John Jarvis, Claudia Johnson

1

John Jarvis, Claudia Johnson Liana Vetter
October 26, 2004
2
Presentation Overview

• Quest Resource Corporation
• Model Development
• Model Implementation Results

3
Quest Resource Corporation

• An oil and gas company whose core business is
  developing, producing and transporting natural
  gas

4
Pipeline Schematic
Pipeline
Well head/site
Delivery/Sale Point
5
Place in the Market
Quest
Pipeline Transportation
Purchaser
In-house Use
External Sales
6
Quests Financial Setting

• Revenues of about 11.7 million
• Access to a 150 million debt facility for
  future opportunities
• Over 900 miles of active pipeline transporting
  gas to sale points, with further construction
  underway.
• 380 wells planned to be drilled in 2005 in
  addition to 900 miles of pipeline construction.

7

• Once discovered, the gas is extracted from the
  Earth and run through condensers that increase
  the pressure of the gas so that its flow through
  the pipelines is swift to one of 10 sale
  points, or gas meters, whereby the ownership is
  transferred to the gas Purchaser

8
Quest Resource Corporation

• Agreements are made between Quest and the
  Purchaser on amounts of gas to be delivered from
  Quest via pipeline
• That monthly agreement distinguishes a daily
  amount guaranteed by Quest to the Purchaser
  (contractual gas) from gas sold at the daily
  market price (swing gas)
• MS/OR methods are used to optimize the solution
  the amount Quest guarantees -- and minimize the
  risk to Quest when guaranteeing gas to the
  Purchaser

9
Current Approach

• Quests current gas marketing strategy
• Sales of gas production
• 85 (anticipated total produced gas) guaranteed
  monthly by Quest
• The remainder sold daily (swing volume) via
  market price
• Pipeline serves as middleman
• Total produced gas gas sold on contract
  gas sold daily

10
Goals of the Project

• Analyze the market trends and forecasting
  accuracy of Quest
• Determine what percentage is optimal to guarantee
  on contract
• Create optimization model Quest can use monthly

11
Model Development
12
Sale Points Evaluated

• Two different sale points
• RH large and unstable
• Housel small and unstable
• Historical data
• Forecasted daily production by sale point (2004)
• Actual daily production by sale point (2004)
• Daily NYMEX prices (2002-2004)

13
Market Prices 2002-2004
14
RH Sale Point Production 2004
15
Problem Constraints

• Maximum days and amount in debt
• Set limit of 2 days in debt based on 2004 data
• Set limit of 10 of production in debt
• Conservative limits to minimize risk in case of
  unexpected changes in production
• Bounds on percentage to guarantee
• Set upper limit as 95, highest Quest has used
• Set lower limit as 30 to protect against sharp
  decrease in production

16
Model Formulation

• Zi Production May vary due to equipment
  failure, geological variations, etc.
• X Forecasted production amount.
• Y Contractual amount decision variable.
• Pi Market Price Affected by many outside
  factors (see NYMEX).

17
Market Price

• Pi Daily price assumption
• Pi P0 (adjustment i)
• P0 initial market price (NYMEX)

1.09
For up-market scenario
1
Adjustments
i
For down-market scenario
0.95
18
Model Formulation

• Over the course of a month, with each day i
• Zi actual units of gas (MCF) produced
• If Zi Y, then, deliver all gas on contract
• If Zi ltY, then, Quest must borrow
  difference from pipeline
• Else Zi gtY, then, Quest repays debt to
  pipeline first, then sells remainder at
  daily market price

Zi production Pi market price Y
contractual X forecasted amt
19
Description of Regret

• Regret difference between optimal revenue and
  actual revenue
• Benefits of regret
• Solution does well in rising and falling market
• Less sensitive to predicted probabilities

20
Market Scenarios

• Up Market Scenario
• Optimal solution has Yup minimum
• Put least amount possible on contract, rest on
  swing volume
• Regretup (Revenueup) - (Revenue)
• Down Market Scenario
• Optimal solution for Ydn maximum
• Put maximum possible on contract, rest on swing
  volume
• Regretdn (Revenuedn) - (Revenue)

21
Regret Objective

• UpRegret Revenueup(Yup) Revenueup(Y)
• DnRegret Revenuedn(Ydn) Revenuedn(Y)
• Min prob(up) UpRegret prob(dn) DnRegret

Zi production Pi market price Y
contractual X forecasted amt
22
Computer Implementation

• User inputs
• Probability the market will rise
• Sale point
• Month to forecast, days in month
• Expected initial NYMEX price
• Forecasted daily production
• Expected beginning debt
• Program output
• Data file for AMPL
• Can be run with regret model to resolve each month

23
Stochastic Model with Regret

• Example case
• User provided data
• Model returns output
• Expected revenue 1,037

24
Sensitivity Analysis

• Optimal monthly guarantee varies little when
  expected production data changes
• Model is more sensitive to changes in market data

25
Model Implementation and Results
26
Analysis and Recommendation

• 50-55 should be guaranteed monthly if no market
  predictions added from Quest
• Consequences of guaranteeing 50-55
• 18,000 additional revenue from January March
  2004 for RH
• 2,400 additional revenue from January March
  2004 for Housel
• Regret model yields more profit than current
  Quest marketing and provides more consistency
  between months

27
Problems and Limitations

• Problems encountered
• Limited historical data
• Multiple daily gas prices (strip price used)
• Large variability of the gas market
• Difference in production records from meter
  inconsistency
• Limitations of the solution
• Dependent on the market, which is unpredictable
• Stochastic variables are based on limited data

28
Letter from Quest

• Thank you for allowing your students to assist us
  on this project. The process we went through was
  in itself beneficial. They have provided us
  information and analysis that we found to be
  helpful and even somewhat unexpected. The
  program they have given us should provide a
  firmer basis for our decision making for gas
  marketing. It should get better as time passes
  and we are better able to provide historical
  information for it. It was an educational
  experience for all parties concerned. Thank you
  for sharing them with us.
• Richard MarlinQuest Cherokee, LLC5901 N.
  Western, Suite 200Oklahoma City, Ok. 73118

29

Questions
30
Normalized Objective Function

• Min prob(up) (UpRegret / Revenueup(Yup))
  prob(dn) (DnRegret / Revenuedn(Ydn))