu’o. It is divided into five submodels. (A flow chart is shown in Fig. 4.) The Cycle Design Routine assimilates resource and plant data such as temperature, salinity. humidity. and the type of plant being developed, and then designs and determines the costs of a plant. If the cost accounts for the plant are already known, the Cycle Design Routine can be superseded by the Power Plant Module. This module combines the actual plant cost data with a schedule for construction and a set of financial parameters, and then calculates a detailed cash flow for the plant construction and operation phases. The Field Design Module, in a similar fashion. takes field development cost data, together with a schedule for development and the necessary financial information. and calculates a detailed cash flow for the field throughout its entire life. The Financial Modules are used in conjunction with the Field and Power Plant Modules when life cycle costs are calculated. Financial parameters such as the cost of borrowed money. required return on equity, escalation rates, tax rates, !ax credits, and depletion allowances are combined vvith the captial expenditures and operating costs to calculate the project cash flows. When a price for the fluid produced by the field or for the electricity generated by the plant is provided. the Field and Plant IModules calculate the rate of return of the venture. Alternatively. the levelized breakeven price of the fluid supplied from the field (or of the electricity produced by the plant) necessary to meet any specified revenue requirements can be calculated. (e)

GEOCOST

GEOCOST is a geothermal electric plant simulation model developed at. PNL. From a limited set of physical, resource. climatic, and reservoir data. the model generates a plant design, a plant construction and operation scenario, and the associated annual cash flows. It also calculates a levelized charge for the electricity generated. The model is composed of two parts: a submodel that simulates the costs associated with the exploration, development, and operation of a geothermal reservoir: and a submodel that simulates the costs associated with the design. construction, and operation of the power plant.’ A flow chart for GEOCOST is shown in Fig. 5. The field submodel. having establsished the total fluid flow that supplies the specified power production requirements of the plant, calculates all costs associated with the development of the reservoir and yields the resultant cash flow. This includes capital costs and operating expenses from the beginning of exploration through the operating life of the power plant. Based on the reservoir cash flow and quantity of energy supplied, the unit cost of energy that equates the present worth of the revenues with the present worth of the expenses for the reservoir is calculated. In the power plant submodel, the unit cost of energy from the reservoir is the energy supply expense to the power plant. This expense is combined with the pouer plant capital costs and operating expenses to generate the cash flow for the plant. Based on the power plant cash flou

INPUTS Heterologlca1 Data Resource Data Engineering Characteristics

Engineering baracteristics Plant Component Capital Costs Plant,O&H Costs -

INPUTS Financial Parameters for Field Developer Resource Data Field Development Costs Field OhH Costs Exploration and Sunk Costs

Fig. 4. Flowchart

for CCFM.

Discounting Rate-of-Return

Price 0; Fluid

t

-Figures

OUTPUTS Cash Flow for Plant of Economic Merit for Plant

Annual

OUTPUTS Annual Cash Flow for Field Figure3 Of Economic Merit for Field

M!j

i\ comparison of economic cvaluatlon modsi,

and the electricity generated, the unit cost of electricity is calculated, as in the reservoir subprogram. by equating the present worth of the revenues and expenses. GEOCOST can simulate the financial and tax structure of various corporate entities through varying rates of return on equity and debt, the debt-equity ratios, and tax rates. Because GEOCOST operates on raw physical data before actual cost accounts are known. it is most useful when estimating the economic characteristics of a proposed geothermal electric power plant at a yet to be developed field.

(f) GEOCITY GEOCITY is a direct-use, district-heating model developed at PNL. In a manner analogous to GEOCOST it takes a limited amount of physical, climatic, load-use, engineering, and resource data, then generates a design, and simulates the development and operation of the system.” The input data include wellhead fluid conditions, density and area of population zones within the city, description of building heat demands within each zone, and climatic conditions. The design submodels develop system components based on the characteristics of the geothermal resource, the climate, city layout, and buildings to be heated. Capital and operating costs for the components are then determined from cost models. Accounting routines generate the annual cash flows from capital and operating costs and other economic factors. e.g., taxes. interest rate, and rate of return. Using a discounted cash flow analysis, GEOCITY calculates the unit cost for district heating. The model can make use of an internal economic and technical data file consisting of design parameters and cost data. The user may access the provided data or supply any desired changes. The flow chart for the model is shown in Fig. 6. The two main elements of the program are the reservoir and distribution system models. Linkage between these two models is provided b:. the system demand and fluid transmission submodels. Total fluid demand at wellhead conditions is computed in the demand submodel. The transmission submodel calculates the number of wells required, pipe size, pumping requirements, as well as the temperature and pressure loss between the reservoir and the distribution system. After the required total fluid flow is established, the reservoir model determines the cash flow associated with the exploration, development, and operation of the reservoir from the beginning of exploration through the life of the distribution system. The distribution system model designs a piping network for each zone within a city and calculates fluid conditions in each segment of the piping network. The cash flow for the construction and operation of the distribution system is then determined. (The revenue to the reservoir model is the energy cost to the distribution system model.) From the distribution system cash flow, the required revenue and levelized breakeven unit cost of heat are determined. As was the case for GEOCOST, the capability to produce a whole design in the absence of extensive cost information is useful in providing an approximate design of a proposed system.

Field Costs Energy Price INPUTS Meteorological Data Resource Data Engineering Characteristics, Plant Type Well Cost Financial Parameters

Cost of Energy

Plant Development Plant Costs Fig. 5. Flowchart

for GEOCOST.

OUTPUTS Develocment Scenario Electric Plant Design Annual Required Revenue Cash Flow for Field and Plant Levelized Cost of Power

*Field

Field Design

INPUTS MeteorologicalData ResourceData Load-Use Data EngineeringCharacteristics

I

OUTPUTS Detailed System Design Cost of Heat

Total F&d Flow Energy,Cost

Piping Network Design Costs and Cash Flow

Fig. 6. Floucharr

4. DESCRIPTION

i

for CEOCITY

OFTHE

C;\SE STUDIES

The case studies were selected to provide a diverse, representative sample of geothermal applications in which the flexibility and adaptability of the various models could be assessed. The descriptions here are, by necessity. brief; interested readers are urged to consult the references. The one-time and recurring cost accounts for the Heber reservoir case, as well as the associated financial parameters and applicable physical data. are given in the Appendix. The data sets for the other cases are available from the authors. (a) Heber Reservoir The Heber Reservoir is the first part of the overall Heber development.’ When complete, the reservoir field v.ill consist of I3 production wells and 7 injection wells and will provide fluid to a 45 MWe binary electric power plant. When the field is fully developed, the knells will supply 1380GPM of geothermal fluid to the plant. As temperature degrades over time, this flow will be increased to 1600GPM. Development of the field occurs in four separate stages. The first stage, resource proving, consists of acquiring leases to the land, exploring the field. and proving the existence of a viable reservoir. The second stage, permit processing, consists of performing all required environmental assessments and obtaining all permits required to fully develop the resource. The third stage, developin g the resource, consists of drilling all production and injection wells, as well as installing pumps and all required surface facilities. The last stage, operating the reservoir, consists of delivering fluid to the binary plant, performing required operating and maintenance activities, and redrilling wells as required. Field exploration is assumed to have started in 1973 with actual drilling of wells and construction of surface facilities (stage 3) beginning in 1980and lasting 3 years. The reservoir will then be pumped for 30 years. (b) Heber Electric Power Plant The plant modeled in conjunction with the Heber Field is a 65 MWe (gross) binary-cycle plant.’ After accounting for parasitic losses and field consumption, the plant will produce 45 MWe. Construction of the plant is scheduled to last 2 years and plant operation is expected to last 30 years. The conceptual design for the power plant is characterized by a single hydrocarbon binary cycle, tailored from a simple Rankine engine. Heat from the geothermal water is used to change the liquid hydrocarbon working fluid to a supercritical fluid. This fluid is then expanded to drive the turbine-generator. The geothermal water enters a conventional shell and tube heat exchangeer where the temperature is reduced from approximately 182°C(360°F)to about71”C (160°F).The water is then pumped to a reinjection island and reintroduced into the reservoir. The geothermal ivater is never exposed to the ambient air or hydrocarbon working fluid. The net cycle efficiency for the power plant is estimated to be 11%. However, reservoir temperature decline over the plant life will cause a slight reduction in efficiency. Therefore, geothermal fluid flow uill increase from 7 to 9 million lb/hr by the end of the project. The fluid production wells are located adjacent to the power plant to minimize heat and pressure losses during fluid transport. The power plant and production island occupy approximately 20 acres.

805

The proposed development of an ethanol production plant at Raft River. ID would use a medium temperature 138’C (280°F) geothermal resource in the production of ethanol to be used for automotive fuels.” The plant would use agricultural feedstocks. such as sugar beets, potatoes. and wheat. in the production of 20 million gallons of ethanol per year. This is considered to be the largest production level that can be supported by the current agricultural resources of the south central Idaho region. The processing facility would use conventional alcohol technology. Geothermal energy bvould be used for all process heating. with the maximum geothermal fluid requirements being approximately 6000 GPM. It is anticipated that this flow will be supplied by nine production wells located on private and Bureau of Land Management lands in the Raft River Known Geothermal Resource Area. The geothermal fluid will be flashed from 138°C (180°F) in three stages to supply process steam at 121, 107, and 95°C (250,225, and 205°F) for various process needs, such as drying and mash preparation. The steam condensate plus the liquid remaining after the third flash will be reinjected to the reservoir by six injection wells. (d) Susancille District Heating System The City of Susanville, CA is currently implementing a system that nil1 provide IOH.temperature 77°C (170°F) geothermal fluid to I-l public buildings in the city for space and domestic hot water heating.g The heating ivill be done by employing heat exchangers. The system is designed to meet 624 of peak demand. One well uill be used to provide 600GPM to the system. No reinjection is required as the water is of good enough quality to be disposed of in the city server system. The engineering and design phases of the project have been completed. and bids have been solicited for the hardware components and the construction ivork. Susanville is also planning to extend the system to an industrial park consisting of a number of agricultural product businesses that would use the geothermal resource. This second phase is in the planning stage and was not modeled in this study. 5. CASE

STUDY

RESULTS

For most cases. close correspondence between the results of the models was attained. Total agreement was not achieved because of inherent differences in the \vay various items (e.g., discounting conventions) were treated and because of the selective input requirements associated with the different modeling approaches. All dollar results in this section are in 1980 dollars, except for the Heber Reservoir Case, which had a 1973 start date. The net present value figures, unless otherwise noted, were calculated using a 15% discount rate. The GCFM runs in this study were performed at DOE/SAN while the GEOCOST and GEOCITY runs were performed at PNL. The remaining models were run at JPL. DOE/SAN and PNL were provided with input data by JPL. Table 1 shows the cases to Lvhich each model was applied and Table 2 presents a summary of the results. (a) Heber Reset-co;, The Heber Reservoir ivas the most extensively modeled case in this study. It was run on all models except GEOCITY uhich is a district heating model. The runs were based on a 30 mills/kWh price for the geothermal fluid except in the cases where a breakeven price was calculated.

CASE

STUDIES

Heber

Reservoir

Heber

Electric

Raft

River

Susanville

x x

Plant

Ethanol District

Facility Heating

System

"

u

x

x

x

x

x

x

x

x

x

x

x

x

ystem

price

price

--

_-

-_

__

--

15.9% $-3.5n

31.0% $16.&l

GCFH

I

in millions (M) of S, or in

mills/kWh’

21.5% $ l.OH $16.4/HUTU

54.8 __

_-

__

t64.UH’

_-

mills/kWb

_-

$15.5M

CEOCOST

$ 2.09lg.d

117.3

14.0% $-3.3H

28.1% $16.OH

CPC

tAll dollx results :Ire given in IYHO dullilrs excepl for the Heber Reservoir case which h;ul ;I I973 clort date. Levrlizrd bre;lkeven price is given in unils of mills/kWh for rleclricily prducliun and $ per million ISTUs (MUTIJ) for he;lt gencrxtion. $I‘hCX prices ilrr redlevclid imd esc;ll;ile ilt IOR/yr. SSec Scclion 5 lor ;I discussinn of lhis rssull.

breakeven

rate of return llet present value

usanville ietrict

Heating

breakeven

aft River than01 Facility

price

breakeven

(CE0C0ST)

rate of return net present value

rate of return net present value

Plant

eber

I

(GCFll)

Reservoir

eber

CASE STUDIES

VENVAL

‘T;~hlc 2. I
mills/kWh

21.6% t I.UH

$ 2.07/gelt

116.7

$ 0.4H

15.2%

27.7% $15.&l

APSEAM

I

$15.4/wuJ

__

-_

__

_-

CEOCL’I’Y

A comparison of economic evaluation models

807

The analysis by the GPC model showed a present value profit for the Reference Scenario using a 15% discount factor of $15.5 million. The Reference Scenario (which was analyzed by the other models) is the most likely case as defined by GPC. The expected present value of profit for all the scenarios generated was $15.3 million. The VENVAL analysis showed a net present value at 15% of 916.0 million and an internal rate of return of 28.1%. The APSEAM results were quite similar. The net present value was $15.8 million and the internal rate of return was 27.7%. GCFM calculated a 38% rate of return and a net present value of $16.8 million. This higher rate of return can be explained by the way in which the exploration and predevelopment phases of the project are collapsed into a sunk cost figure. When the Heber Reservoir was modeled with the GPC model, VENVAL, and APSEAM, the first year of exploration was chosen as the base year for the analysis, and the predevelopment costs were treated explicitly in the cash flow analysis at the times they occurred. In GCFM, all of the preconstruction costs must be aggregated into a single sunk cost figure which occurs at the beginning of project construction. In this study, that aggregation through time was accomplished using a 15% discount rate. The higher rate of return occurs because the 15% rate used to move the preconstruction costs forward through time does not equal, necessarily, the internal rate of return for the project as a whole. If the internal rate of return for the project is 28% (as was the case here), then to maintain that rate of return, the preconstruction cash flows would have to be moved into the sunk cost figure at a 28% rate. Using 15% causes a portion of the costs to be understated, resulting in a higher rate of return (in this case 38%). Since the true rate of return cannot be known beforehand (unless an analysis similar to that given by VENVAL or APSEAM is performed first), GCFM will either overstate or understate the true project rate of return when the early costs are aggregated as described above. Thus, when evaluating a project from its inception, the GPC model, VENVAL, and APSEAM allow consideration of development costs as they occur. GCFM does not and because of this, distortions in the actual rate of return can occur. GEOCOST. as mentioned earlier, is a plant design model, which as an ancilliary function, generates a field exploration and development scenario. Thus, GEOCOST did not use data from the input data set for the Heber reservoir; rather, the costs and their timing were generated by the model as it worked backward from the plant requirements. The GEOCOST generated field scenario estimated that 25 production and 12 injection wells would be used. This contrasts with the actual field design of 13 and 7 wells. Because wells are the principal cost factor in the development of a field, GEOCOST was run a second time with higher flow rates to yield a 13/7 well scenario. The significant effect of the number of wells on cost can be seen in the fall of the model’s levelized breakeven energy cost from 40.0 mills/kWh in the first (25/12) scenario to 24.4 milllkwh in the second (13/7) scenario. In the second scenario, GEOCOST showed a net present value of $64.0 million. However, GEOCOST does not incorporate escalation in its analysis. This has an impact on the results, as any differential rates of escalation for revenues and the various cost categories will not be correctly incorporated. Additionally, the use of a self-generated field scenario, which differed from the actual cost accounts, precluded a direct comparison of GEOCOST with the other models. The Heber Reservoir case, as shown by GCFM, APSEAM, and VENVAL, obtains a high rate of return. The GPC model confirmed this result. This indicates that the Heber Reservoir project would be highly profitable at a sales price of 30 mills/kWh for the geothermal fluid. (b) Heber Electric Power Plant The Heber Electric Power Plant was modeled by GCFM and GEOCOST. GEOCOST used a set of resource characteristics and climatic data, as well as a limited set of cost and financial data, to generate an entire plant construction and operation scenario and its associated cost accounts. In an analogous fashion, GCFM used an abbreviated set of cost data and likewise generated the cost accounts for a plant construction and operation scenario. VENVAL and APSEAM were also run for this case. However, because a detailed set of cost accounts was not available, the cash flows generated by GCFM and GEOCOST had to be used as inputs for VENVAL and APSEAM. Because of this, VENVAL and APSEAM did not run on the Heber

808

G.\1Z!\I\\ andL.S.Ro?,\I%'R<;

plant: rather. they ran on the GEOCOST and GCFM designs for the plant. Therefore, VENVAL and APSEAM were each run for two different pouer plant designs. The GPC model was not run because the engineering scaling equations for the plant and the required probabilistic estimates were not available. .-\s in the Heber Reservoir Case, GEOCITY was not applicable. GCFM generated cash flolvs for the plant construction and operation. It shoived a net present value of S - 3.5 million and a rate of return of 15.9%. There is a major concern with this result. With a 157~discount factor. the net present value of a cash flowwith an internal rate of return of more than 15% should be positive. This problem may be definitional, requiring clarification by the developers of the model. When the annual cost accounts generated by GCFM were used by VENVAL, the net present value at 13% was S - >.3 million and the internal rate of return was 11.0%. APSEAM showed a SO.4million net present value and a rate of return of 15.2%. GEOCOST generated an entire plant design. This run shoived a net output of 52.8 MWe with an internal plant consumption level of 12.2 MWe. The actual plant design is for a 15.0 MWe net output with an average internal consumption of 20.0 MWe. As mentioned before, GEOCOST did not incorporate escalation into its calculation of costs or revenues. Revenues were constrained to equal costs in order to solve for the levelized cost of electricity output. In the GEOCOST analysis. the project could pay all costs. including a 15% rate of return to capital, with a levelized cost of electrity of 54.8 mills/kWh. The VENVAL and APSE.01 runs based on the GEOCOST analysis did consider escalation of costs over time. Thus. their results are not directly comparable to those of GEOCOST. To model the GEOCOST output. the VENVAL and ,IPSEXM runs \vere based on the following approach.’ Project rate of return is defined as the return on both the equity and debt portions of capital. The discount rate tk) used is the Lveighted after-tax opportunity cost of investment, i.e., k = ( I - 7)kJ,, + k,.j,,.

(3

where k = discount rate, T = tax rate. k,, = market rate on debt. f‘, = fraction of total capital that is debt, k, = after-tax return on equity. and fc = fraction of equity. Using this approach. the VENVAL analysis calculated a nominal-levelized breakeven price of 117.3mills/kWh. lvhile APSEAM yielded 116.7mills!kWh. Breakeven prices exactly cover all project costs including a return to capital over the life of the project. The results based on the GCFM run show that the project would not be greatly profitable. It should be mentioned that several plant costs were somewhat speculative due to a lack of detailed input information as well as to constraints imposed by one of the models. Hoivever, this data was used consistently by each model, and the comparisons, if not the absolute results. are valid. The results based on the GEOCOST run were difficult to assess because escalation was not included by GEOCOST. (c) Raft River Ethanol Faciliry

The Raft River ethanol facility was analyzed by VENVAL and APSEAM. since all required cost account information was available. GEOCOST, an electric power plant design model, was not applicable in this case. Similarly, GCFM was not applciable. The GPC model was not utilized because the information required was not available. Once again, GEOCITY was not applicable. This case was modeled from the standpoint of the equity holders of the project. with interest and the repayment of debt being paid out of revenues. In the analysis performed by VENVAL, it was found that a breakeven price of S2.09igal ivas required to pay a 15% return to equity. Similarly, XPSEAM. showed a S2.07/gal breakeven price for the production of ethanol. Both of these prices are real-levelized and escalate at 10% for the life of the project. The results indicate that this would not be a profitable venture under the conditions assumed since a price of SZ.OO/galfor ethanol would be comparable to a S3.30/gal price for gasoline on an equivalent BTU basis.

A comparison

of economic evaluation models

809

(d) Susanuille District Heating System The Susanville Project was run on VENVAL, APSEQI, and GEOCITY. GEOCOST and GCFM were not run for this case because it was a direct use application. The GPC model was not used because probabilistic estimates and the necessary engineering scaling equations were not available. The data sets for VENVAL and APSEAM were the actual costs for the project obtained from the manager of the Susanville Project. GEOCITY did not use these actual costs but rather operated on a data set consisting of load, climatological, engineering, and physical layout data. The VENVAL run (based on actual expenditures) showed a net present value of $1.0 million and a 21.5% rate ,of return. The APSEAM run showed a very close agreement with VENVAL. The net present value was 91.0 million while the rate of return was 21.6%. The GEOCITY model, working from a completely different data set, generated a system design that showed a cost for heat of %15.4/MBTU. The VENVAL cost accounts, when levelized, showed a breakeven cost of $16.4/MBTU. It should be noted that the piping costs for the Susanville system were not identically replicated in GEOCITY. Due to data constraints, the GEOCITY runs used steel pipe costs (which were already contained in the GEOCITY data base), while the actual system is employing asbestos concrete piping. The results for this case show that the project could earn a 21% rate of return which is fairly attractive.

6. SUMMARY

AND CONCLL’SIONS

X11 the models assessed include an economic evaluation capability. They do. however, differ in the following ways: VENVAL and APSEAM only perform economic analyses (VENV&AL also includes a Monte Carlo capability). GEOCOST produces field and plant specifications. However, its economic capability did not include escalation and its design did not correspond to an actual project. GCFM also includes field and plant design; however, difficulties were encountered in interpreting its economic output. GEOCITY is designed only for direct use, district heating. Finally, the GPC model could perform conditional probabilistic analyses of project costs for a geothermal field. As seen in Section 5, it was possible, with some exceptions, to achieve a close correspondence between the results of the various models. The differences that did persist can be attributed to several factors. Some can be attributed to the type of methodology employed by each model. That is, differences in approach result in differing and selective input requirements. Additionally, the varying approaches can restrict the use of a given model based on considerations of data availability. For instance, in the absence of detailed cost accounts for the Heber Electric Power Plant case, VENVAL and APSEAM had to be run on designs created by GEOCOST and GCFM. It should also be stated that these designs did not always have a close correspondence with the actual, real-life project. Finally, inherent differences in the way various factors are treated by the models can create discrepancies. Discounting conventions might be dissimilar, escalation routines for the cost accounts might be omitted, and tax assumptions could vary considerably. For example, GCFM considers a field or plant project as a self-contained entity and thus does not pass any tax losses through to a parent corporation. On the other hand, in VENVAL and APSEAM, the field development and plant construction were assumed to be undertaken by large corporate entities and thus any losses and associated tax benefits were assumed to be passed through to the balance sheet of the parent corporation. The effect of this is to reduce tax liabilities earlier in the cash flow, raising income for those years. In another example. GCFM aggregates preconstruction costs into a single sunk cost figure which occurs at the beginning of project construction. Meanwhile, in the GPC model, VENV.-\L, and APSEAM, these early stages are treated explicitly and are included in the cost analysis when they occur. While the variations in approach, input requirements, and the treatment of various factors resulted in a number of differences, these could in most cases be explained. However, most importantly, some differences were quite significant and would not have been revealed had a comparison between the models not been made.

ECY Vol. 8. Vu IO-E

610

G. M. Zrv~?r and L. S. ROSESBERG REFERENCES

I. J. 3. Muddiman and J. W. Whelan. “VENV;\L L’sers Slanual".DOEIR.V0OOOl~TS Vol. II. DuPont Semours and Co., Wilmington. DE 19801 (July 1980). 7 “The Geothermal Probabilistic _. L. H. Orren, G. M. Ziman, S. C. Jones, T. K. Lee, R. Nell, L. Wilde, and V. Sandanand. Cost Studv”. 5030-491. Jet Propulsion Laboratory. Pasadena. CX 91109 (.Q$ 1981). 3. R. B. Davis and 1. V. V. Kasper, *.Overvie\c of the .-\lternative Power Systems Economic Analysis Siodel”, Proc. Yat. Computer Conf.. Vol. 49. p. 335. AFIPS Press. Arlington, VA 22209 (1980). -1. hi. .\. Keimig. 1.1. Rosenberg, and D. J. Entigh. “Geothermal Loan Guarantee Cash Row Model: Description and Users Guide”, DOE/SFilO494-I, The MITRE Corp.. McLean, V.-2 22101 [Nov. 1980). 5. C. H. Bloomster. “Economic Analysis of Geothermal Energy”, BNWL-SA-5596, Battefle Pacific Northwest Lnboratories. Richland. WA 99352 (July 19%). 6. C. L. McDonald and C. H. Bloomster. “The GEOCITY Sfodel: Description and Applications”. B.c’WL-S.+63-13. Battelle Pacific Northwest Laboratories. Richland. WA 99352 (June 1977). 1. 1. W. Doane. R. P. O’Toole, R. G. Chamberlain, P. B. Bos. and P. D. Maycock, .‘The Cost of Energy from Utility-Owned Solar Electric Systems”. _5040-29. Jet Propulsion Laboratory. Pasadena. CA 91109 (June 1976). 8. R. .A. Stenzel. J. Yu. T. E. Lindemuth. R. Soo-Hoo. S. C. May, Y. J. Yim. and E. H. Houle. “Ethanol Production for Automotive Fuel Usage”. DOEiID/11050-3, The Bechtel Group Inc.. San Francisco. CA 94105 (Aug. 1980). 9. Cost. financial, engineering, and design data were provided by Philip Edwardes of the City of Sujanville and Ken Unmack of the Aerojet Corporation (June 1981).

.\PPENDIX

h

Table Al. Reference scenario cost by stage. 1973 k$ STAGS I - Recwurce Proving (6 years)

Exploration G6A

2,013 55/yr

Lease ecquieition Surface occup*ney Cootingency STAGE II Rent

-

Permit

17d 33 277

Processing

(1

year) 41

EnvironmenceL G&A Contingency

arseasment

109 55 21

STAGE III - Resource Devclopmenc Rent Uell drilling Surface inetelldtion CAA ingeocy

(3

yeer.)

I

34lyr 10.348 5.142 ‘+b/yr

Cont

STAGE IV - Reservoir 06n Pumping Colts Redrilling of Contingency

Operation

538 (30

yeers) 1.145

(*tart)

434/y= 10,897 (year 114/yr

wellr

15)

Table A2. Financial parameters Fluid price Required rate of Debt/equity rstio Fcderel tex rate Scste Locel

tax tax

return

after

30 oillr/kWh 15x 01 LOO 462

fex

rete rate

9f LX

Inveetment tex credit Royalty race (reservoir only) Depletion eLlov*ncc (reservoir Depreciation wehod Eeceletion of energy price Generet ion inf let ion

only)

LOX LOX of groes rcve*ue 152 SUm of yeerl digiCe 10% 9%

Table A3. Field design data used for GCFM.

_

Gmerel Dete Beginning of Project life Power-on-line Plent type Field Brine

conetrucrion (yeare) year

ourpuc ti%e) flow req. (k

Lb/hr)

(yeer)

1980

33 1983 Binary 65 7140

811

A comparison of economic evaluation models Table A3 (Cont.).

Well Well

flow cost

Book life Tax life Intangible Coat per

rate (1980

(k lb/hr) k$)

drilling cost rate makeup well (1980 k$)

Field Plant Data Field plant capital

(1980

Other Data Operating (1980 Sunk wets-construction COneCrUCtiOn invcltwnt

colts

1020

549 896 33 10 .75

I

1043 33 10 .75

945

k$)

945

1921

I

k$)

34241~~ 6849 33Xlyr

(1980 k$) breakdown

Table A4. Additional field financial data used for GCFM. Electricity

price

(milla/kYh)

Fluid price (mills/kUh) Capacity factor Hioimm equity fraction Equity rate of return General inflation Escalation for capital Escalation rate - electricity Discount rate

rate

I I

rate

and

06H costa

and

fluid

price

I

60 30 702 100% 15% 9%

9% 10x 152

250 33 10