Quest Resource: Natural Gas Marketing
Award Winner: Omega Rho National
Student Project Competition
Client: Quest Resource, Inc.
Team: John Jarvis, Claudia Johnson, Liana Vetter
Faculty advisor: Barr Year: 2004
Documents: Final report (Word), final presentation (PPT)
Currently, Quest guarantees about 85% of its gas through monthly contracts, while selling the remaining at the daily price. The amount to guarantee per sale point per month is a major decision within Quest. The motivation for Quest to optimize this process is two-fold. First, by paying close attention to past production, optimization can help assure that Quest will rarely produce under their monthly guarantee and will never incur a penalty for that underproduction. Second, through researching the historical relationship between Quest’s contract prices and the corresponding market prices for each month, Quest can better estimate the contract volume that will maximize revenue.
There are several important decisions involved in seeking a solution. Bounds must be set for minimum and maximum amounts to be sold by Quest on contract. Likewise, limits on both the length and depth of underproduction allowed must be chosen. The effect of a rising market versus a falling market is important. Also, the accuracy of forecasting versus actual production must be assessed. The implementation of the model will allow Quest Resource to manage their marketing schedule with a greater knowledge. Through selling at optimal levels, Quest will materialize greater revenue in selling its gas.
To perform this analysis, we created a series of three linear programming models that build upon each other. We chose to evaluate the sale points separately since each one has different production levels and may have more or less flexibility in terms of debt allowed. Of the ten sale points operated by Quest, we evaluated two, R&H and Housel. Both are relatively unstable sale points. R&H represents the larger sale points; Housel represents the smaller sale points. Data from January – March 2004 production was used as well as market trends from 2002 – 2004.
The challenges we faced in creating a model were 1) the inability to forecast perfectly, 2) the desire to minimize variation between months, and 3) the goal to minimize risk by finding a solution that will perform well in rising and falling markets. Our third model, the stochastic model with regret, achieves these goals. By incorporating future expectations, it can be used without actual production and daily prices for the month. Instead of maximizing profit this model maximizes expected profit. It also provides a robust solution – i.e. a middle-ground solution – that will enable Quest to do well in both a rising and falling market.
Our analysis yields the recommendation that without client input about market predictions, about 50% – 55% of the forecasted gas should be guaranteed on contract. If this model had been implemented from January through March, Quest would have made an additional $18,000 from the R&H sale point and an additional $2,400 from the Housel sale point. With client input concerning market trends, the percentage will change monthly and expected revenue will further increase.