1. How to Build an Energy Market Modeling System
Presented at the 40th International IAEE Conference
June 18-21, 2017
Singapore
Robert Brooks, PhD, RBAC, Inc.

2. Presentation Overview
Scope Approach Application
Presentation Overview

3. Scope of Presentation There are many types of energy models
Focus today is on those models which tie together energy suppliers with energy consumers using a transportation network
Usually described as an energy market model
Also known as a partial equilibrium model of a single energy commodity such as crude oil, natural gas, or electricity
Sometimes multiple energy commodities can be included in the same model
But also focusing on building a modeling system not simply a “model”

4. Presentation Overview
Scope Approach Application
Presentation Overview

5. Modeling System Design Approach
Construct the conceptual design
Node-arc network diagrams, input-output drawings, etc.
Write the model using a modeling language such as AMPL
Programmatically create the application
Test and iterate

6. Global Gas Geo-Political Unit (Gpu) Model
Consumption
Pipeline Exports
LNG Exports
 Injection
 Re-gas
Gas Storage
LNG Storage
Withdrawal 
Liquefaction 
Pipeline Imports
LNG Imports
Gas Production

7. Global Gas Pipeline Model
Pipe A deliveries to Gpu 3
Pipe A deliveries to Gpu 2 N
Pipe Intx
Segment N (Pipe B) 3
Segment 1 (Pipe A)
Pipe Link 2-3
Segment 3 (Pipe A) 1 2
Pipe Link 1-2
Pipe A deliveries to Gpu 4
Segment 2 (Pipe A)
Pipe Link 2-4 4
Pipe A receipts from Gpu 1
Segment 4 (Pipe A)

8. Global Gas LNG Flow Model
LNG Tanker Shipment from Gpu 1 to Gpu 2 2
Import Region (Gpu 2) 1
LNG Tanker diversion from Gpu 3 to Gpu 2
Liquefaction / Export Region (Gpu 1)
LNG Tanker Shipment from Gpu 1 to Gpu 3 3
Import Region (Gpu 3)

9. LNG Transport Cost Model

10. Modeling System Design Approach
Construct the conceptual design
Write the model using a modeling language such as AMPL
Entities: the model’s geographical areas or locations, commodities, facilities, transport modes, periods, etc.
Level and flow variables: the quantities forecast to be made, stored, moved and consumed
Entity data: capacities, costs, yields, loss factors, conversion rates, etc.
Constraints or limitations: production or flow capacities, mass balance at storage facilities and terminals, etc.
Economic objective: market clearing, supply cost minimization, profit maximization, etc.
Programmatically create the application
Test and iterate

11. AMPL Objects Sets Parameters Variables Constraints Objective
Define the basic entities in the model
Parameters
The numerical data representing supply and demand curves, processing, transportation, and storage costs, capacities, yields & losses
Variables
The “decision variables” of the model: levels and flows for production, transport, storage, and consumption of the energy commodities of the model
Constraints
Equations or inequalities describing the limitations on commodity levels and flows, activity matrices defining operations on those commodities such as refining, chemical conversion, etc.
Objective
What is the criterion (or criteria) operating in this market which determine the “best” or most likely economic outcome?

12. AMPL Model Code: Sets
# Sets == indices include geo-political units (GPU) such as countries, periods (months), pipelines, and multi-dimensional combinations of these
set GPU; # countries or sub-country areas
set ORIGIN within {GPU}; # origin gpu for LNG or pipeline
set DEST within {GPU}; # destination gpu for LNG or pipeline
set PIPELINE; # inter-regional pipes
set PERIOD; # this will be integers from 1 to N where N is the number of periods in the scenario
set LNGLINKPER within {ORIGIN,DEST,PERIOD}; # compound set of origin gpu to destination gpu in each time period

13. AMPL Model Code: Parameters
# Parameters: costs, capacities, loss factors, distances associated with flows of gas and LNG; conversion factors for gas to LNG, and LNG to gas; and other numeric values used in constraints
# for example, the number of days in the month for each forecast period
param dpm{PERIOD} >=0; # days per month
# and parameters used in conversions of units of measure
param LNGEXPRATIO default 585; # cubic meters gas per cubic meter liquid (LNG)
param F3perM3 default ; # cubic feet per cubic meter
param LNG2Gas = LNGEXPRATIO * F3perM3 / ; # mmcf gas per cubic meter of LNG
# and parameters used in computing LNG travel time
param lnglink_distance {LNGLINKPER} >=0; # distance from origin to destination by tanker
param lnglink_canaltime {LNGLINKPER} >=0; # uom=days, number of days spent transiting either Suez or Panama canal (zero if no canal)

14. AMPL Model Code: Variables
Variables: the quantities for which optimal values are to be determined, which typically represent flows of natural gas and LNG throughout the global network
var pipesegmentuse_qty{PIPESEGMENT}>=0;
var pipelinkuse_qty{PIPELINK}>=0;
var pipercpts_qty{PIPESUP}>=0;
var pipedelivs_qty{PIPEDEM}>=0;

15. AMPL Model Code: Constraints
limitations on flow variables such as upper bounds (capacities)
balance equations for each node (what comes in is what goes out)
conversion equations (gas to LNG, LNG to gas)
useful definitional equations
The name of the constraint represents its dual variable; e.g. the marginal value of capacity increase
Examples:
pipesegmentuse_mvcap{(pipeline,gpu,t) in PIPESEGMENT}: pipesegmentuse_qty[pipeline,gpu,t] <= pipesegment_cap[pipeline,gpu,t] * dpm[t] ;
gasstorage_price{(gpu,t) in GASSTOR}: gasstorage_endstor[gpu,t] = gasstorage_endstor[gpu,t-1] + gasstorage_injqty[gpu,t] - gasstorage_wdlqty[gpu,t] ;

16. AMPL Model Code: Objective Function
maximize EconValue: sum{(gpu,sector,t,seg) in DEMSEG,s in DEMDSTEP} (demand_price1[gpu,sector,t,seg]+((s-1)/nsteps) * (demand_price2[gpu,sector,t,seg]-demand_price1[gpu,sector,t,seg])) * gasdelivsstep_qty[gpu,sector,t,seg,s] - ( sum{(gpu,gtype,t,seg) in SUPSEG, s in PRODSTEP} (supply_price1[gpu,gtype,t,seg]+((s-1) / nsteps) * (supply_price2[gpu,gtype,t,seg]-supply_price1[gpu,gtype,t,seg])) * gasprodstep_qty[gpu,gtype,t,seg,s] + sum{(gpu,t) in GPUPERIOD} gasflowdata_cost[gpu,t]*gasflow_qty[gpu,t] + sum{(pipeline,gpu,t) in PIPESEGMENT} pipesegment_cost[pipeline,gpu,t] * pipesegmentuse_qty[pipeline,gpu,t] + … )

17. Modeling System Design Approach
Construct the conceptual design
Write the model using a modeling language such as AMPL
Programmatically create the application
Create the required I/O tables in a relational database
Create the user interface for the application (based on “empty” template)
Create test problems to debug the model
Test and iterate

18. Programmatically Create the New App
Function CreateAll()
Call CreateAMPLModelTable
Call CreateDatabaseFromAMPLModelFile
Call WriteInputScript
Call WriteOutputScript
Call PopulateTablelist
Call PopulateRelations
Call RestoreRelations
Call CreateBalanceTables
Call CreateAggregationFields
Call MakeFormQueries
Call BuildForms
Call LoadTestScenario
End Function

19. New App Database Tables and Relations

20. New App User Interface

21. Generated Test Problems
Geographical entities (gpu’s)
Gas supply curves
Gas demand curves
Pipeline segments, supply, demand & interconnect links
LNG production capacities and costs
LNG re-gas capacities and costs
Tanker classes and inventories
Tanker day rates, fuel use, fuel cost, etc.
Maritime distances between LNG origins & destinations

22. Modeling System Design Approach
Construct the conceptual design
Write the model using a modeling language such as AMPL
Programmatically create the application
Test and iterate
Run test problems and examine the results
Modify AMPL code, re-create application, run tests, and analyze results until the model is working exactly as desired

23. Tabular Reports

24. Graphical Reports

25. Presentation Overview
Scope Approach Application
Presentation Overview

26. Application: G2M2 Global Gas Market Model
Monthly time frame: permits modeling of seasonality
Scenario forecast horizon: from now to 2050
Can “back-cast” from 2006 to now
Coverage: more than 100 countries represented
All natural gas producing and /or consuming countries
All LNG exporting and / or importing countries
Large markets divided into sub-country areas
USA -> Census Regions
Canada, Mexico, Russia -> East and West
Nearly 400 gas pipelines: existing and proposed
Multiple LNG tankers classes
Q-Max, Q-Flex, conventional grouped by size

27. G2M2 Data Sources
International Group of Liquefied Natural Gas Imports (GIIGNL)
International Gas Union (IGU)
Gas Infrastructure Europe (GIE)
U.S. Energy Information Agency (EIA)
U.S. Federal Energy Regulatory Commission (FERC)
U.S. Department of Energy (DOE)
Canada National Energy Board (NEB)
LNG Journal
LNG Company websites
Panama Canal Authority (ACP)
Suez Canal Authority (SCA)
Timera Energy
BP Statistical Review
International Energy Agency (IEA)
Joint Organizations Data Initiative (JODI)
European Network of Transmission System Operators for Gas (ENTSOG)
East European Gas Analysis (EEGA)
Gazprom and other company websites

28. How is G2M2 used?
Scenario Design: Forecasting potential impacts of various conditions and events on gas and LNG production, transportation, pricing and demand
Growing shale gas production in new plays and countries
Proposed new gas pipelines or expansions of existing ones
New or expanded LNG import & export terminals
Demand sensitivity to weather and gas price
Effect of economic growth changes on demand
Effect of governmental policies
Scenario Evaluation
Built-in reports
Portal to Excel

29. What issues can G2M2 address?
Natural gas production: How fast & where will it grow?
World gas prices: Will they converge? if so, when?
LNG market share: How will LNG fare against local production and pipeline imports?
North American LNG: Can the US and Canada compete against Qatar, Australia, Russia?
New capacity: LNG liquefaction, re-gas, and tankers; new gas pipelines: how much is needed to satisfy future market growth?

30. Where will gas production grow?

31. Will global prices converge?

32. How much market share for LNG?

33. Can North America compete?

34. Can North America compete?

35. Will new LNG tanker capacity be needed?

36. Thank you! Questions?

37. Contact Information
Robert Brooks, Ph.D. Founder
Liam Leahy, Chief Executive Officer
James Brooks, Director Business Development
Bethel King, Senior Director, Market Analysis
Jill Quick, Senior Analyst
Contact Numbers
Administration (281)
Contracts and Sales (281) ext. 126
More information:

38. About RBAC
RBAC Inc. licenses economic forecasting tools the energy industry, as well as State and Federal government agencies involved with Energy, Transportation and the Environment. RBAC’s principal products include the industry standard GPCM® Natural Gas Market Forecasting System™, the GPCM® Base Case Database for North America, and GPCM Viewpoints® on Natural Gas.
We continuously advance our modeling tools through technology, feature development and regularly released updates based on client requests and energy industry needs. Those needs resulted in our development of RBAC’s North American Natural Gas Liquids Model (NGL-NA™) and GPCM Daily™.
Additional forecasting tools scheduled for release include:
G2M2® Global Gas Market Modeling System .Net Ver. 3.5
GPCM® for MS-SQL
Licensees of RBAC’s systems are involved in natural gas exploration and production, LNG infrastructure development and marketing, natural gas marketing and transportation, electric power generation, natural gas distribution, and commodities trading, as well as most of the major consulting firms that service business and planning needs of the energy industry and its bankers.
Dr. Robert Brooks founded RBAC in 1987 based on experience developing several well respected predictive models since his first work on his doctoral research in natural gas transportation economics at MIT in the 70’s.
Designing forecasting tools for global energy market prices, basis, and flows is RBAC’s core business.
RBAC’s staff includes experts in natural gas supply, demand, marketing and transportation as well as the dynamic global pipeline and LNG markets. Our team applies its world class expertise in mathematical modeling, statistical analysis, mathematical algorithm development, software engineering, and database design to current and future challenges, risks and opportunities in energy.