Loading...

A Risk-averse Policy for the Network Capacity Control Problem

Jalali, Zahra | 2016

358 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 48946 (01)
  4. University: Sharif University of Technology
  5. Department: Industrial Engineering
  6. Advisor(s): Modarres Yazdi, Mohammad
  7. Abstract:
  8. This thesis focuses on the network capacity control problem while, risk-averse criteria is considered for decision making. Network revenue management models have wide applications in various industries such as airlines, hotels, railways, rental halls due to handling multiple resources. All models in the related literature are developed based on risk-neutral criteria. However, in this thesis we consider risk-averse policies for many reasons such as risk reduction, covering fixed costs rather than maximizing revenue, and earning some specific revenue within a specific time period. We apply dynamic programming by adopting two types of utility functions: 1) additive and time-separable utility function, and 2) negative exponential utility function. We show the model which is developed based on the first utility function preserves most features of the similar models with risk-neutral criteria. Therefore, the “bid-price” control policy which is commonly used in the risk-neutral literature is applied in this case too. This policy is calculated by an affine approximation of value function which results in a linear programing model. Numerical results show that using the risk-averse policy increases the expected utility compared with risk-neutral policies. The proposed dynamic programming model with negative exponential utility function has completely different structure from risk-neutral models. Thus, another approximation dynamic programming algorithm is developed to obtain a risk-averse policy. Our numerical results for this model show that the risk-averse policy reduces both the variance and mean of revenue over various time horizons, simultaneously
  9. Keywords:
  10. Capacity Control ; Revenue Management ; Dynamic Programming ; Network Revenue Management ; Risk-Averse Policy ; Exponential Utility Function ; Additive Utility Function

 Digital Object List

 Bookmark

No TOC