Energy Use in Distillation Operation: Nonlinear Economic Effects IETC 2010 Spring Meeting

2010 IETC Meeting PresenterPresenter Doug White Principal Consultant PlantWeb Solutions Group Emerson Process Management Houston, Texas

2010 IETC Meeting Distillation Energy Impact  Over distillation/ fractionation columns in the US alone  Consume 40% - 60% of the total energy used in chemical and refining plants  Consume 19% of the total energy used in manufacturing plants in the US Reference: Office of Industrial Technology: Energy Efficiency and Renewable Energy; US Department of Energy Washington, DC “Distillation Column Modeling Tools”

2010 IETC Meeting Presentation Objectives  Present general approaches to saving energy in fractionation/ distillation through improved control  Present techniques for economic analysis that recognize non-linear character of distillation operation and effects of product blending

2010 IETC Meeting PC FC LC FC TC FC LC Feed, F Bottoms, B Distillate, D Reflux, R Reboiler, E AC AR Steam Gas CW Typical Distillation Column

2010 IETC Meeting Traditional Control Benefit Analysis Improved Profit By Changing Target Better Control, Reduced Variability Poor Control Product Composition ($/ Day Profit) Specification Limit Time Operating Targets When is this valid? When is it not?

2010 IETC Meeting Representation of Variability Product Composition Specification Limit Time Frequency of Occurrence Composition Mean Gaussian Distribution

2010 IETC Meeting Effect of Variability – Linear Objective Function LimitProduct Value; $/ Day Expected Values Move Average Closer To Limit To Increase Value Composition Original Distribution Projected Distribution Valuation Function No Benefit For Better Control At Constant Setpoint!

2010 IETC Meeting PC FC LC FC TC FC LC Feed, F 20,000 BPD $60/ Bbl Bottoms, B 5%C4; $60/ Bbl Distillate, D 3 %C5; $40/ Bbl Reflux, R Reboiler, E AC Case Study – Debutanizer Column AR C3 – 25% nC4 – 25% nC5 – 25% nC6 – 25% Steam 15$/MMBTU

2010 IETC Meeting Case Study – Typical Tiered Pricing With Composition < 3%C5; $60/ Bbl >3%C5; $40/ Bbl < 5%C4; $80/ Bbl > 5%C4; $60/ Bbl On - Spec Product Off - Spec Product

Impact of Material Balance Variability

2010 IETC Meeting Operating Margin – Bottoms Compositional Change – Constant Reflux – No Control Variability Top Product On Spec Bottom Product Off Spec

2010 IETC Meeting Operating Margin – Control Variability Impact – Base Case Spec Initial Operating Target Initial Mean Value Initial Variability

2010 IETC Meeting Operating Margin – Improved Control – Reduced Variability Case Spec Same Operating Target New Variability New Mean Value Increased Margin Improved Control Yields Value At Constant Setpoint!

2010 IETC Meeting Operating Margin – Optimum Target Composition Versus Control Performance Std Dev Optimum Setpoint Optimum Target For Composition Varies with Control Performance and is NOT at the limit!

Energy Balance Control

2010 IETC Meeting Low Energy Cost Optimum Reflux/ Reboiler Operating Margin, $/ Day Low Energy Cost, $/day Product Value, $/day Low Energy Cost Margin $/day High Energy Cost, $/day High Energy Cost Margin $/day High Energy Cost Optimum Min Reflux High Purity Specifications Distillation – Energy and Margin

2010 IETC Meeting Energy Cost versus Reflux Change – Constant Bottom Composition

2010 IETC Meeting Operating Margin – Optimum with Varying Energy Pricing Top Product Specification Limit Control Target Changes from Composition To Reflux (Energy) Depending on Relative Prices

2010 IETC Meeting Non-Linear Objective Functions – Impact of Variability For nonlinear relationship, the expected value of the energy cost is NOT at the value equivalent to the median of the composition; It’s value depends on the standard deviation of the composition Energy Cost Composition Less PureMore Pure Probability Distribution Expected Value

2010 IETC Meeting Energy Cost – Effect of Control Variability New Variability New Mean Value Initial Variability Initial Mean Value Reduced Energy

2010 IETC Meeting Effect of Blending Column Product Shipped Product Proposition: Since actual specification is on shipped product rather than column product directly, small excursions over the specification don’t matter and can be handled by blending. Is this correct?

2010 IETC Meeting Energy Cost – Impact of Control Performance Better Control Performance Pays Even With Blending

Pressure Effects

2010 IETC Meeting Energy Cost – Operating Pressure Impact – Constant Top and Bottom Product Compositions

Non – Symmetric Distributions

2010 IETC Meeting High Purity Columns Often Have Non- Symmetric Compositional Distributions. Aromatics Column Data Data Gaussian Gumbel Gumbel is a two parameter statistical distribution which often fits non- symmetric data well

2010 IETC Meeting Summary – Distillation Economics - Conclusions  For practical cases with tiered product pricing the optimum composition target may not be at the maximum impurity limit  The optimum energy usage depends on energy pricing and may be shift from constrained to unconstrained  Even with product blending there is an incentive for better control performance  Minimizing pressure continues to have value for many separations  High purity columns often have non-symmetric compositional distributions – require special statistical analysis beyond Gaussian distribution assumptions